Factors, Medications Associated With Depression Risk in MS
The methodology for the HOLISM Study has previously been documented in detail. In summary, participants were recruited via posting a link to the survey on Web 2.0 platforms that engaged people with MS. Participants completed a survey via SurveyMonkey® that examined their health and lifestyle behaviours, self-reported disability, disease activity, quality of life and depression, among other factors. Participants were eligible for the study if they were over 18 years of age and had been diagnosed with MS by a medical doctor. Participants were presented with an information page and selected "I agree" at the bottom of the page to provide consent before entering the survey. Ethics approval was granted by St Vincent's Hospital Melbourne Human Research Ethics Committee (LRR 055/12).
For this study the survey data included items assessing the following: socio-demographics; diagnostic history; level of disability; co-morbidities; fatigue; and depression; as well as a range of lifestyle and health behaviours.
Age, age at diagnosis, type of MS, gender, marital status, number of children, education level, employment status, weight and height were assessed and body mass index (BMI) was calculated by dividing weight (in kilograms) by height (in centimeters) and recoded into categories according to the World Health Organization (WHO). Those with BMI below 18.5 were categorized as underweight, those with BMI of 18.5 and up to 25 normal, BMI of 25 and up to 30 overweight and BMI of 30 and over obese.
The Patient Health Questionnaire (PHQ-9) is a screening tool for depression, which has previously been validated in people with MS. In our study we used the Patient Health Questionnaire depression module short version (PHQ-2). The PHQ-2 is scored in response to the question: 'Over the past 2 weeks, how often have you been bothered by any of the following problems: Little interest or pleasure in doing things; Feeling down, depressed, or hopeless' as follows: Not at all [0], Several days [1], More than half the days [2], Nearly every day [3]. The PHQ-2 is a shortened form of the PHQ-9 which has shown good criterion and construct validity. The PHQ-2 has a reported specificity of 92% and sensitivity of 83% for major depression with a score ≥3. This cut off was used in our analysis as a positive screen for depression risk. In addition, participants were asked to indicate whether they had a diagnosis of depression, whether they were receiving treatment for it, and whether depression limited their daily activities. The same questions were asked about an anxiety diagnosis. This question format was derived from the Self-Administered Comorbidity Questionnaire (SCQ) used to assess physical co-morbidities in the survey.
The Fatigue Severity Scale (FSS) consists of nine fatigue-related statements rated on a seven-point scale from 'disagree' to 'agree'. We used a mean score ≥4 to indicate clinically significant fatigue.
The Patient-Determined Disease Steps (PDDS) is a self-reported surrogate tool to the Expanded Disability Status Scale (EDSS). It is an ordinal scale from O (normal) to 8 (bed bound) and has good correlation with the EDSS. It is also used as a practical tool to assess changes in people with MS across time and has been used in numerous studies including the North American Research Committee on MS (NARCOMS) registry. Due to a low number of cases in the 'bed bound' category, participants scoring 7 ('wheelchair/scooter') and 8 were grouped together to enable meaningful analysis. Analyses included people with all types of MS.
We used the SCQ to assess co-morbidities in the absence of access to participants' medical records. The SCQ determines if the co-morbidity limits activities and whether treatment is received. The SCQ has previously been used in a study of people with MS. In our study, two arthritic co-morbidities were combined into one. All listed conditions were summed to determine an estimate of the number of co-morbidities each participant reported. Participants were then categorised as having: 'none', '1', '2', '3', '4', '5 or more', co-morbidities.
For dietary assessment we used the Diet Habits Questionnaire (DHQ). The DHQ is a 24 item tool that has previously been used in an Australian cardiac population and in the HOLISM study research on diet in MS. We removed four items assessing salt use and alcohol intake. The remaining 20 items were scored from 1–5, giving rise to a summary score with a possible range of 20–100, with higher scores indicating more healthy dietary habits. Participants were grouped into quartiles, based on their summary score.
We used a researcher-generated list of disease modifying drugs (DMDs) and common MS drugs. We asked for specific details including past and current use and duration of use. For the purpose of analysis, participants were categorised according to whether they currently took one of the interferons (one of the currently licensed interferon beta preparations: Avonex, Rebif, Betaferon or Betaseron), a disease modifying drug other than an interferon, or "other" (including those who took a non disease modifying drug or who took none). Interferon use was analysed individually given its previously reported association with higher rates of depression.
We explored participants' current vitamin D supplementation and calculated an average daily dosage. Participants were grouped as 'none', '< 5000 international units' (IU) or ≥5000 IU.
We assessed the type and average daily dose of omega-3 fatty acid supplementation used in the last 12 months. Types of omega-3 included flaxseed oil, fish oil, high strength fish oil, and 'other'. Participants were grouped according to whether they took omega-3 supplementation (yes/no) and the type ('flaxseed oil', 'fish oil and high strength fish oil', 'both' or 'none'). These comparisons were made based on findings from our previous work on omega-3s.
We used the International Physical Activity Questionnaire (IPAQ), which assesses the duration and frequency of vigorous and moderate physical activity, walking and sitting over the last 7 days. The IPAQ has been extensively validated, including in MS populations. Participants were categorised as low active, moderate active, or high active, as defined by the IPAQ scoring instructions.
We asked participants how often they meditated on an average weekly basis. Response options were categorised for the purpose of analysis as 'never', 'less than once per week', '1–4 times per week' or '5 or more times per week'.
We used the Single Item Measure of Social Support (SIMSS) to determine how many people provided support to participants. It has previously been used in several studies involving cancer patients. Response options were: none, 1 person, 2–5 people, 6–9 people, or 10 or more people. For the purpose of our study, the latter two categories were collapsed together.
IBM SPSS Statistics 22.0 was used to analyse the data. Univariate analyses were performed and continuous data reported using mean (95% CI) or median (IQR), and categorical data using number and percentage. The denominator for each item varied due to variation in the number of participants completing each item, as per tabulated results.
Bivariate analyses were used to explore the relationship between socio-demographic factors, disease-specific variables, and the depression outcome measure (PHQ-2) to determine factors that should be included as covariates in regression modeling.
The complexity of interactions between the sociodemographic, health-related and lifestyle variables make it difficult to determine which lifestyle variables are uniquely associated with depression. Multivariate analysis examined a range of modifiable risk factors ('exposure variables', i.e. diet, exercise, vitamin D) and their association with screening positive for depression. Medication use could be regarded as a 'modifiable risk factor' for depression and was therefore examined as an exposure variable.
Binary logistic regression was used to predict depressive risk (those scoring ≥3). For each modifiable risk factor individual models were derived adjusting for years since diagnosis, numbers of comorbidities, level of disability, clinically significant fatigue, age, gender and level of education. Type of MS was highly related with level of disability and we therefore chose to include only level of disability in the model. To maintain simplicity, we did not include employment in this model due to the number of categories within this variable. Because depression and fatigue have a complex relationship in people with MS, and fatigue may mediate depressive features, after controlling for fatigue as a potential confounder in multivariate analyses, we additionally compared models both with and without fatigue. There was negligible (up to approximately 10% in either direction) change to parameter estimates for all lifestyle variables when fatigue was removed from the model, and hence these results are not reported.
Methods
Participants and Recruitment
The methodology for the HOLISM Study has previously been documented in detail. In summary, participants were recruited via posting a link to the survey on Web 2.0 platforms that engaged people with MS. Participants completed a survey via SurveyMonkey® that examined their health and lifestyle behaviours, self-reported disability, disease activity, quality of life and depression, among other factors. Participants were eligible for the study if they were over 18 years of age and had been diagnosed with MS by a medical doctor. Participants were presented with an information page and selected "I agree" at the bottom of the page to provide consent before entering the survey. Ethics approval was granted by St Vincent's Hospital Melbourne Human Research Ethics Committee (LRR 055/12).
Data Collection and Tools Used
For this study the survey data included items assessing the following: socio-demographics; diagnostic history; level of disability; co-morbidities; fatigue; and depression; as well as a range of lifestyle and health behaviours.
Demographic Data
Age, age at diagnosis, type of MS, gender, marital status, number of children, education level, employment status, weight and height were assessed and body mass index (BMI) was calculated by dividing weight (in kilograms) by height (in centimeters) and recoded into categories according to the World Health Organization (WHO). Those with BMI below 18.5 were categorized as underweight, those with BMI of 18.5 and up to 25 normal, BMI of 25 and up to 30 overweight and BMI of 30 and over obese.
Depression
The Patient Health Questionnaire (PHQ-9) is a screening tool for depression, which has previously been validated in people with MS. In our study we used the Patient Health Questionnaire depression module short version (PHQ-2). The PHQ-2 is scored in response to the question: 'Over the past 2 weeks, how often have you been bothered by any of the following problems: Little interest or pleasure in doing things; Feeling down, depressed, or hopeless' as follows: Not at all [0], Several days [1], More than half the days [2], Nearly every day [3]. The PHQ-2 is a shortened form of the PHQ-9 which has shown good criterion and construct validity. The PHQ-2 has a reported specificity of 92% and sensitivity of 83% for major depression with a score ≥3. This cut off was used in our analysis as a positive screen for depression risk. In addition, participants were asked to indicate whether they had a diagnosis of depression, whether they were receiving treatment for it, and whether depression limited their daily activities. The same questions were asked about an anxiety diagnosis. This question format was derived from the Self-Administered Comorbidity Questionnaire (SCQ) used to assess physical co-morbidities in the survey.
Fatigue
The Fatigue Severity Scale (FSS) consists of nine fatigue-related statements rated on a seven-point scale from 'disagree' to 'agree'. We used a mean score ≥4 to indicate clinically significant fatigue.
Disability
The Patient-Determined Disease Steps (PDDS) is a self-reported surrogate tool to the Expanded Disability Status Scale (EDSS). It is an ordinal scale from O (normal) to 8 (bed bound) and has good correlation with the EDSS. It is also used as a practical tool to assess changes in people with MS across time and has been used in numerous studies including the North American Research Committee on MS (NARCOMS) registry. Due to a low number of cases in the 'bed bound' category, participants scoring 7 ('wheelchair/scooter') and 8 were grouped together to enable meaningful analysis. Analyses included people with all types of MS.
Co-morbidities
We used the SCQ to assess co-morbidities in the absence of access to participants' medical records. The SCQ determines if the co-morbidity limits activities and whether treatment is received. The SCQ has previously been used in a study of people with MS. In our study, two arthritic co-morbidities were combined into one. All listed conditions were summed to determine an estimate of the number of co-morbidities each participant reported. Participants were then categorised as having: 'none', '1', '2', '3', '4', '5 or more', co-morbidities.
Dietary Habits
For dietary assessment we used the Diet Habits Questionnaire (DHQ). The DHQ is a 24 item tool that has previously been used in an Australian cardiac population and in the HOLISM study research on diet in MS. We removed four items assessing salt use and alcohol intake. The remaining 20 items were scored from 1–5, giving rise to a summary score with a possible range of 20–100, with higher scores indicating more healthy dietary habits. Participants were grouped into quartiles, based on their summary score.
Medication use
We used a researcher-generated list of disease modifying drugs (DMDs) and common MS drugs. We asked for specific details including past and current use and duration of use. For the purpose of analysis, participants were categorised according to whether they currently took one of the interferons (one of the currently licensed interferon beta preparations: Avonex, Rebif, Betaferon or Betaseron), a disease modifying drug other than an interferon, or "other" (including those who took a non disease modifying drug or who took none). Interferon use was analysed individually given its previously reported association with higher rates of depression.
Vitamin D Supplementation
We explored participants' current vitamin D supplementation and calculated an average daily dosage. Participants were grouped as 'none', '< 5000 international units' (IU) or ≥5000 IU.
Omega-3 Fatty Acid Supplementation
We assessed the type and average daily dose of omega-3 fatty acid supplementation used in the last 12 months. Types of omega-3 included flaxseed oil, fish oil, high strength fish oil, and 'other'. Participants were grouped according to whether they took omega-3 supplementation (yes/no) and the type ('flaxseed oil', 'fish oil and high strength fish oil', 'both' or 'none'). These comparisons were made based on findings from our previous work on omega-3s.
Exercise
We used the International Physical Activity Questionnaire (IPAQ), which assesses the duration and frequency of vigorous and moderate physical activity, walking and sitting over the last 7 days. The IPAQ has been extensively validated, including in MS populations. Participants were categorised as low active, moderate active, or high active, as defined by the IPAQ scoring instructions.
Meditation
We asked participants how often they meditated on an average weekly basis. Response options were categorised for the purpose of analysis as 'never', 'less than once per week', '1–4 times per week' or '5 or more times per week'.
Social Support
We used the Single Item Measure of Social Support (SIMSS) to determine how many people provided support to participants. It has previously been used in several studies involving cancer patients. Response options were: none, 1 person, 2–5 people, 6–9 people, or 10 or more people. For the purpose of our study, the latter two categories were collapsed together.
Data Analysis
IBM SPSS Statistics 22.0 was used to analyse the data. Univariate analyses were performed and continuous data reported using mean (95% CI) or median (IQR), and categorical data using number and percentage. The denominator for each item varied due to variation in the number of participants completing each item, as per tabulated results.
Bivariate analyses were used to explore the relationship between socio-demographic factors, disease-specific variables, and the depression outcome measure (PHQ-2) to determine factors that should be included as covariates in regression modeling.
The complexity of interactions between the sociodemographic, health-related and lifestyle variables make it difficult to determine which lifestyle variables are uniquely associated with depression. Multivariate analysis examined a range of modifiable risk factors ('exposure variables', i.e. diet, exercise, vitamin D) and their association with screening positive for depression. Medication use could be regarded as a 'modifiable risk factor' for depression and was therefore examined as an exposure variable.
Binary logistic regression was used to predict depressive risk (those scoring ≥3). For each modifiable risk factor individual models were derived adjusting for years since diagnosis, numbers of comorbidities, level of disability, clinically significant fatigue, age, gender and level of education. Type of MS was highly related with level of disability and we therefore chose to include only level of disability in the model. To maintain simplicity, we did not include employment in this model due to the number of categories within this variable. Because depression and fatigue have a complex relationship in people with MS, and fatigue may mediate depressive features, after controlling for fatigue as a potential confounder in multivariate analyses, we additionally compared models both with and without fatigue. There was negligible (up to approximately 10% in either direction) change to parameter estimates for all lifestyle variables when fatigue was removed from the model, and hence these results are not reported.