Table 4.
Quality of the validation studies*
(A) | (B) | (C) | (D) | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Number of cases identified with study inclusion criteria | Number of cases chosen to undergo validation | Proportion of all cases identified with study inclusion criteria that were chosen to undergo validation | Response rate to GP questionnaire/record request | ||||||||||||||
Validation method | n validations carried out | Median | (Range) | n (%) validations where reported | Median | (Range) | n (%) validations where reported | Mean | (Range) | n (%) validations where reported | Median | (Range) | n (%) validations where reported | ||||
Internal | |||||||||||||||||
Manual review of computerized records or diagnostic algorithm | 31 | 1268 | (50–78 172) | 27 | (87) | 314 | (57–61 097) | 30 | (97) | 89.6 | (0.12–100) | 27 | (87) | n/a | n/a | ||
Sensitivity analysis | 23 | 4732 | (21–36 702) | 19 | (83) | n/a | n/a | n/a | n/a | ||||||||
External | |||||||||||||||||
Comparison of rates | 143 | 1984 | (5–200 000) | 70 | (49) | n/a | n/a | n/a | n/a | ||||||||
Additional information from GP | 160 | 226 | (1–51 688) | 139 | (87) | 100 | (1–3 010) | 153 | (96) | 68.2 | (0.15–100) | 138 | (86) | 90.4 | (55–100) | 134 | (83) |
Questionnaire to GP only | 43 | 1562 | (10–51 688) | 34 | (79) | 159 | (10–2 040) | 40 | (93) | 48.6 | (0.39–100) | 34 | (79) | 92.5 | (55–100) | 31 | (72) |
Questionnaire and record request to GP | 39 | 51.5 | (10–24 131) | 36 | (92) | 40 | (10–795) | 38 | (97) | 78.9 | (0.41–100) | 36 | (92) | 90.4 | (68–100) | 35 | (88) |
Record request to GP only | 78 | 199 | (1–22 195) | 69 | (88) | 88 | (1–3 010) | 75 | (96) | 72.3 | (0.15–100) | 68 | (87) | 90.0 | (56–100) | 68 | (88) |
Total | 357 | 468 | (1–200 000) | 255 | (71) | 104 | (1–61 097) | 183 | (96)* | 71.0 | (0.12–100) | 165 | (86) | n/a | n/a |
For example, in an external validation of glaucoma [29], 24 131 glaucoma patients were identified based on the study inclusion criteria (column A). Of these patients, 100 were chosen to undergo validation by questionnaire and record request (column B). This means that the proportion of cases chosen to undergo validation was just 0.41% of the total (column C). Of these 100 cases, the GP responded in 95, giving a response rate of 95% (column D).