Reservations on today’s report by Nuffield Trust which compares funding and healthcare performance across the UK centre on: Rurality (which not even comparison between Scotland and North East England redresses), Reporting standards, Right measures, Responsiveness, and Repeat attendances.

First, I declare an interest – I am a Scot who works in England. Also, I chaired the Royal Statistical Society’s Working Party on Performance Monitoring in the Public Services which in 2003 called for caution on ‘name and shame’ and that measures of variation be always reported.

The Nuffield Trust’s report is light on reporting standards – hardly a measure of variation in sight. For example, without denominators - including numbers of ambulance calls, and numbers designated category A - Figure 4.2 reports the percentage of category A ambulance calls met within 8 minutes as around 75 per cent in England in 2006 but under 60 per cent in Scotland. Without knowing that methodology for category A designation is identical throughout the UK and without knowing category A workloads per million of population, insightful conclusions cannot be drawn from Figure 4.2.

This New Year, a category A call-out (suspected heart attack) was made to snow-fast Balquhidder Glen in rural Scotland. There arrived a paramedic by car and an RAF SeaKing helicopter which landed in Charlie’s field (air ambulances were elsewhere deployed). The patient was air-lifted to hospital in Glasgow, and remained there more than a week later having undergone tests for an interventional procedure. The 8 minutes target can seldom be met in Balquhidder Glen but the healthcare service responded impressively nonetheless and there was no uprising in Rob Roy country.

The Nuffield Trust gets some measures right. Infant and perinatal mortality rates per 1,000 births are appropriate measures, because changes this century are precisely discernible: on a 3-year basis if not year-to-year.

There are around 50,000 births per annum in Scotland, where perinatal mortality rate in 2006 was 7.4 per 1,000 births. Ridiculously, I’ve had to read this figure off Figure 5.4. If I have read-off correctly, then there were presumably around 370 perinatal deaths in Scotland in 2006 which means that an upper 95 per cent confidence limit for Scotland’s perinatal mortality rate of 8.2.

By contrast, England has over 500,000 births per annum and perinatal mortality rate greater than 8 per 1,000. Upper 95 per cent confidence limit for England would be about 0.2 greater than the level read-off . . . about the same as my accuracy in reading off Figure 5.4. Please, Nuffield Trust, report the data (numbers of perinatal deaths, numbers of births) on which rates are based and which are necessary for statistical inference - not just bar charts.

Singularly unresponsive to short-term changes in policy is life-expectancy, and hence an inappropriate measure in the current context because so historically reliant.

Fewer hospital attendances per hospital doctor may mean better healthcare, if all necessary investigations are co-ordinated within a single visit or attendance. My beloved late husband, Dr A. Graham Bird, was a renowned diagnostician and leading clinical immunologist of his generation. He prided himself on running his clinics in the above manner, and was a strong advocate of nascent NICE’s emphasis on cost-effectiveness.

The Nuffield Trust report is largely concerned with process measures, numbers of attendances or hospitalizations, not with the healthcare outcomes that are delivered per attendance. Another vignette – on repeated attendances to manage a single condition - illustrates this: a friend whose breast cancer was screen-detected in mid December had multiple hospital attendances for a range of separate daily investigations prior to her surgery in the week of 11 January 2010.