Ethics

This study was approved by the Regional Ethics Committee, Stockholm Sweden (approval no. 2007/1762-31).

Study population

In this population-based prospective cohort study, all individuals who were 17–64 years old and lived in Sweden in 2005 were selected from the Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA). This population-based register is held by Statistics Sweden and contains several demographic variables. After exclusion of 65,091 individuals (of which 41 were MS patients) with missing values on place of birth and/or educational level at baseline, the cohort consisted of 10,750 MS-patients, identified through nationwide registers, and 5,553,141 non-MS individuals. The subjects' unique personal identity numbers assigned to all Swedish residents were used to link data from different nationwide Swedish registers.

Covariates

Information for 2005 regarding age-groups (17–34, 35–44, 45–54, 55–64), sex, educational level (compulsory school (≤9 years), high school (10–12 years), university (≥13 years), country of birth (Sweden, the other Nordic countries, other EU 25, all other countries), and type of living area (based on the H-region classification scheme [24] into the following three categories: larger cities (H1–H2), medium-sized (H3–H4) municipalities or smaller municipalities (H5–H6)) was obtained from LISA.

Exposures and outcome variables

Diagnosis-specific data on MS, psychiatric hospitalization and specialized psychiatric out-patient care between 2000 and 2005 was obtained from the National Patient Register, held by the National Board of Health and Welfare. This register contains information on in-patient care since 1964 and specialized out-patient care since 2001. Presence of MS was defined as being hospitalized or receiving specialized care at least once between 2000 and 2005 with a primary or secondary diagnosis of the International Statistical Classification of Diseases and Related Health Problems, 10^th revision (ICD-10) [25] code G35. Psychiatric comorbidity was identified using ICD codes F00–F99, here organized into nine diagnostic categories; Developmental and organic disorders (F00–09, F70–F89), Disorders related to substance abuse (F10–F19), Schizophrenia and non-affective psychoses (F20–29), Bipolar disorder (F31), Depressive disorder (F32–F33), Affective disorders (F30–39, including bipolar disorder and depressive disorder), Neurotic and somatoform disorders (F40–F42, F44–F49), Stress-related mental disorders (F43), Behavioral disorders (F50–F59, F90–99), and Personality disorders (F60–F69).

The individuals were linked to data from the Swedish Prescribed Drug Register, which contains information on all prescribed drugs dispensed at a pharmacy in Sweden, to obtain information on all prescribed and purchased psychiatric medication from July to December 2005 (the register started July 1st 2005). Psychiatric medication was defined as having purchased drugs with the Anatomical Therapeutic Chemical (ATC) [26] codes N03–N07 at least once between July and December 2005. The codes were organized into nine different groups according to pharmacological class and/or clinical use: Selective Serotonin Reuptake Inhibitors (SSRIs; N06AB), Tricyclic antidepressants (TCAs; N06AA), other antidepressants (AD; N06AG02, N06AX), First generation antipsychotics (FGA; N05AA–N05AD, N05AF), Second and third generation antipsychotics (SGA); N05AE, N05AH0, N05AX), Benzodiazepines (N03AE01, N05BA, N05CD), Short-acting sleeping agents (N05CF), Lithium (N05AN01), and Alcohol dependence drugs (N07BB). Anti-convulsants and central stimulants were not included as their indications in MS patients were deemed dominantly somatic.

Information on being granted disability pension, the outcome measure, or old age pension in 2006–2010 was obtained from the national Social Insurance Agency's MiDAS database. In Sweden, all adult residents with a disease or injury that has led to permanent work incapacity can be granted disability pension. Disability pension, which may be granted for part- and full-time, covers up to 64% of the income lost. The customary age for old age pension is 65 years of, but may be granted earlier [27]. Date of death was obtained from the Cause of Death Register.

Analysis

First, descriptive analyses were performed to explore the distribution of the baseline covariates among all MS patients and the non-MS individuals, respectively. Second, logistic regression was used to calculate odds ratios (ORs) with 95% confidence intervals (CI) for psychiatric in- and out-patient care (2001–2005) and for psychiatric drug prescriptions (July–December 2005), for MS-patients compared to non-MS individuals. The analyses were adjusted for age group, sex, educational level, country of birth, and type of living area.

Third, among the 4571 MS patients not on disability pension in 2005, survival analyses were performed with psychiatric diagnoses and medication as exposures, and disability pension as the outcome. The cohort members were followed from 2006 through 2010 or the year the individual turned 65, emigrated, died, or received old age or disability pension, whichever occured first. In each analysis, the reference group included all subjects without the exposure tested, i.e. included all other diagnosis groups and/or medications. Hazard ratios (HRs) with 95% confidence intervals (CI) were estimated by proportional hazards models. The analyses were adjusted for age group, sex, educational level, country of birth, and type of living area. The assumption of proportional hazards was tested and met using log/(−log) plots Each diagnosis group and prescribed drug class were analyzed separately throughout the study, not taking into account any co-occurrence of other diagnoses or drugs in the subjects.

Fourth, two-way interactions between age/sex and the exposure variable were introduced in all analyses to evaluate the fit of the models. These interactions were either deemed non-significant, or did only affect the results marginally, suggesting that all models fitted the data adequately. Furthermore, correction for multiple comparisons was performed using the Holm-Bonferroni method [28].

Main findings

To the best of our knowledge, this study is the first to examine the distribution of specific psychiatric diagnoses as well as medications among MS patients of working age, and their association with future disability pension. MS patients of working age had an elevated risk for being diagnosed with most types of psychiatric disorders, as well as being prescribed most types of psychiatric medications compared to non-MS individuals. Moreover, being diagnosed with a psychiatric disorder or having been prescribed psychiatric medication generally meant an elevated risk for future disability pension among MS patients.

Strengths and limitations

The strengths of this study include the use of data from several nationwide registers of high quality as well as the use of a large nationwide cohort. Missing data was relatively uncommon and there were no loss to follow-up. All Swedish MS patients of working age found in our registers were included in the cohort, which granted sufficient statistical power to analyze most psychiatric diagnoses and medications in detail.

There are also limitations that need to be addressed. First, the method of identifying MS patients means that patients who had not received any specialty care during the five-year interval studied were misclassified. Second, the number of patients with some of the more uncommon psychiatric diagnoses and/or medications was quite low, especially in the survival analysis, which led to wide 95% CIs. Third, pharmacological data only included recipes dispensed in pharmacies during July to December 2005, which means that medication not dispensed during this time was misclassified. Fourth, psychiatric diagnoses given in a primary care setting were not included. While it is likely that many MS patients receive a psychiatric diagnosis from their neurologist in specialty care, a higher rate of the non-MS individuals probably receive their psychiatric diagnoses from a general practitioner. This may lead to a differential misclassification and a skewed comparison regarding psychiatric diagnoses often treated in primary care such as mild to moderate depression and anxiety disorders. On the other hand, in the Swedish health care system, only the registration of a primary diagnosis is mandatory. It is possible that specialty care practitioners, e.g. neurologists, do not specifically register a second or third diagnosis during in- or outpatient care given to MS patients, although psychiatric symptoms may be present. With this in mind, the rate of psychiatric comorbidity could be higher among MS patients than these data reflect. This could be supported by the consistently higher rates in this study of psychiatric medications prescribed compared to their responding diagnoses. This also highlights the question as to whether psychiatric symptoms in MS should be considered comorbid rather than symptoms inherent to the disease itself, an important question both in an epidemiological [29] and clinical [30] context.

Comparison with previous studies

This is, to the best of our knowledge, the first study examining the association of specific psychiatric diagnoses and medications with disability pension among MS patients. It is also the first study comparing the use of psychiatric medications with that of non-MS individuals. Thus, comparisons with previous studies can be made mainly regarding the prevalence of psychiatric diagnoses. Regarding the method of identifying MS and psychiatric comorbidity using administrative data, Marrie et al recently published a validation study using a combined diagnosis and medication model [20]. In this study we chose to analyze diagnoses and medications separately, rather than constructing models of presumed clinical validity.

Depressive disorder was, as expected, the most common psychiatric diagnosis among both MS patients and non-MS individuals of working age. The 4.4% of MS patients who had been diagnosed with depression during the five years studied is a rather low proportion compared with previous studies [13]–[15]. However, most previous estimates of prevalence of psychiatric diagnoses are based on survey or clinical studies, which seem to lead to higher estimates. The exclusion of patients over 65 years of age is another possible contributing factor. The risk for bipolar disorder was nearly doubled in MS patients compared to non-MS individuals, but the prevalence of 0.5% was lower than in other studies [15], [20]. The risk for obtaining a psychotic disorder diagnosis tended to be elevated in MS patients with a 0.8% prevalence, in line with the only previous register study on this topic [22].

Implications of results

In this study, 35% of all MS patients received a psychiatric medication during the six months observed, which was a high number compared to non-MS individuals. Some of these drugs are also used to treat neurological symptoms, e.g. TCAs and duloxetine for neuropathic pain, SSRIs and TCAs for pseudobulbary affect, and benzodiazepines for spasticity. However, that the most commonly prescribed drug class in the study, SSRIs, was prescribed to 17% of MS patients suggests a high prevalence of depressive disorder and anxiety disorders. Of note is that previous studies have shown that depression is generally underdiagnosed and undertreated in MS patients [17], [31], which indicates that the prescription rates should perhaps be even higher if MS patients were to receive adequate antidepressant treatment.

In the survival analysis, psychiatric diagnoses and medications were generally associated with a substantially higher HR for disability pension. Depression was the diagnosis associated with the highest number of patients being granted disability pension. It is possible that this is due to a causative effect of depression, but could also be interpreted as a psychosocial reaction to an increased general disability eventually leading to disability pension. Coping mechanisms have been shown to mediate the relation between depression and disability [32], which suggests that psychosocial interventions could have an preemptive effect on the risk for disability pension in this group of patients, and suggesting depression as a target area for improved diagnosis and management in MS patients.

Personality disorders were the diagnoses with the highest HR for disability pension, which could be mediated by coping difficulties often present in personality disorders [33], but also by the fact that cognitive impairment is overrepresented in this group [34], The diagnosis group of organic and developmental disorders, which includes neuropsychiatric conditions considered to be secondary to an organic disease, also increased the HR for disability pension substantially. This group is heterogeneous, but its effect on the HR for disability pension probably reflects the disabling effects of conditions associated with latter stages of MS disease such as pseudobulbar affect and euphoria sclerotica.

Among the medications, antipsychotics were associated with the highest HR for disability pension, reflecting the serious nature of the conditions often treated with this class of drugs, namely bipolar disease and psychotic disorders. Additionally, the side effect profile of these medications, including various anticholinergic effects and extrapyramidal symptoms, could be especially unwanted in MS patients, and a causal contributing effect cannot be ruled out. MS patients had a risk three times that of the non-MS individuals for being prescribed benzodiazepines, which could, at least partially, be explained by treatment of spasticity, but most likely reflect high rates of anxiety disorders and sleeping problems. These rates must be considered high as modern guidelines do not recommend benzodiazepines at all as treatment for anxiety disorders due to concerns of tolerance and dependence [35], and the evidence for treatment of spasticity is at best uncertain [36]. Even more MS patients, 12%, had been prescribed short-acting sleeping agents. Although 40% of MS patients reportedly suffer from different kinds of insomnia [37], short-term sleeping agents are generally recommended for short-time use only [38], due to concerns of tolerance and dependence [39]. In general, any health problem for which antidepressants or sleep agents are prescribed, be it psychiatric disease, insomnia for non-psychiatric reasons, neuropathic pains or spasticity confer additional burden for patients with MS and may thus contribute to the risk of sick leave and disability pension. Possible contributing effects of the medications themselves on risk for disability pension should, however, not be ruled out.

An interesting finding is that only 0.1% of all MS patients in this study had been prescribed medication for alcohol abuse, less than in non-MS individuals. The one-month prevalence of alcohol dependence in MS patients has been reported to be 14%, and to be more common in working patients [40]. In this study, a diagnosis of a substance related disorder more than doubled the risk for disability pension, again suggesting an area for improved clinical detection and management.

Conclusion

In this population-based register study, MS patients of working age had an elevated risk for most psychiatric diagnoses and medications compared to non-MS individuals. Furthermore, MS patients having a psychiatric diagnosis or medication were at higher risk for future disability pension than MS patients who were not. These findings highlight the need for, and importance of, correct diagnosis and management of psychiatric comorbidity in MS patients. Suggestions for further research include investigations into what extent MS patients actually do receive accurate psychiatric diagnoses and optimal care, prescription patterns of psychiatric drugs in MS patients, and if intensified management of psychiatric comorbidity could decrease the risk for future disability pension and other adverse outcomes.