Summary

Column

SHMI

The SHMI for Mid and South Essex for Oct 2019 to Sep 2020 is 1.063. This is as expected and is higher (worse) than for the previous period (September 2019 to August 2020) when it was 1.055 and as expected.

Across England there were 10 trusts with SHMI higher than expected, and 15 with lower than expected SHMI.

By condition

The conditions with highest number of excess deaths are Other R codes(60 deaths) , Pneumonia (excluding TB/STD) (50 deaths) and Urinary tract infections (45 deaths). Amongst the conditions with least number of excess deaths are Intracranial injury (-15 deaths), Intestinal infection (-10 deaths) and Cardiac arrest and ventricular fibrillation (-10 deaths).

In the NHSD published data there are no conditions that are significantly elevated on the Over Dispersion model.

Changes in counts

The number of deaths at the trust or within 30 days of discharge has risen faster (0%) compared to England (-1%) since March 2011.

The number of spells at the trust has risen faster (9.1%) compared to England (-2.2%) since March 2011.

Data quality

The trust has a higher (worse) percentage of Primary diagnoses (15.2%) that are signs and symptoms (England 13.2%) . Reducing the use of ‘R’ codes will both provide a better insight into what patients are admitted for as well as having a potentially positive effect on the SHMI and other mortality measures.

Coding depth for electives is 4.9, which is lower compared to England (5.2), and for non-electives 5.1 this is lower than England (5.6). Mid and South Essex is in the midrange of trusts for mean coding depth for nonelective admissions. Mid and South Essex is in the lowest 40% of trusts for mean coding depth for nonelective admissions. Coding depth is associated with the Charlson score that adjusts for ‘severity’ of a patient’s condition. Capturing co-mordities will both provide a better insight into the complexity of patients who are admitted as well as having a potentially positive effect on the SHMI and other mortality measures.

0.8% of spells have an invalid diagnosis (R69 codes). Mid and South Essex is in the highest 20% of trusts for percentage of spells with an invalid primary diagnosis_code. This is higher than England (0.6%).

End of life

64.3% Patients died in hospital, this was lower than compared to England (66.2%). Mid and South Essex is in the lowest 40% of trusts for percentage of deaths which occurred in hospital.

Of those patients who died 30.2% had a palliative care code in their last spell, this is lower than England (36.3%). Mid and South Essex is in the lowest 40% of trusts for percentage of deaths with either palliative care specialty or diagnosis_coding. For all spells, 1.7% had a palliative care code, this is lower than England (1.9%). Mid and South Essex is in the lowest 40% of trusts for percentage of spells with either palliative care specialty or diagnosis_coding.

Column

Tables, period up to Sep 2020

Measure Mid and South Essex
SHMI 1.063
SHMI lower limit 0.893
SHMI upper limit 1.120
Measure Mid and South Essex
Spells 170720
Deaths 5780
Expected 5435
Excess deaths 345
Measure England Mid and South Essex
Elective crude mortality 0.66 0.85
Non-elective crude mortality 3.56 3.63
Crude rate 3.24 3.39
Measure England Mid and South Essex
Depth of coding - Elective 5.2 4.9
Depth of coding - Non Elective 5.6 5.1
Primary diagnosis symptom or sign (%) 13.2 15.2
Invalid primary diagnosis (%) 0.6 0.8
Measure England Mid and South Essex
Deaths in hospital (%) 66.2 64.3
Deaths outside hospital (%) 33.8 35.7
Deaths with palliative care (%) 36.3 30.2
Spells with palliative care (%) 1.9 1.7

Overview

Row

SHMI: Oct 2019 to Sep 2020 is 1.063, this is as expected

MID AND SOUTH ESSEX NHS FOUNDATION TRUST

Row

SHMI

Excess deaths by condition, period up to Sep 2020. Click to see Conditions over time

Condition Outlier band SHMI Observed Expected deaths Excess deaths
Other R codes Higher than expected 99.8% limits 1.75 140 80 60
Urinary tract infections Higher than expected 99.8% limits 1.243 230 185 45
Cancer of breast Higher than expected 95% limits 2 20 10 10
Diseases of kidneys and ureters, bladder and urethra Higher than expected 95% limits 2 20 10 10
Other connective tissue disease Higher than expected 95% limits 1.333 80 60 20
Gastrointestinal hemorrhage Higher than expected 95% limits 1.312 105 80 25
COPD & bronchiectasis Higher than expected 95% limits 1.219 195 160 35
Intracranial injury Lower than expected 95% limits 0.769 50 65 -15
Disorders of stomach and duodenum Lower than expected 95% limits 0.4 4 10 -6
Ear and sense organ disorders Lower than expected 95% limits 0.4 4 10 -6
Esophageal disorders Lower than expected 95% limits 0.4 4 10 -6
Other cerebrovascular disease Lower than expected 95% limits 0.4 4 10 -6
Other perinatal conditions Lower than expected 95% limits 0.4 4 10 -6
Pathological fracture Lower than expected 95% limits 0.4 4 10 -6
Short gestation and slow fetal growth Lower than expected 95% limits 0.4 4 10 -6

Row

All trusts

Across England there were 10 trusts with SHMI higher than expected (highlighted in red), and 15 that were lower than expected (highlighted in green). Your trust is in blue. The limits are estimated and may be different to those calculated by NHSD, hence a trust may sit slightly below or above the limits.

Crude death rates

The charts below show the crude (unadjusted) mortality rate per 100 spells, for all patients and split by admission method, compared to England.

Activity

Relative activity changes

The number of deaths at the trust or within 30days of discharge has risen faster (0%) compared to England (-1%) since March 2011. The number of spells at the trust has risen faster (9.1%) compared to England (-2.2%) since March 2011.

Conditions

Coding

The trust has a higher (worse) percentage of Primary diagnoses (15.2%) that are signs and symptoms (England 13.2%) . Reducing the use of ‘R’ codes will both provide a better insight into what patients are admitted for as well as having a potentially positive effect on the SHMI and other mortality measures. Coding depth for electives is 4.9, which is lower compared to England (5.2), and for non-electives 5.1 this is lower than England (5.6). Mid and South Essex is in the midrange of trusts for mean coding depth for nonelective admissions. Mid and South Essex is in the lowest 40% of trusts for mean coding depth for nonelective admissions. Coding depth is associated with the Charlson score that adjusts for ‘severity’ of a patient’s condition. Capturing co-mordities will both provide a better insight into the complexity of patients who are admitted as well as having a potentially positive effect on the SHMI and other mortality measures.

Palliative care

64.3% Patients died in hospital, this was lower than compared to England (66.2%). Mid and South Essex is in the lowest 40% of trusts for percentage of deaths which occurred in hospital. Of those patients who died 30.2% had a palliative care code in their last spell, this is lower than England (36.3%). Mid and South Essex is in the lowest 40% of trusts for percentage of deaths with either palliative care specialty or diagnosis_coding. For all spells, 1.7% had a palliative care code, this is lower than England (1.9%). Mid and South Essex is in the lowest 40% of trusts for percentage of spells with either palliative care specialty or diagnosis_coding.

Site

There are 5 sites with more than 7 deaths and with a calculated SHMI.

Deprivation

The charts show the profile for the crude mortality rate (per 100 spells) by quintile from most deprived to most affluent, along with the percentage of all deaths and percentage of all spells.

Alternate views

These are NOT official measures, and are included only to provide a view of how the SHMI compares to the patient cohorts used in the HSMR. They are compared against the index value of 1 (dashed line). The vertical dashed blue line shows the impact of COVID-19 in March 2020. Click on the legend to see selected measures.

Conditions over time

Row

Changes in SHMI and excess deaths over time. The darker the colour the larger the SHMI, the BIGGER the circle the more excess deaths. Overview

Ribbons

The charts show trusts ranked from the lowest to highest for the period Oct 2019 to Sep 2020. Where it is possible to make a distinction between levels of good or poor performance, this is highlighted with red being poor and blue being good. Where it is not possible to say high is good or bad, the ribbon highlights the highest and lowest in darker shades. The bands are the lowest tenth, 20th centile, 40th centile, mid range (40-60 centiles), 80%th centile, 90th centile and the highest tenth of trusts.

column

Depth of coding: elective

Depth of coding: non-elective

Coding: invalid

Coding: signs and symptoms

column

Place of death: in hospital

Palliative care- deaths

Palliative care- Spells

column

Crude rate of deaths: all

Crude rate of deaths: Non-elective

Crude rate of deaths: elective

Manual

Row

SHMI

Information from NHS Digital, licensed under the current version of the Open Government Licence Source: https://digital.nhs.uk/data-and-information/publications/clinical-indicators/shmi

The SHMI is the ratio between the actual number of patients who die following hospitalisation at the trust and the number that would be expected to die on the basis of average England figures, given the characteristics of the patients treated there. It includes deaths which occurred in hospital and deaths which occurred outside of hospital within 30 days (inclusive) of discharge. Deaths related to COVID-19 are excluded from the SHMI.

The SHMI gives an indication for each non-specialist acute NHS trust in England whether the observed number of deaths within 30 days of discharge from hospital was ‘higher than expected’ (SHMI banding=1), ‘as expected’ (SHMI banding=2) or ‘lower than expected’ (SHMI banding=3) when compared to the national baseline.

(Source: NHS Digital - https://digital.nhs.uk/data-and-information/publications/clinical-indicators/shmi/current/shmi-data)

Excess deaths by condition

The conditions with the highest excess deaths are shown. Where there are more deaths than expected (excess deaths) these are shown in orange, and where there are fewer deaths than expected these are shown in green. The length of the bar is the total deaths (expected + excess deaths).

Conditions with 5 or fewer excess deaths are not shown to ease reading. All conditions are listed in the Conditions over time page.

Row

All trusts

The SHMI is not a measure of quality of care. A higher/lower than expected number of deaths should not immediately be interpreted as indicating poor/good performance and instead should be viewed as a ‘smoke alarm’ which requires further investigation.

The SHMI cannot be used to directly compare mortality outcomes between trusts and it is inappropriate to rank trusts by their SHMI. Instead, the SHMI banding can be used to compare mortality outcomes to the national baseline. If two trusts have the same SHMI banding, it cannot be concluded that the trust with the lower SHMI value has better mortality outcomes

NHS Digital’s interpretation guidance https://files.digital.nhs.uk/96/EE14DB/SHMI%20interpretation%20guidance.pdf

Observed Deaths: This is a count of the number of deaths which occurred in hospital or within 30 days of discharge for each trust. Deaths related to COVID-19 are excluded from the SHMI. If the patient is treated by another trust within those 30 days their death will only be attributed to the last non-specialist acute NHS trust to treat them. Specialist trusts, mental health trusts, community trusts and independent sector providers are excluded from the SHMI. A full list of excluded NHS trusts can be found in Appendix C of the SHMI methodology specification document, which is available on the NHS Digital website at http://digital.nhs.uk/SHMI.

Expected Deaths: The risk of the patient dying in hospital or within 30 days of discharge is estimated from statistical models based on the following variables:

  • the condition the patient is in hospital for other underlying conditions the patient suffers from

  • the age of the patient

  • the sex of the patient

  • the method of admission to hospital (elective/non-elective/unknown)

  • the month of admission

  • the birthweight of the patient (perinatal diagnosis groups only)

The expected number of deaths is obtained by summing the estimated risks for all finished provider spells for a trust. The statistical models are derived using thirty-six months of data from trusts throughout England. The final twelve months of this period are used to calculate the SHMI for each individual trust. Details of the dataset used in the SHMI calculation can be found in the SHMI publication timetable which is available on the NHS Digital website at http://digital.nhs.uk/SHMI.

(Source: NHS Digital https://files.digital.nhs.uk/DB/48BBE9/SHMI%20FAQs.pdf)

Activity

Crude mortality

Number of patients dying in hospital or within 30 days of discharge divided by the number of spells.

England is shown as a guide but due to variations in case mix you cannot draw any statistical comparison between the two.

Deaths for non-elective admissions

This indicator presents crude percentage rates of non-elective admissions where a death occurred either in-hospital or within 30 days (inclusive) of being discharged from hospital.

An adjustment is made for admission method in the SHMI methodology because crude mortality rates for elective admissions tend to be lower than crude mortality rates for non-elective admissions.

Counts are rounded to 5 in line with HES disclosure control.

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

Deaths for elective admissions

This indicator presents crude percentage rates of elective admissions where a death occurred either in-hospital or within 30 days (inclusive) of being discharged from hospital..

An adjustment is made for admission method in the SHMI methodology because crude mortality rates for elective admissions tend to be lower than crude mortality rates for non-elective admissions.

Counts are rounded to 5 in line with HES disclosure control.

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

Spells

Spells Number of spells in a 12 month period.

More detail on how spells are constructed can be found here https://files.digital.nhs.uk/92/4DEC97/Provider%20spells%20methodology.zip

Observed and expected

Count of deaths in hospital or within 30 days of discharge, and the expected deaths as produced by the SHMI model. The difference between the observed and expected is highlighted.

Counts are rounded to 5 in line with HES disclosure control.

To learn more go to: https://files.digital.nhs.uk/02/3013C8/SHMI%20specification%20v1.32.pdf

Conditions

SHMI by diagnosis condition

The SHMI is composed of 142 different diagnosis groups For a subset of diagnosis groups, an indication of whether the observed number of deaths within 30 days of discharge from hospital was ‘higher than expected’, ‘as expected’ or ‘lower than expected’ when compared to the national baseline is also provided. It is possible for conditions to have a statistically high or low SHMI, but if it is not in the subset it is not possible to identify these,

Due to the suppression applied by NHS Digital, the SHMI values may differ slightly compared to those published. The approach here has been to comprimse on precision and present a wider range of conditions.

To learn more go to: https://files.digital.nhs.uk/02/3013C8/SHMI%20specification%20v1.32.pdf

Outliers SHMI by condition

Using a funnel plot it is possible to determine whether there are other conditions that may be statistically elevated, especially where there may be not so many deaths. The table below lists the outliers, either high or low, based on the Poisson limits at 95% or 99.8% as highlighted in the funnel plot.

The funnel plot shows only those conditions where there was a death in the period.

Excess deaths by condition

This chart presents the changes in the SHMI and excess deaths over time. The larger the circle the greater the excess deaths, the darker the colour the higher the SHMI. The SHMI in this presentation is capped to 3 as small increases in the observed deaths can lead to large increases in conditions with low mortality, leading to exceptionally high and meaningless SHMI. Excess deaths are rounded to 5 in line with HES disclosure control.

Coding

Signs and symptoms

This indicator presents the percentage of finished provider spells with a primary diagnosis which is a symptom or sign (identified by ICD-10 codes beginning with the letter ‘R’).

A high percentage may indicate problems with data quality or timely diagnosis of patients, but may also reflect the case-mix of patients or the service model of the trust (e.g. a high level of admissions to acute admissions wards for assessment and stabilisation).

Depth of coding

This indicator presents the mean number of secondary diagnosis codes per finished provider spell (mean depth of coding) by elective and non-elective admission method, for each trust. A higher mean depth of coding may indicate a higher proportion of patients with multiple conditions and/or comorbidities, but may also be due to differences in coding practices between trusts.

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

Palliative care

Palliative care coding This indicator presents crude percentage rates of deaths reported in the SHMI with palliative care coding at either diagnosis or treatment specialty level.

Note that the SHMI makes no adjustments for patients who are recorded as receiving palliative care. This is because there is considerable variation between trusts in the way that palliative care codes are used.

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

Place of death

This indicator presents crude percentage rates of deaths reported in the SHMI which occurred in hospital and deaths reported in the SHMI which occurred outside hospital within 30 days (inclusive) of discharge.

To learn more go to: https://files.digital.nhs.uk/65/85E399/SHMI%20contextual%20indicator%20specifications.zip

Site

Trusts may be located at multiple sites and may be responsible for one or more hospitals.

The SHMI is calculated at the level of the provider spell, which is a continuous period of time spent as a patient within a single trust (provider). A spell may be composed of more than one episode (a single period of care under one consultant). If a patient is moved between hospitals or sites within the same trust, the provider spell continues. Most spells consist of a single episode and so there is no complication when presenting SHMI data at site level because the entire provider spell occurred at a single site. However, spells consisting of multiple episodes may have occurred over multiple sites and only one of these sites can be associated with the spell. This has been chosen to be the site of the first episode in the spell. This may result in hospital deaths being attributed to a site other than the one in which they occurred, with an impact on the SHMI values presented for the sites concerned. This impact is likely to be greater for sites within trusts showing higher percentages for this contextual indicator.

Deprivation

Deprivation is not included as a factor in the SHMI. It is interesting to reflect on how deprivation affects admission patterns and mortality for your trust. The quintiles are order from 1 the most deprived to 5 the most affluent.

Alternative views of SHMI

These charts show four alternative views of the SHMI, alongside the published value:

  • SHMI as published (SHMI)
  • SHMI for the 56 conditions in the HSMR (SHMI56)
  • SHMI for the conditions NOT in the HSMR (SHMIex56)
  • SHMI for those who died in hospital (SHMI in hospital)
  • SHMI for those who died outside hospital (SHMI outside hospital)

These are NOT official measures, and are included only to provide a view of how the SHMI compares to the patient cohorts used in the HSMR.

They are compared against the index value of 1 (dashed line).