Incidence of severe infection in patients with rheumatoid arthritis taking biologic agents: a systematic review

Objective: The objective of this review was to estimate the population-based incidence and determine the types of severe infection and deaths experienced by patients with rheumatoid arthritis taking biologic agents. Introduction: Since the late 1990s, various biologic and synthetic drugs have been developed to treat rheumatoid arthritis. In recent years, the incidence of severe infection in patients with rheumatoid arthritis in Western nations has been determined by observational studies; however, no systematic review has been conducted on this topic. Inclusion criteria: The following inclusion criteria were considered: i) observational studies on patients with rheumatoid arthritis treated with biologic agents; ii) studies reporting the number of severe infections requiring hospitalization for treatment; iii) studies reporting person-years of observation data; and iv) studies based on rheumatoid arthritis registries, medical records from rheumatology centers, or insurance claim databases. Methods: PubMed, CINAHL, Embase, and Web of Science were searched to identify published studies. The reference lists of all studies selected for critical appraisal were screened for additional studies. Unpublished studies were searched on MedNar and OpenGrey databases. All the searches were updated on December 6, 2021. After removing the duplicates, 2 independent reviewers screened titles and abstracts against the inclusion criteria and then assessed full texts against the criteria. Two reviewers independently appraised the study and outcome levels for methodological quality using the critical appraisal instrument for cohort studies from JBI. Two reviewers extracted the relevant information related to severe infection and drugs. Results: Fifty-two studies from 21 countries reported severe infection rates associated with using 8 biologic agents, plus nonbiologic disease-modifying antirheumatic drugs. In total, 18,428 infections with 395,065 person-years of biologic drug exposure were included in the analysis. Thirty-five studies included infections in outpatients receiving intravenous antibiotic therapy. Fifteen studies reported the first episode of infection, and the remaining studies did not specify either the first or all of the episodes of infection. Inclusion of viral infection and/or opportunistic infection varied among studies. Fifteen studies reported the site of infection; respiratory, skin/soft tissue, urinary tract, and sepsis/bacteremia were commonly reported. Ten studies reported the case fatality rates, ranging from 2.5% to 22.2%. Meta-analysis was conducted for 8 biologic agents and conventional disease-modifying antirheumatic drugs. The infection rate varied from 0.9 to 18.1/100 person-years. The meta-analysis revealed an infection rate of 5.0/100 person-years (95% CI 3.8–6.7) among patients receiving tumor necrosis factor inhibitors (heterogeneity 98.2%). The meta-analysis for the other 3 biologic agents revealed a point estimate of 5.5 to 8.7/100 person-years with high heterogeneity. Sensitivity analysis indicated that registry-based studies were less likely to have very low or very high infection rates compared with other data sources. The definition of infection, the patient composition of the cohorts, and the type of databases appeared to be the primary sources of clinical and methodological heterogeneity. Conclusions: Due to high statistical heterogeneity, the meta-analysis was not suited to estimating a summary measure of the infection rate. Developing standardized data collection is necessary to compare infection rates across studies. Systematic review registration number: PROSPERO CRD42020175137 Supplemental Digital Content: A Japanese-language version of the abstract of this review is available as Supplemental Digital Content [http://links.lww.com/SRX/A8].