The analysis of incomplete cost data due to dropout

Abstract

Incomplete data due to premature withdrawal (dropout) constitute a serious problem in prospective economic evaluations that has received only little attention to date. The aim of this simulation study was to investigate how standard methods for dealing with incomplete data perform when applied to cost data with various distributions and various types of dropout. Selected methods included the product-limit estimator of Lin et al. the expectation maximisation (EM-) algorithm, several types of multiple imputation (MI) and various simple methods like complete case analysis and mean imputation. Almost all methods were unbiased in the case of dropout completely at random (DCAR), but only the product-limit estimator, the EM-algorithm and the MI approaches provided adequate estimates of the standard error (SE). The best estimates of the mean and SE for dropout at random (DAR) were provided by the bootstrap EM-algorithm, MI regression and MI Monte Carlo Markov chain. These methods were able to deal with skewed cost data in combination with DAR and only became biased when costs also included the costs of expensive events. None of the methods were able to deal adequately with informative dropout. In conclusion, the EM-algorithm with bootstrap, MI regression and MI MCMC are robust to the multivariate normal assumption and are the preferred methods for the analysis of incomplete cost data when the assumption of DCAR is not justified.

Estimates of patient costs related with population morbidity: can indirect costs affect the results?

“…In this paper we present our results on the concurrent risk adjustment model using morbidity information grouped under the highest level of aggregation, so that average costs for each category are real costs estimates for the population belonging to that risk category during year 2005. Table IV presents direct average costs (Col.2) and total average costs when an absorption hypothesis is chosen (Col. [3][4][5]. Looking at the results, Output and Actual Utilization methodologies explain a very similar amount of variation, although the R 2 coefficient is slightly higher using the Output absorption method for allocating indirect costs.…”

“…Clement et al [4] investigated to what extent the selection of the costing methodology can affect the results of an economic evaluation and produce further wrong decisions. Other problems accepted by the health economics literature in the analysis of individual healthcare costs are the existence of missing data [5,6] or the application of inadequate costing methods [7].…”

Multiple Imputation Methods for Handling Missing Data in Cost-effectiveness Analyses That Use Data from Hierarchical Studies

“…22 In particular, allowing the data to be MNAR was motivated by the general concern in CEA that probability of cost and outcome data may be conditional on the level of the endpoints. 8,10,31 For example, in our case study it may be expected that women with lower WTW may be more likely to have missing WTW. Since the process utility endpoint, WTW, had lower ICC than the cost endpoint, we also hypothesized that there could be smaller differences between methods for handling the missing WTW data than for handling missing costs.…”

Cost‐Utility Analysis of a Multidisciplinary Strategy to Manage Osteoarthritis of the Knee: Economic Evaluation of a Cluster Randomized Controlled Trial Study

“…Therefore, it is important to employ an appropriate technique that will permit complete case analysis using the entire data set. The approach suggested by Briggs and colleagues, Oostenbrink et al, and the International Society for Pharmacoeconomics and Outcomes Research was followed (13)(14)(15)(16). In the PhIT-OA trial, the rate of missing data cases was 14% for costs, 18% for the PAT-5D, and 12% for the HUI3.…”