The main weakness of HSMR is it only captures immediate patient deaths, without capturing incidents that occur during later episodes that result from suboptimal treatment quality at an earlier episode (Bottle, Jarman, and Aylin 2011).
Specifically, the negative of the HSMR rating is included because the less negative the HSMR rating is, the better the hospital quality.
Even if all hospitals were good, half would still have ratios of 100-plus and look bad - and vice versa The Dr Foster guides and website emphasise these points scrupulously: that a high HSMR may be a product of chance, rather than failings in care; that high ratios should be treated as warning signs, with attention focusing mainly on the 'outliers' - those outside, particularly if repeatedly outside, the normal range.
It was wide, and, although all mortality rates have fallen significantly in the past decade - with the HSMR baseline adjusted accordingly - it remains so today.
Dr Foster frequently changes the methodology of the HSMR
which, in our opinion, further reduces its credibility as a comparator.
Much hospital activity is excluded and so alternative formulation of the HSMR could lead to alternative conclusions about a hospital's mortality (Shahian et al.
The expected deaths in an HSMR are the sum of the individual patients' predicted risks derived from a set of logistic regression models, one model for each of the 56 component diagnosis groups (ICD10 codes were grouped using the Agency for Healthcare Research and Quality's Clinical Classification System [CCS]--follow the links on http://www.
are seen as an indicator of patient care and the report showed the trust to be one of the worst 19 of 147 in England.
Overall, the survey found that death rates in hospitals are decreasing, with just 19 of the 147 trusts investigated now recording significantly high HSMRs
, compared with 27 last year.
After the care bundles were introduced, the HSMR
of the trust fell from 89.
He argued that, when combined with other factors, a high HSMR
can indicate broader problems with patient care.
A cloud of residual statistical uncertainty and unmeasured factors of course surrounds every HSMR
estimate, but the observed variation is far greater than the uncertainty in measurement.