Ours is the first population-based study to examine the association between pre-existing depression and stage at diagnosis, treatment, and survival in patients with adenocarcinoma of the pancreas. Patients with depression were less likely to undergo surgical resection (for locoregional disease), chemotherapy, or radiation, and were less likely to see a surgeon or medical oncologist. Depression was associated with poorer 2-year survival in patients with locoregional disease, but this association was partially attributed to failure to undergo surgical resection in the depressed group. In patients who underwent surgical resection for locoregional disease, there was no difference in the receipt of adjuvant chemotherapy and radiation between depression groups. However, in patients who did not undergo resection for their locoregional disease, depressed patients were less likely to receive chemotherapy and radiation. This may indicate an “all or nothing” trend in the treatment of depressed patients with pancreatic cancer. Likewise, the negative association between depression and long-term survival in patients with advanced stage disease was largely explained by failure to receive chemotherapy.
Our results are consistent with previous studies that revealed that depression is associated with advanced stage at presentation, lower odds of receiving definitive treatment, and poorer survival in patients with a variety of cancer types.16,19,25,26,28,32,33
In our population-based study, pre-existing depression was a significant predictor of survival in patients with adenocarcinoma of the pancreas. However, after adjusting for receipt of therapy, this association diminished, signifying that the difference in survival in patients with depression is partly mediated by receipt of appropriate therapy for pancreatic cancer. Our results differ from Sheibani-Rad et al,34
who assessed depression in patients with pancreatic adenocarcinoma in a single-institution study. These authors found no significant difference in stage at presentation or survival between the groups of patients with and without depression.34
Theoretically, many of the patients in this study had depression before the diagnosis of pancreatic cancer. Patients with a nihilistic attitude toward their disease because of depression (or any other reason) would likely not seek consultation at a specialized center. Therefore, this group of patients may not be representative of the general population of patients with pancreatic cancer and pre-existing depression, because all of them sought care at a tertiary referral center and were more likely to undergo treatment. Because some of the observed survival advantage in our study was mediated by receipt of therapy, it might explain the difference between the current study and theirs. In addition, this study is limited by a small sample size and may have been underpowered to detect significant differences between depressed and nondepressed patients.
As mentioned above, our study reveals that patients with depression were less likely to see a surgeon or medical oncologist and less likely to undergo surgical resection, radiation, or chemotherapy, and that these disparities in treatment affected long-term survival. Unfortunately, the reasons behind these disparities in treatment (lack of consultation or referral, noncompliance, or not being offered such therapy even after consultation) cannot be elucidated using claims data. Patients with depression may be less compliant with routine health care visits and screening and may be less likely to adhere to medical recommendations, while physicians may be hesitant about providing appropriate care to patients with mental disorders.28
Our study shows that accurately identifying and treating patients with depression may maximize treatment rates and improve survival and quality of life.
Approximately 8% of patients in our study had a diagnosis of depression before pancreatic cancer diagnosis, which is lower than in previously reported studies but consistent with depression rates in other studies using Medicare claims.16
In our cohort, 7.3% of patients (N
= 1,727) without depression in the 3 to 27 months before their cancer diagnosis subsequently had a code for depression between 3 months before cancer diagnosis and onward. This rate is lower than would be expected, given that the prevalence of depression in pancreatic cancer patients ranges from 33% to 50% in the literature.20,22,35
The observed low prevalence in our study is likely related to both undercoding and underrecognition of depression. By using claims data, a diagnosis of depression is only identified when physicians enter a billing code. When treating patients with cancer, identifying, coding, and treating depression may be lower on the list of priorities for physicians, who may feel that treating the cancer is the main concern. Similarly, the symptoms of depression, including fatigue, pain, weight loss, and anorexia, and those of cancer can overlap, making the diagnosis of either difficult.17,20,21,36-41
Because of this, it is likely that patients with depression went unrecognized and were classified in our study as nondepressed based on billing codes. This would bias our results toward the null hypothesis. However, selection bias may also be affecting our results in the opposite direction. It is possible that only patients with the most severe cases of depression were identified, which would bias our results away from the null hypothesis. We repeated our analyses with the inclusion of the 7.3% of our cohort that was diagnosed with depression from 3 months before cancer diagnosis onward. This analysis indicates that a diagnosis of depression after cancer diagnosis is associated with decreased odds of presenting with advanced stage disease and improved survival. In fact, patients that were diagnosed with depression after cancer diagnosis had higher rates of surgical resection and chemotherapy, supporting our conclusion that the association between depression and survival is mediated by the receipt of appropriate therapy.
Our study shows that depression is a significant predictor of survival in patients with pancreatic adenocarcinoma. Particularly in patients with distant stage disease, where overall survival is notoriously poor, the impact of depression on survival is small---likely on the order of weeks. The influence of depression in patients with locoregional disease is more pronounced. However, the improvement in survival achieved by accurately recognizing and treating patients with depression is on par with the additional weeks attributed to traditional treatments for pancreatic cancer (surgery and chemotherapy).42-45
Improving treating physician (surgeon or oncologist) awareness of the high prevalence of depression and appropriate treatment of depression once it is identified can decrease the impact of depression on outcomes in pancreatic and other cancers. Screening for depression can be easily performed by physicians treating patients with pancreatic cancer. Currently, the most cost effective and efficient method is the use of a brief screening questionnaire to identify cancer patients with depressive symptoms, a variety of which are available.14,19,22,25,32,33,35,38,39,41,46,47
Once identified, patients with depression should be treated with a multidisciplinary approach. Psychiatric or psychological consultation should be sought to optimize the patient’s mental health. Physicians specializing in treating pancreatic cancer (oncologists and surgeons) should be aware of the risks associated with pre-existing depression (decreased odds of receiving appropriate treatment, prolonged hospital stay, and decreased survival) and counsel patients appropriately. The treatment of depression is unlikely to have a significant effect on overall survival but might contribute to improved consultation and treatment rates. In addition, screening for depression is simple and inexpensive and treatment may improve quality of life even without survival benefit.48-50
Our study has several limitations. We evaluate stage at diagnosis as a surrogate for delay in diagnosis. However, the SEER-Medicare data do not allow us to determine the time from onset of symptoms to definitive diagnosis. In addition, while differences in stage at presentation are statistically significant, they may not be clinically significant. Likewise, only a small proportion of the overall cohort have lymph node or tumor size information. This information was most commonly available in locoregional disease. As such, we do not compare these variables in distant disease and limit our conclusions to those with locoregional disease. Incomplete information available with regard to lymph node status and tumor size may introduce bias. Resected tumors are more likely to have lymph node and stage information, so it is possible that those with missing data represent understaging and actually have worse tumors. Regardless, when the group of patients with locoregional disease is considered together, depressed patients were less likely to see specialists and to receive treatment. In addition, in the group of patients who were resected, depression predicted survival even after controlling for tumor size and nodal status, where the incidence of missing data was low.
Our study population was limited to patients ≥67 years of age and those with Medicare data available for linkage. Given that the mean age at diagnosis is 66 years for patients with pancreatic cancer,30
the Medicare population is fairly representative of the overall population with pancreatic cancer. We speculate that younger patients with pancreatic cancer, who are not Medicare eligible, would be more likely to have fewer comorbidities, including depression (beacuse depression is more common in older patients), and are therefore more likely to be offered aggressive therapy. In addition, they would be more likely to be married and have greater social support and perhaps more likely to receive such therapy when offered. Our analysis also excludes patients in Medicare health maintenance organizations, because outpatient claims are not available for these patients. Such patients may experience different referral and treatment patterns. Finally, our follow-up is limited to 2 years. It is possible that if we follow our entire cohort for a longer period of time, the association between depression and survival might be mitigated. We repeated our statistical analysis with a restricted cohort of patients with at least 5 years of follow-up (diagnosed between 1992–2002). The patterns were similar to the 2-year survival patients. Depression predicted 5-year survival in patients with locoregional disease, although this effect was no longer significant after adjusting for surgical resection. In patients with resected locoregional disease, depression was not a significant predictor of 5-year survival. This suggests that in patients with resected locoregional disease, tumor characteristics are much stronger predictors of long-term survival. In those with distant disease, 5-year survival is not likely regardless of treatment or comorbidity, so this endpoint is not relevant.
Our study is also subject to selection bias, with patients being more fit for aggressive therapy (surgical resection or chemotherapy) being more likely to be offered such therapy. The improved survival in treated patients, therefore, is likely reflective of both the effect of treatment and the patient’s overall status. We have attempted to adjust for this by including patient age, marital status, and comorbidities in our statistical models, although it is impossible to completely control for selection bias in a study of this type.
In conclusion, this national, population-based study shows that pre-existing depression in patients with pancreatic cancer is associated with advanced stage at diagnosis, decreased likelihood of receiving adequate treatment, and poor survival. Our study indicates that the prevalence of depression in pancreatic cancer patients is lower than expected, which is, at least in part, related underrecognition. Improved vigilance regarding the diagnosis of depression in patients with pancreatic cancer may improve treatment rates and survival. In addition, the treatment of these patients can enhance quality of life in the face of poor survival.
The collection of the California cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute’s (NCI) Surveillance, Epidemiology and End Results Program under contract N01-PC-35136 awarded to the Northern California Cancer Center, contract N01-PC-35139 awarded to the University of Southern California, and contract N02-PC-15105 awarded to the Public Health Institute; and the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement #U55/CCR921930-02 awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and endorsement by the State of California, Department of Public Health, the NCI, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should it be inferred. The authors acknowledge the efforts of the Applied Research Program (NCI); the Office of Research, Development and Information (CMS); Information Management Services (IMS), Inc; and the Surveillance, Epidemiology and End Results Program tumor registries in the creation of the SEER–Medicare database.