Participants and data collection
The methodology of this study has previously been described in some detail [11]. Briefly, using the online software SurveyMonkey®, we developed a webpage describing the study aims and inviting PwMS to take part. We posted a link to this webpage on websites, blogs, forums, and social media in which the principal investigator (GJ) was actively involved, including his own website (www.overcomingms.org; ‘OMS’). Online groups and pages used by PwMS with over 500 members or followers were targeted. Several follow up notices were posted over the 15 week recruitment period. MS societies were also asked to circulate details to their members.
After reading a participant information sheet, responders were advised that participation in the survey would constitute consent to the study. Inclusion criteria were adults of 18 years of age or more, who had been formally diagnosed with MS by a medical doctor and who could undertake an English language survey. We also sought contact details to enable planned longitudinal follow up. Data were stored in re-identifiable format, and access was restricted to members of the research team. Ethical approval was granted by St Vincent’s Hospital Melbourne Human Research Ethics Committee (LRR 055/12).
Data collected and tools used
Overall, the survey consisted of 163 questions, and took approximately 40 min to complete. Validated tools were used where possible, although for several variables, tools had not previously been validated. Socio-demographic data included age, gender, current location of residence, country of birth, cultural background, marital status, number of children, employment, education, height and weight. Disease-specific data included whether MS diagnosis had been confirmed by a medical doctor, year of diagnosis, first year of symptoms, diagnostic investigations undertaken, type of MS on diagnosis and currently. Participants were asked with a researcher-devised question how many physician-diagnosed relapses they had had in the previous 12 months and the last 5 years. Level of disability was assessed using the Patient-Determined Disease Steps (PDDS), a validated self-reported surrogate tool to the Expanded Disability Status Scale (EDSS) commonly used by neurologists to assess gait disability [12]. PDDS is scored ordinally from 0 (normal) to 8 (bed bound).
Health-related quality of life (HRQOL) was assessed with the validated and widely used Multiple Sclerosis Quality Of Life-54 (MSQOL-54), developed from the RAND 36-Item Health Survey (SF-36) and supplemented with 18 additional items. MSQOL-54 consists of 52 items in 12 scales, and two single items, producing two composite scores – the physical and mental health composites [13]. Whether participants deliberately exposed themselves to the sun to increase vitamin D production (yes/no) and frequency and dosage of vitamin D supplementation were recorded using researcher-developed questions, as there were no validated tools for these data. Vitamin D supplementation dose in International Units (IU) (0, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10,000, 20,000, 30,000, 40,000, 50,000, 100,000, 250,000, 500,000) and frequency (daily, every second day, weekly, monthly) gave rise to an average daily vitamin D supplementation dose which was collapsed into four categories for data analysis purposes (none, 1–2000 IU, 2001–5000 IU and >5000 IU). Latitude (degrees and minutes) was obtained from the city and country of residence.
Data analysis
Data were analysed using IBM SPSS Statistics 22.0 (IBM Corporation). Continuous data were summarised using mean (95 % CI) or median (interquartile range (IQR)) and categorical data using number (N) and percentage. Comparisons involving two continuous variables were undertaken using independent samples Kruskal-Wallis test as the variables latitude and relapse rate were not normally distributed, with separate Mann Whitney U tests for post-hoc analyses. For categorical data Pearson’s Chi Square was used, with assessment of adjusted standardised residuals to indicate under- or over-representation of groups. For all inferential tests, two-tailed tests of significance were used and the criterion for significance was set at .05.
Linear regression was used to predict the Physical Health and Mental Health composite scores and the Health Perception and Energy scale scores of the MSQOL-54 using the three variables of interest: latitude, deliberate sun exposure, and average daily vitamin D supplementation. First these predictor variables were entered separately to obtain crude odds ratios (and 95 % CI), and then together in a multiple linear regression model including the variables age, gender, disability, fish consumption, and physical activity to assess adjusted odds ratios (95 %CI). Relatively stable factors including gender, age, disability and fish consumption were included in the regression models as these can impact on the variables of interest. Physical activity was controlled for as many people exercise outdoors and it may therefore impact on sun exposure and vitamin D levels. Multinomial logistic regression was used to predict disability in a model including the variables gender, age, fish consumption and physical activity. Ordinal logistic regression was used to predict relapse rate in a model including the variables disability, gender, age, fish consumption and physical activity. Testing of assumptions for regression analysis were performed, including an examination of multi-collinearity to ensure that continuous independent variables were not closely correlated (having a bivariate correlation >0.70). All percentages reported were adjusted for missing data (due to item non-completion) on an item by item basis.