Extended Data Fig. 6: Select pre- and post-matching covariate balances (based on normalized mean difference values) in the primary comparisons.

Comparisons of dementia outcomes for cohorts that experienced multiple vs a single HZ episode (a), that were exposed to ZVL vs PPSV23 (b), and that were exposed to RZV (2+ doses) vs PPSV23 (c). The values on the y-axis are the names of covariates. The dots indicate pre- and post-matching normalized mean difference (NMD) for the corresponding covariate. The red vertical lines indicate the bounds of what is considered a good balance (−0.1 < NMD < 0.1). Post-matching dots within these bounds indicate that the corresponding covariate’s mean is well balanced after matching. The covariates are sorted in descending order of their pre-matching NMD. Note for interpretation of covariate names: For covariates named ‘months_since_i’ (where i = 1, 2, 3, 4, 5), the name indicates months between i^th last intervention pre-exposure and the exposure date. For ‘months_since_vaccination’, the term ‘vaccination’ refers to any vaccination except the exposure vaccination. For ‘months_since_diagnosis’ covariate, the term ‘diagnosis’ refers to any diagnosis in the Diagnosis table with a certain status. For covariates named ‘intervention_counts_i’, the name indicates the number of interventions in the i^th pre-exposure year. The covariate named ‘well_visits_counts_all’ indicates the total count of routine medical and gynecological examinations. Multiple and single HZ, individuals experiencing at least 2 (multiple) and 1 (single) episodes of herpes zoster (HZ); PPSV23, recipients of at least 1 dose of a 23-valent pneumococcal polysaccharide vaccine; RZV (2+ doses), recipients of at least 2 doses of the recombinant zoster vaccine (Shingrix, GSK); tdap, receipt of tetanus-diphtheria-acellular pertussis vaccine; ZVL, recipients of at least 1 dose of the live-attenuated zoster vaccine (Zostavax, Merck).