Cost of diabetes and hypertension services delivered in primary and secondary-level government facilities in Punjab, India

Abstract

Background

The recent rise of diabetes and hypertension in India has motivated initiatives to enhance screening and management of these conditions within the public health care system. The associated healthcare costs of screening and management may inform policy planning and scaling initiatives, yet costs are not well documented in this setting. We conducted a micro-costing study to estimate the cost per visit and total annual cost of outpatient diabetes and hypertension care in primary and secondary level health facilities in northern India.

Methods

Data collection took place in rural Punjab state in 2019 and included a facility survey and patient time and motion study at 1 district hospital (DH), 2 community health centres (CHC), and 8 primary health centres (PHC). Costs per visit were compared by visit type and facility level using Dunn’s test. We used one-way, deterministic sensitivity analyses to examine the potential impact of uncertainty on findings. Costs are expressed in 2019 INT$ and INR₹ from a health system perspective.

Results

Average per-visit costs were higher for diabetes (range: INT$12.54–14.36) and co-morbid hypertension and diabetes (range: INT$13.11–15.04) visits than those for hypertension (range: INT$9.04–12.73) across facility levels. The resource categories that drove visit-specific costs were medication (30.4% of costs) and staff (28.9%). At a facility-level, medication and investigation costs tended to be higher for visits at CHCs than those at PHCs and the DH. Total annual costs were highest at the DH (INT$96,114.66) due to the larger number of patients served. Sensitivity analyses confirmed that per-visit costs were most sensitive to medication price and staff salary.

Conclusions

The cost of services delivered for diabetes and hypertension at Indian public sector facilities varied by facility level. Higher costs at CHC and DH levels may reflect the role of these facilities in providing more specialized services, serving medically complex patient populations, or may indicate inefficiencies in service organization. Findings of this analysis may inform health system planning to expand coverage of service delivery.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-026-14058-4.

Background

Cardiovascular disease, including ischemic heart disease and strokes, is now the leading cause of death in India [1]. High blood pressure and high blood glucose, key cardiometabolic risk factors, were estimated to account for 24.4% of deaths in India in 2019 [1]. Timely detection and appropriate treatment of hypertension and diabetes to achieve control targets can significantly decrease risk of cardiovascular events and other complications, reducing mortality and morbidity [2–4]. However, access to these diagnostic and management services remains insufficient. National data indicate that only 8.5% of adults with hypertension and 48.6% of those with diabetes in 2019–2021 had received treatment and achieved control of their blood pressure or glucose [5, 6].

Strengthening health service delivery for non-nommunicable diseases (NCDs) is a priority for the Indian Ministry of Health and Family Welfare [7]. In 2016, the Ministry of Health and Family Welfare introduced operational guidelines for the management of common chronic conditions, including diabetes and hypertension [8]. These guidelines, formulated under the former National Programme for the Prevention and Control of Cancer, Diabetes, Cardiovascular Diseases and Stroke, screening services for hypertension and diabetes were to be conducted in communities and local health facilities (such as sub-centres and primary healthcare centres). Patients requiring more specialized services were to be referred to secondary and higher level facilities (such as community health centres and district hospitals) as necessary, before being referred back to local-level facilities for ongoing management. In 2025, as at the time of this study in 2018–2019, alignment of service provision with these strategies remains ongoing as part of broader health system reforms and efforts to expand care at the community level [9].

Economic evidence relating to the cost of delivering health services can be useful for health sector planning and scale up of services [10]. Estimates of the cost of services can support budget forecasting and inform allocation of resources to disease areas and resource categories. However, the cost of diabetes and hypertension services at government health facilities in India is not well documented, as prior studies have primarily estimated patient out of pocket expenditure [11–15]. To address these knowledge gaps, we conducted a study to estimate the health system cost of delivering diabetes and hypertension services in government healthcare facilities in Shaheed Bhagat Singh (SBS) Nagar district of Punjab state.

Methods

Study design and setting

This micro-costing study was conducted to estimate the economic cost of outpatient diabetes and hypertension services delivered in government health facilities in rural Punjab state, India. The study was conducted as part of the Integrated Tracking, Referral, Electronic decision support, and Care coordination (I-TREC) project baseline assessments [16]. Data collection occurred in primary and secondary health facilities located in Mukandpur and Sujjon blocks of SBS Nagar district between July 2018 and December of 2019. The district and blocks were selected based on consultation with the Punjab Department of Health and Family Welfare for the I-TREC implementation trial.

Three levels of health facilities were included in the study: Primary Health Centre (PHC), Community Health Centre (CHC), and District Hospital (DH) levels. Under the 2012 Indian Public Health Standards, which were in place at the time of this study, PHCs were expected to provide outpatient primary care services, minimal laboratory investigations (e.g. random blood glucose), and basic medicines (e.g. calcium channel blockers and metformin) [17]. CHCs, as secondary level healthcare facilities, were to offer additional investigations (e.g. blood lipid profiles, ECGs) and expanded medications (e.g. statins) [18]. Both PHCs and CHCs were expected to conduct health promotion counselling, screening, diagnosis, and management of uncomplicated hypertension and diabetes, with vertical referral mechanisms in place between the levels to allow CHCs to see individuals requiring additional investigations and management. Finally, DHs were to offer more specialized services for the management of complicated diabetes and hypertension, such as emergency cases of hypoglycemia [19]. For the purpose of this study, we only estimate the cost of facilities’ outpatient services.

Data collection

Data were collected using both top-down and bottom-up approach to estimate health system costs. Bottom-up data collection methods included surveys at the patient and health facility levels. Surveys were conducted by a field team of enumerators, all of whom had completed university and received training on study instruments and methods prior to data collection. All surveys were pretested, first internally by the field team, then in health facilities in adjoining blocks. Instruments were finalized to address any identified issues prior to starting data collection. Table 1 summarizes the scope and methods employed for each survey.

Table 1.

Facility and participant characteristics of surveys utilized for costing analysis

Survey	Scope	Facilities	Participants [n (%), mean ± SD]
Facility Assessment	Facility characteristics, including resource availability	n = 11 • District Hospital (n = 1) • Community Health Centre (n = 2) • Primary Health Centre (n = 8)	None
Facility Cost Survey	Annual facility spending, including staff salaries, overhead, and building value	n = 8 • District Hospital (n = 1) • Community Health Centre (n = 3) • Primary Health Centre (n = 4)	None
Patient Time and Motion Study	Characteristics of patient visits, including providers seen, services received, and duration	n = 11 • District Hospital (n = 1) • Community Health Centre (n = 2) • Primary Health Centre (n = 8)	n = 516 • Male, 245 (47.5%) • Age, 60.1 ± 11.2 years • Visit reason  o Hypertension, 250 (48.4%)  o Diabetes, 184 (35.7%)  o Hypertension & Diabetes, 82 (15.9%)
Community Survey	Patient care seeking behavior, used to estimate annual number of diabetes and hypertension visits	None	n = 3239 • Male, 1499 (46.3%) • Age, 51.5 ± 13.0 years • Prior diagnoses  o Hypertension, 508 (15.8%)  o Diabetes, 156 (4.9%)  o Hypertension & Diabetes, 141 (4.4%)

At the patient level, a time and motion study was conducted. All patients were approached on arrival at the outpatient registration counter to determine if they were seeking services for diabetes or hypertension. Following introduction of the study, eligible patients provided written informed consent to participate. Patients were then shadowed for the duration of their visit at the facility by an enumerator who recorded the services they received and the duration of each activity in the facility (Figure S1). Data collected included details of which providers were present for consultations, which lab investigations were conducted, and whether medications were prescribed. Due to a gap in information on medication dispensing, patient self-report was used to record which medications individuals were prescribed.

At the facility level, a facility assessment and a facility cost survey were conducted by interviewing a designated facility representative (often a senior medical officer) at each facility. The facility assessment captured information on each facility’s resource availability, including equipment inventory and useful life. The facility cost survey captured expenditures over the last fiscal year (2017–2018), as reported by the facility representative. The included expenditure categories were staff salaries, equipment, overhead (including utilities), and other facility infrastructure. In addition, this survey captured patient volume at the facility based on Health Management Information System (HIMS) reports. The cost portion of the facility assessment was also conducted with two facilities (1 PHC and 1 CHC) in the Muzzafarpur block of adjoining Jalandhar district.

Top-town data collection methods drew inputs from government procurement rates and prior costing analyses. Medication costs were extracted from National Health System Cost Database for India (NHSCDI) for medications [20]. We used Central Government Health Scheme (CGHS) rates for laboratory and other investigations [21]. We estimated the cost of non-medical support staff (e.g. cleaning and administrative staff) at 14% of visit cost, non-inclusive of medication, and furniture at 1–2% of total visit cost based on prior cost studies of public healthcare services in India [22, 23]. Additional details regarding the assumptions used in estimating unit costs are available in the Supplementary Appendix.

Data analyses

Costs were estimated from the health system perspective and by facility level (Primary Health Centre, Community Health Centre, and District Hospital) [24]. We estimated the average cost per visit for three visit types: diabetes, hypertension, and co-morbid hypertension & diabetes. Visit types were classified based on patient self-report of their reason(s) for visiting the facility. We then estimated the average annual cost of diabetes and hypertension services for a facility at each level (PHC, CHC, DH). Any costs with a source year other than 2019 were converted using the World Bank Consumer Price Indices and the annual inflation rates for India in the relevant years [25]. We applied a discount rate of 3% to any costs incurred over multiple years (varied 1–5% in sensitivity analyses). Costs are expressed in 2019 International dollars (INT$) and 2019 Indian rupees (INR₹).

To calculate the cost per visit, we first estimated the cost of each patient encounter observed in the time and motion study. For each patient, the unit cost of each resource was multiplied by the observed resource use, and costs were then added across resource categories to estimate the total visit cost [24]. For example, the staff cost associated with each visit depended on which providers interacted with the patient, and the duration of their interactions. While all patients incurred a cost associated with shared overhead expenses, only patients who had their blood pressure measured incurred the cost of using the blood pressure monitor. Cost estimates reflect current levels of capacity utilization at the facilities. Cost estimation methods are further described in the Supplementary Appendix.

Patient-level costs were summarized for each facility level and visit type using descriptive statistics (means and 95% confidence intervals). The mean cost per visit was compared across facility levels and visit types using Kruskal-Wallis test and Dunn’s test, using the Bonferroni correction to adjust for testing of multiple pairwise comparisons at a threshold of α = 0.05. For each facility level and visit type, the mean cost per visit was then multiplied by the estimated number of visits per year to calculate annual expenditure on outpatient diabetes and hypertension services for each facility level.

Sensitivity analyses

To account for potential variability in the cost of resource inputs, we conducted a one-way, deterministic sensitivity analysis. In this analysis we varied the cost of each input category one by one to assess their potential impact on estimated overall cost per visit [22, 26]. The unit cost of most input measures were varied by 25% in each direction. To account for the greater variability of medication costs, these were instead varied by 50%. Equipment and capital expenses are varied using a minimum discount rate of 1% and maximum of 5%. Findings are summarized using tornado plots.

Ethical oversight

The I-TREC data collection protocol was reviewed and approved by the ethics committee at All India Institute of Medical Sciences (AIIMS), New Delhi, India (IEC-361/07.07.2017). Collaboration by Emory University researchers was determined to be Not Human Subjects Research (IRB00098808). Research was conducted in accordance with the Declaration of Helsinki.

Results

Sample characteristics

This costing analysis utilized data from four surveys. Survey scope and facility and participant characteristics for survey instruments are presented in Table 1. The facility assessment included 11 facilities and the facility cost survey included 8, with representation across facility levels. The patient time and motion study included 516 patients with diabetes and/or hypertension (47.5% male, mean age of 60.1 years) attending care at the same 11 health facilities. Finally, the community survey included a sample of 3,239 adults residing in the study blocks, representative of the adult population of the blocks (46.3% male; mean age 51.5 years).

Average cost per visit

The average cost per visit varied by disease type and facility level (Fig. 1). Across facilities, the mean cost per visit was higher for diabetes (range: INT$12.54–14.35) and co-morbid hypertension & diabetes visits (range: INT$13.09–15.04) than those for hypertension (range: INT$9.03–12.73). By facility level, average costs tended to be higher for visits at CHCs than PHCs, and highest for visits at the DH. However, visits for hypertension had greater average costs at the CHC level (INT$12.73 [95% CI:11.51, 13.95]) compared to the PHC (INT$9.04 [95% CI: 8.82, 9.27], p < 0.001) and the DH (INT$9.46 [95% CI: 8.51, 10.40], p < 0.001).

Fig. 1.

Mean cost of diabetes and hypertension visits, by visit type and facility level (2019 INT$)^†. *: p < 0.001 for pairwise comparisons using Dunn’s test with Bonferroni correction; NS: no significant difference. †PHC: primary health centre; CHC: community health centre; DH: district hospital; DM: diabetes visits; HTN: hypertension visits; HTN & DM: comorbid hypertension & diabetes visits

The resource categories that made up the largest share of the visit costs were medication and staff salaries (Fig. 2). The share of visit costs attributable to medications were higher for diabetes-related visits than those for hypertension, with the highest medication costs for visits for diabetes (at DH-level) and co-morbid hypertension & diabetes visits (at PHC and CHC levels). Medication costs accounted for approximately one third of total visit costs for diabetes and co-morbid hypertension and diabetes visits, and 15–20% of total costs per visit for hypertension visits.

Fig. 2.

Share of visit cost attributable to resource cost categories, by visit type and facility level ^†. ^†PHC: primary health centre; CHC: community health centre; DH: district hospital; DM: diabetes visits; HTN: hypertension visits; HTN & DM: comorbid hypertension & diabetes visits

The costs of medical staff for each visit ranged from approximately INT$2 at the PHC level to about INT$5 at the DH level. Across facilities and visit types, almost all patients were seen by a medical officer. The difference in cost between facilities reflects both higher average salaries for providers at the DH level, as well as differences in the providers seen. For example, while only 23% of people seeking diabetes care at PHCs were seen by a nurse before being seen by the medical officer, 100% of patients seeking diabetes care at the DH level first saw a nurse (Supplemental Table S1). Medical staff accounted for the lowest share of the cost per visit for diabetes-only visits at PHCs (16.4%) and the greatest share at hypertension-only visits at the DH (51.1%) (Fig. 2; Supplemental Tables S2-S3).

Visits for diabetes alone or co-morbid hypertension and diabetes had higher costs for laboratory investigations. The cost for laboratory and other investigations tended to be higher for visits at CHCs than those at PHCs and the DH, with the exception of visits for comorbid hypertension and diabetes for which the investigation costs were similar across levels.

Many patients did not receive recommended investigations. Only about two thirds of patients attending visits for hypertension and only one third of those attending visits for diabetes received a blood pressure measurement in the time and motion study (Table S1). Similarly, about half of patients at diabetes-related visits received a blood glucose test (57.1% for diabetes, 54.9% for co-morbid hypertension & diabetes visits, Table S1). The majority of investigations observed in the patient flow survey, including laboratory investigations for blood lipids and other investigations such as electrocardiograms (ECGs), occurred at the CHC level. The exception was blood glucose tests, which were conducted across levels (Table S1). Supplemental Tables S2-S4 present the detailed cost of each visit type in 2019 INR₹ and INT$.

Annual cost of services by disease and facility type

The total annual cost of diabetes and hypertension services was highest at the DH (INT$192,229.26) and lowest for a PHC (INT$5,396.03) (Table 2). The difference in annual cost was largely driven by differences in patient volume, as the DH served an estimated 16,927 people for diabetes and/or hypertension in a year, compared to only 510 for a PHC. Although diabetes related visits had a higher associated cost per visit, annual costs were higher for hypertension due to the larger number of patients served at all facility levels. The annual cost of services in INR₹ is presented in Supplemental Table S5.

Table 2.

Expected annual cost of diabetes and hypertension visits, by visit type and facility level (2019 INT$)

	Primary Health Centre			Community Health Centre			District Hospital
	DM	HTN	HTN & DM	DM	HTN	HTN & DM	DM	HTN	HTN & DM
Mean cost per visit	12.54	9.04	13.11	13.39	12.73	15.03	14.36	9.46	15.04
Mean # visits per year	99.85	303.43	106.79	369.46	651.39	257.97	4,094.00	10,664.95	2,168.25
Expected annual cost by visit type	1,252.55	2,743.90	1,399.58	4,948.64	8,292.96	3,877.12	58,774.53	100,838.20	32,616.49
Expected annual cost, total	5,396.03			17,118.72			192,229.26

Sensitivity analyses

One-way sensitivity analyses confirmed that visit costs were most sensitive to medication price and staff salary (Fig. 3). Varying medication price resulted in the largest change in estimated visit costs in all diabetes-related visits. For hypertension visits, varying staff-related costs resulted in the greatest change in visit costs at CHC and DH level visits. Varying other cost categories, such as overhead (i.e. utilities) and capital (i.e. building infrastructure), did not substantially impact the estimated visit costs for most visit types or facility levels. Results of these sensitivity analyses are also presented in Supplemental Table S6-S7.

Fig. 3.

Results of one-way deterministic sensitivity analyses for visit cost, by facility level and visit type (2019 INT$)^†. ^†Tornado diagrams represent the potential minimum (Lower Bound) and maximum (Upper Bound) cost per visit expected based on uncertainty in the cost parameters. The Upper Bound represents the cost per visit calculated based on maximum unit cost of one resource category (i.e. medications), holding all other resource unit costs constant. Similarly, the Lower Bound is the cost per visit based on the minimum unit cost of one resource category, holding other resource unit costs constant. The plot center line is the cost per visit estimated in primary analyses

Discussion

To our knowledge, this is among the first studies to examine health system costs for diabetes and hypertension services at the primary and secondary care levels in India. Population-based data suggest that hypertension and diabetes remain underdiagnosed and undertreated in the Indian setting. Addressing these population health gaps is among the responsibilities of primary and secondary healthcare services provided by the public sector. Cost data are essential to plan efforts to strengthen delivery of NCD services while contending with resource constraints. We found that the cost of services delivered in one outpatient visit varied from INT$9.04–15.04. The cost tended to be higher for diabetes-related visits than those for hypertension. Costs also tended to be higher at secondary-level healthcare facilities than at primary-level facilities. Medication and staff salaries accounted for the largest share of costs at most visits. However, investigations made up a comparatively larger share of costs at the CHC-level, and staff costs were highest at the DH-level. These results contribute to an important gap in the current research literature on the cost of providing services for diabetes and hypertension in India.

Our findings identified several resource categories that drove total visit costs, with variation based on visit type and facility level. The higher cost we estimated for diabetes-related visits was largely driven by the higher cost of diabetes medications. Overall, medications made up about one third (range 15–44%) of the cost of visits, highlighting the salience of efforts to enhance affordability [27]. Medications are commonly cited as one of the most costly components of NCD care, particularly for diabetes [28, 29]. We also found cost per visit to be higher at CHCs and the DH than at PHCs. The higher cost per visit at CHCs was partially driven by a larger share of patients receiving laboratory and other investigations compared to other levels, including HbA1c tests and ECGs. At the DH, higher costs attributable to staff-related expenses were due to both higher salaries and greater time spent with patients. These patterns may reflect differences in care seeking and specialization of the facility levels, as CHCs are envisioned as the center of NCD care and patients at the DH may be receiving care for complications or other specialized services [30]. The costing protocol developed for this study could be used to estimate service delivery costs across more settings to investigate potential cost differences between facility levels.

Prior facility-based studies have estimated the costs of services delivered at specialized hospitals [31] and those for treatment of complications, such as diabetic retinopathy and foot ulcers [32–34]. Other studies have examined patient out of pocket expenditure for routine services for managing these conditions [11–15]. Patient out of pocket costs are an important component of understanding healthcare costs in India, as this approach has the advantage of capturing spending in both the public and private sectors and allowing for the framing expenditure within patients’ ability to pay. However, such estimates have less direct utility for health system planning and budgets for primary healthcare services. Several studies have examined facility-based costs of primary and secondary health services in multiple Northern Indian states, estimating the average cost of a general outpatient consultation as INR 139 at PHCs and INR 134–172 at CHCs [22, 23]. Our estimates, which range from INR 183–304 per visit, suggest that the cost of diabetes and hypertension services are greater than for general outpatient care, which emphasizes the importance of preventing these conditions for the government sector.

Our estimates reflect the cost of imperfect service delivery. National treatment guidelines for both diabetes and hypertension recommend patients have their blood pressure checked at every visit [35, 36]. However, only about two thirds of patients attending visits for hypertension received a blood pressure measurement and only about half of patients with diabetes received a blood glucose test in our time and motion study. These gaps in care represent continued unmet needs, even among patients attending health facilities, and are an potential source of inefficiency for the system [37]. Without recommended investigations and resources, providers may not have the full picture of a patient’s health that could inform alterations to their treatment plan to improve blood pressure or glucose management. As a result, the health system incurs the cost of resources required for a consultation, but this spending may not result in optimal improvements to health. These inefficiencies in service delivery are a reason to consider a societal perspective in future analyses, particularly to evaluate both costs and health outcomes. In addition, the noted gaps in service delivery and availability may influence patients to seek care in the private sector, where costs (namely out of pocket costs) are as much as 2.5–4.5 times higher [11].

We acknowledge limitations of this analysis. These estimates are based on data collected at facilities in one rural district of Punjab state and may have limited generalizability to other states and settings. Our costs also reflect the current level of capacity utilization at these facilities. It is possible that as a result the share of overhead costs allocated to each individual visit is high due to lower patient footfall, especially in smaller rural clinics. Because we lacked information on dispensing of prescribed medications, our costs assume that all medications prescribed were dispensed to patients. Prior studies have identified frequent stock outs of recommended medications for hypertension and diabetes, so all prescribed medications may not be present at these facilities consistently [38–41]. As a result, the share of the total visit costs attributable to medication might also overestimate current service delivery. Because we costed observed service delivery, we note that the cost of delivering diabetes and hypertension care as per the national treatment guidelines may also differ from the costs we estimate. Finally, while our focus was on public sector services, we do not have visibility into care seeking patterns and costs for patients who may access services in both the public and private sectors, so these estimates are not comprehensive from a patient perspective.

There are also several strengths of our approach to note. The first is the use of primary data collection to inform costing inputs and assumptions. The project included a time and motion study to collect detailed information on the services received by patients and the duration of each visit. Notably, information on which providers were seen and for how long enabled detailed estimation of the contribution of staff salaries to the overall visit cost, a substantial proportion of the total cost. Additionally, through the facility-based data collection we were also able to incorporate facility-reported expenditure on overhead and equipment to complement cost estimation. The focus on primary and secondary level facilities also represents a strength of this approach, as these estimates fill a gap in the literature, as previously described. Finally, we conducted one-way deterministic sensitivity analyses to quantify potential variability in these estimates and the contribution of variability in each cost category to our overall estimates.

Conclusions

We estimated the health system cost of delivering services for diabetes and hypertension in primary and secondary health facilities in a rural district of Punjab state, finding differences in cost based on disease, facility level and visit type. This study used primary data triangulated across multiple dimensions, including time and motion data collection and facility assessments, to characterize the current status of service delivery and offers novel estimates of the costs of services in these facilities. The costs estimated in this study provide an initial picture of the cost of diabetes and hypertension services, a critical step toward sustainably scaling up access to close gaps in service availability and improve health outcomes [42]. The methods employed in this study could be replicated in additional contexts to inform the efforts of the National Program for the Prevention and Control of Noncommunicable Diseases to strengthen of care for these conditions in primary and secondary health facilities across India [30].

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

NCDs

Noncommunicable diseases

DH

District hospital

CHC

Community health centre

PHC

Primary health centre

INR

Indian rupees

INT$

International dollars

NHSCDI

National Health System Cost Database for India

CGHS

Central Government Health Scheme

Author contributions

NT, SAP, SM, and MKA conceptualized the I-TREC program and obtained funding. NT, SAP, and SM designed the baseline needs assessment and evaluation protocol. SAP, HS, PJ, RS, TU, and MA developed data collection protocols, and NT and SM provided additional review of protocols. EK developed the costing analysis protocol, conducted data analysis, and prepared the initial draft of the manuscript. NP and SP provided guidance on economic evaluation methods and data sources for unit costs. All authors provided critical inputs to the manuscript and approved its final submission.

Funding

This study is supported in part by the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health (NIH), award number 5U01HL138635 under the Hypertension Outcomes for T4 Research within Lower Middle-Income Countries (Hy-TREC) program. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Data availability

The datasets used during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The I-TREC data collection protocol was reviewed and approved by the ethics committee at All India Institute of Medical Sciences (AIIMS), New Delhi, India (IEC-361/07.07.2017). Collaboration by Emory University researchers was determined to be Not Human Subjects Research (IRB00098808). All participants provided informed consent prior to participation. Research was conducted in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.