Table 4. Sensitivity analysis adjusting for variation in positive predictive value for HES-only admissions*.
| Positive predictive value, % | HES only | PHE study only | Both sources | Estimated total no. cases (95% CI) | % Completeness (95% CI) |
Estimated incidence | |
|---|---|---|---|---|---|---|---|
| HES† | PHE study | ||||||
| Encephalitis-specific code in any of the 20 diagnostic fields | |||||||
| 30 | 114 | 54 | 59 | 332 (272–391) | 52 (43–61) | 34 (27–41) | 3.42 |
| 40 | 152 | 54 | 59 | 405 (332–477) | 52 (43–61) | 28 (22–34) | 4.17 |
| 50 | 190 | 54 | 59 | 477 (392–562) | 52 (43–61) | 24 (18–29) | 4.91 |
| 54 | 206 | 54 | 59 | 508 (418–598) | 52 (43–61) | 22 (17–27) | 5.23 |
| 60 | 228 | 54 | 59 | 550 (452–648) | 52 (43–61) | 21 (16–25) | 5.67 |
| 70 | 266 | 54 | 59 | 623 (512–734) | 52 (43–61) | 18 (14–22) | 6.42 |
| 80 |
304 |
54 |
59 |
696 (572–819) |
52 (43–61) |
16 (12–20) |
7.17 |
| Encephalitis-specific code in primary field only | |||||||
| 30 | 71 | 60 | 53 | 265 (214–315) | 47 (38–56) | 43 (34–51) | 2.73 |
| 40 | 94 | 60 | 53 | 314 (254–374) | 47 (38–56) | 36 (28–44) | 3.23 |
| 50 | 117 | 60 | 53 | 363 (293–433) | 47 (38–56) | 31 (24–38) | 3.74 |
| 54 | 127 | 60 | 53 | 384 (310–458) | 47 (38–56) | 29 (23–36) | 3.96 |
| 60 | 141 | 60 | 53 | 414 (334–494) | 47 (38–56) | 27 (21–34) | 4.26 |
| 70 | 164 | 60 | 53 | 463 (374–553) | 47 (38–56) | 24 (19–30) | 4.77 |
| 80 | 188 | 60 | 53 | 514 (415–614) | 47 (38–56) | 22 (17–27) | 5.29 |
*Incidence = cases/100,000 population. HES, Hospital Episode Statistics; PHE, Public Health England †Due to the Lincoln-Petersen formula, if the number of HES-only admissions varies, the total HES admissions (HES only plus matched) is proportional to the estimated total number of cases; hence, the completeness of HES (ratio) does not change with varying positive predictive value.