Pages

Tuesday, 25 February 2020

NHS performance can be improved by paying more attention to reliable data and sound analysis

While the NHS and many other public services need more resources this might not be the best path to big improvements in quality and performance. Not least because deploying extra resources in the wrong place might yield only small improvements. Maybe we should start by spending more on getting the data and analysis required to understand where the biggest problems are and what interventions might solve them. And then focus on installing effective performance management throughout the system so improvements happen and stick. This will be expensive, but is demonstrably worth it.


It is almost universally assumed by commentators that the only thing that will improve the performance of the NHS or any other public service is a bigger budget. The bottleneck preventing better performance is austerity. The police need more cops; the NHS needs more doctors and nurses.

But is this true? Is a bigger budget the only way to improve anything? Is there nothing else that we should be doing?

The answer is no. really large improvements in performance are possible even without really big increases in the budget. And we know how because it has been done.

In the early 1990s New York had around 2000 murders a year. It now has fewer than 300, the lowest since records started. Other major crimes show similar reductions. What kicked off the improvement and kept that improvement happening for the next 25 years wasn't a vast increase in police numbers but a management process called Compstat.

The story of Compstat is not well known as lazy journalists have tended to credit the improvement to Broken Windows Theory (see, for example the Wikipedia page on crime in New York or this longer discussion about what really happened). Few accounts even mention the man who developed the Compstat process, Jack Maple, who wrote a book (The Crime Fighter) describing how he developed it and how it works.

What Maple describes ought to be of great interest for other public services like the NHS.

According to Maple The essence of Compstat is built on 4 key principles:

  1. Accurate, timely intelligence
  2. Rapid deployment
  3. Effective tactics
  4. Relentless follow-up and assessment
The goal of the process is to reduce crime rather than to pursue proxy metrics like arrest rates or response times. A key part of the process is the weekly meeting of local police commanders where they are not held to account for missing their targets but for not understanding the patterns of crime and not having effective plans to address the crimes in their area. The focus of performance meeting is to develop and share that understanding and to make sure it is being acted on. Commanders who fiddle their numbers rather than tackling their problems are ruthlessly exposed.

This contrasts sharply with how the NHS manages performance. It is worth comparing the NHS process with Compstat.

The NHS shares with Compstat the idea of regular meetings where the performance is reviewed. But what happens in the meetings is very different. In the NHS the meetings are not based on a wide range of insightful metrics that provide insight to why performance is bad but are based on a small number of less reliable headline metrics that provide no insight into the why of performance. And local managers are berated for failing to meet the target and encouraged to promise future performance without being encouraged to show they understand the key causes of poor performance. Neither the central team nor the local managers have a wide range of shared data that helps anyone understand the causes of problems nor do either group have much of a clue about what effective actions to improve performance look like. Since there is no collective understanding of problems there is no sharing of good ways to tackle them. The system reverts to measuring inputs not outputs (more nurses and doctors sounds good but won't help much if we don't know where the problems are).

In short, all the effective habits of Compstat are undermined from the top down in the NHS because there is no focus on understanding why performance is poor.

In a 2015 HSJ article called The way the NHS manages A&E problems is not fit for purpose, Nigel Edwards described this process as "A significant organisational pathology".

The NHS hasn't always been like this. Though it has never been as thorough as Compstat, there was a period in the 2000s where effective actions to improve performance were understood and performance did improve (eg starting in 2002 A&E performance went from ~70% of patients admitted or discharged in 4hr to 98% in a 3 year period). Most of the knowledge from that time seems to have been lost. The Cabinet Office even used studies on what worked in Compstat to help design the way performance management was done (but the 2004 document describing this–see pages 26-27–is buried deep in the national archives and it is doubtful anyone currently in power has read it).

Part of the reason why the NHS has forgotten what it once knew is that running a management process like Compstat is expensive (Maple estimated it might take 5% of the police budget). That is more than the NHS spends on management. Worse the most consistent NHS strategy over time–from both Labour and Conservative governments–has been to cut management so it can devote more resources to the front line. But adding more resources does little for performance if those resources are not well coordinated and deployed.

The Compstat process has worked for more than 25 years in the NYPD and improvements are continuing. But it also worked in cities like New Orleans where the police were far less well funded.

Perhaps the NHS has something to learn. Reforming its shonky management processes might be the best way to improve performance.