Seventeen studies were included in the review (137,561 patients were reported to be included). Study duration ranged from six months to five years.
Average length of hospital stay was significantly shorter with hospitalist groups than with non-hospitalist groups (MD -0.44 days, 95% CI -0.68 to -0.20; 17 studies). For non-resident services, a similar result was shown in favour of hospitalist services (MD -0.69 days, 95% CI -0.93 to -0.46; eight studies) compared with non-resident non-hospitalist services. No further statistically significant differences were found for analyses that compared average length of hospital stay among hospitalist and non-hospitalist resident and non-resident services.
Statistically significant differences were found between all included studies (Ι²=97.9%); the funnel plot suggested that this heterogeneity was linked to three outlier studies. Statistical heterogeneity remained high (Ι²=85%) after removal of the three outlier studies. No evidence of publication bias was found.
Fifteen further studies not included in the meta-analysis (75,493 patients) had incomplete data; following average weighting by sample size, average length of hospital stay for hospitalist groups was -0.35 days shorter compared with non-hospitalist groups (62,122 patients).