eMethods. Supplemental Methods
eFigure. Study Design of the Self-controlled Analyses
Customize your JAMA Network experience by selecting one or more topics from the list below.
Lafaurie M, Lapeyre-Mestre M, Sailler L, Sommet A, Moulis G. Risk of Immune Thrombocytopenia After Influenza Vaccine. JAMA Intern Med. 2022;182(4):444–445. doi:10.1001/jamainternmed.2021.8523
Immune thrombocytopenia (ITP) has been reported after various vaccines, particularly influenza.1 However, to our knowledge only 2 case-control studies have been conducted to assess this association, leading to contradictory results.2,3 We aimed to assess this risk in a nationwide setting in France.
We conducted a population-based cohort study in France, selecting in the National Health Database the patients aged 65 years or older (a population for which influenza vaccination is recommended and totally reimbursed) who had an incident diagnosis of primary ITP between July 2009 and June 2018 (FAITH cohort).4 Patients and date of ITP onset were identified using a previously validated algorithm combining hospital discharge diagnoses, long-term disease diagnoses (recorded by general practitioners), and drug exposures (positive predictive value of 95.8%; 95% CI, 92.8%-98.8% for the identification of patients with ITP).5 Data on race and ethnicity were not collected or recorded in the database; in accordance with French law, collection of such information is prohibited. Data from January 1, 2009, through December 31, 2018, were included in the analyses. Exposure to influenza vaccine was identified using community pharmacy dispensing data. We used a self-controlled case series (SCCS) design resulting in the control of time-independent confounding factors (eMethods and eFigure, panel A, in the Supplement).6 This study was approved by the Institut des Données de Santé (Health Data Institute) and the Commission Nationale de l’Informatique et des Libertés (French Data Protection Agency). According to French law, no informed patient consent is required for studies using data from the National Health Database.
The incidence of ITP within risk periods (6 weeks following each vaccination) was compared with the incidence of ITP within other time periods, excluding the 2 weeks prior to vaccination to avoid selective survival bias (healthy vaccinee effect). Incidence rate ratios adjusted for seasonality and age were calculated using a Poisson regression model. We also performed a case-crossover analysis (eMethods and eFigure, panel B, in the Supplement) and a case-control analysis with up to 2 controls per case from the general population matched for year of birth, sex, and geographical area of residency (eMethods in the Supplement). Sensitivity analyses were conducted with longer periods of risks (8 and 12 weeks). All analyses were performed using SAS, version 9.4 (SAS Institute Inc).
We selected 4394 patients with incident primary ITP (88% identified with a hospital discharge diagnosis). Of these, 2333 (53.1%) were men, with a median age of 78 years (IQR, 71-84 years) (Table 1). In the SCCS, 3245 patients had received at least 1 dose of influenza vaccine, including 231 (7.1%) who were diagnosed as having ITP during one of the risk periods. The adjusted incident rate ratio was 0.91 (95% CI, 0.79-1.05) (Table 2). The odds ratio (OR) was 0.81 (95% CI, 0.61-1.08) in the case-crossover study and 0.85 (95% CI, 0.71-1.03) in the case-control study (4267 cases matched with 8337 controls). Sensitivity analyses resulted in similar results (Table 2).
This cohort study found results similar to those of the French PGRx case-control study, which included 198 incident cases of ITP and 878 controls between 2008 and 2011 and adjusted for age and sex (OR, 0.9; 95% CI, 0.4-2.1 in the 2 months prior to the index date).3 Conversely, the Berlin Case-Control Surveillance Study (FAKOS), which included 169 cases of ITP and 1031 controls between 2000 and 2009, found increased risk of ITP following influenza vaccination (OR, 3.8; 95% CI, 1.5-9.1).2 However, biases introduced by confounding variables (significant age difference between cases and controls) and memory (exposures recorded using a personal interview) cannot be ruled out.
This study has strengths and limitations. The main strength is its nationwide assessment with recent data, consisting of the largest cohort to date of ITP among patients aged 65 years or older with vaccine exposures. The SCCS is the reference design for vaccine safety assessment. Healthy vaccinee effect and seasonality were considered. Analyses performed with case-crossover and case-control designs as well as the sensitivity analyses gave consistent results, strengthening the findings.
The main limitation was related to the use of a health care database. Patients with ITP were identified by means of an algorithm that used primarily hospital diagnosis codes, which led to the identification of the most severe ITPs. Patients with mild and transient thrombocytopenia were not included. Moreover, despite good positive predictive value of the algorithm to identify ITP and to date ITP onset, we cannot exclude the possibility that some patients were misclassified. Also, the index date might not correspond to the true ITP onset, which is always difficult to ascertain, even in clinical practice. In addition, we defined exposure to influenza vaccine with dispensing data that did not necessarily correspond to the exact date of vaccination. However, sensitivity analyses with longer periods led to similar results.
Overall, the results of this nationwide population-based study, which included more than 4000 patients with ITP and used 3 different designs, found no increased risk of ITP following influenza vaccine.
Accepted for Publication: December 24, 2021.
Published Online: February 21, 2022. doi:10.1001/jamainternmed.2021.8523
Corresponding Author: Margaux Lafaurie, PharmD, Department of Clinical Pharmacology, Faculty of Medicine, Toulouse University Hospital, 37 allées Jules Guesde, 31000 Toulouse, France (email@example.com).
Author Contributions: Drs Lafaurie and Moulis had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: Lafaurie, Sommet, Moulis.
Drafting of the manuscript: Lafaurie, Sommet, Moulis.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Lafaurie.
Obtained funding: Moulis.
Administrative, technical, or material support: Sailler, Sommet, Moulis.
Supervision: Lapeyre-Mestre, Moulis.
Conflict of Interest Disclosures: Dr Moulis reported receiving grants from Amgen and Novartis to attend meetings; research grants from Amgen, CSL Behring, Grifols, Novartis, and Sanofi; personal fees from Amgen, Argenx, Grifols, Novartis, and Sobi; and serving on boards for Amgen, Argenx, Novartis, and Sobi outside the submitted work. No other disclosures were reported.
Trial Registration: ClinicalTrials.gov Identifier NCT03429660