How to Choose the Right Study Type for Your Systematic Review

The study type you choose for your systematic review is not just a label — it is the single decision that determines your entire methodology. It dictates which PRISMA reporting guideline you follow, which risk of bias tool you use, what kind of statistical analysis is appropriate, and how you assess the certainty of evidence.

Most systematic review platforms treat study type as an afterthought. In mapped, it is the first decision you make, and everything downstream adapts accordingly.

This guide explains the five study types supported in evidence synthesis, when each is appropriate, and how your choice cascades through every step of the review.

The Five Study Types

1. Pairwise Reviews

Pairwise reviews ask: does an intervention (treatment, therapy, program) produce a desired effect compared to a control?

This is the most common type of systematic review. The research question follows the PICOS framework:

Population: Who receives the intervention?
Intervention: What is being tested?
Comparison: What is the alternative (placebo, standard care, no treatment)?
Outcome: What effect is measured?
Study design: What types of studies are eligible (typically RCTs)?

Methodology cascade:

Framework: PICOS
PRISMA variant: PRISMA 2020
Risk of bias: RoB 2 (for RCTs) or ROBINS-I (for non-randomized studies)
Analysis: Pairwise meta-analysis (fixed or random effects using metafor)
Certainty: GRADE
Visualizations: Forest plots, funnel plots

Example question: "Does cognitive behavioral therapy reduce anxiety symptoms in adults with generalized anxiety disorder compared to waitlist control?"

2. Prognostic Reviews

Prognostic reviews ask: is a specific factor associated with a clinical outcome over time?

These reviews evaluate prognostic factors (variables that predict outcomes) or prognostic models (combinations of factors that predict outcomes). They differ from pairwise reviews because they do not evaluate the effect of an intervention.

Methodology cascade:

Framework: PICOS (adapted — "I" becomes the prognostic factor, "C" is often absent)
PRISMA variant: PRISMA 2020
Risk of bias: NOS (Newcastle-Ottawa Scale) or ROBINS-I
Analysis: Pairwise meta-analysis with appropriate effect measures (HR, OR)
Certainty: GRADE
Visualizations: Forest plots, funnel plots

Example question: "Is elevated baseline CRP associated with increased mortality in patients with heart failure?"

3. Diagnostic Test Accuracy (DTA) Reviews

DTA reviews ask: how accurately does a test identify a target condition?

Instead of measuring treatment effects, DTA reviews evaluate the sensitivity and specificity of diagnostic tests against a reference standard. The statistical methods and reporting requirements are fundamentally different from pairwise reviews.

Methodology cascade:

Framework: PICOS (adapted — "I" is the index test, "C" is the reference standard)
PRISMA variant: PRISMA-DTA
Risk of bias: QUADAS-2
Analysis: Bivariate meta-analysis using mada (hierarchical summary ROC model)
Certainty: GRADE-DTA
Visualizations: SROC curves, coupled forest plots (sensitivity and specificity side by side)

Example question: "What is the diagnostic accuracy of point-of-care ultrasound for detecting pneumothorax compared to CT scan?"

4. Network Meta-Analysis (NMA)

NMA reviews ask: which of multiple interventions is most effective for a given condition?

NMA extends pairwise reviews by comparing three or more treatments simultaneously, combining direct and indirect evidence. It produces rankings and estimates for all pairwise comparisons.

Methodology cascade:

Framework: PICOS
PRISMA variant: PRISMA-NMA
Risk of bias: RoB 2 or ROBINS-I
Analysis: Network meta-analysis using netmeta (frequentist)
Certainty: CINeMA (replaces standard GRADE)
Visualizations: Network geometry plot, league tables, rankograms, comparison-adjusted funnel plots

Example question: "Which antidepressant is most effective for acute treatment of major depression in adults: SSRIs, SNRIs, TCAs, or placebo?"

5. Scoping Reviews

Scoping reviews ask: what is the breadth and nature of the evidence available on a topic?

Scoping reviews map the literature rather than synthesize effect estimates. They do not perform quality assessment or meta-analysis — they characterize what is known, identify gaps, and inform future reviews.

Methodology cascade:

Framework: PCC (Population, Concept, Context)
PRISMA variant: PRISMA-ScR
Risk of bias: Not applicable
Analysis: Descriptive charting (no meta-analysis)
Certainty: Not applicable (no GRADE)
Visualizations: Evidence maps, frequency tables

Example question: "What is the extent and nature of research on telemedicine for chronic disease management in low-income countries?"

The Decision Tree

What is your research question asking?
│
├─ Does an intervention work? (single comparison)
│  └─ Pairwise Review
│
├─ Which of several treatments works best?
│  └─ Network Meta-Analysis (NMA)
│
├─ How accurate is a diagnostic test?
│  └─ Diagnostic Test Accuracy (DTA)
│
├─ Is a factor associated with a clinical outcome?
│  └─ Prognostic Review
│
└─ What do we know about this topic?
   └─ Scoping Review

How Study Type Affects Every Step

Step	Pairwise	Prognostic	DTA	NMA	Scoping
Framework	PICOS	PICOS (adapted)	PICOS (adapted)	PICOS	PCC
Search	Standard	Standard	DTA-specific filters	Standard	Broad
Screening	Title/abstract + full-text	Title/abstract + full-text	Title/abstract + full-text	Title/abstract + full-text	Title/abstract + full-text
Data extraction	Effect estimates	Prognostic factors, HRs	2x2 tables (TP, FP, FN, TN)	Effect estimates per comparison	Data charting
Risk of bias	RoB 2 / ROBINS-I	NOS / ROBINS-I	QUADAS-2	RoB 2 / ROBINS-I	Not applicable
Analysis	Pairwise MA	Pairwise MA	Bivariate MA	Network MA	None
R engine	metafor	metafor	mada	netmeta	None
Certainty	GRADE	GRADE	GRADE-DTA	CINeMA	None
PRISMA	2020	2020	PRISMA-DTA	PRISMA-NMA	PRISMA-ScR
Manuscript	Standard SR	Standard SR	DTA SR	NMA SR	Scoping review

Why This Matters

Choosing the wrong study type leads to methodological mismatches that reviewers and editors will catch:

Using PICOS for a scoping review (should be PCC)
Applying RoB 2 to diagnostic accuracy studies (should be QUADAS-2)
Pooling sensitivity and specificity in a standard forest plot (should use bivariate models and SROC curves)
Reporting GRADE for a scoping review (scoping reviews do not assess certainty)
Using pairwise meta-analysis when multiple treatments need comparison (should be NMA)

These are not minor issues — they can result in desk rejection or major revisions.

Study Type Selection in mapped

When you create a new project in mapped, the first step is selecting your study type. This single selection configures the entire platform:

The workflow steps that appear (or are hidden) are determined by your study type
The correct PRISMA variant is enforced
The appropriate risk of bias tool is pre-selected
The meta-analysis method matches your study design
The certainty assessment (GRADE vs CINeMA vs none) is set automatically
The manuscript template follows the correct reporting structure

You cannot accidentally apply pairwise review methods to a scoping review, or use QUADAS-2 for a pairwise study. The platform enforces methodological consistency by design.