Skip to main content
NCBI home page
PLOS Medicine logoLink to PLOS Medicine
. 2024 Dec 2;21(12):e1004486. doi: 10.1371/journal.pmed.1004486

Conditional cash transfers and mortality in people hospitalised with psychiatric disorders: A cohort study of the Brazilian Bolsa Família Programme

Camila Bonfim 1,*, Flávia Alves 1,2, Érika Fialho 1, John A Naslund 2, Maurício L Barreto 1, Vikram Patel 2, Daiane Borges Machado 1,2
Editor: Charlotte Hanlon3
PMCID: PMC11649113  PMID: 39621791

Abstract

Background

Psychiatric patients experience lower life expectancy compared to the general population. Conditional cash transfer programmes (CCTPs) have shown promise in reducing mortality rates, but their impact on psychiatric patients has been unclear. This study tests the association between being a Brazilian Bolsa Família Programme (BFP) recipient and the risk of mortality among people previously hospitalised with any psychiatric disorders.

Methods and findings

This cohort study utilised Brazilian administrative datasets, linking social and health system data from the 100 Million Brazilian Cohort, a population-representative study. We followed individuals who applied for BFP following a single hospitalisation with a psychiatric disorder between 2008 and 2015. The outcome was mortality and specific causes, defined according to International Classification of Diseases 10th Revision (ICD-10). Cox proportional hazards models estimated the hazard ratio (HR) for overall mortality and competing risks models estimated the HR for specific causes of death, both associated with being a BFP recipient, adjusted for confounders, and weighted with a propensity score. We included 69,901 psychiatric patients aged between 10 and 120, with the majority being male (60.5%), and 26,556 (37.99%) received BFP following hospitalisation. BFP was associated with reduced overall mortality (HR 0.93, 95% CI 0.87,0.98, p 0.018) and mortality due to natural causes (HR 0.89, 95% CI 0.83, 0.96, p < 0.001). Reduction in suicide (HR 0.90, 95% CI 0.68, 1.21, p = 0.514) was observed, although it was not statistically significant. The BFP’s effects on overall mortality were more pronounced in females and younger individuals. In addition, 4% of deaths could have been prevented if BFP had been present (population attributable risk (PAF) = 4%, 95% CI 0.06, 7.10).

Conclusions

BFP appears to reduce mortality rates among psychiatric patients. While not designed to address elevated mortality risk in this population, this study highlights the potential for poverty alleviation programmes to mitigate mortality rates in one of the highest-risk population subgroups.

Camila Bonfim and colleagues investigate the effect of the Brazilian Bolsa Família Programme, a conditional cash transfer scheme, on rates of mortality in people hospitalized with a psychiatric disorder.

Author summary

Why was this study done?

  • People living with psychiatric disorders have a higher risk of mortality compared to the general population.

  • Poverty contributes to these individuals experiencing more risky behaviours and receiving less healthcare.

  • Conditional cash transfer programmes (CCTPs) have shown an association with reduced mortality in the general population; however, there is a lack of studies investigating this among psychiatric patients.

What did the researchers do and find?

  • We performed a population-based cohort study to investigate the association between the Brazilian Bolsa Família Programme (BFP) and the risk of overall mortality and specific causes, such as natural and unnatural causes, as well as suicide among psychiatric patients.

  • We observed that BFP recipients had lower mortality rates when compared to nonrecipients, especially for overall mortality and natural causes of death.

  • In addition, we found that a considerable number of deaths could have been prevented if this benefit had been present.

What do these findings mean?

  • These are the first results to suggest that a broad programme of social assistance, not directed at psychiatric patients in particular, has major benefits in reducing mortality among psychiatric patients after discharge.

  • It has broader significance too, opening the question of whether such assistance could reduce the well-established high mortality rate among all people with psychiatric disorders.

  • The main limitation of this study is that the results are not generalizable to all hospitalised individuals, as we have accessed severe cases that were hospitalised in public services only. However, these services in Brazil cover 75% of the population.

Introduction

Epidemiological studies show that individuals with psychiatric disorders have a shortened life expectancy [1–3]. Compared to the general population, the mortality rate is nearly double among those living with psychiatric disorders [2,4] and 80% higher among mental health service users [5]. This mortality gap is predicted to worsen, as reflected in recent Global Burden Disease study data, demonstrated by the rise in psychiatric disorders from 13th position to the seventh leading cause of disability-adjusted life-years over the past decade [3].

Individuals with psychiatric disorders commonly face multimorbidity [6], reflected by co-occurring chronic medical conditions such as cardiovascular diseases, respiratory illnesses, diabetes, hepatitis, and obesity [7]. Moreover, this patient population often receives poor quality healthcare, including limited access to effective mental health services, a concern that is particularly severe in low- and middle-income countries (LMICs) [8]. These individuals face numerous barriers to accessing health services, such as transportation difficulties, lengthy waiting lists, language and cultural barriers, and experiences of stigma and discrimination [9]. Inadequate care, prevalent in LMICs, is closely linked to poor physical health outcomes and increased mortality [10,11].

Research has also demonstrated variations in mortality rates among individuals with psychiatric disorders. While the relative risks for unnatural causes of death, such as violence, suicide, road injuries, and falls, were elevated among individuals with psychiatric disorders, when compared to natural causes of death, such as cardiovascular and respiratory diseases and cancer, over two-thirds of deaths in this population group may be attributed to natural causes [12]. We highlight that suicide rates are higher for individuals hospitalised with psychiatric disorders when compared to those in primary care settings, reflecting added vulnerabilities among those seeking care in a hospital environment [13,14]. Although suicide is a multifactorial problem, research shows that people with psychiatric disorders experience a higher risk of suicide, compared to the general population. Psychiatric disorders may increase the risk of suicide 10-fold [15]; it is estimated that approximately 62.2% of the global burden of suicide may be associated with psychiatric disorders [16] even though these global estimates rarely include LMICs. The differences in mortality rates may be influenced by multiple factors, including genetic, behavioural, aspects of lifestyles, access to healthcare, psychiatric treatments, and social determinants of health, such as poverty and lack of social support [12].

There is a vicious cycle between poverty and psychiatric disorders, with longitudinal studies showing that conditional cash transfer programmes (CCTPs) in LMICs, in addition to alleviating poverty, may reduce the burden of mental health problems [17–19], suicidal behaviour [20,21], and premature mortality [22,23]. Cash transfer programmes are social policies commonly implemented in LMICs with the main goal of alleviating extreme poverty and poverty-related outcomes such as food insecurity. These programmes can exhibit considerable heterogeneity among countries in terms of transfer value, frequency and duration of the benefit, presence of conditionalities such as health and schooling attendance, targeting and eligibility criteria, as well as implementation systems (government or private) [24,25].

Therefore, cash transfer programmes have been associated with multiple additional benefits, such as improving financial security and family stability and reducing financial strain, which are factors commonly associated with natural causes of death, such as cardiovascular diseases [22] and unnatural causes of death, such as suicide [20] and violence [26]. In addition, CCTPs include conditions, such as improving primary care services and regulating school attendance, which may improve beneficiary health behaviour [27], thereby contributing to reduced mortality rates [23,28]. Despite these promising tendencies, the potential benefits of CCTPs have not yet been evaluated among the vulnerable population of individuals hospitalised with psychiatric disorders. Therefore, the contribution of CCTPs to this population would be particularly relevant, considering that there is an increased risk of death after discharge [29].

Very few studies have explored the potential impact of economic interventions, such as CCTPs, on natural and unnatural causes of mortality among individuals hospitalised with psychiatric disorders. These insights are valuable for supplementing policies to promote the health and longevity of individuals with psychiatric disorders, while advancing broader efforts to address the disproportionately elevated mortality risk affecting this population group, which has been identified as a major priority for global mental health [3,9,30]. We hypothesise that the Bolsa Família Programme (BFP) could reduce the mortality risk among psychiatric patients. The objective of this study was to test the association of participating in a CCTP through Brazil’s national BFP, and the risk of mortality due to overall, natural, and unnatural causes, as well as suicide in those previously hospitalised with any psychiatric disorders.

Methods

Study design, data sources, and dataset linkage

We conducted this evaluation according to Machado and colleagues‘ protocol [31]. This cohort study uses data from the 100 Million Brazilian Cohort, linked to the Hospitalisation (SIH) and Mortality Information Systems (SIM) (2001–2018) [31] (S6 Text). The 100 Million Brazilian Cohort is a dynamic cohort comprised of individuals registered on CadÚnico, which is the primary system used when applying for social assistance in Brazil [32]. CadÚnico is a database for selecting and including families in the various government-supported social programmes, which is linked with BFP database. It enables families in situations of socioeconomic vulnerability to access these benefits. Covering roughly 55% of the total Brazilian population, it comprises individuals facing poverty and extreme poverty [31,32].

Brazil presents one of the world’s largest and most complex public health systems, ranging from primary to tertiary levels, the latter providing hospital care. Access to this system is comprehensive, universal, and free for the entire population. Although a private supplementary system exists, over 70% of the Brazilian population relies on the public health system, particularly the poorest segments of society [33].

SIH encompasses 75% of Brazil’s total admissions in general or specialised hospitals under this national health system [34]. It focuses on serious morbidity requiring specialised attention and monitoring, often involving advanced resources [35]. SIM records all Brazilian deaths through mandatory, high-quality death certificates [36]. Both systems use standardised forms completed by health professionals, including the cause of a hospital admission and death, according to the International Classification of Diseases 10th Revision (ICD-10) [31].

The databases were connected using nondeterministic linkage and a tool developed by a team of experts at the Centre of Data and Knowledge Integration for Health (CIDACS)/Fiocruz, to link administrative data from Brazil [37] (S1 Text). Further information on data governance and the linkage process has been published elsewhere [37,38]. Ethical approval for the study was obtained from the Federal University of Bahia (UFBA—registration number: 1023107). This study is reported as per STROBE guideline (S1 STROBE Checklist).

Participants

Our study population comprised individuals aged 10 and older who had been registered on the cohort baseline at different times and had at least 1 hospitalisation with a psychiatric disorder (defined by code “F”, according to ICD-10 [39], and registered on SIH) between January 1, 2008 and December 31, 2015 when all the data were available.

We identified all individuals aged 10 and older who had been first hospitalised with psychiatric disorders in the study period. We analysed all the hospital discharge records of patients with primary or secondary diagnoses classified as a psychiatric disorder over the same period. We included secondary diagnoses, considering psychiatric disorders that were also common among people admitted with other primary diagnoses [40]; however, this represented only 5.07% of the total admissions. Cutoff at the age of 10 is justified since suicide, one of the outcomes investigated in this study, is an extremely rare occurrence under this age [20]. Moreover, considering that individuals with a history of multiple hospitalisations due to psychiatric disorders are at a higher risk of mortality [41], we only included individuals with a single hospitalisation in our analysis in order to enhance comparability among participants. In our data, approximately 3% of individuals were readmitted for psychiatric hospitalisation. Subsequently, we selected all individuals registered on CadÚnico following their first hospitalisation in this period to avoid selection bias. Beneficiaries might be less frequently hospitalised than nonbeneficiaries, considering the association between reduction of poverty and improved health conditions [22,23,28], causing an imbalance with the comparison group. A description of the individuals excluded from the analysis can be found in the Supporting information (S2 Text). Finally, we excluded individuals who had anomalous information that could reflect linkage errors (Fig 1) (S2 Text).

Fig 1. Flowchart of the study population.

Fig 1

Open in a new tab

BFP, Bolsa Família Programme.

For the BFP beneficiary subset, these individuals were followed from the time they registered to receive the BFP benefit, and their follow-up ended either due to the individual’s death by any cause, or on December 31, 2015 (reflecting the end of the follow-up period). For the nonbeneficiary subset (i.e., individuals who were not registered for the BFP benefit), the follow-up started when these individuals were registered on CadÚnico. The follow-up ended for nonbeneficiaries either due to their death by any cause, or the end of the follow-up period on December 31, 2015.

Exposure and covariates

BFP is a conditional cash transfer that targets the poorest families [42], with conditions related to healthcare for children and pregnant women, and children’s education. Eligibility is based on CadÚnico system registrations and having a monthly household income of less than BRL 70.00 (USD 17.00), or BRL 140.00 (USD 34.00) for households with a child, adolescent, or pregnant woman. Benefits range from BRL 41.00 (USD 10.00) to a maximum of BRL 300.00 (USD 75.00) per person, utilising 2015 values, adjusted for inflation [32]. Conditions for receiving the BFP benefit include school attendance, vaccinations, and monitoring young people’s growth. Pregnant or breastfeeding women must also follow a health and nutrition protocol [32]. We assumed that individuals receiving BFP were considered exposed throughout the study, covering benefit receipt and adherence to conditions for receiving BFP, impacting their quality of life and health outcomes [32]. While beneficiaries can stop receiving BFP due to noncompliance or increased income, 95% remained in the programme until the last year of the follow-up period. The study compared the group receiving BFP benefits with those who did not, defining beneficiaries as individuals receiving benefits during the follow-up. In addition, only those registered for BFP posthospitalisation with psychiatric disorders were included, focusing on the effect of BFP registration following this type of hospitalisation and reducing potential biases related to better socioeconomic conditions that could improve mental health and reduce hospitalisation rates [29].

Covariates were defined based on a literature review and selected from those registered on databases [14–20]. This included sex, age, race, education level, location of residence (rural or urban), household characteristics (presence of a water supply, waste, sanitation, and construction materials), crowding (number of people in the house by the number of rooms), isolation (people who live alone, or with someone else), Brazilian regions of residence, year of hospitalisation, year of CadÚnico registration, and length of hospitalisation. The latter was calculated in terms of days and categorised based on tertiles, considering the first admission record. Most of these covariates were registered at the cohort baseline, except for hospitalisation-related variables, which were registered at the time of a hospital admission.

Outcomes

This study included overall, natural, and unnatural causes of death, as well as suicide, recorded on SIM between 2008 and 2015. Natural causes were defined by all causes of mortality, according to ICD-10 [39], except for external causes, using violence and suicide codes. Unnatural causes included external causes, such as accidents, falls, suicide, or violence (V010 –Y98) [39], while the overall causes of mortality included all general medical conditions. Furthermore, suicide was analysed separately (codes X60 to X84) [35], given the strong association between suicide and psychiatric disorders [15].

Statistical analyses

Descriptive analyses were carried out to characterise the demographic, socioeconomic, and hospitalisation covariates, comparing BFP beneficiaries to nonbeneficiaries (Table 1). Standardised mean differences (SMDs) were computed before and after inverse probability of treatment weighting (IPTW) estimation, and absolute values higher than 0.1 indicated imbalance in the distribution of covariates between treated and untreated groups (Table 1) [20]. Mortality rates were estimated using person-years as the denominator for each individual’s observation period (S1 Table).

Table 1. Description of BFP nonbeneficiaries and beneficiaries before and after IPTW, 2008 to 2015.

Characteristics Before IPTW
(N = 69,901)		 After IPTW
(N = 57,905)
BFP
N = 26,556 (37.99)		 Non-BFP
N = 43,345 (62.01)		 Diff. 1
(BFP – non-BFP)		 BFP
N = 20,399 (35.22)		 Non-BFP
N = 37,506
(64.78)		 Diff.1
(BFP – non-BFP)
% % % %
Sex
Male 56.92 62.70 0.21 0.52 0.53 0.03
Female 43.08 37.30 0.48 0.47
Age group (years old)
10-24 12.11 7.36 −0.39 0.13 0.12 −0.31
25-59 84.23 80.49 0.83 0.84
>60 3.66 12.14 0.04 0.04
Education Level (years of education)
Never studied 10.43 15.74 0.10 0.10 0.11 −0.01
Preschool 0.76 1.41 0.01 0.07
Primary school or less (<=5 years) 34.87 35.86 0.35 0.34
Junior high school (6-10 years) 27.37 20.57 0.28 0.27
High school (10-12 years) 24.86 23.91 0.25 0.25
College/university (>=13 years) 1.42 2.23 0.01 0.01
Missing data 0.29 0.28 NA NA
Race
White 43.06 47.91 0.08 0.47 0.47 0.01
Black 8.88 6.84 0.09 0.09
Asian 0.41 0.51 0.01 0.01
Mixed race/Brown 41.50 39.16 0.44 0.44
Indigenous 0.23 0.10 0.01 0.01
Missing data 5.93 5.49 NA NA
Location of residence
Rural 9.06 8.17 0.05 0.91 0.92 0.01
Urban 81.18 90.95 0.09 0.08
Missing data 9.76 0.88 NA NA
Brazilian regions
Southeast 49.09 43.83 −0.10 0.49 0.49 −0.01
Northeast 14.57 13.70 0.15 0.14
Central-West 6.89 10.27 0.07 0.07
South 27.28 29.74 0.27 0.28
North 2.17 2.45 0.02 0.02
Household characteristics
Water supply
Public network (running water) 71.57 83.77 0.13 0.87 0.87 0.03
Well, natural sources, or other 12.75 9.60 0.13 0.13
Missing data 15.68 6.63 NA NA
Waste
Public collection system 77.41 88.10 0.09 0.93 0.93 0.02
Burned, buried, outdoor disposal, or other 6.91 5.28 0.07 0.07
Missing data 15.68 6.63 NA NA
Sanitation
Public network 54.63 61.74 0.04 0.67 0.68 0.02
Septic tank 9.94 11.91 0.12 0.12
Homemade septic tank 13.46 16.25 0.16 0.16
Ditch or other 3.60 1.94 0.05 0.04
Missing data 18.36 8.16 NA NA
Construction materials
Bricks/cement 70.85 80.78 0.04 0.85 0.85 0.01
Wood, other plant materials, or other 13.47 12.59 0.15 0.15
Missing data 15.68 6.63 NA NA
Isolation
Lives with someone else 66.58 73.16 −0.16 0.78 0.76 −0.01
Lives alone 33.42 26.84 0.22 0.24
Year of registration on CadÚnico
2008 2.75 0.62 −0.55 0.03 0.03 −0.05
2009 7.50 1.72 0.09 0.09
2010 11.59 3.74 0.15 0.14
2011 11.50 10.63 0.13 0.13
2012 19.58 25.66 0.20 0.21
2013 17.60 17.53 0.13 0.13
2014 17.20 20.79 0.16 0.16
2015 12.28 19.31 0.11 0.11
Open in a new tab

BFP, Bolsa Família Programme; IPTW, inverse probability of treatment weighting.

1The difference in proportions of each category between BFP beneficiaries and nonbeneficiaries (BFP beneficiary proportion minus nonbeneficiary proportion).

In line with the study protocol [31], and other studies using the 100 Million Brazilian Cohort [20,23,27,28], the propensity score (PS)–based method was applied to promote comparability between treated and untreated groups. We estimated the PS using a multivariable logistic regression, adjusted for socioeconomic covariates associated with BFP [43] (S2A Table). The following covariates were considered when estimating the PS: sex, age, education level, race, location of residence (rural or urban), household characteristics (presence of a water supply, waste, sanitation, and construction materials), crowding (number of people in the house by the number of rooms), isolation (people who live alone or with someone else), Brazilian regions of residence, and year of CadÚnico registration. The last item was included considering variations in the CadÚnico register over time. Additionally, we assessed the common support graph, and we compared the range of propensity scores among BFP and non-BFP groups (S3 Fig and S2B Table).

Then, we estimated the IPTW [44,45]. IPTW uses the PS to balance differences among covariates in the treated and untreated groups using weights [45]. Individuals who received the BFP were given weights equal to the inverse of their propensity scores (1/PS), whereas those who did not receive the treatment were given weights equal to the inverse of 1 minus their propensity scores [PS/(1 –PS)] [44,45]. Following this, we identified problems with extreme weights in the estimation. To correct this problem, truncated weights were used with specified thresholds based on weight distribution for the 99th percentile [27]. To estimate the association between BFP and mortality rates, we used an average treatment effect on the treated (ATT) estimator, according to the recommendation of Ali and colleagues [45] and other studies using 100 million Brazilian cohort [20,22,23,27,28], and fitted as a survival analysis model using Cox proportional hazard regression [46]. Hazard ratio (HR) was estimated for the overall mortality with 95% confidence intervals (CIs), and we introduced IPTW weights as a weight function in Stata, in addition to the year and length of the hospitalisation, to control for mortality-related risk factors. We also estimated a competing risks model using a Fine Gray model, which directly models the subdistribution hazard for each cause of death, accounting for the presence of competing risks [47]. In each competing risk model, we considered each cause of death as the failure, the other causes as competing risks, and the individuals who were alive were censored (Table 2). Participants with missing covariates data were excluded from the final model.

Table 2. Association of BFP participation with overall, natural, unnatural, and suicide mortalities, 2008–2015.

Confounder adjustment Overall population Cox model Competing risks model
Overall mortality Natural causes Unnatural causes Suicide
HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)
Cox adjusted with IPTW1 (final model)
Non-BFP 57,905 1.00 1.00 1.00 1.00
BFP 0.93
(0.87, 0.98)		 0.89
(0.83, 0.96)		 1.14
(0.97, 1.33)		 0.90
(0.68, 1.21)
p-value 0.018 0.001 0.112 0.514
Sensitivity analysis
Cox adjusted with SIPTW2
Non-BFP 57,905 1.00 1.00 1.00 1.00
BFP 0.91
(0.86, 0.97)		 0.87
(0.82, 0.93)		 1.21
(1.04, 1.40)		 0.94
(0.71, 1.23)
p-value 0.002 <0.001 0.012 0.642
Cox adjusted with kernel matching3
Non-BFP 57,475 1.00 1.00 1.00 1.00
BFP 0.77
(0.72, 0.81)		 0.74
(0.69, 0.78)		 1.02
(0.89, 1.18)		 0.91
(0.69, 1.20)
p-value <0.001 <0.001 0.795 0.502
Open in a new tab

BFP, Bolsa Família Programme; CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting; SIPTW, stabilised inverse propensity scores.

1HR estimated with IPTW given sex, age, race, education level, household characteristics (water supply, waste, sanitation, and construction materials), living alone, crowding, Brazilian region, location of residence, length and year of hospitalisation, and year of CadÚnico registration.

2HR estimated with SIPTW given sex, age, race, education level, household characteristics (water supply, waste, sanitation, and construction materials), living alone, crowding, Brazilian region, location of residence, length and year of hospitalisation, and year of CadÚnico registration.

3HR estimated with K matching given sex, age, race, education level, household characteristics (water supply, waste, sanitation, and construction materials), living alone, crowding, Brazilian region, location of residence, length and year of hospitalisation, and year of CadÚnico registration.

Furthermore, to better identify the potential effects of a public policy as BFP on reducing mortality, we calculated an equivalent population attributable risk (PAF) to estimate the proportion of death that BFP theoretically could prevent [48]. We used the formula PAF = [P(HR − 1)] / [P(HR − 1) + 1] [48], where “P” was the incidence of the BFP in the population and “HR” is the hazard ratio estimated for the association between BFP and overall mortality. We estimated unadjusted and adjusted PAF using the punafcc package in Stata, which uses Cox regression and a 95% CI according to previous publication [49]. For the adjusted model, we included the same covariates as in the final model.

We used multiple approaches for the sensitivity analysis. First, we employed the stabilised inverse propensity scores (SIPTW) approach, a modification of IPTW that stabilises weights to enhance numerical stability during estimation processes. We estimated SIPTW weights for nonbeneficiaries using the formula (1 − Pt) / (1 − Psmul), and for beneficiaries using the formula Pt / Psmul, where “Pt” represents the marginal probability of treatment in the population, and “Psmul” denotes the propensity score obtained from multivariable logistic regression adjusted for covariates. We applied the same truncation process to extreme weights and repeated the final model. Second, we employed the Kernel matching (KM) approach, which establishes a weighting scheme for all untreated units, assigning greater weights to units closer to those treated units to which they are matched. This method matches pairs based on weights estimated from propensity scores. Beneficiaries and nonbeneficiaries were matched by year of registration in the cohort, and the PS matched with kernel weights for the same covariates used in the final model generating the ATT (S3A–S3D Table). Third, to assess the relevance of IPTW in obtaining unbiased BFP estimates, we conducted crude and adjusted Cox regressions without IPTW weighting (S4 Table). Fourth, the final model was repeated, using missing values as a category in the analysis (S5 Table). Fifth, subgroup analyses were performed by sex and age using IPTW weights, calculated for each subcategory (Table 3). Finally, we repeated these analyses using Poisson models (S6 Table). Stata version 15.0 was used for data analysis.

Table 3. Association of BFP participation with overall, natural, unnatural, and suicide mortalities by subgroups, 2008–2015.

Subgroups Cox model Competing risks model
Overall mortality Natural causes Unnatural causes Suicide
HR1 (95% CI) HR1 (95% CI) HR1 (95% CI) HR1 (95% CI)
Sex
Male 1.05 (0.98, 1.13) 1.02 (0.94, 1.11) 1.24 (1.04, 1.48) 0.86 (0.61, 1.21)
N 33,786 33,786 33,786 33,786
p-value 0.169 0.650 0.016 0.391
Female 0.75 (0.67, 0.85) 0.73 (0.64, 0.83) 0.91 (0.63, 1.35) 1.02 (0.56, 1.82)
N 24,119 24,119 24,119 24,119
p-value <0.001 <0.001 0.666 0.928
Age groups (years old)
10- 24 0.79 (0.57, 1.10) 0.56 (0.33, 0.95) 1.21 (0.79, 1.85) 0.84 (0.33, 2.08)
N 5,589 5,589 5,589 5,589
p-value 0.174 0.033 0.387 0.708
25-59 0.95 (0.89, 1.01) 0.91 (0.84, 0.98) 1.16 (0.98, 1.39) 1.01 (0.73, 1.37)
N 47,261 47,261 47,261 47,261
p-value 0.131 0.015 0.080 0.992
60 or older 0.97 (0.83, 1.11) 0.98 (0.84, 1.14) 0.90 (0.43, 1.89) 0.18 (0.02, 1.29)
N 5,135 5,135 5,135 5,135
p-value 0.554 0.839 0.794 0.087
Open in a new tab

BFP, Bolsa Família Programme; CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting.

1HR estimated with IPTW given sex, age, race, education level, household characteristics (water supply, waste, sanitation, and construction materials), living alone, crowding, Brazilian region, location of residence, length and year of hospitalisation, and year of CadÚnico registration.

Results

We identified 369,959 individuals hospitalised with psychiatric disorders on the SIH database between 2008 and 2015. When these individuals were linked to the CadÚnico database after their first hospitalisation in this period, we identified 71,978 individuals who entered the 100 Million Brazilian Cohort. We then excluded 910 participants (<1%) who did not meet the inclusion criteria. Finally, we excluded 1,167 participants (<1%), due to inconsistent data. The study sample only included 69,901 individuals who had applied for BFP following a single hospitalisation with any psychiatric disorders and met the eligibility criteria (Fig 1).

Twenty-six thousand, five hundred and fifty-six (26,556) (37.99%) individuals who had been hospitalised with psychiatric disorders received BFP. The average time of receiving BFP after discharge was 2.86 years (SD = 1.85). Before IPWT weighting, there were differences in sociodemographic characteristics between beneficiaries and nonbeneficiaries (Table 1). BFP beneficiaries, compared to nonbeneficiaries, were more likely to be aged between 25 and 59 (84.23% versus 80.49%), non-white (56.94% versus 52.09%), live in the Southeast region of Brazil (49.09% versus 43.83%), and reside in more crowded households (0.79 versus 0.57), respectively. In contrast, when comparing nonbeneficiaries to beneficiaries, they were more likely to be male (62.70% versus 56.92%), live in urban areas (90.95% versus 81.18%), and not live alone (73.16% versus 66.58%), respectively (Table 1). There was a progressive increase in CadÚnico registration over the period studied, while psychiatric hospitalisations decreased for both groups, and approximately half (52.81%) of all hospitalisations were for more than 2 weeks (S6 Text). After IPTW weighting, the beneficiary and nonbeneficiary groups had similar sociodemographic characteristics. The difference in proportions of each category between BFP beneficiaries and nonbeneficiaries was lower than 10% (Table 1).

Over the period, 8,118 individuals died for overall causes and most of them were nonbeneficiaries (63.5%) (S1 Table). Mortality rates per 100,000 person-years for natural causes (non-BFP 4,077.98 95% CI 3,958.81, 4,200.74 versus BFP 2,783.79 95% CI: 2,672.95, 2,899.23) were higher when compared to unnatural causes (non-BFP 738.65 95% CI 688.93, 791.96 versus BFP 757.26 95% CI 700.51, 818.61). For suicide, the rates were 176.49 95% CI 153.04, 203.53 for non-BFP and 144.75 95% CI 121.13, 172.98 for BFP recipients (S1 Table).

The subgroup analysis observed lower mortality rates among females for all causes, particularly in beneficiaries. Among individuals aged between 10 and 24, mortality rates were lower among hospitalised nonbeneficiaries, compared to beneficiaries for natural causes (416.28 versus 755.17 per 100,000 person-years) and overall mortality (1,185.49 versus 1,546.89 per 100,000 person-years), respectively. For those aged 60 or older, mortality rates due to unnatural causes were lower for hospitalised nonbeneficiaries compared to beneficiaries (647.74 versus 665.08 per 100,000 person-years). Suicide rates were lower across different age groups, especially among hospitalised nonbeneficiaries aged between 10 and 24 (158.34 per 100,000 person-years) and beneficiaries aged 60, or older (39.12 per 100,000 person-years; S1 Table).

BFP was associated with a reduction in overall mortality (HR 0.93; 95% CI 0.87, 0.98; p = 0.018) and mortality due to natural causes (HR 0.89; 95% CI 0.83, 0.96; p = <0.001; Table 2). The associations between BFP beneficiares and mortality rates were similarly observed in sensitivity analyses and after including missing covariate values as missing categories. The effects of BFP appeared strongest among females and younger individuals (Table 3). When clustering by household level was accounted for, the estimate of the effect of BFP on overall mortality was null (IRR 0.98; 95% CI 0.92, 1.04; p = 0.478) (S4 Text and S7A and S7B Table).

The unadjusted PAF analysis showed that 18% (95% CI 16, 20) of deaths could potentially be prevented if BFP had been present, while the adjusted analysis showed a reduction of 4% (95% CI 0.06, 7.10).

Discussion

To our knowledge, this is the first study to estimate the association of a CCTP with mortality in individuals hospitalised with psychiatric disorders registered on the 100 Million Brazilian Cohort. BFP was associated with a 7% reduction in the overall mortality rate among beneficiaries, primarily driven by lower mortality due to natural causes. For mortality due to unnatural causes and suicide, in particular, results were consistent with an effect, but they were not statistically significant. Furthermore, the effects of BFP were strongest among females and younger individuals. Furthermore, 4% of these deaths could be prevented if BFP were present.

Previous studies have demonstrated how CCTPs contribute to breaking the bidirectional cycle of poverty and psychiatric disorders in the general population [50]. However, less is known about how these programmes could break this cycle among those already affected by severe psychiatric disorders. Therefore, this study contributes to understanding the role of a CCTP in increasing the chance of survival in a population subgroup that disproportionately faces financial hardship and complex mental and physical health care needs. These findings illustrate the potential of BFP in advancing tertiary prevention within this highly vulnerable patient population.

This study highlights the potential impact of BFP in reducing mortality rates among those hospitalised with psychiatric disorders, attesting to the importance of implementing social protection programmes to cover vulnerable population subgroups. Although BFP was not specifically designed to address health and social concerns affecting individuals with psychiatric disorders, the programme appears to have important downstream benefits on their health and on reducing mortality rates. The BFP focus on various aspects of health, coupled with poverty alleviation, may have assisted in facilitating access to primary care services and routine checkups for individuals with psychiatric disorders, thereby resulting in improvements in their health behaviour and, ultimately, reducing natural causes of death [22]. The strong association between receiving BFP and a reduction in natural causes of death suggests a synergistic effect between BFP and the Family Health Strategy [51,52], whereby individuals supported by BFP experience direct health benefits through preventive measures and management of comorbidities, such as hypertension, diabetes, and other chronic diseases. Over time, through increased access to basic and preventive health services, BFP may have had a positive influence on alleviating the elevated mortality rates affecting individuals with psychiatric disorders.

In Brazil, the public healthcare system for individuals with psychiatric disorders is provided through the Psychosocial Care Network, which includes services such as primary care, community-based mental health centers (called psychosocial care centers, or CAPS, in Brazil), emergency and urgent care network, residential services, and general hospitals [53]. The Brazilian health reform has increasingly prioritized mental health care in the CAPS, progressively reducing psychiatric hospital beds, which were the primary option before the reform [54]. However, there is a shortage of these services, with better access found in larger cities, despite some limited expansion into small municipalities [55]. In cases where these services are not available, general hospitals may be the only option for individuals experiencing a psychiatric emergency, such as those living with severe mental disorders. Although the provision of mental health care in general hospitals can bring benefits such as stigma reduction, increased service access, improved physical health care, and the possibility of multidisciplinary team care, the number of psychiatric beds available in these hospitals remains limited [54]. Therefore, these findings may also reflect weaknesses in access to urgent mental health services through the psychosocial care network.

A stronger effect of BFP on reducing mortality due to natural causes and overall mortality rates was observed for women and the younger population. This is consistent with previous research, which found that BFP reduces overall mortality, especially among women and younger populations [22,56]. Although most of the sample was composed of men, BFP had a greater impact on reducing mortality rates among women. This can potentially be explained, given that BFP emphasises women’s important role, and the benefit is provided to women who are, for the most part, heads of households [57]. Previous studies have demonstrated that CCTPs may encourage female empowerment and facilitate decision-making power among women in relation to managing household budgets [57].

Although these findings showed that BFP was associated with an overall reduction in overall and natural cause mortality rates in individuals hospitalised with psychiatric disorders, the association between receiving BFP benefits and a reduction in mortality due to unnatural causes or suicide did not emerge as being statistically significant. There are some possible reasons for these unexpected findings. First, the sample size and lower frequency of deaths due to unnatural causes and suicide could have affected the power of the study, thereby reducing the probability of detecting statistically significant differences between BFP beneficiaries and nonbeneficiaries [58]. Second, unnatural causes of death may be influenced by multifactorial aspects, such as behaviour and the social environment [59], which would require a longer follow-up period to fully observe any benefits that could be attributed to BFP within this specific subgroup.

This study has a number of limitations. While death certificates are mandatory in Brazil, and SIM is recognised for its high-quality standards [36], underreporting is always a possibility. Psychiatric disorders require a comprehensive clinical response, with services delivered by mental health professionals. Although the diagnosis was based on ICD-10, one of the leading classifications for mental health diagnoses, we lacked the means to validate the accuracy of this diagnosis [60]. In addition, the datasets did not provide information on the severity of the psychiatric disorders. However, given that psychiatric hospitalisation is typically reserved for severe cases when outpatient resources have been exhausted [61], it was hypothesised that most cases in this study involved severe mental disorders requiring hospital support. Consequently, these findings cannot be generalised for patients with mild mental disorders who receive care in primary and secondary outpatient services.

The data used in this study are not generalisable to all hospitalised individuals since the system only covers public service information and may be influenced by service availability. Nonetheless, as recorded on CadÚnico, this study population has limited access to private health insurance and cannot afford private healthcare services. Despite this limitation, the system is considered suitable for conducting epidemiological research, since it captures approximately 75% of hospitalisations in Brazil [34]. Also, there are possible issues with missing information and diagnosis misclassification [34]. However, improvements have been made in the quality of SIH records in recent years [34], thereby mitigating this concern.

This study used an administrative database that was not designed for research purposes. Thus, there were a number of issues with missing values, mainly in variables that are not mandatory in the system. However, main variables, such as cause of mortality, diagnosis, individual information (e.g., sex and age), and access to benefits, were mostly complete. Furthermore, covariates were only measured at the cohort baseline and were not updated at a later date. Moreover, the linkage process may generate a bias due to challenges such as computational complexity and the absence of a unique number that may identify health and social systems. However, sensitivity and specificity exceeding 94% was achieved in the linkage validation process, and these errors are likely to be nondifferential (S1 Text). An additional limitation is the potential bias related to unmeasured confounding, especially socioeconomic and behavioural factors that were not available in the routinely collected datasets. Finally, data cannot be generalised for the entire Brazilian population and only reflects the poorest segments since the database only includes those seeking social benefits. Compared to the Brazilian population, this cohort overrepresents young people and women due to the BFP target population [32]. Furthermore, the profile of individuals hospitalised for psychiatric reasons in our study also differed from other studies [54]. This might have happened because in our study, the selection criteria focused exclusively on individuals who registered at CadÚnico after hospitalisation. This criterion was intentionally chosen to reduce potential biases related to better socioeconomic conditions that could improve mental health and reduce hospitalisation rates. Consequently, this may have resulted in a cohort that differs from those in other studies where different selection criteria were employed.

This study is highlighted as the first, to our knowledge, to examine the impact of BFP on mortality rates among individuals hospitalised with psychiatric disorders. Using a robust analysis such as competing risk models, this study stands out by showing how the BFP reduces the overall risk of death, and its impact on specific cause of death, considering the competing risks of death from other causes. These findings reveal a noteworthy effect, indicating that receiving financial assistance intended for poverty alleviation can potentially reduce the mortality risk in this vulnerable population subgroup. These results underscore the importance of considering intersectoral strategies for tertiary prevention posthospitalisation of mental health patients. Collaborative initiatives with BFP not only contribute to financial stability but possibly also address institutional barriers, thereby playing a pivotal role in shaping mortality outcomes among individuals hospitalised with psychiatric disorders. The large size of the cohort allowed an evaluation of mortality due to specific causes among BFP nonbeneficiaries and beneficiaries in this specific population, hospitalised with psychiatric disorders. The large size of the cohort also made it possible to explore the effects of BFP on less common health outcomes and the effect variation among subgroups.

This study also contributes to current knowledge on the role of a large economic intervention in alleviating the elevated mortality risk among people hospitalised with severe psychiatric disorders in a LMIC. BFP has primarily reduced overall and natural causes of mortality, especially for women and young people who are commonly the target of social policies due to several poverty-related vulnerabilities. Importantly, BFP has provided economic support and demonstrated the potential to act as a tertiary prevention intervention. Thus, the results in this study have important practice and policy implications to advance efforts for early mortality prevention in mental health care settings and provide support for this vulnerable population group facing the challenges of serious psychiatric disorders.

Supporting information

S1 Strobe Checklist. STROBE Statement—Checklist of items that should be included in reports of observational studies.

(DOCX)

pmed.1004486.s001.docx (35KB, docx)
S1 Text. Accuracy analysis of the linkage between CadÚnico and the mortality information system in a randomised sample of 10,000 record pairs.

(DOCX)

pmed.1004486.s002.docx (16.4KB, docx)
S2 Text. Detailed information from eligible study population.

(DOCX)

pmed.1004486.s003.docx (15.9KB, docx)
S3 Text. Propensity score: Definition, estimation, summary, and support graphs.

(DOCX)

pmed.1004486.s004.docx (14.8KB, docx)
S4 Text. Intraclass correlation coefficient estimation.

(DOCX)

pmed.1004486.s005.docx (14.4KB, docx)
S5 Text. Description of individuals excluded from the analysis.

(DOCX)

pmed.1004486.s006.docx (13.5KB, docx)
S6 Text. Summary of the dataset’s description.

(DOCX)

pmed.1004486.s007.docx (16.9KB, docx)
S1 Fig. ROC curve of the 100 million Brazilian Cohort and SIH (2001–2018) linkage.

Source: Developed by the CIDACS Data Production Center.

(TIF)

pmed.1004486.s008.tif (639.5KB, tif)
S2 Fig. ROC curve of the 100 million Brazilian Cohort and SIM (2000–2015) linkage.

Source: Developed by the CIDACS Data Production Center.

(TIF)

pmed.1004486.s009.tif (865.6KB, tif)
S3 Fig. Distribution of the propensity score in the sample, 2008–2015.

(TIF)

pmed.1004486.s010.tif (575.3KB, tif)
S1 Table. Mortality rates overall and by subgroups through receipt of the BFP, 2008–2015.

(DOCX)

pmed.1004486.s011.docx (19.6KB, docx)
S2 Table. (A) Logistic regression to estimate propensity scores for receiving Bolsa Familia according to covariables, N = 57,905 (B) Propensity score description in accordance with the confounding covariates observed, Brazil, 2008 to 2015, N = 57,905.

(DOCX)

pmed.1004486.s012.docx (18.1KB, docx)
S3 Table

(A) ATT of overall mortality for BFP receipt between 2008 and 2015 using KM. (B) ATT of natural causes of death for BFP receipt between 2008 and 2015 using KM. (C) ATT of unnatural causes of death for BFP receipt between 2008 and 2015 using KM. (D) ATT of suicide for BFP receipt between 2008 and 2015 using KM.

(DOCX)

pmed.1004486.s013.docx (15.2KB, docx)
S4 Table. Crude and adjusted association of BFP participation with overall, natural, unnatural, and suicide mortalities, 2008–2015.

(DOCX)

pmed.1004486.s014.docx (15.9KB, docx)
S5 Table. Association of BFP participation with overall, natural, unnatural, and suicide mortalities accounting for missing data, 2008–2015.

(DOCX)

pmed.1004486.s015.docx (21.4KB, docx)
S6 Table. Incidence rate ratio of BFP participation with overall, natural, unnatural, and suicide mortalities, 2008–2015.

(DOCX)

pmed.1004486.s016.docx (15.3KB, docx)
S7 Table

(A) Intraclass correlation estimation for the household level. (B) Association of BFP participation with overall mortality considering household level, 2008–2015.

(DOCX)

pmed.1004486.s017.docx (15.4KB, docx)
S8 Table. Description of individuals excluded from the analysis following definition of BFP exposition, 2008–2015.

(DOCX)

pmed.1004486.s018.docx (22.6KB, docx)
S9 Table. Description of year and length of hospitalisation overall and by BFP participation, 2008–2015.

(DOCX)

pmed.1004486.s019.docx (15.3KB, docx)

Acknowledgments

We thank the data production team and all CIDACS/FIOCRUZ collaborators for their work on building the 100 Million Brazilian Cohort. We also thank Kosuke Imai for his invaluable assistance and expertise in providing statistical support for this research.

The content is solely the authors’ responsibility and does not necessarily represent the official views of the National Institutes of Health.

Abbreviations

ATT

average treatment effect on the treated

BFP

Bolsa Família Programme

CCTP

conditional cash transfer programme

CI

confidence interval

HR

hazard ratio

ICD-10

International Classification of Diseases 10th Revision

IPTW

inverse probability of treatment weighting

KM

Kernel matching

LMIC

low- and middle-income country

PAF

population attributable risk

PS

propensity score

SIPTW

stabilised inverse propensity scores

SMD

standardised mean difference

Data Availability

The code used in the analysis is available from Github [https://github.com/profacamilabonfim/Codes-from-the-paper.git] and archived in Zenodo [https://zenodo.org/records/13750552]. The data analyzed in this study is hosted by the Centre of Data and Knowledge Integration for Health (CIDACS). Full access to the data is restricted due to its sensitive nature and the exclusive licensing agreement for its use in this study. The privacy regulations set by the Brazilian Ethics Committee prohibit the public availability of this data. However, upon reasonable request, and provided that all ethical and legal requirements are met, the institutional data curation team can make the data available. Further information can be obtained by emailing cidacs.curadoria@fiocruz.br. Study Protocol available on: https://pubmed.ncbi.nlm.nih.gov/36201469/.

Funding Statement

This study was supported by funding from the National Institute of Mental Health (Grant number: 5R01MH128911, awarded to DBM). The funder had no role in the design of the review; collection, analysis, or interpretation of data; writing of the manuscript; or decision to submit the manuscript for publication.

References

Decision Letter 0

Syba Sunny

3 Apr 2024

Dear Dr Bonfim,

Thank you for submitting your manuscript entitled "Do conditional cash transfers reduce mortality in people hospitalised for psychiatric disorders? A quasi-experimental analysis of the Brazilian Bolsa Família Programme" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by 5th April. If you need extra time, please do let me know.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Syba Sunny, MBBS, MRes, FRCPath

Associate Editor

PLOS Medicine

ssunny@plos.org

Decision Letter 1

Syba Sunny

11 Jun 2024

Dear Dr. Bonfim,

Many thanks for submitting your manuscript "Do conditional cash transfers reduce mortality in people hospitalised for psychiatric disorders? A quasi-experimental analysis of the Brazilian Bolsa Família Programme" (PMEDICINE-D-24-01060R1) for consideration at PLOS Medicine.

Your paper has now been reviewed by 3 subject reviewers and a statistician; it was also discussed with an academic editor with relevant expertise and the wider editorial team. The comments are included below and can also be accessed here:

[LINK]

As you will see, the reviewers found your paper interesting. However, there were requests for further information and clarification of various points. In addition, the statistical reviewer (Reviewer 1) raised several important concerns about the study’s design and analysis. Nevertheless, we remain interested in your work. After discussing the paper with the editorial team, I’m pleased to invite you to revise the paper in response to the reviewers’ comments and our editorial requests (below). We plan to send the revised paper to some or all of the original reviewers*, and of course we cannot provide any guarantees at this stage regarding publication.

When you upload your revision, please include a point-by-point response that addresses all of the reviewer and editorial points, indicating the changes made in the manuscript and either an excerpt of the revised text or the location (e.g. page and line number) where each change can be found. Please submit a clean version of the paper as the main article file and a version with changes marked should as a marked-up manuscript. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper.

We ask that you submit your revision by July 2nd, 2024. However, if this deadline is not feasible, please contact me by email (ssunny@plos.org), and we can discuss a suitable alternative.

Please don’t hesitate to contact me directly with any questions (ssunny@plos.org).

Kind regards,

Pippa - Senior Editor, PLOS Medicine - on behalf of:

Syba Sunny MBBS, MRes, FRCPath

Associate Editor

PLOS Medicine

ssunny@plos.org

*Please note: If your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

-----------------------------------------------------------

EDITORIAL COMMENTS:

We thank you for submitting your valuable work to PLOS Medicine. We note your unique data set, but, equally, agree with all the points raised by the reviewers. As such, we require that you address all the reviewer and editorial comments in full, before we can consider your paper further.

1) Data Availability:

Please revise your data availability section. At present, there is missing information – please replace the parts with XXX with the relevant information.

If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (this cannot be a study author). Further information is available from

http://journals.plos.org/plosmedicine/s/data-availability

2) Reporting guidance:

Please report your study according to the relevant guidance which can be found here https://www.equator-network.org/reporting-guidelines/

3) Statistical reporting:

(Note that not all of the following instructions will apply to your work, but please review all points and action as appropriate.)

Please quantify the main results with 95% CIs and p values.

When reporting p values please report as <0.001 and where higher as p=0.002, for example. When reporting 95% CIs please separate upper and lower bounds with commas instead of hyphens as the latter can be confused with reporting of negative values.

Please include the actual amounts and/or absolute risk(s) of relevant outcomes (including NNT or NNH where appropriate), not just relative risks or correlation coefficients. (example for absolute risks: PMID: 28399126).

Please include any important dependent variables that are adjusted for in the analyses.

4) Prespecified analysis plan/study protocol:

Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

For all observational studies, in the manuscript text, please indicate: (1) the specific hypotheses you intended to test, (2) the analytical methods by which you planned to test them, (3) the analyses you actually performed, and (4) when reported analyses differ from those that were planned, transparent explanations for differences that affect the reliability of the study's results. If a reported analysis was performed based on an interesting but unanticipated pattern in the data, please be clear that the analysis was data-driven.

5) Abstract layout:

Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions), i.e. combine the 2 middle sections together to make a section called ‘Methods and Findings’.

6) Author summary:

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The authors summary should consist of 2-3 succinct bullet points under each of the following headings:

• Why Was This Study Done? Authors should reflect on what was known about the topic before the research was published and why the research was needed.

• What Did the Researchers Do and Find? Authors should briefly describe the study design that was used and the study’s major findings. Do include the headline numbers from the study, such as the sample size and key findings.

• What Do These Findings Mean? Authors should reflect on the new knowledge generated by the research and the implications for practice, research, policy, or public health. Authors should also consider how the interpretation of the study’s findings may be affected by the study limitations. In the final bullet point of ‘What Do These Findings Mean?’, please describe the main limitations of the study in non-technical language.

The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

7) Introduction layout:

Please address past research and explain the need for and potential importance of your study. Indicate whether your study is novel and how you determined that. If there has been a systematic review of the evidence related to your study (or you have conducted one), please refer to and reference that review and indicate whether it supports the need for your study.

8) Discussion layout:

Please present and organize the Discussion as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy; one-paragraph conclusion.

9) Miscellaneous:

Please revise your ‘Short Title’ – the word ‘condicional’ should be replaced with ‘conditional’.

-----------------------------------------------------------

Comments from the reviewers:

Reviewer #1:

Using data from the 100 Million Brazilian Cohort, this paper investigates the effect of conditional cash transfers on mortality among people previously hospitalised for any psychiatric disorder. A propensity score-based approach to analysis is taken. I have several questions about the analysis and other statistical aspects of the study.

1. This study is referred to as "quasi-experimental". However, I cannot determine what the experimental element of this study is. It would seem preferable to refer to this as an observational study.

2. Given that the patients included in the dataset for analysis may not have had a primary psychiatric disorder diagnosis, is it correct that all included patients were hospitalised due to a psychiatric disorder? It seems that some included patients were people hospitalised with a psychiatric disorder (rather than "for a psychiatric disorder").

3. In the analysis presented in Tables 3 and 4, overall mortality is considered, but then mortality is divided up by cause. However, if an individual dies due an unnatural cause, they cannot die due to some other cause (for example). Thus, the results of the analysis presented in Table 3 for all apart from overall mortality do not have a useful interpretation. An appropriate analysis would account for competing risks of death by other causes, using a competing risks model.

4. It is unclear what the propensity score-based analysis that was applied here was - I could not access the Appendix, where additional details may have been supplied. What is clear is that a propensity score model for receipt of BFP was fit. However, the authors mention that IPTW was applied, but the details presented here do not align with a standard IPTW analysis. In such an analysis, all individuals would receive a weight of one over the probability that they received their actual "treatment" (here a cash transfer or not). Instead, here those participants who did receive the cash transfer had a weight of 1, and those participants who did not receive the cash transfer had another weight. I am not sure what is meant by "E(ps) is the probability if receiving the programme in the population": is this the average propensity score? The following approach is recommended:

a. Assess the common support condition by comparing the range of propensity scores in each exposure group, and exclude participants outside the range of common support.

b. Generate IPTWs following the formulas in reference 39.

c. Calculate standardised differences in the weighted space (i.e. after applying the IPTWs). It is impossible to determine how well the propensity score model has performed without the calculation of these standardised differences.

d. If standardised difference have reduced sufficiently, then the weighted outcome regression model can be it. This will include the exposure as the single covariate. Including the IPTWs as a covariate is not a recommended approach.

5. I do not understand how the ATT was calculated, how the stabilised versions of the IPTWs were calculated or how the kernel matching approach was applied. More details are required.

6. How many individuals were excluded due to missing data?

7. How were the standardised mean differences calculated for the categorical characteristics? Generally an SMD would be calculated for each level of the characteristic.

8. Given my comments on the analysis above, it is difficult to determine whether the conclusions drawn in this study are supported by the data.

Reviewer #2:

This is an interesting article using a quasi-experimental study design to examine whether enrollment in a conditional cash transfer program was associated with reduced mortality among Brazilians previously hospitalized for a psychiatric illness. I am generally in agreement with the analysis and its interpretation. However, I did have the following questions:

1) Why was the sample restricted to those with a psychiatric hospitalization only once during the study period? Researchers provide a rationale but I'm unsure how does this differ from other studies that looked at the overall Brazilian Bolsa Família Programme enrollment and mortality risk from different causes.

2) Did you stratify according to whether the participant primary vs. secondary diagnosis was for psychiatric disease to see if results differ according to disease severity?

3) Your propensity score did not control for any health-related covariates? Is health a strong predictor for enrollment in this CCT program?

4) Participants in your study only applied for the CCT program after their psychiatric hospitalization - would the initial hospitalization influence enrollment in CCT in any way?

5) Minor - I am confused when the researchers write about hospitalized beneficiaries in the text because I thought all the participants were hospitalized.

6) Minor - What is the rationale for the age groupings?

Reviewer #3:

This manuscript reports findings from an interesting analysis on a large Brazilian database in which the authors investigated mortality after a psychiatric hospitalisation and investigated associations with conditional cash transfer payments. My comments are as follows:

1. Line 121 - for a journal of this general nature, I think a sentence could usefully be added on what a CCTP entails, beyond simply saying that it's a social policy; in particular, some description is needed as to whether it's a tightly defined intervention internationally or is describing something much broader and heterogeneous. Currently the paper assumes that this will be familiar territory to all readers.

2. Lines 146-157 - likewise, there are quite a lot of assumptions about readers knowledge of Brazilian healthcare. For interpretation, it would probably be important to be clear what healthcare options are available in Brazil and what proportion of people will register for social assistance (and thus, presumably, appear on the source dataset).

3. Figure 1 - the substantial drop from n=152,862 to n=71,068 is a little concerning. Am I correct in assuming that 'registration date after follow-up has ended' represented an indication of data errors? If so, this seems quite a high level and it would be helpful if there was some assurance of data quality for the remainder.

4. Is BFP registration at individual or household level? If at household level, was any procedure deployed to account for clustering?

5. Table 2 - is there any particular reason for the inconsistency in the ordering of columns between the two comparison groups?

6. Is there any information that can be provided on the extent of propensity score overlap between the two comparison groups? Is it appropriate to be including people with 0% or 100% propensity scores (assuming there were some of them)? Or do the sensitivity analyses cover this scenario?

7. As a general comment, for a paper whose analyses are focused on public health and policy, it would be worth considering the inclusion of metrics with stronger communication in that field. From the results available, wouldn't it be possible to include a PAF equivalent (i.e., the proportion of deaths that could theoretically have been prevented with the intervention, assuming causality) and/or a NNT equivalent (number receiving the intervention required to result in one fewer deaths over a given follow-up period)?

Reviewer #4: Manuscript Number: PMEDICINE-D-24-01060R1

This manuscript is based in the 100 Million Brazilian Cohort, a dynamic cohort representing people who registered for CadUnico, a system used to apply for social assistance in Brazil. The authors examined whether a conditional cash transfer benefit, the Bolsa Familia Program (BFP) reduced mortality after a single hospitalization for a primary or secondary psychiatric diagnosis during 2008-2015. Importantly, the conditions for receiving the BFP cash transfer include participation education, health and other social programs.

The study sample comprised those who registered for CadUnico and applied for BFP after their psychiatric hospitalization and before 2015. Within the study sample, they compared mortality rates up to 2015 for those who received vs did not receive BFP. Those who received BFP were followed from the time they received it, and those who did not receive BFP were followed from the time they registered for CadUnico. BFP was associated with a substantially reduced mortality rate overall. This was primarily driven by a reduced mortality for "natural" causes, although there was also a trend for "unnatural" causes such as suicide.

These are the first results to suggest that a broad program of social assistance, not directed at psychiatric patients in particular, has major benefits in reducing mortality among psychiatric patients after discharge. It has broader significance too, opening the question of whether such assistance could reduce the well established high mortality rate among all people with psychiatric disorders.

Overall, the study appears to be solid, once one understands what the authors did. The presentation, however, is somewhat confusing. It took us quite some time to understand what they did. For most readers, who will not be familiar with the 100 million cohort nor the BFP, it would be much more difficult. In that regard we have several comments: 3 main ones and then several smaller ones. We have not included the many ways in which the study could be expanded in scope or in detail, because we recognize that the paper is already reporting an important result, and further results could be in future papers.

Main Comments

First

The authors need to clarify the study sample in the text and in Figure 1. The large number who were excluded due to registering for CadUnico after 2015 is not noted in the text under Participants where Figure 1 is first referred to. In Results, where the authors again refer to Figure 1, they only note that they excluded persons who did not meet inclusion criteria, without noting that the largest number were excluded due to registration after 2015 (this can be found in the Figure 1 but not easily). It is not clear why these persons were considered participants in the first place, but whatever they call them, what is important is to describe the study sample upfront and highlight the key points clearly, such as this one.

Note in the manuscript that the profile of individuals hospitalized for psychiatric reasons in other studies differs markedly from the cohort studied here (see Rocha et al, 2021, Revista de saude publica). The authors should also note how that might be due to the selection criteria for this study.

Other points that need to be clarified about the study sample include:

were these "first ever" psychiatric hospitalizations, or "first during 2008-2015"?

what were the main reasons why some who applied for BFP received it or did not receive it?

for those who received BFP, how long was the period between hospitalization and receipt?

what proportion were in psychiatric versus general hosptials? were there differences in mortality between these two groups? between those with primary vs secondary psychiatric diagnoses?

When the answers to these questions are not available, that should simply be stated.

Second

In the introduction or discussion, it would be helpful to further situate the reader in terms of the unified healthcare system in Brazil, who it serves, and the general psychosocial attention policies (e.g., Law 10.216/2001, implementation of CAPS, reduction of hospital beds in psychiatric hospitals and increase in beds in general hospitals).

Also, clarify how individuals who are registered in the CadUnico differ from the entire population even though they comprise over 60% of Brazil's population. One sentence referencing the original 100 million cohort paper should suffice to situate the reader.

Line 364: this may not be clear to international readers. A potential alternative "community based mental health centers (called psychosocial care centers, or CAPS, in Brazil)"

Space could be made by implementing the third suggestion below.

Third

The main results are in Tables 3 and 4, and Table 2 is not needed for the main text. Since Table 2 has only unadjusted results, it should be made a supplement. Table 2 provides information that might be needed to fully evaluate the study. Anything that needs to be highlighted- such as differences between BFP and non-BFP in person-years at risk- could be briefly alluded to in the text. However, presenting the unadjusted results in detail in the main text and table 2, before the main results, is quite confusing for readers.

Since groups differ on person-years at risk, could the authors explain the choice of IRRs rather than doing a survival analysis such as proportional hazards? We don't insist on doing a survival analysis, but think some rationale for the approach chosen is needed.

Smaller points:

Line 84: we think mortality risk should be mortality rate.

Line 112: note lack of LMIC data on mortality for psychiatric disorders.

Line 113: "approximately 62·2% of the global burden of suicide deaths may be attributed to psychiatric disorders". This statement derives from a paper but might be problematic because "attributing" suicide deaths to a single factor across the globe is fraught with problems. It seems unnecessary here though we leave this to the authors' discretion.

Lines 114-116: include the role of psychiatric treatments in mortality (e.g., long term neuroleptic use)

Line 131: note that the post-discharge period is particularly vulnerable, and that psychiatric hospitalizations themselves may increase all-cause mortality, as shown in many articles.

Line 237 delete statement that SMDS >.1 indicated confounding. This would not be the correct approach, and indeed, they do not follow it in constructing the propensity scores.

Line 260: how was length of hospitalization obtained? Authors may be aware that the typical way to obtain that information is not reliable because administrators often circumvent the need to maintain the same AIH number for hospitalizations longer than 15 days and often generate a new number, which may cause problems calculating length of stay. Other authors have found ways to address this, please clarify how you dealt with this issue or if LOS was calculated in a different way

Line 281: please provide rationale for excluding everyone who received Bolsa Familia before their hospitalization, as it seems that this would be an important group to examine- but might be beyond the scope of the study.

Line 305: interesting finding, could there be group differences because women were more likely to have been excluded from this cohort for being on benefits before hospitalization (e.g., because of pregnancy)?

Line 331: change "there were effects" to "results were consistent with an effect"

Line 365: consider explaining that a national policy gradually reduced the numbers of hospital beds in psychiatric hospitals and implemented community based care nationally starting in 2021.

Lines 367-369: the national policy encourages hospitalizations in general hospitals and discourages psychiatric beds in psychiatric hospitals. Authors may be referring to the fact that smaller cities may not have professionals with the right expertise (psychologists, psychiatrists, occupational therapists, psychiatric nurses), or the appropriate space in general hospitals to serve the complex needs of this population

Lines 455-456: national policy encourages general hospital admission vs psychiatric hospital, please revise.

Ezra Susser and Ana Florence

Any attachments provided with reviews can be seen via the following link:

[LINK]

-----------------------------------------------------------

COMMENTS FROM THE ACADEMIC EDITOR:

The Academic Editor was supportive of your work from the outset. She commented that she believed that you address a very important topic and also commented positively regarding your unique data set. She noted the concerns raised by the statistical reviewer, and encourages you to address these.

-----------------------------------------------------------

GENERAL EDITORIAL REQUESTS:

1. Please upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

2. Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

3. We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

-----------------------------------------------------------

To submit your revised manuscript please use the following link:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

Decision Letter 2

Syba Sunny

23 Aug 2024

Dear Dr Bonfim,

Many thanks for submitting your revised manuscript "Do conditional cash transfers reduce mortality in people hospitalised with psychiatric disorders? A cohort study of the Brazilian Bolsa Família Programme" (PMEDICINE-D-24-01060R2) to PLOS Medicine. Firstly, let me say that we are grateful for your thorough engagement with the reviewer comments and appreciate the efforts made to act on these. The paper has now been reviewed again by a subject expert and a statistician; their comments are included below and can also be accessed here: [LINK]

As you will see, the reviewers were largely positive about the revised paper, but the statistician had a number of recommendations. After discussing the paper with the editorial team and an academic editor with relevant expertise, I'm pleased to invite you to revise the paper in response to the statistician’s comments. Please note that we plan to send the revised paper back to the statistical reviewer, and we cannot provide any guarantees at this stage regarding publication.

When you upload your revision, please include a point-by-point response that addresses all of the reviewer and editorial points, indicating the changes made in the manuscript and either an excerpt of the revised text or the location (e.g.: page and line number) where each change can be found. Please also be sure to check the general editorial comments at the end of this letter and include these in your point-by-point response. When you resubmit your paper, please include a clean version of the paper as the main article file and a version with changes tracked as a marked-up manuscript. It may also be helpful to check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper.

We ask that you submit your revision by Sep 13 2024 11:59PM. However, if this deadline is not feasible, please contact me by email, and we can discuss a suitable alternative.

Don't hesitate to contact me directly with any questions (ssunny@plos.org).

Best regards,

Syba

Syba Sunny, MBBS, MRes, FRCPath

Associate Editor

PLOS Medicine

ssunny@plos.org

-----------------------------------------------------------

Comments from the academic editor:

The academic editor continued to be supportive of your manuscript and hopes that you can address the statistician’s points.

-----------------------------------------------------------

Comments from the reviewers:

Reviewer #1 (statistician):

I thank the authors for their responses to my comments on the previous version of this manuscript. However, I have noticed some inconsistencies throughout the manuscript and supplementary materials that require attention. Please note that line numbers in my comments refer to those on the tracked changes version of the manuscript.

1. Abstract: please change "competitive risk" to "competing risks" (here and on line 356 and Tables 2, 3, and in the Supplementary material)

2. Line 315: IPTW estimation is mentioned before it is defined. Please define this acronym when first used.

3. Is the primary aim to estimate the ATT (average effect in the treated), as stated on line 347? This should be stated early on in the Statistical Analyses section. If this is the case the authors were correct in their previous choice of weights: 1 for those individuals who did receive a cash transfer, and PS/(1-PS) for those who did not. However, I am confused since on line 389 it is stated that estimation of the ATT is done as sensitivity analysis (or is this just referring to one particular method for estimation of the ATT)? I also note that the text in the manuscript regarding the weights should match that in the Appendix.

4. Are the SMDs in Table 1 for weighted or unweighted data? Please provide SMDs for the unweighted and weighted data in Table 1. Why do the SMDs in Table 1 not match those in the before IPTW column in S4 Table 4?

5. S4 Fig 3 does not match the data provided in S4 Table 3 - for example, the maximum propensity score in the non-BFP sample is stated as being 0.767 in the table, but is >0.8 in the figure.

6. Please reduce the number of significant figures provided in the tables in S5 to 2.

7. An intraclass correlation is provided in S6. What is the outcome that this is calculated for? Were multiple members of households hospitalised with psychiatric disorders? How many households were included, and what sizes were these? Is this referred to anywhere in the paper? I would suggest deleting this section.

Reviewer #3:

My comments have been addressed satisfactorily. I have no further comments.

Any attachments provided with reviews can be seen via the following link: [LINK]

--------------------------------------------------------- ---

General editorial requests:

(Note: not all will apply to your paper, but please check each item carefully)

* We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. Please do not add or remove authors without first discussing this with the handling editor. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

* Please upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

* Thank you for providing a data availability statement. We have a few requests for amendments here.

(1) Please note that a study author cannot be the contact person for the data. We ask that details be provided so that data requests can be made to a non-author institutional point of contact, such as a data access or ethics committee, as this helps guarantee long term stability and availability of data. Providing interested researchers with a durable point of contact ensures data will be accessible even if an author changes email addresses, institutions, or becomes unavailable to answer requests.

(2) We note that there are appears to be 3 datasets that have been accessed in this study. Could you provide access details for all three datasets (in the relevant metadata section) please?

* We expect all researchers with submissions to PLOS in which author-generated code underpins the findings in the manuscript to make all author-generated code available without restrictions upon publication of the work. In cases where code is central to the manuscript, we may require the code to be made available as a condition of publication. Authors are responsible for ensuring that the code is reusable and well documented. Please make any custom code available, either as part of your data deposition or as a supplementary file. Please add a sentence to your data availability statement regarding any code used in the study, e.g. "The code used in the analysis is available from Github [URL] and archived in Zenodo [DOI link]" Please review our guidelines at https://journals.plos.org/plosmedicine/s/materials-software-and-code-sharing and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. Because Github depositions can be readily changed or deleted, we encourage you to make a permanent DOI'd copy (e.g. in Zenodo) and provide the URL.

* We note that the datasets referred to in your study have been used by other publications. Nevertheless, it might be useful for readers if you could perhaps provide a short summary of the datasets in the Supporting Information files? This may enable the reader to understand the nature of the data a bit better without having to leave the article page.

* Thank you for including the STROBE checklist in your Supporting Information. Could you kindly revise this so that you only use section and paragraph numbers, rather than page numbers? This is because page numbers can change at a later point in the publication process. Please also add the following statement, or similar, to the Methods: "This study is reported as per STROBE guideline (S1 Checklist)."

SUPPLEMENTARY MATERIAL

* Please cite your Supporting Information as outlined here: https://journals.plos.org/plosmedicine/s/supporting-information

REFERENCES

* Where website addresses are cited, please include the complete URL and specify the date of access (e.g. [accessed: 12/06/2023]).

Decision Letter 3

Syba Sunny

26 Sep 2024

Dear Dr. Bonfim,

Thank you very much for re-submitting your manuscript "Do conditional cash transfers reduce mortality in people hospitalised with psychiatric disorders? A cohort study of the Brazilian Bolsa Família Programme" (PMEDICINE-D-24-01060R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by the statistical reviewer. There are some outstanding requests from the statistical reviewer. However, I am pleased to say that provided the remaining requests are satisfied, and editorial and production issues are dealt with, we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by the reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 2 weeks (extended from the usual 1 week to take into account the statistical reviewer’s requests). Please email us (ssunny@plos.org) if you need more time.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Oct 10 2024 11:59PM.

Sincerely,

Syba

Syba Sunny, MBBS, MRes, FRCPath

Associate Editor

PLOS Medicine

ssunny@plos.org

------------------------------------------------------------

Comments from Statistical Reviewer:

Reviewer #1: I thank the authors for their responses to my comments on the previous version. My only remaining comment is to do with the intracluster correlation and S6 text. I appreciate the additional context the authors provided regarding the inclusion of this material, and agree that it should remain in the Supplementary material. However, the issue is not the intracluster correlation itself (and I note that an ICC of around 0.02 as found here is meaningful and can have a large impact on inference for the outcome). The issue is to do with the estimation of the effect of the exposure when clustering is accounted for in this way - please include the estimated exposure effect and 95% confidence interval for this fitted model in the main paper. That is, the sentence "Our sample had no clustering effects" should be deleted, while a sentence like "When clustering by family was accounted for, the estimate of the effect of BFP on OUTCOME was ??? (S6 Text)" should be included in an appropriate place in the Results section. Please also state in the S6 text which model had mixed effects included: was this the Poisson model for mortality rates per 100,000 person years?

------------------------------------------------------------

Comments from the editor:

Thank you for engaging with the review process so thoroughly. I have some very minor requests:

1) Abstract - Methods and Findings: Please expand the abbreviation ‘ICD-10’.

2) I note that you’ve used the abbreviation ‘CCTP’ for conditional cash transfer programmes, yet it’s referred to as ‘CTP’ in the abstract. Could you revise for consistency across the abstract and main text please?

3) Line 193 of clean revised copy: The first instance of the abbreviation ‘BFP’ is not expanded. (Rather I see that the 2nd instance on line 195 is.) Could you kindly expand on the first instance that ’BFP’ appears in the main text please?

Decision Letter 4

Syba Sunny

9 Oct 2024

Dear Dr Bonfim, 

On behalf of my colleagues and the Academic Editor, Charlotte Hanlon, I am pleased to inform you that we have agreed to publish your manuscript "Do conditional cash transfers reduce mortality in people hospitalised with psychiatric disorders? A cohort study of the Brazilian Bolsa Família Programme" (PMEDICINE-D-24-01060R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process. 

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper. 

Sincerely, 

Syba

Syba Sunny, MBBS, MRes, FRCPath 

Associate Editor 

PLOS Medicine

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Strobe Checklist. STROBE Statement—Checklist of items that should be included in reports of observational studies.

    (DOCX)

    pmed.1004486.s001.docx (35KB, docx)
    S1 Text. Accuracy analysis of the linkage between CadÚnico and the mortality information system in a randomised sample of 10,000 record pairs.

    (DOCX)

    pmed.1004486.s002.docx (16.4KB, docx)
    S2 Text. Detailed information from eligible study population.

    (DOCX)

    pmed.1004486.s003.docx (15.9KB, docx)
    S3 Text. Propensity score: Definition, estimation, summary, and support graphs.

    (DOCX)

    pmed.1004486.s004.docx (14.8KB, docx)
    S4 Text. Intraclass correlation coefficient estimation.

    (DOCX)

    pmed.1004486.s005.docx (14.4KB, docx)
    S5 Text. Description of individuals excluded from the analysis.

    (DOCX)

    pmed.1004486.s006.docx (13.5KB, docx)
    S6 Text. Summary of the dataset’s description.

    (DOCX)

    pmed.1004486.s007.docx (16.9KB, docx)
    S1 Fig. ROC curve of the 100 million Brazilian Cohort and SIH (2001–2018) linkage.

    Source: Developed by the CIDACS Data Production Center.

    (TIF)

    pmed.1004486.s008.tif (639.5KB, tif)
    S2 Fig. ROC curve of the 100 million Brazilian Cohort and SIM (2000–2015) linkage.

    Source: Developed by the CIDACS Data Production Center.

    (TIF)

    pmed.1004486.s009.tif (865.6KB, tif)
    S3 Fig. Distribution of the propensity score in the sample, 2008–2015.

    (TIF)

    pmed.1004486.s010.tif (575.3KB, tif)
    S1 Table. Mortality rates overall and by subgroups through receipt of the BFP, 2008–2015.

    (DOCX)

    pmed.1004486.s011.docx (19.6KB, docx)
    S2 Table. (A) Logistic regression to estimate propensity scores for receiving Bolsa Familia according to covariables, N = 57,905 (B) Propensity score description in accordance with the confounding covariates observed, Brazil, 2008 to 2015, N = 57,905.

    (DOCX)

    pmed.1004486.s012.docx (18.1KB, docx)
    S3 Table

    (A) ATT of overall mortality for BFP receipt between 2008 and 2015 using KM. (B) ATT of natural causes of death for BFP receipt between 2008 and 2015 using KM. (C) ATT of unnatural causes of death for BFP receipt between 2008 and 2015 using KM. (D) ATT of suicide for BFP receipt between 2008 and 2015 using KM.

    (DOCX)

    pmed.1004486.s013.docx (15.2KB, docx)
    S4 Table. Crude and adjusted association of BFP participation with overall, natural, unnatural, and suicide mortalities, 2008–2015.

    (DOCX)

    pmed.1004486.s014.docx (15.9KB, docx)
    S5 Table. Association of BFP participation with overall, natural, unnatural, and suicide mortalities accounting for missing data, 2008–2015.

    (DOCX)

    pmed.1004486.s015.docx (21.4KB, docx)
    S6 Table. Incidence rate ratio of BFP participation with overall, natural, unnatural, and suicide mortalities, 2008–2015.

    (DOCX)

    pmed.1004486.s016.docx (15.3KB, docx)
    S7 Table

    (A) Intraclass correlation estimation for the household level. (B) Association of BFP participation with overall mortality considering household level, 2008–2015.

    (DOCX)

    pmed.1004486.s017.docx (15.4KB, docx)
    S8 Table. Description of individuals excluded from the analysis following definition of BFP exposition, 2008–2015.

    (DOCX)

    pmed.1004486.s018.docx (22.6KB, docx)
    S9 Table. Description of year and length of hospitalisation overall and by BFP participation, 2008–2015.

    (DOCX)

    pmed.1004486.s019.docx (15.3KB, docx)
    Attachment

    Submitted filename: Point by Point_PlosMedicine.docx

    pmed.1004486.s020.docx (56.1KB, docx)
    Attachment

    Submitted filename: Point by Point_PlosMedicine_2_v3.docx

    pmed.1004486.s021.docx (26.7KB, docx)

    Data Availability Statement

    The code used in the analysis is available from Github [https://github.com/profacamilabonfim/Codes-from-the-paper.git] and archived in Zenodo [https://zenodo.org/records/13750552]. The data analyzed in this study is hosted by the Centre of Data and Knowledge Integration for Health (CIDACS). Full access to the data is restricted due to its sensitive nature and the exclusive licensing agreement for its use in this study. The privacy regulations set by the Brazilian Ethics Committee prohibit the public availability of this data. However, upon reasonable request, and provided that all ethical and legal requirements are met, the institutional data curation team can make the data available. Further information can be obtained by emailing cidacs.curadoria@fiocruz.br. Study Protocol available on: https://pubmed.ncbi.nlm.nih.gov/36201469/.

    Articles from PLOS Medicine are provided here courtesy of PLOS

    RESOURCES