Differences in treatment choices for localised prostate cancer diagnosed in private and public health services

To the Editor: In the retrospective study by te Marvelde and colleagues,1 the proportions of men in public and private health services receiving radical prostatectomy and curative external beam radiation therapy were examined in a multivariable logistic regression analysis. However, only age, International Society of Urological Pathology (ISUP) tumour grade, and comorbidity were studied.

Prostate‐specific antigen (PSA) level and T stage are two of the strongest determinants of choice of treatment modality in clinical practice and have not been considered or discussed by the authors. We consider this to be a major flaw in this study and a failure of the peer‐review process to highlight this deficiency, which has a significant impact on the results and subsequent conclusions reached by the authors.

Furthermore, patient comorbidities have not been adequately accounted for. The authors identify comorbidity as a factor influencing treatment, but they fail to assess and account for this variable in a reliable way. Victorian Admitted Episodes Dataset (VAED) data for the year preceding the prostate cancer diagnosis and up to 30 days after diagnosis were assessed to identify comorbid conditions other than cancer according to the Charlson Comorbidity Index, categorised as 0 or at least 1. This variable provided little discriminatory power (3% v 6%), and yet it was the only surrogate variable that accounted for comorbidity in the study’s key multivariable analyses. Additionally, the odds ratios for this variable in these analyses were not reported. It should be noted that 38% of the study population were men older than 70 years, but only 3.8% scored 1 or more on the VAED‐derived Charlson Comorbidity Index. We believe that the method used in this study to account for comorbidity is not adequately robust to provide an accurate picture of the patients’ general health status.

The authors also cite the ProTect trial2 to suggest no major differences between active treatment options exist; however, they did not identify the vast differences in the disease characteristics of men in the ProTect trial compared with those included in their study (77% ISUP 1 and 2% ISUP 4/5 v 35% ISUP 1 and 18% ISUP 4/5).

Moreover, te Marvelde and colleagues did not address clinical outcomes and have not presented evidence that the variation in treatment modalities between public and private services has had a negative impact on the final clinical outcome.

Outcomes data can be helpful in identifying systematic shortcomings, inequities and barriers to just health care, but the authors missed the opportunity to highlight these issues. They concluded that the treatment of people with cancer should be consistent, safe, of high quality and evidence‐based, but did not provide evidence that the current practice is to the contrary.