Preadmission antidepressant use and bladder cancer: a population-based cohort study of stage at diagnosis, time to surgery, and surgical outcomes

As the ninth most commonly diagnosed cancer worldwide, bladder cancer caused a reported 165,000 deaths globally in 2012 [1]. Recommended standard treatment for muscle-invasive and high-risk non-muscle-invasive tumors is radical cystectomy, a procedure with 30-day complication rates between 24% and 73% [2]. Comorbidity raises complication risk [3], and bladder cancer patients are increasingly comorbid [4], with increased medication use including antidepressants as a consequence. The overall use of antidepressants increased by a mean of 40 defined daily doses/1000/day across Europe in the period 1980–2009 [5].

Although increasingly prescribed for neuropathic pain and sleep disorders, antidepressants are mostly prescribed for depression [6]. In a Danish setting, use of antidepressants detects patients suffering from a depression with a specificity of 94% to 97% [7]. This makes prescription data about antidepressants suited for identification of patients with depression in Denmark, where more than 90% of such patients are treated by general practitioners [7]; the treatment in these cases is not recorded in hospital registries.

In a recent Danish cohort study of patients with various types of cancer, prior antidepressant use was associated with a 30% increase in one-year mortality [8]. The study supports prior studies reporting an elevated mortality in patients with depression who developed various cancers [9‐11], pancreatic cancer [12], colon cancer [13], or prostate cancer [14]. Diagnostic delay, reduced likelihood of appropriate treatment, or treatment complications may explain an increased mortality in antidepressant users. However, studies of delay in diagnosis of breast, prostate, colorectal, bladder, lung, skin, uterine, and pancreatic cancer for patients with depression show conflicting results [8‐10, 12]. Prior antidepressant use was not associated with more advanced bladder cancer stage in the recent Danish work [8]. Although previous studies have reported a lower likelihood of appropriate cancer treatment in patients with prior or existing depression in a number of cancers combined [9], as well as pancreatic [12], colon [13], or prostate cancer [14], this has not been examined in patients with bladder cancer. The reported increase in one-year mortality among patients with cancer who used antidepressants before cancer diagnosis [8] is of concern and more detailed studies are needed to clarify underlying mechanisms.

Therefore, the aim of this nationwide population-based cohort study was to investigate whether antidepressant use initiated before invasive bladder cancer diagnosis was associated with more advanced cancer at time of diagnosis or time of surgery, lower rate of cystectomy, and a more complicated postoperative course reflected in length of hospital stay, readmission, postoperative procedures and mortality.

We performed a nationwide register-based cohort study including all patients diagnosed with incident invasive bladder cancer in Denmark from 2005 through 2015. The Danish population, approximately 5.6 million inhabitants, has access to tax-supported healthcare, with free access to hospital-based and primary medical care provided by the Danish National Health Service [15]. All inhabitants in Denmark have a unique Civil Personal Register (CPR) number, which is recorded along with administrative and medical information in registries and databases and allows for person-specific linkage of information [16].

We identified 10,427 patients with incident invasive bladder cancer during 2005 through 2015 (Fig. 1). In the cohort, 1079 (10%) were antidepressant users. At presentation, antidepressant users were less likely to be married and had more comorbidity than non-users (Table 1).

In this nationwide population-based cohort study, we found that patients with bladder cancer who filled at least two antidepressant prescriptions in the year before cancer diagnosis were less likely to present with muscle-invasive disease at diagnosis (i.e. they had lower odds of muscle-invasive disease relative to non-users upon adjustment). Users also had a lower rate of cystectomy in case of muscle-invasive disease. Following surgery, antidepressant users and non-users had comparable length of stay. However, we observed a slightly higher risk of postoperative procedures within 90 days and a higher 30-day readmission rate after surgery in antidepressant users than in non-users. Still, users’ all-cause mortality was not different from non-users at one and three years.

We observed similar proportions of patients with muscle-invasive cancer at diagnosis among antidepressant users and non-users. However, this observation was likely confounded since antidepressant users had lower odds of muscle-invasive disease upon adjustment for age, sex, comorbidity, and marital status. Our result is unexpected since depression has been associated with advanced stage at cancer presentation [8, 12]. One possible explanation is an enhanced likelihood of cancer detection due to increased utilization of health care by antidepressant users [23]. Study setting may add to this: access to the tax-funded Danish health care services is free [15], minimizing the risk of patient delay in diagnosis due to patients’ financial situation or constraints. However, Sun et al. found that Danish patients with breast, colorectal, or prostate cancer who were using antidepressants presented in an advanced stage [8]. This suggests that mechanisms underlying diagnostic delay may differ between cancer types. In bladder cancer, hematuria may be so concerning to patients that it leads to cancer diagnosis even if patients are impaired by depression. In other cancers, symptoms may be less alarming, or perhaps symptoms are considered caused by depression, leading to a diagnostic delay in patients with depression who develop cancer.

Previous studies have found a lower likelihood of treatment in cancer patients with prior depression [9, 12‐14]. This is in line with our finding of antidepressant users’ lower rate of cystectomy. Patient-related factors associated with decline of curative cancer treatment include patients’ perceived low quality of patient-physician communication [24], concern about side effects [25], and expected low quality of life after surgery [24] all of which may be exaggerated in antidepressant users. Another explanation could be that a larger proportion of antidepressant users than non-users received palliative care rather than radical cystectomy. We must note, however, that we did not have information on all clinical features and patient characteristics contributing to treatment decision.

We observed a high rate of postoperative procedures compared with prior studies reporting 90-day reoperation rates in the range 14–33% [2]. The discrepancy is likely explained by the fact that we included all types of procedures (not solely complication related procedures) performed 90 days after cystectomy. We note that the estimate in this analysis is somewhat imprecise, given that the 95% confidence interval includes one. Higher rates of readmission and, potentially, postoperative procedures among antidepressant users suggest a more complicated postsurgical course, while the lack of difference in length of stay speaks against any major difference. In prior studies, surgical patients with depressive symptoms had impaired wound healing and higher risk of infections [26, 27]. Our results are consistent with these findings, if antidepressant users had depressive symptoms at the time of surgery.

In contrast to prior work [8], we did not observe an association between antidepressant use and all-cause mortality in patients with cancer. Possible explanations may be that we restricted the survival analysis to patients undergoing surgery and that selection for surgery differed between antidepressant users and non-users. Antidepressant users and non-users who underwent surgery are a selected group of patients with bladder cancer. Differences in comorbidity are likely to be equalized because of the selection resulting in comparable survival. Also, Sun et al. [8] found that the increased mortality among antidepressant users depended on time of antidepressant initiation. This may contribute to the discrepancy between our studies. Another aspect to consider is severity of depression: generally, antidepressant users were probably less severely depressed than individuals included in earlier studies based on secondary psychiatric care data [10, 11], because these previous studies did not include patients with milder depression treated by general practitioners.

A strength of the current study is that by using register data, we were able to include all patients in Denmark with invasive bladder cancer during the study period. Also, we were able to control for differences in comorbidity in contrast to previous studies [9, 10]. We did not rely on self-report of exposure and we achieved essentially complete follow-up, thereby increasing generalizability. To reduce the risk of misclassification due to medication non-adherence, we required the exposed patients to have filled at least two prescriptions for antidepressants.

The study results must, however, be interpreted bearing in mind a number of limitations. Our analyses did not take into account indications for readmission and postoperative procedures. However, we only observed small differences between antidepressant users and non-users in types of postoperative procedures and readmission, and most procedure codes and readmission diagnosis codes reflected complications related to cystectomy.

We emphasize that use of antidepressants does not always indicate depression, since antidepressants are sometimes prescribed for other conditions. As such, our findings may not be directly comparable to studies investigating patients with a depression diagnosis. Also, some non-users may have suffered from a depression not treated by antidepressants. This could weaken associations between antidepressant use and outcomes such as time to cystectomy or postoperative outcomes. Prior studies based on a diagnosis of depression might have experienced a comparable misclassification of exposure due to undiagnosed depression or depression not recorded in registries [9‐14].

We note that antidepressant users were generally more comorbid than non-users; a potential confounding effect we accounted for by adjusting for CCI. However, we cannot rule out the possibility of residual confounding due to e.g. comorbidity not recorded in registries. Also, we lacked information on certain additional potential confounders, such as detailed clinical characteristics, smoking, alcohol use, and socioeconomic status. A previous study, however, found no clear association between socioeconomic status and radical cystectomy [28]. Although we restricted the analysis of time to surgery to patients with muscle-invasive disease, we cannot rule out that exposure groups differed with regard to lymph node status and/or distant metastases. If more antidepressant users had distant metastases, this could explain their lower rate of cystectomy. However, antidepressant users had less advanced cancer at diagnosis, suggesting that they were not more likely to have metastases compared with non-users. Moreover, we did not observe any mortality difference between groups, which would have been expected in case of unequal proportions of patients with distant metastases. Smoking increases the risk of surgical complications [29], and may in part confound our findings on readmission and postoperative procedures. However, estimates changed very little upon adjustment for other covariates (age, sex, comorbidity, and marital status). This leads us to believe that confounding by smoking does not completely explain the observed associations. Smoking is associated with more advanced stage of bladder cancer at diagnosis [30]. Thus, if antidepressant users were more often smokers, the lower stage at diagnosis observed among them would be unexpected. On the other hand, if antidepressant users were less often smokers, our observation of no difference in mortality after surgery could be confounded by smoking. Altogether, we cannot exclude the possibility that the inability to adjust for smoking leads to residual confounding.

Our findings suggest that patients with bladder cancer using antidepressants before diagnosis may be a subgroup worth special attention—especially in relation to receipt of treatment and postsurgical care. This also points to the importance of differentiated clinical care according to individual characteristics and individual needs. For example, patients with high levels of comorbidities preoperatively, dementia, depression, etc. are subgroups who we should devote more effort to in terms of preoperative preparation and closer postoperative care following discharge. In line with this, future studies may focus on other patient subgroups and evaluate aspects in relation to receipt of treatment and treatment outcomes as such knowledge is a prerequisite for individualized clinical care.

In conclusion, this nationwide population-based cohort study using Danish register data found that antidepressant use before invasive bladder cancer diagnosis was associated with less advanced stage of cancer at diagnosis. Among patients with muscle invasive disease, we observed lower rate of cystectomy in antidepressant users, and after surgery higher rate of readmission and, potentially, higher rate of postoperative procedures, but not prolonged length of stay or increased all-cause mortality. Our findings indicate that antidepressant users who develop bladder cancer may require increased medical attention.