Systematic Reviews of Prevalence and Incidence: A Different Kind of Question

Not every systematic review asks whether an intervention works. Reviews estimating how common a condition, exposure, or characteristic is within a population -- prevalence and incidence reviews -- ask a genuinely different kind of question, and this difference shapes nearly every methodological choice throughout the review.

Framing the question differently

Rather than PICO's intervention-and-comparator structure, a prevalence review typically frames its question around a condition, a population, and a context -- what proportion of a defined population has a given condition, in what setting, over what time period. This is sometimes described using a CoCoPop framework, Condition, Context, Population, reflecting the genuinely different question structure compared to an intervention effectiveness review.

Why study design expectations differ

Prevalence and incidence questions are typically answered by cross-sectional or cohort studies reporting how many people in a defined population meet a specific condition definition, not by randomized controlled trials, which are designed to test an intervention's effect rather than estimate how common something naturally occurring is. A prevalence review's eligibility criteria should reflect this, specifying appropriate observational study designs rather than defaulting to RCT-focused criteria that wouldn't sensibly apply to this kind of question.

A dedicated risk-of-bias tool: the JBI prevalence checklist

Standard risk-of-bias tools built for intervention studies don't map well onto prevalence research's specific concerns. JBI provides a dedicated critical appraisal checklist for prevalence studies, assessing domains including whether the sample frame was appropriate for addressing the target population, whether sampling methods adequately represented that population, whether sample size was adequate, and whether the condition was measured using valid, reliable methods -- concerns specific to prevalence estimation rather than intervention comparison.

Pooling prevalence estimates statistically

Meta-analysis of prevalence data uses proportions as the effect measure rather than risk ratios or mean differences, and raw proportions close to 0 or 1 have statistical properties that can cause problems with standard pooling methods, which is why prevalence meta-analyses commonly apply a variance-stabilizing transformation, such as the Freeman-Tukey double arcsine transformation, before pooling, then back-transform the pooled result to a more directly interpretable proportion.

Heterogeneity in prevalence reviews

Heterogeneity in prevalence estimates across studies is often expected and substantial, reflecting genuine differences in the populations, settings, and time periods studied, rather than necessarily indicating a methodological problem the way unexplained heterogeneity might in an intervention review. Subgroup analysis by region, setting, or population characteristic is frequently a central, planned part of a prevalence review's analysis, rather than a secondary or exploratory addition.

Reporting with PRISMA, appropriately adapted

Standard PRISMA 2020 still provides the overall reporting structure for a prevalence systematic review, though some items need sensible adaptation given the different question type -- your eligibility criteria section should describe your CoCoPop framing rather than PICO, and your risk-of-bias section should describe your prevalence-specific appraisal tool rather than RoB 2 or ROBINS-I.

Incidence reviews: a related but distinct question

Incidence, the rate of new cases over a specified time period, is related to but distinct from prevalence, which is a snapshot proportion at one point in time. A review examining incidence needs eligibility criteria and extraction fields specifically capturing the time period over which new cases were counted, and appropriate statistical methods for pooling rate data, which differ somewhat from the proportion-pooling methods appropriate for straightforward point prevalence.

Why this distinction matters for interpretation

A pooled prevalence estimate answers "how common is this condition right now, in this kind of population," while a pooled incidence estimate answers "how quickly are new cases arising." Conflating the two, or extracting data without being careful about which one a given primary study actually reports, produces a result that doesn't actually answer either question correctly, which is a surprisingly easy mistake to make during extraction if this distinction isn't kept explicitly in mind throughout.

A practical takeaway for planning this kind of review

Recognizing early that a prevalence or incidence question needs its own dedicated appraisal tool, its own statistical pooling approach, and its own adapted PRISMA reporting structure -- rather than defaulting to intervention-review conventions out of habit -- is the single most important methodological decision point for this kind of systematic review, and getting it right from the protocol stage prevents a considerable amount of rework later.

Communicating prevalence findings to a policy audience

Prevalence estimates frequently inform resource allocation and public health planning decisions, and translating a pooled prevalence estimate, along with its confidence interval, into concrete terms relevant to a specific population size or planning context often does more to make a prevalence review practically useful than the raw statistical output alone, which can otherwise remain abstract to a non-specialist policy audience reading your findings. Including a brief, plainly worded summary of practical implications alongside your full statistical results, rather than assuming readers will make this translation themselves, considerably increases how usable your review actually becomes for the audiences most likely to act on a prevalence estimate in practice. This translation step is a relatively small addition to an already completed statistical analysis, but it is frequently the specific part of a prevalence review that a policymaker or program planner actually reads and acts upon, making it disproportionately valuable relative to the modest additional effort it requires. Building this translation step into your standard reporting template for prevalence work, rather than treating it as an optional addition only included when time permits, ensures your findings consistently reach the practical audiences most likely to benefit from and act on them. Ultimately, a prevalence or incidence review's value is measured not just by its statistical rigor but by how effectively its findings actually reach and inform the people making decisions based on how common a given condition genuinely is. Keeping that end use in view throughout the analysis shapes a genuinely more useful final product. That practical connection between statistical output and real-world decision-making is what ultimately gives this kind of review its lasting value. Keeping this end use in mind throughout shapes a genuinely more useful final product.