Results of meta-regression analysis of occurrence (%) of work absence in back pain populations (follow-up 6 months or longer)
| Number of studies | Pooled estimate (95% CI) | Meta-regression | ||
|---|---|---|---|---|
| p Value | Explained variance (%) | |||
| All studies | 17* | 15.5 (9.8 to 23.6) | ||
| Setting | 0.109 | 21.5 | ||
| Insurance database | 2 | 11.8 (3.3 to 34.0) | ||
| Population | 4 | 20.8 (12.1 to 33.6) | ||
| Healthcare | 7 | 7.9 (2.7 to 20.7) | ||
| Workplace | 4 | 35.0 (18.4 to 56.1) | ||
| Geographical area | 0.539 | 0.0 | ||
| North America | 14 | 14.4 (9.6 to 21.1) | ||
| Europe | − | − | ||
| Other | 3 | 21.6 (2.6 to 73.9) | ||
| Study period† | 0.888 | 0.0 | ||
| 1977–1990 | 2 | 14.6 (11.3 to 18.7) | ||
| 1991–2000 | 6 | 10.7 (5.5 to 20.0) | ||
| 2001–2007 | 7 | 14.0 (7.2 to 25.4) | ||
| Assessment | 0.203 | 4.5 | ||
| Electronic record | 5 | 25.2 (11.1 to 47.7) | ||
| Self-reported | 12 | 12.4 (6.8 to 21.5) | ||
| Participation bias | 0.278 | 1.1 | ||
| Low risk of bias | 7 | 10.7 (4.1 to 25.3) | ||
| Moderate risk of bias | 10 | 19.5 (11.2 to 31.8) | ||
| High risk of bias | − | − | ||
*Vingard et al26 reported data separately for men and women. Similarly, Murtezani et al43 reported data separately for white and blue collar jobs. Each data point is entered into the meta-analysis as a separate study.
†Two studies did not report study period.