Results
In the pre-intervention group, 171 patient case-notes were reviewed with 205 in the post-intervention group. Average age was 76 (range 65-95) pre- and 77 (range 65-97) post-intervention (mean difference = -0.4, p=0.607, 95% confidence interval [CI] -1.4 to 1.8). The frequency of admission to intensive care remained stable at 20% following intervention (34/171 pre and 41/204 post-intervention, 95% CI -8.3% to 7.9%), and 30-day readmission rates remained unchanged (22/171 versus 24/204, p=0.746, 95% CI -5.6% to 7.8%). Mortality showed a trend reduction of 1.4% (9/171 to 8/204, p=0.534, 95% CI -2.9% to 5.6%). See Table 1 for further patient demographic details.
Following implementation of our medical liaison service, overall mean LOS showed a trend reduction from 10.75 to 7.95 days (p=0.635, CI 0 to 1 day). When comparing the impact of our intervention on LOS for patients admitted either acutely or electively, there was no significant reduction seen (p=0.103, 95% CI 0 to 5 days and p=0.890, 95% CI -1 to 0 days respectively).
However, a Kaplan-Meier analysis (Figure 1) and a comparison of means showed a significant reduction from 25.12 to 17.28 days in mean LOS for patients admitted for more than seven days (p=0.025, 95% CI for mean difference 1.5 to 14 days). This time period was considered intuitive given the mix of durations of length of stay for these patient groups. Post hoc analyses using other potential thresholds (e.g. longer than 10-days) did not maximise the difference in length of stay between the two groups. Moreover, on further enquiry, demographic data for those with a prolonged stay does not differ between the two cohorts. Specifically, there are no differences in patient sex (p=0.774), age (p=0.923), type of surgery (elective/acute) (p=0.710) or rates of intensive care admission (p=0.696). Lastly, for patients admitted for longer than seven days, 30-day readmission rates reduced from 12/50 (20.0%) to 8/72 (11.1%), (p=0.156, 95% CI -3% to 21%).
The total number of postoperative complications suffered per person demonstrated a trend reduction following service implementation (1.09 to 0.86 per person, p=0.181, CI -0.11 to 0.56). Complications suffered by patients admitted acutely were more frequent compared to those admitted electively (1.35 versus 0.71 per person, p<0.001, mean difference = 0.64, 95% CI 0.29 to 0.98). When evaluating complication frequency following service implementation in the elective patients, there was no significant reduction (0.63 to 0.79 per person, p=0.373, mean difference -0.16, 95% CI -0.60 to 0.19). However, when evaluating complication frequency in the acute patients following service implementation, there was a significant reduction (1.81 to 0.97 per person, p=0.01, mean difference = 0.84, 95% CI 0.21 to 1.46).
Utilising correlation analysis, age did not significantly correlate with LOS either pre-intervention (r= -0.11, p=0.171) nor post-intervention (r= -0.127, p=0.064). Likewise, age did not significantly correlate with frequency of complications either pre-intervention (r= -0.069, p=0.405) nor post-intervention (r= -0.120, p=0.122).