Interactive Voice Response (IVR) based health information for antenatal and postnatal care services in rural Bangladesh

Our treatment intervention provides knowledge on reproductive health issues via information modules delivered via IVR technology, which treated women are able to access for free using their cell phones. The intervention has two treatment arms and a control arm outlined below.
Treatment Arm 1: The control group was not administered any intervention.
Treatment Arm T2: Access to IVR-based modules via toll-free phone number
Treatment Arm T3: Access to IVR-based modules via toll-free phone number + telehealth support by a doctor for common health problems, at participant’s own discretion. each.

Enumerators working with the two treatment arms will have different responsibilities. For the participants in arm T2, we will see them face-to-face a total of 4 times: first, at the time of the information and baseline survey, about 45-60 minutes; second, at the time of the demonstration of how to access IVR information modules, about 45-60 minutes; third, at the 10th to 13th day of delivery, about 30 minutes; and fourth, at endline, that is 6-6.5 months after delivery, about 45-60 minutes. The enumerators explained in a step-by-step manner, how the participant could access the IVR modules using a mobile phone. They highlighted that participants can access the information free of cost and as many times as they like. Participants are expected to listen to each IVR module at least once for a potentially measurable effect. The duration of each module is about 5 to 8 minutes. However, for each participant, we will observe their interaction with the system and the time they spent dialed into each module. Fifteen IVR modules are being delivered to the T2 and T3 participants. Intervention modules cover the following topics: (i) taking care of self during pregnancy (Module 1); (ii) diet during pregnancy (Module 2); (iii) signs of potential pregnancy complications (Module 3); (iv) antenatal care and birth plan (Modules 4-6); (v) postnatal care (Modules 7-8); (vi) maternal mental health (Modules 10-12); and caring for the baby in the first six-moths of life (Modules 9, 13-15).

The telehealth intervention in T3, is provided for individual, one-on-one counseling regarding any pregnancy-related issues including some general health problems only where a doctor can provide support without physically examining pregnant women. This includes referrals, advice on what to do for each individual circumstance, and over-the-counter drugs to support ongoing general health situations. If a physical examination is needed, the patient is referred to a doctor or a nearby hospital. The patient was always advised to visit a doctor or hospital and was told not to think of this telehealth option as an alternative to seeing a doctor face-to-face. It provided flexibility for patients to seek immediate support for consultation. In most cases, women from rural areas do not have access to consulting a doctor, and a visit to a certified doctor is seen as the last option. The situation is even more acute for pregnant women because of trouble commuting far to see a doctor or nurse in a clinic or hospital. Thus, telehealth consultations are less likely to replace any existing medical support they can receive.

2023-03-31

2024-03-31

(1) Antenatal Care (ANC) and Postnatal Care (PNC) Practice
Number of ANC visits to a doctor/midwife
Number of PNC consultations with a doctor/midwife
Tool: Study-specific questionnaire about ANC and PNC visits at endline
Timepoint: 6 months postpartum

(2) Maternal Mental Health and Child Caring Learning
Awareness about maternal mental health
Awareness about nutrition and immunization of infants and babies
Tool: Study-specific questionnaire
Timepoint: 6 months postpartum

(3) Health-Seeking Behavior from Facilities for Mother and Child
Visiting health facilities or consulting a doctor for health problems during pregnancy and postpartum (up to 6 months) (excluding ANC and PNC visits/consultations)
Visiting health facilities or consulting a doctor for health problems of the infant/baby (within the follow-up period) (excluding immunization)
Tool: Study-specific questionnaire
Timepoint: 6 months postpartum

(1) Events of 5 Signs of Pregnancy Complications
Tool: Study-specific questionnaire
Timepoint: At delivery (+ 10 days, as the mother may need time to recover and care for the new baby)
(2) Stillbirth
Tool: Patient medical record
Timepoint: At 6 months after delivery

(3) Episodes of Common Illnesses Among Babies
Tool: Study-specific questionnaire
Timepoint: At 6 months of age

(4) Postnatal Infection
Tool: Study-specific questionnaire
Timepoint: At 10 days post-delivery

(5) Preterm Birth
Tool: Patient medical record
Timepoint: At 10 days post-delivery

(6) Low Birthweight
Tool: Patient medical record
Timepoint: At 10 days post-delivery

(7) Duration and Number of Times Accessed IVR (Composite Outcome)
Tool: Platform analytics
Timepoint: 6 months postpartum

(8) Awareness of ANC and PNC - Nutrition, Rest, Required ANC and PNC Visits (Composite Outcome)
Tool: Study-specific questionnaire
Timepoint: 6 months after delivery

The study leverages an interactive voice response (IVR) system to disseminate health information to pregnant women in rural Bangladesh. The randomized controlled trial includes three groups: control (no intervention), IVR-only (access to information modules), and IVR + telehealth support (access to doctors). Women were recruited based on their pregnancy stage, aiming to reach those in their first or second trimester for optimal impact

Randomization was performed at the office using a computer.

Village-level randomization, with each village serving as a cluster.

Yes

330 villages.

Approximately 4,110 participants (1,370 per treatment arm).

Sample Size by Treatment Arms
T1: Control: 1359 women
T2: IVR-only: 1374 women
T3: IVR + doctor support: 1371 women

We perform a power calculation using our baseline data to determine the minimum detectable effect size (MDE). The key parameters are as follows: α = 0.05 (standard type I error), κ = 0.80 (standard power), J = 110 (number of clusters (villages) per treatment arm), number of respondents under three treatment arms T1 = 1359 (control), T2 = 1374 (IVR only Treatment), and T3 = 1371 (IVR + Access to Doctor Treatment), and we used the low ICC = 0.15 (intra-cluster correlation coefficient). For the actual power analysis, we round off the latter two numbers to 1370. Using one of our main outcome variables, the required number of Ante-Natal Care (ANC) visits, we compute the MDE. According to our baseline, only 2% of respondents in the three treatment arms have access to the required number of ANC visits. Therefore, our study is sufficiently powered to detect an effect size of 0.0331 or 3.31 percentage points. In case we experience very high attrition of roughly 30% (which would reduce the treatment sample to 959 and the control sample to 951), the minimum detectable effect size changes very little to 0.0359. For the MDE, we need less than 4 percentage points improvement in ANC visits, and given such a low base we expect the impact will be much higher. We implement a similar exercise for a second outcome of interest, Post-Natal Care (PNC) visits. According to the baseline, only 5% of women (overall) have access to four required number of PNC visits. Our study is sufficiently powered to detect an effect size of 0.0491, which is 4.91 percentage points (or 0.0528, with 30% attrition). For the MDE, we need around 5 percentage points improvement in PNC visits, and given such a low base we expect a much higher treatment effect. Similarly, we also compute the MDE of the treatment for some other outcome variables such as knowledge of five risk factors during pregnancy, providing additional food to the child after six months, breastfeeding duration, breastfeeding time, and going to Kabiraj (witch doctor) during complexity. Our study is sufficiently powered to detect an effect size of (i) 0.0381 for knowledge of five risk factors (or 0.0412 with 30% attrition); (ii) 0.0611 (or 0.0647) for knowledge of providing additional food to the child after six months of birth; (iii) 0.0882 (or 0.0939) for knowledge of breastfeeding duration (1 = if continue breastfeeding until the age of 6 months); (iv) 0.0775 (or 0.0823) for knowledge of breastfeeding time (=1 if started within an hour); and (v) 0.0865 (or 0.0823) for knowledge of going to Kabiraj (witch doctor) during complexity (=1 if say yes).