24 July 2010

Dicing with death

Once or twice the Jobbing Doctor has commented on how odd it is that he has seen three cases of a condition that he hasn't seen for ages all in one week. The same thing happens in hospitals. Very rare things come in runs. You would think that it couldn't happen by chance but it does. If you listen to radioactive decay using a Geiger counter the blips you hear (listen) are random but they are not evenly spaced. Sometimes the blips seem to occur in runs. So it is with any random event.

If you have ever wondered why it is you have to wait so long for a bus, that too is predictable. Unless the buses are evenly spaced you are more likely to arrive at a big gap between buses - because it is bigger. It's as simple as that. That's why when you travel on the tube they may stop the train to improve the service. It might seem crackers but it is to ensure the trains are evenly spaced which means that people will never have to wait too long. Even spacing is crucial. Dr Grumble has patients whose job it is to control the buses in the same way. Heavy traffic can limit what they can do but GPS gives them the opportunity to try and space the buses evenly along the route. The days of buses coming in threes should be over.




People don't understand statistics. Dr Grumble includes himself in this. It is a difficult subject. Managers certainly don't understand statistics. They spout utter nonsense. We had top management consultants in recently. We paid them a small fortune. They were on about how we should improve our performance to reach that of the best quartile of equivalent hospitals. There was the unspoken assumption that our performance was below par because we are crap. Dr Grumble had been warned in advance to behave. He pointed out serious flaws in the data but was given a stern look. He was tempted to point out that 50% of doctors are below average and should be sacked but he bit his tongue. Mad management speak fools nobody but highlighting the flaws in their thinking is not permitted in today's NHS.

Sometimes though the effects of statistics on individuals are too serious to ignore. Explaining this to managers who just want to cover their backs by sacking poorly performing doctors is a challenge. They cannot understand how sheer chance can make you a killer.

7 comments:

Cockroach Catcher said...

It is a well know fact that Pres. Bill Clinton picked the New York hospital with nearly twice the mortality for his heart surgery.

Crude statistics will say nothing about the kind of cases that were included in the stats. I am not excusing some hospitals that were not doing difficult cases for their bad mortality figures, but what Clinton did spoke tons. And he is still alive.

Private hospitals pick and choose and I have no idea if it is still happening: often in a critical situation, the patient will be transferred to an NHS hospital.

Stuart Jeffery said...

It is a disgrace that half of doctors are below average and it is appalling that there aren't at least half of all hospitals in the top quartile.

Apparently too, if you choose two different numbers, one will be higher than the other. Surely management consultants ought to fix that...

Dr Grumble said...

Getting half of all hospitals into the top quartile would certainly seem to be the aim of our management consultants. Brilliant idea.

Anonymous said...

I can relate to the example cockroach catcher gives cos' I've seen the same thing happen in schools. They avoid allowing some borderline pupils to even sit certain exams in case they fail and bring the league table figures down for that school. The stats for the school therefore don't reflect its performance in how pupils are taught because it still means that borderline pupils don't get even the chance to attain that GCSE.

At least that only affects a person's school grade, what you're referring to is playing with people's lives for the sake of better statistics. A hospital which performs more risky surgeries (cos' the alternative is making the patient 'comfortable' to die) could end up with a higher mortality rate even though that statistic is not a reflection of bad performance whereas a hospital with very low figures may just be avoiding risky surgeries altogether meaning it's not necessarily the best place to go in a medical emergency.

When my dad used to work for a telecommunications company, this sort of stupidity from management consultants was what he referred to as "bottom line finance". I cringe at the thought of using this same approach with hospitals. You were right to answer them back and I like the comment about trying to put half of the hospitals in the top quartile lol:p

Neelu

Anonymous said...

It is quite possible for 99% of doctors to be above average. For example there are some Drs with one leg, and some with none. There are none with three and most have two. The mean could therefore be 1.9999 legs, with 99% of drs above average in this measure. They could not be in the top quartile though as that is similar to the median.

Our management is also keen to have us in the top quartile of similar Trusts, without considering why we are not in it already. It is simpler for them to flog the galley slaves a little harder than to consider that their past decisions put us on the wrong course in the first place.

Arrows Group healthcare said...

Good post.

Clearly, statistics can only be the first clue to discover the correlations.

We need to ask 1,000 questions before accepting statistics, that by themselves, are worth very little

Simon Cole said...

Clusters occur for we surgeons, and are a well recognised phenomenona. I preferthe cluster of appendicitis cases, because I can speculate about the recent change in weather being responsible, on my ward round. The old jokes are the best....
I am currently experiencing a horrible gall bladder cluster, and is months ago I had a cluster of anastomotic leaks. These are stressful clusters, and no surgeon enjoys them.
Try explaining that to a dunderhead manager. Not a hope of making them understand it.
The chief difference between most managers and most doctors, is that back in the day we were the cream of the university educated cadre and they were middling polytechnic students. It stands to reason: if the milk round still existed, do you think decent graduates would take a job in the NHS? When there are jobs in decent nice places to work like Marks and Spencers, no-one in their right mind would take a job in the NHS.
I feel sorry for some hospital managers. There are a few around who are well motivated and patient orientated. Maybe the coming financial cutbacks will serve us well by triggering a cull of the rest. Here's hoping.