Skip to main content
. Author manuscript; available in PMC: 2020 Apr 24.
Published in final edited form as: Pharmacoepidemiol Drug Saf. 2014 Apr 3;23(8):839–848. doi: 10.1002/pds.3618

Table 3.

Descriptive statistics of the length of surveillance in months for two study designs over a range of true risks with misclassification

True risks Signala (%) Mean (months) Median (months) 80th (months) 95th (months)
IRDb IRR Poi Bin Poi Bin Poi Bin Poi Bin Poi Bin
Reference casec: sensitivity = 1.0; PPV0 = 1.0; database size, 564 000 persons
0 1.0 5.0 5.3 16.8 73.3 17 74 17 87 17 102
1 1.5 15.2 15.8 16.4 60.8 17 63 17 75 17 88
2 2.0 29.3 31.0 15.8 50.1 17 53 17 65 17 76
4 3.0 58.6 60.9 14.0 34.6 15 36 17 49 17 59
8 5.0 90.9 89.9 10.9 20.3 10 18 14 29 17 39
18 10.0 99.9 99.5 7.2 11.8 7 11 9 15 11 19
Algorithm 1d: sensitivity = 0.84; PPV0 = 0.54; database size, 564 000 persons
0 1.0 5.0 4.0 28.3 89.1 29 90 29 102 29 114
1 1.5 12.3 11.4 27.4 79.8 29 82 29 94 29 105
2 2.0 23.4 23.9 26.0 69.9 29 74 29 86 29 96
4 3.0 51.5 54.9 22.0 52.2 28 57 29 71 29 82
8 5.0 90.0 91.0 14.7 29.6 12 26 23 44 29 58
18 10.0 >99.9 99.9 7.8 14.8 7 14 9 19 14 25
Algorithm 2e: sensitivity = 0.99; PPV0 = 0.53; database size, 0.40 × 564 000 persons
0 1.0 5.0 4.0 57.4 178.8 59 180 59 205 59 231
1 1.5 12.2 11.2 55.3 158.7 59 164 59 188 59 211
2 2.0 23.3 23.4 52.1 138.2 59 148 59 171 59 193
4 3.0 51.3 53.8 43.7 101.1 58 112 59 142 59 165
8 5.0 90.0 90.4 27.6 54.6 22 47 44 85 59 114
18 10.0 >99.9 99.9 13.1 23.5 12 21 16 31 26 46
Algorithm 3f: sensitivity = 0.59; PPV0 = 0.79; database size, 0.40 × 564 000 persons
0 1.0 5.0 5.5 70.5 280.5 72 285 72 337 72 388
1 1.5 13.8 14.7 67.9 237.6 72 248 72 296 72 343
2 2.0 26.4 28.5 64.1 195.0 72 211 72 258 72 302
4 3.0 55.0 58.4 53.9 133.3 60 144 72 199 72 240
8 5.0 90.0 90.1 36.6 68.6 32 59 54 110 72 155
18 10.0 >99.9 99.7 19.7 28.5 18 24 26 41 35 62

Bin, Binomial MaxSPRT Model; IRD, incidence rate difference; IRR, incidence rate ratio; Poi, Poisson MaxSPRT Model; PPV0, positive predictive value in the comparison population.

a

Signal (%) is the percent of time the null hypothesis is rejected.

b

Incidence rate difference is given in events per 100 000 person-years.

c

Poisson MaxSPRT upper limit = 1.371, Poisson MaxSPRT critical value = 1.663; binomial MaxSPRT upper limit = 20, binomial MaxSPRT critical value = 3.296.

d

Poisson MaxSPRT upper limit = 4.106, Poisson MaxSPRT critical value = 2.507; binomial MaxSPRT upper limit = 39, binomial MaxSPRT critical value = 3.296.

e

Poisson MaxSPRT upper limit = 4.226, Poisson MaxSPRT critical value = 2.522; binomial MaxSPRT upper limit = 39, binomial MaxSPRT critical value = 3.296.

f

Poisson MaxSPRT upper limit = 2.076, Poisson MaxSPRT critical value = 2.032; binomial MaxSPRT upper limit = 25, binomial MaxSPRT critical value = 3.296.