Table 2. Meta-regression and Subgroup Analyses for the Frequency of Penicillin Allergy in Patients With Index Cefazolin Allergy.
| Variablea | No. of studies | No. of patients | Subgroup analyses | Univariable meta-regression, OR (95% CrI) | Probability of OR <1, % | |
|---|---|---|---|---|---|---|
| Severe reaction, % (95% CrI) | I2, % | |||||
| All | 15 | 146 | 3.7 (0-13.3) | 64.9 | NA | NA |
| Year of publication | 15 | 146 | NA | NA | 0.98 (0.95-1.01) | 85 |
| Geographic region | ||||||
| Europe | 9 | 63 | 3.7 (0-11.0) | 11.1 | 3.53 (<0.01-8.11)b | 86 |
| North America | 3 | 33 | NRc | NRc | NA | NA |
| Setting | ||||||
| Surgical patients | 8 | 104 | 4.4 (0-23.0) | 75.5 | NRc | NA |
| Referenced patients to an outpatient clinic | 7 | 42 | 6.4 (0.1-23.1) | 37.3 | NA | NA |
| Sampling method | ||||||
| Convenience | 8 | 64 | 10.0 (0-54.1) | 70.6 | NRc | NA |
| Consecutive | 5 | 77 | 1.1 (0-4.7) | 22.4 | NA | NA |
| Timing of index reactions | ||||||
| Immediate | 12 | 115 | 2.8 (0.8-4.9) | 3.4 | 0.04 (0.01-0.11)b | 99 |
| Immediate and nonimmediate | 3 | 31 | NRc | NRc | NA | NA |
| Clinical manifestations of index reactions | ||||||
| Anaphylaxis onlyd | 7 | 80 | 2.7 (0.8-3.9) | 0.9 | 3.46 (<0.01-10.50)b | 87 |
| Different reactions | 8 | 66 | 10.6 (0-48.8) | 66.4 | NA | NA |
| Assessment of penicillin allergy | ||||||
| Skin tests | 13 | 104 | 3.7 (0-16.6) | 72.1 | NRc | NA |
| Drug challenge or exposure | 9 | 103 | 2.5 (0.7-5.6) | 3.5 | 0.67 (0.01-1.46)b | 96 |
| Studies’ methodologic quality | ||||||
| Classification of most items | ||||||
| Low risk of bias | 11 | 109 | 4.6 (0.3-9.4) | 11.9 | 0.49 (0.09-1.40)e | 95 |
| High risk of bias | 4 | 37 | 15.2 (0-100) | 93.4 | NA | NA |
Abbreviations: CrI, credible interval; NA, not applicable; NR, not reported; OR, odds ratio.
a
For some variables, the number of studies and patients in the categories do not sum to 15 and 146, respectively, because there were studies belonging to other categories (with insufficient numbers of patients to be used in subgroup analyses or meta-regression) or with missing information.
b
Compared with all other categories of the respective variable.
c
Analysis not possible on account of the insufficient number of primary studies or participants.
d
All primary studies concerned surgical patients.
e
Meta-regression on the number of items classified as low risk of bias.