Table 1 'Measuring Hospital Standardised Mortality Rates'

Measuring Hospital Standardised Mortality Rates
PSI 1: Measuring Hospital Standardised Mortality Rates
Origin: PSI by SimPatIE
Dimension Description
Description of Specific
Aspects of Patient Safety
Hospital Standardised Mortality Ration (HSMR) was first developed to
monitor the quality of care delivered. Yet, death is the ultimate harm to
a patient regarding patient safety. Thus HSMR is an adequate PSI
(1;2).
Level of Determination of
Patient Safety
Safety is assessed at the aggregated patient level.
Aim of the PSI The PSI aims at surveillance of institution-wide in-hospital mortality.
Source(s) HSMR was developed by B. Jarman, London.
Extent of Clinically
Testing
Since 1999 HSMR has been used in all National Health Service
hospitals in England, and the results have been published. The method
has also been used and tested in Sweden where it was applied to the
national patients’ registry. Registration of death is statutory, which
increases the data registrations and makes the registration specific.
By testing the method predictive differences in HSMR was found, they
were: the number of in-hospital doctors per 100 beds, number of GPs
per 100000 habitants in the area of interest, the number of acute
admissions, and the part of patients suffering from comobidity like
pneumonia and heart insufficiency (1).
Evidence of Clinically use
of Standards
No evidence of clinically use of standards was found.
PSI category Institution-Wide PSI.

of hospitalisation.
Data Source Administrative data.
Identifying the
institutional context
This PSI is relevant to quality improvement and accreditation.
Care Setting The PSI applies institution-wide.
Professionals Responsible
for Health Care
All authorised health care workers.
Lowest Level of Health
Care Delivery Addressed
Individual clinical units or departments.
Allowance for Patient
Factors
Age (10 years intervals), sex, comorbidity.
Stratification by
Vulnerable Populations
Age (10 years intervals), sex, elective/acute admission and total time of
hospitalisation.
Standard of Comparison No time frame of the comparison set. Hospital-wide surveillance.
Scoring Only the 85 primary admission diagnoses contributing to 80% of all inhospital
deaths are counted. All transfers between hospitals are
excluded.
To know about differences in primary illness and comorbidity the 15
most often discharge diagnosis (covering chronic diseases and acute
cause of admission) are to be found as the primary diagnosis related to
50% of all in hospital deaths (Please see (1) for further definition)
For each of the 85 included primary discharge diagnoses the yearly
cumulative mortality proportion (CMP) of the hospital is assessed, that
is dividing the number of deaths given a specific diagnoses by the total
number of admissions given the specific diagnosis stratified by age (10
years intervals), sex, elective/acute admission and total time of
hospitalisation.
The expected yearly CMP is assessed for each stratum multiplied by
the total strata specific CMP.
HSMR is assessed as the ratio of the observed versus the expected
CMP for each of the primarily discharge diagnosis multiplied by 100.
By means of stepwise regression analysis risk factors for differences in
HSMR are identified. The steps are: 1) general hospital data (e.g. part
of acute admitted patients 2) individual hospital data (e.g. number of
beds) 3) society related data (e.g. number of GPs per 100000 habitants
in the area). Please see (1) for further scoring advise.

of hospitalisation.
Data Source Administrative data.
Identifying the
institutional context
This PSI is relevant to quality improvement and accreditation.
Care Setting The PSI applies institution-wide.
Professionals Responsible
for Health Care
All authorised health care workers.
Lowest Level of Health
Care Delivery Addressed
Individual clinical units or departments.
Allowance for Patient
Factors
Age (10 years intervals), sex, comorbidity.
Stratification by
Vulnerable Populations
Age (10 years intervals), sex, elective/acute admission and total time of
hospitalisation.
Standard of Comparison No time frame of the comparison set. Hospital-wide surveillance.
Scoring Only the 85 primary admission diagnoses contributing to 80% of all inhospital
deaths are counted. All transfers between hospitals are
excluded.
To know about differences in primary illness and comorbidity the 15
most often discharge diagnosis (covering chronic diseases and acute
cause of admission) are to be found as the primary diagnosis related to
50% of all in hospital deaths (Please see (1) for further definition)
For each of the 85 included primary discharge diagnoses the yearly
cumulative mortality proportion (CMP) of the hospital is assessed, that
is dividing the number of deaths given a specific diagnoses by the total
number of admissions given the specific diagnosis stratified by age (10
years intervals), sex, elective/acute admission and total time of
hospitalisation.
The expected yearly CMP is assessed for each stratum multiplied by
the total strata specific CMP.
HSMR is assessed as the ratio of the observed versus the expected
CMP for each of the primarily discharge diagnosis multiplied by 100.
By means of stepwise regression analysis risk factors for differences in
HSMR are identified. The steps are: 1) general hospital data (e.g. part
of acute admitted patients 2) individual hospital data (e.g. number of
beds) 3) society related data (e.g. number of GPs per 100000 habitants
in the area). Please see (1) for further scoring advise.