Graphic Art

11 pages
55 views

The effect of community health worker–led education on women’s health and treatment– seeking: A cluster randomised trial and nested process evaluation in Gujarat, India

Please download to get full document.

View again

of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Description
The effect of community health worker–led education on women's health and treatment– seeking: A cluster randomised trial and nested process evaluation in Gujarat, India Background A community–based health insurance scheme operated by the
Transcript
      V    I    E    W    P    O    I    N    T    S       P    A    P    E    R    S  journal of health global Sapna Desai 1,2 , Ajay Mahal 3 , Tara Sinha 2 , Joanna Schellenberg 1 , Simon Cousens 1 1  Faculty of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK  2  Self Employed Women’s Association,  Ahmedabad, Gujarat, India 3  Nossal Institute for Global Health, University of Melbourne, Melbourne, Australia Correspondence to: Sapna Desai Faculty of Epidemiology and Population Health London School of Hygiene and Tropical Medicine London, UK sapna.i.desai@gmail.com The effect of community health worker–led education on women’s health and treatment–seeking: A cluster randomised trial and nested process evaluation in Gujarat, India Background  A community–based health insurance scheme operated by the Self–Employed Women’s Association in Gujarat, India reported that the leading reasons for inpatient hospitalisation claims by its mem-bers were diarrhoea, fever and hysterectomy – the latter at the average age of 37. This claims pattern raised concern regarding potentially un-necessary hospitalisation amongst low–income women. Methods  A cluster randomised trial and mixed methods process eval-uation were designed to evaluate whether and how a community health worker–led education intervention amongst insured and uninsured adult women could reduce insurance claims, as well as hospitalisation and morbidity, related to diarrhoea, fever and hysterectomy. The 18–month intervention consisted of health workers providing preventive care information to women in a group setting in 14 randomly selected clusters, while health workers continued with regular activities in 14 comparison clusters. Claims data were collected from an administra-tive database, and four household surveys were conducted amongst a cohort of 1934 randomly selected adult women. Results  30% of insured women and 18% of uninsured women report-ed attending sessions. There was no evidence of an intervention effect on the primary outcome, insurance claims (risk ratio (RR) = 1.03; 95% condence interval (CI) 0.81, 1.30) or secondary outcomes amongst insured and uninsured women, hospitalisation (RR = 1.05; 95% CI 0.58, 1.90) and morbidity (RR = 1.09; 95% CI 0.87, 1.38) related to the three conditions. The process evaluation suggested that participants retained knowledge from the sessions, but barriers to behaviour change were not overcome. Conclusions  We detected no evidence of an effect of this health work-er–led intervention to decrease claims, hospitalisation and morbidity related to diarrhoea, fever and hysterectomy. Strategies that capitalise on health workers’ role in the community and knowledge, as well as those that address the social determinants of diarrhoea, fever and the frequency of hysterectomy – such as water and sanitation infrastruc-ture and access to primary gynaecological care – emerged as areas to strengthen future interventions. www.jogh.org  • doi: 10.7189/jogh.07.020404   1   December 2017 • Vol. 7 No. 2 • 020404 Since the Alma–Ata declaration, community health workers (CHWs), also known as lay health workers, have been promoted as a key component of primary health care strategies aimed at women and children [1]. CHWs      V    I    E    W    P    O    I    N    T    S       P    A    P    E    R    S have also been shown to be uniquely positioned to inuence behavior change, through their use of in-digenous knowledge and ability to communicate with empathy and locally appropriate language [2]. Ev-idence synthesised through meta–analyses, qualitative syntheses and disease–focused reviews thus far suggests that CHWs have the potential to improve knowledge, behaviour and health outcomes [3–8]. A 2010 Cochrane systematic review and meta–analysis of interventions involving lay health workers indi-cated moderate evidence of their potential to improve immunisation coverage, breastfeeding and adher-ence to tuberculosis treatment, primarily through one–to–one visits and linking women to health systems [4]. Evidence, albeit limited, also suggests that CHW–led education delivered to women in a group set-ting can improve knowledge and preventive behavior [9–16]. This paper reports on the ndings of a clus- ter randomised trial and nested process evaluation of a CHW–led group health education intervention to improve women’s health and treatment–seeking behaviour in a low–income setting in Gujarat, India. Study setting  Gujarat, though one of India’s wealthier states, performs close to national averages with regards to many health indicators. The last (2015–6) National Family Health Survey reported an infant mortality rate of 34/1000 live births and that only one–half (50.4%) of children between 12–23 months were fully im-munised [17]. Utilisation of health services in Gujarat, as in most of India, is largely nanced by individ-ual households. Outpatient and inpatient care are predominantly sought in the private sector [18]. In 2009, Gujarat initiated roll–out of Rashtriya Swasthya Bima Yojana (RSBY), a government–nanced health insurance scheme that provides hospitalisation coverage up to Rs 30 000 (US$ 442, 12.19.2016) for fam-ilies identied to be below the poverty line [19]. In 2011, Gujarat recruited close to 30 000 village health workers (known as Accredited Social Health Activists, or ASHAs), one per 1000 population to cover its 18 539 villages [20].The intervention was designed with the Self–Employed Women’s Association (SEWA), a trade union of over 1.5 million women workers in India’s informal economy, whose members typically have insecure employment and limited access to social protection. SEWA operates a community health worker–led health program and insurance scheme for its members, VimoSEWA, that provides up to Rs.5000 (74 USD, 19 December 2016) coverage for inpatient hospitalisation that exceeds 24 hours in exchange for annual premium payments by members. Intervention  A 2009 analysis of 12 027 VimoSEWA hospitalisation claims reported that two of the leading reasons for inpatient hospitalisation amongst adult women were diarrhea and fever, the latter considered primarily related to malaria [21]. The third leading reason for insurance claims was hysterectomy, at a relatively low average age of 37. VimoSEWA was surprised by the high proportion of hospitalisation for diarrhoea and fever – seemingly common, preventable ailments. The frequency and age at hysterectomy suggested that some procedures may not have been medically indicated and were thus avoidable. Given that diarrhoea, fever and hysterectomy comprised over 40% of VimoSEWA’s claims, SEWA aimed to design a scalable intervention to reduce claims, hospitalisation and morbidity related to the three conditions. If effective, the intervention would protect members from unnecessary hospitalisation as well as improve VimoSEWA’s nancial sustainability.The aim of the intervention was to (i) raise awareness on prevention and immediate treatment for ma-laria–related fever and diarrhea and (ii) improve knowledge of hysterectomy and its side effects, in order to reduce medically unnecessary procedures. The intervention focused on group health education ses-sions implemented by its CHW team; this approach was viable with respect to the nancial and human resources available. Operationally, SEWA dened health education as a tool to improve knowledge and change women’s attitudes and behavior through information, dissemination and discussion. Further, since SEWA’s CHWs were seasoned local leaders and activists, group education sessions could potentially en-gage women in community action. At the time of the intervention, SEWA CHWs conducted limited group health education programs, none of which addressed diarrhoea, fever/malaria or hysterectomy. Both in-tervention and control areas were exposed to information through government health programs, includ-ing ASHA home visits to mothers and children and limited media messaging. Messages included infor-mation on malaria and diarrhoea and did not address gynaecological ailments. However, since ASHAs were neither trained nor incentivised to conduct health education, SEWA felt a group–based intervention could ll an important gap in existing services. SEWA CHWs in intervention areas implemented three to ve group health education sessions monthly with adult women over an 18–month period, while com-parison area CHWs continued with regular activities ( Table 1 ). Desai et al. December 2017 • Vol. 7 No. 2 • 020404   2   www.jogh.org  • doi: 10.7189/jogh.07.020404      V    I    E    W    P    O    I    N    T    S       P    A    P    E    R    S METHODS  As the intervention was implemented at the CHW level, a cluster randomised trial was designed to eval-uate the effect of the intervention on three outcomes: claims rates (primary outcome), hospitalisation and morbidity related to diarrhoea, fever and hysterectomy. Clusters were dened as the discrete geographi-cal areas served by one CHW. The number of clusters included in the trial, 28 in total, was determined by the number of CHWs available in areas where VimoSEWA operates. The intervention was implement-ed in 14 randomly–selected CHW catchment areas of Ahmedabad city and rural areas of Ahmedabad district, with 14 comparison clusters. Randomisation was stratied by urban and rural location, as urban rates of claim submission had been observed to be higher than in rural areas in two previous studies at  VimoSEWA [21,22]. Treatment allocation was assigned through randomly generated numbers and an- nounced in a CHW meeting. Data collectors were not informed of cluster allocations. Claims Reduction in claims submission was measured by utilising all VimoSEWA insurance claims submitted in the intervention and control areas over the intervention period, as recorded in VimoSEWA’s claims database.  VimoSEWA considered the minimum worthwhile intervention effect to be a 30–40% reduction in claims for the three conditions. The cooperative had moved towards a sustainable model without external funding support; a smaller reduction in claims would not have justied funding a health intervention from its op-erational costs. The between–cluster coefcient of variation ( k ) was estimated to be 0.28 using data on claims submission rates in 2008–2009 [23]. The study was estimated to have 77% power ( P  < 0.05, 2–sided test) to detect a 40% reduction in insurance claims for diarrhea, fever and hysterectomy. Hospitalisation and morbidity Data on hospitalisation and morbidity rates related to the three conditions were collected through house-hold surveys. Both insured and uninsured women were included in the household survey to enable the investigation of predictors of insurance coverage and to investigate whether the effect of the intervention varied with insurance status. A sample size of 35 uninsured and 35 insured households per cluster was chosen – a total of 1960 households across 28 clusters. Household listings of insured women were pro-vided by VimoSEWA. A listing of uninsured households was compiled by following CHWs on daily rounds. Households were randomly selected through computer generated numbering. A baseline survey was conducted from January to March 2010, followed by three survey rounds at six–month intervals fol-lowing implementation of the intervention. An adult woman was selected for interview in each house-hold: the same primary VimoSEWA policy holder or SEWA member in uninsured households was inter-viewed at each round. A total of 980 uninsured and 954 uninsured adult women were surveyed at baseline. Survey data were double–entered into a Microsoft Access database. A supervisor observed a ran-dom sub–set of interviews and checked each survey form manually before data entry. Attrition increased at each round, primarily due to demolition of slum pockets in Ahmedabad city and rural pre–monsoon seasonal migration: a total of 1616 households were surveyed in the nal round ( Figure 1 ). Community health worker–led education on women’s health and treatment–seeking in India www.jogh.org  • doi: 10.7189/jogh.07.020404   3   December 2017 • Vol. 7 No. 2 • 020404 Table 1. CHW activities in comparison and intervention areas A ctivity i ntervention c ompArison Home visits and group education on common illnesses (excluding diarrhoea, malaria and hysterectomy) × × Accompanied referral to health services × ×Medicine sales and insurance promotion × ×Linkages with government providers × × Activate Village Health and Sanitation Committees × ×Group education sessions on hysterectomy with lm viewings ×Communication tools/handouts on hysterectomy ×Group education on diarrhoea with ORS demonstrations ×Group education on fever/malaria with interactive games × Wall paintings on diarrhoea and malaria ×Education sessions on sanitation linkages and programs ×Monthly refresher training for CHWs ×ORS – oral rehydration salts, CHW – community health worker      V    I    E    W    P    O    I    N    T    S       P    A    P    E    R    S Statistical methods  Analysis was by intention to treat. In the initial analysis, women’s insurance status at baseline was used to dene the insured and uninsured groups. A Poisson regression model with cluster–level random ef-fects to account for between–cluster variation was tted to estimate the effect of the intervention on claims rates for the three conditions [24]. Effect estimates were adjusted for rural–urban location and cluster–level baseline claims rates. Likelihood ratio tests comparing models with and without the intervention effect were performed to obtain p values. Analyses of the effect of the intervention on hospitalisation and morbidity rates for the three conditions were conducted using similar methods, adjusting for survey round, insurance status, rural/urban location and cluster–level baseline rates. Effect modication by rural/urban location was examined for all three outcomes and by insurance status for hospitalisation and morbidity. Lastly, a process evaluation collected quantitative and qualitative data at each step in the hypothesised causal chain ( Figure 2 ). Ethics and consent Representatives of the clusters, drawn from SEWA’s membership–based health cooperative, provided ap-proval prior to randomisation. A board constituted by SEWA’s Health Cooperative Executive Committee and the Ethics Committee of the London School of Hygiene and Tropical Medicine granted ethical ap-proval for the intervention, evaluation and qualitative research. In light of low literacy levels in the study area, all households provided oral informed consent to participate in the survey, as approved by the local ethics board. The study was registered as ISRCTN21290274. Reporting follows the CONSORT guidelines and extension for cluster randomised trials. RESULTS Baseline comparability Based on the demographic characteristics recorded in VimoSEWA’s administrative databases, intervention and control arms were generally balanced, with the exception of differences in the proportions of agricul- Desai et al. Figure 1. Cluster and survey participation. Figure 2. Intervention casual chain. December 2017 • Vol. 7 No. 2 • 020404   4   www.jogh.org  • doi: 10.7189/jogh.07.020404      V    I    E    W    P    O    I    N    T    S       P    A    P    E    R    S tural and home–based workers ( Table 2 ). Claims rates based on individual–level data and cluster summaries were similar (5.4 and 5.3 per 100 person–years). The between clus-ter coefcients of variation ( k ) in claims rates, estimated using baseline data, were 0.46 (ur-ban) and 0.66 (rural).Similarly, household survey data indicated that baseline demographic characteristics were largely balanced across intervention and comparison arms, including baseline rates of reported morbidity and hospitalisation ( Ta-ble 3 ). However, latrine ownership was high-er among intervention households than con-trol households. Amongst insured women, a higher proportion had attended school and a higher proportion lived in a concrete home in the intervention arm. The between cluster coefcients of variation ( k ), estimated using baseline hospitalisation data, were 0.49 (urban) and 0.56 (rural). At baseline, the three focus conditions – fever/malaria, diarrheal illness and hysterectomy – comprised approximately half of all hospitalisations in the preceding 6 months amongst both insured and uninsured women (48 of 99 hospitalisations). Hysterectomy was the most common reason for hospitalisation. Hospitalisation rates among insured women were approximately double those among the uninsured. Community health worker–led education on women’s health and treatment–seeking in India Table 2. Overview of baseline demographic characteristics, VimoSEWA membership database i ntervention  ( n  = 1839)c ompArison  ( n  = 1719) Demographic variables:Mean age37.737.1% married83.885.3% widowed10.19.4Occupation:% agricultural34.844.7% service37.336.5% home–based17.710.7% unemployed10.18.0 Baseline claims rate  (/100 person–years)5.75.0 Table 3. Baseline demographic characteristics, by insurance status and treatment arm U ninsUred  ( n  = 980)i nsUred  ( n  = 954) Selected variablesIntervention (n = 490)Comparison (n = 490)Intervention (n = 4698)Comparison (n = 485) Mean age in years37.035.939.839.1Mean household size5.85.86.05.8% concrete home26.124.935.124.1% with toilet60.051.863.146.1% individual drinking tap76.775.576.773.3Mean annual income (INR)82 70780 81282 74776 637% never attended school50.253.9950.162.7% respondents reported illness, past 30 d13.512.015.919.2% respondents reported hospitalization, past 6 mo3.12.97.07.7INR – Indian rupee Intervention coverage In the end line survey, 30.3% of insured women and 18.2% of unin-sured women in intervention clusters reported attending at least one session on diarrhoea, malaria or hysterectomy in the past year ( Table 4 ). A lower proportion of women reported attending hysterectomy sessions compared to diarrhoea and malaria. Of 203 surveyed women who reported participating in a session, women who were insured, currently working and had attended at least primary school were more likely to attend. Intervention effect on claims, hospitalisation and morbidity During the 18–month intervention period, 3340 women residents in the study area were insured at some point, contributing 1436 person–years in the intervention arm and 1227 person–years in the comparison arm. These women submitted 140 claims for the three target conditions over the study period, with a slightly higher claims rate (5.5 per 100–person years) in the intervention arm, compared to 5.0 in com-parison clusters. The estimated rate ratio, adjusted for location and cluster–level baseline claims rate was 1.03 (95% CI: 0.81–1.30, P  = 0.81) ( Table 5 ). There was no evidence that the effect of the intervention differed between rural and urban areas ( P  = 0.84). Table 4. Intervention outreach by insurance status (% women surveyed intervention areas, n = 833) m AlAriA d iArrhoeA h ysterectomy A ny   session Insured23.225.013.230.3Uninsured13.614.16.318.2 www.jogh.org  • doi: 10.7189/jogh.07.020404   5   December 2017 • Vol. 7 No. 2 • 020404
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x