There are many ways of monitoring mortality in general practice. While
calculating standardised mortality ratios for Primary Care Groups we
calculated the 3 year SMRs for individual practices within a health
authority. The data was derived from the GP registrations held by the health
authority which includes details of patients removed the register following
death and the age and sex of all patients registered with GPs. The SMRs
varied between about 30 (for a medical practice looking after healthy young
people) to 200 (for a single handed medical practice in a deprived inner
city area), however the confidence intervals were wide. SMRs for individual
doctors would be more difficult since partners within a practice usually
care for each others patients making it more difficult to choose an
appropriate denominator for individual doctors. Other problems include even
wider confidence intervals. The retrospective nature of the investigation
which means a lot of deaths might have occurred before any investigation
takes place.
Another approach might be to use the cumulative sum approach comparing at
weekly intervals the number of deaths occurring among patients in each
practice with that expected from the age and sex distribution of the
practice population. This would give a rapid warning of when the death rates
are deviating from that expected.
All these approaches would involve lots of resources to avoid what as far as
we know is a unique series of murders. The adverse effects of such
monitoring might include general practitioners avoiding caring for sick
patients and working in disadvantaged areas were the death rates might be
higher. Therefore such monitorng might actually lead to more deaths than
they prevent.
Perhaps the best approach is for people to be more alert to rumours. In 1998
the GPs in a neighbouring practice informed the coroner having noticed that
Shipman had signed 41 cremation certificates compared with 15 for the
average GP. Assuming they are both looking after a list of about 2500
patients this gives a chi sguare of 11.91 df=1 p=0.0005. I am sure that if a
simple statistical analysis had been undertaken at that time, the first
investigation would have been conducted with much more vigour.
Jonathan Higham
SR Public Health Medicine
-----Original Message-----
From: [log in to unmask] <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>
Date: 01 February 2000 22:03
Subject: MONITORING DOCTORS PERFORMANCE
>It looks as if one of the first new 'National Statistics' might be measures
>of performance of General Practioners!
>
>In the aftermath of the conviction for murder of Dr Shipman, one of the
>promises made by Alan Milburn the Minister of Health is to involve the ONS
>in monitoring the performance of GPs.
>
>I suppose it would be relatively easy to calculate the average expectation
>of life of patients of a particular doctor. The data needed is simply
date
>of birth and date of death for all deaths. Is is not likely that, even
>though Shipman's victims were elderly, such calculations made on the basis
>of the deaths of his patients would have shown that something strange was
>going on?
>
>Standardised mortality rates for individual doctors, and individual
>practices, could also be calculated, but that would require data on the age
>distribution of all patients.
>
>This kind of monitoring is has some precedents in the US. New York State
>publishes statistics for the mortality rate for coronary artery bypass
>surgery for individual surgeons every year. See
><http://www.health.state.ny.us>.
>
>
>Ray Thomas, Social Sciences, Open University
>Tel: 01908 679081 Fax 01908 550401
>Email: [log in to unmask]
>35 Passmore, Milton Keynes MK6 3DY
>
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|