Published
The PRISMA-IPD checklist has 31 items, and several of them ask for things that a standard systematic review never has to report, such as how participant-level data were requested, how they were checked and how many participants each trial actually supplied. This guide applies every item to one real, open-access IPD meta-analysis, names the section where each answer sits, and marks the rows that need a judgement call. You will see what a complete answer looks like and where even a careful team leaves a small gap.
The checklist was published by Stewart and colleagues in JAMA in 2015 as a stand-alone extension to the PRISMA statement. If you need the general version first, read our PRISMA 2020 guide and then come back for the IPD rows. For another extension applied row by row, see our PRISMA-DTA worked example. This page covers the individual participant data (IPD) extension only. The full item wording is in the official PRISMA-IPD checklist PDF, which is also listed on the EQUATOR Network page for PRISMA-IPD, and we paraphrase it here.
In Simple Words
PRISMA-IPD is the reporting checklist for systematic reviews that collect and re-analyse the raw, participant-level data from each eligible study instead of relying on published summaries. It keeps the standard PRISMA items and adds rows on how the data were obtained, how they were checked and what the checks found, plus a row on variation between participants. Below, all 31 items are checked against a real published review, so you can see what to write and where to put it.
What the PRISMA-IPD checklist adds to PRISMA
An IPD review asks the team to collect, check and re-analyse individual-level data from the studies that address a question, not just to read their reports. That changes what a reader needs to know. They need to learn which studies were approached, which of them supplied data, what was done to the data before analysis, and how the pooled model treated participants who sit inside studies.
The extension came from a web-based survey and an international workshop of methodologists, clinicians and journal editors, and the group agreed the checklist and a flow diagram by consensus. Compared with standard PRISMA it adds three items, which cover checking the integrity of the data, reporting what the checks found and exploring variation in effects between participants. A fourth added item, A4, was created by rearranging the items on interpreting results, and the wording of 23 existing items was changed to suit the IPD approach. The official PRISMA-IPD checklist therefore numbers its rows 1 to 27 and labels four IPD rows A1 to A4, which gives 31 in all. A1 follows item 11, A2 follows item 14, A3 follows item 18 and A4 follows item 26.
The checklist appeared in 2015, before PRISMA 2020 (Page et al., BMJ 2021), so its base rows follow the earlier statement, and we found no updated version. IPD meta-analyses are now common. A 2026 brief report in BMC Medicine by Dal-Ré and colleagues counted 198 published in the first nine months of 2025 on a title-only PubMed search, against a first one in 1998, and argues that the next PRISMA-IPD update should add items the current checklist lacks.
The scope is wider than trials. The authors developed the checklist mainly for reviews of randomised trials and note that many items apply to reviews of diagnosis and prognosis, which is the situation in our example, a prognostic question answered with trial data. For the reasoning behind the method itself, Riley, Lambert and Abo-Zaid explain the rationale, conduct and reporting of IPD meta-analysis in the BMJ.
How we chose and checked the example
We searched PubMed for IPD meta-analyses that state adherence to PRISMA-IPD and offer full text without a paywall. We kept one whose licence we could confirm from three independent sources. Europe PMC flags the article as CC BY, the licence statement inside the full-text XML names a Creative Commons Attribution licence at CC BY 4.0, and the PubMed Central open-access dataset lists it as CC BY. Crossref holds no licence record for the article, so that source could not be used. The supplement also contains the authors’ own PRISMA-IPD checklist with page numbers, which let us compare our reading with theirs.
Each row of the PRISMA-IPD checklist below carries the section where the answer sits, the evidence in the article, and a call. Reported means the item is answered where the checklist expects it. Judgement call means two careful readers could score the row differently, or the numbers in the article need reconciling before they can be trusted. Every quotation is 25 words or fewer, sits in quotation marks, and was checked against the full text. This is a reporting check, so we did not re-analyse the data or assess the authors’ conclusions. Because the example obtained IPD from every eligible study, the parts of items 10, 17 and 23 that deal with studies lacking IPD were not triggered, and we say so rather than score them.
The IPD meta-analysis behind this example
The article is “Socioeconomic Indicators of Treatment Prognosis for Adults With Depression” by Buckman and colleagues, published in JAMA Psychiatry in 2022. It pools individual patient data from 9 randomised trials in UK primary care, 4864 patients in all, to ask whether employment, financial strain, housing and education predict depressive symptoms 3 to 4 months later, independent of the treatment given. It uses a two-stage random-effects model, QUIPS for risk of bias and GRADE for certainty, and its abstract states adherence to PRISMA-IPD.
The review suits a worked example because it has a real flow diagram that reports study teams contacted and study teams supplying data, a data-integrity section, a per-study characteristics table, pooled estimates with intervals, sensitivity analyses and GRADE tables. A review in which several trials declined to share data would test items 10, 17 and 23 harder. Its gaps are small and typical, which is what makes them useful.
We re-added every box in the figure and the counts reconcile. The database records sum to 164 (53, 50, 39, 22 and 0), the two screening exclusions of 35 and 55 leave 64 and then 9, and the 9 study teams contacted match the 9 that supplied data.
PRISMA-IPD checklist items 1 to 5 (title, abstract, introduction, protocol)
The opening rows are the easiest to satisfy, yet the abstract row asks for more than most journals’ formats allow, and the objectives row asks for hypotheses about participant-level subgroups, which many teams skip.
| Item | What the item asks | Where in the article | Evidence | Call |
|---|---|---|---|---|
| 1 | The report should say in its title that it is a systematic review and meta-analysis of individual participant data. | Title | The published title ends with “A Systematic Review and Individual Patient Data Meta-analysis”. The authors write “patient” where the checklist writes “participant”, which is a wording choice and does not hide the design. | Reported |
| 2 | Give a structured summary that notes IPD were sought, the date of the last search, the numbers obtained, effect estimates with intervals, the registration number and the funder. | Abstract | The structured abstract names the five databases and the last search date of October 8, 2021, states adherence to PRISMA-IPD, and reports 9 studies and 4864 patients with effect sizes and intervals. It gives no registry number or funder, which the item lists under Other, and no heterogeneity measure. The housing result is worded as 18% lower for homeowners in the abstract and as 17.6% higher for other housing in the Results. The two carry the same interval (6% to 30%) and, on a ratio scale, a rise of 17.6% for one group is not a fall of 17.6% for the other. | Judgement call |
| 3 | Describe the rationale in light of what is already known. | Introduction | The Introduction cites an earlier meta-review in which only 2 systematic reviews addressed the question, then states that “none have adjusted for the outcomes of known clinical prognostic factors” (Buckman et al.). That gap is the stated reason for pooling participant-level data. | Reported |
| 4 | State the questions with reference to participants, interventions, comparisons, outcomes and design (PICOS), plus any hypotheses about participant-level subgroups. | Introduction, last paragraph | Two aims are given, the association of four socioeconomic factors with prognosis independent of treatment, and whether those factors add to known prognostic factors. Population, exposures and outcome can be read from the aims. There is no PICOS statement and no hypothesis about participant-level subgroups. | Judgement call |
| 5 | Say whether a protocol exists, where to find it, and give the registration details. | Methods, Identification and Selection of Studies | The Methods point to a PROSPERO record, a general protocol reported to PRISMA-P and a separate preregistration of the analysis. The registry number sits in the reference list, not the running text. The eAppendix adds that registration came after IPD from four studies had been obtained and that the protocol was amended twice, which is full disclosure. | Reported |
PRISMA-IPD checklist items 6 to 11 and A1 (eligibility to IPD integrity)
This block holds most of what makes an IPD review different. Rows 10, 11 and A1 are where readers learn how the data were obtained, harmonised and checked, so read them closely. Row A1 is new in PRISMA-IPD and has no equivalent in the standard statement.
| Item | What the item asks | Where in the article | Evidence | Call |
|---|---|---|---|---|
| 6 | State inclusion and exclusion criteria, whether they were applied to studies or to individual participants, and the rationale. | Methods, Inclusion and Exclusion Criteria | Criteria cover design (randomised trials), age (16 and over), unipolar depression, at least one active arm, a socioeconomic measure and the CIS-R at baseline, plus clear exclusions. The rationale is stated, because the CIS-R guaranteed common data on key disorder characteristics. The text never says whether eligibility was applied to studies or to participants, although “All 4864 participants from the eligible RCTs were included” (Buckman et al.) suggests study level. | Judgement call |
| 7 | Describe every way studies were identified, with coverage dates and the date of the last search. | Methods; Abstract | Five databases were searched from inception to October 8, 2021, reference lists were hand searched and experts were contacted for unpublished or missed studies. | Reported |
| 8 | Print a repeatable electronic search strategy for at least one database, including limits. | eAppendix, eTable 1 | eTable 1 prints the strategy and hit counts for all five databases. Every strategy includes the name of the CIS-R instrument, so the search is reproducible but narrow by design, which links back to item 6. | Reported |
| 9 | State how eligible studies were determined. | Methods, Identification and Selection of Studies | One reviewer screened titles and abstracts, and two reviewers judged full texts, with a third resolving disagreements. The Discussion acknowledges the single-reviewer first stage. | Reported |
| 10 | Describe how IPD were requested, collected and managed, including querying data with investigators, and how studies without IPD were handled. | Methods (one clause); eAppendix | The main text says IPD were sought and that data were cleaned and checked by multiple reviewers. The request process, the cleaning by two reviewers working one study at a time and the cross-check with chief investigators sit in the eAppendix prose and the abstract. The authors’ own checklist points to eTable 3, which lists ethics approvals and trial registrations, so a reader must look elsewhere for the narrative. Aggregate data were not needed because IPD came from every eligible study. | Judgement call |
| 11 | Describe how variables were chosen, define study-level and participant-level data, and explain how variables were harmonised across datasets. | Methods, Measures, Outcomes, Prognostic Indicators, Confounders; eAppendix | Four socioeconomic indicators, the confounders and the outcome scales are listed and defined. Harmonisation is explicit. Depressive symptom scores were converted to z scores within each study, and category sets were collapsed, for example an eight-category employment variable reduced to three. | Reported |
| A1 | Describe which aspects of the IPD were checked (such as sequence generation, consistency, completeness and baseline imbalance) and how. | Discussion (one clause); eAppendix | The main text carries only the statement that data were extracted, cleaned and checked by multiple reviewers. The eAppendix says baseline and endpoint data were checked with each study team and against published details. It does not say whether the randomisation sequence or baseline imbalance were examined. | Judgement call |
PRISMA-IPD checklist items 12 to 16 and A2 (bias, synthesis and variation)
The synthesis row of the PRISMA-IPD checklist, 14, lists eight issues, from one-stage or two-stage choice to missing data within the IPD. The example answers most of them in a single paragraph. Row A2 shows how a checklist built for treatment effects meets a prognostic question.
| Item | What the item asks | Where in the article | Evidence | Call |
|---|---|---|---|---|
| 12 | Describe risk-of-bias methods for each study, whether they were applied per outcome, how IPD checks informed them and how they entered the synthesis. | Methods, Risk of Bias | QUIPS, the Quality in Prognosis Studies tool, was used together with GRADE by two independent reviewers, with consensus for disagreements. QUIPS fits a prognostic question better than a trial tool such as RoB 2. The text does not say whether ratings were made per outcome or how IPD checks informed them, and the only link to the synthesis is a planned sensitivity analysis that could drop low-quality studies. | Judgement call |
| 13 | State the comparisons and outcomes, whether they were prespecified, and the effect measures. | Methods, Primary Outcomes, Secondary Outcomes | The primary outcome is depressive symptoms 3 to 4 months after baseline, expressed as a z score difference and as a percentage difference. Secondary outcomes are remission, with odds ratios, and later follow-up. The eAppendix lists two changes to the statistical plan, attrition dropped as an outcome and no conversion to PROMIS T-scores. | Reported |
| 14 | Describe the IPD synthesis, including one-stage or two-stage analysis, how effects were combined, model assumptions, heterogeneity and missing data. | Methods, Statistical Analysis; eAppendix | A two-stage approach with DerSimonian and Laird random effects in Stata 16 and four nested adjustment models is stated and justified. Missing data were imputed by chained equations, with 50 imputed datasets described in the eAppendix. The Methods say “Heterogeneity was assessed using prediction intervals” (Buckman et al.), yet Tables 3 and 4 show I-squared and study counts and we found no prediction intervals in the tables we read. | Judgement call |
| A2 | Describe methods used to explore variation in effects by study or participant characteristics, and whether they were prespecified. | Methods, Confounders; Primary Analyses | The review asks about prognostic associations, not treatment-by-covariate interactions, so the item fits only loosely. Four nested models adjust for sets of participant characteristics and the four socioeconomic factors are the planned exposures, but no effect-modifier analysis is described. | Judgement call |
| 15 | Specify any assessment of bias across the body of evidence, including not obtaining IPD for particular studies. | Methods, Risk of Bias; Discussion | GRADE ratings for indirectness and publication bias were made for each factor across all studies. The Discussion adds that having IPD from every eligible study avoided a common source of selection bias. | Reported |
| 16 | Describe additional and sensitivity analyses and which were prespecified. | Methods, Secondary and Sensitivity Analyses | Five sensitivity analyses add variables that not every study collected, and further analyses drop influential studies when I-squared reaches 75%. The text does not label each sensitivity analysis as prespecified. The eAppendix discloses that the I-squared threshold was lowered from 80% to 75% by protocol amendment. | Judgement call |
PRISMA-IPD checklist items 17 to 23 and A3 (results)
The results rows of the PRISMA-IPD checklist ask for the numbers that let a reader judge how complete the IPD are. Item 17 and the new row A3 carry most of that weight, and the example’s answers are split between the main text and the supplement.
| Item | What the item asks | Where in the article | Evidence | Call |
|---|---|---|---|---|
| 17 | Give numbers of studies screened, assessed and included with reasons, the studies and participants for which IPD were sought and obtained, and a flow diagram. | Figure 1; Results | Figure 1 reconciles. It shows 164 database records plus 1 from experts, 99 screened after duplicates (the number removed is not printed and is 66 by subtraction), 35 and then 55 exclusions with reasons, and 9 study teams contacted and 9 supplying data. Participants sought are not shown as a count. The Results give 4864 obtained, while the eAppendix says data for 54 CADET participants who withdrew after the trial were not made available. | Judgement call |
| 18 | Present key characteristics of each study, with citations, and of any studies without IPD. | Results, Table 1 | Table 1 lists the 9 trials with citations, size, inclusion criteria, baseline severity, intervention and primary endpoint. There is no funding column. The per-study sizes sum to 4868, whereas the abstract and Table 2 give 4864. The eAppendix lists participant differences for four studies against their original publications but does not say which figure Table 1 shows, and the eAppendix itself gives 4868 for the CIS-R sample across all studies. | Judgement call |
| A3 | Report any important issues found in checking the IPD, or state that there were none. | eAppendix, Data Integrity Checks | Differences are reported for four studies. Two participants were removed from IPCRESS and one from PANDA for missing data on more than 75% of baseline variables, data for 54 CADET participants who withdrew after the trial ended were not released, and ITAS had 36 more complete records than published. The Results neither state that no important issues arose nor point to this section. | Judgement call |
| 19 | Present risk-of-bias data and say whether data checking changed the weighting of assessments. | Results, first paragraph; eTables 4 and 5 | The text states “Study quality was judged to be high, and overall risk of bias was low” (Buckman et al.) and names the studies with high or moderate attrition. eTable 4 matches that text, with IPCRESS rated high and CADET, MIR and REEACT moderate. Interrater agreement was kappa 0.96 for QUIPS and 1.00 for GRADE. | Reported |
| 20 | For each study and main outcome, report participants, simple summary data per group and effect estimates with intervals, in a table or forest plot. | Table 1; eTable 8; Tables 3 and 4 | Participant numbers per study are in Table 1, and per-study means and standard deviations of the z score by category are in eTable 8. Tables 3 and 4 report pooled estimates only, and we found no per-study effect estimates or forest plot in the main text or the supplement. | Judgement call |
| 21 | Present summary effects with intervals and heterogeneity, state prespecification, give the numbers behind each analysis and describe effect size in meaningful terms. | Results, Tables 3 and 4; Discussion | Pooled results such as scores “47.3% (95% CI, 38.4%-56.8%) higher for unemployed patients than for employed patients” (Buckman et al.) come with study counts and I-squared, and the Discussion translates effects into about 4 points on the Beck Depression Inventory II. The loose end is the participant count behind each analysis. The Methods give n of 4864 for employment, 3656 for financial strain, 4397 for housing and 3689 for education, while the observed counts in Table 2 sum to 4861, 3647, 4391 and 3423. The article does not say how the two are defined. The financial strain figure equals the sum of the Table 1 sizes of the seven trials that measured it, which suggests the Methods count all participants in contributing trials while Table 2 counts non-missing answers. | Judgement call |
| 22 | Present results of any assessment of bias across the body of evidence. | eTable 5 | eTable 5 gives GRADE ratings for each factor, including publication bias, with all four factors rated high overall. The main text does not repeat them. | Reported |
| 23 | Give results of additional analyses and, where relevant, analyses that add studies without IPD. | Results, Further Sensitivity Analyses; eTables 14 to 21 | Removing the studies that contributed most to heterogeneity “did not substantively change the magnitude of the associations” (Buckman et al.). No studies lacked IPD, so the aggregate-data part of the item was not triggered. | Reported |
PRISMA-IPD checklist items 24 to 27 and A4 (discussion and funding)
The closing rows of the PRISMA-IPD checklist are answered cleanly. The one call falls on item 24, where the strength of evidence sits in a supplement table rather than in the Discussion.
| Item | What the item asks | Where in the article | Evidence | Call |
|---|---|---|---|---|
| 24 | State the main findings and how strong the evidence is for each main outcome. | Discussion, first paragraph | The first paragraph reports the unemployment and housing findings in percentages and in Beck Depression Inventory II points, with 28%, 30% and 37% higher scores for unemployed patients at the three follow-up points. The GRADE rating of high for each factor is in eTable 5 and is not repeated here. | Judgement call |
| 25 | Discuss strengths and limitations, including the benefits of access to IPD and limits from IPD that were not available. | Discussion, Strengths and Limitations | The IPD benefit is stated plainly, with “a common source of selection bias that can occur when only a subset of eligible trials provide individual patient data was avoided” (Buckman et al.). Limitations cover UK-only trials, participants who had consented to a trial, single-reviewer title and abstract screening and residual confounding. | Reported |
| 26 | Give a general interpretation of the findings alongside other evidence. | Discussion; Conclusions | The effects are set against published estimates of the minimal clinically important difference and against earlier reviews that covered fewer socioeconomic factors. | Reported |
| A4 | Consider relevance to policy makers, service providers and service users, and implications for future research. | Conclusions | The Conclusions recommend assessing employment and housing before treatment, point to interventions that support work and stable housing, and name future work on the order of offering support, longer follow-up and a public health angle. | Reported |
| 27 | Describe sources of funding and other support, including the supply of IPD, and the role of the funders. | Article notes, Funding/Support | Funders are listed, including a Wellcome Trust clinical research fellowship and an MQ Foundation grant, and a Role of the Funder statement says the funders had no part in design, analysis or the decision to submit. The eAppendix adds that the chief investigators of the trials were invited to join the project. | Reported |
What the 31 rows tell you
Across the 31 rows of the PRISMA-IPD checklist we scored 16 as Reported and 15 as Judgement call, and none as missing. That pattern is typical for a careful team. The review answers every section, and most calls concern where an answer sits or how it is phrased, not absent content. Six of the 15 calls fall on rows with IPD-specific content, namely 10, 14, 17, A1, A2 and A3.
Five rows where reviews lose marks
How IPD were requested and managed (item 10)
Say who was approached, who agreed, how queries were resolved and who cleaned each dataset, in the Methods and not only in a supplement. In the example, the narrative is in the eAppendix and the abstract, and the authors’ own page pointer sends readers to a table about ethics approvals.
Integrity checks and what they found (A1 and A3)
Name the aspects you checked, such as the randomisation sequence, consistency, completeness and baseline imbalance, and report the findings even when the answer is that nothing important turned up. The example reports participant differences in four studies, but only in the eAppendix.
Studies and participants in the flow diagram (item 17)
Show participants as well as studies, with the number sought and the number obtained, and print the number of duplicates removed. A reader should not have to subtract to learn that 66 records were duplicates.
The synthesis row and what the Methods promise (item 14)
Print every heterogeneity measure the Methods name. A sentence about prediction intervals invites a reviewer to look for them, and a two-stage analysis should say where the study-level estimates can be seen.
Per-study results (item 20)
The row asks for results for each study, in a table or forest plot. With a two-stage method the study-level estimates already exist, so a forest plot costs little and answers the row completely.
Reconciliation, the pass that clears most judgement calls
Several of the judgement calls above are not missing content at all. They are numbers, names or pointers that differ between the abstract, the text, the tables and the supplement. In this example the abstract and the Results describe one housing estimate from opposite directions, the per-study sizes in Table 1 sum to 4868 against 4864 elsewhere, the participant counts in the Methods differ from the observed counts in Table 2, and the Methods name prediction intervals that the tables do not show. These mismatches are cheap to fix before submission and costly to leave, because a reviewer who finds one mismatch starts checking everything else.
The fix is a single reconciliation pass in which every count is traced from the flow diagram through the abstract, the Results text, the tables and the supplement, and every named method is matched between the Methods and the Results. Our PRISMA flow diagram service checks numbers for internal consistency at every stage, and you can run the same pass yourself with the workflow below.
A PRISMA-IPD checklist workflow you can copy
- Open the PRISMA-IPD checklist at the protocol stage and write the protocol against rows 4 to 14. Our PRISMA-P protocol guide shows how to register and report it.
- Decide and record whether eligibility is applied to studies or to participants, and what you will do if IPD cannot be obtained from a study (items 6 and 10).
- Keep a request log for every eligible study, with the date asked, the answer, the variables received and each query and reply. The log feeds items 10 and 17 and row A3.
- Write the data-checking plan before the data arrive, covering sequence generation, consistency, completeness and baseline imbalance, and report what the checks found even when nothing important turned up (A1 and A3).
- Choose one-stage or two-stage analysis, name the heterogeneity measures you will display and print every measure the Methods promise (item 14). Assess bias with a tool that fits the question, such as RoB 2 for trial effects (see our RoB 2 guide) or QUIPS for prognostic factors.
- Build the flow diagram from the master counts, with studies and participants sought and obtained. Our PRISMA flow chart guide explains each box, and the Flowchart Pro tool below can draw it.
- Reconcile every number across the abstract, text, tables, figure and supplement, and check that each page pointer in your own checklist lands on the content it names.
Frequently asked questions about the PRISMA-IPD checklist
How many items does the PRISMA-IPD checklist have?
It has 31 items, which are the numbered rows 1 to 27 plus four IPD rows labelled A1 to A4. Three of the four are new, and A4 comes from rearranging the interpretation items. The wording of 23 standard items was also changed.
Should authors use PRISMA-IPD or PRISMA 2020 for an IPD meta-analysis?
Use the PRISMA-IPD checklist, because it was written as a stand-alone extension for reviews that collect participant-level data. Many authors also cite PRISMA 2020 for general reporting, and the PRISMA 2020 guide remains useful for background on shared items such as search reporting.
Does the PRISMA-IPD checklist apply outside randomised trials?
The authors developed it mainly for reviews of randomised trials and say many items apply to reviews of diagnosis and prognosis. The example here is a prognostic question, and every row could be scored, although rows 12 and A2 needed interpretation.
Is there a newer version of the PRISMA-IPD checklist?
We found no updated version. A 2026 brief report in BMC Medicine argues that the next update should add items the current checklist lacks, so check the PRISMA website before you submit.
What if IPD cannot be obtained from every eligible study?
Items 10, 17 and 23 cover this. Give the reason IPD were not sought or obtained for each study, report the numbers of studies and participants with and without IPD, and summarise results with and without the studies that lacked IPD. The example review obtained IPD from every eligible trial, so it does not show this case.
Free Research Tools from Editverse
- GraphPro: build publication-ready charts, 25 types, with built-in t-test, ANOVA, correlation, and chi-square p-values. Free, no signup.
- StatPro: free statistical calculators and a test-selector tool covering group comparisons, categorical outcomes, time-to-event analysis, and diagnostic accuracy.
- Flowchart Pro: generate PRISMA 2020, PRISMA-ScR, CONSORT, STROBE, STARD, MOOSE, and ARRIVE flow diagrams directly in your browser.
Professional Support for a PRISMA-IPD Compliant Review
Writing to the PRISMA-IPD checklist is easier when someone has already traced every number and every method name through the manuscript. Two Editverse services map onto the rows covered above.
Evidence synthesis
$3,500
Protocol through publication support for a systematic review and meta-analysis, covering search strategy, data extraction, statistical analysis and manuscript writing, with a dedicated methodologist. 8 to 16 weeks depending on scope.
Flow diagram
$100
A PRISMA 2020 flow diagram built from your screening counts, with the numbers checked for internal consistency at every stage. Vector files and high-resolution PNG, delivered in 1 to 2 business days.
The meta-analysis service page lists pairwise, network, diagnostic accuracy and prognostic synthesis and does not name IPD meta-analysis, so describe your data-sharing situation in your first message. For a complete review from protocol to submission, see our systematic review writing service, which supports PRISMA-compliant evidence synthesis. Our work aligns with COPE, ICMJE and GPP 2022, and follows institutional and country-specific ethics requirements.
Using the checklist well
The PRISMA-IPD checklist rewards teams that treat the data as part of the method. The example review answers every section and still leaves its clearest gains in the rows that describe how IPD were requested, what was checked and what the checks found, how many participants each trial supplied and which heterogeneity measures were reported. It also shows that most judgement calls are reconciliation problems that a single careful pass can clear. Fill the PRISMA-IPD checklist at the protocol stage, log every data request, report your integrity checks, trace every count from the flow diagram to the abstract, and your reviewers will have little left to ask. For the shared items of any systematic review, return to the PRISMA 2020 guide.
References
- Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, et al. Preferred Reporting Items for Systematic Review and Meta-Analyses of individual participant data. The PRISMA-IPD Statement. JAMA. 2015;313(16):1657-1665. doi:10.1001/jama.2015.3656. PMID 25919529.
- Buckman JEJ, Saunders R, Stott J, Cohen ZD, Arundell LL, et al. Socioeconomic indicators of treatment prognosis for adults with depression. A systematic review and individual patient data meta-analysis. JAMA Psychiatry. 2022;79(5):406-416. doi:10.1001/jamapsychiatry.2022.0100. PMID 35262620. PMCID PMC8908224. Licence CC BY 4.0.
- Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data. Rationale, conduct, and reporting. BMJ. 2010;340:c221. doi:10.1136/bmj.c221. PMID 20139215.
- Hayden JA, van der Windt DA, Cartwright JL, Côté P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med. 2013;158(4):280-286. doi:10.7326/0003-4819-158-4-201302190-00009. PMID 23420236.
- Dal-Ré R, Bourgeois FT, Hemkens LG, Holm S, Moher D, Naudet F. Inconsistent ethics reporting in individual participant data meta-analyses. A call for updating PRISMA and PRISMA-IPD. BMC Med. 2026. doi:10.1186/s12916-026-05056-8. PMID 42449420. PMCID PMC13651295.
- Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement. An updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi:10.1136/bmj.n71. PMID 33782057.