Should emergency departments move patients to other wards even when there's no bed space available?

Emergency departments across the country are facing unprecedented problems with overcrowding. The Royal College of Emergency Medicine recently reported that in the first four months of this year, 79,610 patients waited more than 12 hours for an inpatient bed from the time that a decision to admit them was made. For context, this is nearly as many as had previously cumulatively waited more than 12 hours in the 11 years since records began.

These long waits for inpatient beds are undoubtedly detrimental to patients. One recent study found that standardised mortality at 30 days began to rise once patients had waited more than five hours in the emergency department from time of arrival, with an increase in absolute mortality of around 2% (8.2% to 10.1%) at 30 days for those who waited more than 6–8 hours. Given the current number of patients experiencing very long waits, this is estimated to translate into 500 patients dying each week due to A&E overcrowding.

One solution that has gained traction is a ‘continuous flow model’ first introduced in North America. These effectively mandate that a set number of patients should be moved each hour from the emergency department to inpatient wards, regardless of whether a bed is available or the time of day. This means that patients might need to be ‘doubled up’, with an extra patient in a bay and two patients in a side room. In turn, this encourages wards to discharge more patients and allows ambulances to offload patients in the space created in the emergency department. Thus, the whole system is decongested.

Several hospitals across England are currently trialling local versions of the continuous flow model and the word is that the results are highly encouraging. Patients are moved out of the emergency department faster and ambulances are waiting less time to offload – with no increase in the number of safety incidents recorded.

But continuous flow models make me nervous for a number of reasons.

What contributes to patient harm?

The current rhetoric acknowledges that long waits in the emergency department are a by-product of high occupancy elsewhere in the hospital, but it assumes that the emergency department overcrowding is the main, if not sole, cause for the increases seen in mortality.

The literature strongly suggests that overcrowding in other wards ‘downstream’ also carries risk. A study in Australian hospitals found that while overcrowding in emergency departments increased mortality at two, seven and 30 days, an overcrowded hospital and an empty emergency department was also associated with increased mortality. Mortality was highest when both the hospital as a whole and the emergency department were overcrowded. A large American study found that hospitals which ‘speed up’ processes and decrease their length of stay in response to emergency department overcrowding increase their 30-day mortality by 3.8%, which translates to 2.3 additional deaths per 1,000 patients per year.

While protocols for continuous flow models insist that the ‘right’ patients go to the ‘right’ wards, this is not possible in circumstances of hospital overcrowding. ‘Boarding’ patients has been shown to nearly double the mortality on the wards and within 30 days of discharge not only of patients directly experiencing boarding (from 2% to 4.2%), but also for all patients on wards where patients are boarded (from 2% to 3.7%). The practice was also found to increase length of stay for all patients on wards with boarders and to increase readmissions for boarders themselves.

The appeal of the continuous flow protocol is that it ‘makes visible’ the problems within the emergency department to the whole system. In doing so, it shares the risk across the wider hospital. I have looked at just three studies. But they do indicate that the increase in mortality cannot be attributed to long waits in the emergency department alone. One of those studies suggests that, while the risk might be significant, only those patients directly experiencing long waits in the emergency department appear to be affected. However, if overcrowding on the wards has the potential to impact substantially more patients, if not all patients, even a very small risk translates to a very large number who might be potentially harmed.

The need for staff

The other issue is that continuous flow models don’t actually address the nub of the problem, which is the massive shortage of health and social care workers in the community. Less than 20% of the problems with delayed discharge appears to be within the gift of hospital staff on wards beyond the emergency department. Most delays to discharge from hospital (66%) are caused by waiting for community sector services. Pushing harder to discharge may result in patients being sent home early or ending up in the wrong place, simply for the sake of expediency. This, in turn, would drive readmissions, further burdening the emergency department. Which seems like something of an own goal.

There are many that argue something needs to be done now and that continuous flow models offer a potential solution for the coming winter. But there are many other things that could be done now that would help support patient flow through hospitals. And, indeed, hospitals implementing continuous flow models have insisted on a number of these anyway: decision-making earlier in the day; prescriptions for discharge medications done the previous day, with timely pharmacy services; better use of discharge lounges. Could the same results be achieved just with relentless focus on other improvements?

The success of a similar model was found to depend heavily on inpatient wards having sufficient staff to cope with the increased number of patients. With staff shortages and a potential strike in the offing, it is hard to see how most hospitals could support a continuous flow model.

Cause for caution

The NHS has a tendency to grasp for solutions too quickly and to assume that just because something works in one hospital, it must work everywhere else. We know that centrally mandated solutions often end up being implemented in haste, with little fidelity to the original plan, and frequently destabilise already fragile systems. Continuous flow models are acknowledged as being risky and needing careful introduction. So there really is cause for being cautious. We just don’t have enough deep understanding of the really sticky problems facing hospitals this winter – the underlying relative risks, the dynamics of the relationship between the emergency department and the wards, or the pros and cons of continuous flow models over the long-term – to be able to make any kind of good judgement.

There is really an urgent need for good, high-quality research. Given that’s the case, then ‘first, do no harm’ applies.