Did hospital capacity affect mortality during the pandemic’s first wave?

Hospital capacity was a concern at the outset of the pandemic

‘The risk of health care system capacity being exceeded in the EU/EEA and the UK in the coming weeks is considered high.’
European Centre for Disease Prevention and Control, 12 March 2020

A pressing concern both at the outset of the pandemic and now, as case numbers climb again, was the ability of health systems to cope. The fear was that hospital admissions for COVID-19 would outstrip hospital capacity, resulting in deaths that would otherwise have been preventable.

This risk was apparent from early in the evolution of the pandemic. On 11 March 2020, with hospitals overwhelmed, the Italian College of Anaesthesia, Analgesia, Resuscitation and Intensive Care published guidelines for rationing care. Shortly afterwards, cases were rising across Europe, prompting warnings of health system saturation.

If hospital capacity has been an important determinant of mortality rates during the COVID-19 pandemic, this has implications for what is needed in the future. With many countries in Europe now experiencing a second spike in cases, it seems important to ask: to what extent did hospital capacity at the outset of the pandemic determine mortality rates?

UK hospital capacity is low by international standards

Health system capacity is a very broad concept. Here, we focus on measures that are easy to capture and for which there are available data, such as staff numbers and beds. Other elements of capacity, such as stocks of personal protective equipment (PPE), are important but there are less data recorded. Most of the data on capacity we look at comes from OECD statistics (for comparability, we only include countries with a GDP per capita of above $19,999) but we also provide an update of work comparing the number of critical care beds.

Based on the broadest measures, the UK had low capacity for treating patients at the outset of the pandemic compared to other OECD countries. There were fewer practising physicians, as well as one of the lowest rates of total beds in the OECD (Figure 1). Coming into the pandemic, the UK had high levels of bed occupancy and very little spare capacity.

Figure 1

These general measures may not wholly capture what matters for treating COVID-19 patients. For instance, some patients required critical care, for which critical care beds and specialist staff are needed. At the outset, the UK had relatively few critical care beds, but performed better on critical care staff (Figure 2) – we use the rate of doctors in the ‘surgical group’ (which includes intensive care medicine and anesthesiology) as a measure.

Figure 2

Hospital capacity and COVID-19 deaths

Although we have argued that excess death is a better measure of the full impact of COVID-19, here we focus on COVID-19 deaths because there are more data available. Under either metric, it is notable that there was significant variation in outcomes internationally over the first months of the pandemic: between 12 February and 30 June, COVID-19 deaths ranged from 84 per 100,000 in Belgium to 0.45 per 100,000 in New Zealand (Figure 3).

Figure 3

The most obvious way in which hospital capacity could have impacted COVID-19 mortality is if hospitals did not have resources to treat patients presenting with COVID-19 – for example, if there were an insufficient number of beds or staff. It is notable that the UK entered the pandemic with relatively few of either and has had a high death toll; so, is there any link between hospital capacity at the outset and deaths?

We performed a rough test for this by looking at the relationship between different measures of hospital capacity and COVID-19 deaths. We found that on average countries with higher capacity had fewer COVID-19 deaths. The association between deaths and capacity was strongest when looking at hospital beds, followed by surgical staff, and was weakest for doctors (Figure 4).

Figure 4

This would seem to support the idea that hospital capacity, by some measures, helped reduce mortality. However, correlation is not causation. To work out whether having more hospital capacity led to fewer deaths, we need to better understand the links between cause and effect.

Understanding the link between hospital capacity and mortality

If capacity determined mortality rates the way set out above, we would expect health systems in countries with low capacity to have been overwhelmed. However, countries acted to avert such a situation. Two common responses were to:

1. Establish additional capacity. The UK, like many countries, took steps to establish additional surge capacity, including additional beds, by quickly establishing the Nightingale hospitals; more staff, by calling on medical students or recently retired staff; and new equipment, eg ventilators were rapidly sourced. Resources were also redeployed, with staff shifting from other services to focus on COVID-19 patients.

2. Act to slow community transmission. Many countries imposed national lockdowns. Thereafter, the growth in new cases fell and with it the burden on health services. In the UK, protecting the health system was a prominent part of the justification for social distancing and a national lockdown, as reflected in the core government messaging: ‘stay at home, protect the NHS, save lives’.

As a result, countries did not operate at the limits of hospital capacity or did so only briefly. Evidence for this comes from data on hospital and ICU admissions, which show the same pattern: a short peak and then a rapid fall (Figure 5). Indeed, much of the surge capacity in the UK in the form of the Nightingale hospitals was unused during the first spike in cases.

Figure 5

Instead, given the speed at which the virus was spreading, steps to slow transmission – often involving initial national lockdowns – were crucial in determining the trajectory of cases and therefore COVID-19 deaths.

To illustrate this point, we use data from the stringency index developed by the Blavatnik School at the University of Oxford. This is a composite measure of the strictness of the government response on a range of measures (such as imposing travel restrictions, working from home etc). Comparing the stringency index at an early point in the evolution of the pandemic across different countries (we use 100 confirmed cases) helps explain much of the difference in death rates. For instance, countries with a stringency index above 20 at 100 confirmed cases generally had a lower death rate than those with a stringency index below 20 (Figure 6).

Figure 6

Returning to our associations with capacity, to understand how these interacted with other factors we split countries into three groups based on the level of capacity and compared the strictness of the lockdown, the rate of confirmed cases and COVID-19 deaths across these groups (Figure 7). Of the capacity measures, hospital beds had the strongest negative relationship with COVID-19 deaths. We find that countries with more beds imposed a stricter lockdown early in the pandemic, had fewer confirmed cases, and – crucially – had fewer deaths.

Figure 7

This suggests it was not so much capacity to treat patients that resulted in fewer deaths. Rather, somewhat counterintuitively, countries with greater bed capacity were quicker to act to slow transmission. It may be that both a higher number of beds and a stricter response reflect a more cautious or pro-active approach to managing population health generally.

Wider impacts of focusing on hospital capacity

We have discussed the prospect of hospitals being overwhelmed, but there are other, more subtle ways in which hospital capacity may have impacted outcomes. For instance, the initial response to COVID-19 focused on measures to free up capacity for COVID-19 patients and avoid hospitals being overwhelmed. This created a second channel by which mortality might have been impacted: the indirect effect of patients with other urgent health needs (or undiagnosed COVID-19) being unable or unwilling to access care.

This was most immediately visible in falls in emergency department attendances, which were seen in the UK, but also in countries such as Germany that have higher hospital capacity. It is notable that deaths at home have remained higher than the 5-year average in the UK, suggesting barriers to accessing hospital care (a reluctance on the part of the individual and/or services) remain.

In addition, the cancellation or postponement of planned treatments have led to backlogs of care, with concerns that delayed care (eg for cancer) will have impacts on morbidity and death in the future. It is possible that countries with more health system capacity were better able to continue treatment despite a spike in cases, and it is likely they will have more capacity to treat going forward.

Lastly, a focus on boosting hospital capacity may have been at the expense of other areas. As we have written elsewhere, in the UK protecting and strengthening social care services appears to have been given lower priority by national policymakers than protecting the NHS. At the peak of the pandemic, care homes in the UK recorded the biggest proportional increase in deaths (compared with deaths in hospital or at home). The UK is not alone in recording a spike in deaths in care homes, however, data show that the likelihood of a care home resident dying directly or indirectly as a result of COVID-19 was highest in Spain and the UK. Both countries had high overall death rates and low rates of hospital beds.

Conclusions

Hospital capacity was a concern entering the crisis. The COVID-19 case fatality rate early in the pandemic estimated at around 0.4–1%, and speed of transmission, created a strong imperative for governments to act to protect health systems. Indeed, the risk of cases exceeding hospital capacity has been an important part of the justification for imposing the first, and now a second, lockdown.

While the UK entered the pandemic with low hospital capacity (eg beds) compared with other countries, steps were taken to prevent hospital capacity being overwhelmed. The data presented here suggest the timing of interventions to slow transmission (eg to impose a lockdown) was important in determining the death rate from COVID-19. Our data are largely descriptive, but consistent with the argument that locking down a week or two earlier in the UK would have reduced the number of deaths – a suggestion reinforced by subsequent research.

The NHS did not run out of capacity during the first spike in COVID-19 cases, but there are other ways that low hospital capacity could have impacted death rates. Messaging early in the pandemic may have discouraged people from accessing health services; while a focus on hospital capacity seemed to take priority over other areas where action was needed, such as care homes. There are signs, as we enter a second lockdown, that policymakers in the UK are more mindful of the need to ensure health care remains available and of the risks in care homes.

Finally, it is important to acknowledge that a lockdown imposes other societal costs, both directly through increases in isolation and other mental health impacts, and indirectly via the overall economy, that should be considered when weighing up the overall response. Some of these effects will not be fully understood for years.