The study was conducted in two wards of Mumbai, India. Mumbai is a metropolis of 21 million people, with nearly half the population living in slums [9]. The prevalence of hypertension in Mumbai is estimated to be 25.9% for adults above 18 years of age [10]. Populations in Mumbai slums are reported to have poor access to diagnosis and underdiagnosis of non-communicable diseases [11, 12]. PATH, a global non-profit organization, had worked with private service providers in the slums of Mumbai through its public private interface agency model to ensure early and accurate diagnosis of tuberculosis, effective case management and successful treatment for patients through universal access to quality services [13]. Detailed description of this model is available elsewhere [14, 15]. The current hypertension control study leveraged PATH’s experience in working with the private providers in Mumbai.

After conducting a scoping exercise in six wards of Mumbai, two wards were selected for the project (Supplementary Fig. 1). The first ward, G-North, was a dense slum cluster with a population of 602,238. The second ward, the N-ward, had sporadic slum clusters and a population of 622,594.

The project conceptual framework was designed to address anticipated challenges to the successful management of hypertension in India’s private sector. A service delivery model was developed with three key interventions as seen in Fig. 1. The project was designed as an iterative model to enable quick operational changes and its adaptation to project activities to maximize impact. The project on-boarded private-sector service providers (doctors, pharmacists and laboratories) in both the wards.

The framework describes the underlying challenges in hypertension management in the private sector, the service delivery model, the implementation framework, and the expected outcomes.

We established a network of private practitioners in the study area who reported diagnosing more than ten hypertensive patients a month at baseline and were willing to comply with the study protocols. These included practitioners with Doctor of Medicine (MD), Bachelor of Medicine and Surgery (MBBS) and Ayurveda, Homeopathy and Unani degrees (together termed as AYUSH providers). All practitioners in the study received training on hypertension management according to the study protocols. A validated digital sphygmomanometer was provided to all practitioners. The frequency of follow-up visits depended on the number of patients being enrolled by the practitioners—those who enrolled fewer patients were visited more often. A total of 82 practitioners were enrolled in each ward over the course of the project. Among these, in the lean ward, 60 (73%) were AYUSH, 10 (12%) were MBBS and 12 (15%) were MD providers. In the intensive ward, 44 (54%) were AYUSH, 18 (22%) were MBBS and 20 (24%) were MD providers. Engagement of the private providers in the lean and intensive wards is depicted in Supplementary Fig. 2. Additionally, follow-up activities with empaneled private providers to ensure adherence to protocol are depicted in Supplementary Fig. 3.

Participants were enrolled in the study from January 2019 till February 2020. However, due to the COVID-19 pandemic, follow-ups could not be conducted regularly from February 2020 and the study procedures were stopped. While the project activities continued till February 2020, post November 2019, project activities were gradually phased out.

Different intervention models were implemented in the two wards as shown in Fig. 2.

This figure illustrates the interventions offered in the lean and intensive models.

In both the wards, hypertensive patients willing to register under the project received project services such as free drug vouchers, free serum creatinine test, adherence support, self-management counseling and follow-up call services to encourage patients to regularly follow-up with their service provider. The duration of the vouchers was coterminous with the next follow-up date. In addition, in the intensive ward, hub agents were placed in selected facilities with high patient load to improve patient registrations and to support service providers in facilitating uptake of services offered under the project. Some hub agents were full time project staff, while others were clinic staff and were provided an honorarium to conduct additional duties for the project. Of the 82 providers in the intensive ward, 16 facilities were provided project hub agents, while 14 were provided honorarium hub agents. The remaining 52 facilities in the intensive ward did not have any hub agents. Hub agents were given a target for enrollment based on the records and were eligible for incentives if they exceeded those targets.

Along with the hub agents, the project also provided for treatment coordinators. The treatment coordinator would call the patient 48 h after enrollment to provide lifestyle management and adherence counseling. The treatment coordinator would also contact all patients who missed their appointments. In the intensive ward, three treatment coordinators were provided to conduct adherence and monthly follow-up calls to registered patients, whereas, in the lean ward, only one treatment coordinator conducted adherence and quarterly follow-up calls.

Based on the performance of the project sites, over the course of the project, several modifications were made to the project operations. The changes included prioritization of field visits to providers based on their performance and potential to deliver targets, incentivization of hub agents for patient registration, simplification of data collection tools, introduction of fixed dose combinations (FDCs) to increase project hypertension protocol drug prescriptions and voucher utilization, intensive follow-up call week, call center services, and setting up of community kiosks for BP measurements. The FDCs available through the project included Telmisartan 40 mg and Amlodipine 5 mg, Telmisartan 40 mg and Chlorthalidone 6.25 mg, Telmisartan 40 mg and Hydrochlorothiazide 12.5 mg, Amlodipine 5 mg and Telmisartan 40 mg and Chlorthalidone 6.25 mg. In addition, some strategies that did not work were discontinued. The mid-course corrections and their effect on patient enrollment is shown in Supplementary Fig. 4.

Patients above the age of 18 years and diagnosed with hypertension were enrolled after written consent was provided. Hypertension was defined as systolic BP ≥ 140 mmHg or diastolic BP ≥ 90 mmHg on two successive occasions. Patients who were known hypertensives were also included. Patients with secondary hypertension were excluded.

Details of all patients enrolled in the study were entered into a patient card by the practitioners or their staff. Patients were provided lab vouchers for serum creatinine testing and antihypertensive drug vouchers as required. The counterfoils of the patient card, lab voucher and drug vouchers were collected from the empaneled clinics, laboratories and pharmacists respectively on a regular basis by the project field staff and compiled. All data was regularly monitored by the study staff for accuracy and completeness.

The primary outcome was the difference between a change in cohort hypertension BP control rates between the intensive and lean wards. Cohort BP control was based on an indicator of hypertension program performance as recommended in the WHO HEARTS technical package [16]. Cohort control was assessed for each wave of enrolled participants (cohorts) 3–6 months after they were initiated on treatment under the project. For 3–6 monthly cohort BP control rates, analysis was conducted using data of (1) all patients registered; and (2) only those patients who followed up in the quarter.

The secondary outcome was the difference in change in follow-up rates between the two wards. Follow-up rates were analyzed for patients with at least one follow-up visit 1-month after enrollment under the project. Other indicators that were analyzed were the proportion of patients with uncontrolled BP at baseline who achieved control and time to BP control. Kaplan–Meier curves were used to describe the time to BP control among patients with at least one follow-up visit. The cohort enrolled till December 2019 was considered for logistic regression and survival analysis.

Data were analyzed using R. Categorical variables among demographic and clinical variables are summarized using numbers and proportions, and continuous variables are described using the median and interquartile range. Logistic regression controlling for the ward, age, gender, voucher utilization, hub agent, provider type, and follow-up phone calls was used to describe the factors associated with controlled BP and age, gender, voucher utilization, hub agent, provider type, BP measurement at registration, proportion of follow-up phone calls, proportion of follow-up visits, duration of treatment, and ward for patient follow-up (at least once). Irrespective of the number of follow-up visits, the last available BP measurement was taken as final for this analysis.