The NHS, data and management
Imagine you are shopping in a supermarket. You fill your trolley and turn up at the checkout. When you get there, the checkout boy is holding a notebook. There is no scanner or technology. He takes each item from your trolley and notes what it is by hand in his notebook. When he has emptied your trolley, he then looks up all the prices for your goods in a book. Then he adds up your bill manually.
Imagine how frustrating, slow and error-prone this would be for you as a shopper. What may be less obvious is how much it would mean to the backroom systems in the supermarket. It would be hard for them to ensure the shelves were stocked because the knowledge about how fast good sold would flow slowly. Whether some new products were selling well or badly would be hard to discover quickly so the shelves might be full of things you don't want and bereft of those you do. It would be a horrible, expensive, ineffective mess. Shoppers would hate it and so would the checkout staff who would complain about how they were overwhelmed by tedious paperwork (and shopper complaints).
The thing is, that is how we run the NHS.
Supermarkets have invested a lot of money to make shopping easier and automatically and quickly keeping track of the information they need to know. The NHS has tended to assume that spending money on information would detract from money spent on “front line” staff. As a result, to get any useful information at all about what is happening tends to be burdensome, slow and annoying (taking up a great deal of the time front line staff should be spending with patients).
As a result, it is hard for hospitals or the system as a whole to know what is going on, whether it is working or whether the activities of the hospital need to change to match demand from patients. It is also hard to coordinate care for patients. Where information about, for example, what drugs a patient is taking needs to be shared among all the staff caring for the patient, it often isn’t as it hasn’t been properly collected in an electronic form that can be easily shared for all staff. This results in high error rates many of which harm patients. Some hospitals estimate, for example, that 20% of all hospital prescriptions are never given to the patient because there is no automated system to ensure dispensing happens. Hospitals frequently lose track of where patients are delaying their treatment or discharge and blocking the beds for new arrivals (which, in turn, is a major cause of delays in A&E).
What may be less obvious is how this discombobulates a whole range of management decisions from day-to-day staff scheduling to long term planning. How, for example, does a hospital know how many staff it needs or how many free beds it needs to operate its A&E effectively with no delays for patients? How does the system know how much to pay hospitals for the work they do? How, in the long term, do we know how many hospitals we need and with what facilities?
A fair allocation of money to hospitals, for example, requires a good knowledge of what hip replacements or cataracts cost to treat. If this information is poor the money will buy less care than required and hospitals will struggle to balance the books. Even more importantly, they won’t be able to work out whether they could do hip replacements or cataracts more effectively (achieving better quality at lower cost). Some might argue that we shouldn’t be paying hospitals this way, but alternative payment systems don’t solve the problem: they just makes it a hospital problem not a system problem. If hospitals don’t get this right internally, their care will be more expensive and worse.
Perhaps less obviously (as many assume that hospitals are already doing their absolute best with what they have) a lack of information prevents hospitals improving the quality, effectiveness and cost of what they do over time. Small changes to procedures accumulate over time to give big improvements in both cost and quality. But finding and monitoring those improvements requires detailed analysis of tf how things are done and the outcomes that result for patients. If we don’t have reliable data, we can’t drive or even monitor those improvements.
An example of where this matters to a current problem in the system is how hospitals manage their beds and the knock-on effect this has on A&E performance (which is currently very poor mainly because it is hard to find free beds for patients who need to be admitted). Peak need for beds for A&E patients (at least in the UK) occurs around lunchtime but in most hospitals peak discharges from beds occur late in the afternoon (implying that many beds are occupied by patients fit to go home just when they are needed for new patients). Working out how much could be gained if a hospital altered the typical time of its discharges is, however, hard because few collect any information about what time their discharges happen. One reason often given for why the information is not recorded is that it would impose an enormous paperwork burden on nurses who are already busy looking after patients. Well, OK, but only if, like the nightmare vision of a supermarket we started with, we ignore technology and insist everything has to be done manually on paper. We also have good information that the majority of reasons why discharges are delayed are caused by poor internal coordination of hospital discharge processes, again a consequence of a failure to use technology to make administration run smoothly.
The NHS has some very big challenges to meet over the next decade.
But it is turning a blind eye to the need for reliable information to help it improve and the technology investments that would enable that information to be collected.