Introduction

Mortality from prostate cancer in the USA has decreased by roughly 50% in the past two decades, largely due to early diagnosis and improved curative therapies [1, 2]. The 5-year survival rate for prostate cancer among all races is 99% [3]. There are currently more than 3.3 million prostate cancer survivors (45% of all male cancer survivors) in the USA, and this number is expected to reach 4.5 million by 2026 [3]. Cancer survivors, including those with prostate cancer, have poor healthcare outcomes when compared to the general population without cancer [4]. In order to provide better care for the prostate cancer survivors, it is important to understand the causes for poor outcomes and their relationship with socioeconomic and medical comorbidities [5, 6]. Often a diagnosis of cancer dominates the healthcare issues, suppressing all other non-cancer medical comorbidities during the acute and chronic phases of patient care [7]. A majority (62%) of prostate cancer survivors are over the age of 70, and more than 30% of them have multiple comorbidities [1, 8]. These comorbidities progress over time with age and can lead to recurrent hospitalization. In the general population, the most common diagnoses associated with hospital readmission are heart failure, pneumonia, psychoses and COPD [9]. Additional predictors of hospital readmissions include polypharmacy, previous hospital admissions and lower socioeconomic status [10,11,12].

There are numerous studies focused on the health care and quality of life of prostate cancer survivors including genitourinary system complications in the outpatient setting [13,14,15,16]. However, data are limited on what causes recurrent hospital admissions among prostate cancer survivors. A recent study has suggested that the type of prostate cancer treatment did not play a role in the reasons for hospitalizations among prostate cancer survivors [17]. It has also been reported that approximately 84% of hospital admissions among prostate cancer survivors are unrelated to prostate cancer [18]. Therefore, it is important to explore the types of admissions and associated risk factors with recurrent hospitalizations among prostate cancer survivors. Identifying the cause of recurrent hospitalizations might help devise strategies to prevent the number of hospitalizations among prostate cancer survivors. We hypothesize that risk factors for recurrent hospitalization among prostate cancer survivors will be similar to the general population with the exception of admissions related to genitourinary complaints.

Methods

Study design and data source

Detailed methods have been published [17]. Briefly, a retrospective review of medical records was performed for all general medical admissions from January 1, 2008, to December 31, 2010, at a single academic institution. All male patients older than 40 years with a prior history of prostate cancer that is those who had been diagnosed and/or treated for prostate cancer at least 2 years prior to the study observation period (including those who were under active surveillance) were included in the study population. There were 461 admissions eligible for study enrollment, of which 157 admissions were from patients with a single medical admission and 304 admissions were recurrent admissions from 88 patients. Recurrent admission for our study was defined as more than one admission during the 3-year study period. Therefore, the final study population consisted of 245 patients. To avoid possible immediate prostate cancer treatment-related bias, patients who were diagnosed or treated with surgery or radiation therapy within 2 years of the study observation period were excluded.

Study variables and measures

Demographic information (age and race), health behavior (smoking status) and access to health care (health insurance status) were extracted during extensive review of the medical records. Chronic disease burden was evaluated by recording several medical comorbidities, including those needed for the Charlson comorbidity index (CCI). Medical admissions were also categorized according to the systemic involvement as detailed earlier [17]. A detailed abstract sheet was developed for all the needed information. Two independent data abstractors entered the data separately using an abstraction sheet and whenever there was a discrepancy; the data were rechecked by a third person to confirm the correct information. The Institutional Review Board at Johns Hopkins Bayview Medical Center approved the study protocol.

Statistical analysis

Patients’ characteristics are presented as proportions and means. Unpaired t test and Chi-square tests were used to compare study population characteristics and determined significance at p <  0.05. We used logistic regression models for analyses to predict the odds of recurrent medical hospitalization for prostate cancer survivors. Logistic regression models were used for sociodemographic, prostate cancer-related risk factors, prostate cancer treatment/treatment-related side effects and medical comorbidity burden models separately, and the resultant models were then adjusted for all possible risk factors for the final analyses. The study data were analyzed using Stata statistical software (StataCorp LP, version 13.1).

Results

The mean age of the study population was 76 years, and the mean duration since prostate cancer diagnosis was 9.7 years. The mean hospital length of stay was 4.4 days for all study population. Patients were stratified into two groups—those with a single medical admission and those with recurrent admissions—and their characteristics were analyzed. The recurrent admission group had an average of 3.25 (SD = 1.9) admissions during the 3-year study period, approximately 70% (n = 62) of the recurrent admissions were within 90 days of first admission, and 80% were discharged to home. While most variables were similar across the two groups (Table 1), patients with recurrent hospitalization had a higher proportion with comorbidities (age-adjusted CCI > 3) and were more likely to be admitted with cardiovascular conditions. Patients with a single hospitalization were more likely to be admitted due to gastrointestinal problems.

Table 1 Characteristics of study population
Full size table

In unadjusted, bivariate analysis, 4 of the 16 clinical variables were associated with recurrent hospitalization among prostate cancer survivors (Table 2). None of the sociodemographic and prostate cancer or treatment-related variable was independently associated with recurrent hospitalization. Logistic regression analysis (Table 2) showed that odds of recurrent hospitalization among prostate cancer survivors were ~2.5 times higher (OR 2.55; 95% CI 1.35–4.79) among patients with congestive heart failure, twice higher (OR 2.04, 95% CI 1.09–3.81) among patients with chronic lung disease (including chronic obstructive lung disease/interstitial lung disease/asthma) and approximately three times higher (OR 2.87, 95% CI 1.03–8.03) for patients with metastasis after adjustment for all clinical variables and comorbid conditions (Table 2). However, in the multivariable regression analyses where sociodemographic, prostate cancer-related factors, primary prostate cancer treatment-related sequelae and clinical variables and comorbid conditions were simultaneously analyzed, the above-mentioned variables showed no statistical significance except for congestive heart failure (OR 3.90, 95% CI 1.25–12.2) and history of metastasis (OR 8.10, 95% CI 1.10–60.1).

Table 2 Unadjusted and adjusted logistic regression analyses for risk factors associated with hospital readmission among prostate cancer survivors
Full size table

The most common cause of recurrent admissions among prostate cancer survivors was congestive heart failure (14%), followed by upper gastrointestinal bleed (9%), COPD exacerbation (8%), acute renal failure (8%) and pneumonia (7%) (Table 3).

Table 3 Most common reasons for admission and recurrent admissions
Full size table

Discussion

This study is the first attempt to identify and quantify risk factors associated with recurrent hospitalization among prostate cancer survivors. Using multivariable analysis model accounting for both sociodemographic and clinical variables simultaneously, congestive heart failure and a history of metastasis were each independently associated with recurrent hospitalization among prostate cancer survivors. Admissions from cardiovascular causes were significantly higher in the recurrent admissions group (p value—0.007) compared to the single admission group. These results corroborate previous studies reporting that the most common cause of hospital admission among the elderly was cardiovascular disease (20) and the Healthcare Cost and Utilization Project (HCUP) report that showed CHF among the most frequent reasons for hospital readmissions among Medicare population age 65 years or older, followed by septicemia and pneumonia [19]. The average length of stay for prostate cancer survivors in our study was also similar in both the groups reiterating previous studies that length of stay does not affect readmission rates [20, 21]. These results indicate that the major factors that influence hospital readmissions among prostate cancer survivors are not very different from those among the general population.

However, upper gastrointestinal bleeding, which constitutes about 9% of the total recurrent admissions in our study, has not been reported among the most frequent causes for readmission [19]. This cannot be explained by the side effects of prostate cancer treatment. Radiation therapy associated gastrointestinal side effects are commonly seen after the first 3 years of prostate cancer treatment and unlikely to present a decade post-treatment among prostrate cancer survivors. The radiation-related side effects include enteritis, bleeding telangiectasia and secondary cancers [22]. However, when all GI-related problems were considered, it represented the initial presenting complaint for 19% of the hospitalization among the prostate cancer survivors group with single admissions and only 5% of the hospitalization among those with recurrent admissions (Table 1).

Cardiovascular diseases are the most common non-cancer-related cause of death in prostate cancer survivors [23], and specifically androgen deprivation therapy (ADT) has shown to increase the risk of coronary artery disease and myocardial infarction [24]. Although our study did not show acute myocardial infarction as a predictor of recurrent admission, congestive heart failure was found to be a strong predictor of recurrent hospitalization. Current recommendations are that prostate cancer patients with cardiac disease, in whom ADT is initiated, receive appropriate secondary preventive measures for coronary artery disease (including lipid-lowering therapy, antihypertensive therapy, glucose-lowering therapy, and antiplatelet therapy) [25].

Patients with disseminated prostate cancer are commonly treated with androgen deprivation treatment [26], which in turn can influence volume status among patients with congestive heart failure. Thus, it is not surprising that metastatic disease was a significant predictor for recurrent hospital admissions even after adjusting for other variables. However, the significance of metastatic disease should be interpreted with caution due to wide confidence interval likely because of small number of prostate cancer patients with metastasis. Surprisingly chronic lung disease, which is the fourth major cause for hospital readmissions, was not a risk factor for recurrent admissions among prostate cancer survivors, and this may be due to the fact that smoking (a major risk factor for lung disease) is not prevalent in this age group [27].

Several limitations of this study should be considered. First, this study was conducted at a single hospital. Second, in this retrospective chart review (heavily relied on medical documentation by clinicians) while we attempted to evaluate a wide spectrum of variables including sociodemographic, clinical comorbidities, prostate cancer treatment and related sequelae that can potentially influence recurrent hospitalization, we may not have accounted for all relevant factors. Third, we did not compare the type of hospital admissions with patients without prostate cancer to better evaluate influence of prostate cancer treatment and treatment-related side effect on recurrent hospitalization among prostate cancer survivors. Fourth, our study population was small; however, it was large enough to provide a reliable estimate for predictors of recurrent hospitalization among prostate cancer survivors. Finally, it can be argued that average time since diagnosis of prostate cancer was short, although 9 years seemed to be enough time to give us insight into reasons of recurrent admission prostate cancer survivors during first decade in their cancer survivorship.

Prostate cancer survivors have a long survivorship time and are frequently admitted to the hospital for chronic medical conditions (congestive heart failure, COPD exacerbation, etc.) or for acute illness like gastrointestinal bleeding, pneumonia and acute renal failure. Understanding the nature of these admissions and associated medical comorbidities may help us in developing screening or preventive strategies to reduce the readmissions for this group of cancer survivors.