Individual Participant Data Meta-Analysis. Группа авторов
programme of IPD meta-analyses of therapies for metastatic prostate cancer.94"/>
Figure 4.3 Excerpt from a trial‐level data collection form for the STOPCAP M1 programme of IPD meta‐analyses of therapies for metastatic prostate cancer.94
Source: Based on Tierney JF, Vale CL, Parelukar WR, et al. Evidence Synthesis to Accelerate and Improve the Evaluation of Therapies for Metastatic Hormone-sensitive Prostate Cancer. Eur Urol Focus 2019;5(2):137–43.
Given this set of eligible trials, how much IPD should be sought from them? As a general rule, the aim should be to maximise the quantity and quality of IPD available, in order to fulfil the project objectives and complete the planned analyses reliably. For conventional reviews, aggregate data would ordinarily be sought for all studies relevant to the question of interest. Similarly, and ideally, IPD should be sought from all the eligible trials, for all participants recruited to those trials, and for all relevant outcomes, even if they were not published or included in the original analyses. This will help circumvent the risk of publication bias, outcome reporting bias, attrition bias, and other data availability biases (Chapter 9).46.58,95 For example, in trials of the effectiveness of recombinant human bone morphogenetic protein‐2 for spinal fusion, the adverse event data were not reported sufficiently to allow a rigorous evaluation of safety,96 whereas the collection of IPD allowed a complete, detailed, and in‐depth analysis.65 If it is not feasible or practical to seek IPD from all trials, the potential impact of these ‘missing’ trials should be taken into account (Chapter 9).
4.2.6 Deciding Which Variables Are Needed in the IPD
As for all systematic reviews, the pre‐specified outcomes for the IPD meta‐analysis project should be those judged to be most important and relevant to the objectives, even if ultimately there are insufficient data available to analyse all of them. While consideration of the participant‐level variables required begins when the IPD meta‐analysis questions are formulated, this should be re‐visited before requesting IPD from trial investigators, so that they can be specified more precisely and ensure that the planned analyses can be completed satisfactorily. Trial publications can provide an initial guide as to which data might be available, but more variables may have been collected in a trial than is evident from a trial report. Often the trial protocol and the associated case report forms can provide a more reliable indication of which data have been recorded. Irrespective of whether these documents are available or not, it is useful to supply trial investigators with a provisional list of desired variables via a paper or online form, or as a detailed data dictionary (Section 4.2.7), so that trial teams can clarify precisely which data items they can provide. An example of the typical types of data requested from trial investigators is shown in Box 4.3.
In addition, it is important to anticipate what supplementary analyses might be needed in the IPD meta‐analysis project to explore the main results. For example, for a question about the effects of chemotherapy on long‐term cancer survival, it may be helpful to collect data which would allow the investigation of the effects of treatment on different (competing) causes of death, such as those due to cancer, treatment‐related side effects or co‐morbid conditions.
In many cases, it will only be necessary to collect outcomes and participant characteristics as defined in the individual trials. However, additional variables might be required to provide greater granularity (e.g. sub‐scales in quality of life instruments), or to allow outcomes or other variables to be defined in a consistent way for each trial. For example, in an IPD meta‐analysis of anti‐platelet therapy for pre‐eclampsia in pregnancy, data on systolic and diastolic blood pressure plus presence of proteinurea were collected. This was to allow the central research team to analyse pre‐eclampsia according to both a pre‐defined meta‐analysis definition, as well as the individual trial definitions (of which there were many variations).97 Furthermore, if the IPD are to be maintained in perpetuity, to address new questions that might arise, additional data may be requested to effectively ‘future‐proof’ the database. For example, if there is a plan to use the IPD collected to produce conditional treatment effects (Chapter 5), to identify predictors of treatment effect (Chapter 7), or to identify prognostic factors (Chapter 16), then it would be sensible to request more detailed baseline data than might be necessary if just the overall (unadjusted) effects of treatments were of interest. Having said that, it is important to avoid collecting extraneous data, as these will still need to be checked and managed, and if not used, this represents an unnecessary burden for the trial teams who have spent time preparing data. Of course, it may be easier for trial teams to provide a complete trial data file, and let the IPD meta‐analysis research team extract what they need.
Box 4.3 Example of typical data obtained for trials to be included in an IPD meta‐analysis project
At a minimum, the IPD requested for each trial would typically include variables that:
‘Identify’ participants, e.g.De‐identified participant ID (Section 4.4.1), centre ID
Describe the participant population, facilitate data checking and allow analyses by participant characteristics, e.g.Age, sex, demographic variables, disease or condition characteristics and key prognostic factors
Describe the intervention, e.g.Date of randomisationIntervention allocationIf appropriate, the interventions participants received and the dates of administration
Record all outcomes of interest and relevant to the objectives, e.g.Survival, toxicity, pre‐eclampsia, healing, hospital stay, last follow‐up date
Describe whether participants were excluded from the primary trial analysis and reasons, e.g.Ineligible, protocol violation, missing outcome data, withdrawal, ‘early’ outcome
Source: Jayne Tierney and Lesley Stewart.
4.2.7 Developing a Data Dictionary for the IPD
In addition to preparing a list of variables that will be required for the analyses, it is important to consider carefully how best to define, collect and store these in an appropriate and unambiguous manner. The development of a detailed data dictionary for an IPD meta‐analysis project effectively establishes the structure of the meta‐analysis database, facilitates processing of IPD from each trial and ensures that the analyses can proceed as planned, with the greatest degree of flexibility. It also helps guide the trial teams in the preparation of IPD prior to transfer, and gives them the responsibility for modifying variables, lessening the likelihood of misinterpretation or coding errors. However, trial teams may not have the time to adhere to the data dictionary, and they should not be compelled to do so, particularly if their resources for preparing the IPD are limited. In such instances, it is advisable that the central research team accepts trial IPD in any (reasonable) workable form, and take responsibility for reformatting and re‐coding it themselves, according to the data dictionary.
Table 4.1 provides an excerpt from a data dictionary used in an IPD meta‐analysis examining the effects of chemoradiation for cervical cancer.93 Age at randomisation