Review rejects claim of higher hospital death rates in England than in USBMJ 2015; 350 doi: https://doi.org/10.1136/bmj.h787 (Published 10 February 2015) Cite this as: BMJ 2015;350:h787
All rapid responses
My response, dated 16 March 2015, to this paper (the previous Rapid Response to the paper) seems to be missing graph 2 and table 2 so I am attaching them.
18 March 2015
Competing interests: I am employed part-time by Imperial College and advise Dr Foster about their international work
Reply to “Review rejects claim of higher hospital death rates in England than in US” BMJ 2015;350:h787
The Academy of Medical Royal Colleges working group February 2015 report to Professor Sir Bruce Keogh on International HSMRs (Hospital Standardised Mortality Ratios) stated four main reasons why they considered that US and English hospital mortality rates could not be directly compared.
Taking the Academy’s main points in turn, and referring to 2004 data, firstly I agree that in the US there may be greater financial incentives to put patients into a higher risk category and that, to use the Academy’s example of septicaemia, the more vague Clinical Classification System (CCS) group diagnosis of eg ‘pneumonia’ (CCS=2) may be more likely to be classified as ‘septicaemia’ (CCS=122) in the US than in England. However, if the two diagnoses are combined there is very little change in the countries’ HSMRs (the US overall HSMR goes from 76.0 to 75.0 and England HSMR from 122.5 to 122.6).
Secondly, I agree that more secondary diagnoses are recorded in the US than in England but the effect of the Charlson Index of secondary diagnoses on the overall HSMRs is relatively small and there is less than 2% change if it is removed from the calculation. The graph 1 and table 1 show the International HSMRs for (1) the ‘regular’ model (which includes the Charlson index and does not include length of stay), (2) for a model including length of stay, and (3) for a model without the Charlson index. The differences between the HSMRs of the three models are small for both the US and England, but there are highly significant differences between the US and England HSMRs for each model.
Thirdly, regarding the Academy’s claim that patients in England are sicker on admission to hospital based on an ‘acute physiology score’, it is difficult to comment because that measure is not available for all patients admitted to English and US hospitals. An argument could be made that there would be a financial disincentive for poorer patients and those without insurance, who may be sicker than average, to be admitted to hospitals in the US compared with England, which has the National Health Service covering virtually the whole population, but in the US 70% of hospital in-patient deaths are of patients, mainly over age 65, covered by Medicare, which is a national social insurance programme, administered by the US Federal Government. On the other hand, in the US, of those patients who are not covered by Medicare or private insurance, it would be the sicker patients who would be more likely to be admitted.
It is not correct to say, as the Academy does, that “hospital episode statistics in England do not include patients who choose to pay for treatment by private providers.” Hospital Episode Statistics include private patients treated in NHS hospitals, patients resident outside of England and care delivered by treatment centres (including those in the independent sector) funded by the NHS.
Finally, I agree that lengths of stay are longer in English hospitals and that a higher proportion of elderly patients die in acute care hospitals rather than in intermediate care facilities (such as hospices and nursing homes) in England than in the US. However, an analysis, discussed with the Academy, of the effect of the proportion of patients dying in English hospitals shows that it would only account for about 4.8% of the difference between the US and England HSMRs. In a sensitivity analysis of the impact of variants of the HSMR methodology it was found that comparing the usually calculated or ‘regular’ HSMR, covering only in-patients, with an HSMR including deaths up to 30 days after discharge, the difference between the ‘regular’ and 30-day HSMR was less than 10% for 92% of the NHS acute non-specialist hospital Trusts. I agree the greater use of different forms of intermediate care (such as in hospices) in the US than in England may mean that 30-day mortalities differ from the in-hospital mortalities more than in England and that it would also be useful to analyse 30-day HSMRs if the record linkage information were available. However, the graph of the International HSMRs using a model including length of stay indicates only small change in the US and England HSMR values.
The crude 2004 death rate for patients aged 70 or more is higher in English hospitals (see graph 2 and table 2). This is consistent with the census data from the US and England & Wales censuses, which show that the overall national death rate (covering both hospital and community deaths), in year 2001, in people aged 75 to 84 was 5.5% higher in England & Wales than in the US, and 7.7% higher in E&W in the 85+ agegroup. Although the analyses do not specifically cover the Liverpool Care Pathway (LCP) for the Dying Patient, which gave palliative care options for patients in the final days or hours of life, it is probably worth noting that the LCP was in operation in some English hospitals in 2004, and the 2013 English Department of Health-commissioned Neuberger review of the LCP concluded “it is clear to us…that there have been repeated instances of patients dying on the LCP being treated with less than the respect that they deserve.” The review stated “Whether true or not, many families suspected that deaths had been hastened by the premature, or over-prescription of strong pain killing drugs or sedatives, and reported that these had sometimes been administered without discussion or consultation. There was a feeling that the drugs were being used as a “chemical cosh” which diminished the patient’s desire or ability to accept food or drink. The apparently unnecessary withholding or prohibition of oral fluids seemed to cause the greatest concern.… Relatives and carers felt that they had been “railroaded” into agreeing to put the patient on a one-way escalator.” In 2013 the English Department of Health stated that the LCP should be phased out.
The HSMR is merely a measure of adjusted hospital death ratios and is not a direct measure of the quality of care given in a hospital or healthcare system such as the UK National Health Service. The US Commonwealth Fund study of some aspects of 11 developed countries’ healthcare systems, showed the UK overall ranking as the best and the US as the worst, although they found that the “U.S. and U.K. had much higher death rates in 2007 from conditions amenable to medical care than some of the other countries.” That study also found that “A higher percentage of people in the U.S. go without needed care because of cost than in any other surveyed nation…Patients in the United Kingdom” were “the least likely to report having these cost-related access concerns.”
In our BMJ 1999 paper on HSMRs we suggested that a “matched pair study of patients admitted to hospitals with high and low standardised mortality ratios could help to elucidate these findings. In such an investigation detailed data would have to be collected to allow for accurate adjustment of case mix.” I discussed that approach with the Academy as a possible means of assessing hospital care in the US and England.
Competing interests: Imperial College Dr Foster Unit calculates England HSMRs and I also calculate International HSMRs. I am employed part-time by Imperial College and I also advise Dr Foster about their international work.