Background
Although depression and anxiety are considered common in MS [
1,
2], population-based prevalence estimates for these conditions are rare. Even fewer prevalence estimates exist for bipolar disorder and schizophrenia in the MS population, and they vary widely [
3,
4]. The paucity of population-based studies of mental comorbidity may reflect the challenges of conducting such studies. However, such studies are needed given the impact of mental comorbidity in MS, including lower quality of life and reduced adherence to treatment [
5,
6]; and to minimize the biases from using clinic-based samples.
Studies of mental comorbidity could potentially use one of several data sources including medical records review, self-report, interviews, or administrative data. Administrative data are population-based in publicly funded health systems such as Canada and are cost-effective and accessible [
7]. Such data are useful for assessing the burden of disease at the population level, including health services use and costs [
8]. Mental comorbidities can be assessed in clinical samples using structured diagnostic interviews such as the Composite International Diagnostic Interview (CIDI) although these are time consuming and depend heavily on recall of past episodes [
9]. Administrative data have the advantage that they are recorded during an episode and need not be recalled later. Administrative data, however, are collected for health system management and are often inadequately validated [
7,
10]. Indeed, few published case definitions for mental comorbidity have been validated, and efforts to develop and validate case definitions for depression have identified poor concordance with the CIDI Short Form [
11], and difficulties distinguishing depression from anxiety [
12].
We aimed to validate administrative case definitions for several mental comorbidities in MS, and to describe their prevalence among persons with MS versus a matched cohort from the general population. We hypothesized that the prevalence of depression, anxiety, bipolar disorder and schizophrenia would be higher in the MS population than in the general population.
Discussion
Few population-based studies have evaluated the prevalence of mental comorbidity in MS [
4]. To facilitate such studies, we validated case definitions for mental comorbidities based on hospital, physician and prescription claims. Our case definitions showed almost perfect agreement versus medical records for schizophrenia, and moderate agreement for any mental comorbidity, any mood or anxiety disorder, and depression. The case definition for bipolar disorder had lower agreement, but acceptable sensitivity of 75% and high specificity of 97.5%. The case definition for anxiety had the lowest agreement but a specificity of 82%. Further, persons with MS were at increased risk of depression, anxiety and bipolar disorder, but not schizophrenia when compared to the general population.
Previous validation studies of administrative case definitions for mental comorbidity were often disappointing, and have highlighted the challenges of distinguishing depression from anxiety when using 3-digit ICD codes [
11,
12]. In a Manitoba study, agreement was only fair (k = 0.26) between surveys and administrative definitions for depression which used hospital, physician and prescription claims [
11] although this lower agreement may reflected their use of survey data and a broader range of prescription claims than in our study. Among persons newly treated with antidepressants in Saskatchewan, Canada, agreement between depression identified on physician claims and medical records was moderate (k = 0.54), similar to our findings [
29]. We could not identify any published, validated case definitions for anxiety. Thus these validated case definitions augment the ability to conduct population-level surveillance of depression and anxiety. Despite challenges in developing case definitions for depressive and anxiety disorders sensitive and specific definitions were available for bipolar disorder (sensitivity 75%, specificity 97%). Among 225 Americans, inpatient diagnoses of bipolar disorder, outpatient diagnoses of bipolar disorder by mental health providers, and outpatient diagnoses of bipolar disorder by non-mental health providers that were accompanied by a prescription for lithium, carbamazepine or valproate, had false positive rates below 10% [
30]. However, we found that bipolar disorder could be identified without such claims. Consistently, administrative case definitions for schizophrenia have performed well, with agreement between hospital claims for schizophrenia and medical records of 93.9-100% [
31,
32]. In American Medicaid data, the case definition that we validated of either one hospital or two physician claims for schizophrenia in two years identified only 6% false positives (k = 0.76) [
33].Collectively, this suggests that administrative data can accurately identify bipolar disorder and schizophrenia in the MS and general populations.
Our approach is informative for researchers wishing to study mental comorbidity in other chronic neurologic diseases, which share the potential problem of under-reporting of comorbidities due to coding biases [
34]. While prescription claims may add sensitivity, their use for mental comorbidity is challenging in chronic neurologic diseases because of the frequent off-label use of therapies. By restricting the breadth of prescriptions used, and using them in combination with a physician claim for mental comorbidity we successfully created valid case definitions.
Prior studies suggest that the annual prevalence of depression in MS is up to 14% with a lifetime prevalence of up to 50% [
1], and that anxiety disorders affect more than 30% of persons with MS [
2,
22]. Our crude prevalence estimates of 33% for depression and 37% for anxiety based on two years of administrative data are consistent with those observations. The age-standardized prevalence of bipolar disorder in the MS population was 5.83% (crude prevalence 6.3%), 70% higher than in the general population. Studies in hospital or clinic populations suggested that bipolar disorder affects 0.30% to 13% of the MS population [
24‐
28]. The only one of these studies that used a true general population control group reported that hospitalized persons with MS had bipolar disorder twice as often as hospitalized controls (1.97% vs. 0.92%) [
25]. Since that study was limited to hospitalized persons, the prevalence of bipolar disorder may have been underestimated, although the increased risk of bipolar disorder in MS was similar to our findings.
The prevalence of schizophrenia was the same in the MS and general populations (0.93%). Two population-based studies, both using administrative data, evaluated the prevalence of psychosis, not limited to schizophrenia. In Taiwan, psychosis affected 7.5% of the MS population and 2.0% of the general population (odds ratio 4.0) [
35]. In Alberta, Canada only 0.8% of MS patients had non-organic pyschoses including schizophrenia-spectrum disorders, and other non-organic psychoses, but this was more than in the general population [
4]. Our findings of an absence of an increased prevalence of schizophrenia in our MS population suggest a lack of increased risk which may reflect differences in the psychotic disorders studied (all versus schizophrenia alone), as well as the small number of persons with schizophrenia.
Medical records review for the validation cohort did not involve all records of all health care providers over the lifetime of study participants. For practical reasons we also compared medical records to administrative data for the 1 to 5 year period ending in fiscal year 2005/06, rather than from 1984 onward. Both factors may have reduced agreement between the data sources. Like medical records, administrative data only allow us to identify mental comorbidities for which the patient has been treated; undiagnosed mental comorbidity cannot be captured without a direct patient assessment. This study had several strengths, however. We validated the case definitions in a population similar to the one in which it was applied, the design was population-based, we used matched general population controls, and we used multiple types of administrative data.
Competing interests
Ruth Ann Marrie receives research funding from: Canadian Institutes of Health Research, Public Health Agency of Canada, Manitoba Health Research Council, Health Sciences Centre Foundation, Multiple Sclerosis Society of Canada, Multiple Sclerosis Scientific Foundation, Rx & D Health Research Foundation, and has conducted clinical trials funded by Bayer Inc. and Sanofi-Aventis.
John Fisk is the Director of the endMS Atlantic Regional Research and Training Centre which is funded by the Multiple Sclerosis Society of Canada. He receives research funding from the Canadian Institutes of Health Research (CIHR) and in the past has received grants, honoraria and consultation fees from AstraZeneca, Bayer, Biogen-Idec Canada, Heron Evidence Development Limited, Hoffmann-La Roche, MAPI Research Trust, Novartis, Sanofi-Aventis, Serono Canada, and QualityMetric Incorporated.
Nancy Yu receives research support from the Canadian International Development Agency, the Multiple Sclerosis Society of Canada, CIHR, and Manitoba Health and Healthy Living.
Stella Leung reports no disclosures.
Lawrence Elliott receives research support from the Canadian Institutes of Health Research, Health Sciences Centre Foundation, Public Health Agency of Canada, and the Multiple Sclerosis Society of Canada.
Patricia Caetano has worked on a research project funded by Amgen.
Charity Evans reports no disclosures.
Sharon Warren receives research funding from the CIHR, the Canadian Health Services Research Foundation, Alberta Health Services and SSHRC.
Christina Wolfson receives research funding from the Multiple Sclerosis Society of Canada, Canadian Institutes of Health Research, Canada Foundation for Innovation, and Public Health Agency of Canada.
Larry Svenson reports no disclosures.
Helen Tremlett currently receives funding from: the Multiple Sclerosis Society of Canada [Don Paty Career Development Award]; US National MS Society [#RG 4202-A-2 (PI)]; Canadian Institutes of Health Research [MOP: #190898 (PI) and MOP-93646 (PI)]; Michael Smith Foundation for Health Research (Scholar award) and the Canada Research Chair program. She has received speaker honoraria and/or travel expenses to attend conferences from: the Consortium of MS Centres, US National MS Society, Swiss Multiple Sclerosis Society, the University of British Columbia Multiple Sclerosis Research Program, Teva Pharmaceuticals and Bayer Pharmaceutical (honoraria declined) and ECTRIMS. Unless otherwise stated, all speaker honoraria are either donated to an MS charity or to an unrestricted grant for use by her research group.
James Blanchard receives research support from the Multiple Sclerosis Society of Canada, CIHR, Bill & Melinda Gates Foundation, Canadian International Development Agency and the United States Agency for International Development.
Scott Patten was a member of an advisory board for Servier, Canada. He has received honoraria for reviewing investigator-initiated grant applications submitted to Lundbeck and Pfizer and has received speaking honoraria from Teva and Lundbeck. He is an Associate Editor for the Canadian Journal of Psychiatry and a member of the editorial board of Chronic Diseases and Injuries in Canada. He is the recipient of a salary support award (Senior Health Scholar) from Alberta Innovates, Health Solutions and receives research funding from the Canadian Institutes for Health Research, the Institute of Health Economics and the Alberta Collaborative Research Grants Initiative.
Authors’ contributions
RAM, JDF, SW, SBP, and HT conceived of and designed the study initially. RAM, JDF, LE, PC, CE, SW and SBP reviewed and selected diagnostic codes and pharmacotherapies for algorithm development. RAM, NY and SL analyzed the data. All authors assisted in the interpretation of the data. RAM drafted the manuscript. All authors revised the manuscript and approved the final version for publication.