Let's Cut the NHS a bit of Slack
When I was a junior doctor, I rotated through the surgical specialties. Being on-call for surgical emergencies, particularly on the weekends or at night, the workload could vary dramatically. It could be an extra night's sleep or a complete nightmare. When things were calm for too long, the doctors' mess became a superstitious place to hang out, the word 'quiet' could not be spoken out loud for fear of jinxing our collective good fortune (this phenomenon prompted a brilliant Christmas BMJ RCT).
Weekly variations in workload were inevitable in the hospital and also predictable. Just as inevitable although less predictable, were the crises, when the system itself became overwhelmed. This usually meant a superhuman effort from all staff, which often included some extraordinary bed (management) juggling to keep the show on the road.
I moved from medicine to industry and observed what I now believe to be the same phenomenon but in different contexts. I was once given a tour of a UK call centre, or more accurately a service desk, that provided a single point of contact (SPOC) for customer issue resolutions. The team needed to stick to pre-agreed service level agreements (SLAs) or incur a financial penalty. An older team member told me stories of the early days, when they did not get the staffing numbers right. A small reduction in the ideal number of staff at the wrong time could result in a dramatic increase in the time it took to resolve call issues and subsequent SLA breaches.
I also experienced this on my new commute along the M25 where I became mildly obsessed with trying to predict the flow of traffic. After a few months, it occurred to me that while the outer lanes were open to overtake slower vehicles when traffic was light, as traffic became heavier, the outer lanes usually became the first to stop moving. The truck drivers who stuck to the inside lane appeared to always move faster than their more agile overtaking fellow travellers when traffic was heavier. And then less predictably, all traffic could then just stop, for no apparent reason, other than the sheer weight of traffic.
All three of these seemingly distinct phenomena can be explained by what Tom DeMarco refers to as Slack (worth a read). His core idea is that when a system attempts to maximise productivity and efficiency it can sometimes have the opposite effect.
An extreme example makes the point. Let's take the M25 and consider maximising its efficiency in terms of the passage of cars per stretch of road. If we calculate the maximum capacity of the M25; a length of 188 km, 4 lanes in both directions, gives us a total of 188x4x2 =1,504km or 1,504,000m of space. The average car length is say 5m (for easy maths), which makes the maximum car capacity of the M25 1,504,000/5 = 300,800 cars. You don't need to be a town planner to see the issue here. If we aimed for 100% utilisation of the M25, presumably to maximise the return on our collective investment, we would create the country's biggest car park.
But why? At what point did we actually make our mistake? In theory, we could get 300,800 cars to drive at 70 miles an hour around the M25. That is theoretically possible. The issue is that if just one of them tapped the brakes, it would cause the one behind to brake harder, and the one behind that to brake even harder, until at last a car has to stop. Like dominoes, a line of stationary cars would form very quickly behind that stationary car.
To keep things running smoothly we need to cut the system a bit of Slack. Here braking distance is the slack that's needed. It's not just about safety. It keeps the traffic moving. By leaving room in front of me, I can accommodate slight changes in the speed of the car in front. Like a buffer that can grow and shrink, it absorbs the random changes in the system, to keep the overall flow of cars moving. When traffic is heavy, cars in the overtaking lanes typically leave less braking distance than trucks in the slow lane - which explains what many of us will observe on a daily basis.
You can see this in action with a simulation by Martin Treiber here (adjust the 'Density/lane' value).
The same principle is true for our service desk. Some slack is required to adjust to sudden peaks in demand. When that buffer is exhausted the effect of a sudden additional peak dramatically increases the call resolution time.
And again, our hospital on-call team is equally affected by this. When there is no more slack in the system, a crisis is more likely, and when it does happen, it's more likely to escalate rapidly.
In all these examples, our intuition is challenged, because we expect things to progress in a linear fashion: for every additional car on the road, the speed of traffic is reduced by the same amount. But instead these phenomena, at a certain point, exhibit non-linear behaviour: for every additional car on the road, the speed of traffic is reduced by the same proportion.
Our lack of intuition for understanding these types of non-linear phenomena was noted by the physicist Albert Allen Bartlett:
"The greatest shortcoming of the human race is our inability to understand the exponential function."
There are, of course, other reasons why an organisation's productivity might be poor, and the cause may not be the absence of an appropriate buffer in the system. So this is not an excuse to abdicate management responsibilities. Teams still need to be led and motivated. Many of us have worked in environments where morale is low and productivity suffers. The absence of proper management, a clear unifying goal, and the general supporting infrastructure to get the job done, can be incredibly demoralising to any team. But I have also seen many other examples, first-hand, where the absence of a buffer, or a bit of slack, is a far more credible explanation for the underperformance of the system.
The question here for organisations is how do we calculate the ideal number of staff to respond to an incident, meet an SLA or manage an emergency department. The answer would typically be a balance between the additional cost for a larger buffer versus the economic and reputational impact of the system being overwhelmed. The service desk provider can make a very reasonable cost-benefit calculation.
For a hospital, management and staff have an additional legal duty of care, for patient safety, quality of care and safeguarding against risk. These should all be taken into account when determining the size of the buffer.
Our current healthcare system is set up to give us the outcomes we have. To improve this, we should consider adding, by design, an appropriate buffer. Let's debate what it should be, let's agree how we pay for it, let's make it a national target, but let's not wait for the next crisis before we start the conversation. Let's cut the NHS a bit of slack.