A continuing contribution to the promotion of sanity and decency in life with a few delights do lighten the load.
Thursday, 26 March 2015
A Good News Story
A small child got flu and was seriously ill. She recovered. Geneticists studied both her and her parents DNA. They discovered that she had a genetic mutation whereby she failed to make quantities of Interferon in response to infection. In this regard both her parents were genetically normal. The genetic finding has been confirmed by testing blood samples from the clild and her parents with influenza virus. Interferon is readily available to treat patients. How many more people are at risk of serious illnes from flu due to this mechanism and how many might be helped if they got flu by giving them Interferon.? Time will tell but the news is good.
Friday, 13 February 2015
Unnecessary Deaths
The UK health secretary says 1000 people are dying unnecessarily each week in Britain's NHS hospitals. Aside from the fact that this is intended to drive people in greater numbers to the (no more safer or efficient) private sector he is of course wrong. but then politicians often are and on healthcare issues they have a noble tradition of being wrong!
but he must have got this number form somewhere! Indeed he did. And the source was wrong as well.
How come?
Its easy to be wrong. There are lots of ways of being wrong but usually only one way of being right!
Hospitals make records of all the patients they see. The amount of information on each patient is quite small - typically it would include age, sex, times and dates of admissions and discharges, whether the patient died in hospital and a list of the diagnoses reached by the doctors. This sounds fairly straightforward but recording the diagnoses is often wrong. The doctors may not have made the correct diagnoses(or indeed any at all), they may not have recorded their findings properly and what they have recorded is not correctly transmitted to the hospitals information system - this is done by applying a code to each diagnosis reached (and each major procedure carried out). This coding is done by hospital staff reading the doctors diagnoses and converting this into an agreed code. Sounds easy? It does -but it is not. There a re many reasons for it being wrong! However let us assume all that is correct. What about the 1000 deaths?
Each hospital makes its records available to a central system, often a private company who analyses the information and compares hospitals one against the others.
For each diagnosis there will be a death rate for every hospital that had patients with that diagnosis. Sort these hospitals in order and one hospital will have the lowest death rate. Every other hospital will have higher death rates and if the lowest hospital is doing everything right why can't all the others. By this test all the deaths in all the hospitals apart from the lowest that are above the lowest rate are "avoidable". Do this for hundreds of diagnoses and in no time you have hundreds of "avoidable" deaths. Except you don't. No one hospital will have the lowest rate for everything - indeed it is entirely possible that every hospital might have the lowest death rate for something!
Why? Things vary in ways the numbers collected cannot explain. Sure they can allow for the fact that one hospital might on average treat older or younger patient, more men or more women, more patients with multiple diagnoses and so on and this does help to make the information more understandable. But the collected information leaves out far more than it contains. it does not tell much about the patients' past lives, how fit they were before coming ill and crucially how ill they were on this occasion. Not all heart attacks are of the same severity, Not all bouts of pneumonia make the patient as unwell. Not all patients with a given diagnoses come to hospital - some die at home - and so on.
There are not 1000 avoidable deaths each week. There are od course avoidable deaths - no system is perfect - but for now we do not know how many or indeed which deaths are avoidable. One day information gathering will improve and we may know the answer to that question. But not yey.
but he must have got this number form somewhere! Indeed he did. And the source was wrong as well.
How come?
Its easy to be wrong. There are lots of ways of being wrong but usually only one way of being right!
Hospitals make records of all the patients they see. The amount of information on each patient is quite small - typically it would include age, sex, times and dates of admissions and discharges, whether the patient died in hospital and a list of the diagnoses reached by the doctors. This sounds fairly straightforward but recording the diagnoses is often wrong. The doctors may not have made the correct diagnoses(or indeed any at all), they may not have recorded their findings properly and what they have recorded is not correctly transmitted to the hospitals information system - this is done by applying a code to each diagnosis reached (and each major procedure carried out). This coding is done by hospital staff reading the doctors diagnoses and converting this into an agreed code. Sounds easy? It does -but it is not. There a re many reasons for it being wrong! However let us assume all that is correct. What about the 1000 deaths?
Each hospital makes its records available to a central system, often a private company who analyses the information and compares hospitals one against the others.
For each diagnosis there will be a death rate for every hospital that had patients with that diagnosis. Sort these hospitals in order and one hospital will have the lowest death rate. Every other hospital will have higher death rates and if the lowest hospital is doing everything right why can't all the others. By this test all the deaths in all the hospitals apart from the lowest that are above the lowest rate are "avoidable". Do this for hundreds of diagnoses and in no time you have hundreds of "avoidable" deaths. Except you don't. No one hospital will have the lowest rate for everything - indeed it is entirely possible that every hospital might have the lowest death rate for something!
Why? Things vary in ways the numbers collected cannot explain. Sure they can allow for the fact that one hospital might on average treat older or younger patient, more men or more women, more patients with multiple diagnoses and so on and this does help to make the information more understandable. But the collected information leaves out far more than it contains. it does not tell much about the patients' past lives, how fit they were before coming ill and crucially how ill they were on this occasion. Not all heart attacks are of the same severity, Not all bouts of pneumonia make the patient as unwell. Not all patients with a given diagnoses come to hospital - some die at home - and so on.
There are not 1000 avoidable deaths each week. There are od course avoidable deaths - no system is perfect - but for now we do not know how many or indeed which deaths are avoidable. One day information gathering will improve and we may know the answer to that question. But not yey.
Subscribe to:
Comments (Atom)