Descriptive statistics over the study period
Figures 2(a) and 2(b) contain, respectively, the timeseries of the attendances at UCLH ED and the percentage of people that left within four hours of arrival in the period between April 1, 2017 and October 14, 2020. We observe that there was almost constant level of attendance at this ED in the period before March 2020 with a notable drop in attendances that coincides with the onset of the first national lockdown (23/03/2020).
Over the entire study period, a total of 458,456 patients attended UCLH ED of whom 92,362 (20.1%) arrived by ambulance and 366,094 (79.8%) patients walked-in. Daily, on average 355 patients attended UCLH ED, with 72 patients arriving by ambulance and an average of 283 patients walking-in. The average time spend at UCLH ED was 3.5 hours (95%CI=[3.3,3.6]), with times spend by patients coming by ambulance spending on average 4.6 hours (95%CI=[4.3,4.9]), while patients walking-in spending, on average, 3.2 hours (95%CI=[3.0,3.4]) at UCLH ED. Finally, the average proportion of people leaving UCLH ED within 4 hours of arrival was 84.8% (95%CI=[84.4%,85.2%]).
Impact of trialed interventions in the pre-COVID-19 era
The results of the first analysis showing the impact of the overcrowding interventions A-C on the overcrowding metrics are contained in Table 2.
Table 2
Outcomes from the ITS analysis projecting the impact of interventions A, B and C on the overcrowding indicators. The study period is truncated into periods before and after the start of each intervention A, B and C; respectively starting on February 01, 2018, October 01, 2017 and April 01, 2019. We calculate the changes over a period of six months before and after the start of each intervention.
Changes in number of people attending UCLH ED |
Intervention | IRR (95% CI) | p-value |
interventions A | 1.058 (1.041,1.075) | < 0.001 |
Interventions B | 1.053 (1.034,1.071) | < 0.001 |
Interventions C | 1.074 (1.057,1.091) | < 0.001 |
Changes in % of people leaving within 4 hours of arrival |
Interventions A | 0.935 (0.927,0.943) | < 0.001 |
Interventions B | 0.926 (0.919,0.934) | < 0.001 |
Interventions C | 0.938 (0.929,0.947) | < 0.001 |
Changes in average waiting time |
Interventions A | 0.834 (0.734,0.949) | 0.005 |
Interventions B | 0.913 (0.763,1.092) | 0.322 |
Interventions C | 0.832 (0.773,0.896) | < 0.001 |
The number of people attending UCLH ED increased slightly, albeit statistically significant, in the period after implementation of interventions A, B and C (Table 2, rows 3–5). The increase in numbers of patients attending UCLH ED after implementation of interventions A (row 3 of Table 2) was 5.8%, for interventions B was 5.3% (row 4 of Table 2) and for interventions C was 7.4% (row 5 of Table 2).
There was a small but statistically significant decrease in number of people leaving within four hours of arrival (rows 7–9 of Table 2). For interventions A this reduction was 6.6% (p < 0.001), from 88.5–82%. For interventions B this reduction was 7.4% (p < 0.001) from 90.4–83% of people leaving within 4 hours of arrival (Table 2). Finally, for interventions C this reduction was 6.2% (p < 0.001), from 86.2–80% of people leaving within four hours of arrival.
The average waiting time spend by patients at UCLH ED was reduced by 38 minutes (17%) after implementation of interventions A (p = 0.005) and C (p < 0.001) (rows 11 and 13 of Table 2 respectively). With implementation of intervention B, the average waiting time was reduced by 20 minutes (9%) but this was not statistically significant (p = 0.322 in row 12 of Table 2).
Impact of COVID-19 on overcrowding
The results of the second and third analysis showing the impact of the epidemic and the lockdown on overcrowding metrics in this ED are shown in Table 3.
Table 3
Outcomes from the ITS analysis projecting the impact of the first national lockdown (‘lockdown’) to suppress COVID-19 during the spring of 2020 on the overcrowding metrics and considering all attendances and those arriving by ambulance or walking in. The study period (over the period January 12, 2020-August 11, 2020) is split into three time periods of 71 days defined as before the first national lockdown (January 12, 2020-March 22, 2020) (‘before’), during the first national lockdown (March 23, 2020-May 31, 2020) (‘lockdown’) and after the first national lockdown (June 01, 2020-August 11, 2020)(‘after’).
Impact of COVID-19 lockdown on all people attending UCLH ED |
Intervention | IRR (95% CI) | p-value |
Lockdown VS before lockdown | 0.346 (0.324,0.369) | < 0.001 |
After lockdown VS lockdown | 1.496 (1.417,1.576) | < 0.001 |
Impact of COVID-19 lockdown people arriving by ambulance to UCLH ED |
Lockdown VS before lockdown | 0.473 (0.443,0.505) | < 0.001 |
After lockdown VS lockdown | 1.285 (1.211,1.362) | < 0.001 |
Impact of COVID-19 lockdown people walking-in the UCLH ED |
Lockdown VS before lockdown | 0.308 (0.287,0.331) | < 0.001 |
After lockdown VS lockdown | 1.596 (1.506,1.692) | < 0.001 |
Impact of COVID-19 lockdown in % of people leaving within 4 hours of arrival |
Lockdown VS before lockdown | 1.059 (1.034,1.085) | < 0.001 |
After lockdown VS lockdown | 1.075 (1.053,1.098) | < 0.001 |
Impact of COVID-19 lockdown on average waiting time |
Lockdown VS before lockdown | 0.876 (0.839,0.915) | < 0.001 |
After lockdown VS lockdown | 0.868 (0.837,0.899) | < 0.001 |
Our second analysis suggested that in the presence of the COVID-19 epidemic, overcrowding in UCLH ED was reduced. During 2020, there was a notable drop in the number of attendances at UCLH ED that coincides with the onset of the first national lockdown (Fig. 2(a) and Table 3). While drop in attendances was evident in both people walking in UCLH ED and people arriving with ambulances, the decrease in those walking in was larger (Fig. 2(a) and Table 3)
Specifically, during the period January 01, 2020-October 14, 2020, a total of 69,491 patients attended UCLH ED of which 14,837 (21.4%) arrived by ambulance and 54,654 (78.63%) patients walked-in. This was a 39.3% reduction in visits to this ED in comparison to the same period in 2019. In addition, comparing these periods in 2019 and 2020, we determined that there was 8% reduction in waiting time and 6% increase in number of people leaving within four hours of arrival (Table 3).
Considering only the period of the first national lockdown (March 23-June 01) in 2019 and 2020, these changes were even more dramatic: 65.3% reduction in attendance and 21.9% reduction in waiting time and 8.3% increase in number of people leaving within four hours of arrival (Table 3).
For the third analysis, considering 2020 only, our results suggest that the lockdown significantly affected the three overcrowding metrics. In comparison to the period before the lockdown, the lockdown led to 65.4% reduction in attendance (52.7% in those arriving by ambulance and 69.2% in those walking in), 12.4% reduction in waiting time and 5.9% increase in number of people leaving within four hours of arrival; all statistically significant (Table 3).
Once the strict social distancing measures were lifted, there was an 49.6% increase in attendances to UCLH ED (28.5% in those arriving by ambulance and 59.6% in those walking in) (Table 3). But interestingly, compared to during lockdown, in the period afterwards, the waiting time was reduced by 13.2%, while the number of people leaving within four hours of arrival was increased by 7.5%; all changes statistically significant (Table 3).