Artigos - Doenças Crônicas.pdf
Documento PDF (2.9MB)

Exibir conteúdo Baixar

                    Original Investigation | Psychiatry

Mortality Risk Following a Household Suicide
Flávia Alves, PhD; Elisângela da Silva Rodrigues, PhD; Lidiane Toledo, PhD; Júlia M. Pescarini, PhD; Rodrigo Lins Rodrigues, PhD; John A. Naslund, PhD;
Maurício L. Barreto, MD, PhD; Vikram Patel, MD, PhD; Daiane B. Machado, PhD

Abstract
IMPORTANCE The broader mortality risks faced by household members following a suicide remain
poorly understood, particularly in low- and middle-income countries.
OBJECTIVES To estimate the risk of all-cause and cause-specific mortality among surviving
household members after a suicide within the household and to identify individual and

Key Points
Question Does household exposure to
suicide increase mortality among
surviving members, and what risk
factors and temporal patterns are
involved?

contextual factors associated with mortality risk using data from a national population-based

Findings In this cohort study of over

cohort in Brazil.

100 million Brazilians, household suicide
exposure was associated with a 32%

DESIGN, SETTING, AND PARTICIPANTS This nationwide cohort study used data from the 100

higher risk of all-cause mortality and a

Million Brazilian Cohort linked to the Mortality Information System (SIM) (2001-2018). All individuals

4-fold higher risk of suicide, with over

who lived in a household with a suicide index case were considered exposed. Data were analyzed

half of suicides occurred within 2 years

from March 2023 to September 2025.

following the index case. Risks were
greatest when the index cases were

MAIN OUTCOMES AND MEASURES All-cause mortality, cause-specific mortality, and suicide.

young or female and among those in

Outcomes were derived from the national mortality tracking system. The adjusted hazard ratio of

poor housing conditions.

all-cause and specific-cause mortality was calculated using a multivariate, time-varying Cox
regression. The risk factors associated with increased mortality, including characteristics of the index
case, surviving household members, household conditions, and the timing since the suicide event
were also analyzed.

Meaning These findings suggest that
household suicide exposure confers
substantial time-sensitive mortality
risks, underscoring the need for
targeted early postvention strategies.

RESULTS The cohort included 101 million individuals, of which there were 47 982 suicide index cases
identified. Household members exposed to a suicide (11 070 [7%] Black, 82 407 [53%] Parda, and
57 726 [37%] White) had a 32% higher risk of all-cause mortality (adjusted hazard ratio [aHR], 1.32;
95% CI, 1.28-1.36). The risk of suicide among exposed individuals was more than 4 times higher (aHR,
4.42; 95% CI, 3.86-5.07), with 101 of these deaths (44%) occurring within 2 years of the index suicide

+ Supplemental content
Author affiliations and article information are
listed at the end of this article.

case. The population attributable fraction for suicide was 77%. Elevated risks were also observed for
other external causes (eg, assault and falls) and nonexternal causes (eg, neoplasms and circulatory
and respiratory diseases). Mortality risk was highest when the index case was female and younger,
among male survivors, and individuals aged 25 to 59 years. Better household conditions were
associated with lower risks of both suicide and all-cause mortality.
CONCLUSIONS AND RELEVANCE Exposure to suicide within the household was associated
with a substantial increase in both suicide-specific and all-cause mortality among surviving
household members, particularly in the immediate aftermath. These findings underscore the
urgent need to incorporate targeted postvention strategies into comprehensive suicide
prevention efforts.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286

Open Access. This is an open access article distributed under the terms of the CC-BY License.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

1/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

Introduction
Suicide is a major global public health issue with profound impacts that extend beyond the deceased
to surviving family members.1-4 The psychological, social, and economic consequences of suicide
ripple through families and communities.2-4 Surviving family members are frequently confronted
with a complex mix of grief, stigma, and structural vulnerability5-7 that adversely affects their
mental8-11 and physical well-being.9 These challenges are heightened in low-income settings, where
the burden of suicide intersects with barriers to mental health care, socioeconomic hardship, and
cultural stigma.12,13
Research has largely focused on suicide risk among bereaved relatives,2,14-25 though less is
known about how many people are exposed and need support in the broader household19 or about
their overall mortality risk from nonsuicide causes.9,26-28 Prior studies have lacked unexposed
comparison groups, reducing causal inference and limiting the ability to attribute observed outcomes
to suicide exposure. Furthermore, there is a need to consider when excess mortality arises and how
it varies over time following a suicide in the household. Most evidence comes from high-income
countries,2,9,14-24,26-28 where more robust health systems may buffer harms. In contrast, household
socioeconomic stressors and sociodemographic characteristics remain understudied in low-income
settings where risks may be amplified.17
Understanding the long-term consequences of suicide exposure on mortality is essential for
guiding public health and postvention efforts.29,30 However, few population-based studies have
examined these outcomes.17,19 This study used nationwide administrative data from Brazil to
investigate mortality outcomes among household members exposed to suicide in socioeconomically
disadvantaged settings. Objectives were to (1) estimate the increased risk of all-cause and causespecific mortality among household members exposed to suicide, (2) identify risk factors for
subsequent suicide among these household members, and (3) assess timing of mortality after
exposure. It was hypothesized that exposure to suicide increases mortality, with risks varying by
timing, sociodemographic factors, and household conditions.

Methods
Ethical Considerations
Ethical approval was obtained from the Federal University of Bahia and Centro de Pesquisa Gonçalo
Moniz, Fundação Oswaldo Cruz, Bahia. As no personally identifiable information was included,
informed consent was waived. The study followed the Strengthening the Reporting of Observational
Studies in Epidemiology (STROBE) reporting guideline.

Study Design and Data Sources
This nationwide longitudinal study used data from the 100 Million Brazilian Cohort (100MCohort)
linked with the Brazilian Mortality Information System (SIM) (2001-2018).31 The 100MCohort,
established by the Centre for Data and Knowledge Integration for Health (CIDACS),32 involves linkage
between the Unified Registry for Social Programs (CadÚnico) and Brazilian Health Information
Systems.33 CadÚnico eligibility requires per capita monthly family income equal to half of the
minimum wage or less.34 From 2011 to 2018, CadÚnico included 131 697 800 individuals, with higher
proportions of young people, female individuals, and urban residents. SIM is a nationwide health
information system for recording mortality data, encompassing all deaths in the country and
their causes.35

Data Linkage
Linking the 100MCohort baseline and SIM datasets (2001-2018) used a 2-step process based on 5
individual identifiers (name, date of birth, sex, mother’s name, and municipality) via the CIDACS
record linkage tool.36 Exact matching was followed by similarity-score linkage for unmatched entries.

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

2/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

Accuracy was assessed through manual validation of a random subset and receiver operating
characteristic analysis, with sensitivity and specificity exceeding 92% (eFigure 1 and eTable 1 in
Supplement 1).

Study Population
This study included individuals aged 10 years or older at registration or who reached this age during
follow-up (January 1, 2001, to December 31, 2018). Excluded individuals were (1) those aged above
110 years at registration and (2) individuals who registered in CadÚnico on the last day of the
follow-up (December 31, 2018), died, or became exposed on the day of cohort enrollment (eFigure 2
in Supplement 1).

Exposure and Follow-up
Suicide survivors were defined as household members of the decedent, given their exposure to a
suicide and likelihood of being personally affected by it.19,29 In the 100MCohort, household members
share a family and/or household code, and each individual has a unique identifier linked to
socioeconomic and demographic characteristics. Using these codes, we identified all members of
each household and the first suicide within it, referred to as the suicide index case. The exposed
individuals were those who had a suicide index case in their household, identified by sharing the
same household code. Unexposed individuals were classified as: (1) those who never had a suicide
index case during the follow-up period or (2) those who had an suicide index case but were
considered unexposed until the occurrence of that event.
Unexposed individuals were followed up from the time of their registration in the cohort
baseline until 1 of the following occurrences: (1) death due to any cause (including suicide), (2) end of
the follow-up (December 31, 2018), or (3) the date of the occurrence of the suicide index case within
the same household. Exposed individuals were followed up from the occurrence of the suicide index
case until 1 of the following: (1) death due to any cause (including suicide) or (2) end of the follow-up
(December 31, 2018) (eFigure 3 in Supplement 1).

Primary and Secondary Outcomes
The primary outcome was all-cause mortality, defined as death from any cause as classified by the
International Statistical Classification of Diseases and Related Health Problems, Tenth Revision
(ICD-10).35 As suicide is an expected outcome in families with a history of suicide, the main analyses
focused on all-cause mortality excluding suicide (ICD-10 codes X60-X84) to avoid confounding.
Separate analyses were conducted for all-cause mortality including suicide and for suicide-specific
mortality to enable comparison and interpretation. Secondary outcomes included (1) other mortality
due to external causes (ICD-10 codes V00-Y99, excluding suicide); (2) the 5 most frequent specific
causes of external mortality in our dataset: violence (X85-Y09), transport injuries (V01-V99), falls
(W00-W19), accidental drowning (W65-W74), and accidental poisoning (X43); (3) mortality due to
natural causes (all other causes excluding external causes); and (4) the most frequent specific natural
causes of death, including diseases of the circulatory system (I00-I99), neoplasms (C00-D48),
metabolic diseases (E70-E89), diseases of the respiratory system (J00-J99), and parasitic diseases
(B65-B83).

Statistical Analysis
For the first objective, sex-specific age-standardized rates were estimated using person-year as the
denominator and using the Brazilian 2015 official population projection as the standard.33 A
multivariable, time-varying Cox regression model was used to assess whether exposure to a
household suicide was associated with subsequent mortality among surviving members. This
approach considers changing variables and time-dependent covariates,37 allowing exposure to be
treated dynamically; individuals contributed person-time as unexposed until a household suicide
occurred, after which they were classified as exposed (eFigure 3 in Supplement 1). Crude and

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

3/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

adjusted hazard ratios (HRs) were estimated with 95% CIs. Models were adjusted for confounders
identified in prior literature, including sex, age, race, region, location of residence, urbanicity,
unemployment, housing materials, water supply, sanitation, and waste. Race was derived from the
CadÚnico database and self-reported as Asian, Black, Indigenous, Parda, or White. Parda (Portuguese
for brown) denotes individuals of predominantly Black or mixed ancestry, including European,
African, and Indigenous origins. The attributable risk percentage (%AR) was calculated using the
Levin formula: %AR_exp = [(Incidence_exp – 1) ÷ (Incidence_exp – 1) + 1] × 100, to quantify the
proportion of deaths attributable to exposure (exp).38
For the second objective, the focus was on individuals in households with a prior suicide. A
multivariable Cox regression was used to examine factors associated with subsequent suicide or
all-cause mortality. The final model, optimized through stepwise selection, included variables for (1)
the sex and age of the index case and (2) the sex and age of surviving household members (measured
at the time of the index case), as well as household conditions and geographic region. Household
conditions were captured using a composite score from 0 to 4, where 0 indicates no access and 4
indicates full access to essential services (water supply, sanitation, adequate housing materials, and
waste disposal). Two interaction terms (sex and age of the index case) were tested to evaluate
whether the outcomes associated with other risk factors varied according to these characteristics.
Details of model specification and covariates are provided in eAppendices 1, 2, 3, and 4 in
Supplement 1.
Finally, the proportional mortality of subsequent deaths and suicide events by year of
occurrence after the index case were calculated. Nearly half of these deaths occurred within the first
2 years, which guided the definition of the immediate period as 2 or fewer years and the distant
period as 3 or more years. The analysis further assessed whether risk factors varied between
immediate (ⱕ2 years) and distant (ⱖ3 years) follow-up periods by including multiplicative
interaction terms between each covariate and a dichotomous variable indicating follow-up time. The
model can be represented as follows: log[hazard(t)] = β0 + β1 (ⱖ3 years period) + β2 (characteristic
X) + β3 (ⱖ3 years period × characteristic X), where characteristic X refers to an explanatory variable
of interest, such as the sex of the index individual, age at death, or sex of surviving household
members, among others. The interaction term (β3) allows us to assess whether the association of
characteristic X with the risk of the outcome (all-cause mortality or suicide) differs between the
period shortly after the death (ⱕ2 years) and the later period (ⱖ3 years). Although the model
includes a single variable, multiple factors were included simultaneously, each with its respective
main association and interaction term with time since death, allowing us to evaluate time-varying
associations. The joint significance of the interaction terms was tested using the Wald test.
HRs and 95% CIs were reported separately for each time interval. All analyses were performed
using Stata version 15.1 (StataCorp LLC). A 2-sided P < .05 was considered statistically significant.
To address potential bias from differences in follow-up time between exposed and unexposed
groups, we performed sensitivity analyses using (1) a doubly robust approach adjusting the main
analysis by adding exposure time as a covariate, (2) restricting follow-up to cohort entry for exposed
and unexposed, and (3) simulating equal follow-up durations. For analyses of immediate and distant
periods, we also tested alternative cutoffs and time scales (ⱕ1 year vs ⱖ2 years; ⱕ3 years vs ⱖ4
years). Data were analyzed from March 2023 to September 2025.

Results
The cohort consisted of 101 346 669 individuals from 26 594 713 households, with 167 475
individuals in the exposed group. There were 47 982 suicide index cases (eFigure 2 in Supplement 1).
Compared with the nonexposed group, individuals exposed to suicide were more likely to be female
(90 405 individuals [54%]), young people (aged 10-24 years: 111 791 individuals [67%]), Parda
(82 407 individuals [53%]), and unemployed (162 916 individuals [97%]) (Table 1). The exposed
group also included 11 070 (7%) Black and 57 726 (37%) White individuals.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

4/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

Table 1. Description of Study Population by Exposure to an Index Suicide Case in the Same Household
(N = 101 346 669)
Participants, No. (%)
Characteristic

Nonexposure (n = 101 179 194)

Exposure (n = 167 475)

P valuea

Sex
Male

47 658 742 (47)

77 070 (46)

Female

53 520 452 (53)

90 405 (54)

10-24 y

58 804 961 (58)

111 791 (67)

25-59 y

37 278 090 (37)

50 809 (30)

60-110 y

5 096 143 (5.0)

4875 (2.9)

Asian descendants

372 702 (0.4)

506 (0.3)

Black

7 036 884 (7.5)

11 070 (7.1)

Indigenous

538 453 (0.6)

3627 (2.3)

Pardac

55 370 996 (59)

82 407 (53)

White

31 068 881 (33)

57 726 (37)

Unknown

6 791 278

12 139

Northeast

40 065 572 (40)

61 754 (37)

North

10 465 502 (10)

16 459 (9.8)

Southeast

32 322 337 (32)

45 742 (27)

South

11 522 233 (11)

31 126 (19)

<.001

Age cohort

<.001

Raceb

<.001

Region

Central-West

6 710 100 (6.6)

12 216 (7.3)

Unknown

93 450

178

Urban

73 144 808 (74)

113 151 (69)

Rural

25 327 550 (26)

51 598 (31)

Unknown

2 706 836

2726

Yes

96 309 167 (95)

162 916 (97)

No

4 870 027 (4.8)

4559 (2.7)

Uninformed

3 630 860 (3.6)

3644 (2.2)

Bricks or cement

72 657 604 (72)

109 128 (65)

Wood, vegetal materials, and other

24 890 730 (25)

54 703 (33)

4 301 559 (4.3)

4391 (2.6)

<.001

Location residence

<.001

Unemployed
<.001

Construction materials

<.001

Sanitation
Uninformed
Public network

42 391 809 (42)

59 256 (35)

Septic tank

14 932 195 (15)

27 551 (16)

Homemade septic tank

24 675 659 (24)

45 457 (27)

Ditch or other

14 877 972 (15)

30 820 (18)

<.001

Water supply
Uninformed

3 630 469 (3.6)

3634 (2.2)

Public network (running water)

68 163 602 (67)

107 549 (64)

Well, natural sources, or other

29 385 123 (29)

56 292 (34)

<.001

Waste
Uninformed

3 630 901 (3.6)

3642 (2.2)

Public collection system

71 129 518 (70)

108 884 (65)

Burned, buried, outdoor disposal,
or other

26 418 775 (26)

54 949 (33)

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

<.001

a

Pearson χ2 test.

b

Race was self-reported.

c

Parda, which translates from Portuguese as brown,
is used to denote individuals whose racial
background is predominantly Black and those with
multiracial ancestry, including European, African,
and Indigenous origins.

November 25, 2025

5/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

Risk of All-Cause and Specific-Cause Mortality
Exposure to a suicide index case was associated with a 32% higher risk of mortality when including
suicide (HR, 1.32; 95% CI, 1.28-1.36), and 27% higher risk when excluding suicide (HR, 1.27; 95% CI,
1.23-1.31), compared with nonexposed individuals (Figure 1). Suicide-specific mortality risk was over
4 times higher among exposed household members (HR, 4.42; 95% CI, 3.86-5.07). Exposed
members also had a higher risk of death from external causes (excluding suicide) (HR, 1.35; 95% CI,
1.27-1.44) and natural causes (HR, 1.23; 95% CI, 1.18-1.27). Among external causes, higher risks were
observed for deaths due to falls (HR, 1.80; 95% CI, 1.31-2.44), drowning (HR, 1.49; 95% CI,
1.09-2.03), assault (HR, 1.38; 95% CI, 1.27-1.51), and transport crashes (HR, 1.26; 95% CI, 1.12-1.43). For
natural causes, higher risks were found for circulatory diseases (HR, 1.20; 95% CI, 1.12-1.29),
neoplasms (HR, 1.22; 95% CI, 1.11-1.33), metabolic conditions (HR, 1.27; 95% CI, 1.11-1.46), and
respiratory diseases (HR, 1.27; 95% CI, 1.08-1.48), with no clear association for infectious diseases
(HR, 1.09; 95% CI, 0.92-1.29) (Figure 1).

Main Risk Factors Among Surviving Household Members
Exposure to a female index case was associated with higher all-cause mortality excluding suicide
among surviving household members (HR, 1.27; 95% CI, 1.16-1.40) compared with a male index case,
but not with higher suicide (HR, 1.39; 95% CI, 0.94-2.06). Exposure to a younger index case (aged
10-24 years) was associated with higher all-cause mortality (HR, 1.16; 95% CI, 1.08-1.26) and suicide
(HR, 1.68; 95% CI, 1.22-2.31) compared with an older index case (aged 60 years or older). An
interaction showed that a young female index case was associated with lower risk of all-cause
mortality (HR, 0.72; 95% CI, 0.61-0.84) relative to older male cases (Table 2).
Among surviving household members, male individuals had a 75% higher risk of all-cause
mortality excluding suicide (HR, 1.75; 95% CI, 1.64-1.86) and over 3 times higher risk of suicide (HR,
3.58; 95% CI, 2.65-4.84) compared with female individuals. Compared with young survivors, risk of
all-cause mortality was higher among those aged 25 to 59 years (HR, 3.81; 95% CI, 3.51-4.15) and
those aged 60 years or older (HR, 24.13; 95% CI, 22.09-26.37), whereas suicide risk was higher

Figure 1. Cox Multivariate Model of the Association Between Suicide Index Case and the Risk of Mortality
Age-standardized rate, per person-years (95% CI)
Exposed

Adjusted HR
(95% CI)

Cause of death
All-cause mortality

Unexposed

PAF (95% CI)

With suicide

541.92 (541.92-541.93) 646.68 (646.67-646.68) 1.32 (1.28-1.36)

0.24 (0.22-0.27)

Without suicide

537.60 (537.60-537.61) 625.22 (625.22-625.23) 1.27 (1.23-1.31)

0.21 (0.18-0.24)

Cause-specific mortality
Suicide

4.32 (4.31-4.32)

21.45 (21.45-21.46)

4.42 (3.86-5.07)

0.77 (0.74-0.80)

External cause (without suicide)

57.32 (57.31-57.32)

87.65 (87.65-87.66)

1.35 (1.27-1.44)

0.26 (0.21-0.31)

Nonexternal cause

480.29 (480.29-480.30) 537.56 (537.56-537.57) 1.23 (1.18-1.27)

0.18 (0.15-0.21)

External cause-specific mortality
Assault

26.12 (26.12-26.13)

43.69 (43.69-43.70)

1.38 (1.27-1.51)

0.28 (0.22-0.34)

Transport related

17.68 (17.68-17.69)

23.38 (23.38-23.39)

1.26 (1.12-1.43)

0.21 (0.11-0.31)

Falls

4.74 (4.73-4.74)

7.01 (7.01-7.02)

1.80 (1.31-2.44)

0.44 (0.27-0.61)

Accidental drowning

2.17 (2.17-2.18)

3.60 (3.60-3.61)

1.49 (1.09-2.03)

0.33 (0.12-0.54)

Exposure smoke

0.34 (0.34-0.35)

0.45 (0.45-0.46)

1.45 (0.54-3.86)

0.31 (-0.37-0.99)

Accidental poisoning

0.23 (0.23-0.24)

0.22 (0.22-0.23)

0.93 (0.30-2.89)

-0.07 (-1.29-1.14)

Nonexternal cause-specific mortality
Diseases in the circulatory system

165.83 (165.83-165.84) 184.03 (184.03-184.04) 1.20 (1.12-1.29)

0.17 (0.11-0.22)

Neoplasms

89.92 (89.92-89.93)

100.41 (100.41-100.42) 1.22 (1.11-1.33)

0.18 (0.11-0.25)

Metabolic diseases

37.52 (37.52-37.53)

45.30 (45.30-45.31)

1.27 (1.11-1.46)

0.21 (0.11-0.32)

Diseases in the respiratory system

27.55 (27.55-27.56)

30.86 (30.86-30.87)

1.27 (1.08-1.48)

0.21 (0.09-0.33)

Infectious disease

22.83 (22.83-22.83)

23.64 (23.64-23.64)

1.09 (0.92-1.29)

0.08 (-0.07-0.24)
0.3

1

4

Adjusted HR (95% CI)

Model adjusted for sex, age cohort, race, region, location residence, unemployed, construction materials, water supply, and waste. HR indicates hazard ratio; PAF, population
attributable fraction.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

6/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

among those aged 25 to 59 years (HR, 1.64; 95% CI, 1.25-2.14) but not among those aged 60 years or
older (HR, 1.04; 95% CI, 0.52-2.06).
Better household conditions, reflected by having all 4 essential services, had lower risk of
all-cause mortality (HR, 0.76; 95% CI, 0.69-0.85) and suicide (HR, 0.49; 95% CI, 0.33-0.73)
compared with households without these essential services. Regionally, suicide risk was higher in the
South (HR, 2.13; 95% CI, 1.51-3.01) and Central-West (HR, 2.89; 95% CI, 1.89-4.41) compared with the
Northeast. For all-cause mortality excluding suicide, the South showed increased risk (HR, 1.17; 95%
CI, 1.07-1.28) (Table 2).

Table 2. Cox Regression Model by Characteristics of the Index Suicide Case, Sociodemographic Characteristics of the Members, and Household Conditions
Among Individuals Who Have Experienced a Previous Suicide Within the Same Household, 2001 to 2018
All-cause mortality outcome (without suicide)

Suicide outcome

HR (95% CI)

Adjusted HR (95% CI)

HR (95% CI)

Male

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

Female

1.13 (1.05-1.22)

1.27 (1.16-1.40)

1.48 (1.12-1.96)

1.39 (0.94-2.06)

≥60 y

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

10-24 y

1.19 (1.11-1.27)

1.16 (1.08-1.26)

1.70 (1.30-2.21)

1.68 (1.22-2.31)

25-59 y

1.73 (1.55-1.94)

1.04 (0.92-1.19)

1.22 (0.69-2.16)

1.09 (0.54-2.17)

Male ≥60 y

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

Female 10-24 y

0.64 (0.55-0.75)

0.72 (0.61-0.84)

0.57 (0.32-1.03)

0.70 (0.39-1.27)

Female 25-59 y

1.47 (1.12-1.96)

0.94 (0.72-1.24)

1.78 (0.52-6.16)

1.88 (0.54-6.51)

Female

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

Male

1.50 (1.41-1.60)

1.75 (1.64-1.86)

3.54 (2.63-4.77)

3.58 (2.65-4.84)

10-24 y

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

25-59 y

3.57 (3.29-3.89)

3.81 (3.51-4.15)

1.42 (1.09-1.85)

1.64 (1.25-2.14)

≥60 y

22.74 (20.84-24.82)

24.13 (22.09-26.37)

0.92 (0.47-1.82)

1.04 (0.52-2.06)

0

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

1

1.02 (0.94-1.11)

1.03 (0.94-1.12)

0.43 (0.29-0.65)

0.47 (0.31-0.71)

2

0.96 (0.87-1.05)

0.99 (0.90-1.10)

0.62 (0.43-0.91)

0.61 (0.42-0.89)

3

0.78 (0.70-0.86)

0.79 (0.71-0.88)

0.34 (0.23-0.51)

0.33 (0.22-0.50)

4

0.73 (0.66-0.81)

0.76 (0.69-0.85)

0.43 (0.30-0.62)

0.49 (0.33-0.73)

Northeast

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

North

0.85 (0.75-0.97)

1.09 (0.96-1.23)

1.51 (0.94-2.43)

1.22 (0.75-1.98)

Southeast

1.04 (0.96-1.12)

1.09 (1.00-1.18)

1.00 (0.68-1.47)

1.15 (0.76-1.73)

South

1.09 (1.01-1.19)

1.17 (1.07-1.28)

2.05 (1.45-2.88)

2.13 (1.51-3.01)

Central-West

0.97 (0.85-1.10)

1.09 (0.95-1.24)

2.80 (1.84-4.25)

2.89 (1.89-4.41)

Characteristic

Adjusted HR (95% CI)

Characteristics of the index case
Sex of the index case

Age at death of the index case

Sex of the index case and age at death
of the index case

Sociodemographic characteristics of
surviving members
Sex of surviving household members

Mortality age of the surviving household members
at the date of occurrence of the Index case

Household characteristics
Living conditionsa

Region

Abbreviation: HR, hazard ratio.
a

Living conditions defined as household conditions measured using a composite score from 0 to 4, where 0 indicates no access and 4 indicates full access to essential services,
including water supply, sanitation, adequate housing materials, and waste disposal.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

7/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

Timing of Subsequent Deaths After the Index Case: Comparison of Immediate
(≤2 Years) and Distant Events (>3 Years)
Most subsequent deaths, both all-cause and by suicide, occurred within 2 years of the index suicide.
Among all-cause mortality (excluding suicide; 4009 individuals), 1429 (35.6%) occurred within 2
years (931 [23.2%] in year 1; 498 [12.4%] in year 2) (Figure 2). Among suicides (232 individuals), 101
(43.6%) occurred in the same period (79 [34.1%] in year 1; 22 [9.5%] in year 2). For all-cause
mortality, no differences were observed between immediate (ⱕ2 years) and distant (ⱖ3 years)
periods. For suicide, a time-dependent association was found; during the first 2 years, having a
female index case was associated with higher suicide risk among survivors (HR, 1.72; 95% CI,
1.03-2.84; P = .03), an outcome not observed after 3 years (HR, 0.86; 95% CI, 0.58-1.28) (Figure 3).
All sensitivity analyses yielded results consistent with those of the main model (eTables 6-9,
eFigure 5 in Supplement 1). A summary of key findings and associated risk factors is provided in
eFigure 6 in Supplement 1.

Discussion
This large-scale cohort study is the first we know of to examine timing and risk factors for all- and
cause-specific mortality after household exposure to suicide. Exposure was associated with a 27%
increased in all-cause mortality (excluding suicide) and over 4-fold higher suicide risk, with a
population attributable fraction of 77%. Risks were elevated when the index case was younger or
female, and when surviving household members were male, aged 25 to 59 years, or living in
households with poor infrastructure. Over half of subsequent suicides occurred within the first 2
years, reinforcing the urgency of targeted early interventions (eFigure 7 in Supplement 1).
Prior research shows elevated suicide risk among individuals with a family history of
suicide,2,14-25 yet broader mortality outcomes remain underexplored.9,26-28 In Taiwan, higher rates of
suicide (rate ratio [RR], 4.61; 95% CI, 4.02-5.29) and accidental deaths (RR, 1.62; 95% CI, 1.43-1.84)
were reported among first-degree relatives of suicide decedents.27 Another study found increased
risks of suicide (HRs up to 15.67; 95% CI, 2.09-117.41), homicide (HRs up to 23.26; 95% CI,
3.10-174.56), and dementia (HR, 4.41; 95% CI, 1.14-17.05) in suicide-exposed individuals, but not for

Figure 2. Proportion of Subsequent All-Cause Mortality (Excluding Suicide) and Suicide in Years
B

Suicide

40

40

30

30

Proportion following index case, %

Proportion following index case, %

A All-cause mortality

20

10

0

20

10

0
1

2

3

4

5

6

7

8

9

10

11

12

13

14 ≥15

1

Time to subsequent death, y

2

3

4

5

6

7

8

9

10

11

12

13

14 ≥15

Time to subsequent death, y

All-cause mortality (4009 individuals) and suicide (232 individuals).
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

8/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

all-cause mortality relative to unnatural deaths (HR, 0.95; 95% CI, 0.87-1.04).28 The current study
expands on existing evidence, showing that household suicide exposure is linked to higher risks of
both suicide and all-cause mortality, including from metabolic, respiratory, circulatory diseases, and
neoplasms. All major external and natural causes of death were elevated except parasitic diseases,
which are likely more tied to environmental and socioeconomic factors than psychosocial stressors.
The 442% increased risk of suicide observed in this study exceeds estimates from high-income
countries such as Denmark (RR, 2.58; 95% CI, 1.84-3.61)21 and South Korea (RR, 2.75; 95% CI,
2.55-2.97),16 suggesting greater impact in settings characterized by heightened social and structural
vulnerability.
The mechanisms linking family suicide history and increased mortality likely involve multiple
pathways.9,26-28 Suicide exposure is traumatic and often results in grief, guilt, stigma, and family
disruption.5-7 These factors are associated with suicide behaviors,2,14-25 and contribute to

Figure 3. Factors Associated with Immediate (≤2 Years) and Distant (≥3 Years) Time-to-Event Mortality
A All-cause mortality

Characteristic

HR (95% CI)

Lower
risk

B

Higher
risk

P value

0.97 (0.88-1.07)
Female index case

.87

1.01 (0.89-1.14)

.72

0.91 (0.46-1.81)

.50

0.56 (0.22-1.47)

.50

0.97 (0.52-1.83)
1.03 (0.63-1.67)

.50

0.98 (0.85-1.14)
1.23 (1.03-1.44)

Central-west region

.72

0.92 (0.45-1.85)

0.97 (0.53-1.77)
.50

1.01 (0.88-1.17)
1.10 (0.99-1.23)

South region

.72

1.08 (0.55-2.11)

1.20 (0.63-2.27)
.50

0.88 (0.73-1.08)
1.16 (1.04-1.30)

Southeast region

.72

1.31 (0.69-2.47)

0.51 (0.29-0.90)
.86

0.96 (0.80-1.14)
1.18 (1.00-1.39)

North region

.64

1.30 (0.52-3.24)

0.80 (0.43-1.46)
.86

1.19 (0.99-1.43)
1.00 (0.87-1.14)

Households with 4 essential services

.64

0.87 (0.48-1.55)

0.74 (0.44-1.26)
.86

0.96 (0.81-1.15)
1.03 (0.90-1.17)

Households with 3 essential services

.96

0.89 (0.54-1.47)

0.87 (0.47-1.62)
.86

0.96 (0.83-1.12)
1.02 (0.90-1.16)

Households with 2 essential services

.61

0.83 (0.40-1.72)

1.30 (0.52-3.24)
.71

1.12 (0.95-1.31)
1.03 (0.92-1.16)

Households with 1 essential service

.61

1.46 (0.88-2.41)

1.03 (0.67-1.56)
.71

1.01 (0.87-1.19)
1.29 (1.16-1.45)

Survivor household member aged ≥60 y

.03

1.72 (1.03-2.84)

0.87 (0.51-1.50)
.34

1.02 (0.91-1.14)
1.12 (1.00-1.24)

Survivor household member aged 25-59 y

P value

1.06 (0.24-4.60)
.24

1.14 (0.96-1.36)
1.13 (1.04-1.23)

Male survivor household member

Higher
risk

1.09 (0.75-1.60)
.24

0.92 (0.82-1.04)
0.90 (0.78-1.05)

Index case aged 25-59 y

Lower
risk

HR (95% CI)
0.86 (0.58-1.28)

0.98 (0.90-1.06)
Index case aged 10-24 y

Suicide

.50

1.34 (0.76-2.36)
Period
≤2 y
≥3 y

0.98 (0.52-1.86)
.50

0.96 (0.77-1.19)
0.6

1

0.68 (0.32-1.45)

2

HR (95% CI)

0.2

1

.50

5

HR (95% CI)

P value represents the test of interaction between the time since the index case's death (ⱕ2 years vs ⱖ3 years) and the respective characteristic under evaluation.
HR indicates hazard ratio.
JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

9/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

depression, anxiety, PTSD, and substance use,3-11,18 thereby leading to early death. Chronic stress
may disrupt the hypothalamic-pituitary-adrenal axis, suppress immune function, and increase the
risk of cardiovascular and metabolic disorders,39 explaining elevated deaths from circulatory,
respiratory, and metabolic diseases. Social and economic stressors may also play a role.2,18 Suicide
loss can trigger financial hardship, social withdrawal, family conflict, and caregiving burdens.5-7 Male
survivors and adults aged 25 to 59 years were especially affected, suggesting that adult men may
face distinct challenges in coping with household suicide. Death of a younger index case (aged 10-24
years) also resulted in higher mortality risk, reflecting the destabilizing impact of an unexpected
death and the profound grief it provokes.20,21,40
Over a third of all-cause deaths (35.6%) and nearly half of suicides (43.6%) occurred within 2
years of the index suicide. In the immediate aftermath of the suicide, family members often
experience intense grief marked by guilt, blame, and stigma,2,5-7,18 leading to enduring physical,
psychological, and psychosomatic difficulties.18 These challenges highlight the urgent need for
family-centered interventions after suicide loss.12 Although the bereaved may express a notable
desire to participate in suicide support groups,2,18 access to mental health services or specialized care
for survivors remains scarce, especially in lower-resource contexts such as Brazil.41 In the current
study, survivors living in better housing conditions showed lower mortality risk, reinforcing the role
of socioeconomic factors in mitigating outcomes after suicide loss.42,43

Strengths and Limitations
Previous studies have been limited to high-income settings with small samples and lacked
population-based comparison groups. This study leveraged a large population-based cohort to
overcome these gaps. This design provides sufficient power to examine rare outcomes and
exposures, including suicide occurrence and recurrence within households. Most earlier studies used
case-control designs, which precluded estimation of population-level measures such as suicide,
relative risks, and attributable risks. Other studies focused only on individuals with recorded deaths,
limiting comparisons by family suicide history. The current approach enabled the estimation of these
rates and risks and the assessment of the interval between an index suicide and subsequent suicide
within the same household, offering important insights for prevention strategies.
Some limitations are unmeasured confounders that could introduce bias and the lack of
information regarding parental status, as well as access to mental health care. It is noteworthy that
some causes of death, such as ill-defined or unknown causes, events of undetermined intent, and
accidental deaths, may conceal hidden suicides, potentially influencing the findings. Moreover, this
study population was composed of low-income individuals, limiting generalizability. Additionally,
suicide may be underregistered; however, in Brazil, external causes of death undergo systematic
review by medical-legal institutes, and the SIM database is recognized for its high quality, minimizing
misclassification.44

Conclusions
This cohort study of 101 346 669 individuals in Brazil found that household exposure to suicide
resulted in over 4-fold increase in subsequent suicide risk and a 27% rise in all-cause mortality
(excluding suicide) among surviving members (eFigure 7 in the Supplement). These far-reaching
health consequences, associated with external and nonexternal causes, were most pronounced in
the first 2 years, highlighting the urgency for early intervention. These findings call for integrated
postvention strategies including bereavement care, psychosocial support, and clinical follow-up,
especially in lower-resource settings, to advance global mental health equity and support Sustainable
Development Goal 3.4: “reducing premature mortality from noncommunicable diseases and
promoting mental well-being.”45

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

10/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

ARTICLE INFORMATION
Accepted for Publication: October 2, 2025.
Published: November 25, 2025. doi:10.1001/jamanetworkopen.2025.45286
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2025 Alves F
et al. JAMA Network Open.
Corresponding Author: Flávia Alves, Center for Data and Knowledge Integration for Health (CIDACS), Rua Mundo,
Sem número, Parque Tecnológico da Bahia–Trobogy, Salvador, Bahia CEP 41.745-715, Brazil (Flavia_Alves@hms.
harvard.edu).
Author Affiliations: Center of Data and Knowledge Integration for Health, Instituto Gonçalo Moniz, Fundação
Oswaldo Cruz, Salvador, Bahia, Brazil (Alves, E. d. S. Rodrigues, Toledo, Pescarini, Barreto, Machado); Department
of Global Health and Social Medicine, Harvard Medical School, Boston, Massachusetts (Alves, Naslund, Patel,
Machado); Federal University of Ceará, Campus Jardins de Anita, Itapajé, Ceará, Brazil (E. d. S. Rodrigues);
Department of Infectious Disease Epidemiology, and Epidemiology and Population Health, Faculty of
Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, United Kingdom
(Pescarini); Federal Rural University of Pernambuco, Dois irmãos Campus, Recife, Pernambuco, Brazil
(R. L. Rodrigues).
Author Contributions: Drs Rodrigues and Alves had full access to all of the data in the study and take
responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Oliveira Alves, Toledo, Pescarini, Barreto, Patel, Machado.
Acquisition, analysis, or interpretation of data: Oliveira Alves, E. Rodrigues, Pescarini, R. Rodrigues, Naslund,
Barreto, Machado.
Drafting of the manuscript: Oliveira Alves, E. Rodrigues, R. Rodrigues, Patel.
Critical review of the manuscript for important intellectual content: Oliveira Alves, Toledo, Pescarini, Naslund,
Barreto, Machado.
Statistical analysis: Oliveira Alves, E. Rodrigues, Pescarini, R. Rodrigues.
Obtained funding: Machado.
Administrative, technical, or material support: Naslund.
Supervision: E. Rodrigues, Barreto, Patel, Machado.
Conflict of Interest Disclosures: Dr Naslund reported receiving grants from HE Butt Foundation, John Templeton
Foundation, Tepper Foundation, Wellcome Trust, and Lyda Hill Philanthropies outside the submitted work. No
other disclosures were reported.
Funding/Support: Research reported in this publication was supported by the National Institute of Mental Health
of the National Institutes of Health under Award No. R01MH128911.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and
decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official
views of the National Institutes of Health.
Data Sharing Statement: See Supplement 2.
Additional Contributions: We extend our gratitude to the CIDACS/Fundação Oswaldo Cruz data production team
for their invaluable assistance in linking the data used in this study, as well as for providing data quality information.
Additionally, we acknowledge the dedicated efforts of the information technology team in facilitating seamless
access to data and analytical tools. They were not compensated for this work outside of their normal salaries.
REFERENCES
1. World Health Organization. World health statistics 2019: monitoring health for the SDGs, sustainable
development goals. Accessed April 2, 2024. https://www.who.int/publications-detail-redirect/9789241565707
2. Spillane A, Larkin C, Corcoran P, Matvienko-Sikar K, Riordan F, Arensman E. Physical and psychosomatic health
outcomes in people bereaved by suicide compared to people bereaved by other modes of death: a systematic
review. BMC Public Health. 2017;17(1):939. doi:10.1186/s12889-017-4930-3
3. Brent D, Melhem N, Donohoe MB, Walker M. The incidence and course of depression in bereaved youth 21
months after the loss of a parent to suicide, accident, or sudden natural death. Am J Psychiatry. 2009;166(7):
786-794. doi:10.1176/appi.ajp.2009.08081244

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

11/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

4. Levi-Belz Y, Gilo T. Emotional distress among suicide survivors: the moderating role of self-forgiveness. Front
Psychiatry. 2020;11:341. doi:10.3389/fpsyt.2020.00341
5. Tal Young I, Iglewicz A, Glorioso D, et al. Suicide bereavement and complicated grief. Dialogues Clin Neurosci.
2012;14(2):177-186. doi:10.31887/DCNS.2012.14.2/iyoung
6. Mitchell AM, Kim Y, Prigerson HG, Mortimer MK. Complicated grief and suicidal ideation in adult survivors of
suicide. Suicide Life Threat Behav. 2005;35(5):498-506. doi:10.1521/suli.2005.35.5.498
7. Mitchell AM, Kim Y, Prigerson HG, Mortimer-Stephens M. Complicated grief in survivors of suicide. Crisis.
2004;25(1):12-18. doi:10.1027/0227-5910.25.1.12
8. Mitchell AM, Terhorst L. PTSD symptoms in survivors bereaved by the suicide of a significant other. J Am
Psychiatr Nurses Assoc. 2017;23(1):61-65. doi:10.1177/1078390316673716
9. Erlangsen A, Runeson B, Bolton JM, et al. Association between spousal suicide and mental, physical, and social
health outcomes: a longitudinal and nationwide register-based study. JAMA Psychiatry. 2017;74(5):456-464. doi:
10.1001/jamapsychiatry.2017.0226
10. Bolton JM, Au W, Leslie WD, et al. Parents bereaved by offspring suicide: a population-based longitudinal casecontrol study. JAMA Psychiatry. 2013;70(2):158-167. doi:10.1001/jamapsychiatry.2013.275
11. Ljung T, Sandin S, Långström N, Runeson B, Lichtenstein P, Larsson H. Offspring death and subsequent
psychiatric morbidity in bereaved parents: addressing mechanisms in a total population cohort. Psychol Med.
2014;44(9):1879-1887. doi:10.1017/S0033291713002572
12. World Health Organization. Preventing suicide: a global imperative. Accessed April 2, 2024. https://www.who.
int/publications-detail-redirect/9789241564779
13. Lichtenthal WG, Roberts KE, Donovan LA, et al. Investing in bereavement care as a public health priority.
Lancet Public Health. 2024;9(4):e270-e274. doi:10.1016/S2468-2667(24)00030-6
14. Kõlves K, Milner A, Värnik P. Suicide rates and socioeconomic factors in Eastern European countries after the
collapse of the Soviet Union: trends between 1990 and 2008. Sociol Health Illn. 2013;35(6):956-970. doi:10.1111/
1467-9566.12011
15. Kõlves K, Zhao Q, Ross V, Hawgood J, Spence SH, de Leo D. Suicide and sudden death bereavement in
Australia: a longitudinal study of family members over 2 years after death. Aust N Z J Psychiatry. 2020;54
(1):89-98. doi:10.1177/0004867419882490
16. Jang J, Park SY, Kim YY, et al. Risks of suicide among family members of suicide victims: a nationwide sample
of South Korea. Front Psychiatry. 2022;13:995834. doi:10.3389/fpsyt.2022.995834
17. Calderaro M, Baethge C, Bermpohl F, Gutwinski S, Schouler-Ocak M, Henssler J. Offspring’s risk for suicidal
behaviour in relation to parental death by suicide: systematic review and meta-analysis and a model for familial
transmission of suicide. Br J Psychiatry. 2022;220(3):121-129. doi:10.1192/bjp.2021.158
18. Spillane A, Matvienko-Sikar K, Larkin C, Corcoran P, Arensman E. What are the physical and psychological
health effects of suicide bereavement on family members? An observational and interview mixed-methods study
in Ireland. BMJ Open. 2018;8(1):e019472. doi:10.1136/bmjopen-2017-019472
19. Andriessen K, Rahman B, Draper B, Dudley M, Mitchell PB. Prevalence of exposure to suicide: a meta-analysis
of population-based studies. J Psychiatr Res. 2017;88:113-120. doi:10.1016/j.jpsychires.2017.01.017
20. Qin P, Mortensen PB. The impact of parental status on the risk of completed suicide. Arch Gen Psychiatry.
2003;60(8):797-802. doi:10.1001/archpsyc.60.8.797
21. Qin P, Agerbo E, Mortensen PB. Suicide risk in relation to family history of completed suicide and psychiatric
disorders: a nested case-control study based on longitudinal registers. Lancet. 2002;360(9340):1126-1130. doi:
10.1016/S0140-6736(02)11197-4
22. Runeson B, Asberg M. Family history of suicide among suicide victims. Am J Psychiatry. 2003;160(8):
1525-1526. doi:10.1176/appi.ajp.160.8.1525
23. Agerbo E. Risk of suicide and spouse’s psychiatric illness or suicide: nested case-control study. BMJ. 2003;327
(7422):1025-1026. doi:10.1136/bmj.327.7422.1025
24. Agerbo E. Midlife suicide risk, partner’s psychiatric illness, spouse and child bereavement by suicide or other
modes of death: a gender specific study. J Epidemiol Community Health. 2005;59(5):407-412. doi:10.1136/jech.
2004.024950
25. Tidemalm D, Runeson B, Waern M, et al. Familial clustering of suicide risk: a total population study of 11.4
million individuals. Psychol Med. 2011;41(12):2527-2534. doi:10.1017/S0033291711000833
26. Rostila M, Saarela J, Kawachi I. The forgotten griever: a nationwide follow-up study of mortality subsequent to
the death of a sibling. Am J Epidemiol. 2012;176(4):338-346. doi:10.1093/aje/kws163

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

12/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

27. Chen YY, Gunnell D, Wu CK, Hu YH, Lee PC. All-cause and cause-specific mortality in parents after the death of
a child in Taiwan: a population-based cohort study. Psychosom Med. 2023;85(3):221-230. doi:10.1097/PSY.
0000000000001182
28. Ho HY, Chen YL. Causes of death in individuals exposed to spousal, parental, and child suicide: a nationwide
population-based cohort comparison study. Eur J Epidemiol. 2023;38(11):1165-1174. doi:10.1007/s10654-02301055-8
29. Jordan JR. Lessons learned: forty years of clinical work with suicide loss survivors. Front Psychol. 2020;11:766.
doi:10.3389/fpsyg.2020.00766
30. Jordan JR. Postvention is prevention—the case for suicide postvention. Death Stud. 2017;41(10):614-621. doi:
10.1080/07481187.2017.1335544
31. Barreto ML, Ichihara MY, Pescarini JM, et al. Cohort profile: the 100 Million Brazilian Cohort. Int J Epidemiol.
2022;51(2):e27-e38. doi:10.1093/ije/dyab213
32. Barreto ML, Ichihara MY, Almeida BA, et al. The Centre for Data and Knowledge Integration for Health
(CIDACS): linking health and social data in Brazil. Int J Popul Data Sci. 2019;4(2):1140.
33. Informações de Saúde (TABNET). DATASUS. Accessed April 3, 2024. https://datasus.saude.gov.br/informacoesde-saude-tabnet/
34. VIS DATA 3 beta. Accessed April 3, 2024. https://aplicacoes.cidadania.gov.br/vis/data3/v.php?q[]=r6JtZI%
2B0gbBtxKW25rV%2FfmdhhJFkl21kmK19Zm11ZmqmaX7KmZO20qfOnJm%
2B6IianbSon7Stv8OcaJLHlNawmJi2wKmpa5Rwr2%
2BGf2uMvNSn06qU0eTDlZuulqfipbavqozH0JXcoqa83bOTn62jrehrfX9naL3Cn92ibtHtwpl3g5ub5ayyiXKgz
M5W5V5bnd2zk628mZnfmrp9d4yNl2mjbme84MCopbWWruKvrq2djsTKn9OeprzrvJauraiZ25%
2BssZybytBlmm5jl7W7qamtp6PcYnd%2FZ1143m7Qnp%2FQ4IianbSon7R0s6%
2BjoLycbt2yoNnMwpWqvJ6e2p2ybpuSd6eU1wDgyeSup1yqmqjen7axoI67wqaKrZjJ6m12q7Som5l%
2Frrv62sPKlI2CptHkupWwsaubmZ2ybn2OxCTg1qaU0Judo566mq2ZZm2RnJvK0FOzf3qim39kbXhYit6rs
LOloczCn4qhmH3evJahuqmv65ptspigd6eU1wDgyeSup1yqmqjen7axoPD405zLsFPB6m2Efo6xqrQ%3D
35. World Health Organization. International Statistical Classification of Diseases, Tenth Revision (ICD-10). World
Health Organization; 1992.
36. Barbosa GCG, Ali MS, Araujo B, et al. CIDACS-RL: a novel indexing search and scoring-based record linkage
system for huge datasets with high accuracy and scalability. BMC Med Inform Decis Mak. 2020;20(1):289. doi:10.
1186/s12911-020-01285-w
37. Zhang Z, Reinikainen J, Adeleke KA, Pieterse ME, Groothuis-Oudshoorn CGM. Time-varying covariates and
coefficients in Cox regression models. Ann Transl Med. 2018;6(7):121. doi:10.21037/atm.2018.02.12
38. Mansournia MA, Altman DG. Population attributable fraction. BMJ. 2018;360:k757. doi:10.1136/bmj.k757
39. Sic A, Cvetkovic K, Manchanda E, Knezevic NN. Neurobiological implications of chronic stress and metabolic
dysregulation in inflammatory bowel diseases. Diseases. 2024;12(9):220. doi:10.3390/diseases12090220
40. Rostila M, Saarela J, Kawachi I. Suicide following the death of a sibling: a nationwide follow-up study from
Sweden. BMJ Open. 2013;3(4):e002618. doi:10.1136/bmjopen-2013-002618
41. Oliveira Alves FJ, Fialho E, Paiva de Araújo JA, et al. The rising trends of self-harm in Brazil: an ecological
analysis of notifications, hospitalisations, and mortality between 2011 and 2022. Lancet Reg Health Am. 2024;31:
100691. doi:10.1016/j.lana.2024.100691
42. Machado DB, Williamson E, Pescarini JM, et al. Relationship between the Bolsa Família national cash transfer
programme and suicide incidence in Brazil: a quasi-experimental study. PLoS Med. 2022;19(5):e1004000. doi:10.
1371/journal.pmed.1004000
43. Alves FJO, Machado DB, Barreto ML. Effect of the Brazilian cash transfer programme on suicide rates:
a longitudinal analysis of the Brazilian municipalities. Soc Psychiatry Psychiatr Epidemiol. 2019;54(5):599-606.
doi:10.1007/s00127-018-1627-6
44. Rebouças P, Alves FJ, Ferreira A, et al. Avaliação da qualidade do Sistema Brasileiro de Informações sobre
Mortalidade (SIM): uma scoping review. Cien Saude Colet. 2025;30(1):e08462023. doi:10.1590/141381232025301.08462023
45. United Nations Department of Economic and Social Affairs. Goal 3: Ensure healthy lives and promote wellbeing for all at all ages. Accessed October 31, 2025. https://sdgs.un.org/goals/goal3
SUPPLEMENT 1.
eAppendix 1. Data Sources and Linkage of the Datasets

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

13/14

JAMA Network Open | Psychiatry

Mortality Risk Following a Household Suicide

eTable 1. Accuracy Analysis of the Linkage Between CadUnico and Mortality Information System in a Sample of
10 000 Record Pairs
eFigure 1. ROC Curve of the Linkage Between Mortality and CadUnico From 2001 to 2015
eFigure 2. Flowchart of the Selected Population
eTable 2. Summary Measures of the Variable Number of People in the Household
eTable 3. Descriptive Analysis by Index Case
eAppendix 2. Statistical Modeling
eTable 4. Test of Proportional-Hazards Assumption
eFigure 3. Follow-Up
eFigure 4. Log-Log Survival Probability
eTable 5. All-Cause Mortality and Suicide Rates by Characteristics of the Index Suicide Case Among Individuals Who
Have Experienced a Previous Suicide Within the Same Household, 2001 to 2018
eAppendix 3. Sensitivity Analyses
eTable 6. Hazard Ratios for Suicide and All-Cause Mortality Including Time-Dependent Exposure and Duration of
Follow-Up
eTable 7. Hazard Ratios for Suicide and All-Cause Mortality Using a Common Definition of Follow-up Entry for
Exposed and Unexposed
eTable 8. Distribution of Follow-Up Time (Years) by Definition of Baseline Date and Exposure Status
eTable 9. Hazard Ratios for Suicide and All-Cause Mortality Using Common Follow-up Windows for Exposed and
Unexposed
eFigure 5. Factors Associated with Immediate (ⱕ1 year), Intermediate (2-4 years), and Distant (ⱖ5 years) Deaths
by All-Cause Mortality (Excluding Suicide) and Suicide, 2001 to 2018
eFigure 6. Summary of Statistically Significant Risk and Protective Factors for All-Cause Mortality (Excluding
Suicide) and Suicide
eAppendix 4. Pathways Linking Household Suicide Exposure to Subsequent Mortality and Target Groups for
Postvention Interventions
eFigure 7. Pathways Linking Household Suicide Exposure to Subsequent Mortality and Targets for Postvention
Interventions
eReferences
SUPPLEMENT 2.
Data Sharing Statement

JAMA Network Open. 2025;8(11):e2545286. doi:10.1001/jamanetworkopen.2025.45286 (Reprinted)

Downloaded from jamanetwork.com by guest on 12/15/2025

November 25, 2025

14/14

T h e n e w e ng l a n d j o u r na l o f m e dic i n e

original article

Early Palliative Care for Patients with
Metastatic Non–Small-Cell Lung Cancer
Jennifer S. Temel, M.D., Joseph A. Greer, Ph.D., Alona Muzikansky, M.A.,
Emily R. Gallagher, R.N., Sonal Admane, M.B., B.S., M.P.H.,
Vicki A. Jackson, M.D., M.P.H., Constance M. Dahlin, A.P.N.,
Craig D. Blinderman, M.D., Juliet Jacobsen, M.D., William F. Pirl, M.D., M.P.H.,
J. Andrew Billings, M.D., and Thomas J. Lynch, M.D.

A bs t r ac t
Background

Patients with metastatic non–small-cell lung cancer have a substantial symptom
burden and may receive aggressive care at the end of life. We examined the effect
of introducing palliative care early after diagnosis on patient-reported outcomes
and end-of-life care among ambulatory patients with newly diagnosed disease.
Methods

We randomly assigned patients with newly diagnosed metastatic non–small-cell
lung cancer to receive either early palliative care integrated with standard oncologic care or standard oncologic care alone. Quality of life and mood were assessed
at baseline and at 12 weeks with the use of the Functional Assessment of Cancer
Therapy–Lung (FACT-L) scale and the Hospital Anxiety and Depression Scale, respectively. The primary outcome was the change in the quality of life at 12 weeks.
Data on end-of-life care were collected from electronic medical records.

From Massachusetts General Hospital,
Boston (J.S.T., J.A.G., A.M., E.R.G., V.A.J.,
C.M.D., J.J., W.F.P., J.A.B.); the State University of New York, Buffalo (S.A.); Adult
Palliative Medicine, Department of Anesthesiology, Columbia University Medical
Center, New York (C.D.B.); and Yale University, New Haven, CT (T.J.L.). Address
reprint requests to Dr. Temel at Massachusetts General Hospital, 55 Fruit St.,
Yawkey 7B, Boston, MA 02114, or at
­jtemel@partners.org.
N Engl J Med 2010;363:733-42.
Copyright © 2010 Massachusetts Medical Society.

Results

Of the 151 patients who underwent randomization, 27 died by 12 weeks and 107
(86% of the remaining patients) completed assessments. Patients assigned to early
palliative care had a better quality of life than did patients assigned to standard
care (mean score on the FACT-L scale [in which scores range from 0 to 136, with
higher scores indicating better quality of life], 98.0 vs. 91.5; P = 0.03). In addition,
fewer patients in the palliative care group than in the standard care group had
depressive symptoms (16% vs. 38%, P = 0.01). Despite the fact that fewer patients in
the early palliative care group than in the standard care group received aggressive
end-of-life care (33% vs. 54%, P = 0.05), median survival was longer among patients
receiving early palliative care (11.6 months vs. 8.9 months, P = 0.02).
Conclusions

Among patients with metastatic non–small-cell lung cancer, early palliative care led
to significant improvements in both quality of life and mood. As compared with
patients receiving standard care, patients receiving early palliative care had less
aggressive care at the end of life but longer survival. (Funded by an American
Society of Clinical Oncology Career Development Award and philanthropic gifts;
ClinicalTrials.gov number, NCT01038271.)
n engl j med 363;8

nejm.org

august 19, 2010

733

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

T h e n e w e ng l a n d j o u r na l o f m e dic i n e

T

he quality of care and the use of
medical services for seriously ill patients
are key elements in the ongoing debate
over reform of the U.S. health care system.1 Oncologic care is central to this debate, largely because anticancer treatments are often intensive
and costly.2 Comprehensive oncologic services for
patients with metastatic disease would ideally
improve the patients’ quality of life and facilitate
the efficient allocation of medical resources. Palliative care, with its focus on management of
symptoms, psychosocial support, and assistance
with decision making, has the potential to improve the quality of care and reduce the use of
medical services.3,4 However, palliative care has
traditionally been delivered late in the course of
disease to patients who are hospitalized in specialized inpatient units or as a consultative service for patients with uncontrolled symptoms.5,6
Previous studies have suggested that late referrals to palliative care are inadequate to alter the
quality and delivery of care provided to patients
with cancer.7,8 To have a meaningful effect on
patients’ quality of life and end-of-life care, palliative care services must be provided earlier in the
course of the disease.
Metastatic non–small-cell lung cancer, the
leading cause of death from cancer worldwide,9
is a debilitating disease that results in a high
burden of symptoms and poor quality of life; the
estimated prognosis after the diagnosis has been
established is less than 1 year.10-12 We previous­
ly found that introducing palliative care shortly
after diagnosis was feasible and acceptable among
outpatients with metastatic non–small-cell lung
cancer.13 The goal of the current study was to
examine the effect of early palliative care integrated with standard oncologic care on patientreported outcomes, the use of health services,
and the quality of end-of-life care among patients
with metastatic non–small-cell lung cancer. We
hypothesized that patients who received early
palliative care in the ambulatory care setting, as
compared with patients who received standard
oncologic care, would have a better quality of life,
lower rates of depressive symptoms, and less
aggressive end-of-life care.

Me thods
Study Design

static non–small-cell lung cancer in a nonblinded, randomized, controlled trial of early palliative
care integrated with standard oncologic care, as
compared with standard oncologic care alone.
The study was performed at Massachusetts General Hospital in Boston. Eligible patients were
enrolled within 8 weeks after diagnosis and were
randomly assigned to one of the two groups in a
1:1 ratio without stratification. Patients who were
assigned to early palliative care met with a member of the palliative care team, which consisted
of board-certified palliative care physicians and
advanced-practice nurses, within 3 weeks after
enrollment and at least monthly thereafter in the
outpatient setting until death. Additional visits
with the palliative care service were scheduled at
the discretion of the patient, oncologist, or palliative care provider.
General guidelines for the palliative care visits in the ambulatory setting were adapted from
the National Consensus Project for Quality Palliative Care and were included in the study protocol.14
Using a template in the electronic medical record, palliative care clinicians documented the
care they provided according to these guidelines
(see Table 1 in the Supplementary Appendix, available with the full text of this article at NEJM.org).
Specific attention was paid to assessing physical
and psychosocial symptoms, establishing goals
of care, assisting with decision making regarding treatment, and coordinating care on the
basis of the individual needs of the patient.14,15
Patients who were randomly assigned to standard
care were not scheduled to meet with the palliative care service unless a meeting was requested
by the patient, the family, or the oncologist; those
who were referred to the service did not cross
over to the palliative care group or follow the
specified palliative care protocol. All the participants continued to receive routine oncologic care
throughout the study period. Before enrollment
in the study was initiated, the protocol was approved by the Dana Farber/Partners CancerCare
institutional review board. All participants provided written informed consent. The protocol,
including the statistical analysis plan, is available at NEJM.org. All the authors attest that the
study was performed in accordance with the
protocol and the statistical analysis plan.
Patients

From June 7, 2006, to July 15, 2009, we enrolled Patients who presented to the outpatient thoracic
ambulatory patients with newly diagnosed meta- oncology clinic were invited by their medical on734

n engl j med 363;8

nejm.org

august 19, 2010

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

Early Palliative Care for Metastatic Cancer

cologists to enroll in the study; all the medical
oncologists in the clinic agreed to approach, recruit, and obtain consent from their patients.
Physicians were encouraged, but not required, to
offer participation to all eligible patients; no additional screening or recruitment measures were
used. Patients were eligible to participate if they
had pathologically confirmed metastatic non–
small-cell lung cancer diagnosed within the previous 8 weeks and an Eastern Cooperative Oncology Group (ECOG) performance status of 0, 1, or
2 (with 0 indicating that the patient is asymptomatic, 1 that the patient is symptomatic but
fully ambulatory, and 2 that the patient is symptomatic and in bed <50% of the day)16 and were
able to read and respond to questions in English.
Patients who were already receiving care from
the palliative care service were not eligible for
participation in the study.

had to be present for more than half the time,
except for the symptom of suicidal thoughts,
which was included in the diagnosis if it was
present at any time.

Patient-Reported Measures

Data Collection

Health-related quality of life was measured with
the use of the Functional Assessment of Cancer
Therapy–Lung (FACT-L) scale, which assesses
multiple dimensions of the quality of life (physical, functional, emotional, and social well-being)
during the previous week.17 In addition, the lungcancer subscale (LCS) of the FACT-L scale evaluates seven symptoms specific to lung cancer. The
primary outcome of the study was the change
from baseline to 12 weeks in the score on the
Trial Outcome Index (TOI), which is the sum of
the scores on the LCS and the physical well-being
and functional well-being subscales of the FACT-L
scale.
Mood was assessed with the use of both the
Hospital Anxiety and Depression Scale (HADS)
and the Patient Health Questionnaire 9 (PHQ9).18,19 The 14-item HADS, which consists of two
subscales, screens for symptoms of anxiety and
depression in the previous week. Subscale scores
range from 0, indicating no distress, to 21, indicating maximum distress; a score higher than
7 on either HADS subscale is considered to be
clinically significant. The PHQ-9 is a nine-item
measure that evaluates symptoms of major depressive disorder according to the criteria of the
fourth edition of the Diagnostic and Statistical Manual
of Mental Disorders (DSM-IV). A major depressive
syndrome was diagnosed if a patient reported at
least five of the nine symptoms of depression on
the PHQ-9, with one of the five symptoms being
either anhedonia or depressed mood. Symptoms

Participants completed baseline questionnaires
before randomization. Follow-up assessments of
quality of life and mood were performed at 12
weeks (or at an outpatient clinic visit within
3 weeks before or after that time point). Participants who had no scheduled clinic visits within
this period received the questionnaires by mail.
When responses on questionnaires were incomplete, research staff documented the reasons for
which the participant did not give a full response.

n engl j med 363;8

Measures of Health Care Use

Data were collected from the electronic medical
record on the use of health services and end-oflife care, including anticancer therapy, medication
prescriptions, referral to hospice, hospital admissions, emergency department visits, and the date
and location of death. Patients were classified as
having received aggressive care if they met any of
the following three criteria: chemotherapy within
14 days before death, no hospice care, or admission to hospice 3 days or less before death.20-22
Finally, we assessed whether patients’ resuscitation preferences were documented in the outpatient electronic medical record.23

Statistical Analysis

Data obtained through December 1, 2009, were
included in the analyses. The primary outcome
was the change in the score on the TOI from
baseline to 12 weeks. We estimated that with 120
patients, the study would have 80% power to detect a significant between-group difference in the
change in the TOI score from baseline to 12
weeks, with a medium effect size of 0.5 SD.24 The
protocol was amended in August 2008 to allow
for the enrollment of an additional 30 participants
in order to compensate for the loss of any patients
to follow-up.
Statistical analyses were performed with the
use of SPSS software, version 16.0 (SPSS). Descriptive statistics were used to estimate the frequencies, means, and standard deviations of the study
variables. Differences between study groups in
baseline characteristics and clinical outcomes
were assessed with the use of two-sided Fisher’s
exact tests and chi-square tests for categorical

nejm.org

august 19, 2010

735

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

T h e n e w e ng l a n d j o u r na l o f m e dic i n e

variables and independent-samples Student’s t-tests
for continuous variables. Multivariate linear regression analyses, adjusted for baseline scores,
were used to examine the effect of early palliative
care on quality-of-life outcomes. For intentionto-treat analyses, we used the conservative method
of carrying baseline values forward to account
for all missing patient-reported outcome data,
including data that were missing owing to death.
Survival time was calculated from the date of
enrollment to the date of death with the use of
the Kaplan–Meier method. Data from patients who
were alive at the last follow-up (December 1,

2009) were censored on that date. A Cox proportional-hazards model was used to assess the effect of early palliative care on survival, with adjustment for demographic characteristics and
baseline ECOG performance status.

R e sult s
Baseline Characteristics of the Patients

A total of 151 patients were enrolled in the study
(see the figure in the Supplementary Appendix).
The percentage of patients enrolled was similar
for each of the thoracic oncologists in the clinic.

Table 1. Baseline Characteristics of the Study Participants.*
Variable

Standard Care
(N = 74)

Early Palliative Care
(N = 77)

Age — yr

64.87±9.41

64.98±9.73

0.94

36 (49)

42 (55)

0.52

White

70 (95)

77 (100)

Black

3 (4)

0

Asian

1 (1)

0

1 (1)

1 (1)

45 (61)

48 (62)

Female sex — no. (%)
Race — no. (%)‡

P Value†

0.06§

Hispanic or Latino ethnic group‡
Marital status — no. (%)

1.00
1.00

Married
Single

9 (12)

9 (12)

Divorced or separated

12 (16)

12 (16)

Widowed

8 (11)

8 (10)

ECOG performance status — no. (%)¶

0.24

0

30 (41)

26 (34)

1

35 (47)

46 (60)

2

9 (12)

5 (6)

19 (26)

24 (31)

Presence of brain metastases — no. (%)
Initial anticancer therapy — no. (%)

0.48
0.87‖

Platinum-based combination chemotherapy

35 (47)

35 (45)

Single agent

3 (4)

9 (12)

Oral EGFR tyrosine kinase inhibitor

6 (8)

6 (8)

Radiotherapy

26 (35)

27 (35)

Chemoradiotherapy

3 (4)

0

No chemotherapy

1 (1)

0

20 (27)

16 (21)

0.45

16/73 (22)

18/76 (24)

0.85

Anxiety subscale

24/72 (33)

28/77 (36)

0.73

Depression subscale

18/72 (25)

17/77 (22)

0.70

12/72 (17)

9/76 (12)

0.48

Receipt of initial chemotherapy as part
of a clinical trial — no. (%)
Never smoked or smoked ≤10 packs/yr — no./
total no. (%)
Assessment of mood symptoms — no./total no. (%)
HADS**

PHQ-9 major depressive syndrome††

736

n engl j med 363;8

nejm.org

august 19, 2010

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

Early Palliative Care for Metastatic Cancer

Table 1. (Continued.)
Standard Care
(N = 74)

Early Palliative Care
(N = 77)

P Value†

FACT-L scale

91.7±16.7

93.6±16.5

0.50

Lung-cancer subscale

18.7±4.4

20.1±4.4

Trial Outcome Index

55.3±13.1

56.2±13.4

Variable
Scores on quality-of-life measures‡‡

* Plus–minus values are means ±SD. Percentages may not total 100 because of rounding. ECOG denotes Eastern
Cooperative Oncology Group, EFGR epidermal growth factor receptor, FACT-L Functional Assessment of Cancer
Therapy–Lung, HADS Hospital Anxiety and Depression Scale, and PHQ-9 Patient Health Questionnaire 9.
† P values were calculated with the use of two-sided chi-square and Fisher’s exact tests for categorical variables and the
independent-samples Student’s t-tests for continuous variables.
‡ Race or ethnic group was self-reported.
§ The P value is for the between-group comparison of the proportions of patients who were white and those who were
members of a minority group (black and Asian), calculated with the use of Fisher’s exact test.
¶ An ECOG performance status of 0 indicates that the patient is asymptomatic, 1 that the patient is symptomatic but
fully ambulatory, and 2 that the patient is symptomatic and in bed less than 50% of the day.
‖ The P value is for the between-group comparison of the proportion of patients receiving platinum-based combination
chemotherapy and the proportion receiving other treatments, calculated with the use of Fisher’s exact test.
** The HADS consists of two subscales, one for symptoms of anxiety and one for symptoms of depression. Subscale
scores range from 0, indicating no distress, to 21, indicating maximum distress; a score higher than 7 indicates clinically meaningful anxiety or depression.
†† The PHQ-9 is a nine-item measure that evaluates symptoms of major depressive disorder according to the criteria of
the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). A major depressive syndrome
was diagnosed if a patient reported at least five of the nine symptoms of depression on the PHQ-9, with one of the
five symptoms being either anhedonia or depressed mood. Symptoms had to be present for more than half the time,
­except for the symptom of suicidal thoughts, which was included in the diagnosis if it was present at any time.
‡‡ The quality of life was assessed with the use of three measures: the FACT-L scale, on which scores range from 0 to 136,
with higher scores indicating a better quality of life; the lung-cancer subscale of the FACT-L scale, on which scores range
from 0 to 28, with higher scores indicating fewer symptoms; and the Trial Outcome Index, which is the sum of the scores
on the lung-cancer, physical well-being, and functional well-being subscales of the FACT-L scale (scores range from 0
to 84, with higher scores indicating a better quality of life).

No significant differences in demographic characteristics or overall survival were seen between
the study participants and eligible patients who
were not enrolled in the study. The baseline characteristics were well matched between the two
study groups (Table 1). Known prognostic factors,
including age, sex, ECOG performance status,
presence or absence of brain metastases, smoking
status, and initial anticancer therapy, were also
balanced between the study groups. Although
genetic testing was not routinely performed, the
proportions of patients with mutations in the
epidermal growth factor gene (EGFR) were similar between the study groups among the patients
who underwent testing (9% in the palliative care
group and 12% in the standard-treatment group,
P = 0.76). No significant between-group differences were seen in baseline quality of life or mood
symptoms.

palliative care service by the 12th week. The average number of visits in the palliative care group
was 4 (range, 0 to 8). Ten patients who received
standard care (14%) had a palliative care consultation in the first 12 weeks of the study, primarily to address the management of symptoms, with
seven patients having one visit and three having
two visits.
Quality-of-Life and Mood Outcomes

A comparison of measures of quality of life at 12
weeks showed that the patients assigned to early
palliative care had significantly higher scores
than did those assigned to standard care, for the
total FACT-L scale, the LCS, and the TOI, with
effect sizes in the medium range (Table 2). Patients in the palliative care group had a 2.3-point
increase in mean TOI score from baseline to 12
weeks, as compared with a 2.3-point decrease in
the standard care group (P = 0.04) (Fig. 1). With
Palliative-Care Visits
the use of linear regression to control for baseAll the patients assigned to early palliative care, line quality-of-life values, the group assignment
except for one patient who died within 2 weeks significantly predicted scores at 12 weeks on the
after enrollment, had at least one visit with the total FACT-L scale (adjusted difference in mean
n engl j med 363;8

nejm.org

august 19, 2010

737

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

T h e n e w e ng l a n d j o u r na l o f m e dic i n e

Table 2. Bivariate Analyses of Quality-of-Life Outcomes at 12 Weeks.*

Variable
FACT-L score

Standard Care
(N = 47)

Early Palliative Care
(N = 60)

Difference between Early
Care and Standard Care
(95% CI)

P Value†

Effect Size‡

91.5±15.8

98.0±15.1

6.5 (0.5–12.4)

0.03

0.42

LCS score

19.3±4.2

21.0±3.9

1.7 (0.1–3.2)

0.04

0.41

TOI score

53.0±11.5

59.0±11.6

6.0 (1.5–10.4)

0.009

0.52

* Plus–minus values are means ±SD. Quality of life was assessed with the use of three scales: the Functional Assessment
of Cancer Therapy–Lung (FACT-L) scale, on which scores range from 0 to 136, with higher scores indicating better quality of life; the lung-cancer subscale (LCS) of the FACT-L scale, on which scores range from 0 to 28, with higher scores
indicating fewer symptoms; and the Trial Outcome Index (TOI), which is the sum of the scores on the LCS and the
physical well-being and functional well-being subscales of the FACT-L scale (scores range from 0 to 84, with higher
scores indicating better quality of life).
† The P value was calculated with the use of two-sided Student’s t-tests for independent samples.
‡ The effect size was determined with the use of Cohen’s d statistic, which is a measure of the difference between two
means (in this case, the mean in the group assigned to early palliative care group minus the mean in the group assigned
to standard care) divided by a standard deviation for the pooled data. According to the conventional classification, an
effect size of 0.20 is small, 0.50 moderate, and 0.80 large.

[±SE] scores, 5.4±2.4; 95% confidence interval
[CI], 0.7 to 10.0; P = 0.03) and the TOI (adjusted
difference in mean scores, 5.2±1.8; 95% CI, 1.6 to
8.9; P = 0.005), but not on the LCS (adjusted difference in mean scores, 1.0±0.6; 95% CI, –0.2 to 2.3;
P = 0.12). In addition, the percentage of patients
with depression at 12 weeks, as measured by the
HADS and PHQ-9, was significantly lower in the
palliative care group than in the standard care
group, although the proportions of patients receiving new prescriptions for antidepressant drugs
were similar in the two groups (approximately
18% in both groups, P = 1.00) (Fig. 2). The percentage of patients with elevated scores for symptoms of anxiety did not differ significantly between the groups.
The figure in the Supplementary Appendix includes an explanation of missing data according
to study group. There was no significant association between missing data on patient-reported
outcomes at 12 weeks and any baseline characteristic (although there was a trend toward a
significant association between missing data and
assigned treatment [P = 0.07]). When we carried
the baseline scores of the participants forward
for the missing data on patient-reported outcomes, all primary treatment effects were replicated with respect to quality of life (P = 0.04 for
the 12-week FACT-L score, P = 0.01 for the 12-week
LCS score, P = 0.04 for the 12-week TOI score,
and P = 0.04 for the mean change from baseline
to 12 weeks in the TOI score) and mood (P = 0.04
for the comparison of patients with elevated scores
on the HADS depression subscale, and P = 0.02
738

n engl j med 363;8

for the comparison of patients with symptoms
of major depression on the PHQ-9).
End-of-Life Care

At the time of the analysis of end-of-life care, 105
participants (70%) had died; the median duration
of follow-up among participants who died was
5.7 months. Within this subsample, a greater percentage of patients in the group assigned to standard care than in the group assigned to early
palliative care received aggressive end-of-life care
(54% [30 of 56 patients] vs. 33% [16 of 49 patients], P = 0.05). In addition, fewer patients in the
standard care group than in the palliative care
group had resuscitation preferences documented
in the outpatient electronic medical record (28%
[11 of 39 patients who had preferences documented during the course of the study] vs. 53%
[18 of 34 patients], P = 0.05). The study did not
have adequate power to examine specific indicators of aggressive care at the end of life. However, analyses of various measures of utilization,
such as rates of hospitalization and emergency
department visits (Table 2 in the Supplementary
Appendix), as well as the duration of hospice care
(median duration, 11 days in the palliative care
group vs. 4 days in the standard care group;
P = 0.09 with the use of the Wilcoxon rank-sum
test), suggested an improvement in the quality of
care with early palliative care. Despite receiving
less aggressive end-of-life care, patients in the palliative care group had significantly longer survival
than those in the standard care group (median
survival, 11.6 vs. 8.9 months; P = 0.02) (Fig. 3).

nejm.org

august 19, 2010

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

Early Palliative Care for Metastatic Cancer

A FACT-L
7.5

Mean Change

5.0
2.5
0.0
−2.5
−5.0

Standard Care

Early Palliative
Care

Standard Care

Early Palliative
Care

Standard Care

Early Palliative
Care

B LCS
2.0
1.5

Mean Change

Figure 1. Mean Change in Quality-of-Life Scores
from Baseline to 12 Weeks in the Two Study Groups.
Quality of life was assessed with the use of the Functional Assessment of Cancer Therapy–Lung (FACT-L)
scale, on which scores range from 0 to 136, with higher
scores indicating a better quality of life; the lung-cancer
subscale (LCS) of the FACT-L scale, on which scores
range from 0 to 28, with higher scores indicating fewer
symptoms; and the Trial Outcome Index (TOI), which
is the sum of the scores on the LCS and the physical
well-being and functional well-being subscales of the
FACT-L scale (scores range from 0 to 84, with higher
scores indicating a better quality of life). With study
group as the independent variable, two-sided independent-samples Student’s t-tests showed a trend toward
a significant between-group difference in the mean
(±SD) change in scores from baseline to week 12 on
the FACT-L scale (−0.4±13.8 in the standard care group
vs. 4.2±13.8 in the palliative care group; difference between groups, 4.6; 95% confidence interval [CI], −0.8
to 9.9; P = 0.09) (Panel A), no significant betweengroup difference in the mean change in scores on the
LCS (0.3±4.0 and 0.8±3.6 in the two groups, respectively; difference between groups, 0.5; 95% CI, −1.0 to 2.0;
P = 0.50) (Panel B), and a significant between-group
difference in the mean change in scores on the TOI
(−2.3±11.4 vs. 2.3±11.2; difference ­between groups, 4.6;
95% CI, 0.2 to 8.9; P = 0.04) (Panel C). Data are from
the 47 patients in the standard care group and the 60
patients in the palliative care group who completed the
12-week assessments. I bars indicate 95% confidence
intervals.

1.0
0.5
0.0
−0.5
−1.0

C TOI
5.0

This study shows the effect of palliative care
when it is provided throughout the continuum of
care for advanced lung cancer. Early integration
of palliative care with standard oncologic care in
patients with metastatic non–small-cell lung cancer resulted in survival that was prolonged by approximately 2 months and clinically meaningful
improvements in quality of life and mood. Moreover, this care model resulted in greater documentation of resuscitation preferences in the
outpatient electronic medical record, as well as
less aggressive care at the end of life. Less aggressive end-of-life care did not adversely affect
survival. Rather, patients receiving early palliative care, as compared with those receiving standard care alone, had improved survival. Previous
data have shown that a lower quality of life and
depressed mood are associated with shorter survival among patients with metastatic non–smallcell lung cancer.25-27 We hypothesize that improvements in both of these outcomes among
patients assigned to early palliative care may ac-

2.5

n engl j med 363;8

Mean Change

Discussion

0.0
−2.5
−5.0

count for the observed survival benefit. In addition, the integration of palliative care with standard oncologic care may facilitate the optimal
and appropriate administration of anticancer
therapy, especially during the final months of
life. With earlier referral to a hospice program,
patients may receive care that results in better
management of symptoms, leading to stabilization of their condition and prolonged survival.
These hypotheses require further study.
Improving quality of life and mood in patients

nejm.org

august 19, 2010

739

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

T h e n e w e ng l a n d j o u r na l o f m e dic i n e

Patients with Mood Symptoms (%)

50

Standard care

Early palliative care

40

30

20

10

0

HADS-D

HADS-A

PHQ-9

Figure 2. Twelve-Week Outcomes of Assessments
of Mood.
Depressive symptoms were assessed with the use of
the Hospital Anxiety and Depression Scale (HADS),
which consists of two subscales, one for symptoms
of anxiety (HADS-A) and one for symptoms of depression (HADS-D) (subscale scores range from 0, indi­
cating no distress, to 21, indicating maximum distress;
a score higher than 7 on either HADS subscale is considered to be clinically significant) and with the use of
the Patient Health Questionnaire 9 (PHQ-9). The PHQ-9
is a nine-item measure that evaluates symptoms of
major depressive disorder according to the criteria
of the fourth edition of the Diagnostic and Statistical
Manual of Mental Disorders (DSM-IV). A major depressive syndrome was diagnosed if a patient reported at
least five of the nine symptoms of depression on the
PHQ-9, with one of the five symptoms being either
­anhedonia or depressed mood. Symptoms had to be
present for more than half the time, except for the
symptom of suicidal thoughts, which was included in
the diagnosis if it was present at any time. The percentages of patients with mood symptoms, assessed on the
basis of each of these measures, in the group assigned
to standard treatment and the group assigned to early
palliative care, respectively, are as follows: HADS-D,
38% (18 of 47 patients) versus 16% (9 of 57), P = 0.01;
HADS-A, 30% (14 of 47 patients) and 25% (14 of 57),
respectively; P = 0.66; and PHQ-9, 17% (8 of 47 patients)
versus 4% (2 of 57); P = 0.04. The analyses were performed with the use of a two-sided Fisher’s exact test.

with metastatic non–small-cell lung cancer is a
formidable challenge, given the progressive nature of the illness.28 The improvement we observed in the quality of life among patients assigned to early palliative care, as indicated by a
mean change in the TOI score by 12 weeks that
was approximately 5 points higher in the palliative care group than in the standard care group,
is similar to the improvement in the quality of
life that has been observed among patients who
have a response to cisplatin-based chemotherapy.29 Most studies show that there is a deteriora740

n engl j med 363;8

tion in the quality of life over time, which is
consistent with the results in the standard care
group in our study.30-32 Despite similar cancer
therapies in our two study groups, the patients
assigned to early palliative care had an improved
quality of life, as compared with those receiving
standard care. Rates of depression also differed
significantly between the groups, with approximately half as many patients in the palliative care
group as in the standard care group reporting
clinically significant depressive symptoms on the
HADS, and this effect was not due to a betweengroup difference in the use of antidepressant
agents.
To date, evidence supporting a benefit of palliative care is sparse, with most studies having
notable methodologic weaknesses, especially with
respect to quality-of-life outcomes.8 One study
with sufficient power to examine quality-of-life
outcomes showed that among patients receiving
radiation therapy, a multidisciplinary intervention
focused on education, behavioral modification,
and coping style resulted in improvements in the
quality of life.33 A recent study showed that Project ENABLE (Educate, Nurture, Advise, Before Life
Ends), a telephone-based, psychoeducational program for patients with advanced cancer, significantly improved both quality of life and mood.34
However, the percentage of patients who completed the study assessments was somewhat low,
and the study did not use a traditional palliative
care model.
Our study also showed that early outpatient
palliative care for patients with advanced cancer
can alter the use of health care services, including
care at the end of life. Other studies of outpatient palliative care have failed either to investigate these outcomes or to show an effect on the
use of resources.5,34,35 In our trial, significantly
more patients in the group assigned to early palliative care than in the standard care group had
resuscitation preferences documented in the outpatient electronic medical record, an essential
step in clarifying and ensuring respect for patients’ wishes about their care at the end of
life.36 Early introduction of palliative care also
led to less aggressive end-of-life care, including
reduced chemotherapy and longer hospice care.
Given the trends toward aggressive and costly
care near the end of life among patients with
cancer, timely introduction of palliative care may
serve to mitigate unnecessary and burdensome
personal and societal costs.20,37

nejm.org

august 19, 2010

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

Early Palliative Care for Metastatic Cancer

n engl j med 363;8

100

80

Patients Surviving (%)

Our study has several advantages over previous studies, in which investigators have often relied on referrals to palliative care instead of using a recruitment approach designed to obtain a
representative sample.5,35 Because all patients with
a new diagnosis of metastatic non–small-cell
lung cancer were eligible for enrollment in our
study, we extended the generalizability of our
findings. Another strength of our trial was the
low rate of loss to follow-up and the high percentage of participants who completed the study
assessments. In addition, the dropout rate by
week 12 was less than 1%, further supporting
the feasibility and acceptability of early palliative
care. Finally, the trial was adequately powered to
detect changes in both quality of life and mood,
and we prospectively collected data on end-oflife care.
Several limitations of the study deserve mention. It was performed at a single, tertiary care
site with a specialized group of thoracic oncology providers and palliative care clinicians, thereby limiting generalization of the results to other
care settings or patients with other types of
cancer. In addition, because the sample lacked
diversity with respect to race and ethnic group,
we were unable to assess the effect of these
important factors on study outcomes. Although
we used a randomized, controlled design, both
the patients and the clinicians were aware of the
study assignments. To account for possible influences of care that are not specific to the palliative care provided, follow-up investigations
should include a control group that receives a
similar amount of attention. In addition, we did
not deny palliative care consultations to participants receiving standard care, and a small minority of patients in the standard care group was
seen by the palliative care team. The data from
these patients were analyzed with the data from
their assigned study group (standard care), a factor that may have diluted our findings. Finally,
carrying the last observation forward for all missing data in the intention-to-treat analyses is a
conservative approach; therefore, the actual treatment effect of early palliative care may be greater
than we report.
Early integration of palliative care for patients
with metastatic non–small-cell lung cancer is a
clinically meaningful and feasible care model
that has effects on survival and quality of life
that are similar to the effects of first-line chemotherapy in such patients.28,38,39 As compared with

60

40
Early palliative care

20
Standard care
0

0

10

20

30

40

Months

Figure 3. Kaplan–Meier Estimates of Survival According to Study Group.
Survival was calculated from the time of enrollment to the time of death,
if it occurred during the study period, or to the time of censoring of data on
December 1, 2009. Median estimates of survival were as follows: 9.8 months
(95% confidence interval [CI], 7.9 to 11.7) in the entire sample (151 patients),
11.6 months (95% CI, 6.4 to 16.9) in the group assigned to early palliative
care (77 patients), and 8.9 months (95% CI, 6.3 to 11.4) in the standard
care group (74 patients) (P = 0.02 with the use of the log-rank test). After
adjustment for age, sex, and baseline Eastern Cooperative Oncology Group
performance status, the group assignment remained a significant predictor
of survival (hazard ratio for death in the standard care group, 1.70; 95% CI,
1.14 to 2.54; P = 0.01). Tick marks indicate censoring of data.

the study participants who received standard
care, those who were assigned to early palliative
care had improved mood, more frequent documentation of resuscitation preferences, and less
aggressive end-of-life care. Although our findings must be replicated in a variety of care settings and cancer populations, the results nonetheless offer great promise for alleviating distress
in patients with metastatic disease and addressing critical concerns regarding the use of health
care services at the end of life.
Supported by an American Society of Clinical Oncology Career
Development Award and philanthropic gifts from the Joanne
Hill Monahan Cancer Fund and Golf Fights Cancer.
Dr. Temel reports receiving payment for developing continuing medical education (CME) programs from InforMEDical; and
Dr. Lynch, serving on the board of Infinity Pharmaceuticals, receiving consulting fees from Roche, Boehringer Ingelheim,
Merck, AstraZeneca, Bristol-Myers Squibb, and Sanofi-Aventis,
royalties from Partners HealthCare, and payment for developing
CME programs from InforMEDical. No other potential conflict
of interest relevant to this article was reported.
Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.

nejm.org

august 19, 2010

741

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

Early Palliative Care for Metastatic Cancer
References
1. Mongan JJ, Ferris TG, Lee TH. Options

for slowing the growth of health care
costs. N Engl J Med 2008;358:1509-14.
2. Weinstein MC, Skinner JA. Comparative effectiveness and health care spending — implications for reform. N Engl J
Med 2010;362:460-5.
3. Ferris FD, Bruera E, Cherny N, et al.
Palliative cancer care a decade later:
­accomplishments, the need, next steps —
from the American Society of Clinical Oncology. J Clin Oncol 2009;27:3052-8.
4. Levy MH, Back A, Benedetti C, et al.
NCCN clinical practice guidelines in oncology: palliative care. J Natl Compr Canc
Netw 2009;7:436-73.
5. Follwell M, Burman D, Le LW, et al.
Phase II study of an outpatient palliative
care intervention in patients with metastatic cancer. J Clin Oncol 2009;27:20613.
6. Jordhøy MS, Fayers P, Loge JH, AhlnerElmqvist M, Kaasa S. Quality of life in
palliative cancer care: results from a cluster randomized trial. J Clin Oncol 2001;
19:3884-94.
7. Morita T, Akechi T, Ikenaga M, et al.
Late referrals to specialized palliative care
service in Japan. J Clin Oncol 2005;23:
2637-44.
8. Zimmermann C, Riechelmann R,
Krzyzanowska M, Rodin G, Tannock I.
Effectiveness of specialized palliative care:
a systematic review. JAMA 2008;299:
1698-709.
9. Jemal A, Siegel R, Ward E, Hao Y, Xu J,
Thun MJ. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225-49.
10. Hopwood P, Stephens RJ. Depression
in patients with lung cancer: prevalence
and risk factors derived from quality-oflife data. J Clin Oncol 2000;18:893-903.
11. Hopwood P, Stephens RJ. Symptoms
at presentation for treatment in patients
with lung cancer: implications for the
evaluation of palliative treatment. Br J
Cancer 1995;71:633-6.
12. Lutz S, Norrell R, Bertucio C, et al.
Symptom frequency and severity in patients with metastatic or locally recurrent
lung cancer: a prospective study using the
Lung Cancer Symptom Scale in a community hospital. J Palliat Med 2001;4:15765.
13. Temel JS, Jackson VA, Billings JA, et al.
Phase II study: integrated palliative care
in newly diagnosed advanced non-smallcell lung cancer patients. J Clin Oncol
2007;25:2377-82.
14. Clinical practice guidelines for quality
palliative care. 2nd ed. Pittsburgh: National Consensus Project, 2009. (Accessed July
23, 2010, at www.nationalconsensusproject
.org.)
15. Jackson VA, Jacobsen JC, Greer J, et al.
Components of early intervention out­
patient palliative care consultation in pa-

742

tients with incurable NSCLC. J Clin Oncol
2009;27:Suppl:20635. abstract.
16. Oken MM, Creech RH, Tormey DC, et
al. Toxicity and response criteria of the
Eastern Cooperative Oncology Group. Am
J Clin Oncol 1982;5:649-55.
17. Cella DF, Bonomi AE, Lloyd SR, Tulsky
DS, Kaplan E, Bonomi P. Reliability and
validity of the Functional Assessment of
Cancer Therapy-Lung (FACT-L) quality of
life instrument. Lung Cancer 1995;12:199220.
18. Zigmond AS, Snaith RP. The Hospital
Anxiety and Depression Scale. Acta Psychiatr Scand 1983;67:361-70.
19. Kroenke K, Spitzer RL, Williams JB.
The PHQ-9: validity of a brief depression
severity measure. J Gen Intern Med 2001;
16:606-13.
20. Earle CC, Landrum MB, Souza JM,
Neville BA, Weeks JC, Ayanian JZ. Aggressiveness of cancer care near the end of
life: is it a quality-of-care issue? J Clin Oncol 2008;26:3860-6.
21. American Society of Clinical Oncology Quality Oncology Practice Initiative.
ASCO QOPI summary of measures. Fall
2009. (Accessed July 23, 2010, at http://
qopi.asco.org/Documents/QOPIFall2009
MeasuresSummary8-18-09.pdf.)
22. Tang ST, Wu SC, Hung YN, Chen JS,
Huang EW, Liu TW. Determinants of aggressive end-of-life care for Taiwanese
cancer decedents, 2001 to 2006. J Clin
Oncol 2009;27:4613-8.
23. Temel JS, Greer JA, Admane S, et al.
Code status documentation in the outpatient electronic medical records of patients with metastatic cancer. J Gen Intern Med 2010;25:150-3.
24. Cella D. The Functional Assessment of
Cancer Therapy–Lung and Lung Cancer
Subscale assess quality of life and meaningful symptom improvement in lung cancer. Semin Oncol 2004;31:Suppl 9:11-5.
25. Maione P, Perrone F, Gallo C, et al.
Pretreatment quality of life and functional status assessment significantly predict
survival of elderly patients with advanced
non-small-cell lung cancer receiving chemotherapy: a prognostic analysis of the
Multicenter Italian Lung Cancer in the
Elderly Study. J Clin Oncol 2005;23:686572.
26. Movsas B, Moughan J, Sarna L, et al.
Quality of life supersedes the classic prognosticators for long-term survival in local­
ly advanced non-small-cell lung cancer:
an analysis of RTOG 9801. J Clin Oncol
2009;27:5816-22.
27. Pirl WF, Temel JS, Billings A, et al.
Depression after diagnosis of advanced
non-small cell lung cancer and survival:
a pilot study. Psychosomatics 2008;49:
218-24.
28. Brown J, Thorpe H, Napp V, et al. Assessment of quality of life in the support-

n engl j med 363;8

nejm.org

ive care setting of the Big Lung Trial in
non-small-cell lung cancer. J Clin Oncol
2005;23:7417-27.
29. Cella D, Eton DT, Fairclough DL, et al.
What is a clinically meaningful change
on the Functional Assessment of Cancer
Therapy-Lung (FACT-L) Questionnaire? Results from Eastern Cooperative Oncology
Group (ECOG) Study 5592. J Clin Epidemiol 2002;55:285-95.
30. Bonomi P, Kim K, Fairclough D, et al.
Comparison of survival and quality of life
in advanced non-small-cell lung cancer
patients treated with two dose levels of
paclitaxel combined with cisplatin versus
etoposide with cisplatin: results of an
Eastern Cooperative Oncology Group trial.
J Clin Oncol 2000;18:623-31.
31. Langer CJ, Manola J, Bernardo P, et al.
Cisplatin-based therapy for elderly patients
with advanced non-small-cell lung cancer:
implications of Eastern Cooperative Oncology Group 5592, a randomized trial.
J Natl Cancer Inst 2002;94:173-81.
32. Sandler AB, Nemunaitis J, Denham C,
et al. Phase III trial of gemcitabine plus
cisplatin versus cisplatin alone in patients
with locally advanced or metastatic nonsmall-cell lung cancer. J Clin Oncol 2000;
18:122-30.
33. Rummans TA, Clark MM, Sloan JA, et
al. Impacting quality of life for patients
with advanced cancer with a structured
multidisciplinary intervention: a randomized controlled trial. J Clin Oncol 2006;
24:635-42.
34. Bakitas M, Lyons KD, Hegel MT, et al.
Effects of a palliative care intervention
on clinical outcomes in patients with advanced cancer: the Project ENABLE II randomized controlled trial. JAMA 2009;302:
741-9.
35. Rabow MW, Dibble SL, Pantilat SZ,
McPhee SJ. The comprehensive care team:
a controlled trial of outpatient palliative
medicine consultation. Arch Intern Med
2004;164:83-91.
36. Walling A, Lorenz KA, Dy SM, et al.
Evidence-based recommendations for information and care planning in cancer
care. J Clin Oncol 2008;26:3896-902.
37. Emanuel EJ, Ash A, Yu W, et al. Managed care, hospice use, site of death, and
medical expenditures in the last year of
life. Arch Intern Med 2002;162:1722-8.
38. Non-Small Cell Lung Cancer Collaborative Group. Chemotherapy in non-small
cell lung cancer: a meta-analysis using
updated data on individual patients from
52 randomised clinical trials. BMJ 1995;
311:899-909.
39. Spiro SG, Rudd RM, Souhami RL, et
al. Chemotherapy versus supportive care
in advanced non-small cell lung cancer:
improved survival without detriment to
quality of life. Thorax 2004;59:828-36.
Copyright © 2010 Massachusetts Medical Society.

august 19, 2010

The New England Journal of Medicine is produced by NEJM Group, a division of the Massachusetts Medical Society.
Downloaded from nejm.org on November 12, 2025.
Copyright © 2010 Massachusetts Medical Society. All rights reserved, including those for text and data mining, AI training, and similar technologies.

theqtz_email)

USA (https://www.emjreviews.com/en-us/amj/flagship-journal/article/does-gut-microbiome-composition(https://www.emjreviews.com)
influence-the-efficacy-of-psychiatric-drugs-j190125/?
utm_campaign=6%20at%206&utm_medium=email&_hsenc=p2ANqtz-9CrKAPjrBgZ2vbF6UeZ4dgW5t9jusATMl6rbULWt0qkEnUK7UTJlG1t_QBQ_ltfhnnhgIdFPJN8_LXO3WgvrwORRTdn15c0J_wxWQcaK1aMXWy8&_hsmi=368055006&utm_content=368046508&utm_source=hs_email)

Does Gut Microbiome Composition Influence the Efficacy of Psychiatric
Drugs?
7 Mins 13 Mar 2025
Flagship Journal (https://www.emjreviews.com/therapeutic-area/flagshipjournal/) Microbiology & Infectious Diseases (https://www.emjreviews.com/therapeuticarea/microbiology-infectious-diseases/) Gastroenterology
(https://www.emjreviews.com/therapeutic-area/gastroenterology/) Neurology

(https://www.emjreviews.com/therapeutic-area/neurology/)
View All Articles (https://www.emjreviews.com/type/article/)

Download PDF (https://www.emjreviews.com/wp-content/uploads/2025/03/Does-Gut-

Microbiome-Composition-Influence-the-Efficacy-of-Psychiatric-Drugs.pdf)
Authors:

* Nadia Suryawinata (https://www.emjreviews.com/contributor/nadia-suryawinata/) , 1 Sarkis Mazmanian
(https://www.emjreviews.com/contributor/sarkis-mazmanian/) 1

1. California Institute of Technology, Pasadena, California, USA
*Correspondence to nsuryawi@caltech.edu (emailto:nsuryawi@caltech.edu)
Disclosure:
Mazmanian has received consulting fees from Axial Therapeutics and Nuanced Health for work unrelated to the
current manuscript. The authors have declared no conflicts of interest.
Received:
03.02.25
Accepted:
24.02.25
Citation:
EMJ. 2025;10[1]:40-46. https://doi.org/10.33590/emj/CSEB2440 (https://doi.org/10.33590/emj/CSEB2440).
Keywords:

Has our content helped YOU

elevate the quality of healthcare
Anxiety, biomarkers, depression, gut microbiome, microbiome composition.
for YOUR patients?
Submit your one-click score

Each article is made available under the terms of the Creative Commons Attribution-Non
Commercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/).

SUMMARY
Altered gut microbiome profiles correlate with anxiety and depression in humans, and work in animal
models has identified specific bacterial taxa and/or microbiome-derived metabolites that influence complex
emotional behaviours. Intriguingly, many pharmaceuticals, including widely used oral treatments for anxiety
and depression, can be chemically modified by microbes in the gastrointestinal tract, which may lead to
drug inactivation. The authors highlight the importance of integrating research across microbial culture
systems, animal models, and multi-omics analyses of clinical cohorts to gain mechanistic insights into
whether microbiome composition determines efficacy, bioavailability, and tolerability of neuropsychiatric
medications. This hypothesis, if validated, may have profound implications for personalised drug treatment
plans and microbiome-based biomarker development.

THE RECIPROCAL RELATIONSHIP BETWEEN THE GUT MICROBIOME AND
MEDICATIONS
The gut microbiome, comprising a staggering 3.8×1013 bacteria along with microscopic fungi, archaea, and
viruses in humans,1 plays crucial roles in shaping and maintaining host health. Gut microbes support a wide
range of physiological functions including digestion, immune modulation, metabolism, and neuronal
signaling. Disruptions in host-microbe interactions are associated with a range of human diseases, such as
inflammatory bowel disease (IBD),2 cancer,3 Type 2 diabetes,4 and neurological disorders.5
The gut microbiome is highly dynamic, with community composition influenced by intrinsic factors such as
host genetics,6 but also strongly determined by extrinsic/environmental contributors,7 including diet and
medication.8 Because diet and drugs are modifiable, understanding the interactions between environmental
factors and the gut microbiome offers an exciting and tractable opportunity for development of personalised
medicines.
Most pharmaceuticals are administered orally. These substances are either absorbed in the small intestine,
where the microbiome is sparse, or pass to the colon, where the densest and most diverse microbial
communities reside. Additionally, drugs absorbed in the small intestine may be modified (or not) and
secreted back into the intestine, creating new opportunities for exposure to the gut microbiome.9
Consumption of antibiotics, unsurprisingly, has profound effects on the gut microbiome. Acute exposure to
a single course of antibiotics can result in the transient reduction or loss of microbial taxa that are important
for basic metabolic functions such as carbohydrate fermentation,10 energy production, bile acid
transformation,11 and lipid absorption. While most individuals treated with antibiotics experience a rapid
recovery of microbiome composition, for some it may take up to 6 months to fully recover their original (predrug) microbiome.12 Loss of community stability and, consequently, compromise of normal metabolic
functions of the microbiome may lead to opportunistic infections,12,13 deficits in gut barrier integrity,14

weakening of the immune system,15,16 and other unintended consequences. While antibiotics likely have the
most profound impact on microbiome function, emerging evidence suggests that other medications may
also compromise the microbiome, albeit to a subtler degree.
The vast majority of pharmaceutical drugs were developed against human targets (e.g., proteins,
molecules, metabolic pathways), are diverse in structure, and are often consumed for extended periods of
time, making it challenging to predict their direct or indirect effects on the microbiome. However, some
drug–microbiome interactions have been uncovered. The common Type 2 diabetes medication metformin
alters gut microbiome composition in patients, increasing microbial taxa that promote glucose metabolism
and thereby increasing its therapeutic effect.17 Methotrexate, a first-line treatment for rheumatoid arthritis,
alters microbiome composition in patients and in human microbiome colonised mice, with transplantation of
a drug-modified microbiome into drug-naïve mice being sufficient to reduce immune activation.18 The
benzisoxazole ring structure in risperidone, an atypical antipsychotic used for schizophrenia and bipolar
disorder, is chemically modified by gut microbes, leading to its rapid excretion and thus potentially reducing
efficacy and altering dosing regimens in ways that may vary between patients.19
Informed by these findings, there is growing interest in understanding how the gut microbiome may be
influenced by, and may influence the efficacy of, various drug classes. Emerging evidence has identified
novel microbial transformations of drugs that may alter the intended outcomes of medications.9,20 Given the
emerging and likely intricate relationship between gut bacteria and brain function, drug–microbiome
interactions in the context of neuropsychiatric disorders represent a particularly interesting area of study.
This perspective will first examine how the gut microbiome influences drug metabolism in vivo, drawing
from studies in mice and humans with anxiety and major depressive disorder (MDD). The authors will then
review known drug–microbiome interactions, primarily through examples beyond neuropsychiatric
medications, as these well-characterised cases provide insights into the methodologies needed for future
study of microbial metabolism of psychiatric drugs. Finally, the authors will discuss how integrating these
approaches can provide an actionable framework for understanding the role of microbial influences on the
efficacy and other features of psychiatric drugs.

THE GUT MICROBIOME AND BRAIN HEALTH: INSIGHTS FROM ANXIETY
AND DEPRESSION
Anxiety disorders represent the most common class of neuropsychiatric conditions, and are characterised
by a persistent avoidance response even in the absence of imminent danger.21 Often co-occurring with
depression,22 which is marked by a prolonged loss of interest in activities, anxiety disorders significantly
impact quality of life in up to a quarter of the USA and European populations.23,24
In recent years, studies conducted in both mouse models and human cohorts have described a functional
role for the gut microbiome in the development of anxiety and depression (Figure 1A). The gut microbiome
is stereotypically altered in individuals with anxiety or MDD,25,26 and is speculated to affect symptoms via
altered neurotransmitter production,27 inflammation/cytokines,28 the hypothalamic–pituitary–adrenal (HPA)
axis,29 vagus nerve,27 and other potential mechanisms. These associations are supported by animal

models. Germ-free mice, which are raised without any exposure to microbes, exhibit reduced anxiety-like
behaviour, and the reintroduction of a normal microbiome early in life is sufficient to restore anxiety-like
traits of standard laboratory mice.30 Transplantation of microbiomes from mice that have experienced
chronic stress into naïve recipient mice induces behaviours consistent with depression, and
supplementation with Lactobacillus alleviates this effect.31
The gut and brain communicate through various pathways (neuronal, endocrine, immunological) and these
interactions involve factors that can be influenced by a diverse array of microbes and their products. For
instance, treatment with the bacterium Lactobacillus rhamnosus (JB-1) has been shown to alleviate anxiety
and depression-like behaviours in mice.27 This effect occurs through the differential regulation of GABA
receptors in the brain and is dependent on the vagus nerve, as vagotomised mice do not exhibit the same
behavioural improvements when treated with L. rhamnosus (Figure 1B). Gut bacteria can also produce small
molecule metabolites that then travel to the brain and alter cell function: the gut microbial metabolite 4ethylphenyl sulfate (4-EPS) impairs oligodendrocyte differentiation in mice and increases anxiety-like
behaviour.32 In contrast, treating mice with the human commensal Bacteroides fragilis is able to alleviate
anxiety-like and autism-associated features (Figure 1C).33

Figure 1: The microbiome and microbially-derived metabolites modulate host nervous system
function.
Figure 1A is generated on Biorender.com.
In humans, large cohort studies surveying the microbiomes of depressed patients have revealed
stereotypical alterations. Notably, MDD patients often show depletion of genera such as Subdoligranulum
and Coprococcus, and an increase in Eggerthella, alongside changes in their metabolomes, particularly
increased lipid metabolism.26,34 A recent study integrated microbiome sequencing data from faecal
samples of individuals with anxiety and depressive disorders, including those taking medications, to train
machine learning algorithms that could successfully predict both the presence of these disorders and
medication use based on microbiome profiles alone.35 While effect sizes in human studies remain modest
and may necessitate further replication, research to date on the potential pathogenic or protective effects of
the gut microbiome in neuropsychiatric conditions represents an exciting frontier of research at the
intersection of microbiology, neuroscience, and human health.

Microbiome Modulation of Neuropsychiatric Drugs
Selective serotonin reuptake inhibitors (SSRIs) and selective norepinephrine reuptake inhibitors (SNRIs) are
first-line treatments for both anxiety and MDD.36 While these drug classes have been shown to be more
effective than placebo for generalised anxiety disorder, their benefits are often accompanied by a
therapeutic lag and significant variability in response rates, particularly in terms of long-term acceptability
and sustained efficacy.37 Given that SSRIs and SNRIs are orally administered, an important question is
whether their interaction with the gut microbiome contributes to the substantial differences in therapeutic
acceptability observed across patient populations. Interindividual variations in gut microbiome composition
may influence drug metabolism and bioavailability, potentially explaining why some patients respond better
to treatment than others.
Recent studies with high-throughput, in vitro culture-based screening systems have revealed extensive
drug–microbiome interactions (Figure 2). In one study, researchers exposed 76 individual strains of diverse
human gut bacterial taxa to over one hundred commonly prescribed drugs, including medications for
anxiety.38 This work found that metabolic reactions were taxon-specific; i.e., Bacteroidetes primarily
hydrolysed drugs with ester or amide groups, while most other strains metabolised drugs containing a nitro
or azide group. Of note, 10% of the strains chemically transformed anxiolytics, significantly reducing active
drug levels in culture, with the SSRI fluoxetine emerging as the most widely metabolised anxiolytic across
isolated bacterial strains. Another study incubated different complex communities derived from human
faecal samples with drugs used to treat anxiety, again finding that microbiome composition can broadly
influence drug metabolism.39 Some communities had the metabolic capacity to degrade specific drugs,
while others did not, highlighting interindividual variability in microbiome-driven drug metabolism.

Figure 2: The microbiome modulates drugs, potentially affecting their therapeutic function in the host.
Created on Biorender.com.
Researchers have also leveraged publicly available repositories to develop models to predict drug–
microbiome interactions, such as SIMMER (Similarity algorithms that Identify MicrobioMe Enzymatic
Reactions)40 and AGORA2 (Assembly of Gut Organisms through Reconstruction and Analysis, version 2).41
SIMMER combines metagenome-assembled genomes, protein homology, and enzyme databases to
predict bacterial drug metabolism. This tool identified candidate gut bacterial enzymes, primarily
carboxypeptidase G2-like enzymes, with sequence similarity to an environmental enzyme known to
hydrolyse methotrexate.40 Experimental testing of strains containing these enzymes confirmed
methotrexate degradation. AGORA2 provides a resource for reconstructing metabolic pathways from
metagenomic datasets, and incorporates clinical parameters such as BMI and age to facilitate rapid
prediction of drug metabolism in epidemiological cohorts.41 Both SIMMER and AGORA2 provide interactive
frameworks, allowing researchers to prioritise microbial species, gene products, and pathways of particular
relevance for a given disorder and drug class.
While high-throughput screening and large-scale dataset analyses have provided valuable insights, efforts
to fully characterise drug–microbiome interactions remain ongoing, with only a few examples to date that
have identified products of drug metabolism and even fewer cases tested functionally. For instance,
Levodopa (L-DOPA), the first-line treatment for Parkinson’s disease, is degraded by Eggerthella lenta and
Enterococcus faecalis.42 These bacterial species were shown to contain enzymes for conversion of LDOPA into m-tyramine through decarboxylation and dihydroxylation, which may reduce L-DOPA
bioavailability and impact treatment efficacy. Another well-known example of a drug–microbiome interaction
is 5-aminosalicylic acid (5-ASA), used to treat IBD, whose efficacy is reduced by microbial metabolism.43 By
longitudinally monitoring IBD patients on 5-ASA treatment using metagenomics, metatranscriptomics, and

metabolomics, researchers identified twelve previously uncharacterised microbial acetyltransferases that
were upregulated in non-responders. In vitro assays confirmed that these enzymes acetylate 5-ASA into an
inactive form, providing a mechanistic link between microbial metabolism and drug response.
In addition to metabolising drugs, some gut bacteria have been shown to actively transport and
bioaccumulate drugs in vitro without modifying their chemical structure (Figure 2).44 Duloxetine, an SNRI,
bioaccumulates in diverse gut species, including many from the Firmicutes phylum (Streptococcus
salivarius, Clostridium bolteae, Clostridium saccharolyticum, Ruminococcus gnavus, Lactobacillus
plantarum, and Lacticaseibacillus paracasei), resulting in altered endogenous metabolism and secretion
profiles. Duloxetine modulates Caenorhabditis elegans movement in a dose-dependent manner, and
colonisation with the Escherichia coli IAI1 strain that is capable of bioaccumulating duloxetine attenuates this
behaviour, highlighting that drug–microbiome interactions can impact behavioural outcomes.44
Finally, the gut microbiome can regulate host drug transporters, thus influencing pharmacokinetics.
Differences in microbiome composition, such as between conventionally-raised and germ-free animals,
alter the expression of the efflux transporter P-glycoprotein (P-gp/ABCB1),45 which may contribute to
pharmacokinetic variability for P-gp substrate drugs, including the SSRI sertraline and the antipsychotic
risperidone. However, whether degradation, modification, bioaccumulation and/or altered transport of SSRIs
or SNRIs impact anxiety or depression-like behaviours in mammalian model systems remains unexplored to
date, defining a frontier of future research.

TOWARDS A HOLISTIC UNDERSTANDING OF DRUG–MICROBIOME
INTERACTIONS
While microbial cell culture-based experiments offer rigorous insights into drug–microbiome interactions,
these systems are unable to capture the physiology of an organism and its associated microbiome, with
studies in freely behaving animals required to advance this research toward understanding effects on
emotional behaviours. Recent in vitro findings have also revealed that reductions in drug levels do not
necessarily indicate microbial metabolism.20 Abiotic factors, including spontaneous degradation, ion
suppression, surface adsorption, and bioaccumulation, can have strong effects on drug activity.
To ensure reproducible and clinically relevant results, it is important to test drug–microbiome interactions
within their native host context, minimising artefacts introduced by culture conditions. Moreover, it is
possible that long-term medication use can reshape the gut environment and microbiome composition,
which then secondarily influences symptoms or treatment outcomes, though this concept remains
hypothetical in the absence of empiric evidence. Disentangling these factors requires an integrated
approach, combining multi-omics analyses of diverse human cohorts with rodent models or non-human
primate models that are amenable to experimental approaches to define functional outcomes. Given that
microbial bio-transformations largely fall within a defined set of reaction types, such as reduction,
hydrolysis, decarboxylation, and dealkylation, identifying overarching principles governing these
transformations may be feasible. Leveraging large-scale machine learning models trained on high-

resolution microbiome and metabolomics datasets could offer a powerful strategy to predict drug
modifications and their downstream effects, ultimately guiding the design of more precise and effective
therapeutic interventions.
As our understanding of drug–microbiome interactions becomes more refined, the development of
predictive frameworks for drug efficacy and tolerability based on an individual’s symptoms, lifestyle,
medication history, and microbiome status will be increasingly feasible. Such tools could one day help tailor
pharmacological treatments to maximise therapeutic benefit, ultimately advancing precision medicine. It is
conceivable that gut microbiome variations explain inter-individual responses to numerous classes of oral
drugs, beyond those for neuropsychiatric conditions, and potentially even injectables via microbiome
modulation of immune profiles (e.g., immune checkpoint inhibitors)46-48 and metabolic states (e.g., weight
loss drugs).49,50 Identifying microbiome-based markers that quantitatively predict variance in drug response
in defined patient populations may streamline drug discovery and development, improve efficacy rates and
response times, and reduce side effects.
References
1. Sender R et al. Revised estimates for the number of human and bacteria cells in the body. PLoS Biol.
2016;14(8):e1002533.
2. Xavier RJ, Podolsky DK. Unravelling the pathogenesis of inflammatory bowel disease. Nature.
2007;448(7152):427-34.
3. Louis P et al. The gut microbiota, bacterial metabolites and colorectal cancer. Nat Rev Microbiol.
2014;12(10):661-72.
4. Qin J et al. A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature.
2012;490(7418):55-60.
5. Morais LH et al. The gut microbiota-brain axis in behaviour and brain disorders. Nat Rev Microbiol.
2021;19(4):241-55.

Share:

(https://www.facebook.com/sharer/sharer.php?u=&t=)
(https://twitter.com/intent/tweet?text=)
(https://www.linkedin.com/shareArticle?mini=true&url=&title=&summary=&source=)
More great content like this
- straight to your inbox >
(https://www.emjreviews.com/mailing-list)

Rate this content's potential impact on patient outcomes

Average rating 5 / 5. Vote count: 2

Related To This Subject

(https://www.emjreviews.com/microbiology-infectious-diseases/news/rising-bedaquilineresistance-threatens-tuberculosis-control-efforts/)
(https://www.emjreviews.com/microbiology-infectious-diseases/news/rising-bedaquilineresistance-threatens-tuberculosis-control-efforts/)
Microbiology & Infectious Diseases
(https://www.emjreviews.com/therapeutic-area/microbiology-infectious-diseases/)
26th June

Rising Bedaquiline Resistance Threatens Tuberculosis Control Efforts

(https://www.emjreviews.com/microbiology-infectious-diseases/news/rising-bedaquilineresistance-threatens-tuberculosis-control-efforts/)

(https://www.emjreviews.com/microbiology-infectious-diseases/news/sharp-rise-in-syphilisfound-in-turkiye-hiv-study/)
(https://www.emjreviews.com/microbiology-infectious-diseases/news/sharp-rise-in-syphilisfound-in-turkiye-hiv-study/)
Microbiology & Infectious Diseases
(https://www.emjreviews.com/therapeutic-area/microbiology-infectious-diseases/)
25th June

Sharp Rise in Syphilis Found in Türkiye HIV Study

(https://www.emjreviews.com/microbiology-infectious-diseases/news/sharp-rise-in-syphilisfound-in-turkiye-hiv-study/)

Elevating the quality of healthcare globally

Therapy Area
About Us

(https://twitter.com/EMJReviews)
(https://www.facebook.com/emjreviews)
(https://www.youtube.com/channel/UCaOYR5zTz5XcB-ZGNwj467g)
(https://www.linkedin.com/company/3321578/)
This site is intended for healthcare professionals.
Copyright © 2025 European Medical Group LTD trading as European Medical Journal. All rights reserved. European Medical Journal is for
informational purposes and should not be considered medical advice, diagnosis or treatment recommendations.
Ts & Cs (https://www.emjreviews.com/terms-conditions/)
(https://www.emjreviews.com/cookie-policy/)
Website by Vibe Agency (https://www.vibeagency.uk)

Privacy Policy (https://www.emjreviews.com/privacy-policy/)

Cookie Policy

OPEN ACCESS

PSICO
Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022
e-ISSN: 1980-8623 | ISSN-L: 0103-5371

http://dx.doi.org/10.15448/1980-8623.2022.1.37342

SEÇÃO: ARTIGO

Suicidal ideation in mothers of asthmatic children and
adolescents in a subspecialty outpatient practice
Ideação suicida em mães de crianças e adolescentes asmáticos em ambulatório
especializado
Ideación suicida en madres de niños y adolescentes asmáticos en consulta externa de
subespecialidad
Auxiliadora Damianne
Pereira Vieira da Costa1
orcid.org/0000-0003-3072-8273
doradami@gmail.com

Leticia Marques dos Santos2
orcid.org/0000-0001-5963-2166
marquesleticia@hotmail.com

Abstract: We aimed to investigate prevalence and factors associated with

Suicide ideation (SI) in mothers of asthmatic children. This cross-sectional study
included 362 dyads of mothers and children with asthma aged 2 to 14 years who
attended two pediatric outpatient clinics in Brazil. We assessed the presence of SI
(Self-Report Questionnaire-20), the occurrence of stressful events and maternal
social support. The prevalence of SI was 8.6%. Low maternal education, exposure
to serious illness, and low perception of social support in its affective–social interaction dimension remained significantly associated with SI in the final model.
Thus, life stressors, social support and low maternal education accounted for
most of the variation in prevalence of maternal SI. There were no effects of child
asthma severity on maternal SI in this study.

Mércia Lamenha Medeiros1

Keywords: suicidal ideation, mental health, maternal behavior, asthma

orcid.org/0000-0002-1776-3181
mercia.medeiros@famed.ufal.br

Resumo: Nosso objetivo foi investigar a prevalência e os fatores associados

Camila Oliveira Hansen1
orcid.org/0000-0002-5994-0772
milahansen_@hotmail.com

Yasmin Cardoso Monteiro
Formiga1
orcid.org/0000-0002-7221-5225
yascardoso09@gmail.com

Claudio Torres de Miranda1
orcid.org/0000-0002-9602-6736
mirandaclaudio@gmail.com

Recebido em: 11 mar. 2020.
Aprovado em: 13 dez. 2021.
Publicado em: 9 ago. 2022

à ideação suicida (IS) em mães de crianças asmáticas. Este estudo transversal incluiu 362 díades de mães e crianças com asma de 2 a 14 anos em dois
ambulatórios pediátricos no Brasil. Avaliamos a presença de IS (Self-Report
Questionnaire-20), a ocorrência de eventos estressantes e o suporte social
materno. A prevalência de IS materna foi de 8,6%. Escolaridade materna inferior
a oito anos, doença materna grave e a baixa percepção de suporte social em
sua dimensão afetivo-social permaneceram significativamente associadas à IS
no modelo final. Portanto, eventos estressores maternos, suporte social e baixa
escolaridade materna foram os responsáveis pela maior parte da variação na
prevalência de IS materna. Não houve efeitos da gravidade da asma infantil na
IS materna neste estudo.

Palavras-chave: ideação suicida, saúde mental, comportamento materno, asma
Resumen: Este estudio investigo la prevalencia y los factores asociados com

ideación suicida (IS) en madres de niños asmáticos. Participaron 362 díadas de
madres y niños con asma de 2 a 14 años en dos clínicas pediátricas ambulatorias en Brasil. Evaluamos la presencia de IS (Self-Report Questionnaire-20), la
ocurrencia de eventos estresantes y el apoyo social materno. La prevalencia
de IS materno fue del 8,6%. La educación materna de menos de ocho años, la
enfermedad materna grave y la baja percepción de apoyo social en su dimensión afectivo-social se mantuvieron significativamente asociadas con el SI en el
modelo final. Entonces, los eventos de estrés materno, el apoyo social y la baja
educación materna explicaron la mayor parte de la variación en la prevalencia
materna de IS. No hubo efectos de la gravedad del asma infantil en el IS materno
en este estudio.

Palabras clave: ideación suicida, salud mental, conducta materna, asma
Artigo está licenciado sob forma de uma licença
Creative Commons Atribuição 4.0 Internacional.

1
2

Universidade Federal de Alagoas (UFAL), Maceió, AL, Brasil.
Universidade Federal da Bahia (UFBA), Salvador, BA, Brasil.

2/11

Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022 | e-37342

Every year, over 800,000 people die from

generally consistent (Liu & Miller, 2014).

suicide worldwide (World Health Organization,

As caring for patients with chronic mental or

2014). Suicide is the 15 most common cause of

physical illness can be a stressful event, a growing

death and accounts for 1.4% of all deaths glo-

number of studies have investigated aspects of

bally; 75.5% of these deaths occur in developing

the caregiver’s mental health, including SI, whose

countries (Cha et al., 2018; Turecki & Brent, 2016;

prevalence may range from 10.3 to 18% (Huang

World Health Organization, 2014, 2017). Suicidal

et al., 2018; Koyama et al., 2017; O’Dwyer et al.,

behavior is a spectrum that includes suicidal

2016; Park et al., 2013; Skeen et al., 2014). Studies

ideation (SI), planning, attempt, and the action of

revealed much heterogeneity, with SI prevalence

committing suicide itself (World Health Organiza-

depending on child’s illness type and other fac-

tion, 2014). However, suicidal ideation (SI) inclusion

tors associated with the caregiver, such as the

in suicidal behavior is controversial because the

presence of CMD, depression, and anxiety, social

factors associated with SI may differ from the

support, age, and associated chronic disease

factors underlying suicide attempt and suicide

(Huang et al., 2018; O’Dwyer et al., 2016; Park et

itself (Klonsky et al., 2016; Wetherall et al., 2018;

al., 2013; Skeen et al., 2014). Being single, female,

World Health Organization, 2014). The SI preva-

and unemployed; having low perception of social

lence was 9.2% in a multicenter study involving

support; and presenting a mental disorder have

17 countries. Studying SI and its determinants is

been frequently identified as factors for increased

important because up to 60% of the transitions

risk of SI among caregivers (Huang et al., 2018;

to suicide planning and attempt can occur in the

O’Dwyer et al., 2016). In the particular case of chil-

first year after suicidal thoughts are developed

dren, studies evaluating the presence of suicidal

(Nock et al., 2008).

thoughts in their caregivers, especially mothers

th

As SI can be an important predictor of death by

(if we consider that they are the main caregivers

suicide (Nock et al., 2008; World Health Organiza-

in this age group), are lacking. (Lise et al., 2017).

tion, 2014), it is extremely important to identify SI

Moreover, no studies have tested any type of

as a topic for planning suicide prevention. Some

socioeconomic and/or psychosocial model for

theories regarding suicide and suicidal behavior

SI in mothers of children with asthma, the most

are rooted in the ideation-to-action framework.

prevalent chronic disease in childhood. (Asher &

These theories consider SI development and SI

Pearce, 2014; Global Initiative for Asthma, 2015)

transition to suicide attempt as distinct processes

Thus, this study aims to investigate the socioeco-

(Joiner, 2005; Klonsky & May, 2015; Klonsky et

nomic and psychosocial factors that are associa-

al., 2017; O’Connor, 2011). The Three-Step The-

ted with an increased prevalence of SI in mothers

ory (3ST), for example, hypothesizes that: 1) SI

of asthmatic children in subspecialty outpatient

results from the combination of pain (mainly

practice. Since having a child with asthma can be

psychological) and hopelessness; 2) among those

a stressful event for the mother, we will consider

who experience one or both, connectedness is

the additional stress of severe asthma in the child

a crucial protective factor against the escalating

as a determinant for SI in these mothers.

SI and 3) progression from ideation to attempts
depends on dispositional, acquired, and practical

Methods

contributors to the capacity to attempt suicide
(Klonsky & May, 2015; Klonsky et al., 2017). Regarding the first and second steps of 3ST, stressful
life events (SLE) may act as a trigger and social
support as a protective factor on the escalating
SI. In the first case, association between negative stressful situations and SI and behavior was

Participants
This study was conducted in two public pediatric pulmonology outpatient clinics that are reference for attendance of children and adolescents
in the state of Alagoas, Brazil. Eligible participants
were mothers of asthmatic children aged 2 to 14

Auxiliadora Damianne Pereira Vieira da Costa • et al.
Suicidal ideation in mothers of asthmatic children and adolescents in a subspecialty outpatient practice

years selected by convenience.

3/11

(Lopes & Faerstein, 2001).

A total of 481 eligible mothers were invited

Maternal Social Support. Maternal social su-

to participate in the study. Once they agreed to

pport was assessed using the Social Support

participate and signed a consent form, a face-

Scale (Medical Outcomes Study Questions; MOS-

-to-face interview was conducted in a private

-SSS). The instrument consisted of 19 questions

room before medical appointment. Seventeen

and answers on a five-item Likert scale (“never”

mothers refused to participate, and 102 question-

– 1, “rarely” – 2, “sometimes” – 3, “almost always”

naires were excluded due to inconsistencies in

– 4 and “always” – 5), validated for the Brazilian

their completion or by noting that mothers were

population (GRIEP et al., 2005). For the purposes

embarrassed to complete one or more items in

of this study, the items were organized to cover

the questionnaire.

three dimensions of social support: 1) affective–

Mother-child dyads included mainly male

positive social interaction (7 items), 2) emotional–

children, mothers with low education, and se-

informational (8 items), and 3) material support (4

vere financial problems (see table 1 for detailed

items) (Griep et al., 2005). The higher the score,

information on socio-demographic characte-

the greater the perception of social support. The

ristics, maternal factors, and asthma severity in

scores in each dimension were dichotomized at

the sample).

high and low, using the first distribution quartile
as a cut-off point.

Measurements
Maternal SI. Information about maternal SI
was obtained through the question “Have you
had thoughts about ending your life in the past
30 days?” from the Self Report Questionnaire
(SRQ-20). This is a common mental disorder (CMD)
screening questionnaire comprising 20 questions
and dichotomous answers (yes/no) on symptoms
over the previous 30 days and validated in Brazil
(Harding et al., 1980; Mari & Williams, 1986). Participants were divided into groups of mothers
with and without SI.
Maternal stressful life events (SLE). The occurrence of maternal stressful life events (SLE) in
the previous 12 months was measured through
nine close-ended questions about events or
unpleasant situations taking place over the previous 12 months, with dichotomized answers in
yes or no – exposed or non-exposed, respectively
(Lopes & Faerstein, 2001). Each one of the nine
events was evaluated as an isolated variable
(serious illness, hospitalization, death of close
family member, severe financial problems, change
of residence, separation/divorce, physical aggression, mugging/robbery). This questionnaire
showed good test-retest reliability, with most
of the questions having a good stability when
reported by adults in a previous study in Brazil

Support from relatives and friends. Support
from relatives and friends was measured through the questions: 1) “How many relatives do you
feel comfortable with and can talk about almost
everything?” and 2) “How many friends do you
feel comfortable with and can talk about almost
everything?”, also extracted from MOS-SSS mentioned above. Both variables were dichotomized
based on presenting or not support from relatives
or friends, at least one.
Covariates. Child’s asthma severity was considered an adjustment variable, as this sample
consists of mothers of asthmatic children. Child’s
gender, child’s age, maternal age, maternal educational level, economic classification, and maternal smoking were considered as possible
adjustment variables if they achieved significance
(p<0.05) in the bivariate analysis for predictors of
SI in the mothers. Asthma severity was defined
based on the type of medication that was being
used (Global Initiative for Asthma, 2015) and was
categorized as mild or moderate to severe for
analysis. For economic classification, the ABEP
(Associação Brasileira de Empresas de Pesquisa)
questionnaire was used, which categorizes economic classes as A, B, C, D, and E, in witch A was
the highest and E was the lowest. For analysis, the
economic classes were dichotomized in A/B/C

4/11

Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022 | e-37342

and D/E.

association at a level of p ≤ 0.05 in the bivariate
analysis with SI entered a multivariate regression

Procedures
This was a cross-sectional study conducted
between June 2015 and December 2017 in two
outpatient pediatric clinics in the State of Alagoas,
Brazil. Participants included dyads of mothers
and their respective asthmatic child aged 2 to 14
years. The data were collected through structured
interviews with the mothers and by using data
from patients’ medical records.
After the Informed and direct consent was
obtained from the mothers, questionnaires were
applied. Application of the instrument (face-to-face), as well as the interview environment (waiting
room) were used to minimize the gaps in filling
in the questionnaires. However, the interviewees
might have felt embarrassed with one or more
items in the questionnaire. When this happened, we invalidated the instrument entirely and
the participant was excluded from the research.
When the mothers answered positively to the
question “Have you had thoughts about ending
your life in the past 30 days?”, they were referred

model. Estimates by point (prevalence ratio; PR)
and adjusted 95% confidence intervals (95% CIs)
were calculated by using Poisson regression with
robust variance in order to produce point and
interval estimates that were lower than those
obtained using logistic regression which would
overestimate the associations for outcomes (Coutinho et al., 2008). A value of p <0.05 was considered as statistically significant in the multivariate
regression model.
To quantify the additional contribution of each
group of independent variables to the variation
observed in the dependent variable (maternal SI)
explained by the model, such groups of interest
variables were progressively added. Contribution
with the addition of each group of variables to
the model was measured by change in R2. Successive models for association with maternal SI
were: 1) socio-demographic factors; 2) Model 1 +
perception of maternal social support; 3) Model 2
+ maternal exposure to SLE in the previous year;
4) Model 3 + asthma severity.

to the clinic’s psychology service for follow-up.
This assurance complies with items III, III.2 of the

Results

Guidelines and Regulatory Norms for Research

The maternal SI prevalence in the previous

Involving Humans, RESOLUTION No. 466, of the

month was 8.6%. All mothers with SI also had

Brazilian NATIONAL HEALTH COUNCIL.

evidence of common mental disorder (CMD),

The Research Ethics Committee of the Federal

with eight or more positive answers to SRQ-20, a

University of Alagoas approved this study with

cut-off point defined in the Brazilian validation of

the protocol number 1.091.863.

the questionnaire (Mari & Williams, 1986). Half of

All the analyses were conducted using the

the mothers in the sample had CMD. Almost half

STATA version 13.0 program. Association between

of the dyads belonged to economic classes D

the presence of maternal SI and categorical in-

and E. Exposure to severe financial problems was

dependent variables was analyzed by means of

the most frequently reported SLE in the previous

the qui square test. The variables that presented

year, followed by loss of a close relative (Table 1).

Auxiliadora Damianne Pereira Vieira da Costa • et al.
Suicidal ideation in mothers of asthmatic children and adolescents in a subspecialty outpatient practice

5/11

Table 1 – Socio-demographic characteristics, maternal factors and asthma severity of the sample (n=362)
Characteristics

N (%)
31 (8,6%)
183 (50,55%)
232 (64,09%)
161 (44,48%)
146 (40,33%)
116 (32,04%)
212 (58,56%)
175 (48,34%)
18 (4,97%)
82 (22,65%)
91 (25,14%)
103 (28,45)
64 (17,68%)
166 (45,86%)
190 (52,49%)

Maternal SI*
Maternal CMD**
Gender of the child, male
Age of child (≤ 5 years)
Asthma severity in the child (moderate / severe)
Maternal age (> 35 years)
Maternal education (less than eight years)
Economic class (D/E)
Maternal smoking
Affective – positive social interaction support, low
Emotional – informational support, low
Material support, low
Relatives support (none)
Friends support
Maternal stressful life events (SLE) in the last year (2 or more)

105 (29,01%)
29 (8,01%)
116 (32,04%)
206 (56,91%)
63 (17,40%)
50 (13,81%)
39 (10,77%)
17 (4,70%)
14 (3,87%)

Serious illness
Hospitalization
Death of close family member
Severe financial problems
Forced change of residence
Separation/divorce
Mugging/robbery
Physical aggression
Discrimination

* CMD: Common mental disease; **SI: Suicidal ideation
Bivariate Relationships between maternal

classes D and E showed a positive association

SI, sociodemographic characteristics, mater-

with SI (PR: 2,95; CI95%: 1,24 – 7,02 and PR: 2,24;

nal psychosocial factors, and Child’s Asthma

CI95%: 1,09 – 4,63, respectively). Child’s asthma

severity. Mothers with less than eight years of

severity was not associated with maternal SI

schooling and mothers belonging to economic

report (Table 2).

Table 2 – Clinical and sociodemographic characteristics of the mother-child dyads, according to the
presence of SI (n=362)
Characteristics

Without SI (n=331)

With SI (n=31)

PR

CI (95%)

p

Gender of the child, Male

213 (64.3%)

19 (61.3%)

0,89

0,44 – 1,77

0,73

Age of child, ≤ 5 years

149 (45.0%)

12 (38.7%)

0,79

0,39 – 1,58

0,50

Asthma severity in the child, Moderate /
severe

131 (39.6%)

15 (48.4%)

1,39

0,71 – 2,72

0,34

Maternal age, > 35 years

105 (31.7%)

11 (35.5%)

1,17

0,58 – 2,35

0,67

Maternal education, Less than eight years

187 (56.5%)

25 (80.6%)

2,95

1,24 – 7,02

0,01*

Economic class (ABEP), D/E

154 (46.5%)

21 (67.7%)

2,24

1,09 – 4,63

0,03*

Maternal smoking, Yes

16 (4.8%)

2 (6.4%)

1,32

0,34 – 5,10

0,69

* p < 0,05
* ABEP: Associação Brasileira de Empresas de Pesquisa

6/11

Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022 | e-37342

Among the maternal psychosocial factors,

aggression and discrimination) were significantly

perception of low social support in its three di-

associated with maternal SI (Table 3). Interper-

mensions and exposure to most stressors in the

sonal support provided by relatives was also

previous year (serious illness, hospitalization,

inversely associated with maternal SI.

separation/divorce, and being victim of physical
Table 3 – Maternal psychosocial characteristics, according to the presence of SI (n=362)
Characteristics

Without SI
(n=331)

With SI
(n=31)

PR

CI (95%)

P

Serious illness

90 (27.2%)

15 (48.4%)

2,29

1,18 – 4,47

0,01*

Hospitalization

21 (6.3%)

8 (25.8%)

3,99

1,96 – 8,13

<0,01*

Death of close family member

103 (31.1%)

13 (41.9%)

1,53

0,78 – 3,02

0,22

Severe financial problems

184 (55.6%)

22 (71.0%)

1,85

0,88 – 3,91

0,11

Forced change of residence

58 (17.5%)

5 (16.1%)

0,91

0,36 – 2,29

0,84

Separation/divorce

41 (14.4%)

9 (29.0%)

2,55

1,25 – 5,23

0,01*

Mugging/robbery

34 (10.3%)

5 (16.1%)

1,59

0,65 – 3,91

0,31

Physical aggression

11 (3.3%)

6 (19.3%)

4,87

2,31 – 10,28

<0,01*

Discrimination

10 (3.0%)

4 (12.9%)

3,68

1,49 – 9,10

0,01*

Low Affective – positive social interaction support

62 (18.7%)

20 (64.5%)

6,21

3,10 – 12,43

<0,01*

Low Emotional – informational support

78 (23.6%)

13 (41.9%)

2,15

1,10 – 4,22

0,02*

Low Material support

88 (26.6%)

15 (48.4%)

2,36

1,21 – 4,59

0,01*

No support from relatives

53 (16.0%)

11 (35.5%)

2,56

1,29 – 5,08

0,01*

No support from Friends

151 (45.6%)

15 (48.4%)

1,11

0,56 – 2,17

0,77

* p < 0,05
Maternal SI, Stressors, Social Support and

multivariate model, we accounted for elements

Child’s Asthma severity: a series of Multivaria-

of the socio-demographic factors of mother-child

te Models. To predict the dependent variable

dyads first, describing the relationships of social

maternal SI, we estimated a series of regression

support and stressors to the mother’s SI before

models to identify the independent contributions

we incorporated the additional stress of child’s

of socio-demographic factors, social support,

asthma severity (Table 4). Support from relatives

life stressors and childhood asthma severity to

and perception of social support added 16% of

maternal SI. The variables that presented asso-

explanation to the SI model in this study measured

ciation at a level of p < 0.05 with SI entered the

by change in R2. Thus, social support explained

multivariate regression model. Although asthma

a considerable amount of additional variation in

severity was not associated with maternal SI in

maternal SI, over and above socio-demographi-

the bivariate analysis, we added this variable to

cs factors. The included SLE added 7% to the

the model to identify a possible strengthening or

explanation provided by socio-demographic

buffering effect of it on the other variables. In this

factors and social support. Inclusion of asthma

Auxiliadora Damianne Pereira Vieira da Costa • et al.
Suicidal ideation in mothers of asthmatic children and adolescents in a subspecialty outpatient practice

7/11

severity did not explain additional variation in the

significantly associated with SI in the final model,

cumulative model of maternal SI to the model, as

even with the addition of child’s asthma severity

observed by unaltered R . Maternal exposure to

to the model. The complete model accounted

severe illness and the low social support in their

for 27% of maternal SI in this sample.

2

affective–social interaction dimension remained
Table 4 – Multivariate regression model for SI asthmatic children and adolescents’ mothers
SI

Model 1
socio-demographic
factors
(R2 0,04)

Model 2
(Model 1 +
perception of
maternal social
support)
(R2 0,20; Change
in R2:0,16)

Model 3
(Model 2 + maternal exposure
to SLE in the
previous year)
(R2 0,27; Change
in R2:0,07)

Model 4
(Model 3 + asthma
severity)
(R2 0,27; Change in
R2:0,00)

PR (CI 95%)

PR (CI 95%)

PR (CI 95%)

PR (CI 95%)

Economic classification
(D/E)

1.36 (0.56 – 3.29)

1.18 (0.54 – 2.60)

1,03 (0.43 – 2.51)

1.03 (0.40 – 2.66)

Maternal educational
level
< 8 years of schooling

2.85 (0.96 – 8.44)

2.51 (0.97 – 6.52)

2,59 (1.05 – 6.38)*

2.59 (1.04 – 6.47)*

Low Affective – positive
social interaction support

6.43 (2.37 – 17.46)*

6.86 (2.66 – 17.71)*

6.86 (2.67 – 17.65)*

Low Emotional – informational support

1.08 (0.49 – 2.38)

1.10 (0.54 – 2.22)

1.10 (0.53 – 2.27)

Low Material support

0.98 (0.44 – 2.18)

0.73 (0.29 – 1.80)

0.73 (0.30 – 1.79)

No support from relatives

2.09 (0.99 – 4.42)

2.05 (0.94 – 4.48)

2.05 (0.91 – 4.64)

Serious illness

3.25 (1.27 – 8.27)*

3.24 (1.25 – 8.44)*

Hospitalization

1.02 (0.36 – 2.87)

1.02 (0.36 – 2.93)

Separation/divorce

1.11 (0.45 – 2.76)

1.11 (0.45 – 2.75)

Physical aggression

2.61 (0.89 – 7.65)

2.62 (0.83 – 8.25)

Discrimination

1.66 (0.74 – 3.74)

1.66 (0.69 – 3.99)

Asthma severity

1.00 (0.45 – 2.30)

* p < 0,05

Discussion

previous year). For example, a study involving

This study including mothers of asthmatic

young adults in Scotland revealed up to 23% of

children in a subspecialty outpatient practice

suicidal thoughts at one time in life and 10.6% in

regarding suicidal thoughts in the previous month

the previous 12 months (O’Connor et al., 2018). In

found a SI prevalence of 8.6%, which was similar to

Brazil, a study carried out with pregnant women

that found in a previous multicenter study – 9.2%

and using a methodology like the methodolo-

(Nock et al., 2008). Thus, there does not seem to

gy described here found SI prevalence of 6.3%

exist a higher prevalence of SI in this population.

(Huang et al., 2012).

SI prevalence can differ from one study to another

Based on the assumption that context matters,

depending on the evaluation method (response

we built the mother’s social context by crea-

to only one question or a specific questionnai-

ting a model of demographics, social support

re, population characteristics, and assessment

and life stress that served as a backdrop for the

time – if ever in life, the previous month, or the

additional stress of having a child with severe

8/11

Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022 | e-37342

asthma. By looking further into the model, we

with a chronic condition.

found that demographics mattered somewhat,

Several factors, mainly psychosocial factors,

but the lack of social support and experience of

are associated with SI in caregivers including

life stressors explained more of the presence of

CMD, depression, anxiety, low perception of social

maternal SI. Moreover, the child’s actual asthma

support, exposure to stressful events, low quality

severity had not a relationship to the mother’s SI

of life, and coping strategies (Huang et al., 2018;

and did not make an additional contribution to the

O’Dwyer et al., 2016; Park et al., 2013; Skeen et

relationship between the psychosocial variables

al., 2014). For SLE, studies involving caregivers

and maternal SI.

are lacking. The most common SLE identified in

Different from previous studies evaluating the

a British study involving 1066 patients with psy-

presence of SI in caregivers of individuals with

chological morbidity and SI were loss of family

chronic diseases such as HIV, chronic kidney

or friend, interpersonal conflicts, severe illness,

disease, cerebrovascular disease, mental disor-

financial crisis, and interpersonal violence (Mc-

ders, and cancer (Huang et al., 2018; Park et al.,

Feeters et al., 2015). In an epidemiological study

2013; Skeen et al., 2014), we assessed whether

conducted in the United States involving 34,653

the severity of the disease was associated with

adults with major depressive disorder, SI was

the presence of SI. We observed a lack of as-

associated with stressors loss or victimization,

sociation between disease severity in child and

problems with interpersonal relationships, serious

the presence of maternal SI, including child’s

problems with neighbors, friends, or relatives, and

asthma severity to the multivariate analysis mo-

financial difficulties (Wang et al., 2015). Although

del. Although prospective studies with a larger

the above-mentioned stressors were associated

number of subjects should confirm or not this lack

with SI in other studies, in the present study, only

of association, it seems that having a child with

the event severe illness remained associated with

severe asthma per se may not be causally related

SI after multivariate analysis.

to maternal SI. Asthma severity is a possible stres-

In relation to social support, previous studies

sor, but other events related to interaction with

shows that it exerts an effect on SI among adoles-

the environment and social support may exert a

cents by mediating its relationship with stressors

greater influence on suicidal thinking, possibly

(Kang et al., 2017). Further, the lack of support

related to the individual’s inherent feeling of

from relatives has also been associated (in the

non-belonging and non-connectivity regardless

unadjusted analysis) with the presence of SI in

of the amount of care that is required.

caregivers of patients with physical and mental

In line with the 3ST theory, which considers that

illness in a tertiary hospital in Taiwan (Huang et

events that cause pain and hopelessness trigger

al., 2018). The present study adds that perceived

IS, we identified in our study that the SLE serious

lack of social support in their affective–social in-

illness increased the risk of SI in mothers of chil-

teraction dimension was significantly associated

dren with asthma. In addition, low social support

with SI in mothers of asthmatic children, in an

on its affective and social interaction dimension

adjusted model.

remained associated with SI in an adjusted analy-

Among the socio-demographic factors as-

sis, strengthening the idea contained in the 3ST

sociated with SI occurrence, only the maternal

hypothesizing that connectedness may act as a

schooling factor remained statistically significant

protective factor against progression of SI. Thus, in

after adjusted analysis. A study in Korea evaluated

mothers of asthmatic children, the event serious

the risk factors for SI in migraine patients and,

illness (a trigger) and the lack of affective and

through logistic regression analysis, found that

social interaction support (the connectedness

patients with SI were more likely to have a low

protective factor) are important determinants of

educational level - measured in years of study -

suicidal thinking in mothers caring for children

as compared to patients without SI (Kim & Park,

Auxiliadora Damianne Pereira Vieira da Costa • et al.
Suicidal ideation in mothers of asthmatic children and adolescents in a subspecialty outpatient practice

9/11

2014). Among caregivers of cancer patients, being

of a simplified instrument with rapid application

female, single, and unemployed were also socio-

(20-30 minutes). There was no need for the in-

-demographic factors associated with higher risk

terviewers to go through complex training. The

of SI (Park et al., 2013).

instrument could be useful mainly in the context

This study had some limitations. First, because

of primary healthcare, could be applied in an

of its cross-sectional design, we were not able to

outpatient waiting room, is easily accessible, and

determine cause and effect relationship betwe-

has low cost for the health system. Once the risk

en SI and the considered factors. Because both

is identified, the caregiver should be referred to

the exposure and outcome are determined at

specialized care for appropriate diagnosis and

the same time, no temporality between these

treatment. The importance of identifying SI and

variables can be inferred. However, the present

associated factors could enable early interven-

study was useful to provide a snapshot of the

tion and prevention or block the process leading

association between SI and maternal psychoso-

from ideation to suicidal behavior through active

cial at one point in time whose direction of effect

search in an interview with mothers. In addition,

can be better indicated in a prospective study.

providing support to the mother will probably

Second, some possible confounding factors may

improve their children’s asthma status.

have affected self-reporting of SI and attempts,
including educational level, understanding and
interpretation of the questionnaire, and respondent’s willingness to disclose this information
(World Health Organization, 2014). Application
of the instrument (face-to-face), as well as the
interview environment (waiting room) were used
to minimize the gaps in filling in the questionnaires. However, the interviewees may have felt
embarrassed, which may have altered their responses and consequently underestimated the
results of the questions related to the stressor
physical aggression event and the presence of
SI for example. The researchers in relation to the
risks of constraints offered guarantee of data
confidentiality and withdrawal from the study
without consequences for the treatment of the
child to the research participants. Finally, there
is no questions about frequency, intensity, or
duration of SI in the suicide item of the SRQ-20,
thus limiting its ability to assess the severity of
suicidality.
Despite these concerns, the present study
has many important strengths, such as demons-

Conclusions
The SI prevalence in asthmatic children’s mothers was the same as in the general population.
Previously described psychosocial factors, as
maternal education, exposure to stressors (in this
case, serious illness) and low perception of social
support in their affective–social interaction was
also significantly associated with SI in mothers
of asthmatic children. Our results points to the
lack of association between severity of disease
in children and SI in their mothers. In general, the
presence of SI is better explained by the experience of life stressors and low perception of social
support. Finally, the current study suggests that it
is possible to identify maternal SI and associated
factors with the aid of a simplified instrument
during a routine visit to a child outpatient unit.

Conflicts of interest
The authors declare that they have no conflict
of interest.

trating a model where the psychosocial context

References

explains much more IS in mothers of asthmatic

Asher, I., & Pearce, N. (2014). Global burden of asthma
among children. The international journal of tuberculosis
and lung disease: the official journal of the International
Union against Tuberculosis and Lung Disease, 18(11),
1269–78. https://doi.org/10.5588/ijtld.14.0170

children than the severity of the disease in the
child. Moreover, it raises the possibility of identifying maternal SI and associated factors in a
routine outpatient child consultation by means

10/11

Psico, Porto Alegre, v. 53, n. 1, p. 1-11, jan.-dez. 2022 | e-37342

Cha, C., Franz, P., Guzmán, E., Glenn, C., Kleiman, E.,
Nock, M. (2018). Annual Research Review: Suicide
among youth – epidemiology, (potential) etiology, and
treatment. Journal of Child Psychology and Psychiatry,
59(4), 460-482. https://doi.org/ 10.1111/jcpp.12831
Coutinho, L., Scazufca, M., Menezes, P. (2008). Métodos
para estimar razão de prevalência em estudos de corte
transversal. Revista de Saúde Pública, 42(6), 992–998.
https://doi.org/10.1590/S0034-89102008000600003
Global Initiative For Asthma. (2015). Global strategy
for asthma management and prevention. GINA report.
http://ginasthma.org/wp-content/uploads/2016/01/
GINA_Report_2015_Aug11-1.pdf
Griep, R., Chor, D., Faerstein, E., Werneck, G., Lopes, C.
(2005). [Construct validity of the Medical Outcomes Study’s social support scale adapted to Portuguese in the
Pró-Saúde Study]. Cadernos de saude publica, 21(3), 70314. https://doi.org/10.1590/S0102-311X2005000300004
Harding, T., Arango, M., Baltazar, J., Climent, C., Ibrahim,
H., Ladrido-Ignacio, L., Murthy, R., Wig, N. (1980). Mental disorders in primary health care: a study of their
frequency and diagnosis in four developing countries. Psychological medicine, 10(2), 231–41. https://doi.
org/10.1017/s0033291700043993
Huang, H., Faisal-Cury, A., Chan, Y., Tabb, K., Katon, W.,
Menezes, P. (2012). Suicidal ideation during pregnancy:
prevalence and associated factors among low-income women in São Paulo, Brazil. Archives of Women’s
Mental Health, 15(2), 135-138. https://doi.org/10.1007/
s00737-012-0263-5
Huang, Y., Hsu, S., Hung, C., Wang, L., Chong, M. (2018).
Mental health of caregivers of individuals with disabilities: Relation to Suicidal Ideation. Comprehensive
psychiatry, 81, 22-27. https://doi.org/10.1016/j.comppsych.2017.11.003
Joiner, T. E. (2005). Why People Die By Suicide. Harvard
Univ. Press.
Kang, B., Kang, J., Park, H., Cho, Y., Hur, Y., Sim, W.,
Byeon, G., Kim, K.(2017). The Mediating Role of Parental
Support in the Relationship between Life Stress and
Suicidal Ideation among Middle School Students. Korean journal of family medicine, 38(4), 213-219. https://
doi.org/10.4082/kjfm.2017.38.4.213
Kim, S.-Y., & Park, S.-P. (2014). Suicidal ideation and
risk factors in Korean migraine patients. Journal of
Clinical Neuroscience, 21(10), 1699-1704. https://doi.
org/10.1016/j.jocn.2014.03.016
Klonsky, E. D., & May, A. M. (2015). The Three-Step
Theory (3ST): A New Theory of Suicide Rooted in the
“Ideation-to-Action” Framework. International Journal
of Cognitive Therapy, 8(2), 114-129. https://www2.psych.
ubc.ca/~klonsky/publications/3ST.pdf
Klonsky, E., May, A., Saffer, B. (2016). Suicide, Suicide
Attempts, and Suicidal Ideation. Annual Review of Clinical Psychology, 12(1), 307-330. https://doi.org/10.1146/
annurev-clinpsy-021815-093204

Klonsky, E., Qiu, T., Saffer, B. (2017). Recent advances in
differentiating suicide attempters from suicide ideators.
Current opinion in psychiatry, 30(1), 15-20. https://doi.
org/10.1097/YCO.0000000000000294
Koyama, A., Matsushita, M., Hashimoto, M., Fujise, N.,
Ishikawa, T., Tanaka, H., Hatada, Y., Miyagawa, Y., Hotta,
M., Ikeda, M. (2017). Mental health among younger and
older caregivers of dementia patients. Psychogeriatrics :
the official journal of the Japanese Psychogeriatric Society, 17(2), 108-114. https://doi.org/10.1111/psyg.12200
Lise, F., Schwartz E., Milbrath, V., Santos, B., Feijó, A.,
Garcia, R. (2017). Criança Em Tratamento Conservador
Renal: Experiências Das Cuidadoras Familiares. Texto
& Contexto - Enfermagem, 26(3), e1110016. https://doi.
org/10.1590/0104-07072017001110016
Liu, R. T., & Miller, I. (2014). Life events and suicidal
ideation and behavior: A systematic review. Clinical Psychology Review, 34(3), 181–192. https://doi.org/10.1016/j.
cpr.2014.01.006
Lopes, C., & Faerstein, E. (2001). Confiabilidade do relato
de eventos de vida estressantes em um questionário
autopreenchido: Estudo Pró-Saúde. Revista Brasileira
de Psiquiatria, 23(3), 126-133. https://doi.org/10.1590/
S1516-44462001000300004.
Mari, J., & Williams, P. (1986). A validity study of a psychiatric screening questionnaire (SRQ-20) in primary
care in the city of Sao Paulo. British Journal of Psychiatry,
148(1), 23-26. https://doi.org/10.1192/bjp.148.1.23
Mcfeeters, D., Boyda, D., & O’neill, S. (2015). Patterns
of stressful life events: distinguishing suicide ideators
from suicide attempters. Journal of affective disorders,
175, 192-8. https://doi.org/10.1016/j.jad.2014.12.034
Nock, M., Borges, G., Bromet, E., Alonso, J., Angermeyer,
M., Beautrais, A., Bruffaerts, R., Chiu, W., de Girolamo, G.,
Gluzman, S., de Graaf, R., Gureje, O., Haro, J., Huang, Y.,
Karam, E., Kessler, R., Lepine, J., Levinson, D., Medina-Mora, M., Ono, Y., … Williams, D. (2008). Cross-national
prevalence and risk factors for suicidal ideation, plans
and attempts. British Journal of Psychiatry, 192(2), 98-105.
https://doi.org/10.1192/bjp.bp.107.040113
O’connor, R. C. (2011). Towards an Integrated Motivational-Volitional Model of Suicidal Behaviour. In R.
C. O’Connor, S. Platt, & J. Gordon (Eds.), International
handbook of suicide prevention: Research, policy and
practice (pp. 181–198). Wiley Blackwell. https://doi.
org/10.1002/9781119998556.ch11
O’Connor, R., Wetherall, K., Cleare, S., Eschle, S., Drummond, J., Ferguson, E., O’Connor, D., O’Carroll, R. (2018).
Suicide attempts and non-suicidal self-harm: national
prevalence study of young adults. BJPsych open, 4(3),
142-148. https://doi.org/10.1192/bjo.2018.14
O’Dwyer, S., Moyle, W., Zimmer-Gembeck, M., & De Leo,
D. (2016). Suicidal ideation in family carers of people
with dementia. Aging & mental health, 20(2), 222–30.
https://doi.org/10.1080/13607863.2015.1063109

Auxiliadora Damianne Pereira Vieira da Costa • et al.
Suicidal ideation in mothers of asthmatic children and adolescents in a subspecialty outpatient practice

Park, B., Kim, S., Shin, J., Sanson-Fisher, R., Shin, D.,
Cho, J., Park, J. (2013). Suicidal Ideation and Suicide
Attempts in Anxious or Depressed Family Caregivers
of Patients with Cancer: A Nationwide Survey in Korea.
PLoS ONE, 8(4), e60230. https://doi.org/10.1371/journal.
pone.0060230
Skeen, S., Tomlinson, M., Macedo, A., Croome, N., Sherr,
L. (2014). Mental health of carers of children affected
by HIV attending community-based programmes in
South Africa and Malawi. AIDS care, 26(Suppl 1), S11-20.
https://doi.org/10.1080/09540121.2014.906559
Turecki, G., & Brent, D. A. (2016). Suicide and suicidal
behaviour. Lancet, 387(10024), 1227–39.
Wang, Y., Sareen, J., Afifi, T., Bolton, S., Johnson, E.,
Bolton, J. (2015). A population-based longitudinal study
of recent stressful life events as risk factors for suicidal
behavior in major depressive disorder. Archives of suicide
research : official journal of the International Academy
for Suicide Research, 19(2), 202–17. https://doi.org/10.
1080/13811118.2014.957448
Wetherall, K., Cleare, S., Eschle, S., Ferguson, E., O’Connor, D., O’Carroll, R., & O’Connor, R. From ideation to
action: Differentiating between those who think about
suicide and those who attempt suicide in a national study of young adults. Journal of Affective Disorders, 241(0),
475-483, 2018. https://doi.org/10.1016/j.jad.2018.07.074
World Health Organization. (2014). Preventing suicide:
a global imperative. WHO.
World Health Organization. (2015). Disease and injury
country mortality estimates, 2000–2015 [Data files].
Acessado em 23 set. 2017, de http://www.who.int/
healthinfo/global_burden_disease/estimates/en/
index1.html.

11/11

Mércia Lamenha Medeiros
Doutora em Ciências Aplicada a Pediatria pela Universidade Federal de São Paulo (UNIFESP), em São
Paulo, SP, Brasil; mestre em Saúde da Criança pela
Universidade Federal de Alagoas (UFAL), em Maceió,
AL, Brasil. Professora da Universidade Federal de
Alagoas (UFAL), em Maceió, AL, Brasil.

Camila Oliveira Hansen
Graduada em medicina pela Universidade Federal de
Alagoas (UFAL), em Maceió, AL, Brasil.

Yasmin Cardoso Monteiro Formiga
Graduada em medicina pela Universidade Federal de
Alagoas (UFAL), em Maceió, AL, Brasil.

Claudio Torres de Miranda
Doutor em Psiquiatria e Psicologia Médica pela Universidade Federal de São Paulo (UNIFESP), em São Paulo,
SP, Brasil; mestre em Clinical Epidemiology - University
of Pennsylvania, United States. Professor da Universidade Federal de Alagoas (UFAL), em Maceió, AL, Brasil.

Endereço para correspondência
Auxiliadora Damianne Pereira Vieira da Costa
Universidade Federal de Alagoas
Faculdade de Medicina,

Auxiliadora Damianne Pereira Vieira da
Costa
Doutora em Ciências da Saúde pela Universidade
Federal de Alagoas (UFAL), em Maceió, AL, Brasil;
mestre em Saúde da Criança e do Adolescente pela
Universidade Federal de Pernambuco (UFPE), em
Recife, PE, Brasil. Professora da Universidade Federal
de Alagoas (UFAL), em Maceió, AL, Brasil.

Leticia Marques dos Santos
Doutora em Saúde Pública pela Universidade Federal
da Bahia (UFBA), em Salvador, BA, Brasil; mestre em
Psicologia pela Universidade Federal da Bahia (UFBA),
em Salvador, BA, Brasil. Professora da Universidade
Federal da Bahia (UFBA), em Salvador, BA, Brasil.

Campus A. C. Simões
Av. Lourival Melo Mota, s/n
Cidade Universitária, 57072-970
Maceió, AL, Brasil

Os textos deste artigo foram conferidos pela Poá
Comunicação e submetidos para validação do(s)
autor(es) antes da publicação.