All those included in the cohorts were SLE patients who were new to initiating immunosuppressive therapy with one of the three drugs included in the study. The population involved patients who had been hospitalized with serious infection.
You Might Also Like
Explore This IssueJune 2017
Previous controlled trials and academic studies have described increased rates of infections related to immunosuppressive drugs. But those studies were limited by such factors as small sample size, disease severity exclusions and short follow-up periods, the article authors note.
Further, most clinical trials are powered to determine differences in efficacy and not on adverse events, such as serious infection, Dr. Feldman says. Therefore, there may be an insufficient number of people in a clinical trial to detect clinically meaningful difference in infections.
Looking for Meaningful Difference
“A high percentage of lupus patients in this population had serious infections, which allowed us to look for what we thought would be meaningful differences in serious infections,” Dr. Feldman says.
Greater than 90% of the serious infections reviewed in the study were bacterial, which reflects findings from previous studies, notes Dr. Feldman. The remainder were viral or fungal, with the smallest percentage being mycobacterial.
The research team set out to choose a population with a similar chance of having received one drug or the other and a goal to learn whether there was a difference in serious infection between the two compared drugs. To extract a fair comparison, the study used matched cohorts who were most likely to have interchangeably received one or the other drug examined in the analysis, Dr. Feldman explains.
“In our analysis, we did propensity score matching, meaning we included only people whom we felt had a similar chance of receiving one or the other drug,” Dr. Feldman says.
Using 1:1 propensity score matching, researchers narrowed the number of patients for the first cohort to two groups of 1,350 each who were new to using either MMF or AZA. Likewise, the second cohort had score matching groups of 674 patients new to either MMF or CYC. Estimated propensity scores were based on sociodemographic, comorbidity and medication use information.
The study estimated incidence rates of serious infections up to six and 12 months after a patient began using the drug. Cox regression was used to estimate hazard ratios (HRs) of first infection and death, with 95% confidence intervals (95% CIs).
“We really did not find significant differences in serious infection risks,” defined as serious infection that results in hospitalization, Dr. Feldman says.