For an MPH or M.Sc Public Health dissertation in India, the only nationally representative dataset you can download tonight with unit-level records and district identifiers is NFHS-5 (2019–21): 636,669 households, 724,115 women and 101,839 men, free for academic use after registration with the DHS Program. Every other national source either arrives as published tables (SRS, HMIS, Census) or needs its own registration and a longer wait (NSS, LASI, GBD).
That single fact settles most data-chapter decisions, and it is the reason NFHS appears in a large share of Indian public health dissertations. The rest of this guide gives you the comparison behind it — what each source holds, how old its latest release is, whether you get individual records or only tables, and how you get in — followed by a matching table from dissertation question to dataset, the NFHS access procedure step by step, and the mistakes that examiners at Indian schools of public health flag most often. Every access point was tested on 30 August 2026; where a portal was unreachable, the table says so.
The seven national sources compared
| Source | Custodian | What it holds | Latest release (year) | Unit-level records? | How you get it |
|---|---|---|---|---|---|
| National Family Health Survey, NFHS-5 | IIPS Mumbai for the Ministry of Health and Family Welfare; distributed by the DHS Program | Household survey: fertility, family planning, maternal and child health, nutrition, anaemia, NCD screening, domestic violence, household assets | Fieldwork June 2019 – April 2021; final report and datasets released 2022 | Yes — household, woman, man, child and birth recode files, plus GPS cluster files | Free registration on dhsprogram.com; access granted per country for a stated research project |
| Sample Registration System, SRS | Office of the Registrar General and Census Commissioner | Annual birth rate, death rate, infant mortality, total fertility, age-specific rates, life tables and maternal mortality bulletins, by state and rural–urban | SRS Statistical Report 2024 listed in the Census Digital Library (accessed August 2026) | No — published tables only | Open download, no registration |
| Health Management Information System, HMIS | Ministry of Health and Family Welfare | Monthly facility-reported service data: antenatal registrations, institutional deliveries, immunisation, OPD and IPD counts, down to district and block | Monthly; data continuously updated | No — aggregate reports by facility, block, district | Public portal at hmis.mohfw.gov.in returned an error page on 30 August 2026; NDAP republishes HMIS tables but is a JavaScript-only site |
| NSS Household Social Consumption: Health | National Statistics Office, MoSPI | Household survey on illness episodes, hospitalisation, out-of-pocket expenditure, insurance coverage, source of care | NSS 80th round, January – December 2025, public-use unit data posted on microdata.gov.in (page updated August 2026); earlier round NSS 75th, 2017–18 | Yes — public-use unit-level files | Free registration on the NSO microdata catalogue (microdata.gov.in) |
| Longitudinal Ageing Study in India, LASI | IIPS Mumbai with Harvard and USC | Panel of adults aged 45 and above and their spouses: health, cognition, biomarkers, work, family, pensions | Wave 1 fieldwork 2017–18; Wave 2 fieldwork 2024–25 (IIPS); Wave 1 dried-blood-spot biomarker report released | Yes — on application | Data request through IIPS; harmonised file through the Gateway to Global Aging Data |
| Global Burden of Disease, GBD 2023 | Institute for Health Metrics and Evaluation | Modelled estimates for 375 diseases and injuries and 88 risk factors, 1990–2023, for 204 countries and 660 subnational locations including Indian states | GBD 2023 results published November 2025 | No — modelled estimates, not respondent records | Free account on the GBD Results Tool |
| Census of India 2011 | Office of the Registrar General | Population denominators, houselisting and amenities, disability, migration, literacy, down to village and ward | 2011 tables; Census Digital Library catalogue lists 31,245 tables and 8,832 documents (accessed August 2026); the next Census is in the field and no new tables are out | No — tables; the sample microdata is not public | Open download from the Census Digital Library |
Read the fifth column first. A dissertation that runs its own regression, computes its own prevalence with confidence intervals, or tests an association needs unit-level records, and in India that means NFHS, the NSS health round or LASI. A dissertation that describes trends, compares states or builds a denominator uses the table sources — SRS, HMIS, Census and GBD — and its analysis chapter is about comparison and visualisation rather than inference. Mixing the two is normal; confusing them in the methodology chapter is where trouble starts.
Which source answers which dissertation question
| If your dissertation asks… | Use | Why |
|---|---|---|
| Which household or maternal factors are associated with child stunting, anaemia or full immunisation in a state or district? | NFHS-5 child and woman recode files | Individual records with district codes, sampling weights and the outcome measured on the child |
| How has infant mortality or total fertility changed in a state over ten years? | SRS Statistical Reports, year by year | The only annual, registration-based series; NFHS is five-yearly and survey-based |
| Did institutional deliveries or antenatal registrations in a district change after a programme was introduced? | HMIS monthly reports | Facility-reported counts month by month, at block level, before and after the intervention date |
| What did households spend out of pocket on a hospitalisation, and who was insured? | NSS health round (80th round, 2025; 75th round, 2017–18) | NFHS does not collect expenditure; NSS is the national expenditure survey for health |
| What predicts hypertension, depression or functional limitation among older adults? | LASI Wave 1, with Wave 2 for change over time | Only national panel of people aged 45 and above with measured biomarkers |
| What share of the disease burden in a state is attributable to a named risk factor? | GBD 2023 Results Tool | Risk-attributable DALYs by Indian state; no household survey produces these |
| What is the denominator for a rate in a block, tehsil or ward? | Census 2011 tables, projected forward | The only source at that geography; state the projection method |
Two of these are not interchangeable even though students try: NFHS anaemia and NCD figures are survey measurements on a sample, while HMIS figures are counts of people who came to a facility. A rise in HMIS hypertension screenings says something about the programme, not about prevalence. Your guide will ask which of the two your research question actually needs, and the answer decides the chapter.
How to get NFHS-5 unit-level data, step by step
- Register on dhsprogram.com. Registration is required for every dataset; the DHS Program states that it is authorised to distribute unrestricted survey data files at no cost for legitimate academic research.
- Create a project and request India. Access is requested and granted by country, so approval for India opens every unrestricted India dataset — NFHS-5, NFHS-4 (2015–16), NFHS-3 (2005–06) and the earlier rounds — for that one project. The request must state the research purpose; a new purpose needs a new request.
- Choose the recode file that carries your outcome. Household (HR) for assets, water, sanitation and the wealth index; individual woman (IR) for maternal care, family planning and violence; child (KR) for anthropometry, immunisation and anaemia in children under five; birth (BR) for every birth in the five years before the survey; man (MR) for the men’s sample; household member (PR) for every listed person, including adult blood pressure and glucose measurements.
- Pick a format your software reads. Files come as Stata, SPSS and SAS, with the flat ASCII version and a codebook. Apply the sampling weight (the weight variable divided by one million) and the cluster and strata variables in every estimate — an unweighted prevalence from NFHS is a methodological error, not a stylistic one.
- Request the GPS file separately if you need geography beyond the district. NFHS-5 cluster coordinates are available with displacement; the acknowledgement of conditions is a separate online step.
- Cite the dataset and the report. The final report is DHS Final Report FR375 (India 2019–21), with state reports in FR374 and district fact sheets published for every district. Cite the dataset with its file name and version, and the report for any figure you quote rather than compute.
The other registration-based sources follow the same logic with different forms. The NSO microdata catalogue asks for a login and states the conditions of use before a public-use file downloads. LASI requires an application to IIPS, and its harmonised file — built to match the Health and Retirement Study family of surveys — is served through the Gateway to Global Aging Data. The GBD Results Tool needs a free IHME account and returns a CSV you can build a table from in an afternoon.
What is not available, stated plainly
- NFHS-6 unit data. IIPS tender notices place NFHS-6 fieldwork in 2023–24. At the time of writing the DHS Program lists India 2019–21 as the most recent survey with datasets available, so NFHS-5 remains the release you can download. Check the DHS survey list before you commit a dissertation to NFHS-6.
- SRS microdata. The SRS is a continuous registration system in sample units; its individual records are not released. Anything below the state and rural–urban level is not in the reports.
- A public HMIS extract on the day you need it. The portal was unreachable from outside a campus network when tested for this guide. Plan to download on campus, save every extract with its date, and keep the exported file with the dissertation, because the same query can return revised figures a month later.
- A district-level mental health prevalence series. The National Mental Health Survey 2015–16 was a twelve-state study and its state reports are the finest geography it offers; NFHS does not measure mental health. State that no national dataset exists rather than borrowing a figure from a single-city study.
- Census 2021 or later tables. None are published. Use 2011 denominators and say how you projected them.
The five mistakes examiners flag in secondary-data dissertations
- Unweighted estimates from a weighted survey. NFHS, NSS and LASI are all multistage samples with weights; every prevalence and every regression must use the survey design. Name the survey-set command you used.
- A district analysis with too few clusters. Some NFHS districts have a small number of clusters and a handful of children in a specific age band; the district fact sheet reports an indicator, but your subgroup estimate may rest on twenty observations. Report the unweighted n for every cell and consider pooling adjacent districts. The general method for deciding whether a subsample is large enough is in our guide to calculating sample size for a thesis.
- Treating HMIS counts as prevalence. Facility counts have a denominator problem — people who did not attend are invisible. Describe them as service statistics and compare them with the survey source where both exist.
- Mixing reference periods. NFHS-5 fieldwork spanned the pandemic; phase-one and phase-two states were surveyed at different times. Name the phase your state belongs to and the months of fieldwork in the methodology chapter.
- No ethics statement because the data is public. Most institutional ethics committees treat secondary analysis of anonymised public datasets as exempt from full review, but the exemption is granted by the committee, not assumed by the scholar. Submit the protocol; the process is set out in our walkthrough of obtaining ethics committee clearance for an Indian thesis.
Once the dataset is chosen the analysis follows from the design. A binary outcome across NFHS households points to survey-weighted logistic regression; a comparison of state rates from SRS points to a trend description with a clearly stated absence of inference — the choice is mapped in our decision table for choosing a statistical test, and the reporting format in our guide to writing the results chapter with APA tables.
Where this sits among the site’s data guides
This is the public health entry in a set of discipline-specific data guides. The general map of official Indian sources by discipline is in where to find data for your Indian thesis, and the economics counterpart — RBI, MoSPI, EPWRF and CMIE — is in our guide to data sources for an M.A. Economics dissertation. If your public health question is about health financing or household expenditure, the NSS material in that economics guide applies directly.
Documenting the source — custodian, round, fieldwork dates, file name, weight variable, download date — is the part of the methodology chapter that secondary-data dissertations most often leave thin. Tesify drafts that data-source section from your own notes in the structure Indian schools of public health expect, and keeps the citation for each dataset consistent from synopsis to submission; the analysis and the interpretation remain yours.
Draft your public health data-source section with Tesify — free to start
Frequently asked questions
Is NFHS-5 data free for students in India?
Yes. The DHS Program distributes unrestricted survey data files at no cost for legitimate academic research. You register, describe the research project and request India; approval opens all unrestricted India datasets for that project.
Can I get NFHS data for a single district?
Yes. Every NFHS-5 record carries a district code, and district fact sheets were published for every district. Check the unweighted number of observations in your district before planning a subgroup analysis, because some districts have few clusters.
Is NFHS-6 data available yet?
IIPS tender notices place NFHS-6 fieldwork in 2023–24, but at the time of writing the DHS Program lists India 2019–21 as the latest survey with datasets available. Until a release appears on the DHS survey list, NFHS-5 is the unit-level dataset you can download.
Which source has data on out-of-pocket health expenditure?
The NSS Household Social Consumption: Health rounds — the 80th round covering January to December 2025 and the 75th round covering 2017–18 — with public-use unit-level files on the NSO microdata catalogue. NFHS does not collect health expenditure.
Can I use HMIS data in a dissertation?
Yes, as service statistics rather than prevalence. Download on a campus connection, record the extraction date, keep the exported file, and describe the figures as facility-reported counts with the denominator limitation stated.
Do I need ethics clearance for secondary analysis of NFHS or NSS data?
Most institutional ethics committees treat anonymised public datasets as exempt from full review, but the exemption is granted by the committee. Submit the protocol and keep the exemption letter for the appendix.
Where do I get older-adult health data for India?
LASI, the Longitudinal Ageing Study in India, run by IIPS: Wave 1 was conducted in 2017–18 and Wave 2 fieldwork in 2024–25. Unit-level data is available on application to IIPS, and a harmonised version through the Gateway to Global Aging Data.
Is GBD data acceptable as a primary source in an Indian dissertation?
It is acceptable for describing burden and risk attribution by state, provided you state that GBD figures are modelled estimates rather than survey measurements and cite the GBD 2023 round and the Results Tool query you ran.