Table 3. Self-controlled Case Series Results From Analysis of GBS Risk Following RZV Administration.
| Analysis | Risk case | Control cases | Risk days | Control days | Rate ratio (95% CI) | Attributable riska | |||
|---|---|---|---|---|---|---|---|---|---|
| Per million doses | Per 100 000 person-years | ||||||||
| Unadjusted (95% CI)b | PPV-Adjusted (95% CI)c | Unadjusted (95% CI)b | PPV-Adjusted (95% CI)c | ||||||
| Original study period (October 1, 2017 to March 31, 2019) | |||||||||
| Primary analysis | |||||||||
| Claims-based (all doses) | 13 | 8 | 1092 | 2928 | 4.30 (1.76-10.53) | 6.47 (2.50-10.45) | 5.08 (1.04-9.11) | 1.98 (0.77-3.20) | 1.56 (0.32-2.79) |
| Secondary analysis | |||||||||
| Claims-based (first dose only) | 13 | 4 | 714 | 2090 | 9.30 (3.00-28.84)d | 12.20 (7.41-17.00)d | 9.50 (4.35-14.66)d | 4.10 (2.49-5.72)d | 3.20 (1.46-4.93)d |
| Medical record–confirmed analysis | |||||||||
| All doses (risk: 1-42; control: 43-183) | 7 | 4 | 546 | 1557 | 4.96 (1.43-17.27) | 5.17 (1.50-8.84) | 1.59 (0.46-2.71) | ||
| Extended study period (October 1, 2017 to February 29, 2020) | |||||||||
| Primary analysis | |||||||||
| Claims-based (all doses) | 24 | 20 | 2489 | 6157 | 2.84 (1.53-5.27) | 4.17 (1.65-6.69) | 3.13 (0.62-5.64) | 1.25 (0.49-2.00) | 0.93 (0.18-1.68) |
| Secondary analysis | |||||||||
| Claims-based | |||||||||
| First dose only | 21 | 8 | 1191 | 3675 | 7.72 (3.39-17.60)d | 8.65 (5.74-11.56)d | 6.48 (3.36-9.60)d | 3.06 (2.03-4.08)d | 2.29 (1.19-3.39)d |
| Second dose only | 3 | 12 | 626 | 1453 | 0.22 (0.04-1.22) | ||||
Abbreviations: GBS, Guillain-Barré syndrome; PPV, positive predictive value; RZV, recombinant zoster vaccine.
When calculating second-dose only attributable risks (ARs), we observed a high variance among the estimates from the bootstrapping simulations; the AR estimate for the individual who dies before the beginning of the control window can be infinity, resulting in drastically inflated CI estimates. We have suppressed second-dose only AR estimates for this reason.
Unadjusted AR estimates are derived using only the claims-based case classifications and do not take into account the outcome PPV or record return rate.
PPV-adjusted AR was calculated by imputing GBS status for all claims-identified cases and cases without medical records using the PPV as the PPV-based quantitative bias sensitivity analysis (eAppendix in Supplement 2).
P < .001.