I have lost count of how many digital health pilot projects promised to revolutionize healthcare access in LMICs. Most delivered a few thousand consultations before quietly disappearing. Then there’s eSanjeevani, India’s national telemedicine platform, which just delivered 163 million consultations across 28 states in under four years.
The scale alone demands attention. However, what fascinates me about the new research published in Oxford’s Digital Health journal is what it reveals about which digital health models actually survive contact with reality. The findings challenge nearly every assumption we hold about digital health adoption in resource-constrained settings.
The Data Nobody Expected
The Indian Ministry of Health and Family Welfare launched eSanjeevani to improve access to primary healthcare and advance digital health equity towards Universal Health Coverage. eSanjeevani has two models of usage:
- eSanjeevaniOPD: Using a patient’s mobile phone to access to doctors and medical specialists via a mobile application.
- eSanjeevani AB-HWC: Visiting any of the 208,000+ Ayushman Arogya Mandirs to have a teleconsultation with doctors and medical specialists in 15,000 District Hospitals.
In an era where we obsess over smartphone adoption, the winning strategy turned out to be low-tech intermediation through community health workers at rural health centers.
- 93% of consultations happened through the provider-assisted model.
- 57 to 70% of consultations were with women, depending on platform.
- 25-45 age group dominated usage.
- 70% of total volume was generated by 5 states.
The study’s most provocative finding: while acute conditions like fever dominated early consultations, by 2023 chronic disease management surged, with over 327,000 diabetes follow-ups recorded. This challenges the criticism that digital health solutions only work for minor ailments.
7 Lessons Learned for Digital Health Practitioners
eSanjeevani’s trajectory confronts the prevailing orthodoxy that direct-to-patient, smartphone-based solutions represent healthcare access in LMICs. The data suggests otherwise.
The future, at least for the next decade in most low-resource settings, looks like community health workers with basic devices connecting patients to remote physicians through government-managed platforms integrated into public health systems.
Kinda like the Intelehealth approach, eh?
This contradicts the market-driven, app-centric models that dominates global discussions. It also challenges assumptions that women’s limited smartphone access will prevent telemedicine adoption. What actually prevents adoption is inadequate state capacity, fragmented systems, and platforms designed for urban, digitally literate users.
Here are the seven lessons learned in deploying eSanjeevani according to the research:
1. Assisted models are best for LMIC contexts.
The 93% utilization through hub-and-spoke demonstrates that intermediated telemedicine, where a health worker facilitates the consultation, overcomes digital literacy barriers, trust deficits, and connectivity challenges simultaneously. This is not sexy, but it works.
We need to prioritize assisted telemedicine models with CHW facilitation over standalone patient apps in contexts with low digital literacy and connectivity challenges.
2. Infrastructure expansion is a must.
eSanjeevani scaled from 6,868 spokes to 108,610 by 2023. The platform succeeded because it embedded within existing Health and Wellness Centers aligned with the WHO’s Global Strategy on Digital Health. Too many pilots fail because they operate in isolation.
We must embed platforms within existing health infrastructure rather than creating parallel systems. Integration with national digital health architecture is non-negotiable for sustainability, as detailed in recent locally-owned health data systems analysis.
3. Geographic concentration reflects state capacity.
The top five states generated over 70% of consultations. Rather than inequity, I see evidence that political commitment and implementation capacity matter more than technology design. This suggests governance capacity building should receive equal investment to platform development.
We need to invest in state-level implementation capacity including training for community health officers, technical support for hub facilities, and governance structures. Technology is necessary but insufficient.
4. Women will use digital health services.
Despite India’s gender digital divide, women were majority users. The assisted model, delivered through local health centers, eliminated mobility and permission-based barriers restricting women’s healthcare access. This contradicts assumptions that telemedicine will reproduce existing gender gaps.
Still, we need to account for gender disparities when designing digital health solutions. For example, are their women on the application development team, and crucially, are women developers leading conversations with female end users and patients?
5. Chronic disease management requires tech maturity.
The shift toward NCD follow-ups in version 2.0 correlates with improved audio-video stability, prescription synchronization, and point-of-care device integration. Most LMIC telemedicine platforms still operate at version 1.0 functionality.
We should design for chronic disease management from the beginning by incorporating longitudinal records, point-of-care diagnostic integration, and medication tracking functionality.
6. Short consultation times are not inherently problematic.
The average 1 minute 15 seconds reflects active interaction time, not total turnaround. For medication refills or follow-up check-ins, brief consultations are appropriate and enable volume. Quality should be measured by clinical appropriateness, not arbitrary time thresholds.
We should focus on quality use of digital health solutions, not just the absolute quantity of usage.
7. Aggregate data obscures critical equity questions.
The study’s primary limitation is reliance on programmatic data without patient-level tracking. We cannot determine unique users or repeat visit patterns. Impressive aggregate numbers may mask persistent inequities.
We can implement individual patient tracking using unique health identifiers immediately to enable equity analysis, continuity of care assessment, and outcome evaluation.
Ecosystem Development Before Software Deployment
The eSanjeevani experience reveals that successful digital health scaling in LMICs requires addressing foundational inequities rather than simply deploying technology.
While government commitment and indigenous development enabled rapid deployment, the platform’s challenges mirror broader LMIC digital health patterns: infrastructure dependencies, workforce capacity constraints, and socio-cultural adoption barriers.
