Meta-analysis of randomized controlled trials is generally considered the most reliable source of estimates of relative treatment effects. However, in the last few years, there has been interest in using non-randomized studies to complement evidence from randomized controlled trials. Several meta-analytical models have been proposed to this end. Such models mainly focussed on estimating the average relative effects of interventions. In real-life clinical practice, when deciding on how to treat a patient, it might be of great interest to have personalized predictions of absolute outcomes under several available treatment options. This paper describes a general framework for developing models that combine individual patient data from randomized controlled trials and non-randomized study when aiming to predict outcomes for a set of competing medical interventions applied in real-world clinical settings. We also discuss methods for measuring the models' performance to identify the optimal model to use in each setting. We focus on the case of continuous outcomes and illustrate our methods using a data set from rheumatoid arthritis, comprising patient-level data from three randomized controlled trials and two registries from Switzerland and Britain.

Individual participant data (IPD) from multiple sources allows external validation of a prognostic model across multiple populations. Often this reveals poor calibration, potentially causing poor predictive performance in some populations. However, rather than discarding the model outright, it may be possible to modify the model to improve performance using recalibration techniques. We use IPD meta-analysis to identify the simplest method to achieve good model performance. We examine four options for recalibrating an existing time-to-event model across multiple populations: (i) shifting the baseline hazard by a constant, (ii) re-estimating the shape of the baseline hazard, (iii) adjusting the prognostic index as a whole, and (iv) adjusting individual predictor effects. For each strategy, IPD meta-analysis examines (heterogeneity in) model performance across populations. Additionally, the probability of achieving good performance in a new population can be calculated allowing ranking of recalibration methods. In an applied example, IPD meta-analysis reveals that the existing model had poor calibration in some populations, and large heterogeneity across populations. However, re-estimation of the intercept substantially improved the expected calibration in new populations, and reduced between-population heterogeneity. Comparing recalibration strategies showed that re-estimating both the magnitude and shape of the baseline hazard gave the highest predicted probability of good performance in a new population. In conclusion, IPD meta-analysis allows a prognostic model to be externally validated in multiple settings, and enables recalibration strategies to be compared and ranked to decide on the least aggressive recalibration strategy to achieve acceptable external model performance without discarding existing model information.

Aims: Use of prediction models is widely recommended by clinical guidelines, but usually requires complete information on all predictors, which is not always available in daily practice. We aim to describe two methods for real-time handling of missing predictor values when using prediction models in practice.

Methods and results: We compare the widely used method of mean imputation (M-imp) to a method that personalizes the imputations by taking advantage of the observed patient characteristics. These characteristics may include both prediction model variables and other characteristics (auxiliary variables). The method was implemented using imputation from a joint multivariate normal model of the patient characteristics (joint modelling imputation; JMI). Data from two different cardiovascular cohorts with cardiovascular predictors and outcome were used to evaluate the real-time imputation methods. We quantified the prediction model's overall performance [mean squared error (MSE) of linear predictor], discrimination (c-index), calibration (intercept and slope), and net benefit (decision curve analysis). When compared with mean imputation, JMI substantially improved the MSE (0.10 vs. 0.13), c-index (0.70 vs. 0.68), and calibration (calibration-in-the-large: 0.04 vs. 0.06; calibration slope: 1.01 vs. 0.92), especially when incorporating auxiliary variables. When the imputation method was based on an external cohort, calibration deteriorated, but discrimination remained similar.

Conclusions: We recommend JMI with auxiliary variables for real-time imputation of missing values, and to update imputation models when implementing them in new settings or (sub)populations.