Treatments and Corresponding Costs of Prostate Cancer
This study has several strengths and limitations associated with its study design and data analysis. First, a retrospective claims data analysis limits the amount of clinical variables available (e.g., cancer grade), which would be valuable in further understanding of the treatment patterns. Second, the analysis relies on selecting claims based on ICD-9 codes that accurately represent the specific population of interest. Accordingly, there is a possibility that patients with PCa could be omitted if a diagnosis indicative of PCa was not included in standard diagnosis fields of the claims database, resulting in misclassification of patients; however, our analysis included 211,440 encounters for men who received PCa treatment in a nationally representative hospital database, giving a good sample for evaluation. It should be noted that patients having ICD-9 codes for rectal (ICD-9: 154) or bladder cancer (ICD-9: 188) or secondary cancer codes (ICD-9: 196–198) in addition to the PCa diagnosis were not excluded given that PCa may invade into these areas. Posthoc analyses indicated that including these patients had little to no influence on parameter estimates. Also, patients having cancer diagnosis codes of 173.xx or 209–239.xx were not excluded, as these cancers were thought to have no clinical impact on patient survival. Inclusion of these patients was also shown to have no influence on parameter estimates.
Additionally, our analysis sought to describe utilization and costs among patients who received treatment during their hospital encounter. As such, encounters that were not associated with treatment were excluded. As a result, encounters associated with a watchful waiting or active surveillance approach may have been excluded. Last, the total costs of each hospital encounter represent the total costs of treating a patient with metastatic PCa, which includes the costs of treating comorbid conditions and complications that may have occurred during the hospitalization.
In an evaluation of the patient demographics for the inpatient-treated PCa population, the mean age was 69 years, which corresponds well with SEER data, which indicate that the mean age of diagnosis for PCa is 67 years. This similarity in demographics gives our analysis good generalizability to the overall PCa population. However, 68% of the population was Caucasian men, while African–Americans represented approximately 12% of patients. Previous studies have demonstrated that PCa disproportionately affects African–American men and, worldwide, African–American men have the highest incidence of PCa. Therefore, the difference in racial demographics in our population compared with the general PCa population could be considered a study limitation.
Finally, newer therapies that entered the market shortly before (cabazitaxel) and after (abiraterone) the end date of this study were not reflected in the results. Future research re-evaluating this topic once these agents have become a part of the treatment paradigm will be necessary to determine emerging treatment patterns.
Strengths & Limitations
This study has several strengths and limitations associated with its study design and data analysis. First, a retrospective claims data analysis limits the amount of clinical variables available (e.g., cancer grade), which would be valuable in further understanding of the treatment patterns. Second, the analysis relies on selecting claims based on ICD-9 codes that accurately represent the specific population of interest. Accordingly, there is a possibility that patients with PCa could be omitted if a diagnosis indicative of PCa was not included in standard diagnosis fields of the claims database, resulting in misclassification of patients; however, our analysis included 211,440 encounters for men who received PCa treatment in a nationally representative hospital database, giving a good sample for evaluation. It should be noted that patients having ICD-9 codes for rectal (ICD-9: 154) or bladder cancer (ICD-9: 188) or secondary cancer codes (ICD-9: 196–198) in addition to the PCa diagnosis were not excluded given that PCa may invade into these areas. Posthoc analyses indicated that including these patients had little to no influence on parameter estimates. Also, patients having cancer diagnosis codes of 173.xx or 209–239.xx were not excluded, as these cancers were thought to have no clinical impact on patient survival. Inclusion of these patients was also shown to have no influence on parameter estimates.
Additionally, our analysis sought to describe utilization and costs among patients who received treatment during their hospital encounter. As such, encounters that were not associated with treatment were excluded. As a result, encounters associated with a watchful waiting or active surveillance approach may have been excluded. Last, the total costs of each hospital encounter represent the total costs of treating a patient with metastatic PCa, which includes the costs of treating comorbid conditions and complications that may have occurred during the hospitalization.
In an evaluation of the patient demographics for the inpatient-treated PCa population, the mean age was 69 years, which corresponds well with SEER data, which indicate that the mean age of diagnosis for PCa is 67 years. This similarity in demographics gives our analysis good generalizability to the overall PCa population. However, 68% of the population was Caucasian men, while African–Americans represented approximately 12% of patients. Previous studies have demonstrated that PCa disproportionately affects African–American men and, worldwide, African–American men have the highest incidence of PCa. Therefore, the difference in racial demographics in our population compared with the general PCa population could be considered a study limitation.
Finally, newer therapies that entered the market shortly before (cabazitaxel) and after (abiraterone) the end date of this study were not reflected in the results. Future research re-evaluating this topic once these agents have become a part of the treatment paradigm will be necessary to determine emerging treatment patterns.