• Hospitals report data irregularities which could mean SHMI needs to be “recalculated”
  • Increase in “same day emergency care” patients has led to hike in mortality scores despite no real-world change in patients’ prospects
  • Problem follows concerns raised last week that A&E data was “contaminated”

An increase in so-called same day emergency care patients and confusion over how they should be counted has caused an artificial hike in some hospitals’ mortality scores, senior sources have told HSJ.

The mortality score irregularities represent the second major concern over the accuracy of a major NHS data set in as many weeks, following NHS bosses warning last week the four-hour target data was “contaminated”.

The two issues are both connected to what the NHS’ emergency care chief called an “explosion” in patients being treated and discharged on the same day, often in ambulatory care units, for conditions which up until relatively recently would have warranted an overnight admission.

Although this development is seen as both clinically and financially advantageous and better for patient experience, national bosses have not yet rolled out robust counting methodologies so trusts can properly code these patients.

Removing these lower risk patients from the hospital setting, and the corresponding admitted patient care data set, leaves hospital trusts with a higher risk population. This inadvertently pushes up trusts’ summary hospital-level mortality indicator score, despite there being no real-world change in a patient’s chances of death (See box: SHMI explainer).

HSJ understands NHS England and NHS Digital are aware of the problem, which they are looking to resolve as part of a wider review of the SHMI being carried out by the two bodies. It is, however, unclear how many trusts are impacted and when the review will be concluded.

A senior NHS source familiar with efforts to address the problems told HSJ while it was a statistical problem and not a reflection of the reality, the SHMI could need to be “recalculated” to take into account the issue.

The source said: “NHSE is aware of the problem, which is a statistical issue not reflecting real mortality performance. I understand they are working on guidance on how to count same day emergency care cases [which] is going out soon.”

Mortality rates expert Chris Sherlaw-Johnson explained: “If patients are counted in the outpatient data or the non-admitted [accident and emergency] data then they will not appear in the SHMI, so it is certainly plausible that problems could arise from a lot of low-risk patients being taken out of the denominator.”

The senior research analyst at the Nuffield Trust added: “Without clear guidelines for how to report patients moving through new care pathways, there is no assurance trusts will be reporting the same activity in the same way.

“This, in turn, could have an impact on the reliability of statistics that are used to compare and monitor their outcomes.”

NHSE was approached for comment. 

SHMI explainer

The summary hospital-level mortality indicator is the ratio between the actual number of patients who die following hospitalisation at a trust and the number that would be expected to die based on average England figures and the characteristics of the patients treated by the trust.

It covers patients admitted to hospitals in England who died either while in hospital or within 30 days of being discharged.

Trusts are categorised as having a mortality rate which is “higher than expected”, “as expected” or “lower than expected”.

NHS Digital guidance says the SHMI is not a quality measure. A higher than expected number of deaths should not immediately be interpreted as indicating poor performance. But it is an alarm which requires further investigation, it says.

There is a significant lag in when SHMI data becomes publicly available. The most recent SHMI data published by NHS Digital in February covers October 2017 to September 2018. It recorded that there had been 299,000 deaths out of 9.1 million discharges.

Fifteen trusts had a higher than expected number of deaths and, of these 15 trusts, seven also had a higher than expected number of deaths for the same period in the previous year.

There were 101 trusts which had a number of deaths within the expected range and 15 trusts which had a lower than expected number of deaths.