Effects of community-based rehabilitation on caregivers of people with schizophrenia in Ethiopia in the RISE trial

doi:10.21203/rs.3.rs-3573904/v1

Download PDF

Research Article

Effects of community-based rehabilitation on caregivers of people with schizophrenia in Ethiopia in the RISE trial

https://doi.org/10.21203/rs.3.rs-3573904/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Schizophrenia is a severe mental health condition that is burdensome for those affected and their families. Community-based rehabilitation (CBR) is a recommended treatment component for schizophrenia in low- and middle-income countries (LMICs), but there is little evidence on the effects of CBR on caregivers of people with schizophrenia. RISE was the first randomised controlled trial of CBR for schizophrenia in a low-income country. In this paper, we extend our previous examination of caregiver impact by (1) investigating the impact of CBR on caregiver stigma and burden, (2) assessing effect modification of outcomes, and (3) determining predictors of caregiver outcomes at 12 months.

Methods

Data are from the cluster-randomised controlled RISE trial, which investigated CBR and facility-based care versus facility-based care alone among 166 people with schizophrenia and 166 linked caregivers in 48 sub-districts in Ethiopia. We analyse the effect of CBR on caregiver stigma, unemployment and burden measured with the WHO Family interview Schedule-Impact at 6 and 12 months; and caregiver depression, reduction in work due to caregiving and caregiver burden measured with the Involvement Evaluation Questionnaire at 6 months. Logistic and linear regression models adjusted for clustering by sub-district and health centre were used for binary and continuous outcomes respectively. Effect modification by caregiver sex, age, baseline of the outcome, and baseline disability were assessed. Baseline factors associated with caregiver outcomes across the whole cohort at 12 months were investigated using hierarchal regression modelling.

Results

Data were available for 112 caregivers at 6 months (67%), and 149 caregivers at 12 months (90%). There was evidence that CBR was associated with greater tendency to reduce work due to caregiving at 6 months (OR:2.40, 95%CI:1.06–5.45). No evidence of an intervention effect was found on unemployment, depression, stigma or other aspects of caregiver burden. There was no evidence for effect modification. Higher baseline disability was independently associated with greater caregiving burden at 12 months (β:0.26, 95%CI:0.14–0.37).

Conclusions

To improve outcomes of people with schizophrenia and their caregivers in LMICs, interventions and research addressing the needs of caregivers are needed.

Trial registration

: Clinical Trials.gov Identifier NCT02160249. Registered on 10/06/2014.

community-based rehabilitation

psychosocial intervention

community mental health services

schizophrenia

caregivers

randomized controlled trial

Schizophrenia spectrum disorders are severe mental health conditions that place a high burden on those affected and their families (1, 2). These conditions are associated with high levels of disability and poverty, premature mortality, and human rights abuses (3, 4). Evidence-based treatment for mental health conditions is largely inaccessible in low- and middle-income countries (LMICs), due to a lack of investment in services and a shortage in trained mental health workers (5).

Community-based psychosocial interventions for schizophrenia can reduce symptom severity and improve functioning of people with schizophrenia in LMIC, yet access to such services is extremely limited (6). The World Health Organization (WHO) recommends Community Based Rehabilitation (CBR) as a component of the treatment of people with schizophrenia in LMIC (7). CBR is an approach aimed at improving the quality of life of people with disabilities by involving not only the person with the disability, but also families and communities (8). It can be delivered by lay workers, making it suitable for low-resource settings. Potential elements of CBR are promoting health, education, livelihoods, and social life, as well as a focus on empowerment. Counselling, problem-solving techniques, and community mobilisation may be used (9).

In Ethiopia, there is an estimated treatment gap for schizophrenia of 90% (10, 11). In the past decade, mental health has been integrated into primary care in pilot sites in Ethiopia, which has increased the availability of basic facility-based care (anti-psychotic medication and psychoeducation) for some people with schizophrenia (12, 13). However, family members are responsible for most care, leading to a heavy financial, social, and emotional burden (2, 14). Furthermore the CBR for people with Schizophrenia in Ethiopia (RISE) cluster-randomised trial found that CBR is effective in reducing disability and symptom severity among people with schizophrenia in a rural district in south-central Ethiopia who had not improved of engaged with primary mental health care services (15).

Caregivers are important actors in CBR but there is an evidence gap on the effects of CBR, and other community-based psychosocial interventions, on the experience of caregivers of people with schizophrenia. We hypothesised that improved functioning and symptoms in people with schizophrenia would lead to improvements in caregivers’ experience. This hypothesis is supported by a previous observational study in Ethiopia showing reduced symptoms and longer periods in remission in people with severe mental health conditions was associated with reduced caregiver burden (16). Furthermore, findings from the RISE trial demonstrated a beneficial intervention effect of CBR on caregiver burden at 12 months, measured by the Involvement Evaluation Questionnaire (IEQ), in the domains of tension and worrying. However there was no evidence of an effect on depression or reduction in work due to caregiving at 12 months (15). In this paper we explore the effect of CBR on the additional exploratory outcomes of caregiver stigma, unemployment and burden measured with the WHO Family interview Schedule-Impact at 6 and 12 months; and caregiver depression, reduction in work due to caregiving and caregiver burden measured with the IEQ at 6 months, among participants in the RISE trial. Secondary aims were to assess (i) effect modification of any CBR effect by participant and caregiver characteristics and (ii) the determinants of caregiver outcomes across the whole cohort.

Study design and participants

This study uses data on caregiver outcomes from the parallel-arm cluster-randomised controlled RISE trial, which compared CBR plus facility-based care to facility-based care alone (15, 17). The RISE trial was conducted in Sodo District, Gurage Zone, Southern Nations, Nationalities and People’s Region, Ethiopia. The study protocol and main trial results were previously published (15, 17). Sodo District has an estimated total population of 170,000 people in 58 sub-districts. Most of the population live in remote rural areas. The district has high levels of poverty, and the main economic activity is subsistence farming. Primary care is delivered by health officers and nurses at one primary hospital and seven health centres, each catering to a mean of six sub-districts.

The unit of randomisation in the trial was sub-district, and a total of 54 sub-districts were allocated using minimization in a 1:1 ratio to the intervention arm (facility-based care plus CBR) and the control arm (facility-based care alone). Randomization and masking have been described previously (15). Trial participants were individuals with a diagnosed schizophrenia spectrum disorder, recruited from the PRIME (PRogramme for Improving Mental healthcarE) cohort study, which had already implemented facility-based mental health care in primary health centres for 6 months in the area (18). Only participants with high levels of disability after 6 months access to facility-based care were included in the RISE trial. Participants were followed up at 6 months (midline) and 12 months (endline) after recruitment. Data were collected in the health centre or at participants’ homes.

For each trial participant, one primary caregiver was identified. Caregivers were eligible if they were aged 18 years or older and provided regular support to the person with schizophrenia. If the original caregiver was unavailable at endline, caregiver-reported data were collected from a different caregiver meeting the criteria.

At recruitment, written informed consent was taken from both the person with schizophrenia and the caregiver by a psychiatric nurse. In this process verbal information was given to the person with schizophrenia and the caregiver and an information sheet was provided that was elaborately discussed and explained. Time for questions was given and non-technical language was used. Sufficient time was allowed for participants to think about the information, with extra time if the person with schizophrenia was unwell or seemed to have cognitive difficulties. If the participant was unable to write, a thumb impression and a witness’s signature were recorded. These processes meant all caregiver participants were able to participate fully in informed consent procedures despite the majority having no formal education.

Procedures

The interventions have been described in detail previously (15, 17). All participants had access to facility-based care, which is a stepped care model in which most care is delivered in primary care. It primarily comprised prescription of anti-psychotic medication and psychoeducation by nurses and health officers trained in the WHO mental health Gap Action Programme-Intervention Guide (mhGAP-IG) (7).

Intervention clusters additionally received a CBR intervention which had been shown to be acceptable and feasible in the rural Ethiopian context (12, 19, 20). CBR workers were lay people from the local area with at least ten years of education but no prior experience in delivering mental healthcare. They received five weeks initial training in CBR delivery, guided by a manual, including basic counselling and problem-solving techniques (20). Training was split between classroom teaching and fieldwork.

CBR was delivered by 11 CBR workers, each supporting a median of seven participants and their families (range 4–11). CBR visits lasted 30–90 minutes and took place at the participants’ home. Caregivers were requested to participate in every visit. The intervention emphasized human rights, social inclusion and personal recovery. Visits were every 1–2 weeks for the first three months, then every 2 weeks for the next five months, and monthly for the final four months. Two supervisors oversaw the home visit content and frequency. Core modules covered understanding schizophrenia, access to health services, crisis management and human rights. Optional modules included adherence support, family intervention, support returning to work and social activities, and dealing with stigma. Caregivers were intimately involved in CBR, for example, contributing to goal setting, reminding their relative to take medication, accompanying them to community activities, and reducing unhelpful communication (such as excessive criticism) in the home. The aim was to empower caregivers to keep up the positive effects of CBR after the intervention had terminated. CBR workers were trained to be aware of potential distress in caregivers and to propose solutions such as speaking with, or sharing the practical or financial burden with, relatives and neighbours; and attending the health centre for mental health support. CBR workers met with community members to mobilise resources for individual participants, conducted public awareness-raising meetings and ran family support groups. These groups were intended as places for caregivers to gain mutual emotional support.

Instruments and measurements

The primary trial outcome was participant disability (proxy-reported 36-item WHODAS) at 12 months (17). The present paper focuses on the following pre-specified exploratory trial outcomes in caregivers: caregiver employment at 6 and 12 months, caregiver stigma at 6 and 12 months (WHO Family Interview Schedule (WHO FIS)- Stigma total score), caregiver burden measured with WHO FIS- Impact at 6 and 12 months, caregiver burden measured with IEQ at 6 months, reduction in work due to caregiving at 6 months (binary question) and caregiver depression at 6 months (PHQ-9 ≥ 5 and total score). The appendix (A1) includes details on the validity of these measures. In the Ethiopian setting, a PHQ-9 score ≥ 5 indicates possible major depressive disorder (21). We have previously presented the intervention effect on IEQ, caregiver depression and reduced work due to caring at 12 months (15); the IEQ analysis was repeated in order to provide a foundation for the analysis of effect modification and determinants of caregiver outcomes.

Sample size

Of the 166 caregivers recruited into the RISE trial, 149 (89.8%) were seen within the pre-specified window for analysis (+/- 10 weeks of endline) at 12 months. This provided 86% power to detect a standardised mean difference (SMD) of 0.5 for continuous outcomes, assuming Type 1 error of 5%, and using the intra-cluster correlation (ICC) of 0.02 from the observed data. Similarly, there was 80% power to detect a difference in proportions of e.g. 50% vs 28% with the same assumptions. For the 6 month outcomes, 112 (67.4%) caregivers were seen within the pre-specified window for analysis (+/- 10 weeks of endline), giving 80% power to detect an SMD of 0.56 with a Type 1 error of 5%.

Statistical analysis

Statistical analysis was done with Stata (version 16). The appendix (A2) includes a table of the coding of all variables used in this report. To assess differences between treatment arms at baseline, cross-tabulations for categorical variables and means and medians for continuous variables were used. Baseline caregiver and participant characteristics were compared between those with and without complete 6 and 12-month outcome data with random-effects linear and logistic regression for continuous and binary outcomes respectively, adjusting for clustering by sub-district as a random effect and health center as a fixed effect. Missing data were recoded as a separate category for exposure variables so no observations would be dropped.

The effectiveness analyses used intention-to-treat principles. For all binary outcomes, Generalized Estimating Equation (GEE) models were used instead of random-effects models, as the differences between the quadrature points was > 0.01% in the random-effects models (22). For continuous variables, a linear regression random-effects model was fitted, including sub-district cluster as a random effect and health centre as a fixed effect. Minimally-adjusted analyses were first fitted, including only the baseline score of the outcome of interest and the clustering variables. Next, fully-adjusted models were fitted that additionally adjusted for baseline WHODAS score, variables associated with missingness, and variables unbalanced at baseline (23, 24). We created a final model through a backwards modelling strategy where all covariates that did not change the effect estimate by 10% or more when taken out of the model were removed. If the model had computational issues based on collinearity of the included variables one of the variables was dropped.

Effect-modification of intervention arm with caregiver’s age and sex, baseline WHODAS score (dichotomized with a cut-off of 40), and baseline of the outcome of interest was assessed for those outcomes associated with CBR by fitting an interaction term with the effect-modifier of interest and intervention arm in those final models. Validity checks and sensitivity analyses were done, such as using a log-transformation on the outcome variable when that variable was not normally distributed. Other sensitivity analyses included outcome data that were collected outside the specified +/- 10-week window, excluding those records that related to a different caregiver at that timepoint than at baseline, multiple imputation and re-running the models without missingness coded as a category for the covariates. See appendix A3 for details of analyses.

We examined factors associated with caregiver burden (IEQ total score) and depression (PHQ9 ≥ 5) at 12 months using a hierarchical conceptual framework (25) based on the peer-reviewed literature (appendix A4, Fig. 1) (2, 14, 16, 26–29). Based on the conceptual framework, we identified the following levels for the hierarchical analysis:

Level 1: Factors relating to the person with schizophrenia, social norms, and caregiver working away from home (most distal)

Level 2: Time spent with person with schizophrenia and stigma

Level 3: Caregiver coping and capacity

Level 4: Burden (most proximal)

Univariable analyses were conducted to assess the association between each exposure variable with depression and burden separately. We used GEE logistic regression models for depression and random effects linear regression for burden. Factors that had a Wald-test p-value lower than 0.1 in the univariable analysis were included as covariates for the multivariable analysis. First, models were run including all the Level 1 variables that had a p < 0.1 in the univariate analysis (model 1). Variables were retained if they were independently associated with the outcome (p < 0.1) (model 1b). The subsequent levels were then added stepwise, to indicate whether variables in the distant levels are mediated through variables in the proximate levels.

Ethical approval
The study was approved by the LSHTM Research Ethics Committee (reference 0735-2), the Addis Ababa University College of Health Sciences Institutional Review Board (reference 083/13/Psy), and the Ethiopian National Research Ethics Review Committee (reference 310/048/2015). The trial was registered on Clinical Trials.gov (NCT02160249) on 10/06/2014.

Participant characteristics

Trial participants and caregivers were enrolled between September 16, 2015 and March 11, 2016. Of the 54 available sub-districts, 27 were randomised to the intervention arm and 27 to the control arm. A total of 294 potential participants were pre-screened, of whom 91 were excluded. A further 37 individuals were not enrolled. Of these, 22 participants (10·8%) did not meet the inclusion criteria, one participant and six caregivers declined (3·4%) and eight participants were excluded due to already reaching sufficient numbers in the cluster (3·9%). Three sub-districts were excluded at each of the pre-screening and enrolment stages. Hence of 54 potential sub-districts for inclusion, 48 were included. Twenty-four sub-districts (79 participants and linked caregivers) were assigned to the intervention arm and 24 sub-districts (87 participants and linked caregivers) were assigned to the control arm.

The 166 caregivers had a mean age of 41 years (SD=14.7). Approximately half were male (n=77; 46.4%), and half were aged >35 years (n=93; 56.0%). Two thirds of caregivers were married (n=111; 66.9%), and almost all were either a first degree relative or a spouse of the person with schizophrenia (N=149; 89.8%). The majority had no education (N=96; 57.8%), whilst 11.4% (N=19) had more than 8 years of education. About one third (N=63; 38.0%) were illiterate. Most caregivers had reduced their work due to the caring burden (N=134; 80.7%) and 44.0% of caregivers had depression (PHQ≥5) (N=73). 11 (6.6%) had missing data on at least one baseline characteristic.

Caregiver depression (PHQ≥5) and caregiver employment were considered unbalanced by arm at baseline (Table 1), along with several participant characteristics (sex, education level, socioeconomic status, employment status, the level of social support, and occupation; reported previously (15)).

In each arm, 27 caregivers (31.0% for the control arm and 34.2% for the intervention arm) did not complete the outcome assessment at 6 months. At 12 months, 11 (12.6%) caregivers in the control arm and 6 (7.6%) in the intervention arm did not complete the assessment. Characteristics associated with missingness at 12 months were urban residence (p=0.06), higher socioeconomic status (p=0.02), higher caregiver burden (IEQ score) (p=0.03), and shorter illness duration (p=0.04), relapse in preceding 6 months (p=0.04), older age (p=0.05), male sex (p=0.06) amongst participants with schizophrenia. Female caregivers (p=0.07) with longer travel time to the facility (p=0.07) were more likely to have missing data at 6 months.

[Table 1 here]

Effect of CBR on caregiver outcomes

Table 2 shows the results of the minimally adjusted and final models respectively for caregiver outcomes. In the final model there was evidence that caregivers who were randomised to the CBR arm were more likely to have reduced their work due to undertaking caring activities at 6 months compared to those in the control arm (aOR=2.40, 95%CI:1.06-5.45, p=0.04). There was no evidence of an intervention effect for other outcomes. The sensitivity analyses were broadly consistent with the main analysis (appendix A5 and A6).
There was no evidence of effect modification by caregiver’s age and sex, baseline WHODAS score and baseline worrying (dichotomized with a cut-off of 10), tension (dichotomized with a cut-off of 10), or reduced work.

[Table 2 here]

Factors associated with caregiver outcomes

Disability, disorganized symptoms, lower caregivers’ education, being a first-degree family member or spouse, the caregiver not going out of the home to work, stigma, and time spent with the person with schizophrenia met the p<0.10 threshold in the univariable analysis related to burden (see appendix A7). The result of the hierarchical linear regression models for burden at 12 months are shown in Table 3. Higher levels of disability were consistently associated with higher caregiver burden in all models (p<0.001). First-degree relatives and spouses reported higher levels of burden than more distant relatives or unrelated caregivers. Caregiver stigma was also independently associated with greater burden.

[Table 3 here]

Disability, social support, caregiver gender, residence, relationship to the person with schizophrenia, caregiving burden, time spent with person with schizophrenia, and stigma met the p<0.10 threshold in the univariable analysis relating to depression (see appendix A8). Table 4 shows the results for the hierarchical logistic regression models of depression at 12 months. Disability was initially associated with depression, but not after adjusting for caregiving burden (p=0.89) suggesting this was largely mediated through burden, assuming the framework is correct. Living in a rural area was independently associated with higher depression (OR=8.56, 95%-CI:1.33-44.08). Adult children of the person with schizophrenia had the highest odds of depression, and siblings and other relatives had the lowest. Stigma and burden were also independently associated with higher odds of depression (OR=1.04, 95%-CI:1.00-1.08; OR=1.03, 95%-CI:1.01-1.05 respectively for a unit increase in score).

[Table 4 here]

This analysis of exploratory data from the RISE trial looked at depression, burden, employment, reduction in work due to caring and stigma among caregivers of people with schizophrenia in Ethiopia. RISE is the first randomised trial of CBR for people with schizophrenia in a low-income country, which means our results offer new insights into caregiver outcomes in such settings. With a few exceptions (30-32), most previous evaluations of psychosocial interventions for people with schizophrenia in LMICs have not assessed the impact on caregivers and few, to our knowledge, have found a positive effect (6, 30). We have previously demonstrated that, there was evidence that CBR is effective in reducing worrying and tension in caregivers at 12 months (15). In the current analysis, CBR appeared to increase caregivers’ need to reduce their work due to caring at 6 months. This may be because caregivers were influenced by CBR to undertake more frequent or intensive caring activities, such as accompanying their relative to the health centre or to social activities. Alternatively, physically attending CBR sessions may have meant caregivers had less time to work, thereby indicating a possible negative effect of CBR. The fact that this effect was seen only at 6 months, when sessions were more frequent, but not at 12 months (15), when session frequency was monthly, supports this hypothesis. We found in the RISE pilot study that CBR participation was sometimes less disruptive for female compared to male participants because women could continue their usual work (for example, preparing food or handicrafts) whilst speaking with the CBR worker (12). However men’s typical work roles, for example, tending to livestock, were less compatible.

Our finding of no intervention effect on caregiver stigma, depression or employment is likely to be because CBR does not adequately address those specific aspects of caregivers’ experiences. We also previously found CBR to be ineffective in reducing discrimination, improving employment or work functioning in people with schizophrenia (15). CBR is therefore unlikely to decrease poverty levels. Our formative work identified poverty as the most pressing need experienced by families affected by schizophrenia (19); CBR’s limited ability to affect this underpinning concern may explain the absence of impact on several aspects of caregiver burden. Furthermore, only 7/24 (30%) of sub-districts ran a family support group and 3/24 (13%) subdistricts had only 1 or 2 meetings (15). The absence of opportunities for mutual support amongst caregivers may have limited the impacts of CBR on caregiver outcomes. The lack of effect on several caregiver outcomes reflects findings from the COPSI trial of community-based care for people with schizophrenia in India, which also did not demonstrate effects of CBR on caregiver burden or stigma (32). However, similar to the qualitative results from the COPSI trial, qualitative data from the RISE pilot study suggested that caregivers worried less and felt less need to supervise their relative, giving them more time for daily tasks (12). It is therefore also possible that the instruments used to measure caregiver outcomes were not sensitive enough to change during the intervention period.

There was an independent association between greater baseline disability and increased burden at 12 months. This aligns with previous studies indicating that greater functional impairment is associated with higher caregiver burden (16, 26, 33). However, we found weaker evidence for an association between baseline disability and depression and, assuming the hierarchical framework is correct, this relationship appeared to be fully mediated through burden. This suggests that it is the high burden caused by greater disability that leads to depression in caregivers. However, it is possible the model does not reflect the likely more complex relationship between burden and depression well enough, as it could also be that caregivers experiencing depressive symptoms are inclined to report higher burden. There was evidence that the relationship the caregiver had to the person with schizophrenia was important. First degree relatives and spouses experienced more burden than more distantly related or unrelated caregivers, and this relationship appeared to be partly mediated by stigma. This suggests that the stigma associated with schizophrenia was felt more by people that are more closely related. We also found that caregivers who were children and parents of the person with schizophrenia experienced the most depression. Whilst a previous Ethiopian study did not find an association between type of relative and caregiver depression (14), a further study in Ethiopia did find other severe intergenerational impacts in families of people with severe mental conditions, such as an increased risk of food insecurity, lower school attendance and even an increased risk of death (34).

Strengths and Limitations
This study had a few strengths. One is the use of multiple sensitivity analyses, which confirmed the validity of the analyses. The use of longitudinal data for the exploratory analysis of factors influencing caregiver outcomes, in contrast to previous cross-sectional studies (35), made it less likely that associations with the outcomes are due to reverse causality.

This study also had a number of limitations. The missing data at the 6-month assessment might have introduced some selection bias, which might not have been fully mediated by including variables associated with missingness in the analysis. However, loss to follow-up was low at the 12-month assessment. Another issue was that the caregiver that attended the 6 month or 12-month assessment was not always the same person that was interviewed at baseline, though sensitivity analysis showed that omitting those caregivers did not change the results of the main analysis. The participants and caregivers were only followed up for 12 months, making it hard to say anything about the effect of the treatment in the longer term. There was also a lack of power to detect smaller changes for most of the outcomes and the effect modification analysis. There were also computational and reliability issues running the random effects logistic regression models, meaning ORs are based on GEE and thus represent population average rather than cluster-specific ORs (36). As population average ORs tend to be closer to 1, this might have under-estimated the true effects, or contributed to not detecting evidence for an association with binary outcomes.

There might be factors influencing caregiver outcomes that we were unaware of or did not have data for and thus failed to include in the conceptual framework. Using a p-value based threshold to include factors in the model also means that there is likely to be insufficient correction for many of the confounding factors, as some variables that did not meet the threshold might still have confounded the relationship between the outcomes and other variables in the model. This method was chosen as it allowed for modelling many different parameters on the four hierarchical levels, while preventing having too many parameters in the model (25).

Implications

Several factors suggest the individual with schizophrenia together with their family should be the unit for delivery of psychosocial support in LMICs. First, caregivers can play an essential role in the rehabilitation of people with schizophrenia particularly in LMICs like Ethiopia, where access to mental health services is limited and the burden of care falls almost entirely on caregivers. Second, caregivers also have the right to health and their needs should be considered. Addressing caregivers’ needs may mean they are better equipped to support their relative, including encouraging engagement with health services and interventions such as CBR. Whilst psychosocial interventions directly targeting family members of people with schizophrenia are less successful in improving distress, depression or burden, they are effective in improving knowledge and coping (37, 38). Yet in settings with limited mental health resources, stand-alone caregiver interventions may not be feasible. Finally, in rural Ethiopia, in common with other LMIC, family members have a prominent role in their relative’s care and treatment decisions (39). Involving caregivers in psychosocial interventions is therefore likely to increase acceptability and feasibility.

We therefore propose that assessment of caregiver needs, including mental health and burden, should be incorporated into psychosocial interventions for people with schizophrenia. Though directly addressing the mental health needs of caregivers may be beyond the scope of a psychosocial intervention targeting people with schizophrenia, referral pathways to available services should be in place. The planned trial qualitative evaluation will hopefully aid future initiatives by illuminating which aspects of CBR impacted on burden and where the gaps are: for example, whether the light touch support provided directly to caregivers was helpful, and why the family support groups did not gain traction. Workers delivering psychosocial interventions such as CBR should be mindful of the greater likelihood of burden and depression in caregivers of people with high levels of disability and should offer additional support and/or referrals as needed. Workers should also be aware of the greater likelihood of high burden and depression amongst children and spouses of people with schizophrenia. CBR workers should be aware of the time burden of CBR participation and fit sessions around participants and caregivers’ schedules, particularly if it is not possible for them to do the work simultaneously. The forthcoming qualitative evaluation will shed light on whether caregivers’ participation in earlier phases of CBR actually decreased their ability to work and whether the benefits of CBR were perceived as an acceptable pay off.

The importance of reducing stigma in mental health was recently highlighted by the Lancet commission on ending stigma and discrimination in mental health (40). This commission also highlighted the knowledge gap on strategies to reduce stigma in LMIC. Social contact anti-stigma approaches, co-produced and delivered with people with lived experience (41), could be a candidate area for development in future iterations of CBR for schizophrenia. Our findings suggest incorporating caregivers into stigma reduction efforts could be beneficial.

We previously confirmed the positive effect of CBR on two subdomains of caregiver burden, but the present study did not show such an effect on depression, stigma, employment or other aspects of burden. Caregivers of people with high levels of disability have the highest levels of need. As a result of CBR, caregivers may also reduce their own work due to caring, possibly due to the time commitment of participation. Future research should explore how caregivers’ needs can be best addressed within psychosocial interventions targeting people with schizophrenia.

95%-CI	95 percent confidence interval
aOR	Adjusted Odds Ratio
CBR	Community Based Rehabilitation
COPSI	COmmunity-based intervention for People with Schizophrenia and their caregivers in India
FIS	Family Interview Schedule
GEE	Generalised Estimating Equation
ICC	Intra-Cluster Correlation
IEQ	Involvement Evaluation Questionnaire
IQR	Inter Quartile Range
LMIC	Low and Middle Income Countries
LSHTM	London School of Hygiene and Tropical Medicine
MD	Mean Difference
mhGAP-IG	WHO mental health Gap Action Programme-Intervention Guide
N	Number
N/A	Not applicable
OR	Odds Ratio
PHQ-9	9 item Patient Health Questionnaire
PRIME	PRogramme for Improving Mental healthcarE
RISE	Community-based Rehabilitation Intervention for people with Schizophrenia in Ethiopia
SD	Standard Deviation
SMD	Standardized Mean Difference
SMI	Severe Mental Illness
WHO	World Health Organisation
WHODAS	World Health Organisation Disability Assessment Schedule
β	Beta coefficient for linear regression models

Ethics approval and consent to participate

The study was approved by the LSHTM Research Ethics Committee (reference 0735-2), the Addis Ababa University College of Health Sciences Institutional Review Board (reference 083/13/Psy), and the Ethiopian National Research Ethics Review Committee (reference 310/048/2015).

In the trial informed consent was taken from both the person with schizophrenia and the caregiver by a psychiatric nurse. In this process verbal information was given to the person with schizophrenia and the caregiver and an information sheet was provided that was elaborately discussed and explained. Time for questions was given and non-technical language was used. Sufficient time was allowed for participants to think about the information, with extra time if the person with schizophrenia was unwell or seemed to have cognitive difficulties. Although many caregivers and participants were illiterate, written information was still given to review with assistance at a later time. The nurse used clinical judgement to determine if the person with schizophrenia has capacity to provide informed consent. When the person with schizophrenia was deemed not to have capacity, consent was sought from the caregiver. Assent was then sought from the person with schizophrenia. If the person with schizophrenia regained capacity during the trial, they were asked for consent at that time.

Consent for publication

Not applicable

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests

Funding

LA was funded by a Wellcome Trust PhD Fellowship in International Health (grant number: 100142/Z/12/Z). The RISE project was part of the PRogramme for Improving Mental health care (PRIME), which was funded by the UK Department for International Development (DfID) for the benefit of LMIC (HRPC10). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of any of the funders. LD worked on this paper as part of an MSc program for which she received scholarships from the Prins Bernard Cultuurfonds and VSBfonds.

CH receives support through a National Institute for Health and Social Care Research (NIHR) RIGHT grant (NIHR200842) and an NIHR global health research group on homelessness and mental health in Africa (HOPE; NIHR134325). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health and Care Research or the Department of Health and Social Care, England. CH is also funded by the Wellcome Trust through grants 222154/Z20/Z (SCOPE) and 223615/Z/21/Z (PROMISE). HW receives support from an award jointly funded by the UK Medical Research Council (MRC) and the UK Foreign Commonwealth and Development Office (FCDO) under the MRC/FCDO Concordat agreement and is also part of the EDCTP2 programme supported by the European Union (MR/R010161/1).

Authors' contributions

LGD and LA drafted the report, which all the authors reviewed and approved. LA, AF, CH, MDS, GM, and HAW designed the trial. LA, AF, CH, MDS, and RB devised the intervention content and data collection instruments. LA, AF, and RB were responsible for trial conduct. LA and RB were responsible for database design and management. LGD and HAW did the statistical analyses. LA and HAW have verified the underlying data.

Acknowledgements

We would like to thank all the participants and caregivers that took part in the RISE trial, as well as the CBR workers, CBR supervisors and research staff.

Kebede D, Alem A, Shibre T, Negash A, Deyassa N, Beyero T, et al. Short-term symptomatic and functional outcomes of schizophrenia in Butajira, Ethiopia. Schizophrenia research. 2005;78(2-3):171-85.
Shibre T, Kebede D, Alem A, Negash A, Deyassa N, Fekadu A, et al. Schizophrenia: illness impact on family members in a traditional society--rural Ethiopia. Soc Psychiatry Psychiatr Epidemiol. 2003;38(1):27-34.
Laursen TM, Nordentoft M, Mortensen PB. Excess early mortality in schizophrenia. Annual review of clinical psychology. 2014;10:425-48.
Patel V. Universal Health Coverage for Schizophrenia: A Global Mental Health Priority. Schizophrenia bulletin. 2016;42(4):885-90.
Kakuma R, Minas H, van Ginneken N, Dal Poz MR, Desiraju K, Morris JE, et al. Human resources for mental health care: current situation and strategies for action. Lancet (London, England). 2011;378(9803):1654-63.
Asher L, Patel V, De Silva MJ. Community-based psychosocial interventions for people with schizophrenia in low and middle-income countries: systematic review and meta-analysis. BMC Psychiatry. 2017;17(1):355.
WHO Guidelines Approved by the Guidelines Review Committee. mhGAP Intervention Guide for Mental, Neurological and Substance Use Disorders in Non-Specialized Health Settings: Mental Health Gap Action Programme (mhGAP): Version 20. Geneva: World Health Organization Copyright © World Health Organization 2016.; 2016.
WHO Guidelines Approved by the Guidelines Review Committee. In: Khasnabis C, Heinicke Motsch K, Achu K, Al Jubah K, Brodtkorb S, Chervin P, et al., editors. Community-Based Rehabilitation: CBR Guidelines. Geneva: World Health Organization Copyright © World Health Organization 2010.; 2010.
Iemmi V, Gibson L, Blanchet K, Kumar KS, Rath S, Hartley S, et al. Community‐based rehabilitation for people with disabilities in low‐and middle‐income countries: A systematic review. Campbell Systematic Reviews. 2015;11(1):1-177.
Alem A, Kebede D, Fekadu A, Shibre T, Fekadu D, Beyero T, et al. Clinical course and outcome of schizophrenia in a predominantly treatment-naive cohort in rural Ethiopia. Schizophrenia bulletin. 2009;35(3):646-54.
Fekadu A, Thornicroft G. Global mental health: perspectives from Ethiopia. Global health action. 2014;7:25447.
Asher L, Hanlon C, Birhane R, Habtamu A, Eaton J, Weiss HA, et al. Community-based rehabilitation intervention for people with schizophrenia in Ethiopia (RISE): a 12 month mixed methods pilot study. BMC Psychiatry. 2018;18(1):250.
Fekadu A, Hanlon C, Medhin G, Alem A, Selamu M, Giorgis TW, et al. Development of a scalable mental healthcare plan for a rural district in Ethiopia. Br J Psychiatry. 2016;208 Suppl 56(Suppl 56):s4-s12.
Derajew H, Tolessa D, Feyissa GT, Addisu F, Soboka M. Prevalence of depression and its associated factors among primary caregivers of patients with severe mental illness in southwest, Ethiopia. BMC Psychiatry. 2017;17(1):88.
Asher L, Birhane R, Weiss HA, Medhin G, Selamu M, Patel V, et al. Community-based rehabilitation intervention for people with schizophrenia in Ethiopia (RISE): results of a 12-month cluster-randomised controlled trial. The Lancet Global health. 2022;10(4):e530-e42.
Shibre T, Medhin G, Teferra S, Wakwoya A, Berhanu E, Abdulahi A, et al. Predictors of carer-burden in schizophrenia: a five-year follow-up study in Butajira, Ethiopia. Ethiopian medical journal. 2012;50(2):125-33.
Asher L, De Silva M, Hanlon C, Weiss HA, Birhane R, Ejigu DA, et al. Community-based Rehabilitation Intervention for people with Schizophrenia in Ethiopia (RISE): study protocol for a cluster randomised controlled trial. Trials. 2016;17(1):299.
Hanlon C, Medhin G, Selamu M, Birhane R, Dewey M, Tirfessa K, et al. Impact of integrated district level mental health care on clinical and social outcomes of people with severe mental illness in rural Ethiopia: an intervention cohort study. Epidemiol Psychiatr Sci. 2019;29:e45.
Asher L, Fekadu A, Hanlon C, Mideksa G, Eaton J, Patel V, et al. Development of a Community-Based Rehabilitation Intervention for People with Schizophrenia in Ethiopia. PLoS One. 2015;10(11):e0143572-e.
Asher L, Birhane R, Teferra S, Milkias B, Worku B, Habtamu A, et al. "Like a doctor, like a brother": Achieving competence amongst lay health workers delivering community-based rehabilitation for people with schizophrenia in Ethiopia. PLoS One. 2021;16(2):e0246158.
Hanlon C, Medhin G, Selamu M, Breuer E, Worku B, Hailemariam M, et al. Validity of brief screening questionnaires to detect depression in primary care in Ethiopia. Journal of affective disorders. 2015;186:32-9.
StataCorp. Stata 16 Base Reference Manual. 2019. In: Stata 16 Base Reference Manual [Internet]. College Station, TX: Stata Press. Available from: https://www.stata.com/manuals/xtquadchk.pdf.
Steyerberg EW, Bossuyt PM, Lee KL. Clinical trials in acute myocardial infarction: should we adjust for baseline characteristics? American heart journal. 2000;139(5):745-51.
Hauck WW, Anderson S, Marcus SM. Should we adjust for covariates in nonlinear regression analyses of randomized trials? Controlled clinical trials. 1998;19(3):249-56.
Victora CG, Huttly SR, Fuchs SC, Olinto MT. The role of conceptual frameworks in epidemiological analysis: a hierarchical approach. International journal of epidemiology. 1997;26(1):224-7.
Ayalew M, Workicho A, Tesfaye E, Hailesilasie H, Abera M. Burden among caregivers of people with mental illness at Jimma University Medical Center, Southwest Ethiopia: a cross-sectional study. Ann Gen Psychiatry. 2019;18:10.
Ergetie T, Yohanes Z, Asrat B, Demeke W, Abate A, Tareke M. Perceived stigma among non-professional caregivers of people with severe mental illness, Bahir Dar, northwest Ethiopia. Ann Gen Psychiatry. 2018;17:42.
Shibre T, Negash A, Kullgren G, Kebede D, Alem A, Fekadu A, et al. Perception of stigma among family members of individuals with schizophrenia and major affective disorders in rural Ethiopia. Soc Psychiatry Psychiatr Epidemiol. 2001;36(6):299-303.
Teferra S, Hanlon C, Alem A, Jacobsson L, Shibre T. Khat chewing in persons with severe mental illness in Ethiopia: a qualitative study exploring perspectives of patients and caregivers. Transcult Psychiatry. 2011;48(4):455-72.
Luo X, Law SF, Wang X, Shi J, Zeng W, Ma X, et al. Effectiveness of an Assertive Community Treatment program for people with severe schizophrenia in mainland China - a 12-month randomized controlled trial. Psychol Med. 2019;49(6):969-79.
Sungur M, Soygür H, Güner P, Üstün B, Çetin I, Falloon IR. Identifying an optimal treatment for schizophrenia: a 2-year randomized controlled trial comparing integrated care to a high-quality routine treatment. International journal of psychiatry in clinical practice. 2011;15(2):118-27.
Chatterjee S, Naik S, John S, Dabholkar H, Balaji M, Koschorke M, et al. Effectiveness of a community-based intervention for people with schizophrenia and their caregivers in India (COPSI): a randomised controlled trial. The Lancet. 2014;383(9926):1385-94.
Peng M-M, Xing J, Tang X, Wu Q, Wei D, Ran M-S. Disease-Related Risk Factors for Caregiver Burden among Family Caregivers of Persons with Schizophrenia: A Systematic Review and Meta-Analysis. International Journal of Environmental Research and Public Health. 2022;19.
Fekadu W, Craig TKJ, Kebede D, Medhin G, Fekadu A. Multidimensional and intergenerational impact of Severe Mental Disorders. eClinicalMedicine. 2021;41.
Fekadu A. Living with the high burden of care: experience of caregivers of people with severe mental illness in rural Ethiopia. Ethiopian medical journal. 2020;58:115-23.
Neuhaus JM, Kalbfleisch JD, Hauck WW. A comparison of cluster-specific and population-averaged approaches for analyzing correlated binary data. International Statistical Review/Revue Internationale de Statistique. 1991:25-35.
Sin J, Norman I. Psychoeducational interventions for family members of people with schizophrenia: a mixed-method systematic review. The Journal of clinical psychiatry. 2013;74(12):e1145-62.
Napa W, Tungpunkom P, Pothimas N. Effectiveness of family interventions on psychological distress and expressed emotion in family members of individuals diagnosed with first-episode psychosis: a systematic review. JBI database of systematic reviews and implementation reports. 2017;15(4):1057-79.
Souraya S, Hanlon C, Asher L. Involvement of people with schizophrenia in decision-making in rural Ethiopia: a qualitative study. Global Health. 2018;14(1):85.
Thornicroft G, Sunkel C, Alikhon Aliev A, Baker S, Brohan E, el Chammay R, et al. The <em>Lancet</em> Commission on ending stigma and discrimination in mental health. The Lancet. 2022;400(10361):1438-80.
Gronholm PC, Bakolis I, Cherian AV, Davies K, Evans-Lacko S, Girma E, et al. Toward a multi-level strategy to reduce stigma in global mental health: overview protocol of the Indigo Partnership to develop and test interventions in low- and middle-income countries. International Journal of Mental Health Systems. 2023;17(1):2.

Table 1: Baseline characteristics of caregivers and caregiver outcomes at baseline

	Intervention	Control
Sex (N [%])*
Male	33 (41.8%)	44 (50.6%)
Female	42 (53.2%)	36 (41.4%)
Missing	4 (5.1%)	7 (8.1%)
Age group (N [%])*
<25	15 (19.0%)	14 (16.1%)
5-35	18 (22.8%)	15 (17.2%)
35>	42 (53.2%)	51 (58.6%)
Missing	4 (5.1%)	7 (8.1%)
Marital status (N [%])
Single	14 (17.7%)	12 (13.8%)
Married	49 (62.0%)	62 (71.3%)
Divorced or Widowed	16 (20.3%)	12 (13.8%)
Missing	0	1 (1.1%)
Education in years (N [%])*
0	46 (58.2%)	50 (57.5%)
1-8	17 (21.5%)	23 (26.4%)
9-15	12 (15.2%)	7 (8.1%)
Missing	4 (5.1%)	7 (8.1%)
Literacy (N [%]) *
Illiterate	32 (40.5%)	31 (35.6%)
Can read and write	14 (17.7%)	19 (21.8%)
Formal education	29 (36.7%)	30 (34.5%)
Missing	7 (8.1%)	4 (5.1%)
Relationship to person with schizophrenia (N [%])
Parent	18 (22.8%)	27 (31.0%)
Sibling	14 (17.7%)	22 (25.3%)
Child	21 (26.6%)	11 (12.6%)
Spouse	16 (20.3%)	18 (20.7%)
Other	10 (12.6%)	9 (10.4%)
Outcomes at baseline
Reduced work due to caring
No	15 (19.0%)	17 (19.5%)
Yes	64 (81.0%)	70 (80.5%)
Caregiver unemployment (N [%])
Employed	43 (54.4%)	55 (63.2%)
unemployed	36 (45.6%)	32 (36.8%)
Depression (PHQ>4) (N [%])
No	36 (45.6%)	57 (65.5%)
Yes	43 (54.4%)	30 (34.5%)
PHQ total (median [IQR])	5 (2-8)	3 (2-6)
IEQ
Urging (median [IQR])	14 (9-20)	16 (10-20)
Supervision (median [IQR])	9 (4-13)	8 (4-12)
Worrying (median [IQR])	13 (6-17)	11 (8-15)
Tension (median [IQR])	8 (4-11)	7 (4-11)
Total (median [IQR])	41 (26-53)	41 (29-50)
FIS
Total stigma (median [IQR])	10 (5-18)	8 (3-17)
Total impact (median [IQR])	4 (0-7)	3 (1-7)

*These were collected at PRIME baseline.

Table 2: Caregiver outcomes at 6 and 12 months

Outcome	Control N (%) or mean (SD)	Intervention N (%) or mean (SD)	Minimally adjusted mean difference or odds ratio (95%CI)^a	p- value	Final model mean difference or odds ratio (95% CI)^b	p-value	Standardized mean difference (95% CI)
6 months (N=112)
Reduced work due to caring	45 (75.0%)	44 (84.6%)	OR 1.92 (0.79,4.68)	0.15	OR 2.40 (1.06,5.45)	0.04	-
Caregiver unemployment	22 (36.7%)	28 (53.9%)	OR 2.16 (0.56,8.38)	0.26	OR 2.09 (0.55,7.95)^ijn	0.28	-
PHQ
Depression (>4)^e	21 (35.0%)	20 (38.5%)	OR 1.26 (0.46 ,3.47)	0.65	OR 1.58 (0.52,4.84)^iln	0.42	-
Total score	3.52 (2.91)	4.04 (3.51)	0.14 (-1.50 ,1.78)	0.87	0.61 (-0.87,2.09)^ln	0.42	0.19 (-0.18,0.56)
FIS
Total stigma	10.2 (8.84)	9.96 (9.17)	0.38(-3.46 ,4.21)	0.85	0.54 (-4.26,5.33)^ceijlqn	0.83	0.06 (-0.31,0.43)
Total impact	4.23 (4.01)	3.53 (3.63)	-1.01 (-2.95,0.94)	0.31	-0.69 (-2.79-1.41)^cejln	0.52	-0.18 (-0.55,0.19)
IEQ
Urging	14.68 (7.88)	13.46 (6.57)	-0.15 (-3.13,2.84)	0.92	-0.76(-4.09,2.58)^ehijln	0.65	-0.10 (-0.48,0.27)
Supervision	7.67(6.58)	7.54 (5.96)	-0.88 (-3.58,1.82)	0.52	-1.71 (-4.94,1.53)^ejln	0.30	-0.27 (-0.64,0.10)
Worrying	10.28 (6.86)	8.68 (5.86)	-1.12 (-3.60,1.35)	0.37	-1.42 (-3.94,1.10)^el	0.27	-0.22 (-0.59,0.15)
Tension	6.3 (5.64)	6.96 (6.84)	0.98 (-1.83,3.79)	0.49	1.07 (-1.83,3.97)^eln	0.47	0.17 (-0.20,0.54)
Total	36.47 (19.84)	34.22 (18.41)	-0.33 (-9.29, 8.62)	0.94	-1.67 (-11.65,8.31)^ecijm	0.74	-0.09 (-0.46,0.28)
12 months (N=149)
Caregiver unemployment	43 (56.6%)	40 (54.8%)	OR 1.09 (0.60,1.97)	0.78	OR 0.72 (0.35,1.50)^cdefk	0.38	-
FIS
Total stigma	10.89 (10.15)	9.05 (9.30)	-2.52 (-6.08,1.03)	0.16	-2.57 (-6.28,1.13)^c	0.17	-0.27 (-0.60,0.05)
Total impact	4.21 (3.69)	5.25 (3.93)	1.17 (-0.31,2.65)	0.12	0.98 (-0.55-2.52)^cdfj	0.21	0.26 (-0.06,0.58)

Socio-economic status, urban vs rural residence, caregiver sex, illness duration and travel time to the facility were coded to have missingness as a category.
^aAdjusted for sub-district (cluster) as random effect and health centre and baseline score of outcome as fixed effects
^b All are adjusted for sub-district (cluster) as random effect and health centre, baseline score of outcome, and baseline WHODAS score as fixed effects in addition to the variables that were found to have a confounding effect.
^cAdjusted for socioeconomic status
^dAdjusted for place of residence
^eAdjusted for baseline depression
^fAdjusted for disease course prior to study
^gAdjusted for baseline total IEQ score at baseline
^hAdjusted for the patients sex
ⁱAdjusted for the patients employment status
^jAdjusted for social support
^kAdjusted for the patients age
^lAdjusted for the caregivers sex
^m Adjusted for the caregivers employment
ⁿAdjusted for travel time to the facility

Table 3: Hierarchical linear regression models for burden measured by total IEQ score at 12 months

	1 (N=147)		1b (N=149)		2 (N=149)		2b (N=149)		3 (N=138)		3b (N=149)
	β (95%-CI)	p-value	β (95%-CI)	p-value	β (95%-CI)	p-value	β (95%-CI)	p-value	β (95%-CI)	p-value	β (95%-CI)	p-value
Level 1
WHODAS	0.26 (0.14-0.38)	<0.001	0.29 (0.18-0.40)	<0.001	0.26 (0.15-0.37)	<0.001	0.26 (0.15-0.37)	<0.001	0.26 (0.14-0.38)	<0.001	0.26 (0.14-0.37)	<0.001
Disorganisation		0.15
Absent	0
Present	4.93 (-1.75 -11.61
Relationship		0.003		0.01		0.03		0.01		0.05		0.01
1^st degree/spouse	0		0		0		0		0		0
Other relationship	-13.78 (-23.02- -4.54)		-13.97 (-22.31 - -3.82)		-10.26 (-19.74 - -0.77)		-11.92 (-21.12- -2.71)		-9.61 (-19.29-0.05)		-11.89 (-21.19- -2.59)
Works away from home		0.02		0.03		0.05		0.05		0.24
No	0		0		0		0		0
Yes	-6.23 (-11.59- - 0.87)		-6.03 (-11.36- -0.70)		-5.33 (-10.63 - -0.03)		-5.31 (-10.63-0.00)		-3.45 (-9.21-2.30)
Level 2
Stigma					0.29 (-0.02 – 0.59)	0.06	0.31 (0.01-0.61)	0.04	0.30 (-0.01-0.61)	0.06	0.35 (0.05-0.65)	0.02
Time spent						0.17
<32 hours/week					0
>32 hours/week					4.84 (-2.14-11.82)
Level 3
Caregiver years of education										0.16
0									0
1-8									-2.06 (-8.62-4.50)
9-15									-9.06 (-18.34-0.22)

All models are adjusted for sub-district (cluster) as random effect and health centre as a fixed effect. All other covariates in the model are measured at baseline. The model number represents the level of variables included, and the b-versions are those with only the variables independently associated with the outcome (p<0.10)

Table 4: Hierarchical logistic regression models for depression measured by PHQ>4 at 12 months

	1		1b		2		2b		3		3b		4		4b
	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value	OR (95%-CI)	p-value
Level 1
WHODAS	1.01 (1.00-1.03)	0.05	1.01 (1.00-1.03)	0.05	1.01 (1.00-1.03)	0.10	1.01 (1.00-1.03)	0.10	1.01 (1.00-1.03)	0.06	1.01 (1.00-1.03)	0.10	1.00 (0.98-1.02)	0.89
Sex		0.56
Male	1
Female	1.25 (0.60-2.60)
Place of residence		0.02		0.02		0.03		0.03		0.02		0.03		0.02		0.02
Urban	1		1		1		1		1		1		1		1
Rural	8.90 (1.48-53.65)		8.61 (1.41-52.47)		7.89 (1.24-50.08)		8.02 (1.25-51.41)		9.99 (1.52-65.69)		8.02 (1.25-51.41)		8.59 (1.32-56.02)		8.56 (1.33-55.08)
Relationship		0.01		0.01		0.01		0.01		0.01		0.01		0.01		0.01
Parent	1		1		1		1		1		1		1		1
Sibling	0.13 (0.03-0.58)		0.12 (0.03-0.55)		0.14 (0.03-0.71)		0.12 (0.02-0.61)		0.11 (0.02-0.57)		0.12 (0.02-0.61)		0.13 (0.02-0.67)		0.12 (0.02-0.67)
Child	1.58 (0.54-4.64)		1.64 (0.57-4.68)		1.84 (0.61-5.52)		1.69 (0.59-4.86)		1.71 (0.57-5.14)		1.69 (0.59-4.86)		1.94 (0.65-5.75)		1.95 (0.66-5.78)
Spouse	0.52 (0.16-1.71)		0.52 (0.16-1.72)		0.52 (0.16-1.73)		0.54 (0.17-1.74)		0.58 (0.16-2.06)		0.54 (0.17-1.74)		0.51 (0.15-1.75)		0.52 (0.15-1.78)
Other	0.14 (0.02-0.83)		0.12 (0.02-0.77)		0.20 (0.03-1.24)		0.15 (0.02-1.01)		0.13 (0.02-0.91)		0.15 (0.02-1.01)		0.24 (0.04-1.48)		0.24 (0.04-1.42)
Level 2
Stigma					1.05 (1.00-1.09)	0.04	1.05 (1.00-1.09)	0.03	1.05 (1.01-1.10)	0.02	1.05 (1.00-1.09)	0.03	1.04 (1.00-1.09)	0.06	1.04 (1.00-1.08)	0.06
Time						0.23
<32 hours/week					1
>32 hours/week					1.85 (0.67-5.07)
Level 3
Social support										0.32
Strong/intermediate									1
Poor									0.65 (0.27-1.53)
Level 4
Burden													1.03 (1.00-1.06)	0.02	1.03 (1.01-1.05)	0.01

All models are adjusted for sub-district (cluster) as random effect and health centre. All other covariates in the model are measured at baseline. N=137 for all models. The model number represents the level of variables included, and the b-versions are those with only the variables independently associated with the outcome (p<0.10)

No competing interests reported.

Download PDF

Editorial decision: Revision requested
02 May, 2024
Reviews received at journal
13 Apr, 2024
Reviews received at journal
05 Apr, 2024
Reviewers agreed at journal
05 Apr, 2024
Reviewers agreed at journal
30 Mar, 2024
Reviewers agreed at journal
25 Mar, 2024
Reviewers invited by journal
15 Mar, 2024
Editor assigned by journal
14 Mar, 2024
Editor invited by journal
12 Dec, 2023
Submission checks completed at journal
12 Dec, 2023
First submitted to journal
07 Nov, 2023

You are reading this latest preprint version

Effects of community-based rehabilitation on caregivers of people with schizophrenia in Ethiopia in the RISE trial

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Trial registration

Figures

Background

Methods

Study design and participants

Procedures

Instruments and measurements

Sample size

Statistical analysis

Results

Discussion

Conclusions

Abbreviations

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1