Coronary revascularisation in Spain - ECHO · 2017-01-24 · Exposure to surgical revascularization...
Transcript of Coronary revascularisation in Spain - ECHO · 2017-01-24 · Exposure to surgical revascularization...
Coronary
revascularisation
in Spain
February 2014
García-Armesto S, Comendeiro-Maalooe M, Martinez-Lizaga N, Angulo-Pueyo E, Bernal-Delgado E on behalf of the ECHO consortium
ECHO Atlas on Coronary Revascularisation EUROPEAN COLLABORATION FOR HEALTHCARE OPTIMIZATION (ECHO)
Edited by
Seral Rodríguez M., García Armesto S., Bernal Delgado E. Instituto Aragonés
de Ciencias de la Salud- Instituto de Investigación Sanitaria Aragón.
Acknowledgment and disclaimer
The ECHO Consortium appreciates the key role of the national institutions
that kindly provided the data used in this research project.
ECHO Consortium strives to keep the content of this Atlas accurate according
to rigorous professional standards. Their institutions do not necessarily share
the contents of this report, which is entirely the responsibility of the authors.
Funding
The research leading to these results has
received funding from the European
Community's Seventh Framework Programme
(FP7/2007-2013). Sole responsibility lies with
the authors. The EC is not responsible for any
use that might be made of the information
contained there in.
This ECHO Atlas has been elaborated by the Institute for Health Sciences
In Aragon in partnership with the following organisations:
-
This publication should be cited as
European Collaboration for Healthcare Optimization (ECHO) Project. www.echo-
health.eu . Zaragoza (Spain): Instituto Aragonés de Ciencias de la Salud - Instituto
Investigación Sanitaria Aragón; c2010. García-Armesto S, Comendeiro-Maalooe
M, Martinez-Lizaga N, Angulo-Pueyo E, Bernal-Delgado E on behalf of the ECHO
consortium . ECHO Atlas on Coronary Revascularisation; February 2014
[accessed: date]; Available from: www.echo-health.eu/echo-atlas-reports
More details and ECHO Atlases available at www.echo-health.eu
© ECHO Consortium. Reproduction, distribution, transmission, re-publication,
display or performance, of the Content delivered in this ECHO Atlas is only
allowed for non-commercial purposes and shall acknowledge the source.
1
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
I. EXECUTIVE SUMMARY
This report analyses the magnitude and the variation of the surgical
management of ischaemic coronary disease. The analysis is twofold: on the
one hand it examines population exposure to revascularisation surgery, and
on the other, evaluates hospital quality in terms of hospital differences in case
fatality rates.
Percutaneous Coronary Intervention (PCI, commonly known as coronary
angioplasty) and Coronary Artery Bypass Graft (CABG) are effective and safe
revascularization procedures that have improved survival and quality of life in
recent decades. PCI has been proven to be a better option to reduce the risk
of death, especially when few blood vessels are blocked/affected. In
particular, primary PCI supersedes all other alternatives. Nevertheless, CABG
is still considered more effective when dealing with multivessel disease
(involving 3 or more vessels).
Exposure to surgical revascularization
Spain registered 78,585 CID hospitalisations in 2009 (1 admission per 485
adult inhabitants). This Figure was among the lowest of all ECHO countries.
There was a 2.6-fold difference between healthcare areas with the highest
and lowest rates of CID admissions, and systematic variation was moderate,
at 10% higher that randomly expected. More than half of all CID admissions
were labelled as AMI. For these admissions there was an almost 3-fold
difference between healthcare areas with the highest and lowest rates (EQ5-
95).
In the same year, 48,368 PCI interventions and 7,068 CABG surgeries were
recorded. These figures were among the lowest of all ECHO countries. The
PCI rate was similar to that of Portugal and less than half the rate of Slovenia,
the country with the highest rate. Spain´s CABG rate was the lowest of all
ECHO countries, and 3 times lower than that of Denmark.
The differences between healthcare areas with the highest and lowest rates
were 4.5-fold and almost 9.3-fold for PCI and CABG, respectively. Variation
not deemed random was moderate in both cases: 19% and 22% higher than
Mortality and morbidity
from cardiovascular disease
are considered a public
health issue. In fact,
coronary ischaemic disease
is one of the leading causes
of death in Europe.
The study of systematic
variation in its surgical
management, and
associated outcomes, offers
a critical view on how
healthcare organizations
provide care to patients.
2
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
expected, respectively. For both PCI and CABG, region accounted for over
40% of the observed variation, which suggests a relevant role of regions in
modulating the provision of this intervention.
The geographical approach examined the mismatch between patterns of
burden of coronary ischaemic disease (CID) and the intensity of use of
revascularization procedures. Marked variation in exposure to
revascularisation interventions was observed across healthcare areas.
Moreover, this variation was largely unaffected by the burden of ischemic
disease.
At the regional level an inverse relationship between CABG and PCI
procedures was detected. This may be due to early adopting regions in which
progressively higher levels of PCI led to a decrease in CABG utilisation.
From 2002 to 2009, coronary ischaemic disease admissions decreased slightly
(by 11%), from 1 admission per 392 to 1 admission per 429 adult inhabitants.
Of these hospitalisations, the number corresponding to AMI remained almost
constant.
During the same period, PCI utilisation rates increased by 75%, from 1
admission per 791 to 1 admission per 413 inhabitants. The CABG rate
remained stable over the same period (from 1 admission per 2,899 to 1
admission per 2,857 inhabitants aged 40 or older). Heterogeneous exposure
was observed across the territory for both interventions, as evidenced by
constant, moderate systematic variation over time.
From 2002 to 2009, significantly more CID admissions were recorded in more
deprived healthcare areas than in wealthier ones. Specifically, AMI
admissions were more frequent in worse-off versus better-off areas, with the
gap between the lowest and highest quintiles widening over time. This
finding may reflect a proper response to the population’s needs.
PCI utilisation increased at all income levels, but most of all in more deprived
areas, leading to a significantly higher PCI utilisation rate in those areas.
Conversely, CABG exposure was higher in wealthier areas, and remained
stable in all quintiles.
3
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Differences in hospital case-fatality rates
Differences in the risk-adjusted case fatality rates (CFR) after both
revascularisation procedures were marked, with huge variation across
hospitals. “Volume” (number of interventions carried out) has been proposed
as a possible contributor to these differences.
After a sharp decrease, the Spanish risk-adjusted CFR for AMI in 2009 was
93.75 per 1,000 patients aged 18 and older, the lowest rate of all ECHO
countries, and 5.3 per thousand points below the ECHO average. In terms of
exposure, only 5.2% of all Spanish AMI patients were treated at poorly
performing hospitals, the lowest proportion of patients of all ECHO
countries. Almost 40% of AMI patients were admitted to hospitals flagged as
good or even excellent performers, a proportion slightly higher than the
ECHO average.
In-hospital mortality after PCI in Spain underwent a net increase of 5 per
thousand points. In 2009, the risk-adjusted CFR was 25.6 per 1,000 patients
aged 40 and older, by far the highest of all ECHO countries, and almost 6 per
thousand points above the ECHO average. Furthermore, 34% of patients
undergoing PCI were treated at alarm hospitals, while only 4% of patients
underwent interventions at hospitals flagged as good performers (the lowest
share for this procedure of all ECHO countries).
Although in-hospital mortality after CABG appeared to decrease markedly,
Spain´s risk-adjusted CFR after CABG surgery in 2009 was by far the highest
of all ECHO countries, at 66 per 1,000 patients aged 40 and older. This was
over twice the English rate and 16 per thousand points above the ECHO
average. Only 21% of all Spanish patients undergoing CABG surgery in 2009
underwent interventions at high-volume centres (more than 250 procedures
per year), by far the lowest proportion of all ECHO countries, and 60
percentage points below the ECHO average. Moreover, 20.8% of patients
underwent interventions at alert/alarm hospitals, by far the highest
proportion of all ECHO countries.
4
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
All through this section, paired dot plots are used to show results. The chart on the
left is always intended to give the reader a sense of the magnitude of burden of
disease or utilisation of revascularisation procedures in each country; the image on
the right provides an idea of the actual variation comparable across countries.
Note that each dot represents the relevant health care geographic unit in each
country.
II. INTERNATIONAL COMPARISON
This chapter offers a perspective on the behaviour of Spain as compared with the
other ECHO countries when it comes to ischaemic coronary disease and its
clinical management and treatment. The analysis is two-fold:
a. Geographic approach: this compares the population burden of disease
and the exposure to treatment depending on place of residence (both
the magnitude and the within-country variation) .
b. Hospital approach: this examines the quality of hospital care in terms of
case fatality rates for patients with acute myocardial infarction (AMI) and
for the procedures of choice for this condition. These outcomes are used
to benchmark all hospitals across the ECHO region, providing an overview
of the performance of Spanish hospitals relative to those of other ECHO
countries.
a. Geographic approach
This section provides a broad overview of the incidence of coronary ischaemic
disease (CID) and AMI admissions, taken as a proxy of burden of coronary
disease. It also examines the intensity of use of alternative revascularization
procedures in Spain as compared with other ECHO countries.
The geographic approach focuses on population exposure. The key issue
addressed by this analysis is the correlation between the risk of coronary disease
and access to revascularisation procedures, depending on an individual´s place of
residence.
The cross-country
comparison of the
geographical distribution of
population exposure to
burden of disease and to
intensity of use of
procedures allows the
flagging of situations of
over- and under-use of
revascularisation.
The benchmarking of
hospital case fatality rates
provides an additional
perspective on the quality
and safety of the care
provided and its variation
within each country.
Accounting for specific
organisational features, the
international comparison
provides a wider
perspective.
5
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Coronary Ischaemic Disease (CID)
In 2009, Spain recorded the fourth CID admission rate among ECHO countries, at
1 admission per 420 adult inhabitants. This was 33% higher than the
corresponding rate in Portugal, the country with the lowest rate, and 30% lower
than the corresponding rate in England, the country with the highest rate (see
Table 1 in Appendix 1a).
In Spain, residents living in areas with the highest rates were three times more
likely to be admitted for CID than those living in areas with the lowest rates. This
difference was about 2-fold in Denmark, Slovenia, England and Portugal.
Systematic variation not deemed random was moderate to low in all countries,
ranging from 9% higher than expected(Slovenia) to 24% higher than expected
(England).
Figure 1a. Age and sex-standardised CID hospitalisation rates per 10,000 inhabitants (natural scale to compare actual rates).
Year 2009
Figure 1b. Age and sex-standardised CID hospitalisation rates per 10,000 inhabitants (normalised scale to compare degree of
variation). Year 2009
Each dot represents the relevant healthcare administrative area in each ECHO country (Healthcare areas in Spain). The y-axis charts the standardised rate per 10,000 inhabitants (age 18+) for the administrative area. The figure is generated based on the total number of CID hospitalisations recorded in 2009 in ECHO countries. In Figure 1b admission rates have been normalised to facilitate comparison of the degree of variation across countries
6
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Acute Myocardial Infarction (AMI)
AMI admission rates were similar to those of other ECHO countries: the Spanish
rate was the lowest (1 hospitalisation per 726 adults), close to that of Portugal.
Slovenia recorded the highest rate, at 1 admission per 449 adult inhabitants,
followed by England, at 1 per 597 adults. Differences between areas with the
highest and lowest rates of AMI hospitalisations were around 2-fold in all ECHO
countries.
Variation not attributable to chance was low to moderate, except in Slovenia
where it reached 34% higher that randomly expected. In Spain 11% of variation
exceeded that which could be randomly expected (see Table 2 in Appendix 1a).
Figure 2a. Age and sex-standardised AMI hospitalisation rates per 10,000 inhabitants (natural scale to compare actual rates).
Year 2009
Figure 2b. Age and sex-standardised AMI hospitalisation rates per 10,000 inhabitants (normalised scale to compare degree of variation).
Year 2009
Each dot represents the relevant healthcare administrative area in each ECHO country (Healthcare areas in Spain). The y-axis charts the standardised rate per 10,000 inhabitants (age 18+) for the administrative area. The figure is generated based on the total number of AMI hospitalisations recorded in 2009 in ECHO countries. In Figure 2b admission rates have been normalised to facilitate comparison of the degree of variation across countries.
7
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Percutaneous Coronary Interventions (PCI)
Spain recorded the lowest PCI rate among ECHO countries, at 1 admission per
419 inhabitants aged 40 or older, close to that of Portugal, the country with the
lowest rate. This was half the rate recorded in Slovenia, the country with the
highest rate. Despite its low PCI rate, the difference recorded between
healthcare areas with the highest and lowest rates was close to 5-fold, indicating
marked differences in PCI utilisation across the Spanish territory. Slovenia,
England and Portugal showed similar differences between healthcare areas with
the highest and lowest rates (2.2 to 2.6-fold).
Systematic variation ranged from just 8% higher than expected by chance in
England and Portugal to 1.8 times higher than expected in Slovenia. In Spain this
value was close to the ECHO average, at 22% higher than expected by chance
(see Table 3 in Appendix 1a).
Figure 3a. Age and sex-standardised PCI utilisation rates per 10,000 inhabitants (natural scale to compare actual rates).
Year 2009
Figure 3b. Age and sex-standardised PCI utilisation rates per 10,000 inhabitants (normalised scale to compare degree of
variation). Year 2009
Each dot represents the relevant healthcare administrative area in each ECHO country (Healthcare areas in Spain). The y-axis charts the standardised rate per 10,000 inhabitants (age 40+) for the administrative area. The figure is generated based on the total number of PCI procedures recorded in 2009 in ECHO countries. In Figure 3b admission rates have been normalised to facilitate comparison of the degree of variation across countries.
8
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Coronary Artery Bypass Grafting (CABG)
Spain recorded the lowest CABG rate among ECHO countries, at 1 admission per
2959 inhabitants aged 40 or older. This was 3 times less the rate recorded in
Denmark, the country with the highest rate.
Conversely, at the local level, the difference between areas with the highest and
lowest CABG rates was the greatest in all ECHO countries, at close to 10-fold. In
Denmark and England, this difference was around 2-fold.
The systematic part of this variation was high in all countries, reaching 27%
higher that randomly expected in Spain (see Table 4 in Appendix 1a).
Figure 4a. Age and sex-standardised CABG utilisation rates per 10,000 inhabitants (natural scale to compare actual rates).
Year 2009
Figure 4.b. Age and sex-standardised CABG utilisation rates per 10,000 inhabitants (normalised scale to compare degree of
variation). Year 2009
Each dot represents the relevant healthcare administrative area in each ECHO country (Healthcare areas in Spain). The y-axis charts the standardised rate per 10,000 inhabitants (age 40+) for the administrative area. The figure is generated based on the total number of CABG interventions recorded in 2009 in ECHO countries. In Figure 4b intervention rates have been normalised to facilitate comparison of the degree of variation across countries.
9
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Funnel plots allow the assessment of the performance of individual hospitals
against the international benchmark. Each hospital (dot) is charted by its risk-
adjusted case fatality rate and the volume of patients or procedures per year. The
benchmark is built based on the average ECHO hospital CFR (risk-adjusted) and its
95% and 99% CIs. The solid grey line represents the ECHO CFR, red lines correspond
to the 95% confidence interval control limits and the dashed blue lines represent to
the 99% confidence interval limits. These thresholds represent the boundary
between expected variation in outcomes (not significantly different from the
average) and significant variation. Outcomes lying beyond the upper thresholds are
indicative of poor performing hospitals (alert or alarm position); outcomes below
the lower limits indicate good or excellent performing hospitals. Outliers (in either
direction) should be investigated further to identify the underlying factors and
should be addressed or used as examples of good practice.
b. Hospital approach
The analysis in the following section focuses on providers, benchmarking for 3
quality outcome indicators. The main parameters of interest are the actual value
of the hospital case-fatality rate (CFR), and the relative position compared to the
ECHO benchmark and its confidence interval limits (95 and 99% levels), built into
a funnel plot. This relative position allows for the classification of hospital
performance as average, good, excellent, alarm and alert.
ECHO benchmarks are generated based on the expected average behaviour,
using data from all hospitals in the 5 countries analysed (multilevel regression
modelling). All CFRs are risk-adjusted for sex, age, severity of the underlying
condition and co-morbidity (Elixhauser index). This way, differences across
providers should not be attributable to patient characteristics, affecting their
inherent probability of dying after admission or surgery (Appendix 4 provides
details about the variables included in the risk-adjustment procedure).
Hospitals with less than 30 patients or procedures/year have been excluded from
the analysis to avoid noise when modelling (Table 5, Appendix 1b, details the
number of hospitals, per indicator, excluded under this criterion, and the
percentage of treated patients). In fact, the amount of interventions conducted
at each hospital, or so-called "volume", is one of the significant explanatory
variables when analysing the risk-adjusted CFR, and has been proposed as a
potential factor contributor to the observed differences in rates across hospitals.
The threshold for high and low-volume hospitals has been empirically set at 250
patients or procedures/year.
Different healthcare
systems across Europe,
with different
organizational features,
may obtain different
outcomes in dealing with
ischaemic coronary
disease. Comparing
outcomes across
individual hospitals in
each country provides
insights as to where
interventions could be
aimed in order to improve
case fatality rates for
patients with coronary
conditions.
International comparison
adds a complementary
perspective to the usual
national-based
benchmarks.
10
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
In-hospital mortality: Acute Myocardial Infarction (AMI).
In-hospital risk-adjusted CFR per 1,000 AMI patients (urgent admission of
patients aged 18 or older) is a widely used indicator of the quality and safety of
hospital care.
In 2009 in the ECHO area, 146,859 hospital admissions of patients aged 18 or
older were flagged as AMI. Of these, 12,582 died. After risk-adjusting modelling,
the ECHO average CFR was 99.03 per 1,000 hospitalised patients, which means
that 1 in 10 AMI admissions resulted in death.
In Spain in 2009, 1 in every 10.7 AMI patients admitted to a hospital died (risk-
adjusted CFR, 93.75 per 1,000), the lowest rate among ECHO countries, at 5.3 per
thousand points below the ECHO average.
The total number of ECHO hospitals analysed was 435; 55% of these were flagged
as high volume hospitals (more than 250 AMI patients in a year) and cared for
82.5% of the total number of hospitalised AMI patients.
In 2009, 82 out of 202 centres were classed as high volume hospitals, and cared
for 70.6% of all hospitalised AMI patients. In fact, Spain, together with Denmark,
had the second lowest percentage of AMI patients treated at high volume
hospitals in the ECHO area.
Of the 202 Spanish centres, 15 were flagged as alert or alarm performers in
terms of adjusted CFR. In terms of exposure, 5.2% of all AMI patients were
treated at these alert/alarm hospitals, the lowest percentage among all ECHO
countries. Nonetheless, 39.82% of all AMI patients were admitted to hospitals
classed as good or even excellent performers (see Table 6, Appendix 1b, for
further details).
Figure 5 shows the risk-adjusted CFR for each of the ECHO hospitals. Their
position relative to the ECHO benchmark is indicated in the funnel plot.
11
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
The outcomes shown in the funnel plot indicate good performance; 67% of
hospitals were close to the average, indicating that the risk-adjusted in-hospital
mortality did not differ significantly from the ECHO benchmark.
All but one of the Spanish hospitals flagged as alarm or alert treated less than 250
AMI patients per year.
In-hospital mortality after Percutaneous Coronary Intervention (PCI)
In 2009, 132,737 patients aged 40 or older underwent a PCI procedure at one of
the ECHO country hospitals. Of these, 2,623 died (1 per 51 intervened patients).
The ECHO risk-adjusted CFR was 19.86 per 1,000 patients (age 40+) undergoing a
PCI procedure. That year, Spain had by far the highest risk-adjusted CFR, at 5.7 per
thousand points above ECHO benchmark, and almost 12 per thousand points
above the CFR of England, the country with the lowest rate.
Within the ECHO framework, 80% of the hospitals performing PCI procedures
were high volume and cared for 95.44% of patients undergoing that procedure. In
Figure 5. In-hospital case fatality rate for AMI admissions across hospitals in ECHO countries. Year 2009.
Each dot represents an ECHO hospital that treated more than 30 AMI cases in that year. The expected number of deaths per 1,000 hospitalised patients is based on the average calculated across ECHO hospitals. Spanish hospitals are indicated in blue.
12
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Spain the corresponding proportion was 94.5% (see Tables 5 and 7 in Appendix
1b).
Contrary to expectations, the bulk of Spanish alarm hospitals performed more
angioplasties and had worse risk-adjusted case fatality rates than those
performing fewer interventions. All but one of those alert/alarm hospitals were
classed as high volume, and cared for almost 34% of all patients, while only a 4%
of patients underwent interventions at hospitals flagged as good performers, the
lowest share of all ECHO countries (See Table 7, Appendix 1b, for further details).
Figure 6. In-hospital case fatality rate after Percutaneous Coronary Intervention across hospitals in ECHO countries. Year 2009.
Each dot represents an ECHO hospital that performed more than 30 PCIs in that year. The expected number of deaths per 1,000 hospitalised patients is based on the average calculated across ECHO hospitals. Spanish hospitals are indicated in blue.
13
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
In-hospital mortality after Coronary Artery Bypass Graft (CABG)
In 2009, the 89 ECHO hospitals performing CABG surgery performed
interventions in 33,683 patients aged 40 and older. Of these, almost 4% died. In
terms of risk-adjusted CFR, this death rate represents 1 in every 20 patients
undergoing the procedure. More than half of those 89 centres were categorised
as "high volume", and conducted 82.16% of all CABG procedures performed that
year in ECHO countries.
It is worth highlighting that 61.26% of all patients were treated at hospitals
flagged as alert/alarm, while 5.61% were treated at hospitals flagged as
good/excellent performers
The percentage of Spanish patients undergoing CABG surgery who were treated
at higher volume hospitals was 20.9%. None of these hospitals were flagged as
excellent performers, while 26% of hospitals performing CABG were flagged as
alert or alarm performers.
The risk-adjusted case fatality rate after CABG shown in Figure 7, and indicates
poor performance for Spain in 2009. All but two of the alert/alarm hospitals in
this international comparison are Spanish. Relative to the ECHO benchmark, the
Spanish risk-adjusted CFR for CABG was by far the highest, at 15.7 per thousand
points above the ECHO average. This is over twice rate recorded in England, the
country with the lowest rate.
14
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 7. In-hospital case fatality rate after CABG across hospitals in ECHO countries. Year 2009.
Each dot represents an ECHO hospitals that performed more than 30 bypass surgeries in that year. The expected number of deaths per 1,000 hospitalised patients is based on the average calculated across ECHO hospitals. Spanish hospitals are indicated in blue.
15
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
III. IN COUNTRY VARIATION
In this section, the incidence of coronary ischaemic disease and the intensity of
use of alternative revascularization procedures performed in Spain is analysed
from an internal perspective, comparing results in healthcare-relevant
administrative areas (geographic approach) or hospitals (providers approach)
within the country.
The analysis is two-fold, and follows the same structure as described in the
previous chapter:
a. Geographic approach: this compares the burden of disease and the
intensity of exposure to treatment of the population, depending on the
place of residence (i.e., both the magnitude and the within-country
variation across local authorities (áreas sanitarias) and regions.
b. Hospital approach: this examines the quality of hospital care in terms of
case fatality rates for patients with acute myocardial infarction (AMI) and
for the procedures of election in those cases. These outcomes are used
to benchmark individual Spanish hospitals
a. Geographic approach
The magnitude and the variation in coronary conditions and/or revascularization
procedures across the country is mapped out for two healthcare-relevant
administrative tiers: 199 Healthcare Areas and 17 Regions or Comunidades
Autónomas. While healthcare area would represent local provision of care,
regions are used as a surrogate for regional policies affecting all the healthcare
areas they encompass.
Coronary Ischaemic Disease admissions (CID)
In 2009, Spain recorded 78,585 CID admissions, 1 admission per 485 Spanish
adult inhabitants.
There was a 2.6-fold difference in CID admissions between healthcare areas with
the highest and lowest rates. Systematic variation was just 10% higher that
randomly expected, and region accounted for 28% of variation not accounted for
by healthcare areas (see Tables 9 and 10 in Appendix 2a).
CID admissions are
considered a proxy of the
burden of cardiovascular
disease at the geographical
level.
In the ECHO framework
this indicator is used as a
“calibrator” and aids the
interpretation of results
about the intensity of
population exposure to
revascularization
procedures (coronary
artery bypass graft and
percutaneous coronary
intervention).
16
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 10. Age and sex-standardised CID hospitalisation rate per 10,000 inhabitants by regions. Year 2009
Figure 11. Observed/expected CID admissions ratio by regions. Year 2009
Maps on the left (standardised rates) show the number of admissions flagged as CID; the darker the colour, the higher the number of admissions (always per
10,000 adult inhabitants). Areas are clustered into 5 quintiles according to their rate value (Q1 to Q5). Legend indicates the range of standardised rates within each quintile. Maps on the right represent the relative risk of hospitalization for CID in each area using the observed to expected ratio of CID hospitalisation as a proxy. Populations living in areas with values >1 (bluish) are overexposed to the risk of CID hospitalisation, while those in areas with a ratio <1 (pink) are underexposed.
Figure 8. Age and sex-standardised CID hospitalisation rate per 10,000 inhabitants by healthcare areas. Year 2009
Figure 9. Observed/expected CID admissions ratio by healthcare areas. Year 2009
17
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Healthcare areas with high CID admission rates were located in the southern half
of Spain (Figure 8). Residents in most of these areas had a risk of CID admission
that was at least 20% above the national average (bluish areas in Figure 9). By
contrast, local authorities with low rates, in which residents had a lower risk of
CID hospitalisation, were located in the north-eastern part of the country.
At the regional level, the risk of hospitalization for CID was higher than the
national average for residents in Andalucia, Murcia and Asturias (dark blue areas
in Figure 11). Populations living in País Vasco, Navarra and Madrid had the lowest
risk of CID admissions in the territory (purple areas in Figure 11).
Percutaneous Coronary Interventions (PCI) compared with burden
of Coronary Ischaemic Disease (CID)
During 2009, Spain recorded 48,368 PCI interventions; 1 procedure per 495
inhabitants aged 40 or older.
There was an over 4-fold difference in exposure to the procedure between
healthcare areas with the highest and lowest rates. Moreover systematic
variation was 19% higher than randomly expected, and region explained up to a
44% of this variation, which suggests a role of regions in modulating the provision
of this intervention (see Tables 9 and 10 in Appendix 2a).
Some overlap between the intensity of PCI utilisation and the risk of CID
admission could be expected, given that CID admission is a proxy of the burden of
coronary disease. However, burden of ischemic disease only accounted for 16%
of PCI variation across healthcare areas. At the regional level, some correlation
was observed in the Andalucia, Murcia, Extremadura and Castilla-La Mancha
regions (where high PCI rates coincided with an increased risk of CID admission)
and in País Vasco, Navarra, Aragón, Castilla-León (where low PCI rates coincided
with a reduced risk of CID admission). However, residents in Madrid, Cataluña,
Galicia and Canary Island had high PCI rates together with a low risk of CID
admission, while Asturias showed an increased risk of CID admission together
with low exposure to PCI (Figures 14 and 15).
18
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 14. Age and sex-standardised PCI utilisation rate per 10,000 inhabitants by regions. Year 2009
Figure 15. Observed/expected CID admissions ratio, by regions. Year 2009
Maps on the left (standardised rates) show the number of admissions flagged as PCI; the darker the colour, the higher the number of procedures performed, per 10,000 inhabitants aged 40+. Areas are clustered into 5 quintiles according to their rate value (Q1 to Q5). Legend indicates the range of standardised rates within each quintile. Maps on the right represent the relative risk of hospitalization for CID in each area using the observed to expected ratio of CID hospitalisation as a proxy. Populations living in areas with values >1 (bluish) are overexposed to the risk of CID hospitalisation, while those in areas with a ratio <1 (pink) are underexposed.
Figure 12. Age and sex-standardised PCI utilisation rate per 10,000 inhabitants by healthcare areas. Year 2009
Figure 13. Observed/expected CID admissions ratio, by healthcare areas. Year 2009
19
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Coronary Artery Bypass Graft (CABG) compared with burden of
Coronary Ischaemic Disease (CID)
In 2009, Spain recorded 7,068 CABG procedures; 1 surgery per 3,378 inhabitants
aged 40 or older.
There was a 9.3-fold difference in exposure to the procedure between healthcare
areas with the highest and lowest rates, and 22% of this variation could not be
deemed random. As seen for PCI utilisation, variation in CABG surgery was highly
influenced by regions; up to 43% of the observed variation was explained by a
regional effect (see Tables 9 and 10 in Appendix 2a).
A certain pattern of high-rate healthcare areas was observed in the north-
western part of the country. With some exceptions, CABG utilisation did not
correlate with burden of disease (Figure 16 and 17). At the regional level, CABG
procedures and the risk of CID hospitalisation appeared to be inversely related,
with the exception of Asturias, whose residents had a higher risk of CID
admission but which had one of the highest CABG rates in the country. In general
however, lower CABG rates were observed in regions with increased risk of CID
hospitalisation, such as Andalucia, Murcia and Castilla-La Mancha (Figures 18 and
19).
20
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 18. Age and sex-standardised CABG utilisation rate per 10,000 inhabitants by regions. Year 2009
Figure 19. Observed/expected CID admissions ratio by regions. Year 2009
Maps on the left (standardised rates) show the number of admissions flagged as CABG; the darker the colour, the higher the number of procedures performed, per 10,000 inhabitants aged 40+. Areas are clustered into 5 quintiles according to their rate value (Q1 to Q5). Legend indicates the range of standardised rates within
each quintile. Maps on the right represent the relative risk of hospitalization for CID in each area using the observed to expected ratio of CID hospitalisation as a proxy. Populations living in areas with values >1 (bluish) are overexposed to the risk of CID hospitalisation, while those in areas with a ratio <1 (pink) are underexposed.
Figure 16. Age and sex-standardised CABG utilisation rate per 10,000 inhabitants by healthcare areas. Year 2009
Figure 17. Observed/expected CID admissions ratio by healthcare areas. Year 2009
21
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Percutaneous Coronary Interventions (PCI) vs. Coronary Artery
Bypass Graft (CABG)
PCI and CABG are effective and safe revascularization procedures that have
improved survival and quality of life in recent decades. PCI has been proven to be
the best option at reducing the risk of death, particularly when the number of
affected blood vessels is low (primary PCI has superseded all other alternatives).
However, CABG is still considered more effective when dealing with multivessel
disease (3 or more vessels involved).
These procedures can be considered as distinct interventions with different
clinical indications, or, alternatively, as “substitute” approaches for the same
clinical condition. Therefore, analysis of their respective patterns of utilisation
provides some indication as to how populations are being served. Trends in the
same direction for both procedures would rule out the “substitution” hypothesis,
whereas opposing patterns would suggest a certain degree of compensation
across procedures, although the two procedures are not equally effective.
In Spain, although a substitution effect was observed at the regional level, this
phenomenon was near-negligible at the healthcare area level (Figures 22 and 23),
as evidenced by a negative correlation of -0.09.
22
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 21. Observed/expected CABG utilisation ratio by healthcare areas. Year 2009
Figure 20. Observed/expected PCI utilisation ratio by healthcare areas. Year 2009
Figure 22. Observed/expected PCI utilisation ratio by regions. Year 2009
Figure 23. Observed/expected CABG utilisation ratio by regions. Year 2009
Maps represent the level of performance of each area, using the “observed to the expected” ratio of the number of revascularisation procedures as a proxy of the risk of cardiovascular intervention. Residents living at areas with values >1 (bluish) are overexposed to the risk of certain cardiovascular interventions, while those in areas with a ratio <1 (pink) are underexposed to the risk of these cardiovascular interventions.
23
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
b. Hospital approach
The following sections will deal with in-hospital case fatality rates (CFR) across
Spanish hospitals after admission for Acute Myocardial Infarction (AMI) and after
one of the following revascularization procedures; percutaneous coronary
intervention (PCI) and coronary bypass surgery (CABG).
When analysing data on a provider basis, it should be noted that a large
proportion of the observed variation in outcomes may be due to different meso
and micromanagement arrangements used to address coronary ischaemic
disease.
In-hospital case fatality rate for Acute Myocardial Infarction patients
In 2009, 52,683 admissions in 271 Spanish hospitals were flagged as Acute
Myocardial Infarctions. Of these hospitals, 69 (25.5%, the largest share of any
ECHO country) treated less than 30 patients each (1.38% of total AMI patients),
and were thus excluded from the analysis.
Higher hospital risk-
adjusted case fatality
rates may signal lower
quality of care for
coronary ischemic
conditions.
Funnel plots are used throughout this section to represent the performance of
Spanish hospitals as compared with national standards or benchmarks.
Each hospital (dot and numerical code) is charted according to its risk-adjusted
case fatality rate and the volume of patients or procedures in a year. The
benchmark is based on the average CFR (risk-adjusted) of Spanish hospitals and
the corresponding 95% and 99% CIs. The solid grey line represents the English CFR,
red lines correspond to the 95% confidence interval control limits and the dashed
blue lines represent the 99% confidence interval limits. These thresholds represent
the boundary between expected variation in outcomes (not significantly different
from the average) and unwarranted variation. Outcomes lying above the upper
thresholds are indicative of poor performing hospitals (alert or alarm position);
outcomes below the lower limits indicate good or excellent performing hospitals.
Outliers (in either direction) should be investigated further to identify the
underlying factors and should be addressed or used as examples of good practice.
For methodological reasons, hospitals treating less than 30 episodes or procedures
per year have been excluded from the analysis.
24
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Of the 51,955 admissions analysed across the 202 remaining hospitals, 4,210 (8%)
patients died. The overall risk-adjusted CFR was 1 death per 10.7 AMI admissions,
resulting in a countrywide average of 93.75 per 1,000 patients aged 18 and older,
5.3 per thousand points below the ECHO benchmark.
The risk-adjusted CFRs of individual hospitals ranged from 23.23 (5th percentile)
to 181.2 (95th percentile) per 1,000 AMI patients. Thus, depending on the centre
in which they were treated, an AMI patient´s probability of dying could differ by
up to 7.8-fold (see Table 11 in Appendix 2b for further details).
The funnel plot in Figure 24 shows that the results of national benchmarking
differ slightly from those yielded by the international comparison (Figure 5,
Section IIb). Using the national benchmark, 28 hospitals were flagged as
alert/alarm performers (versus 15 according to ECHO standards) and 42 as
good/excellent performers (versus 52 according to ECHO standards).
In 2009, less than three quarters of Spanish hospitals (70.6%) recorded an annual
volume of AMI patients of over 250. This value was set as the threshold for low
Figure 24. In-hospital mortality after AMI admission at Spanish hospitals. Year 2009.
Each dot represents a hospital that treated more than 30 AMI cases. The expected number of deaths per 1,000 hospitalised patients is based on the average calculated for all Spanish hospitals.
25
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
vs. high volume. A trend towards a decreased share of AMI patients was seen in
higher volume hospitals. In fact, in all but 5 of the 28 cases, the poorest
performers (with a risk-adjusted CFR up to 4 times higher than the national
average) were far below the low-volume threshold (Table 12 in Appendix 2b
provides detailed information on each hospital).
However, with the lowest in-hospital risk-adjusted mortality rate for AMI of all
ECHO countries, the outcomes for Spanish hospitals in 2009 indicate reasonably
good performance: only 10.1% of patients were hospitalised at alert/alarm
performers, while 31.4% of patients were treated at good or excellent centres. A
total of 65.5% of hospitals were classed as average performers, indicating a risk-
adjusted CFR that did not differ significantly from the benchmark.
In-hospital case fatality rate for Percutaneous Coronary
Interventions
In 2009, 43,868 PCI procedures were performed across 81 Spanish hospitals, with
a risk-adjusted case fatality rate of 1 death per 39 interventions in patients aged
40 or older.
PCI CFRs varied widely across hospitals, ranging from zero to 79 deaths per 1,000
patients. Thus, depending on the hospital in which the procedure was performed,
the risk of death could differ by up to 17-fold (EQ5-95) (see Tables 11 and 13 in
Appendix 2b for further details).
Since the Spanish in-country benchmark for PCI was higher than the ECHO
benchmark, in-country benchmarking of Spanish hospital performance was less
demanding than the international comparison. As shown in Figure 25, when
nationally benchmarked, 11 hospitals were flagged as alert/alarm (versus the 25
that were flagged according to ECHO benchmarking), while 14 were deemed
good or excellent performers (versus just 3 according to ECHO benchmarking).
The 11 alert/alarm hospitals (13.6% of the total) treated 15% of all patients
undergoing PCI, while hospitals flagged as good/excellent treated 20% of PCI
patients.
Contrary to expectations, the “volume effect” seemed to be all but reversed for
this particular procedure in Spain: there was a wide dispersion of high-volume
activity hospitals that were flagged as alert. The proportion of hospitals carrying
26
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
out 500-1,000 procedures/year was the same for alert/alarm centres as for good
or excellent performers.
In-hospital case fatality rate for Coronary Artery Bypass Graft
procedure
In 2009, 71,147 CABG surgeries were performed at 46 Spanish hospitals, of which
5.9% resulted in death, yielding a risk-adjusted hospital CFR 1 death per 15
interventions in patients aged 40 or older.
In terms of individual hospitals, CABG CFRs ranged from 11 (5th percentile) to 145
(95th percentile) deaths per 1,000 interventions. Thus, the risk-adjusted
probability of death for patients undergoing CABG surgery could differ by up 13-
fold, depending on the hospital in which they were treated (see Tables 11 and 14
in Appendix 2b for further details).
Figure 25. In-hospital mortality after undergoing PCI in Spanish hospitals. Year 2009.
Each dot represents a hospital performing more than 30 interventions during the period of analysis. The expected number of deaths per 1,000 hospitalised patients is based on the average calculated for all Spanish hospitals.
s.
Each dot represent each of the hospitals in the country with cases during the period of analysis. The expected number of deceases per 1,000 patients discharged is built on the national average across hospitals.
27
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
The average hospital risk-adjusted CFR for CABG in Spain in 2009 was much
higher than the ECHO average, as described in Section IIb. Thus, once again, in-
country benchmarking appeared to be less demanding than the international
comparison. As shown in Figure 26, national benchmarking flagged 7 hospitals as
alert/alarm performers (versus 12 labelled as such according to ECHO
benchmarking) while 8 were deemed good or excellent performers (versus 2
labelled as such according to ECHO benchmarking). A total of 67.4% of hospitals
were classed as average performers, indicating a risk-adjusted in-hospital
mortality rate that did not significantly different from the benchmark. In total,
13% of patients underwent interventions at alert/alarm centres, while 20.5%
underwent surgery at hospitals deemed good or excellent performers.
A notable finding was the extremely low percentage of high-volume hospitals
(11%), and hence the very low percentage of patients who underwent CABG
surgery at those centres (only 21%). Moreover, all but one of the high-volume
hospitals were classed as average performers, indicating that that the risk-
adjusted CFR did not differ significantly from the benchmark (see Table 14 in
Appendix 2b for further details).
Figure 26. In-hospital mortality after undergoing CABG surgery in Spain hospitals. Year 2009.
Each dot represents a hospital performing more than 30 interventions during the period of analysis. Given the limited number of centres the risk-adjusted case fatality rates per 1,000 patients undergoing CABG surgery are depicted relative to the ECHO average.
.
28
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Identifying trends (upwards, downwards or unchanged) in revascularisation
surgery over time is helpful in understanding the overall dynamic of the
adoption/established use/withdrawal of this medical procedure in relation to the
evolution of the burden of ischemic disease.
Increases in both PCI and CABG could be considered an indication of growing
overexposure if the burden of ischemic disease does not undergo a corresponding
increase. A decrease in PCI rates that is not accompanied by a corresponding
reduction in CID rates could be indicative of underexposure. In turn, decreases in
CABG rates should be observed in tandem with alterations in PCI rates. If PCI rates
rise, further analysis should be performed to determine whether PCI adoption is
substituting CABG. If this is not the case, any decrease should be considered
indicative of underexposure, provided that CID admissions do not show a
corresponding reduction.
The degree of systematic variation indicates how homogeneous a population’s
exposure to the procedure is at each point in time: the higher the SCV, the more
the unwarranted variation in residents´ exposure to the procedure across
healthcare areas.
IV. EVOLUTION OVER TIME
a. Geographic approach
From 2002 to 2009, coronary ischaemic disease admissions decreased slightly (by
11%), from 1 admission per 392 to 1 admission per 429 adult inhabitants. The
systematic portion of this variation remained moderate, at around 11% higher
that randomly expected (see Table 15 in Appendix 3a).
CID admissions labelled as AMI remained almost constant, ranging from 1
admission per 764 to 1 admission per 742 adult inhabitants. Variation not
deemed random also remained moderate and stable (see Table 16 in Appendix
3a).
During the same period, PCI rates increased by 75%, with values almost doubling,
from 1 admission per 791 to 1 admission per 413 inhabitants. The associated
systematic variation remained constant over this period, at around 22% higher
than expected by chance (see Table 17 in Appendix 3a). Thus, differences in
exposure across the territory remained, despite a near doubling of the overall PCI
rate.
During the period 2002-
2009, hospitalisations for
coronary ischaemic disease
remained quite stable, in
contrast to the huge
increase seen in PCI
utilisation.
In terms of hospital
outcomes, average risk-
adjusted CFRs for AMI and
CABG patients improved
over the study period. By
contrast, PCI CFRs
continually increased from
2006. However, individual
analysis reveals several
hospital with evolutionary
patterns that warrant
further investigation to
identify the factors
underlying their success or
failure.
29
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
The CABG rate remained stable over the same period, moving from 1 admission
per 2,899 to 1 admission per 2,857 inhabitants aged 40 or older. As seen for PCI,
systematic variation in CABG utilisation remained constant and moderate
throughout the study period.
CORONARY ISCHAEMIC DISEASE
ASTHMA
ACUTE MYOCARDIAL INFARCTION
0
5
10
15
20
25
30
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dar
dis
ed
isc
hae
mic
dis
eas
e
ho
spit
alis
atio
ns
pe
r 1
0,0
00
in
hab
itan
ts
0
0.2
0.4
0.6
0.8
1
Syst
em
atic
Co
mp
on
en
t o
f V
aria
tio
n
Ischaemic disease hospitalisations
Systematic variation in ischaemic disease hospitalisations across healthcare areas
0.0
3.0
6.0
9.0
12.0
15.0
2002 2003 2004 2005 2006 2007 2008 2009St
and
ard
ise
d A
MI
ho
spit
alis
atio
ns
pe
r 1
0,0
00
inh
abit
ants
0
0.2
0.4
0.6
0.8
1
Syst
emat
ic C
om
po
nen
t o
f
Var
iati
on
AMI hospitalisations Systematic variation in AMI hospitalisations across healthcare areas
PERCUTANEOUS CORONARY INTERVENTION CORONARY ARTERY BYPASS GRAFT
0
5
10
15
20
25
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dar
dis
ed
PC
I ra
te p
er
10
,00
0
inh
abit
ants
0.0
0.2
0.4
0.6
0.8
1.0
Syst
emat
ic C
om
po
nen
t o
f V
aria
tio
n
PCI utlisation Systematic variation in PCI utilisation across healthcare areas
a<
0
1
2
3
4
5
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dar
dis
ed
CA
BG
rat
e p
er
10
,00
0 i
nh
abit
ants
0
0.2
0.4
0.6
0.8
1
Syst
em
atic
Co
mp
on
en
t o
f
Var
iati
on
CABG utilisation Systematic Variation in CABG utilisation across healthcare areas
Figure 27. Temporal evolution of cardiovascular indicators (geographical approach)
30
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
This section describes several selected examples.
In addition these specific examples of changes in revascularisation utilisation, it is
useful to consider the distribution of bubbles in 2009: as all began in the same
utilisation quintile in 2002, the variety of colours they have acquired by the final
year (one colour for each quintile of utilisation intensity) provides an indication of
the degree to which the medical practice underlying this utilisation pattern has
become established, and of its homogeneity over time and across healthcare
areas.
Trends in healthcare administrative areas within the lowest and
highest quintiles of PCI and CABG utilisation.
As mentioned above, the Spanish PCI rate increased sharply over the period
2002-2009. Analysis of the evolution of healthcare areas whose PCI rates were
among the lowest at the beginning of the study period (Q1) revealed subsequent
spreading across all utilisation quintiles over time (Figure 28). For example, in one
rates increased over time, reaching the fourth quintile in 2009. By contrast, the
other retained the lowest rates for the entire study period. The same
phenomenon is illustrated in Figure 29, but in this case for healthcare areas that
began at the top of the utilisation range (Q5). Most areas remained in the higher
utilisation quintiles (Q5 or Q4), as seen in the case of one. The other, showed
marked decreases in their rates, and ended the study period in the lowest
quintile of PCI utilisation (Q1).
Similar patterns were observed for CABG surgery. For example, both showed low
rates in 2002, but evolved quite differently over time. While one remained
among the lowest quintiles, the other reached the highest utilisation levels by the
end of the study period (Figure 30).
Areas with the highest rates of CABG utilisation in 2002 (Q5, in orange) also
showed uneven evolution over the study period, although most remained in the
top utilisation quintiles. Conversely, several areas, showed steady decreases in
their CABG rates over time, ending the study period in the lowest quintile of
exposure (Figure 31).
31
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Figure 28. Trends in PCI utilisation rates in healthcare areas
showing the lowest rates at the beginning of the study period. Figure 29. Trends in PCI utilisation rates in healthcare areas
showing the highest rates at the beginning of the study period.
Figure 30. Trends in CABG utilisation rates in healthcare areas showing the lowest rates at the beginning of the study period.
Figure 31. Trends in CABG utilisation rates in healthcare areas showing the highest rates at the beginning of the study period.
32
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
b. Hospital approach
To study in-hospital mortality throughout the period of analysis, several examples
are provided showing the evolution of hospitals with the lowest or highest rates
at the beginning of the study period.
Dynamic bubble plots show the sequence of results from funnel plots assessing
annual outcomes throughout the period of analysis. The size of the bubble is
proportional to the number of patients or interventions. Hospitals flagged as
good or even excellent performers (blue bubbles) in 2002 are expected to remain
blue for the duration of the study period. However, hospitals identified at the
beginning of the period as poor performers in the alert/alarm position (orange
bubbles) should show improvements over time (turning to green (average) or,
ideally, bluish).
Departures from this pattern are considered undesirable, and warrant further
analysis.
33
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
In-hospital case fatality rate trends for Acute Myocardial Infarction
patients, 2002-2009.
Figure 32 shows four examples of hospitals whose risk-adjusted CFR for AMI
patients improved or worsened, or remained unchanged during the period
analysed.
One hospital began with an average rating, but improved to excellent over the
study period. Both hospitals (high activity volume) and the other (low activity
volume - note that the size of the bubble is proportional to the number of cases
treated) began with excellent and good performance ratings, respectively. The
former, after maintaining its rating for several years, dropped to alert, while the
latter showed fluctuations to average and less safe. the other, on the other hand,
is an example of a hospital that improved its performance throughout the study
period, starting at alert and ending at average. Further details on the evolution of
the relative performance of Spanish hospitals for AMI admissions throughout this
period are provided in Appendix 3b, Table19.
Figure 32. In-hospital mortality trends for AMI, 2002-2009, showing some of the highest and lowest rates and their evolution.
Bubbles represent individual hospitals. The larger the bubble, the higher the number of hospitalised AMI patients at that hospital. Dark-blue bubbles represent hospitals with risk-adjusted case fatality rates below the CI-99% control limit, and thus classed as “excellent”. Light-blue bubbles represent hospitals with risk-adjusted case fatality rates below the CI-95% control limit, and thus classed as “good”. Yellow bubbles represent hospitals with risk-adjusted case fatality rates above the CI-95% control limit, and thus classed as alert. Orange bubbles represent hospitals with risk-adjusted case fatality rates above the CI-99% control limit, and thus classed as alarm.
34
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
In-hospital case fatality rate for Percutaneous Coronary Intervention,
2002-2009.
As shown in Figure 33, the analysis revealed some hospitals beginning and ending
the study period with an alert/alarm rating, after some fluctuations to average,
others showing non-significant fluctuations to end the study period with an alarm
rating, an yet others showing improvements in performance, moving from alert
to an excellent rating while increasing their volume of activity. The other hospital
is an example of fluctuating evolution, moving between average and excellent.
Further details on the evolution of the relative performance of Spanish hospitals
for PCI throughout this period are provided in Appendix 3b, Table 20.
Figure 33. In-hospital mortality trends for PCI, 2002-2009, showing some of the highest and lowest rates and their evolution
Bubbles represent individual hospitals. The larger the bubble, the higher the number of hospitalised PCI patients at that hospital. Dark-blue bubbles represent hospitals with risk-adjusted case fatality rates below the CI-99% control limit, and thus classed as “excellent”. Light-blue bubbles represent
hospitals with risk-adjusted case fatality rates below the CI-95% control limit, and thus classed as “good”. Yellow bubbles represent hospitals with risk-adjusted case fatality rates above the CI-95% control limit, and thus classed as alert. Orange bubbles represent hospitals with risk-adjusted case fatality rates above the CI-99% control limit, and thus classed as alarm.
35
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
In-hospital case fatality rate trends for Coronary Artery Bypass Graft
surgery, 2002-2009.
Analysis of coronary artery bypass CFRs revealed marked changes in performance
in some hospitals during the period of analysis. Figure 34 shows two examples:,
whose risk-adjusted CFR decreased dramatically after several years in the alert
position, ending the study period with a good/excellent rating; and the other,
whose performance started at average, fluctuated to excellent, returned to
average, and ended up at alarm. One hospital remained in the alert/alarm
position, while the other remained steady in the excellent position throughout
the period analysed. Further details on the evolution of the relative performance
of Spanish hospitals for CABG throughout this period are provided in Appendix
3b, Table 21.
Figure 34. In-hospital mortality trends for CABG, 2002-2009, showing some of the highest and lowest rates and their evolution
Bubbles represent individual hospitals. The larger the bubble, the higher the number of hospitalised CABG patients at that hospital. Dark-blue bubbles represent hospitals with risk-adjusted case fatality rates below the CI-99% control limit, and thus classed as “excellent”. Light-blue bubbles represent hospitals with risk-adjusted case fatality rates below the CI-95% control limit, and thus classed as “good”. Yellow bubbles represent hospitals with risk-adjusted case fatality rates above the CI-95% control limit, and thus classed as alert. Orange bubbles represent hospitals with risk-adjusted case fatality rates above the CI-99% control limit, and thus classed as alarm.
36
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
The graphs in this section provide some sense of the behaviour of CID admissions
and revascularization procedures as a function of the average level of affluence in
a healthcare area. At a glance, these graphs indicate whether differences exist
between better-off and worse-off areas, and whether these differences vary over
time.
The wider the gap between the most and least deprived quintiles, the more
inequitably distributed the exposure to revascularisation surgery. It is also useful
to observe the 95% confident intervals (whiskers) above and below the annual
rates estimated for quintiles 1 and 5. Only those that do not overlap signal a
statistically significant difference between wealthier and deprived areas.
The desired pattern is one which shows no statistically significant differences
across healthcare areas as a function of income, or at least the disappearance of
any existing gap over time. atributable
V. SOCIAL GRADIENT
Significantly more CID admissions were recorded in the most deprived healthcare
areas (Q1) than in the wealthiest ones (Q5). Analyis of CID admissions labelled as
AMI revealed that worse-off areas recorded more admissions than better-off
ones, with the gap between the lowest and highest quintiles widening over time.
When analysing PCI utilisation, the most deprived areas showed significantly
higher rates than the more affluent areas from 2006 onwards. Moreover, the gap
between the lowest and highest quintiles became wider over time. It is worth
noting that the increase in utilisation rates in the most deprived areas was 2.5
times greater than that seen in the most affluent areas (see Table 17 in Appendix
3a).
The opposite pattern was observed for CABG surgery, which was significantly
more frequent in wealthier areas than in worse-off ones between 2004 and 2009.
Again, the gap between the lowest and highest quintiles became slightly wider
over time, due this time to a slight increase in CABG utilisation in better-off areas.
Nevertheless, although significant, the difference in the absolute value between
the lowest and highest quintiles was quite small, at 1.6 admissions/per 10,000
inhabitants (see Table 18 in Appendix 3a)
The most deprived
healthcare areas had higher
CID admissions and
significantly higher PCI
utilisation rates than the
most affluent areas. By
contrast, CABG utilisation
was higher in wealthier
areas than in worse-off
areas.
37
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
The variation in CID admissions across areas described in previous sections thus
appears to be related to income at the area level. Since worse-off areas bear a
greater burden of CID admissions and, specifically, of AMI hospitalisations, we
could expect that these areas have a greater need for health care than high-
income areas. This was observed for PCI use, but not for CABG, which was more
frequent in wealthier areas. Despite the huge increase in PCI utilisation in
deprived areas, the higher number of CID admissions and the growing rate in AMI
in these areas warrant further investigation to determine whether there are
barriers to equity of access to CABG procedures, or other treatments, in these
areas.
CORONARY ISCHAEMIC DISEASE ACUTE MYOCARDIAL INFARCTION
0
10
20
30
40
2002 2003 2004 2005 2006 2007 2008 2009
Sta
nd
ard
ise
d R
ate
Q1 Q2 Q3 Q4 Q5
0
4
8
12
16
20
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dar
dis
ed R
ate
Q1 Q2 Q3 Q4 Q5
PERCUTANEOUS CORONARY INTERVENTION CORONARY ARTERY BYPASS GRAFT
0
10
20
30
40
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dard
ised
Rat
e
Q1 Q2 Q3 Q4 Q5
0
1
2
3
4
5
6
2002 2003 2004 2005 2006 2007 2008 2009
Stan
dard
ised
Rat
e
Q1 Q2 Q3 Q4 Q5
Figure 35. Trends in standardised rates by income quintile.
38
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
VI. POLICY IMPLICATIONS
Coronary ischaemic disease is one of the leading causes of death, disability and
decreased quality of life in Europe, and together with cancer was the main cause
of death in Spain in 2009. It is also a leading cause of premature death in men,
generating important social costs associated with potential years of life lost.
Mortality and morbidity associated with cardiovascular disease have thus become
an important issue for all health systems in Europe, as well as an important driver
of health expenditure.
Several studies in the last decade have shown that the incidence of coronary
heart disease in the northern half of Europe, particularly Scandinavian countries,
is higher than in the south. Even though hospitalisations for ischemic heart
disease are decreasing overall, higher rates were recorded in England and
Denmark (in this order) than in Spain and Portugal. This factor should be taken
into account when assessing and comparing hospitalisation rates and the
intensity of consequent interventions. This section highlights specific factors in
the Spanish healthcare system and/or the organisational processes that may
contribute to the observed results and thus may warrant further examination.
The mapping of burden of coronary disease and PCI intensity of use produces in
some cases contradicting patterns: reveals contradicting patterns in some cases:
some Healthcare Areas with the highest PCI utilisation rates had a lower relative
risk of CID hospitalisation, while others had a very high relative risk. Given the
potential benefit of primary PCI, two hypotheses, which may be concomitant
rather than alternative, should be considered. A higher number of early
interventions could prevent hospitalisation at later disease stages, and thus,
reduce the corresponding admission rate. However, the local risk of being
hospitalised due to CID should drive the need for PCI procedures and hence the
local intensity of use. If that were not the case, such a high intensity of PCI
revascularisation unrelated to need could indicate over-utilisation of the
procedure, i.e., over-exposure of the population and hence inadequate provision
of care.
The geographical analysis revealed a significant role of regions in explaining the
observed variation in the population’s exposure to PCI or CABG: regions
accounted for more than 40% of the variation not attributable to areas. This may
be due to the application of different regional health plans or different
39
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
implementations of national strategies at the local level1 highlighting the
important role of regions in modulating certain health policies, and consequently,
the provision of healthcare services, including these interventions. Alternatively,
the existence of a concentration of early adopters (high-tech hospitals acting as
referral hospitals for all citizens within a region) could explain the large
component of variation attributable to regions.
At the Healthcare area level, only PCI utilisation appeared to be positively
associated with the burden of disease, although this association was small. The
joint analysis of the utilisation patterns for both revascularisation procedures (PCI
and CABG) provides some evidence to suggest a general (in terms of regions)
substitution effect. Nonetheless, a case-by-case analysis of discrepant trends
could provide further insight. In such cases, it would be important to identify
factors other than need or technological changes that may contribute to the
observed revascularisation rates.
Analysis of case fatality rates at the hospital level revealed that Spanish risk-
adjusted in-hospital mortality for AMI patients significantly after 2002, and was
among the lowest of all ECHO countries in 2009. Detailed analysis revealed that
most Spanish hospitals (almost 65.5%) treated AMI patients within the expected
(average) levels of quality and safety. However, detailed analysis revealed two
extreme patterns of care provision. On one hand, 13.8% of hospitals (treating
almost 10% of all Spanish AMI patients) recorded in-hospital mortality rates that
were significantly higher than expected, and were consequently flagged as alert
or alarm. On the other hand, 20.7% of hospitals (treating 31.4% of all AMI
patients) had significantly lower than expected risk-adjusted CFRs and were
flagged as good or excellent. The quantified effect on variation attributable to the
hospital of treatment was 1.252. In the case of AMI, this means that pairs of
similar, randomly chosen patients from two different hospitals would have a 25%
difference in the risk of dying depending on where they were hospitalised.
Analysis of in-hospital mortality after revascularization reveals a more worrying
picture. Risk-adjusted CFRs for both PCI and CABG in 2009 were the highest
across ECHO countries. Two specific findings were of concern: the increase in risk-
adjusted CFR after PCI, and the fact that Spain had the highest proportion of
ECHO hospitals with an alarm rating performing PCI and CABG. In the case of PCI,
similar patients treated in different hospitals could have a differential risk of
dying as high as 44%. This figure increased to 65% for CABG.
1 Ministerio de Sanidad y Consumo. Estrategia en Cardiopatía Isquémica del SNS. 2006
2 Calculated throughout the Median Odds Ratio [MOR]
40
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Volume has been proposed as one potential factors underlying these differences.
In fact, the vast majority of Spanish hospitals registered a volume of annual
patients well below the ECHO high-volume threshold. The lower the volume the
higher the probability of poorer outcomes. Other countries in ECHO, such as
Denmark, have recently implemented reforms addressing this issue, encouraging
centralisation of specialised interventions into fewer centres, such that the
minimum amount of procedures per centre is guaranteed to maintain high level
of expertise and quality. In the case of England, a provision was introduced
whereby hospitals were required to perform more than 300 CABG a year to
acquire accreditation.
The literature recommends assessing a number of critical factors that may
account for differences in hospital outcomes (at both local and global levels).
These include pre-hospital diagnosis and planning of urgent transportation to the
appropriate medical centre. Assessing the relationship with the eventual hospital
of reference could provide important insights as to whether there is a well-
defined, stable and fluid bypass circuit or special techniques for severe patients,
and whether transfer to reference centres occurs immediately or within 24 hours,
depending on the severity of the situation. Analysis of these are key components
of successful treatment could be very helpful in improving patient outcomes as
well as overall health system costs.
The analysis conducted suggests that there is room to improve outcomes in the
Spanish system. The burden of disease and revascularisation rates are generally
lower as compared with other ECHO countries, but do not appear to relate to one
another, suggesting that factors other than need or technological change may
drive revascularisation intensity.
On the other hand the surgical outcomes of Spanish hospitals were rather poor
when assessed by international benchmarking. The comparatively poor results of
some Spanish hospitals for PCI and CABG patients, as assessed by national
standards, also warrant closer analysis. The fact that only 21% of patients
undergoing CABG were treated in high-volume hospitals deserves further
consideration.
41
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 1. General descriptive statistics for burden of disease: CID admissions
Stand. Rate: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: ECHO countries, 2009); EQ: extremal quotient; SCV: systematic component of variation.
Table 2. General descriptive statistics for burden of disease: AMI admissions
Stand. Rate: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: ECHO countries, 2009); EQ: extremal quotient; SCV: systematic component of variation.
Table 3. General descriptive statistics for utilisation of PCI procedure
Stand. Rate: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: ECHO countries, 2009); EQ: extremal quotient; SCV: systematic component of variation.
Table 4. General descriptive statistics for utilisation of CABG surgery
Stand. Rate: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: ECHO countries, 2009); EQ: extremal quotient; SCV: systematic component of variation.
CORONARY ISCHAEMIC DISEASE
DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Cases 13225 141167 14526 4288 78585
Stand. Rate 30.68 34.32 17.86 32.40 23.79 EQ5-95 2.32 2.16 2.12 1.89 3.04
SCV 0.14 0.24 0.15 0.09 0.10
ACUTE MYOCARDIAL INFARCTION
DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Cases 6711 69713 11365 2911 46206
Stand. Rate 15.90 16.76 13.80 22.29 13.78
EQ5-95 1.91 2.63 2.37 1.67 2.98
SCV 0.05 0.15 0.05 0.34 0.11
PERCUTANEOUS CORONARY INTERVENTION
DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Cases 9253 63220 10587 5025 48368
Stand. Rate 37.50 27.18 21.37 60.16 23.89
EQ5-95 1.86 2.20 2.24 2.61 4.71
SCV 0.33 0.08 0.08 1.97 0.22
CORONARY ARTERY BYPASS GRAFT
DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Cases 2371 20434 2446 774 7068
Stand. Rate 9.99 9.00 4.77 9.77 3.38
EQ5-95 1.71 2.33 7.42 5.32 9.83
SCV 0.50 0.41 0.19 0.74 0.27
APPENDIX 1a:
International
Comparison across
ECHO countries
GEOGRAPHICAL
APPROACH
Year 2009
42
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 5. Data for hospitals and patients included* in the analysis.
Hospitals recording less than 30 patients or procedures per year have been excluded from the analysis to avoid noise
when estimating risk-adjustment by logistic multivariate modelling.
Table 6: ECHO hospitals' description and relative performance per country for
hospitalised AMI patients (ECHO benchmark estimation)
Hospitals>250: Hospitals above the activity threshold of 250 AMI hospitalisations a year; Alarm position: hospitals above the CI-99 limit; Alert position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers: hospitals below the CI-99 limit. The percentage of AMI patients in the country hospitalised at each hospital type is shown in brackets.
ECHO DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
ACUTE MYOCARDIAL INFARCTION
Total discharges 147670 8124 71001 12391 3471 52683
Total nº hospitals 522 35 154 46 16 271
Hospitals excluded 87 5 5 6 2 69
(% patients excluded) 0.55% 0.48% 0.01% 0.28% 0.06% 1.38%
Discharges analysed 146859 8085 70994 12356 3469 51955
Nº Hospitals analysed 435 30 149 40 14 202
PERCUTANEOUS CORONARY INTERVENTION
Total discharges 133161 9306 64253 10760 4817 44025
Total nº hospitals 283 25 97 39 9 113
Hospitals excluded 84 18 24 9 1 32
% patients excluded 0.32% 0.43% 0.18% 0.92% 0.29% 0.36%
Discharges analysed 132737 9266 64139 10661 4803 43868
Nº Hospitals analysed 199 7 73 30 8 81
CORONARY ARTERY BYPASS GRAFT
Total discharges 33765 2390 21036 2496 678 7165
Total nº hospitals 145 17 53 10 2 63
Hospitals excluded 56 11 24 4 --- 17
% patients excluded 0.24% 1.26% 0.14% 0.16% --- 0.25%
Discharges analysed 33683 2360 21006 2492 678 7147
Nº Hospitals analysed 89 6 29 6 2 46
ACUTE MYOCARDIAL INFARCTION
ECHO DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Discharges 146859 8085 70994 12356 3469 51955
Deceased 12582 674 6281 1183 240 4204
Nº Hospitals 435 30 149 40 14 202 Hospitals > 250 (% patients treated)
239 (82.47%)
6 (70.3%)
125 (93.9%)
23 (79%)
3 (66.59%)
82 (70.59%)
Average expected Risk-adjusted CFR
99.03 133.45 94.41 109.57 101.58 93.75
Hosp. Alarm position (% patients treated)
40 (5.83%)
10 (21.13%)
9 (4.30%)
10 (20.31%)
3 (7.81%)
6 (1.09%)
Hosp. Alert position (% patients treated)
18 (3.19%)
3 (3.45%)
6 (3.54%)
1 (1.45%)
1 (1.59%)
9 (4.09%)
Hosp. Good performers (% patients treated)
42 (11.42%)
2 (3.15%)
14 (10.65%)
3 (9.43%)
2 (5.85%)
20 (13.97%)
Hosp. Excellent performers (% patients treated)
67 (26.7%)
5 (60.63%)
22 (23.6%)
5 (19.06%)
3 (51.14%)
32 (25.85%)
APPENDIX 1b:
International
Comparison across
ECHO countries
HOSPITAL
APPROACH
Year 2009
43
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 7: ECHO hospitals: description and relative performance per country for
patients undergoing PCI (ECHO benchmark estimation)
Hospitals>250: Hospitals above the activity threshold of 250 PCI a year; Alarm position: hospitals above the CI-99 limit; Alert position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers: hospitals below the CI-99 limit. The percentage of patients in the country undergoing a PCI procedure at each hospital type is shown in brackets.
Table 8: ECHO hospitals: description and relative performance per country for
patients undergoing CABG (ECHO benchmark estimation)
Hospitals>250: Hospitals above the activity threshold of 250 CABG a year; Alarm position: hospitals above the CI-99 limit; Alert position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers: hospitals below the CI-99 limit. The percentage of patients in the country undergoing CABG surgery at each hospital type is shown in brackets.
PERCUTANEOUS CORONARY INTERVENTION
ECHO DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Discharges 132737 9266 64139 10661 4803 43868
Deceased 2623 255 924 188 143 1113
Nº Hospitals 199 7 73 30 8 81
Hospitals > 250 (% patients treated)
159 (95.44%)
7 (100%)
64 (97.17%)
15 (84.05%)
5 (97.04%)
68 (94.53%)
Average expected Risk-adjusted CFR
19.86 22.78 13.70 20.77 15.61 25.59
Hosp. Alarm position (% patients treated)
28 (17.26%)
4 (67.47%)
1 (1.55%)
3 (9.69%)
2 (74.47%)
18 (25.19%)
Hosp. Alert position (% patients treated)
10 (3.9%)
--- ---
2 (1.80%)
1 (1.76%)
--- ---
7 (8.74%)
Hosp. Good performers (% patients treated)
17 (4.8%)
2 (7.52%)
13 (7.80%)
--- ---
1 (5.58%)
1 (0.92%)
Hosp. Excellent performers (% patients treated)
15 (15.51%)
--- ---
12 (28.27%)
1 (9.80%)
--- ---
2 (3.20%)
CORONARY ARTERY BYPASS GRAFT ECHO DENMARK ENGLAND PORTUGAL SLOVENIA SPAIN
Discharges 33683 2360 21006 2492 678 7147
Deceased 1212 96 571 87 37 421
Nº Hospitals 89 6 29 6 2 46
Hospitals > 250 (% patients treated)
46 (82.16%)
5 (93.43%)
29 (100%)
6 (100%)
1 (70.06%)
5 (20.93%)
Average expected Risk-adjusted CFR
50.33 44.54 27.81 33.55 44.97 66
Hosp. Alarm position (% patients treated)
9 (3.58%)
--- ---
--- ---
--- ---
--- ---
9 (16.87%)
Hosp. Alert position (% patients treated)
4 (2.03%)
--- ---
--- ---
1 (16.21%)
--- ---
3 (3.92%)
Hosp. Good performers (% patients treated)
13 (20.65%)
--- ---
8 (26.09%)
2 (32.58%)
1 (29.94%)
2 (6.46%)
Hosp. Excellent performers (% patients treated)
18 (40.61%)
1 (24.79%)
16 (60.32%)
1 (16.97%)
--- ---
--- ---
APPENDIX 1b:
International
Comparison across
ECHO countries
HOSPITAL
APPROACH
Year 2009
44
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 9: Descriptive statistics for burden of coronary disease and use of revascularisation procedures across healthcare areas.
Stand. Rate: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: national 2009); sR Px: percentile x of Stand. Rate distribution; EQ: Extremal quotient; ICC: Intra class
correlation coefficient
Table 10: Relative risk of exposure to coronary disease and revascularisation procedures across healthcare areas.
SUR: Standardised admission/Utilization Ratio (observed/expected); SUR Px: percentile x of the SUR distribution; SCV: Systematic Component of Variation.
CID AMI PCI CABG
Cases 78,585 46,206 48,368 7,068
Population 39,808,144 39,808,144 22,988,560 22,988,560
Crude Rate 21.22 12.22 20.21 2.97
Stand. Rate 20.63 11.91 20.22 2.96
sR Min. 4.6 1.5 0.13 0.16
sR Max. 43.06 23.15 70.02 7.83
sR. P5 12.45 6.36 7.92 0.66
sR. P25 16.37 9.45 14.96 1.88
sR. P50 20.03 11.59 19.27 2.82
sR. P75 23.58 13.96 23.65 3.84
sR. P95 32.56 18.46 35.74 6.09
EQ5-95 2.61 2.9 4.51 9.28
EQ25-75 1.44 1.48 1.58 2.04
ICC 0.28 0.14 0.44 0.43
CID AMI PCI CABG
SUR Min. 0.25 0.21 0.01 0.07
SUR Max. 2.17 2 3.28 2.54
SUR P5 0.63 0.54 0.39 0.21 SUR P25 0.83 0.81 0.7 0.6 SUR P50 1.02 0.99 0.91 0.92
SUR P75 1.19 1.19 1.14 1.25
SUR P95 1.63 1.57 1.72 1.96
SCV 0.1 0.09 0.19 0.22
APPENDIX 2a:
Tables Spain
WITHIN-Country
analysis
GEOGRAPHICAL
APPROACH
Year 2009
45
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 11: Descriptive statistics for hospital activity and outcomes.
CFR: Case Fatality Rate per 1,000 hospitalised patients or patients undergoing procedure; R-adj CFRx: risk-adjusted CFR of percentile x of the CFR distribution; Rho Statistic: cluster effect; MOR (median odds ratio): variation between clusters.
Table 12: Hospital outcomes for Acute Myocardial Infarction patients* (national benchmark estimation)
AMI in-hospital
mortality PCI in-hospital
mortality CABG in-hospital
mortality
Deceased 4210 1113 421
N. hospitals 203 81 46
Crude CFR 88.60 24.62 61.57
Risk-adjusted CFR 93.75 25.59 66.00
R-adj CFR5 23.23 3.41 11.29
R-adj CFR95 181.20 56.75 144.72
Rho Statistic 0.016 0.042 0.077
MOR 1.25 1.44 1.65
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
46
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
47
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
48
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
(i) Total number of AMI admissions per hospital during the period of analysis. * Hospitals with less than 30 AMI admissions per year are excluded from the analysis. CFR: Crude case fatality rate per 1,000 hospitalised AMI patients; sCFR: Risk-adjusted Case Fatality Rate per 1,000 hospitalised AMI patients. Hospitals above the CI-99 control limit are considered in “Alarm position”; hospitals above the CI-95 control limit are considered in an “Alert position”; hospitals below the
CI-95 control limit are considered “Good performers” and hospitals below the CI-99 control limit are considered “Excellent performers”.
Table 13: Hospital outcomes for Percutaneous Coronary Interventions, year 2009. (national benchmark estimation)*
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
49
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
(i) Total number of interventions per hospital accumulated during the period of analysis. Hospitals performing less than 30 interventions per year were excluded from the analysis
CFR: Crude case fatality rate per 1,000 patients undergoing a PCI procedure; sCFR: Risk-adjusted Case Fatality Rate per 1,000 patients undergoing a PCI procedure. Hospitals above the CI-99 control limit are considered in “Alarm position”; hospitals above the CI-95 control limit are considered in an “Alert position”; hospitals below the CI-95 control limit are considered “Good performers” and hospitals below the CI-99 control limit are considered “Excellent performers”.
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
50
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 14: Hospital outcomes for Coronary Artery Bypass Graft, year 2009. National benchmark estimation*
(i) Total number of interventions per hospital during the period of analysis. CFR: Crude case fatality rate per 1,000 patients undergoing a PCI procedure; sCFR: Risk-adjusted Case Fatality Rate per 1,000 patients undergoing a CABG procedure. Hospitals above the CI-99 control limit are considered in “Alarm position”; hospitals above the CI-95 control limit are considered in an “Alert
position”; hospitals below the CI-95 control limit are considered “Good performers” and hospitals below the CI-99 control limit are considered “Excellent performers”.
APPENDIX 2b:
Tables Spain
WITHIN-Country
analysis
HOSPITAL
APPROACH
Year 2009
51
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 15. Spanish descriptive statistics for burden of disease over time: CID
Stand. Rate & sR: Age-sex Standardised Rate (Reference population: national 2002); sR Qx: quintile x of sR distribution; SCV: Systematic Component of Variation.
Table 16. Spanish descriptive statistics for burden of disease over time: AMI
Stand. Rate & sR: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: national 2002); sR Qx: quintile x of sR distribution; SCV: Systematic component of variation.
Table 17. Spanish descriptive statistics for procedure utilisation over time: PCI
Stand. Rate & sR: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: national 2002); sR Qx: quintile x of sR distribution; SCV: Systematic component of variation.
Table 18. Spanish descriptive statistics for procedure utilisation over time: CABG
Stand. Rate & sR: Age- and sex-standardised rate per 10,000 inhabitants (Reference population: national 2002); sR Qx: quintile x of sR distribution; SCV: Systematic component of variation.
CORONARY ISCHAEMIC
2002 2003 2004 2005 2006 2007 2008 2009
Cases 88670 89838 88899 86158 85239 83756 81265 78585
Stand. Rate 25.5 26.22 26.29 25.28 25.29 24.7 24.05 23.31 sR Q1. 27.38 28.60 28.61 28.59 28.98 28.43 27.83 27.02 sR Q5. 21.54 21.89 22.22 19.45 19.90 19.69 18.89 18.48
SCV 0.1 0.1 0.1 0.11 0.11 0.11 0.11 0.13
ACUTE MYOCARDIAL INFARCTION
2002 2003 2004 2005 2006 2007 2008 2009
Cases 45834 47225 47427 47144 46020 45684 46447 46206
Stand. Rate 13.09 13.68 13.81 13.59 13.35 13.28 13.55 13.48 sR Q1. 13.58 14.55 14.48 14.79 14.48 14.31 14.72 15.08 sR Q5. 11.79 12.41 12.42 11.15 10.97 11.28 10.89 11.14
SCV 0.09 0.09 0.08 0.09 0.09 0.1 0.11 0.12
PERCUTANEOUS CORONARY INTERVENTION
2002 2003 2004 2005 2006 2007 2008 2009
Cases 27566 31919 35837 39624 42696 45320 45557 48368
Stand. Rate 12.64 15.07 17.37 19.19 20.95 22.45 22.75 24.19 sR Q1. 11.96 14.85 17.06 21.26 24.43 27.07 27.23 28.95 sR Q5. 12.79 14.20 16.20 16.17 16.67 18.53 18.83 19.88
SCV 0.22 0.21 0.2 0.22 0.23 0.24 0.23 0.22
CORONARY ARTERY BYPASS GRAFT
2002 2003 2004 2005 2006 2007 2008 2009
Cases 7396 7756 7663 7175 7078 7264 7326 7068
Stand. Rate 3.45 3.63 3.73 3.39 3.46 3.63 3.63 3.5 sR Q1. 2.95 3.04 2.89 2.58 2.70 2.81 2.70 2.51 sR Q5. 3.70 4.09 4.43 3.93 3.88 4.60 4.16 4.08
SCV 0.22 0.17 0.17 0.19 0.19 0.3 0.24 0.22
APPENDIX 3a3a:
Tables Denmark
Evolution over
time
GEOGRAPHICAL
APPROACH
Period of
analysis: 2002-
2009
APPENDIX 3a:
Tables Spain
Evolution over time
GEOGRAPHICAL
APPROACH
Period of analysis:
2002-2009
52
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 19. Evolution of the relative performance of Spanish hospitals for AMI admissions (in-country benchmark estimation).
Hospitals>250: Hospitals above the activity threshold of 250 AMI hospitalisations a year; Alarm position: hospitals above the CI-99 limit; Alert position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers:
hospitals below the CI-99 limit. The percentage of patients in the country undergoing AMI at each hospital type is shown in brackets.
Table 20. Evolution of the relative performance of Spanish hospitals for patients undergoing a PCI procedure (in-country benchmark estimation)
Hospitals>250: Hospitals above the activity threshold of 250 PCI a year; Alarm position: hospitals above the CI-99 limit; Alert
position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers: hospitals below the CI-99 limit. The percentage of patients in the country undergoing a PCI procedure at each hospital type is shown in brackets.
ACUTE MYOCARDIAL INFARCTION
2002 2003 2004 2005 2006 2007 2008 2009
Discharges 46732 53303 53245 53183 51719 51790 52388 51985
Deceased 5224 5810 5432 5271 4675 4673 4584 4210
Nº Hospitals 184 200 201 203 195 197 203 203
Hospitals > 250 (% patients treated)
78 (73.32%)
88 (73.5%)
88 (73.65%)
89 (73.97%)
90 (74.58%)
82 (70.62%)
86 (72.64%)
82 (70.55%)
Average expected Risk-adjusted CFR
120.43 117.16 110.26 107.22 98.70 101.35 94.70 93.75
Hosp. Alarm position (% patients treated)
18 (10.1%)
18 (6.62%)
20 (7.34%)
18 (9.23%)
15 (5.04%)
17 (6.29%)
17 (6.28%)
16 (5.23%)
Hosp. Alert position (% patients treated)
9 (4.36%)
9 (3.43%)
6 (3.39%)
11 (4.79%)
16 (6.05%)
6 (1.67%)
9 (3.75%)
12 (4.91%)
Hosp. Good performers (% patients treated)
10 (6.94%)
14 (5.46%)
10 (4.45%)
13 (6.6%)
16 (9.86%)
8 (4.72%)
15 (10.76%)
17 (12.92%)
Hosp. Excellent performers (% patients treated)
18 (18.3%)
21 (17.24%)
23 (21.31%)
21 (19.37%)
22 (17.26%)
25 (20.51%)
15 (9.03%)
25 (18.52%)
PERCUTANEOUS CORONARY INTERVENTION
2002 2003 2004 2005 2006 2007 2008 2009
Discharges 22662 30826 34026 38022 40580 42161 42287 43868
Deceased 465 705 737 895 888 956 1033 1113
Nº Hospitals 61 75 78 77 79 79 80 81
Hospitals > 250 (% patients treated)
35 (83.92%)
48 (87.71%)
51 (87.82%)
60 (92.16%)
66 (95.23%)
65 (94.11%)
68 (95.24%)
68 (94.53%)
Average expected Risk-adjusted CFR
20.22 23.07 20.53 23.55 22.14 23.31 24.70 25.59
Hosp. Alarm position (% patients treated)
9 (16.77%)
11 (16.32%)
10 (14.92%)
9 (14.13%)
10 (13.79%)
11 (13.93%)
12 (14.44%)
10 (13.75%)
Hosp. Alert position (% patients treated)
2 (3.76%)
1 (1.47%)
3 (7.66%)
4 (4.84%)
3 (3.98%)
4 (4.31%)
3 (3.1%)
1 (1.19%)
Hosp. Good performers (% patients treated)
4 (8.19%)
8 (13.67%)
4 (4.83%)
5 (5.8%)
8 (12.20%)
9 (9.47%)
8 (11.2%)
8 (10.18%)
Hosp. Excellent performers (% patients treated)
2 (7.82%)
3 (5.02%)
2 (5.34%)
5 (9.99%)
3 (4.93%)
3 (5.66%)
4 (5.26%)
6 (9.78%)
APPENDIX 3b:
Tables Spain
Evolution over time
HOSPITAL
APPROACH
Period of analysis:
2002-2009
53
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Table 21. Evolution of the relative performance of Spanish hospitals for patients undergoing CABG surgery (in-country benchmark estimation)
Hospitals>250: Hospitals above the activity threshold of 250 CABG a year; Alarm position: hospitals above the CI-99 limit; Alert position: hospitals above the CI-95 limit; Good performers: hospitals below the CI-95 limit; Excellent performers: hospitals below the CI-99 limit. The percentage of patients in the country undergoing CABG surgery at each hospital type is shown in brackets.
CORONARY ARTERY BYPASS GRAFT
2002 2003 2004 2005 2006 2007 2008 2009
Discharges 6732 8000 7881 7355 7265 7257 7457 7147
Deceased 531 616 580 539 524 510 534 421
Nº Hospitals 35 44 44 45 45 45 45 46
Hospitals > 250 (% patients treated)
9 (38.19%)
9 (32.59%)
8 (30.97%)
5 (19.71%)
4 (15.36%)
7 (26.87%)
6 (23.63%)
5 (20.93%)
Average expected Risk-adjusted CFR
88.12 85.54 81.31 80.58 81.38 77.25 82.44 66.00
Hosp. Alarm position (% patients treated)
4 (7.77%)
5 (9.09%)
6 (10.3%)
4 (8.67%)
4 (6.28%)
3 (4.46%)
4 (4.69%)
6 (11.59%)
Hosp. Alert position (% patients treated)
3 (7.58%)
2 (3.79%)
1 (1.42%)
2 (2.38%)
3 (5.2%)
2 (5.43%)
2 (3.66%)
1 (1.33%)
Hosp. Good performers (% patients treated)
4 (12.4%)
7 (22.55%)
6 (15.05%)
4 (10.65%)
--- ---
2 (5.43%)
6 (17.58%)
5 (10.98%)
Hosp. Excellent performers (% patients treated)
4 (15.29%)
6 (18.11%)
4 (12.65%)
6 (17.59%)
5 (14.21%)
4 (11.97%)
4 (10.31%)
3 (9.47%)
APPENDIX 3b:
Tables Spain
Evolution over
time
HOSPITAL
APPROACH
Period of
analysis: 2002-
2009
54
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Cardiovascular Ischaemic Disease and AMI, as well as the revascularisation
procedures PCI and CABG, are used as geographical and hospital-specific
indicators within the ECHO performance model.
This has certain implications both for methodology and the interpretation of
results. This report is based on ecological analyses; data aggregated at a certain
geographical level serve as the unit of analysis for this report. Thus, correct
interpretation of the findings provides an indication of the risk of exposure to
hospitalisation for cardiovascular conditions or revascularisation procedures for a
population living in a given area (as opposed to the risk for an individual patient).
Individual data are subsequently analysed from a provider perspective and risk-
adjusted within multivariate logistic 2-level hierarchical model, and then clustered
according to hospitals. Interpretation of these data provides an indication of the
risk of dying after being hospitalised and/or treated in a specific hospital relative
to national or ECHO country benchmarks.
Main endpoints:
This report maps standardised utilisation rates per geographical area as well as
risk-adjusted case fatality rates per provider, analysing events affected by
healthcare quality. As a summary measure of variation, the report includes the
classical statistics Ratio of Variation between extremes, Component of Systematic
Variation and Rho Statistic (cluster effect) and the Median Odds Ratio (MOR)
statistic, a measure of the variation among hospitals that compares pairs of
patients with comparable risk selected from two, randomly chosen hospitals.
MOR provides information as to how heterogeneity across hospitals increases the
individual odds of experiencing the outcome of interest, in this case case-
fatalities.
Instruments:
In the geographical approach, as this was an ecological study, each admission was
allocated to the place of residence of the patient, which in turn was linked to a
meaningful geographic unit; the 199 healthcare areas and the 17 Regions that
comprise the Spanish National Health System.
For risk-adjustment of the hospital approach within the multivariate logistic 2-
level hierarchical model, the following variables were included:
– Age and sex
APPENDIX 4:
Technical note
55
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
– For patients with a primary diagnosis of AMI, classification as transmural (with
ST segment elevation, STEMI), non-STEMI or unclassified. Whether the patient
underwent heart valve replacement and/or implantation of a cardiac or
circulatory assistance device was also noted.
– Whether the intervention consisted of major structural surgery (including
repair or revision of atrial and ventricular septa, cardiotomy, pericardiotomy,
pericardiectomy and excision of a heart lesion).
– Specific measures of the severity of the underlying condition (42 co-morbidities
included in the Elixhauser index), listed below:
Cardiac arrhythmias Drugs abuse
Valvular disease Lymphoma
Congestive heart failure Solid tumor without metastasis
Chronic lung disease Metastatic cancer
Hypertension, uncomplicated Weight loss
Hypertension, complicated Psychoses
Total hypertension disease Depression
Pulmonary circulation disorders AIDS/HIV
Renal failure Fluid and electrolyte disorders
Pre-existing hypertension complicating pregnancy
Peptic ulcer disease excluding bleeding
Other hypertension in pregnancy Deficiency anemia
Diabetes, without chronic complications Blood loss anemia
Diabetes, with chronic complications Coagulopathy
Hypothyroidism Rheumatoid arthritis/collagen vascular diseases
Liver disease Peripheral vascular disorders
Obesity Paralysis
Alcohol abuse Other neurological disorders
For both approaches, the operational definitions for each indicator are detailed in
the coding Table in Appendix 5. The indicators used are based on those in use in
the international arena, as proposed by AHRQ and OECD. For the analysis of
variation across countries indicators were subjected to a construct validity process
developed by the Atlas VPM project in Spain. Furthermore, as part of a task within
the ECHO project, cross-walks across different diseases and procedure
classifications were constructed in order to face-validate the indicators used.
APPENDIX 4:
Technical note
56
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
This report is based on hospital admissions registered in the National Discharges Dataset (CMBD). Cross- and in-country sections were built based on 2009 discharge data, whereas time-trends and social gradient analyses used data from the period 2002-2009.
Social gradient data were obtained from the Atlas VPM dataset, based on original data from the La Caixa 2003 annual report.
APPENDIX 4:
Technical note
57
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Diagnosis and procedures codes ICD9-CM
Primary diagnosis Secondary diagnosis2-30 Procedures
Inclusions Exclusions Inclusions Exclusions Inclusions Exclusions
Ischaemic Disease Age 18+ Type of admission unplanned
410.*, 411.1, 411.8, 413.* 414.01 (IF DX2-30 411.1)*
410.*, 411.1, 411.8, 413.* 414.01 (IF DX2-30 411.1)*
Acute Myocardial Infarction (AMI) Age 18+ Type of admission unplanned
410.* 410.*
Percutaneous Coronary Interventions (PCI) Age 40+
36.01, 36.02, 36.05, 36.06, 36.07, 36.08, 36.09, 00.66
Coronary Artery Bypass Grafting (CABG) Age 40+
36.10, 36.11, 36.12, 36.13, 36.14, 36.15, 36.16, 36.17, 36.19
APPENDIX 5:
Definitions
58
EUROPEAN COLLABORATION FOR
HEALTHCARE OPTIMIZATION
Diagnosis and procedures codes ICD9-CM
Primary diagnosis Secondary diagnosis2-30 Procedures
Inclusions Exclusions Inclusions Exclusions Inclusions Exclusions
Acute Myocardial Infarction in Hospital Mortality Age 18+
410.* 630.*-677.* 630.*-677.*
Percutaneous Coronary Interventions in Hospital Mortality Age 40+
630.*-677.* 630.*-677.*
36.01, 36.02, 36.05, 36.06, 36.07, 36.08, 36.09, 00.66
Coronary Artery Bypass Grafting in Hospital Mortality Age 40+
630.*-677.* 630.*-677.*
36.10, 36.11, 36.12, 36.13, 36.14, 36.15, 36.16, 36.17, 36.19
APPENDIX 5:
Definitions