Methodological challenges to assess effectiveness of treatments with retrospective studies : the case of localized prostate cancer

Article type
Authors
Larocque B1, Asselin G1, Coulombe M1, Rhainds M1
1UETMIS, CHU de Québec, Canada
Abstract
Background: Radical prostatectomy (RP), external beam radiation therapy (EBRT), and brachytherapy (BT) are the main curative options for localized prostate cancer (PCa). However, divergence still remains between experts about which option should be preferred in terms of benefits and harms.

Objectives: To assess the comparative effectiveness of treatments for localized PCa.

Methods: A systematic review was conducted in in PubMed, Embase, the Cochrane Library and the grey literature until October 2012 to identify systematic reviews (SRs), randomized controlled trials (RCTs), and observational studies. The primary outcomeswere PCa-specific mortality and all-cause mortality. Selection, quality assessment and data extraction were performed by two independent reviewers. Synthesis review was shared with an interdisciplinary group of experts.

Results: No RCT comparing RP to other therapeutic options was found. A total of 12 retrospective population-based cancer studies, including claims databases and hospital-based observational studies were included. Results suggest that patients treated by RP had a lower risk of PCa-specific mortality (hazard ratio (HR): 0.38–0.67) or all-cause mortality (HR: 0.49–0.63) than those receiving EBRT. Results concerning BT compared with RP and EBRT remain unclear. Use of retrospective databases t was associated with major methodological issues. Limitations include validity of the causes of death, age-dependant PCa treatment, staging, individual risk assessment (PSA, Gleason score), treatment description such as surgical procedures, radiation dose, isotope for BT, inability to separate EBRT than BT in databases, and change over time for radiation therapy.

Conclusions: Results from retrospective studies suggest that RP is associated with better survival outcomes compare with EBRT. However, considering the methodological weakness related to selection, misclassification and confounding by indication bias, the level of evidence remains low. Due to the uncertainty of comparative effectiveness, clinicians should emphasize on patient’s values, quality of life and side effects to support shared-decision making in PCa.