2027 Public Health PhD Research: Assessing Competing Risks in a Population Cohort
Audited and Verified:
Our Editorial and Data Verification Process
Research.com uses government datasets, official education statistics and primary institutional information to support its career and degree guides. Our review process checks factual accuracy, occupational definitions, source dates and the requirements associated with the credentials discussed.
How This Article Was Fact-Checked
Our editorial team verifies salary estimates, employment projections, education requirements and program information against original data providers and official organizations. Salary figures are checked against the relevant occupational definitions and reporting periods. Information about licensure, accreditation and required in-person training is reviewed separately.
Fact-checked and verified on:
Co-Founder and Chief Data Scientist
A cohort study can give a misleading picture of disease risk if participants die from another cause before they could experience the outcome being studied. For prospective public health PhD researchers, that makes competing risks both a methodological problem and a test of whether a dissertation is feasible.
This guide explains how to define the research question, choose and interpret an analysis, and assess the supervision, data access, training, and funding needed to carry it out. It is intended for applicants and graduate methods learners deciding whether doctoral research is the right next step.
Key Things You Should Know
- A competing event prevents the outcome of interest from occurring; for example, death from another cause precludes a later death from the cause under study.
- Choose the estimand before the model: a cause-specific hazard addresses an event rate among people still event-free, while cumulative incidence estimates an outcome's probability by a stated time in the presence of competing events.
- A viable PhD project needs more than a cohort dataset: confirm event definitions, follow-up, data permissions, appropriate methods supervision, and the terms of doctoral funding before committing.
What Is Competing Risks Research in a Public Health Population Cohort?
A population cohort follows a defined group from a specified starting point to observe later outcomes. A competing event is an event that makes the outcome of interest impossible. In a study of death from cardiovascular disease, death from another cause is competing: a participant cannot subsequently die from cardiovascular disease. A participant who is merely lost to follow-up is instead censored; their later outcome is unknown, not impossible.
This distinction matters when estimating the probability that an outcome will occur within a period such as 10 years. Treating competing deaths as ordinary censoring in a standard Kaplan-Meier calculation can overstate the observed-world probability of the outcome of interest. The size and direction of an exposure's association with different event types also matter: an exposure associated with both cardiovascular and other-cause mortality may change who remains able to experience either outcome.
The relevant cohort question is narrower than "Does this exposure affect health?" It specifies the population, entry date, exposure or comparison, mutually exclusive first-event categories, follow-up period, and quantity to be estimated. Applicants still choosing between broad population-health training and a research doctorate can use an overview of public health graduate study areas and career outcomes to clarify whether sustained original research is their goal.
How Does a Public Health PhD Study Competing Risks and Cohort Outcomes?
A strong dissertation starts with an answerable public health decision. A doctoral student might ask whether an exposure is associated with cardiovascular death by a defined age, while accounting for deaths from other causes. Before fitting a model, the student and advisor must decide whether the primary objective is to describe observed risk, investigate an association, predict individual outcomes, or estimate a causal effect. Those goals require different assumptions and should not be presented as interchangeable.
The research plan then links a cohort's recruitment and follow-up procedures to an analysis: who enters the risk set, when follow-up begins, how event dates and causes are established, and whether the data contain important confounders. A dissertation may also examine differences across populations, but subgroup analyses require adequate event counts and careful interpretation; a difference in modeled estimates is not automatically evidence of a biological difference.
Doctoral fit depends on a supervisor who can critique the epidemiologic question and access to statistical expertise in survival analysis. Ask prospective advisors who would review the estimand and code, whether the cohort permits the proposed analyses, and what happens if data access is delayed. If the immediate goal is to learn population-health analysis rather than develop an original research program, an MPH may be more proportionate; compare its actual curriculum and the CEPH accreditation for online MPH study rather than assuming it offers PhD-level methods training.
What Methods Are Used to Analyze Competing Risks in Population Cohorts?
Method choice follows the estimand, not a preference for a familiar software command. The comparison below distinguishes common approaches for a first event when competing event types have been defined clearly.
| Approach | What it answers | Interpretive limit |
| Aalen-Johansen cumulative incidence estimate | What proportion is estimated to experience a specified event by a given time, accounting for competing events? | It describes observed-world risk; it does not by itself adjust for confounding. |
| Cause-specific hazard model | How does a covariate relate to the instantaneous event rate among people still free of any event? | Its hazard ratio is not a direct ratio of cumulative probabilities. |
| Fine-Gray subdistribution hazard model | How does a covariate relate to the subdistribution hazard underlying cumulative incidence? | Its coefficient is not an ordinary event-rate effect or automatically a causal effect. |
A cause-specific analysis can be valuable for studying event processes, while cumulative incidence is usually more directly useful for communicating observed-world probability. Fine-Gray regression can model covariate associations with cumulative incidence, but a cause-specific hazard ratio and a subdistribution hazard ratio answer different questions and may differ without either model being erroneous. If the question concerns what risk would be under an intervention, neither regression alone resolves confounding or defines the intervention.
Check whether event categories are reliably recorded, censoring is plausibly independent under the analysis conditions, model assumptions fit the data, and missing exposure or covariate values could distort results. Delayed entry, changing exposures, survey weights, or repeated nonterminal events can require more specialized designs. For researchers who chiefly need these analytical skills rather than a dissertation, biostatistics master's format and scheduling comparisons provide a more focused educational alternative.
Public health schools offer epidemiology and biostatistics PhD pathways, but a program title alone does not establish competing-risks expertise. An epidemiology program may emphasize cohort design, substantive disease research, and causal inference; a biostatistics program may emphasize statistical theory and method development. Either can support this project if the relevant faculty, courses, computing resources, and accessible cohort data are present.
Build a shortlist from current university catalogs and faculty research, then ask whether a prospective supervisor is accepting students and has time to guide this particular question. Check who controls the cohort, whether a data-use agreement permits student work and publication, and whether a second committee member can review the survival methods. An advertised dataset or related faculty publication is not a promise of access or supervision.
Verify the exact degree, university recognition, and any claimed public health-specific accreditation against the accreditor's stated scope; accreditation of a school or MPH should not be assumed to cover every doctorate. Lists of affordable distance learning public health degrees can help identify institutions to investigate, but they cannot establish that a listed school offers a suitable, funded PhD research group.
Online coursework may reduce relocation and commuting, but a research doctorate still depends on sustained meetings, committee feedback, secure data systems, and often teaching or research appointments. Campus study can make collaboration and seminars easier to access, yet its schedule and relocation costs may be prohibitive. Neither format is inherently better for competing-risks research; supervision and permitted data access carry more weight than the classroom delivery method.
Ask each program which courses meet live, whether residencies or campus visits are required, where assistantship duties must be performed, and whether restricted cohort data can be analyzed remotely. Working applicants should also obtain a realistic estimate of weekly research and meeting commitments. An accelerated online course may help someone complete earlier preparation, but budget-conscious accelerated public health education is not a substitute for the time required to obtain approvals, conduct defensible analyses, and write a dissertation.
Relevant doctoral training connects epidemiologic study design with survival analysis, longitudinal data, statistical inference, and responsible research conduct. Depending on the project, students may also need causal inference, missing-data methods, measurement of exposures and outcomes, or health equity approaches. Course titles vary, so compare syllabi and prerequisites rather than assuming that one "advanced epidemiology" course covers competing risks.
A useful preparation sequence moves from regression and probability through survival analysis, then to a supervised cohort analysis in which the student defines the time origin, event coding, estimand, diagnostics, and reproducible workflow. Statistical programming matters, but competence means being able to explain why an analysis answers the question - not simply running a package.
Applicants from computing, social science, clinical, or other fields need not have a public health bachelor's degree if a program accepts their preparation. Those still completing an undergraduate credential can compare fast-paced online public health bachelor's courses, while checking whether the resulting statistics and mathematics courses actually meet prospective PhD prerequisites. An applicant who already has a bachelor's degree may be better served by targeted quantitative coursework or supervised research.
Requirements differ by institution and discipline. Programs commonly assess academic preparation, research interests, references, and evidence that an applicant can undertake sustained inquiry. Some specify prior degrees or quantitative prerequisites; policies on tests, theses, publications, and contacting faculty vary. Meeting minimum eligibility is different from showing that a proposed cohort project fits available supervision.
For this topic, a persuasive application connects prior coursework or research to a feasible question. It should demonstrate how the applicant has handled data or evaluated evidence, identify remaining methods gaps honestly, and explain why particular faculty and cohorts are relevant. Before applying, confirm prerequisite courses, application procedures, faculty availability, and whether admission comes with a funding offer.
An MPH can provide a bridge for an applicant missing population-health foundations, but it is not a universal prerequisite. Applicants considering that route because of academic-record concerns can review guidance on addressing GPA concerns in an MPH application; they should still ask the PhD program which evidence of recent quantitative and research performance it values.
A research PhD typically spans multiple years of coursework, qualifying or candidacy milestones, analysis, and dissertation writing. The actual timeline depends on program policy, prior preparation, event-data access, advisor feedback, and whether the research question remains feasible. Restricted data approvals or a change in cohort availability can delay a project even after coursework is complete.
Compare written funding offers rather than advertised tuition alone. Check which terms include tuition remission, required fees, a stipend, health coverage, and summer support; whether support depends on teaching, grant renewal, or satisfactory progress; and what charges apply if dissertation work continues beyond funded years. Relocation and forgone earnings also affect the decision. Guidance on reducing epidemiology doctorate costs through assistance can help frame the questions, but only the institution's offer establishes an applicant's actual terms.
Competing-risks research can prepare a graduate for epidemiology, biostatistics, health-services research, and other roles involving cohort design, analysis, and communication of population risks. Academic positions may also require teaching, grant development, and a publication record; a dissertation does not guarantee a faculty appointment. In government, health systems, and research organizations, responsibilities and degree requirements vary by employer.
The clearest career evidence is a portfolio of authorized, reproducible work: a well-defined research question, defensible analysis, transparent limitations, and writing for audiences beyond a methods committee. Applicants aiming mainly for an analyst role should compare employer requirements with the time and opportunity cost of a doctorate. A relevant MS, supervised research position, or low-cost public health master's degree may reach that goal more directly than a PhD.
How Much Do Public Health Researchers Earn and Where Do They Work?
There is no single salary for "public health researcher." US Bureau of Labor Statistics occupation categories, such as epidemiologists and statisticians, describe different jobs and include workers with varied degrees and experience. Their occupation-wide wages should not be presented as the expected earnings of a public health PhD graduate. A prospective student should compare current, location-specific postings for the roles they want, noting required methods skills, seniority, and whether a doctorate is necessary.
Potential workplaces include universities, government agencies, health systems, and private or nonprofit research organizations. Compensation and advancement can differ substantially between a grant-funded academic appointment and an experienced quantitative role elsewhere. When weighing those possibilities against years in doctoral study, review the actual funding offer and costs; a guide to low-cost PhD study in public health is a starting point for comparison, not evidence of a particular program's net price or career return.
Other Things You Should Know About Public Health
No. Death is a competing event when it prevents the outcome being studied, such as a first hospitalization. Whether an event competes depends on the precisely defined outcome and follow-up question.
Only if the combined outcome answers a meaningful research question. A composite can obscure important differences when its components have different public health implications or associations with an exposure.
No. It addresses how events and risks are estimated. Causal claims additionally require a defensible comparison, attention to confounding and bias, and assumptions appropriate to the study design.
Possibly, but access does not automatically grant publication rights. Confirm data-use and disclosure rules, required reviews, and authorship expectations with the data steward and research team before starting.
- Survival analysis in the presence of competing risks - PMC - NIH https://pmc.ncbi.nlm.nih.gov/articles/PMC5326634/
- Practical recommendations for reporting Fine-Gray model analyses for competing risk data - PMC https://pmc.ncbi.nlm.nih.gov/articles/PMC5698744/
- Competing Risk Analysis: What Does It Mean and When Do We Need It in Orthopedics Research? - PMC https://pmc.ncbi.nlm.nih.gov/articles/PMC8478701/
- Competing Risk Analysis | Columbia University Mailman School of Public Health https://www.publichealth.columbia.edu/research/population-health-methods/competing-risk-analysis
- A practical guide to handling competing events in etiologic time-to-event studies - ScienceDirect https://www.sciencedirect.com/science/article/pii/S2590113322000104
