Impact of isolation on hospitalised patients who are infectious: systematic review with meta-analysis

Purssell, Edward; Gould, Dinah; Chudleigh, Jane

No patient involved. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 F o r p e e r r e v i e w o n l y Results A total of 3 879 papers were retrieved from the three databases; of which 34 were assessed for eligibility by reading the full text. Of these 13 studies provided data suitable for the calculation of risk ratio, 5 giving psychological outcomes, [17] [18] [19] [20] [21] and 12 non-psychological; [19, [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] and 8 provided data for the calculation of standardised mean differences, 6 giving psychological outcomes, [21, 30, [33] [34] [35] [36] and 2 non-psychological. [29, 37] A further 6 studies did not provide raw data but are included in the results; 3 each giving psychological outcomes [38] [39] [40] and nonpsychological outcomes. [17, 41, 42] Meta-analyses were possible on two outcomes: anxiety and depression from 8 studies using standardised mean difference. [19] [20] [21] 30, [33] [34] [35] [36] Where only risk ratio data were given [20, 21] conversion to standardised mean difference was undertaken using the Campbell Collaboration calculator (https://campbellcollaboration.org/research-resources/effect-sizecalculator.html). [43] Where it was not possible to pool outcome data because of methodological and clinical heterogeneity, the data from studies are shown as forest plots but without meta-analysis. The forest plots contain results from the studies where sufficient data were given to calculate either the risk ratio or standardised mean difference. A number of studies provided data on those under contact precautions, but no comparative data and so were not included. [44] [45] [46] [47] Because of the large number of non-psychological outcomes for which RR could be calculated, it was decided that a change of 20% (i.e. a RR of 0.8 or less, or 1.2 or more) would be clinically significant, regardless of the statistical significance. This appear to be significant cause for concern regarding bias within the limitations inherent in these study designs. Most studies used established or validated tools [17] [18] [19] [20] [21] [23] [24] [25] 27, 29, 30, [33] [34] [35] [36] [37] or clinical outcomes. [22, 26, 28, 31, 32] The data from the comparative studies suggest that although in many cases infective isolation precautions make little difference to psychological outcomes, where it does make a difference this is primarily negative. There were significant declines in mean scores related to control and self-esteem, and in many studies increases in the mean scores for risk of anxiety and depression. However, these findings were not consistent, and some larger studies showed little or no difference between the groups for these outcomes. These are shown in Figures 1 and 2 respectively. Studies not reporting the raw data showed that contact precautions were associated with depression OR 1.4 (95% CI 1.2 to 1.5) but not anxiety OR 0.8 (95% CI 0.7 to 1.1) in a non-ICU population. [41] There was also an association with delirium OR For non-psychological outcomes, using a difference in the risk of +/-20% of an event as being a measure of clinical significance it appears there was a trend for less attention to be given to, and for more errors to occur in those who were isolated.