We conducted a population-based cohort study. We described the outbreak and response efforts. We used survival analysis to evaluate vaccine effectiveness.
The cohort included residents of Communities A and B. We excluded individuals who were in the communities transiently (e.g., healthcare workers on two-to-three week rotations). We created the cohort by extracting a population list from the communities’ immunization record database. The extracted lists were validated for completeness by community leadership. We extracted demographic and immunization information from the database.
We used the provincial confirmed mumps case definition and modified the provincial probable case definition to include individuals with clinically compatible signs and symptoms (i.e., parotitis) but with negative laboratory test results. This modification was made to improve sensitivity and in recognition of the challenge of mumps diagnostic testing in highly vaccinated populations (12). We included as cases all community residents meeting the confirmed and probable case definitions with symptom onset during the outbreak period. The outbreak period differed between the communities and was defined as the date of the first case’s symptom onset to the date of the last case’s symptom onset (Dec 18, 2017–Mar 30, 2018, for Community A; Jan 15, 2018—Jun 27, 2018, for Community B).
Pre-outbreak Immunization Status
For the complete cohort, we extracted dates of receipt for all mumps-containing vaccines received prior to the start of the outbreak, hereafter referred to as “pre-outbreak” doses. We excluded any invalid doses, defined as those received prior to the first birthday and within 28 days following another dose. We conducted chart reviews on a purposive selection of records to confirm the accuracy of the immunization database. In the survival analysis, to account for immune response, we excluded the 14 days following the receipt of a pre-outbreak dose from the at-risk period for individuals who received a pre-outbreak dose within 14 days of the start of the outbreak (Supplemental Figure 1).
Targeted Outbreak Response
As part of the outbreak response, an MMR vaccine dose was offered as a “targeted outbreak response” to individuals aged 8 to 48 years from both communities, regardless of their past immunization history, unless they were a confirmed case of mumps, were confident they had received an MMR vaccine within the last 28 days, or had medical contraindications. The lower end for the targeted age range was selected based on immunization schedule considerations. Due to changes in Ontario’s immunization schedule in 2011, children aged 8 older likely received their second MMR vaccine dose at 18 months while children under 8 years would have received their second dose at ages 4 to 6 years. (Figure 1). The upper age limit was selected to reflect the Canadian Immunization Guide’s recommendations to consider people born before 1970 as immune (13).
The targeted outbreak response supplemented opportunistic immunization at the time of healthcare encounters for other reasons. In addition, nurses systematically reviewed charts to identify and recall individuals who were not up-to-date with their immunizations. Mass immunization clinics were held in locations such as the community school, band office, and the Northern Store on March 8, 2018, in Community A and March 7–9, 2018, in Community B. Immunization was only one component of the outbreak response (Supplemental Table 1).
We defined an ’outbreak dose’ as any dose of mumps-containing vaccine received during the outbreak period. We excluded any doses received prior to an individual’s first birthday and doses received within 28 days following another dose.
In the survival analysis, to account for immune response, the post-outbreak-dose at-risk period began 14 days following receipt of an outbreak dose (Supplemental Figure 1). For individuals who went on to develop mumps, we excluded any “outbreak dose” received in the 14-day period prior to, or after, symptom onset. We conducted sensitivity analysis with shorter and longer periods (7 days and 28 days).
We described the outbreak and characteristics of the cases and non-cases prior to the outbreak (for list, see Table 1). We calculated attack rates per 1,000 person-days. Bivariable analyses were performed using Fisher’s exact or Wilcoxon rank sum test, as appropriate. We calculated unadjusted hazard ratios using Firth’s penalized-likelihood bias reduction method for Cox regression.
Outbreak Dose Evaluation
We included only individuals eligible to receive a valid ‘outbreak dose’ of MMR in the VE analyses (all individuals ≥1 year-of-age at the start of the outbreak).
We conducted univariable and multivariable survival analyses using Firth’s method for Cox regression. Confidence intervals were based on the profile penalized log likelihood. Follow-up time started on the first day of the outbreak period and ended on the last day of the outbreak period or on a case’s symptom onset date. We treated receipt of an outbreak dose as a time-varying covariate (Supplemental Figure 1).
In the multivariable survival analysis, our primary variable of interest was receipt of an outbreak dose. Our initial model was selected using subject-matter knowledge and included 6.8 events-per-degree-of-freedom including receipt of an outbreak dose, age, gender, and time since most recent receipt of pre-outbreak vaccine. All covariables were retained in the final model since each variable maintained the precision of the primary estimate.
We calculated VE for receipt of an outbreak dose as 1-hazard ratio. We conducted sensitivity analysis by excluding probable cases and by examining effects by age group and by pre-outbreak immunization status (see Supplemental Table 2 for list of models and associated estimates of VE).
We completed all analyses using Stata 13.1, SAS 9.3 or R 3.3.2. P-values <0.05 were considered significant.
The outbreak response and evaluation were undertaken with the approval and support of community leadership, including the Chiefs, Band Council Members and the Health Directors. The communities provided their data as part of the response efforts so at-risk individuals could be identified and offered vaccination and so the response could be evaluated. Community leadership supported the sharing of their experiences and the publication of their data. This evaluation was conducted as part of the outbreak response, was considered non-research and was not subject to an institutional review.