Study design/type
This study was approved by the Research Ethics Board at the University of Toronto. Written informed consent was obtained from all patients before sampling.
After enrolment, patients who required antibiotic treatment for infections were asked to provide the date of onset of their symptoms, as well as the date of first dose of antibiotic. To minimize recall bias, details regarding infection timing were collected at the first and all subsequent IVIg infusions. In addition, sputum cultures and chest X-rays were obtained if appropriate as dictated by standard of care.
Over the duration of the study, monthly IVIg infusions took place and the timing of each infection relative to the preceding infusion was collected. The number of infections was recorded, with emphasis on the onset and treatment of infection (i.e. day 1 to day 28 post infusion).
Subject inclusion criteria were: a diagnosis of CVID (based on the current ESID classification) [1, 2] or XLA; currently receiving or previously received IVIG at 4 week intervals, for at least 6 months; a stable clinical course; over the age of 18; willing to comply with all trial procedures, and sign an Informed Consent Form.
Subjects were excluded if: they had an anaphylactic or severe systemic response to Immune Globulin (Human); were a drug or alcohol abuser; were HIV PCR positive, Hepatitis C PCR positive, or hepatitis B DNA assay positive; had a concomitant disease, the course of which might reasonably be expected to change during the study period; or had other chronic conditions or treatments which, in the investigator’s opinion, may confound the interpretation of the results of the study questionnaires, such as chemotherapy or prophylactic antibiotics. The investigators felt that the inclusion of individuals on prophylactic antibiotics may confound results and lead to under-representation of infections, and in such a small sample size in a rare condition, it may skew results.
A history of past comorbidities and those present at enrollment was obtained through patients’ recollection as well as through clinical documentation.
Statistical analysis
Descriptive analyses were conducted to explore the patterns of the data at each month and then over the course of the study year. Descriptive statistics were calculated as mean, standard deviation, median, inter-quartile range for continuous variables and proportions for categorical variables.
To evaluate the primary endpoint, the mean number of days to infection among patients who had at least one infection between infusions, we performed a mixed model with random intercepts to account for multiple visits and multiple infections per patient and estimate the mean number of days after infusion.
To estimate infection rates, we categorized days to each new infections since infusion as Week 1 ≤ 7 days, Week 2 8–14 days, Week 3 15–21 days, Week 4 > 21 days. For each patient, we summed up all new infections within each category across all visits. We analyzed the data assuming the Poisson distribution for total repeated counts of new infections at weeks 1, 2, 3, and 4 post-infusion, accounting for within-patient correlations.
Secondary outcomes including the average number of infections per year, types of infections and the use of antibiotics to treated documented infections, were analyzed assuming the Poisson distribution for total counts.
SAS PROC GENMOD was used to fit a generalized estimating equations Poisson model assuming the exchangeable correlation structure and including an offset as the natural log of total years of follow-up. Rates/year within each week post-infusion and 95% confidence intervals were estimated. A time shifted analysis was not conducted in this study.
All analyses were conducted using SAS 9.4 (SAS Inc. Cary, NC). All tests were two-sided and statistical significance was defined if the p value was 0.05 or less.
PROC GENMOD in SAS will be used fit a Poisson model to the data. Patient baseline characteristics such as age, gender, time since diagnosis, and IgG trough levels may be incorporated in all models.