Individual participant data meta-analysis of prognostic factor studies: state of the art? (original) (raw)
Related papers
Archives of Physical Medicine and Rehabilitation, 2014
Prognostic studies of mild traumatic brain injury (MTBI) can serve many purposes. First, they are used to describe paths and outcomes of patients with MTBI. Second, they provide information on which characteristics are associated with the occurrence of outcomes. Third, they provide insight into the causation of poor or favorable course of the disease. Finally, they can assess how differences in the probability of outcomes can help predict the course of patients. In this article, we summarize methodologic principles used by the International Collaboration on MTBI Prognosis to appraise the prognostic literature. Differentiating prognostic factors (causally linked with outcome), prognostic markers (associated but not causally), and predictors is important to guide interventions, public health policy, and research. Ideally, prognostic studies need a clear statement of the type of question (hypothesis-generating descriptive, exploration of possible prognostic variables, confirmatory modeling of prognosis); a cohort study design with standardized follow-up of a representative population of patients with MTBI; a standardized data collection using reliable and accurate tools to capture clinically, biologically, psychologically, or socially relevant variables and outcomes; and an analysis of data based on survival methods. Interpretation of prognostic studies should consider biases related to differential inclusion of nonrepresentative samples of patients, poor measurements of outcomes, and poor control for confounders. Transferring prognostic results into clinical practice should be based on estimates of the predictive performance of models and on a demonstration that patient outcomes can be improved by the use of prediction rules.
Individual patient data meta-analysis of prognostic factor studies
2011
Prognostic factors (PFs) are patient characteristics (e.g. age, biomarker levels) that are associated with future clinical outcomes in patients with a disease or health condition. Evidence-based PF results are paramount, for which individual patient data (IPD) metaanalysis is thought to be the 'gold-standard' approach, as it synthesises the raw data across related studies (in contrast to an aggregate data meta-analysis, that just uses reported summary data). In this Ph.D. thesis, I investigate statistical issues and develop methodological recommendations for individual patient data meta-analysis of prognostic factor studies (IMPF) projects. First, I investigate the benefits and limitations of IPD meta-analyses of PF studies through a systematic review and in-depth evaluation of existing IPD meta-analyses of PFs; 48 IMPF articles were found and an in-depth evaluation of a random sample of 20 IMPF articles was undertaken to identify how such projects are initiated, conducted, and reported, and to identify the benefits and challenges of the IPD approach. I found that although IMPF articles have many advantages, they still face a number of challenges and pitfalls such as different methods of measurements, ignoring clustering of patients across studies, missing data, and potential publication bias, unachieved linearity assumption of PFs, poor reporting, and potentially not protocol driven. To improve IMPF articles and projects guidelines were developed, and an array of methodological research questions identified. Secondly, I undertook an empirical study to compare between the IPD and aggregated data approach to assess PFs in breast cancer. I showed that the IPD approach is preferable over aggregated data, as it allows one to adjust the PF by other confounding factors, examine PFs in subgroups of patients and assess the interaction between two PFs as an additional PF. It also allowed more studies and more patients to be included. However, the IPD approach still faced challenges, such as potential publication bias, missing data, and failed model assumptions in some studies. Thirdly, I developed eleven IPD meta-analysis models to investigate whether accounting for clustering of patients within studies should be undertaken and which approach is the best to use. The models differed by using either a one-step or two-step approach, and whether they accounted for parameter correlation and residual variation. An IPD meta-analysis of 4 studies for age as a PF for 6 month Firstly, I am heartily thankful to my supervisor, Richard Riley, whose encouragement, supervision and support from the preliminary to the concluding level enabled me to develop an understanding of the subject. I am particularly grateful for how he has helped develop this career path for me, and for the large amount of time he has spent reading various drafts and providing constructive feedback over the last few months, without his guidance and persistent help this dissertation would not have been possible. I also want to greatly thank Prof. Jon Deeks for their encouragement, guide, and advice throughout the duration of the thesis, especially for the final stage in my thesis, it was really helpful. Furthermore, I thank Prof. Willi Sauerbrei who provided helpful direction and feedback at some important stages of the thesis. I also would like to thank Dr. Prakash Patil, Boliang Guo and Sergi for their help and advice in some particular area in my thesis. I would like to show my gratitude to my sponsor, Embassy of the Arab Republic of Egypt cultural centre and Educational Bureau London, for awarded me with this scholarship. I would like also to thank the MRC Midlands Hub for Trials Methodology Research for their support and provision, especially Prof. Lucinda Billingham for her support, help and advice that she gave to me. I owe my deepest gratitude to my parents, my brothers and my sister for their support and encouragement throughout the duration of the thesis. Last but not the least, I offer my regards and blessings to all of those who supported me in any respect during the completion of the thesis.
Prognosis and clinical trial design in traumatic brain injury: the IMPACT study
Journal of …, 2007
Many randomized controlled trials (RCTs) have been performed to investigate the effectiveness of new therapies, but none have convincingly demonstrated benefit. Clinical trials in TBI pose complex methodological challenges and meeting these requires new approaches. The challenges are related to the heterogeneity of head injuries, to optimum analysis of outcome and to aspects of the design of trials. To address these, we have created the IMPACT database on TBI through merging individual patient data from eight RCTs and three observational surveys. This database forms a culture medium in which innovative approaches to improving trial design and analysis are being explored. We hypothesize that the statistical power of TBI trials may be increased by adjusting for heterogeneity with covariate adjustment and/or prognostic targeting, by exploiting the ordinal nature of the Glasgow Outcome Scale and by relating the outcome obtained in individual patients to their baseline prognostic risk. Extensive prognostic analysis was required as a first step towards our aim of optimizing the chance of demonstrating benefit of new therapies in future trials. The fruits of this analysis are reported in detail in the subsequent reports in this issue of the Journal of Neurotrauma. The results will lead to the development and validation of new prognostic models, which will be applied to deal with heterogeneity. The findings will be synthesized into recommendations for the design and analysis of future RCTs, with the expectation of increasing the likelihood of demonstrating the benefit of a truly effective new therapy or therapeutic agent in victims of a head injury.
Systematic Reviews, 2012
Background: Mild traumatic brain injury (MTBI) is a major public-health concern and represents 70-90% of all treated traumatic brain injuries. The last best-evidence synthesis, conducted by the WHO Collaborating Centre for Neurotrauma, Prevention, Management and Rehabilitation in 2002, found few quality studies on prognosis. The objective of this review is to update these findings. Specifically, we aim to describe the course, identify modifiable prognostic factors, determine long-term sequelae, and identify effects of interventions for MTBI. Finally, we will identify gaps in the literature, and make recommendations for future research. Methods: The databases MEDLINE, PsychINFO, Embase, CINAHL and SPORTDiscus were systematically searched (2001 to date). The search terms included 'traumatic brain injury', 'craniocerebral trauma', 'prognosis', and 'recovery of function'. Reference lists of eligible papers were also searched. Studies were screened according to pre-defined inclusion and exclusion criteria. Inclusion criteria included original, published peer-reviewed research reports in English, French, Swedish, Norwegian, Danish and Spanish, and human participants of all ages with an accepted definition of MTBI. Exclusion criteria included publication types other than systematic reviews, meta-analyses, randomized controlled trials, cohort studies, and case-control studies; as well as cadaveric, biomechanical, and laboratory studies. All eligible papers were critically appraised using a modification of the Scottish Intercollegiate Guidelines Network (SIGN) criteria. Two reviewers performed independent, in-depth reviews of each eligible study, and a third reviewer was consulted for disagreements. Data from accepted papers were extracted into evidence tables, and the evidence was synthesized according to the modified SIGN criteria. Conclusion: The results of this study form the basis for a better understanding of recovery after MTBI, and will allow development of prediction tools and recommendation of interventions, as well as informing health policy and setting a future research agenda.
Diagnostic and Prognostic Research Evidence synthesis in prognosis research
Diagnostic and Prognostic Research, 2019
Over the past few years, evidence synthesis has become essential to investigate and improve the generalizability of medical research findings. This strategy often involves a meta-analysis to formally summarize quantities of interest, such as relative treatment effect estimates. The use of meta-analysis methods is, however, less straightforward in prognosis research because substantial variation exists in research objectives, analysis methods and the level of reported evidence. We present a gentle overview of statistical methods that can be used to summarize data of prognostic factor and prognostic model studies. We discuss how aggregate data, individual participant data, or a combination thereof can be combined through meta-analysis methods. Recent examples are provided throughout to illustrate the various methods.
Evidence synthesis in prognosis research
Diagnostic and Prognostic Research
Over the past few years, evidence synthesis has become essential to investigate and improve the generalizability of medical research findings. This strategy often involves a meta-analysis to formally summarize quantities of interest, such as relative treatment effect estimates. The use of meta-analysis methods is, however, less straightforward in prognosis research because substantial variation exists in research objectives, analysis methods and the level of reported evidence. We present a gentle overview of statistical methods that can be used to summarize data of prognostic factor and prognostic model studies. We discuss how aggregate data, individual participant data, or a combination thereof can be combined through meta-analysis methods. Recent examples are provided throughout to illustrate the various methods.
BMJ open, 2017
Reports on the association between comorbidity and functional status and risk of death in patients with traumatic brain injury (TBI) have been inconsistent; it is currently unknown which additional clinical entities (comorbidities) have an adverse influence on the evolution of outcomes across the lifespan of men and women with TBI. The current protocol outlines a strategy for a systematic review of the current evidence examining the impact of comorbidity on functional status and early-term and late-term mortality, taking into account known risk factors of these adverse outcomes (ie, demographic (age and sex) and injury-related characteristics). A comprehensive search strategy for TBI prognosis, functional (cognitive and physical) status and mortality studies has been developed in collaboration with a medical information specialist of the large rehabilitation teaching hospital. All peer-reviewed English language studies with longitudinal design in adults with TBI of any severity, pub...
Comorbidity in adults with traumatic brain injury and all-cause mortality: a systematic review
BMJ Open
ObjectivesComorbidity in traumatic brain injury (TBI) has been recognised to alter the clinical course of patients and influence short-term and long-term outcomes. We synthesised the evidence on the effects of different comorbid conditions on early and late mortality post-TBI in order to (1) examine the relationship between comorbid condition(s) and all-cause mortality in TBI and (2) determine the influence of sociodemographic and clinical characteristics of patients with a TBI at baseline on all-cause mortality.DesignSystematic review.Data sourcesMedline, Central, Embase, PsycINFO and bibliographies of identified articles were searched from May 1997 to January 2019.Eligibility criteria for selecting studiesIncluded studies met the following criteria: (1) focused on comorbidity as it related to our outcome of interest in adults (ie, ≥18 years of age) diagnosed with a TBI; (2) comorbidity was detected by any means excluding self-report; (3) reported the proportion of participants wit...
Statistics in Medicine
Clinical prediction models aim to provide estimates of absolute risk for a diagnostic or prognostic endpoint. Such models may be derived from data from various studies in the context of a meta-analysis. We describe and propose approaches for assessing heterogeneity in predictor effects and predictions arising from models based on data from different sources. These methods are illustrated in a case study with patients suffering from traumatic brain injury, where we aim to predict 6-month mortality based on individual patient data using meta-analytic techniques (15 studies, n = 11 022 patients). The insights into various aspects of heterogeneity are important to develop better models and understand problems with the transportability of absolute risk predictions.