Married status has an established association with better treatment access and even better survival in multiple malignancies [7,8,12–14]. In localized muscle-invasive bladder cancer, this effect has also been validated in patients treated with radical cystectomy, where married patients have better access to radical cystectomy based on more favorable stage at surgery [7,15]. However, the association between married status and access to TMT, an emerging treatment modality for localized muscle-invasive UCUB, is unknown. We addressed this knowledge gap and made several noteworthy observations.
First, between 2004 and 2020, as many as 3,709 stage II UCUB patients underwent TMT. The current study represents the largest and most contemporary cohort of TMT patients. The second most contemporary cohort was reported by Deuker et al. which included 2,675 TMT patients treated from 2004 to 2016 [3]. In consequence, the current study provides the most contemporary image of TMT and compares favourably with respect to contemporaneity and sample size relative to any other reports on that modality [16,17].
Second, of included patients with localized muscle-invasive UCUB, 7,112 (25%) were female and 20,923 (75%) were male. These proportions are highly consistent with previous analyses of patients with localized muscle-invasive UCUB [18–20]. This observation validates the current study cohort regarding female vs. male sex which represents the first important stratification variable in the current analysis. Additionally, the proportions of married vs. unmarried patients in both females (34 vs. 66%) and males (65 vs. 35%) are also highly comparable to previous studies [6,9]. These observations further validate the current study population regarding married status which represents an equally important stratification variable. To the best of our knowledge, no previous study examined married status in TMT patients. Consequently, no direct comparison can be made with previous reports.
Third, comparison of patient characteristics according to married status revealed very important differences. Specifically, married females were significantly younger than unmarried females. Moreover, race/ethnicity distribution in married vs. unmarried females also revealed important differences. Specifically, across all race/ethnicities (Caucasians, Hispanics, African Americans and Other/unknown), the proportion of unmarried females was higher compared to their married counterparts. Conversely, the proportion of married males was higher among Caucasians, Hispanics, and Other/unknown race/ethnicity compared to their unmarried counterparts. Finally, the distribution of African American males was equal between married and unmarried cohorts. It is essential to acknowledge and account for the presence of these differences in comparisons of TMT between married and unmarried females as will be shown below. To the best of our knowledge, this topic was not addressed in any previous study. In consequence it cannot be directly compared to existing data.
Fourth, rates of TMT seemingly did not differ between married vs. unmarried females (14 vs. 13%, p=0.2). However, after multivariable adjustment, the odds ratio associated with married status and TMT rates increased from an insignificant value to an odds ratio of 1.2 and a multivariable p-value of 0.02. This important change in unadjusted vs. adjusted association between married status and TMT use in females is noteworthy. It represents an example of negative confounding, whereby unadjusted rates reveal no statistically significant differences [21,22]. After careful consideration of very important age as well as race/ethnicity differences, the association between female married status and TMT rates becomes clinically meaningful and statistically significant. The current observations regarding the association between female married status and TMT are consistent with other data validating the importance of married status in UCUB [7,12,23]. Interestingly, absence of an association between male married status and TMT also highlights the importance of stratified analyses according to sex. Indeed, married status is well known to exert a differential effect according to sex in a number of settings within multiple primaries [8,24]. For example, Rosiello et al. showed that unmarried males have worse cancer-specific mortality in metastatic renal cell carcinoma, while unmarried females do not [8].
The clinical implications of the current study are important. Our findings indicate that married females may benefit from better access to TMT. Conversely, unmarried females may experience barriers to TMT access. In consequence, careful consideration regarding whether a female represents a candidate for TMT should be recommended. Such recommendations are especially important in unmarried females, where potential barriers to TMT access may exist. In agreement with Rosiello et al., this recommendation is consistent with similar suggestions from other analyses where access to treatment was higher in married patients than in their unmarried counterparts [8,25,26].
Our observations are also of importance when study design and conduct decisions are made. Specifically, our observations indicate that various socioeconomic status (SES) variables may operate differently according to patient sex. Moreover, our observations indicate that the magnitude or even presence of patient characteristics differences may vary significantly according to patient sex. In consequence, strict multivariable adjustment is essential when SES-characteristics are examined in epidemiological or clinical risk factor analyses. Presence of negative confounding was exemplified by absence of TMT rate differences according to unadjusted married status. This phenomenon was recorded in females prior to multivariable adjustment. The differences, according to the definition of negative confounding, emerged after adjustment for age and race/ethnicity covariates [21]. This phenomenon should alert investigators to the need for detailed and thorough considerations of various confounders, that may contribute to a similar scenario as the one recorded in the current study.
Several limitations of the study must be addressed. The first and foremost consists of its retrospective nature. This limitation is shared with similar studies that were based on the SEER database and relied on retrospective data designs [6,27,28]. The second important limitation consists of the lack of detail offered by the SEER database. For example, additional SES-characteristics should ideally be analysed. These may include income and education. Unfortunately, the current SEER database no longer offers these variables [29]. Lack of detail also applies to TMT. Specifically, its components lack granularity. For example, the extent and completeness of the performed TURB is unknown. The type, dose, and fractionation of radiotherapy are also unknown. Finally, type, dose, and duration of chemotherapy are also not provided. Last but not least, detailed description of tumor characteristics is also unavailable. These include tumor multifocality on example basis. Despite the presence of these limitations, all similar large-scale epidemiological analyses that are based on the SEER database or the National Cancer Database suffer from the same limitations [27,30,31].