Assessing the impact of a healthcare-inclusive graduation program on health and livelihoods in Bangladesh

The UTTARAN program is distinguished by its unique focus on promoting women's empowerment within marginalized communities. This is achieved through two intervention streams: (1) the livelihood intervention stream, which transfers productive assets, provides financial literacy training, link beneficiaries to different social safety networks, and matches savings; (2) the health intervention stream, which provides door-to-door healthcare services with telemedicine support and a strong referral linkage with formal health facilities, and funds healthcare.

The details of livelihood and health interventions, provided by UTTARAN are as follows:
1) Livelihood Interventions
The livelihood interventions target one adult female member from each household. The bottom 250 households (ranked based on their Poverty Probability Index (PPI) scores) of a branch are selected as beneficiaries. However, the households are classified into three groups based on the extent of livelihood supports they receive. Group 1 includes the first 10 poorest households, group 2 includes the next 75 poorest households, and group 3 includes the remaining 165 poorest households of the 250-list. The following benefits are received by the beneficiaries under UTTARAN's livelihood intervention:

(a) Social Inclusion and Social Safety Net Linkage
For each branch, a Community Development Committee (CDC) is formed with 7-9 locally influential persons and the Program Officer (PO). The CDC members and beneficiaries hold a meeting every month with an aim to safeguard beneficiaries’ assets and create an inclusive social network. The CDC also assists the listed households linking up with different social safety net benefits provided by the government (e.g., allowance for pregnant women, disability benefits, widow allowance, allowance for homeless). Beneficiaries from all three groups receive these livelihood supports.

(b) Microbusiness Support Package and Enterprise Management Training
The program selects an adult female from each household as their primary beneficiary and assesses her microbusiness preference; commonly chosen options include livestock rearing, agriculture, tea-stalls, handicrafts etc. First, the beneficiaries receive an enterprise management training tailored to their chosen microbusiness type. Following the training the program decides to transfer assets to the beneficiaries based on their preference and performance in the training. The support package differs based on poverty status. If belongs to group 1, the beneficiary receives an asset worth 10 thousand Bangladeshi taka as in-kind grant. An adult female from group 2 receives an asset worth 20 thousand taka, of which 10 thousand is given as in-kind grant, and the remaining 10 thousand is given as a service charge free loan. The load is payable within 18 Months. Group 3 females also receive an asset worth 20 thousand taka; however, the entire amount is given as a service charge free loan, payable within 18 Months.

(c) Bi-weekly Follow-up and Financial Literacy Training
The program visits the beneficiaries twice a month to follow up with their progress. Moreover, to support beneficiaries developing their microbusiness, the program provides them with hands-on financial literacy training as per their need. Beneficiaries from all three groups receive these livelihood supports.

(d) Matched Savings
To encourage savings, the UTTARAN program helps its beneficiaries open a microbusiness savings account with SAJIDA Foundation’s microfinance program. The amount they save is matched by the program, with a limit of 100 Bangladeshi taka per month. This support is continued for 18 months for all three beneficiary groups.


2) Health Intervention
UTTARAN’s health intervention covers all household members of program beneficiaries. The services are provided both at the community and facility levels through four distinct cadres of healthcare providers: SAJIDA Shastho Bondhu (SSB), Program Officer (PO) (Health), Telemedicine Medical Officer, and Sector Specialist. Beneficiaries of UTTARAN receive the following healthcare services:

(a) Health profiling and tracking
The SSBs are responsible for health profiling and tracking within communities. They maintain regular contact with couples of reproductive ages, enquire about their contraceptive use and identify suspected pregnancies, which are reported to the PO (Health). The SSBs conduct monthly follow-ups with each pregnant woman through door-to-door visits, ensuring timely referrals for ANC and PNC services when necessary. Beyond pregnancy care, SSBs also conduct regular health check-ups, focusing on the general and reproductive health of women and adolescents, as well as providing care for newborns and children. Additionally, they measure blood pressure and glucose levels to detect potential cases of diabetes and hypertension at the community level, referring these cases to tele-health doctors or other appropriate referral facilities for further management.

(b) Awareness Raising Campaigns
Awareness raising campaign consists of three level of initiatives, i.e., household level door-to-door awareness by SSBs, health forum facilitated by PO (Health), and popular theatre for mass awareness. SSBs provide daily door-to-door health awareness at household level. Health forum is conducted twice a week by PO (Health) through yard meetings on various health issues. And popular theatre throughout the branches is organized once in a year for creating mass awareness. In health forum, PO (Health) use flipchart to raise awareness of nutrition, basic health and hygiene, RMNCAH (reproductive, maternal, newborn, child, and adolescent health), NCD (non-communicable disease) issues, and the services available to address them.

(c) Static Clinics
PO (Health) operated static clinics are organized five times a week, where quality primary health care, MNCH, reproductive health services are provided. These static clinics cover all the 250 beneficiary households of a branch. All the beneficiaries and their household members are eligible to receive various free healthcare services from the static clinics, including medications, telemedicine consultations, referrals to other healthcare facilities, as well as outpatient services such as consultations, prescriptions, and general health checkups.
Health services at the static clinics are provided primarily by the PO (Health) with support from SSBs to maintain queue, check patient’s vitals and distribute medicines. The static clinics provide the following services – general illnesses, pregnancy care (ANC and PNC), newborn and child care, adolescent health care, elderly care, NCD care (hypertension and diabetes), diagnostic strip tests (for diabetes), etc. Essential medicines by the PO (Health) are also distributed from the static clinics. Patients who cannot be treated at the static clinics are referred first to the telemedicine service, and then to Govt. hospital or nearby SF empanelled health facilities based on need.

(d) Telemedicine or tele-consultation service
Telemedicine is the first point of referral from the static clinic. The teleconsultation service is provided during the static clinics with a doctor appointed through SAJIDA Foundation. During the Static Clinic, patients are first queued for the telemedicine service. The SSB then checks patient’s vitals such as body temperature, blood pressure, saturation and pulse rate. The PO (Health) aids the communication between the patient and the telemedicine doctor at the other end of the service. Based on the symptoms and complaints, the telemedicine doctor either prescribes medicine, advises for further diagnostic tests or refers the patient to a nearby SAJIDA enlisted hospital or government healthcare centre. The telemedicine service is available free of cost for all registered beneficiaries of the program.

(e) Health financing through health card
The health financing initiative aims to enhance healthcare access for extremely poor households. Each registered extreme poor household is eligible for up to BDT 25,000 annually, benefiting all family members within the 250 registered households in each branch. This scheme covers inpatient and medication costs for individuals not covered by other health financing models. The BDT 25,000 allocation for each beneficiary household is divided into two main services: (a) BDT 16,000 for maternal delivery services (including spontaneous vaginal delivery at BDT 4,000 and Caesarean Section at BDT 12,000), and (b) BDT 9,000 for outpatient services for any household member (covering diagnostic tests at BDT 5,000 and medicines at BDT 4,000). In households without pregnant women, the BDT 16,000 allocation can be used for inpatient service costs.

Health services under this scheme are provided by healthcare centres pre-selected by SAJIDA Foundation upon referral by telemedicine doctor. This is a co-payment scheme where 90% of the total cost (until max. limit) is borne by the program and 10% of the payment is made by the patient.

2023-07-16

2025-05-04

To estimate the impact of UTTARAN on livelihood and health of the targeted population, we identified a set of indicators as primary outcomes, including- poverty status; income; consumption; and healthcare-seeking behavior.

> Poverty status
Poverty status was assessed using the Poverty Probability Index (PPI). The data collected through the PPI questionnaire will enable us to estimate the likelihood of participants falling below the expenditure thresholds of $1.90 and $3.20 per day per capita (2011 PPP). These estimations will be made at both the household level and the study arm level.

> Income
Respondents were asked about their household income, its source(s), and the frequency or timeline within which they earn the stated amount. These data will allow us to estimate their monthly household income. Using their household size data, we will also estimate respondents’ per capita per day income. Moreover, since UTTARAN expects to provide the beneficiaries with an alternative source of income, we will also look at the number of income sources contributing to the household earnings.

> Consumption
Using the World Food Programme’s food consumption scale, participants were asked about the frequency at which different food groups were consumed by their household in the week prior to the survey. The food groups include cereals and tubers, potatoes and sweet potatoes, pulses, vegetables, fruits, meat and fish, dairy products, sugar and oil. Based on their food consumption score, household food consumption was classified into poor, borderline, and acceptable.

> Healthcare seeking behaviour
We assessed respondents’ healthcare seeking behaviour related to NCD (specifically for hypertension and diabetes), acute and chronic illness by collecting information on medicine intake and seeking treatment from formal (i.e., MBBS doctors, SACMO, registered nurse, SAJIDA Shastho Bondhu, etcetera) and informal (i.e., kabiraj, local drug seller, village doctors, etcetera) healthcare providers. In addition, we assessed health seeking behaviour of pregnant women and mothers by asking them their place of delivery and number of antenatal care visits.

Food security; wealth and assets; financial inclusion, formal savings and access to social protection; women's empowerment; reduced morbidity and sickness-related absenteeism; knowledge on non-communicable diseases (NCDs), and out-of-pocket health expenditure

> Food security
Food security status was assessed using the Household Food Insecurity Access Scale (HFIAS) which asked respondents whether their households experienced food insecurity-related incidents in the past four weeks. The incidents are broadly categorized as: anxiety and uncertainty about food supply, insufficient food quality, and insufficient food intake and physical consequences.

> Wealth and Assets
We gathered information about households' ownership of various productive and non-productive assets. Productive assets include items such as agricultural land, livestock, poultry, rickshaws or vans, and motor vehicles. Non-productive assets include items like televisions, mobile phones, refrigerators, electric fans, toilet facilities, and sources of drinking water. Using the baseline data, we created a wealth index and classified participants into wealth quintiles. These indicators were used to track participants' progress in wealth and asset acquisition as the impact of UTTARAN program.

> Financial inclusion, formal savings and access to social protection
We measured financial inclusion through respondents’ access to and utilization of financial services. Respondents were asked whether they had previously taken loans and, if so, whether loans were from banks or microfinance institutions. In addition, respondents were asked if they had access to bank accounts or mobile banking accounts. Respondents were also inquired about their savings habits, including whether they had formal savings accounts and their saving pattern. Moreover, access to social protection was identified through self-reported receipt of social protection such as old age allowance, disability allowance, etcetera.

> Women's empowerment
Women’s input into household decision-making includes their decisions regarding food expenditure, food to cook, investment in children, buying clothes for the family, buying household durables, taking a loan, and seeking healthcare. The responses were collected using a 5-point Likert scale, where 3 stands for an equal input; anything below 3 represent lower than equal input and higher than 3 represent more than equal input.

> Reduced morbidity and sickness-related absenteeism
We asked participants whether they or any of their household members suffered from any acute illness (like cold, fever, headache, diarrheal disease, etc.) within the last 30 days prior to the survey. We also inquired whether these illnesses prevented them from performing their regular activities and the number of days they were absent due to the illness during that time. By tracking these indicators, we aim to observe changes in productivity at the endline.

> Knowledge on Non-Communicable Diseases (NCDs)
We assessed respondents’ knowledge on NCDs, focusing specifically on hypertension and diabetes. Our knowledge assessment covered three aspects: the risk factors, management and consequences of uncontrolled hypertension and/or diabetes.

> Out-of-pocket health expenditure
We collected data on the costs incurred by respondents for acute and chronic illnesses and childbirth. We asked the cost of hospitalization for respondents hospitalized due to an acute illness, the monthly treatment costs for respondents with chronic illnesses, and expenses associated with childbirth (at home or healthcare facility) for pregnant women.

We adopted a three-arm Cluster Randomized Controlled Trial (CRCT) to comprehensively assess both the combined and marginal impacts of UTTARAN’s healthcare and livelihood interventions. The villages were treated as clusters, and we randomly assigned them into one of the three intervention arms.

We applied a two-stage sampling technique to recruit the study participants. At the first stage, we equally distributed 90 villages into the three arms using a simple random sampling technique. However, in the second stage, we applied a purposive sampling technique to select the households. In the sub-section, ‘Study sites’, we described how the program used PRA approach to identify locations of poor households in each village and conducted a survey to assess the PPI scores of the households residing in those locations. We used that list of potential beneficiaries as our sampling frame and purposively recruited the households that met our inclusion criteria. Instead of selecting them randomly, we prioritized the households with lowest PPI scores.

However, for 66 villages, the list fell short in providing 36 households from each village. Therefore, to recruit the remaining households, we replicated the PRA approach that the program applied to recruit their potential beneficiaries. First, using PRA approach, we identified additional locations in the village where comparatively poor household could be found. Then, we conducted the PPI survey in those locations and identified households meeting our inclusion criteria. We included them in the sampling frame along with the program-identified households. Finally, from each village-list, we selected 36 households that met our inclusion criteria.

The first arm was pure control. UTTARAN did not provide any intervention in these 30 villages during the study period. The villages in the second arm received only the livelihood support. In the third-arm villages, the program provided both the healthcare and livelihood supports.

We have selected this study design for several compelling reasons. First, it will help us estimate both the combined and marginal impact of SAJIDA’s health and livelihood support packages among poor households. Second, randomizing the clusters will allow us to minimize the risks of contamination. Third, the intervention is expected to bring change at the community level, where our design will be useful in capturing different community-level dynamics and impacts. Finally, randomizing at the cluster-level helps balance potential confounding factors and establish socioeconomic and geographic comparability between the treatment and control arms. By employing this rigorous study design, we aim to provide robust evidence to understand UTTARAN’s effectiveness in uplifting the poor and contribute valuable insights to evidence-based policies aimed at poverty reduction in vulnerable rural communities.

Villages/clusters were randomized in the office using Stata

Village

Yes

90 villages

3240 households

The total sample size is 3240 households, where each arm requires a sample size of 1080 households.
Therefore, we recruited 36 households from each of the 90 villages/clusters.

Minimum detectable effect size: 7% reduction in the average poverty rate, at 5% level of significance and 80% power of the test. According to the report titled 'Poverty Maps of Bangladesh 2016', the average poverty rate of our study sites was 28%. We hypothesized that UTTARAN’s intervention will bring the poverty rate down to 21%. Therefore, we assumed the probability that a household from the treatment arm and the control arm lives below the poverty line would be 0.28 and 0.21, respectively. After applying the sample size calculation formula for binary outcomes (Rutterford et al. 2015), we chose to recruit 32 households per cluster, assuming the intra-cluster correlation coefficient (ICC) to be approximately 0.02. Our assumption about the ICC was based on similar randomized controlled trials conducted in Bangladesh; their ICCs ranged from 0.01 – 0.05 (Amin et al. 2016, Buchman et al. 2018). Using these estimates, we calculated the sample size required to detect whether our hypothesized difference between the poverty rates of the treatment and control arms is statistically significant at 5% level of significance and 80% power of the test. The required sample size for each arm was 960 households, and the number of clusters required for each arm was 30. However, we inflated the sample size to adjust for possible attrition. Drawing upon similar literature, we assumed the attrition rate to be around 10%. Thus, the adjusted sample size for each arm was 1067 households. We rounded the figure to 1080 households to equally distribute among the 30 clusters (i.e., villages). Therefore, we needed to recruit 36 households from each cluster.