We defined capacity in terms of staff, beds and ventilators (herein referred to as resources). Data inputs and sources can be found in Additional file 2 [4, 9,10,11,12,13,14,15,16,17,18,19,20,21]. The analysis considered changes to resources across three different time points: the pre-pandemic phase, the surge phase and the post-surge phase (Fig. 1a, Additional file 1).
The pre-pandemic phase considered capacity before the onset of the COVID-19 pandemic in England (pre-March 2020). During this phase, we assumed baseline capacity, which is estimated as the average number of resources, and baseline patient occupancy, which is the number of these baseline resources occupied, to be constant.
The surge phase referred to the period of March–April 2020, during which there was a large increase in the numbers of hospitalised COVID-19 cases, and interventions to increase hospital capacity were implemented. Throughout this second phase, we considered the impact of interventions on the spare capacity of resources, which is a function of the capacity and patient occupancy, to determine whether patients could access treatment. For this, we developed a model to estimate the corresponding number of COVID-19 patients that would have been able to be accommodated on top of expected non-COVID-19 demand in the pre-pandemic phase. To determine the threshold numbers of COVID-19 patients at which capacity requirements would be exceeded with implemented interventions, we used the model to evaluate the impact of these, both individually and in combination, on top of the baseline capacity and patient occupancy.
Finally, the post-surge phase began in May 2020. At this point, the number of hospitalised COVID-19 cases has been observed to gradually decline, and hospitals have considered how to safely provide care again for all patients requiring it, whilst also planning for possible future surges in COVID-19 case numbers. In this part of the analysis, we used the model to determine how the re-introduction of elective surgery could be enabled by changes to the hospital provision interventions.
Throughout, spare capacity was defined by the difference between the total resources available and the capacity to accommodate a given demand, as determined by patient occupancy numbers (Fig. 1b; Additional file 3). If negative, this reflects a deficit in capacity.
Estimation of baseline capacity in pre-pandemic phase
The baseline capacity of overnight beds, nurses, junior doctors and senior doctors, split by general and acute (G&A) and critical care (CC), and ventilators, was estimated for England using National Health Service (NHS) data in the pre-pandemic phase [9,10,11, 13]. In England, hospital capacity and patient occupancy data are available by NHS trust level (Additional file 1). To account for seasonal fluctuations in capacity, adjusted with respect to seasonal fluctuations in expected demand, we assumed average daily numbers of beds and staff from April–June 2019. This period is most representative of what current capacity and occupancy would have been, without implementation of hospital provision interventions. CC bed numbers include beds in intensive care and high dependency units. We included G&A and CC beds and staff from all acute and community provider NHS trusts but excluded children’s trusts. CC paediatric beds and occupancy are distinguished from adult beds which was reflected in our estimates, but this distinction could not be made for G&A [9, 10]. However, the majority of hospitalised COVID-19 cases are adults and while some hospitals may have converted paediatric beds to treat adults, we do not anticipate this substantially altering the outcome of the analysis [22]. We further distinguished between senior and junior doctors to reflect the requirement of senior clinical decision-makers on wards. Staff numbers are considered in units of full-time equivalents (FTEs) to account for staff employed on a part-time basis or absent due to illness and the possibility of staff working in various wards. Electronic Staff Records (ESR) data were filtered for staff categories normally working on these wards. For example, midwives, general practitioners and paediatric staff were excluded. According to the number of beds in each trust, a weighted average of daily FTE was calculated for each staff category at a national level.
Staff-to-beds ratios specified by the Royal College of Nursing, the Royal College of Physicians and the Faculty of Intensive Care Medicine [16,17,18] were used to quantify required safe staffing levels per category. These were kept constant throughout the analysis. The baseline capacity of ventilators and other parameters in the model were derived from various sources (Additional file 2 [4, 9,10,11,12,13,14,15,16,17,18,19,20,21]).
Capacity during the surge phase
COVID-19 variables
The observed peak number of hospitalised patients with confirmed COVID-19 recorded (as of 31 May 2020) was set as the maximum number of COVID-19 patients in this analysis [4, 23]. This occurred on 12 April 2020, when approximately 3100 and 15,700 COVID-19 patients were occupying CC and G&A beds, respectively (Additional file 2 [4, 9,10,11,12,13,14,15,16,17,18,19,20,21]). We estimated the absence rate of staff due to COVID-19 during this period from surveys of union members for nurses and doctors [19]. These rates were coupled with baseline absence rates, to calculate the number of available staff during the surge.
Hospital provision interventions
Interventions implemented in England during the surge phase were previously identified [24] through a review of NHS sources, the European Observatory’s Health System Response Monitor [25] as well as the public press and were included in the model if they could be quantified at a national level.
The expected impact of each intervention across all resources was calculated as percentage changes of the baseline based on an analysis of NHS England data [26, 27] and from various sources [28,29,30] (Additional file 3). The expected proportion of occupied beds freed up through cancellation of elective surgery was estimated from Hospital Episode Statistics (HES) data of the busiest month in hospitals in January 2019 [27]. This is considered a conservative estimate because this month is the busiest in terms of demand for care. Elective patients requiring hospital care on any average day pre-COVID-19 (herein referred to as elective patients) were defined as those classified as non-emergency, non-maternity and non-cancer in the dataset and considered only if admitted to hospital overnight. They were also stratified into CC and G&A.
Analysis
For the surge phase, the model was used to calculate the spare capacity of resources under varying numbers of adult COVID-19 and non-COVID-19 patients on a given day, accounting for COVID-19-related staff absence, staff-to-bed ratios and the proportion of CC patients requiring ventilation (Fig. 1; Additional file 2 [4, 9,10,11,12,13,14,15,16,17,18,19,20,21]; Additional file 3). The maximum number of COVID-19 patients that could be accommodated by each resource under different scenarios, namely, no interventions, each individual intervention and the combination of hospital provision interventions that was implemented (herein referred to as the implemented intervention package), was determined. This was compared with the estimated maximum number of COVID-19 patients at the observed peak number of hospitalised COVID-19 patients during the first pandemic wave in England. The limiting resources in national baseline capacity were identified as the resources accommodating the smallest number of COVID-19 patients in the absence of interventions. We further compared the magnitude of spare capacity or deficits in different resources under the different scenarios of interventions for the observed peak number of hospitalised COVID-19 patients.
Reintroduction of elective patients in the post-surge phase
For the post-surge phase, we estimated the number of elective patients who could be accommodated under decreasing numbers of COVID-19 patients, for different intervention scenarios. This is referred to as post-surge reintroduction of elective surgery patients. This was facilitated by splitting non-COVID-19 patients into emergency patients, who continue to receive care throughout the pandemic, and elective patients (Fig. 1b). The number of patients that can be accommodated was determined by the number of patients for which all necessary resource categories displayed spare capacity (i.e. a non-negative value). Hospital provision interventions were assessed for their potential long-term feasibility based on official recommendations for the second phase of the NHS response to COVID-19 [4].
Both the number of COVID-19 patients and number of elective patients were varied, with the number of COVID-19 patients being reduced from the observed maximum in 10% intervals. This was done to consider scenarios of 0 to 100% of the maximum applied to both CC and G&A COVID-19 patients. We assumed that elective patients requiring G&A and CC will be introduced simultaneously. Using the previous analysis of HES and baseline occupancy data [9, 10, 27], we derived the expected number of elective patients that could be accommodated based on pre-pandemic demand and quantified a linear relationship between the number of elective patients in G&A and in CC (Additional file 3). Therefore, the daily number of G&A elective patients was varied in bands of 500, and the equivalent value for CC derived via this relationship.
All analysis was undertaken on R and is available publicly on Github.Footnote 1
Patient and public involvement
This research involved evaluating the impact of strategies already adopted by the NHS, and therefore, research questions, outcome measures and dissemination of study results were not developed or informed by patient or public involvement.