Professional Documents
Culture Documents
In 1962, the Director General of the the health system on medical care use
National Board of Health in Sweden, and health outcomes. The centralized
Arthur Engel, received a visit from and socialized National Health Ser-
two American researchers, Odin W. vice, introduced by Aneurin Bevan in
Anderson from the University of the U.K. after World War II, should
Chicago and Olser L. Peterson from be contrasted with the American plu-
Harvard Medical School. Odin An- ralistic and mainly private system. In
derson, a sociologist known for na- this comparison, Sweden could offer
tionwide social surveys of medical a system with an intermediate posi-
costs and voluntary health insurance tion on the scale from central finan-
in the U.S., and Osler Peterson, a pio- cing and control as in England to the
neer in the study of health care quality American insurance based and multi-
through a classical study of general faceted health care system, sometimes
practitioners in North Carolina, sha- described as a “non-system”.
red an interest in what today is called
health systems research. They sug- Dr. Engel contacted Ragnar
gested to Dr. Engel an international Berfenstam, then newly appointed
comparative study of the health care professor of Social Medicine at the
systems in the U.S., U.K. and Sweden Uppsala University, and asked him
aiming at elucidating the effects of to take responsibility for the Swedish
Socialmedicinsk tidskrift 5 - 6/2010 439
tema
part of the study. He, in turn, drew some challenges as well as justifica-
Björn Smedby into the project at the tion for systems comparisons appear
same time as Odin Anderson recrui- consistent (Andersen 1976, Schiötz et
ted Ronald Andersen for the Amer- al. 2010). International comparisons
ican part, both to work as project are often used as arguments in the
directors for national health surveys health policy debate. Comparisons of
in the two countries using the re- health systems performance in differ-
sults as part of their doctoral theses. ent countries may generate hypothe-
Other persons were also involved in ses and explanations that can be used
our three-country project, which was for strategy development at national
the first or at least among the earliest level. Health policy norms and goals
comparisons of health care systems to must be grounded on reality and not
be undertaken, a field that later has only on theories and ideology. What
developed into a research field of its is possible to achieve in one country
own called health systems research. is at least reality based. Comparative
system studies can reveal successful
In this paper we build on the elements of one system that might
health care systems work of Ragnar be applicable in another and also
Berfenstam and his colleagues begin- problems that seem generic to most
ning in the 1960’s and the many de- systems that will be most difficult to
velopments and changes over the past solve everywhere (Anderson 1972).
40 or 50 years. We will comment on,
and document these changes with a The policy implications are perhaps
limited, selective literature review ba- more challenging from international
sed on published international com- comparisons than from correspond-
parisons. We depart from the model ing national studies. No health care
for analysis and understanding that system can be directly transferred
we employed in our early work – “the from one country to another, how-
original model” – and continue to ever, for historic, political and cul-
what we describe as “the developed tural reasons, but there are, of course,
model” which is our attempt to de- experiences in one country that could
scribe what can be said to be valid be thought-provoking for others and
for current international comparative provide a useful perspective of one’s
health care systems work. While we own health services system.
will not provide detailed comparisons
of international systems, we do plan Early international
to provide some examples to elucidate
comparisons of health
the original model and the developed
model. care systems
Odin Anderson was an original for-
Whether we consider the early ef- mulator of the concept of “health care
forts at international comparisons of systems” and the opportunities for
the 1960’s or the more current work, international comparisons of those
systems (Anderson 1963, 1967). His en countries (Kohn & White 1976).
thinking was central to the early com- A cadre of other researchers in the
parative studies of Sweden, the U.S. 1960’s were also at work on various
and the U.K. (Peterson et al. 1967, aspects of international health sys-
Pearson et al. 1968, Anderson 1972). tems comparisons. Some of these
We also performed two compar- were mainly descriptions of general
able social surveys in Sweden and the characteristics of health care systems
U.S. with the objective of revealing (Abel-Smith 1965, Follman 1963, Ho-
some of the structural, demogra- garth 1963, Mechanic 1968; Roemer
phic, social and biological variables 1963). Other work began to show
accounting for the striking internatio- patterns of use and expenditures as
nal differences in health services use related to other characteristics of the
(Andersen, Smedby & Anderson countries considered (Abel-Smith
1970). 1967, Andersen & Hull 1969, Bice &
Kalimo 1969, Lembcke 1959, Logan
Our Swedish-American compari- 1968, NCHS 1969, White et al. 1967).
son was the first based on national
representative samples. It should be The original model – com-
noted, however, that at same time a
ponents and measures
WHO supported study of medical
Figure 1 displays the health systems
care utilization was initiated in twelve
comparative model that guided our
study areas representing parts of sev-
pital beds and fewer admissions than Effectiveness (arrow 3) was consid-
the other countries and longer average ered by comparing the services the pop-
length of stay. ulation received to its mortality rates.
4. Outcomes 1. Efficiency
The original model restricted out- Average number of physician visits
come measures that might be related per person per year were much high-
to the health system to those associa- er in the U.K. than in Sweden and the
ted with mortality. Variables included U.S. but the number of physicians per
infant mortality, age specific mortal- 100,000 population was much low-
ity and life expectancy. Comparisons er in the U.K. than in the U.S. and
among the countries showed some was similar to the number in Sweden.
variation with Sweden tending to have These gross comparisons of resour-
the lowest mortality rates and the U.S. ces to volume of services were at least
the highest. suggestive that the English system
might be more efficient by these pro-
Some linkages among ductivity related measures (Anderson
1972).
the original model
components 2. Equity
The arrows in Figure 1 suggest the link- We found in our social survey
ages among the system components comparisons in 1963 that income was
considered in the original model and much more important in determining
the expected causal direction of these who received health services in the
linkages. We realize that the direction U.S. than in Sweden. Further, how
of the arrows in Figure 1 (and in Fig- people perceived their health (a mea-
ure 2 to follow) can ”go both ways” sure of need) was more closely related
and, empirically, establishing causality to the kinds and amounts of medical
will always be a challenge. However, care they received in Sweden than in
these directional arrows suggest the the U.S. These findings suggested
rationale for assessing and comparing more equity in the Swedish system
health care systems’ performance. than in the U.S. (Andersen, Smedby
& Anderson 1970).
Efficiency (arrow 1) – sometimes
referred to as productivity – of the 3. Effectiveness
When we compared medical care use
system was judged by comparing the (measured by ambulatory visits and
resources used by the system to the hospital admissions) we found the
kinds and volume of services it pro- United States to be highest on use but
vided. Equity (arrow 2) – or equality also highest on mortality rates. This,
– was assessed by comparing the ser- at least, calls into question the effec-
vices received by people of different tiveness of the U.S. system (Peterson
social and economic characteristics. et al. 1967). Of course, many other de-
Socialmedicinsk tidskrift 5 - 6/2010 443
tema
terminants of death rates need to be (Saltman 2002) and new aspects such
taken into account in trying to draw as quality of care and patient safety
conclusions about effectiveness of have been added to the comparisons
medical care systems. (OECD 2010a). The methods for mea-
suring performance, results and ef-
Conclusions regarding the ficiency of the systems have been
developed (Hollingsworth 2008,
original model Murray & Frenk 2000, Tandon et al.
One conclusion from our very first
2003). Model building has also pro-
comparisons (Peterson et al. 1967)
gressed (Andersen 2008).
was that there were good reasons to
ask if the three countries got value for
In 2000 WHO published its World
money for the resources they spent
Health Report 2000, a major effort to
on health care. For an answer to that
establish a comprehensive framework
question more profound studies were
for health systems performance as-
needed, based on better and more
sessment and to develop tools to mea-
comparable data. Our further studies
sure performance (WHO 2000). That
were a first step on this road. Another
report generated considerable media
conclusion was that the model that we
attention, not least because it ranked
used turned out to function well for
countries by health care performance,
analyzing and comparing health sys-
which evoked political protests from
tems in different countries.
low ranked countries. Also in acade-
mic journals the report was citicized
The development of for poor data quality, methodological
international health shortcomings and underlying ideo-
systems comparisons logy (Braveman et al. 2001, McKee
The development of international 2001, Navarro 2000, Williams 2001).
comparative health systems research As a result WHO established an inde-
since the 1960’s has moved towards pendent scientific peer review group
more comprehensive analyses that to review the techniques proposed by
has emphasized partly new aspects. the report. It was also recommended
The different components of the sys- not to rank countries in the future.
tem are described in more detail and
with additional new measures. The The positive effects of the report were
availability of statistical data has in- that it showed the difficulties in com-
creased greatly. It has been emphasi- paring health systems performance
zed that the health of the population and activated health services resear-
is determined by many other factors chers toward improved theory and
than the efforts of the health services. methods in health care systems com-
Inequality in health and utilization parisons. Of special importance was
has been increasingly observed (Fox the fact that it lead to a subsequent
1989, WHO 2008). The importance 900 pages book bringing together in
of health policy has been emphasized one place many of the debates and re-
444 Socialmedicinsk tidskrift 5 - 6/2010
tema
(Gerdtham & Jönsson 2000, Reinhart veloped model have greatly expanded
et al. 2004). Better definitions of ty- beyond medical care. Furthermore,
pes of hospitals and kind of health services may be specified as primary
personnel through the OECD System health care and specialist care, day
of Health Accounts (2000) have made care and other ambulatory care (Scho-
available statistics on these resour- en et al. 2009). Groupings of diagno-
ces more comparable (OECD 2010b, ses adapted for international compar-
Reinhardt et al. 2002). System resour- isons have been developed (ISHMT
ces now include health policy and 2008) as well as more homogeneously
policy implementation. Health policy defined surgical peocedures. One new
includes the laws, plans and protocols measure includes responsiveness of
made at the national or local level con- the system (Valentine et al. 2003).
cerning how the expenditures, per- Responsiveness measures the quality
sonnel and facilities of the health care of basic amenities of the system, its
system should be used. Measures of client orientation and to what extent
policy implementation have been in- patients’ needs and expectations are
corporated in the model because how met – beside the production of health.
policy is implemented or not imple- It assesses experiences patients actu-
mented can have substantial impact ally have with the system but is not
on how the system influences services the same as patient satisfaction. Other
provided and outcomes (Magnussen measures that are part of the compre-
et al. 2009, Saltman et al. 2007). hensive effort to include all major de-
terminants of health in a broader per-
2. Population characteristics spective of “health care system” are:
Important advances have been made (a) personal health practices – includ-
with respect to describing and mea- ing diet, exercise and self care; and
suring the predisposing socioecono- (b) public health practices – including
mic factors and their distribution in environmental health programs and
the population. Increasing attention population based treatment and pre-
is paid to measures of health beliefs vention activities.
as a predisposing population charac-
teristic (Salomon et al. 2003). These 4. Outcomes
beliefs include population values, at- In developing models traditional mor-
titudes and knowledge about health tality measures have been refined to
and the health care system. Measures better reflect the effect of medical
of population need have also been ex- care through the concept of avoid-
panded to include those provided by able or amenable mortality as has been
health professionals through tests and described more thoroughly by Wester-
examinations (evaluated need) as well ling in this issue (Charlton & Welez
as perceptions of patients. 1986; Nolte & McKee 2003; Nolte
& McKee 2008, Westerling 2010).
3. Use of services Outcome measures also emphasize
The services emphasized in the de- a much broader array of potential
446 Socialmedicinsk tidskrift 5 - 6/2010
tema
products of health care systems than same arrow in Figure 2 refers in a cor-
traditional measures of mortality and responding way to inequality in uti-
life expectancy. They emphasize that lization of care in relation to income
health care systems have responsibili- and other socioeconomic conditions
ties beyond quantity of life for quality (Rasmussen et al. 2004, van Doorslaer
of life as well. Studies include measu- et al. 1997, 2004, Whitehead et al.
res of functioning and disability and 1997).
combined measures such as quality-
adjusted life years (QALYs) and, at the In the developed model equity is es-
population level, disability-adjusted tablished according to two additional
life years (DALYs), disability-free distributions. One relationship (arrow
life expectancy (DFLE) and health- 4 in Figure 2) represents the extent to
adjusted life expectancy (HALE) which the system resources are equi-
(Mathers et al. 2003). The distribution tably distributed in the population
of health within populations is seen served according to predisposing and
as an increasingly important aspect enabling characteristics of that popu-
(Murray & Frenk 2000). Satisfaction lation, e.g., expenditures in the system
with services received from the health are equal for individuals of different
care system is also seen as an out- education and income levels control-
come measure (Blendon et al. 1990, ling for need (Andersen 2008). An-
Blendon et al. 2002). other equity relationship in the de-
veloped model (arrow 5 in Figure 2)
Some linkages among represents increasing efforts to inclu-
de in comprehensive systems analyses
components in developed the relationship between population
models characteristics and outcomes that
Additional linkages increasingly mod- might be independent of the medical
eled in comparative health systems care services people receive (Macken-
work include two associated with bach et al. 2008, Vågerö & Lundberg
equity and one with efficiency. They 1989, WHO 2008, Wilkinson & Pick-
have been shown as new arrows in ett 2009). There is a long tradition of
Figure 2. research linking social and economic
factors to mortality, morbidity and
1. Equity functioning, sometimes emphasi-
The equity linkages in both the orig- zing the importance of these factors
inal and developed model are based compared to the possible influence of
on some notion of “fair” distribution medical care (McKeown 1979). What
according to population characteris- is newer are the comprehensive ef-
tics. In the original model (arrow 2 forts to systematically partial out the
in Figure 1) equity was determined relative contributions of medical care
by the extent to which services were and socioeconomic characteristics
distributed according to the needs of (Murray & Evans 2003).
individuals in the population. The
tive for improving international com- Andersen R, Hull J. Hospital utilization and cost
parisons had been “only partially re- trends in Canada and the United States. Health
alized” (Murray & Evans 2003, p. 5). Services Research 1969; 4(3):198-222.
That certainly remains true today. We Andersen R, Smedby B, Anderson OW. Medical
concur that these remain important care use in Sweden and the United States – A
objectives in future work of inter- comparative analysis of systems and behavior.
Chicago: Center for Health Administation
national comparisons of health care Studies, University of Chicago, 1970. (Research
systems. We hope that future assess- Series No. 27).
ments of work in the field can con-
clude that the objectives have been at Anderson OW. Medical care: Its social and
organizational aspects. N Engl J Med 1963; 269:
least “more completely realized”. 839-843, 896-900.
OECD Health data. OECD 2010b. Schoen C, Osborn R, Squires D, Doty MM, Pierson
R, Applebaum S. How health insurance design
Pearson RJ, Smedby B, Berfenstam R, Logan RF, affects access to care and costs, by income, in
Burgess AM, Peterson OL. Hospital caseloads eleven countries. 2010 Commonwealth Fund
in Liverpool, New England, and Uppsala. International Health Policy Survey. Health
An internatoional comparison. The Lancet Affairs Web First, November 18, 2010.
1968;2(September 7):559-566.
Smith PC. Measuring and improving health-system
Peterson OL, Burgess AM, Berfenstam R, Smedby productivity. The Lancet 2010;376(9748):1198-
B, Logan RL, Pearson RJ. What is value for 2000.
money in medical care? Experiences in England
and Wales, Sweden and the U.S.A. The Lancet Tandon A, Lauer JA, Evans DB, Murray CJL. Health
1967;1(April 8):771-776. System Efficiency Concepts. In: Murray CJL,
Evans D (eds.). Health systems performance
Rasmussen NK, McCallum A, Haglund B. Equal assessment: Debates, methods and empiricism.
access to care. In: Health statistics in the Nordic Geneva: World Health Organization, 2003.
countries 2002. Copenhagen: NOMESCO 2004 (Chapter 50, pp. 683-692).
(pp.191-243).
Valentine N, de Silva A, Kawabata K, Darby C, Murray
Reinhardt UE, Hussey PS, Anderson GF. Cross- C, Evans D. Health systems responsiveness:
national comparisons of health systems Concepts domains and operationalization. In:
using OECD data, 1999. Health Affairs Murray CJL, Evans D (eds.). Health systems
2002;21(3):169-181. performance assessment: Debates, methods
and empiricism. Geneva: World Health
Reinhardt UE, Hussey PS, Anderson GF. U.S. Health Organization, 2003. (Chapter 43, pp. 573-596).
Care Spending in an International Context.
Health Affairs 2004;23(3):10-25.