Skip to main content

Table 3 Power (%)* of selected clinical record review (CRR) scenarios to detect a reduction (R) in the real harm rate (rHR) over a twelve month period

From: Can we quantify harm in general practice records? An assessment of precision and power using computer simulation

Practices (n) Records reviewed (n)** Power (%)
rHR = 2 rHR = 5 rHR = 10 rHR = 20
   R = 50% R = 20% R = 50% R = 20% R = 50% R = 20% R = 50% R = 20%
1 40 0 0 0 0 0 0 0 0
50 0 0 0 0 0 0 0 0
100 0 0 0 0 0 0 1 1
200 0 0 0 0 2 1 9 5
300 0 0 0 0 3 2 19 7
400 0 0 1 1 11 4 27 7
10 40 0 0 1 0 8 4 30 8
50 0 0 2 1 16 5 37 8
100 1 0 14 5 38 7 64 14
200 9 4 36 8 66 13 93 22
300 21 5 57 10 84 16 98 32
400 28 8 65 12 91 22 100 41
20 40 0 0 10 4 28 7 53 11
50 0 0 17 6 37 8 66 15
100 10 4 37 10 63 13 93 24
200 28 7 67 13 91 22 100 38
300 43 9 81 19 98 31 100 54
400 56 11 92 20 100 40 100 69
50 40 9 3 34 8 65 13 93 24
50 14 5 45 10 75 16 96 28
100 36 10 74 16 96 25 100 48
200 64 12 97 27 100 48 100 79
300 84 $ 18 $ 100 41 100 64 100 92
400 94 21 100 48 100 79 100 97
100 40 31 6 64 11 92 20 100 40
50 35 10 76 14 96 29 100 53
100 66 12 97 26 100 47 100 77
200 90 22 100 48 100 76 100 96
300 98 34 100 64 100 92 100 100
400 100 38 100 78 100 97 100 100
150 40 45 10 84 16 99 33 100 56
50 51 11 89 21 100 38 100 64
100 82 15 100 39 100 64 100 91
200 99 32 100 64 100 91 100 100
300 100 46 100 81 100 99 100 100
400 100 57 100 92 100 100 100 100
200 40 53 12 91 23 100 38 100 68
50 64 12 96 28 100 47 100 77
100 92 22 100 49 100 78 100 97
200 100 36 100 79 100 96 100 100
300 100 56 100 92 100 100 100 100
400 100 67 100 96 100 100 100 100
250 40 64 14 98 25 100 49 100 77
50 76 13 99 36 100 58 100 85
100 97 30 100 55 100 86 100 99
200 100 50 100 87 100 99 100 100
300 100 67 100 95 100 100 100 100
400 100 79 100 99 100 100 100 100
300 40 74 15 99 31 100 55 100 84
50 82 18 99 37 100 64 100 90
100 98 29 100 65 100 91 100 100
200 100 54 100 91 100 100 100 100
300 100 73 100 99 100 100 100 100
  400 100 85 100 100 100 100 100 100
  1. *Power below the type I error rate of 5% is possible because analyses where the estimates failed to converge are counted as failure to detect change in the harm rate.
  2. **The total number of records reviewed during the twelve month period is shown. Each patient record was reviewed twice during this time.
  3. $Example: In this CRR scenario, 50 practices each reviewed 300 records (150 at the beginning and 150 at the end of twelve months) and had a baseline rHR of 2 incidents/100patients/year. Reductions of 50% and 20% over a twelve month period were detected with 84% (adequate) and 18% (inadequate) power respectively.
  4. Scenarios vary by numbers of practices reviewing records, number of records reviewed in each practice and rHR. Median rate ratios (MRR) between patients and practices are 2 and 1.2 respectively.
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us