Original Article
Patterns of Psychiatric Medication Use Among Nationally Representative Long-Term Cancer Survivors and Controls Ilana M. Braun, MD1; Sowmya R. Rao, PhD2,3; Fremonta L. Meyer, MD1; and Giuseppe Fedele, MD1,4
BACKGROUND: Investigations of long-term cancer survivors (LTCS) indicate that this population is not appreciably different from cancer-naive peers with respect to several neuropsychiatric domains. The current study sought to determine whether differences in psychiatric medication use might help to explain the negative findings. METHODS: In a nationally representative sample, 5692 subjects were queried for cancer history, psychiatric diagnoses, and psychotropic medication use. The LTCS were defined as those individuals who were 5 years from diagnosis and whose cancer was in remission or cured. Odds ratios and 95% confidence intervals were obtained from multivariable logistic regression models evaluating the relationship between cancer status and use of psychiatric medications. The interaction between case/control status and psychiatric diagnoses was also tested in a logistic regression model to predict psychotropic medication use. RESULTS: A total of 225 participants met the criteria for LTCS and 3953 met the criteria for cancer-naive controls (CNC). The LTCS were no more likely than CNC to carry a psychiatric diagnosis. Despite the LTCS reporting somewhat greater psychotropic medication use compared with the CNC (28.8% vs 22.3%), unadjusted and adjusted differences did not reach statistical significance, possibly due to sample size. The interaction between case/control status and carrying a psychiatric diagnosis was not found to be significantly associated with receiving a psychiatric medication. CONCLUSIONS: LTCS and CNC demonstrated comparable rates of psychiatric prescription medication use. The relationship between taking a psychiatric medication and carrying a psychiatric diagnosis was not found to be significantly different between the case and control groups. These findings contribute to an emerging hypothesis that in general LTCS are not a particularly psychiatrically vulnerable group. Cancer 2015;121:132-8. C 2014 American Cancer Society. V KEYWORDS: psychosocial oncology, epidemiology, cancer survivors, psychopharmacology, psychiatric diagnosis.
INTRODUCTION Advances in cancer diagnosis and treatment have contributed to a sharp increase in the number of individuals surviving cancer. Between 1971 and 2007, cancer survivors, defined as individuals living with a history of cancer, increased from 1.5% to 3.9% of the US population.1 In 2012, approximately 64% of US cancer survivors had been diagnosed with cancer within the previous 5 years, and 15% within the previous 20 years.2 Expansion of the cancer survivor population has justifiably prompted inquiries into whether survivors carry a distinct symptom burden that invites tailored medical or mental health care. In fact, in a previous publication, the Centers for Disease Control and Prevention, although not specifically addressing mental health issues, called for the implementation of “evidence-based cancer plans that include all stages of cancer survivorship.”3 In general, cancer survivors require mental health treatment at higher rates than cancer-naive controls (CNC).4-6 Although this appears to be the case for cancer survivors in general, investigations of long-term cancer survivors (LTCS), strictly defined as individuals surviving >5 years after diagnosis with cancer in remission or cured, indicate that this subpopulation is not appreciably different from their cancer-naive peers with respect to several neuropsychiatric domains. When compared with controls in large nationally representative databases, cancer survivors who are 4 years from diagnosis do not appear to demonstrate increased rates of current major depressive disorder, generalized anxiety disorder, posttraumatic stress disorder, panic disorder, or cognitive symptoms.7-10 In the single instance in which a positive association was found, the results were potentially confounded by the inclusion of LTCS who were still undergoing active cancer treatment, as
Corresponding author: Ilana M. Braun, MD, Division of Adult Psychosocial Oncology, Dana-Farber Cancer Institute, 450 Brookline Ave, Boston, MA 02215; Fax: (617) 632-6180; [email protected] 1 Division of Adult Psychosocial Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts; 2Department of Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, Massachusetts; 3Center for Health Quality, Outcomes and Economic Research, Veterans Administration Medical Center, Bedford, Massachusetts; 4Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, New York.
See editorial on pages 6-7, this issue. DOI: 10.1002/cncr.29014, Received: November 6, 2013; Revised: May 30, 2014; Accepted: June 23, 2014, Published online September 10, 2014 in Wiley Online Library (wileyonlinelibrary.com)
132
Cancer
January 1, 2015
Psychiatric Drugs in Cancer Survivors vs Controls/Braun et al
well as a reliance on broadly defined psychological distress, a term that does not necessarily map onto psychiatric diagnoses.11 This body of findings raises a fundamental question that to the best of our knowledge remains unexplored: namely, whether LTCS take more psychotropic medications, thereby contributing to the explanation for the absence of increased rates of psychiatric symptoms noted in this population. Evidence suggests that a substantial percentage of patients with cancer who are in active treatment receive psychotropic drugs. A large case-control study based on insurance data in the Netherlands demonstrated that patients with cancer are significantly more likely than CNC to receive prescriptions for psychiatric medications.12 Another study determined that >50% of a waiting room sample of patients with breast cancer had been prescribed either an antidepressant or anxiolytic.13 Plausibly, some patients with cancer might continue to receive psychiatric medications after the completion of active cancer treatment, or remain better connected than their cancer-naive peers to systems of providers that might prescribe them. The National Comorbidity Study Replication (NCS-R) provides a rare opportunity to probe this question. In the study, a total of 9282 individuals in a probability sample (weighted to approximate the 2000 US census) were interviewed regarding psychiatric disorders, treatments, and risk factors. A subset of 5692 participants was also queried about cancer status and history. The current study harnesses these rich data to compare LTCS and CNC with regard to use of psychiatric medications.
MATERIALS AND METHODS National Comorbidity Study Replication
The NCS-R is a psychiatric, cross-sectional, cluster area probability study conducted by trained lay interviewers in the contiguous United States. Complete methodological information is available elsewhere.14 Investigators identified potential participants through a multistage sampling process that consisted of probability samples, first of counties or metropolitan areas, then units of 50 to 100 households, and then a single household. Once a household was identified, researchers selected a potential participant randomly from among English-speaking individuals aged 18 years. Residents of military bases, nursing homes, and chronic care facilities were excluded. The overall participation rate was 74.6%. All NSC-R participants (n 5 9282) received part 1 of the survey, which covered lifetime psychiatric disorders Cancer
January 1, 2015
and demographic data; a subset of participants (n 5 5692) also received part 2. By design, this subset oversampled those with psychopathology and included all participants with a lifetime psychiatric disorder, 59% of those who were below the threshold for a psychiatric disorder or who reported ever having received mental health services, and 25% of those who neither met the criteria for a lifetime psychiatric disorder nor reported ever having received mental health services. Part 2 of the survey collected data regarding risk factors, additional psychiatric diagnoses, and medical comorbidities. The final data were weighted to adjust for differential probabilities in selection, nonresponse rates, and poststratification differences in demographic variables. For these analyses, we used the public access NCS-R database.15 Sample
Participants who indicated that they had a history of cancer were queried regarding the type of cancer, date of diagnosis, and current status (“in treatment,” “cured,” “in remission,” or “don’t know”). In the current analyses, we considered an LTCS to be someone who reported having been diagnosed with cancer after the age of 18 years (because of the effects cancer might have on psychological maturation) and who at time of the interview was either cured or in remission and at least 5 years from diagnosis. The CNC group consisted of individuals who denied cancer histories. Medications
In addition to the number of prescription and nonprescription medications (psychiatric or otherwise) a subject had used in the 7 days before the interview, the investigators gathered detailed information regarding psychotropic medication use under the supervision of a physician within the year before the interview. Interviewers provided subjects with a list of 215 psychotropic drugs, both brand name and generic, and were instructed both to continue probing until all drugs were mentioned and to consult medication bottles for names when participants encountered memory lapses (discussions were held in a respondent’s home).14 These medications were categorized in the original NCS-R database under the following umbrella headings: “antidepressants,” “sleeping pills/sedatives” (ie, nonbenzodiazepine, benzodiazepine receptor agonists), “tranquilizers” (ie, benzodiazepines), “amphetamines/stimulants,” and “antipsychotics.” Because chronic pain is a common report among LTCS, we added an umbrella term of “psychotropic drugs with possible pain indications.” These medications occasionally are used, either on-label or off-label, to manage 133
Original Article
neuropathic pain and include tricyclic antidepressants, fluoxetine, fluvoxamine, gabapentin, and venlafaxine (duloxetine was not in widespread use when the NCS-R was administered). For each psychotropic medication, subjects were asked to identify the reason for taking it and the prescriber (“psychiatrist,” “general or family physician,” “some other physician,” “some other health professional,” and “no one prescribed the medication”). The latter 4 categories were combined as “other” in the current analyses. Covariates Demographics
Age, sex, marital status (married/cohabiting, divorced/ separated/widowed, or never married), race (white, black, Hispanic, or other), education (
Patterns of Psychiatric Medication Use Among Nationally Representative Long-Term Cancer Survivors and Controls Ilana M. Braun, MD1; Sowmya R. Rao, PhD2,3; Fremonta L. Meyer, MD1; and Giuseppe Fedele, MD1,4
BACKGROUND: Investigations of long-term cancer survivors (LTCS) indicate that this population is not appreciably different from cancer-naive peers with respect to several neuropsychiatric domains. The current study sought to determine whether differences in psychiatric medication use might help to explain the negative findings. METHODS: In a nationally representative sample, 5692 subjects were queried for cancer history, psychiatric diagnoses, and psychotropic medication use. The LTCS were defined as those individuals who were 5 years from diagnosis and whose cancer was in remission or cured. Odds ratios and 95% confidence intervals were obtained from multivariable logistic regression models evaluating the relationship between cancer status and use of psychiatric medications. The interaction between case/control status and psychiatric diagnoses was also tested in a logistic regression model to predict psychotropic medication use. RESULTS: A total of 225 participants met the criteria for LTCS and 3953 met the criteria for cancer-naive controls (CNC). The LTCS were no more likely than CNC to carry a psychiatric diagnosis. Despite the LTCS reporting somewhat greater psychotropic medication use compared with the CNC (28.8% vs 22.3%), unadjusted and adjusted differences did not reach statistical significance, possibly due to sample size. The interaction between case/control status and carrying a psychiatric diagnosis was not found to be significantly associated with receiving a psychiatric medication. CONCLUSIONS: LTCS and CNC demonstrated comparable rates of psychiatric prescription medication use. The relationship between taking a psychiatric medication and carrying a psychiatric diagnosis was not found to be significantly different between the case and control groups. These findings contribute to an emerging hypothesis that in general LTCS are not a particularly psychiatrically vulnerable group. Cancer 2015;121:132-8. C 2014 American Cancer Society. V KEYWORDS: psychosocial oncology, epidemiology, cancer survivors, psychopharmacology, psychiatric diagnosis.
INTRODUCTION Advances in cancer diagnosis and treatment have contributed to a sharp increase in the number of individuals surviving cancer. Between 1971 and 2007, cancer survivors, defined as individuals living with a history of cancer, increased from 1.5% to 3.9% of the US population.1 In 2012, approximately 64% of US cancer survivors had been diagnosed with cancer within the previous 5 years, and 15% within the previous 20 years.2 Expansion of the cancer survivor population has justifiably prompted inquiries into whether survivors carry a distinct symptom burden that invites tailored medical or mental health care. In fact, in a previous publication, the Centers for Disease Control and Prevention, although not specifically addressing mental health issues, called for the implementation of “evidence-based cancer plans that include all stages of cancer survivorship.”3 In general, cancer survivors require mental health treatment at higher rates than cancer-naive controls (CNC).4-6 Although this appears to be the case for cancer survivors in general, investigations of long-term cancer survivors (LTCS), strictly defined as individuals surviving >5 years after diagnosis with cancer in remission or cured, indicate that this subpopulation is not appreciably different from their cancer-naive peers with respect to several neuropsychiatric domains. When compared with controls in large nationally representative databases, cancer survivors who are 4 years from diagnosis do not appear to demonstrate increased rates of current major depressive disorder, generalized anxiety disorder, posttraumatic stress disorder, panic disorder, or cognitive symptoms.7-10 In the single instance in which a positive association was found, the results were potentially confounded by the inclusion of LTCS who were still undergoing active cancer treatment, as
Corresponding author: Ilana M. Braun, MD, Division of Adult Psychosocial Oncology, Dana-Farber Cancer Institute, 450 Brookline Ave, Boston, MA 02215; Fax: (617) 632-6180; [email protected] 1 Division of Adult Psychosocial Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts; 2Department of Quantitative Health Sciences, University of Massachusetts Medical School, Worcester, Massachusetts; 3Center for Health Quality, Outcomes and Economic Research, Veterans Administration Medical Center, Bedford, Massachusetts; 4Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, New York.
See editorial on pages 6-7, this issue. DOI: 10.1002/cncr.29014, Received: November 6, 2013; Revised: May 30, 2014; Accepted: June 23, 2014, Published online September 10, 2014 in Wiley Online Library (wileyonlinelibrary.com)
132
Cancer
January 1, 2015
Psychiatric Drugs in Cancer Survivors vs Controls/Braun et al
well as a reliance on broadly defined psychological distress, a term that does not necessarily map onto psychiatric diagnoses.11 This body of findings raises a fundamental question that to the best of our knowledge remains unexplored: namely, whether LTCS take more psychotropic medications, thereby contributing to the explanation for the absence of increased rates of psychiatric symptoms noted in this population. Evidence suggests that a substantial percentage of patients with cancer who are in active treatment receive psychotropic drugs. A large case-control study based on insurance data in the Netherlands demonstrated that patients with cancer are significantly more likely than CNC to receive prescriptions for psychiatric medications.12 Another study determined that >50% of a waiting room sample of patients with breast cancer had been prescribed either an antidepressant or anxiolytic.13 Plausibly, some patients with cancer might continue to receive psychiatric medications after the completion of active cancer treatment, or remain better connected than their cancer-naive peers to systems of providers that might prescribe them. The National Comorbidity Study Replication (NCS-R) provides a rare opportunity to probe this question. In the study, a total of 9282 individuals in a probability sample (weighted to approximate the 2000 US census) were interviewed regarding psychiatric disorders, treatments, and risk factors. A subset of 5692 participants was also queried about cancer status and history. The current study harnesses these rich data to compare LTCS and CNC with regard to use of psychiatric medications.
MATERIALS AND METHODS National Comorbidity Study Replication
The NCS-R is a psychiatric, cross-sectional, cluster area probability study conducted by trained lay interviewers in the contiguous United States. Complete methodological information is available elsewhere.14 Investigators identified potential participants through a multistage sampling process that consisted of probability samples, first of counties or metropolitan areas, then units of 50 to 100 households, and then a single household. Once a household was identified, researchers selected a potential participant randomly from among English-speaking individuals aged 18 years. Residents of military bases, nursing homes, and chronic care facilities were excluded. The overall participation rate was 74.6%. All NSC-R participants (n 5 9282) received part 1 of the survey, which covered lifetime psychiatric disorders Cancer
January 1, 2015
and demographic data; a subset of participants (n 5 5692) also received part 2. By design, this subset oversampled those with psychopathology and included all participants with a lifetime psychiatric disorder, 59% of those who were below the threshold for a psychiatric disorder or who reported ever having received mental health services, and 25% of those who neither met the criteria for a lifetime psychiatric disorder nor reported ever having received mental health services. Part 2 of the survey collected data regarding risk factors, additional psychiatric diagnoses, and medical comorbidities. The final data were weighted to adjust for differential probabilities in selection, nonresponse rates, and poststratification differences in demographic variables. For these analyses, we used the public access NCS-R database.15 Sample
Participants who indicated that they had a history of cancer were queried regarding the type of cancer, date of diagnosis, and current status (“in treatment,” “cured,” “in remission,” or “don’t know”). In the current analyses, we considered an LTCS to be someone who reported having been diagnosed with cancer after the age of 18 years (because of the effects cancer might have on psychological maturation) and who at time of the interview was either cured or in remission and at least 5 years from diagnosis. The CNC group consisted of individuals who denied cancer histories. Medications
In addition to the number of prescription and nonprescription medications (psychiatric or otherwise) a subject had used in the 7 days before the interview, the investigators gathered detailed information regarding psychotropic medication use under the supervision of a physician within the year before the interview. Interviewers provided subjects with a list of 215 psychotropic drugs, both brand name and generic, and were instructed both to continue probing until all drugs were mentioned and to consult medication bottles for names when participants encountered memory lapses (discussions were held in a respondent’s home).14 These medications were categorized in the original NCS-R database under the following umbrella headings: “antidepressants,” “sleeping pills/sedatives” (ie, nonbenzodiazepine, benzodiazepine receptor agonists), “tranquilizers” (ie, benzodiazepines), “amphetamines/stimulants,” and “antipsychotics.” Because chronic pain is a common report among LTCS, we added an umbrella term of “psychotropic drugs with possible pain indications.” These medications occasionally are used, either on-label or off-label, to manage 133
Original Article
neuropathic pain and include tricyclic antidepressants, fluoxetine, fluvoxamine, gabapentin, and venlafaxine (duloxetine was not in widespread use when the NCS-R was administered). For each psychotropic medication, subjects were asked to identify the reason for taking it and the prescriber (“psychiatrist,” “general or family physician,” “some other physician,” “some other health professional,” and “no one prescribed the medication”). The latter 4 categories were combined as “other” in the current analyses. Covariates Demographics
Age, sex, marital status (married/cohabiting, divorced/ separated/widowed, or never married), race (white, black, Hispanic, or other), education (
