Multiple Sclerosis Journal - October 2017 - 1507

JMA Wijnands, E Kingwell et al.
pan-provincial comprehensive data platform (http://
www.popdata.bc.ca). The linked data included
PharmaNet7 (drug prescriptions filled in the outpatient or community pharmacy setting), Medical
Service Plan Payment Information8 (physician visits),
Discharge Abstract Database9 (hospital admissions),
Census Geodata10,11 (aggregated data on neighborhood income, representing an estimate of socioeconomic status (SES)), Vital Statistics12 (death dates),
and the BC Ministry of Health's Registration and
Premium Billing files10 (indicating residency in the
province via registration status in the mandatory provincial health care plan). Combined, these linked
databases provided information on the dates and diagnostic codes (recorded as International Classification
of Disease (ICD)-9-CM or ICD-10-CA) of hospitalizations and physician visits, the dates and drug identification numbers (DIN) of prescriptions filled
(grouped according to the Anatomical Therapeutic
Chemical (ATC) Classification System), and demographics (sex, date of birth, and the first three digits of
the residential postal code). Linked data were available from 1986 to 2013.
The study was approved by the University of British
Columbia's Clinical Research Ethics Board (approval
H14-00448) and the provincial health agencies regulating the use of administrative data.
Study population
Individuals aged 18 years or older with MS were
identified based on a previously validated algorithm
requiring at least three MS-specific medical encounters defined as an ICD-9-CM/10-CA code related to
a hospitalization or a physician visit, or a DIN for a
MS immunomodulator (Supplementary Appendix
e-1, Table e-1a).13 The index date was defined as the
earliest of the first MS-specific or demyelinating
disease-related code (Supplementary Appendix e-1,
Table e-1a). To select incident MS cases, five full
years without any MS or demyelinating disease code
while resident in BC were required prior to the index
date. Since ICD codes from physician visits were
available starting from 1991, the earliest index date
was in 1996.
Every person with MS was matched with up to five
controls by sex, exact year of birth, and postal code at
the index year. Controls were randomly selected without replacement from the general population, and
were assigned the same index date as their matched
individual with MS. Individuals were only eligible to
be selected as a control if they had never had a

journals.sagepub.com/home/msj

diagnostic code for MS or a demyelinating disease,
never had a prescription filled for any of the
MS-specific immunomodulators, and were resident in
BC from at least 5 years before the index date until the
date of their matched case's third MS-specific diagnostic code. All individuals were followed until death,
emigration from BC, or study end (31 December
2013).
Study outcomes
The primary outcome was health care utilization specifically related to an infection. Health care utilization
was assessed separately for hospitalizations, physician visits, and prescriptions filled in each month over
the study period. Infection-related hospitalizations
and physician visits were identified using the primary
(most responsible) ICD-9/10 diagnostic code
(Supplementary Appendix e-1, Table e-1b), regardless of the other codes recorded.14 Prescriptions filled
for antimicrobials for systemic use (i.e. oral or parenteral) were identified by ATC coding groups
(Supplementary Appendix e-1, Table e-1c). Secondary
outcomes included hospitalizations and physician visits according to the five most commonly recorded
infections in each setting (Supplementary Appendix
e-1, Table e-1d).4,15-18 Prescriptions were similarly
grouped and analyzed separately for antibiotics, antivirals, or antimycotics.
Covariates
Baseline characteristics (measured at the index date)
included sex and SES. Age and comorbidities were
assessed longitudinally and updated on a monthly
basis. Comorbidities included diabetes mellitus,
ischemic heart disease, chronic lung disease, inflammatory bowel disease, rheumatoid arthritis, and psoriasis. These comorbidities have previously been
associated with MS19 and may increase the risk of
infection.20-24 Once a comorbidity was identified (via
one hospitalization or ⩾2 physician visits for the relevant diagnostic code;25-29 Supplementary Appendix
e-1, Table e-2a), an individual was considered affected
thereafter. Prescriptions for immunotherapies, not specifically indicated for MS, but known to increase
infection risk, were identified by the ATC groups L01
and L04A (Supplementary Appendix e-1, Table e-2b)
and assessed as present or absent within the 3 months
before the beginning of each month of follow-up.
Since MS-specific immunomodulatory drugs may lie
in the causal pathway between MS and infections,
these drugs were considered in a separate complementary analysis (see below).

1507
http://www.popdata.bc.ca http://www.popdata.bc.ca https://journals.sagepub.com/home/msj
Table of Contents for the Digital Edition of Multiple Sclerosis Journal - October 2017
