Medicinska resultat i hälso- och sjukvården – några

Calculation of adjusted death
rates, HSMR
Experience in Sweden
Hans Rutberg, Senior medical adviser
Swedish National Board of Health and Welfare
Sweden
Inhabitants ~ 9 million
County Councils: 21
Hospitals ~ 70
ÖSTERGÖTLANDS LÄN
Linköping
Age, sex, admission method, los,
diagnosis, HSMRs - hospital standardised
mortality ratios 1995-2001 (99% CIs)
HSMRs (95% CIs) 1995/6 to 2000/1
140
120
100
80
60
40
20
Source: Sir Brian Jarman
TP
R
VL
R
NH
R
DE
R
EX
R
E7
R
JH
R
H
2
R
D3
Q
Q
R
NQ
R
R
JS
R
AJ
R
CC
R
TK
R
RK
R
DZ
R
VW
R
LZ
M
F
R
CD
R
R
EF
R
CF
R
BD
R
R
T3
0
Walsall change of HSMR
Observed – expected deaths (for top 80% all deaths) by 19 March 2005
= a reduction of 303 deaths (379 death if scaled to 100% deaths, 0.147/bed/year
reduction)
HSMR (with 95% confidence intervals)
150
First publication of HSMRs Jan 2001
Start of improvement interventions
140
130
120
110
100
90
80
70
60
50
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Source: Sir Brian Jarman
Walsall Hospital Standardised Mortality Ratio,
HSMR, England 2000-01distribution and Walsall 19/3/2005
Walsall 2000/1
120
100
80
60
40
Source: Sir Brian Jarman
RMC
RAS
RBN
RFW
RDU
RM4
RWD
RAP
RC9
RWP
RTG
RE9
RJ2
RVW
RWF
RJN
RM3
REM
RA9
RCD
RCC
RR2
RXC
RLQ
RJ7
RGT
RPR
RH8
0
RWG
20
RNJ
HSMRs 2000-01 English acute Trusts
Walsall 19 March 2005, but with wide CIs
140
Evaluation of HSMR in Sweden
• A project initiated by the National Board of Health and
Welfare, the Federation of County Councils and the
Southeast health care region.
• A national interest to use HSMR to compare hospitals and
county councils
• A possible indicator for comparison of the Nordic
countries?
Swedish Hospital Discharge Register started in the
1960:s. Since 1987 it covers all public, in-patient care in
Sweden
• Data on patient:
Personal id-number, sex, age, place of residence
• Data on hospital:
•
County council, hospital, department
Administrative data
Date of admission/discharge, LoS, acute/planned
admissions, admitted from, discharged to
•
Medical data
Main and secondary diagnoses, surgical procedures,
external cause of injury and poisoning
HSMR in Sweden
Variables:
• Sex
•
•
•
•
•
Age
Length of stay
Way of admission, transfer from other hospital
Acute/planned admission
Main diagnosis
HSMR in Sweden
Main diagnosis
• The main diagnoses accounting for
80 % of hospital mortality
• 58 diagnoses
• 370 000 admissions yearly, ~ 25 % of all
admissions
• 27 000 deaths annually
HSMR in Sweden
Discharged as dead
• Good quality
• 27 000 registrations per year
• 30 days mortality ??
Swedish deaths 2001 by CCS* group
(*Clinical Classification System. Elixhauser A, Andrews RM, Fox, S. Clinical classifications for health policy
research: Discharge statistics by principal diagnosis and procedure. Provider Studies Research Note 17.
Rockville, MD:
Agency for Health Care Policy and Research; 1993. AHCPR Pub. No. 93-0043. www.ahrq.org )
Acute cerebrovascular disease
Acute myocardial infarction
Congestive heart failure, nonhypertensive
Pneumonia (except that caused by tuberculosis or sexually
transmitted disease)
Cancer of bronchus, lung
Septicemia (except in labor)
Cancer of prostate
Cancer of colon
Cancer of pancreas
Aortic, peripheral, and visceral artery aneurysms
Cancer of breast
Secondary malignancies
Respiratory failure, insufficiency, arrest (adult)
Peripheral and visceral atherosclerosis
Cardiac arrest and ventricular fibrillation
Chronic obstructive pulmonary disease and bronchiectasis
Leukemias
Pulmonary heart disease
Non-Hodgkin's lymphoma
Coronary atherosclerosis and other heart disease
0
Source: Sir Brian Jarman
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
HSMR in Sweden
Length of stay,1998-2004
0-7 days
8-14 days
15-28 days
29-365 days
OR
0.52
0.48
0.71
1.00
95 % CI
0.51
0.53
0.47
0.49
0.70
0.72
--
HSMR in Sweden
1998-2004
• Transfer from other hospital
OR 95 % Ci
1,26 1,24-1,28
• Not planned
2,53 2,50-2,56
HSMR in Sweden
Admission data
Transfer from other
clinic, same hospital
Transfer from an
other hospital
Total number of
admissions 2003-2004
Sahlgrenska
1,2%
2,8%
212 706
Ryhov
4,4%
4,8%
56 700
Norrlands universitetssj
5,6%
7,6%
65 865
Linköping
5,4%
10,6%
83 700
Huddinge sjukhus
8,5%
6,2%
96 000
MAS
8,4%
2,2%
98 900
10,1%
43,5%
105 600
4,1%
8,0%
2 819 000
Example:
Akademiska sjukhuset
( Uppsala )
All hospitals
50
Universitetssjukhuset
Huddinge sjukhus
Danderyds sjukhus
Norrlands
Akademiska
Karolinska sjukhuset
Falu lasarett
Universitetssjukhuset
S:t Görans sjukhus
Södersjukhuset
Sundsvalls sjukhus
Kristianstads sjukhus
Ryhov, länssjukhus
SÄ-sjukvården
Sahlgrenska
Växjö lasarett
Sunderbyns sjukhus
Enköpings lasarett
Hässleholms sjukhus
Landskrona lasarett
NU-sjukvården
Östersunds sjukhus
Ljungby lasarett
Simrishamns sjukhus
Skaraborgs sjukhus
Värnamo sjukhus
Ludvika lasarett
Länssjukhuset
Kungälvs sjukhus
Örebro regionsjukhus
Mora lasarett
Blekingesjukhuset
Ängelholms sjukhus
Varbergs sjukhus
Västerviks sjukhus
Sala lasarett
Halmstads sjukhus
Arvika sjukhus
Gävle sjukhus
Visby lasarett
Säffle sjukhus
Bollnäs sjukhus
Vrinnevisjukhuset
Hudiksvalls sjukhus
Örnsköldsviks
Lindesbergs lasarett
Kalix lasarett
Karlstads sjukhus
Lycksele lasarett
Piteå Älvdals sjukhus
Avesta lasarett
Köpings lasarett
Trelleborgs lasarett
Norrtälje sjukhus
Södertälje sjukhus
Härnösands sjukhus
Universitetssjukhuset
Västerås lasarett
Mälarsjukhuset
Nyköpings sjukhus
Helsingborgs lasarett
Hälsinglands sjukhus
Ystads lasarett
Alingsås lasarett
Oskarshamns sjukhus
Kullbergska sjukhuset
Kristinehamns
Torsby sjukhus
Kiruna lasarett
Skellefteå lasarett
Sollefteå sjukhus
Gällivare lasarett
Motala lasarett
Nacka sjukhus
Höglandssjukhuset
Karlskoga lasarett
Fagersta lasarett
HSMR (with 95% confidence intervals)
Swedish preliminary HSMRs 1998-2004
(some exclusions will be necessary)
190
170
150
130
110
90
70
Swedish preliminary county HSMRs 19982004 (some exclusions will be necessary)
HSMR, County Council, The Swedish Hospital Discharge Register,1998-2004.
OR
95 % CI
Stockholm
0,85
0,84
0,86
Uppsala
0,87
0,84
0,89
Dalarna
0,91
0,89
0,94
Jönköping
0,93
0,90
0,95
Västerbotten
0,96
0,94
0,99
Östergötland
0,98
0,95
1,00
Halland
0,98
0,95
1,01
Västra Götaland
0,98
0,97
1,00
Kronoberg
1,00
0,97
1,04
Jämtland
1,02
0,98
1,05
Kalmar
1,04
1,01
1,06
Gotland
1,05
1,00
1,12
Västernorrland
1,06
1,03
1,08
Skåne
1,06
1,04
1,07
Blekinge
1,09
1,05
1,13
Norrbotten
1,09
1,06
1,12
Gävleborg
1,14
1,11
1,17
Värmland
1,15
1,12
1,18
Örebro
1,18
1,15
1,21
Södermanland
1,18
1,15
1,21
Västmanland
1,21
1,18
1,25
OR
95 % CI
Stockholm
0,85
0,84
0,86
Dalarna
0,90
0,88
0,93
Östergötland
0,95
0,92
0,98
Kronoberg
0,95
0,92
0,99
Halland
0,96
0,93
0,98
Västra Götaland
0,98
0,96
0,99
Jämtland
0,98
0,95
1,02
Västerbotten
0,99
0,96
1,02
Jönköping
1,00
0,97
1,03
Gotland
1,02
0,96
1,08
Västernorrland
1,04
1,01
1,06
Skåne
1,04
1,02
1,05
Uppsala
1,04
1,01
1,07
Norrbotten
1,05
1,02
1,08
Kalmar
1,06
1,03
1,09
Gävleborg
1,08
1,06
1,11
Värmland
1,09
1,06
1,12
Blekinge
1,09
1,05
1,13
Västmanland
1,17
1,14
1,21
Örebro
1,20
1,17
1,23
HSMR, County Council,
compensated for transfer
within hospital
HSMR in Sweden
Data quality problem:
• Transfer within hospital (1,2 - 10,1%)
• Transfer between hospitals (2,2 - 43,5%)
• Underreporting planned/not planned
• Large vs. small hospitals?
HSMR in Sweden
Future steps:
• Present the method and publish preliminary results on county council
level in the Swedish Medical Journal
• Discuss data quality with those responible on county council level
• Consider to use HSMR as a national indicator
• Pilot project in the South-east healthcare region on structured patient
record review on hospital deaths with IHIs Trigger Tool