The value of the hospital episode statistics to study practice and outcome of urological surgery

Abstract

Introduction: Despite the growing emphasis on the collection, monitoring and reporting of outcomes of surgical care within the UK, few data exist for such purposes. Equally, few data are available on disease occurrence and surgical practice in England. In this thesis, the hypothesis that Hospital Episode Statistics (HES) data can be used for such purposes is tested. Methods: The objectives of this thesis were twofold. Firstly, to review the literature to establish to what extent lIES data has been used to study incidence, surgical treatments and their outcomes. Secondly, to evaluate - using a number of individual lIES-derived case studies, to what extent the lIES database can be utilised to answer clinical questions concerning incidence, practice and outcome of urological surgery. Strengths and weaknesses of the lIES database were subsequently sought on which to generate recommendations concerning the future clinical use of the database. Results: The review highlighted the completeness of the database identifying very low underascertainment when compared to local audit data although did identify that use of Finished Consultant Episodes (FCE'S) resulted in over-estimation of disease incidence. lIES-derived case studies suggested, first, the incidence of hypospadias was considerably higher than previously reported and furthermore, did appear to be on the increase. Second, boys continue to undergo clinically inappropriate foreskin surgery. Third, the shift away from surgery for men with symptomatic BPlI has not resulted in more men experiencing an episode of acute urinary retention. Fourth, high-volume cancer centres appear to achieve improved outcomes following cystectomy by reducing the risk of failure-to-rescue following an adverse event. Recommendations on the future clinical use of HES data drawn from the HES-derived case studies were first, if incidence is to be calculated using HES data, there must be no or minimal ambiguity regarding diagnostic or procedural coding. Second, a number of diagnostic and procedural codes can be operationalised to define an event provided coding practice is not identified in advance. Third, the incidence of a condition can be identified using HES data provided all patients or at least the majority of patients undergo hospital treatment for that condition. Fourth, for HES data to comment on clinical appropriateness of treatment, the prevalence of the condition requiring treatment must be known. Fifth, if HES data is to be used to analyse trends in disease incidence over time, there must be no changes in coding practice over the study period. Sixth, HES data should not be used in isolation to report long-term oncological outcomes. Seventh, HES data is suitable to investigate the determinants of short-term surgical outcome such as mortality. Conclusion: Evaluation of the use of HES data to answer specific questions concerning the incidence, practice and outcome of urological surgery suggests HES is a useful source of data provided caveats concerning the strengths and weaknesses of the database are considered at the time of data analysis.