GRADE evidence profile
| Study | Quality assessment | Reported effect | Overall quality of evidence* | |||
| Risk of bias | Consistency | Directness | Precision | |||
| Outcome: Restraint orders per 100 patients over a 12-month period (specialised model of care vs standard care) | ||||||
| Grover and Lee29 | Not serious | NA | Direct | Precise | MD −0.9 orders (95% CI −1.64 to −0.16)† | ◉◉○○ Low |
| Outcome: Security man-hours per 100 patients over a 12-month period (specialised model of care vs standard care) | ||||||
| Grover and Lee29 | Not serious | NA | Direct | Precise | MD −36.8 hours (95% CI −48.3 to −25.3)† | ◉◉○○ Low |
| Outcome: Change in patient management or disposition over an 18-month period (screening laboratory tests vs no tests) | ||||||
| Donofrio et al 27 | Not serious | NA | Direct | Precise | RR 1.07 (95% CI 0.82 to 3.50) | ◉◉○○ Low |
| Outcome: Length of stay (screening laboratory tests vs no tests) | ||||||
| Donofrio et al 27 | Not serious | No inconsistency | Direct | Precise | 117 min (95% CI 109.7 to 124.4)† | ◉◉○○ Low |
| Santillanes et al 28 | Not serious | Direct | Precise | 120 min† | ||
| Outcome: Length of stay (specialised models of care vs standard care) | ||||||
| Grover and Lee29 | Not serious | No inconsistency | Direct | Precise | −45 min† | ◉◉○○ Low |
| Mahajan et al 30 | Not serious | Direct | Precise | −43.1 min (95% CI −63.1 to −23.1)† | ||
| Outcome: Monthly admissions to inpatient psychiatry over a 46-month period (specialised model of care vs standard care) | ||||||
| Parker et al 26 | Serious | NA | Direct | Imprecise | Second RRM versus RRM termination: MD −4.9†‡ | ◉○○○ Very low |
| Outcome: Hospitalisation over a 12-month period (specialised model of care vs standard care) | ||||||
| Greenfield et al 31 | Serious | NA | Direct | Precise | RR 0.56 (95% CI 0.46 to 0.69)† | ◉○○○ Very low |
| Outcome: ED return (specialised model of care vs standard care) | ||||||
| Greenfield et al 31 | Serious | NA | Direct | Precise | RR 0.61 (95% CI 0.32 to 1.17) | ◉○○○ Very low |
*As the evidence to which GRADE was applied was from studies that were not randomised controlled trials, the quality of evidence started at moderate/low and was downgraded to low/very low if there were concerns for any of the four domains.
†Statistically significant.
‡Other comparisons are not presented due to insufficient information reported in the study to determine whether results were statistically significant.
MD, mean difference; NA, not applicable; RR, relative risk; RRM, rapid response model.