Remote Care for Remote Areas: The Impact of Telehealth in Rural India

Telehealth connects patients to qualified health care professionals via phone. Although its popularity dramatically increased since the onset of the COVID-19 pandemic, there is limited rigorous evidence of its impact. In particular, the impact of telehealth on health access and healthcare inequality in low-income countries is ex-ante ambiguous. On the one hand, telehealth could allow individuals currently out of reach of the official health system -- e.g., because of their remote location or because of prevailing norms -- to access quality healthcare providers, thus improving the quality of care and equalizing access. On the other hand, telehealth might crowd out in-person care, with potential negative consequences on health outcomes for those individuals who do not engage with technology, who are not e-literate, or who have little trust in modern medicine to start with (e.g., women, the elderly, the poor). This may result in telehealth excluding individuals who might need care most, exacerbating inequality in healthcare.

Given the potential barriers to the utilization of telehealth in low-income settings, the presence of local “facilitators” may be crucial to its success. First, local facilitators may attenuate the digital divide by allowing marginalized individuals, including women with low power in the household, to contact the doctor and nurse through a smartphone that has access to the internet. Second, local facilitators may allow for more “continuity of care”, an aspect which is often lacking in telehealth programs that prioritize the “speed of care” (i.e., patients meeting with the first available provider even if that provider is unknown to them) and which might matter especially in contexts with low trust in modern medicine. Finally, local facilitators may complement telehealth services by monitoring the evolution of health conditions of patients that require frequent follow-ups (e.g., noncommunicable diseases).

India provides an ideal study setting, both because of the vast potential user base in need of telehealth services and because recent years have seen a rapid diffusion of telehealth providers. We will study the impact of the arrival of a new telehealth provider in a random subset of the 400 study villages in rural Bihar. The new health provider (Healing Fields Foundation, HFF) will introduce two (randomized) alternative versions of telehealth: with and without a facilitator, i.e., a community health worker that will help community members connect with a doctor remotely. We aim to study:

• What is the causal impact of telehealth on access to health services and health outcomes among people in rural areas?
• Does the presence of a facilitator increase the spreading and utilization of telehealth?
• Which types of patients are more likely to use telehealth with vs. without the facilitator? Does the impact differ across e.g. socioeconomic status and age groups?

2025-10-01

2026-12-15

1) Access to and utilization of healthcare. Measures will capture household contact with health actors; care-seeking in response to recent illness; the timing of care; referrals and follow-up; and use of telehealth, community health entrepreneurs (CHEs), and other public, private, informal, pharmacy, traditional, and government telehealth providers. We will examine both overall utilization and the composition of care in order to assess whether the study interventions expand access to healthcare or substitute for existing services.

2) Detection of chronic conditions and continuity of care for chronic conditions, particularly diabetes, hypertension, and anaemia. Measures will capture whether relevant health conditions are recognized or diagnosed and whether affected individuals are linked to appropriate providers and receive regular monitoring and follow-up.

We will summarize closely related measures within each primary outcome family in standardized domain-level indices, while substantively important component outcomes will also be reported separately. Provider-specific utilization measures will be used to distinguish changes in total healthcare use from substitution across types of care.

The main analyses will follow the randomized assignment of the telehealth and CHE components, separately and jointly. We will also examine whether treatment effects differ across relevant demographic and socioeconomic groups, including by gender (for questions where that is possible), age, socioeconomic status, baseline health needs, local availability of healthcare providers, and access to digital technology.

Household-level:
1) physical health: Self-reported health and recovery from recent illness; ability to live well with chronic health conditions; health-related work disruption or loss;
2) psychological health: psychological wellbeing and access to psychological support;
3) health knowledge: knowledge of common illnesses, chronic conditions, and danger signs requiring referral;
4) preventive and treatment-related behaviours, including medication adherence, hygiene, diet, and care-seeking practices, including healthcare access during periods of extreme heat;
5) household decision-making over healthcare; costs and perceived affordability of care; perceived access to a trusted healthcare provider; trust in and satisfaction with available providers; understanding of health advice received;
6) awareness, experience, perceived barriers, preferences, and demand for telehealth. We will include unincentivized willingness to pay questions and hypothetical choices between telehealth and in-person providers.

CHE-level:
1) retention in the CHE role;
2) work effort and service activity: household outreach; health education, screening, vital-sign measurement, referrals, facilitated teleconsultations, and patient follow-up;
3) training and supervision;
4) availability and use of equipment and any digital tools;
5) earnings and pricing;
6) health knowledge and competence;
7) confidence, motivation, satisfaction, and perceived standing in the community.

Outcomes on the level of other healthcare providers:
1) provider entry and exit; staffing and availability;
2) service and activity: hours worked and patient volumes; home visits and remote-care provision; the range of services and tests offered; referrals;
3) fees, revenues, and earnings;
4) health knowledge and competence;
5) motivation and satisfaction.

Secondary outcomes will be measured using the household, CHE, and healthcare-provider surveys and, where available, administrative data. Closely related outcomes will be combined into domain-level indices, with important individual outcomes also reported separately.

The study will take place in 400 villages in the state of Bihar, India. The study will be a 2x2 field experiment, which will be implemented after the baseline data collection, in collaboration with Healing Fields Foundation (HFF), an NGO which recently launched a telehealth program that relies on community health workers (CHWs) as “facilitators”. We will take advantage of the planned expansion of HFF activities in the study region, to randomly select the villages that will be reached first by the program. Each study village will be randomly assigned to one of four groups of equal size:
• Control (C): status quo (no HFF telehealth and no HFF CHW);
• Telehealth + CHW (T1): HFF will provide telehealth and will recruit and train a CHW, who will be in charge of facilitating it;
• Telehealth only (T2): HFF will provide telehealth services that community members can directly utilize, but won’t recruit any CHW;
• CHW only (T3): HFF will recruit and train a CHW, but there will be no HFF telehealth services.

Not available

Randomization will be done in office by a computer.

Randomization will be done at the village level.

Yes

400 villages.

15 households per village (~6000 households in total) are the units of observation for the household outcomes, and about 5 health providers per village (~2,000 in total) for the provider outcomes. These are spread across the 400 villages.

100 villages in control group (~1500 households, ~500 health providers)
100 villages in T1 (~1500 households, ~500 health providers)
100 villages in T2 (~1500 households, ~500 health providers)
100 villages in T3 (~1500 households, ~500 health providers)

The study is designed to detect at the 5% significance level with 80% power, an effect on health service utilization of each intervention equal to 0.135 standard deviations or larger.