WORKING PAPER
Disease Incidence and Mortality Among Older
Americans and Europeans
Aïda Solé-Auró, Pierre-Carl Michaud, Michael D. Hurd, and Eileen Crimmins
RAND Labor & Population
WR-1006
July 2013
This paper series made possible by the NIA funded RAND Center for the Study of Aging (P30AG012815) and the NICHD funded RAND
Population Research Center (R24HD050906).
RAND working papers are intended to share researchers’ latest findings and to solicit informal peer review.
They have been approved for circulation by RAND Labor and Population but have not been formally edited or
peer reviewed. Unless otherwise indicated, working papers can be quoted and cited without permission of the
author, provided the source is clearly referred to as a working paper. RAND’s publications do not necessarily
reflect the opinions of its research clients and sponsors. RAND® is a registered trademark.
Disease Incidence and Mortality Among Older Americans and
Europeans
Aïda Solé-Auró
INED- Institut National d’Études Démographiques
133 boulevard Davout, 75980 Paris cedex 20
Tel.(33) 01 56 06 43 10
Corresponding author: [email protected]
Pierre-Carl Michaud
Université du Québec à Montréal & RAND
Département des sciences économiques
Case postale 8888, Succ. Centre-Ville
Montréal, (Québec) CANADA H3C 3P8
Michael Hurd
RANDandNBER
1776 Main Street
Santa Monica, CA 90407
Eileen Crimmins
Andrus Gerontology Center
University of Southern California
3715 McClintock Ave, Room 218
Los Angeles, CA 90089-0191
Corresponding author:
Aïda Solé-Auró
INED- Institut National d’Études Démographiques
e-mail: [email protected]
Phone number: (33) 01 56 06 43 10
1
Abstract: Recent research has shown a widening gap in life expectancy at age 50 between the
U.S. and Europe, as well as large differences in the prevalence of diseases at these ages. Little is
known about the processes determining international differences in the prevalence of chronic
diseases. Higher prevalence of disease could result from either higher incidence or longer
disease-specific survival. This paper uses comparable longitudinal data from 2004 and 2006 for
populations aged 50 to 79 from the U.S. and a selected group of European countries to examine
age-specific differences in prevalence and incidence of heart disease, stroke, lung disease,
diabetes, hypertension, and cancer as well as mortality associated with each disease.
Not
surprisingly, we find that Americans have higher disease prevalence. However, incidence of
most diseases and survival conditional on disease is higher in Europe at older ages, in particular
after age 60. The survival advantage in Europe tends to disappear when we control for comorbidities but does not suggest a survival advantage in the U.S. Therefore, the origin of the
higher disease prevalence at older ages in the U.S. is to be found in higher incidence and
prevalence earlier in the life course.
Keywords: disease incidence, disease-specific survival, Europe, United States.
2
Introduction
Recent evidence indicates that the prevalence of chronic diseases in old age is higher in the
United States (U.S.) than in Europe (Banks et al. 2006; Thorpe et al. 2007; Crimmins et al.
2010b). Americans have also been shown to have lower life expectancy at age 50 than most
European countries with the gap growing over time (Glei et al. 2010). The interpretation of higher
national levels of disease prevalence in the U.S. is not clear; however, as higher prevalence could
result from either higher disease incidence or longer survival with health conditions.
The
interpretation of the meaning of the differences would be very different depending upon the
underlying processes that prevail. Little research has decomposed prevalence at older ages into
these components across countries and by age groups.
Cause specific mortality rates show that differences in heart disease mortality account for
about half of the differential between the U.S. and other high mortality countries, differences in
mortality due to diabetes also account for some of the differences. On the other hand, the U.S.
does better in cancer mortality (Glei et al. 2010). However, the overall differences in mortality
rates by cause do not provide information on how the relative likelihood of dying differs across
countries among those with specific conditions because the rates are relative to the total
population and thus overall mortality differences by country do not give a complete picture of the
dynamics involved.
Another important shortcoming of using mortality and prevalence data to understand health
processes is the effect of heterogeneity or selection. For example, incidence rates could be
invariant with age but differ across individuals. Because disease incidence is positively related to
3
mortality, average incidence rates of survivors would decrease with age, creating a spurious
negative age pattern. Omitting heterogeneity biases inferences on the age pattern of incidence
and mortality rates because of the potential dynamic selection of the fittest or less risk-prone
groups over time (Manton et al. 1986). If the selection process differs across countries, incorrect
inferences could be drawn from differences in average incidence and mortality rates. Another
important consequence of heterogeneity is that if disease incidence varies with age differentially
across countries, it is possible to observe a negative relationship between current disease
prevalence and incidence across countries. If the frailty distribution is the same across countries
but incidence rates are higher in one country than another, the remaining pool of undiagnosed
individuals will be at higher risk in the low incidence country compared to the high incidence
country. This is one of the explanations for why the U.S. may have a survival advantage past age
80 relative to European countries (Manton and Vaupel 1995) or why there is mortality crossover
by race in the U.S. at older ages (Lynch et al. 2003).
Evidence on differences in incidence is scarce, in part because of lack of comparable
longitudinal data across countries. Aggregate computation of incidence rates from prevalence by
age is complicated by the fact that individuals can have several diseases. One exception is Banks
et al. (2010) who compare the U.S. and England, and find generally higher disease incidence rates
post age 50 for Americans, caused not only by higher past cumulative disease risk, but also
higher new onset of diseases and higher survival. However, it is unclear whether these patterns
hold in other European countries where even larger prevalence differences are observed.
Furthermore, the comparisons of incidence in Banks et al. (2010) focus on aggregate differences
across age groups. In light of potential differences in conditional incidence and mortality rates,
we investigate whether aggregate differences are still observed when we control for selection.
4
This paper uses longitudinal data for populations aged 50 to 79 for the U.S. and five European
countries to examine age-specific differences in disease incidence and disease-specific mortality
from 2004 to 2006. We focus on the most common and frequent health conditions, which are
major causes of mortality at older ages: heart disease, stroke, lung disease, diabetes, hypertension,
and cancer. We investigate incidence of these conditions using survival models and include
indicators of socio-demographic factors and health behaviors to determine how these factors
relate to disease onset and how differences affect the differences between Europe and the U.S.
We then investigate survival conditional on the presence of these conditions.
DataandMethods
This analysis uses two longitudinal datasets designed to be comparable: the Health and
Retirement Study (HRS) and the Survey of Health, Ageing and Retirement in Europe (SHARE)
collected in 2004 and 2006 for individuals aged 50 to 79. HRS is an ongoing longitudinal survey
of the U.S. population age 50+ begun in 1992 and regularly refreshed at younger ages with new
cohorts.1 SHARE began collecting data in 2004 on the age 50+ non-institutionalized population
with a target sample size of 2,000 respondents per country for 12 European countries. In our
analysis we use a subset of five countries: Denmark, France, Italy, the Netherlands, and Spain.
We focus on those countries because of their relatively high response rate in the SHARE baseline
1
The original sampling frame for each entering HRS cohort is the non-institutionalized population; however,
respondents who moved to nursing homes are retained in the sample (HRS, 2011). We eliminated nursing home
residents in order to be comparable to the SHARE sample.
5
interview and of a close match between country-specific life table mortality and observed
mortality between waves 1 and 2 of SHARE. We will refer to these five countries as “Europe.”2
For both the U.S. and Europe we use 2004 as the base year and examine health transitions
occurring in the interval between the initial interview and re-interview in 2006. Table 1 shows
the national sample sizes for those aged 50-79 in 2004 and indicates the 2006 interview status of
2004 respondents: interviewed, died, or lost to follow-up. Sample size in 2004 was more than
16,000 individuals in the U.S. and over 11,000 individuals in Europe (Table 1). The HRS
response rate is higher at follow-up in 2006. We discuss further the issues of non-response and
evaluate the completeness of mortality information below. The number of known deaths between
the two waves is higher for HRS (2.9%) than for SHARE (2.0%). Analyses are weighted using
sample weights.
[Insert Table 1 about here]
Evaluation of Mortality Reports and Effect of Differential Response Rates
We compared mortality rates estimated from survey data to life-table mortality rates calculated
from the Human Mortality Database (HMD) data for 2004.3 Two-year mortality rates from the
life-tables weighted by gender are computed to match the average time interval between
interviews. Results indicate that survey estimates are relatively close to the life table estimates
between ages of 50 and age 80 for the U.S., the Netherlands, Spain, Italy, France and Denmark
2
Almost none of the SHARE respondents are living in nursing homes; we eliminate the small numbers who are. For
most of the sample, records are matched to the National Death Index for identification of deaths; in addition, deaths
are reported in exit interviews.
3
For most of the HRS sample, records are matched to the National Death Index for identification of deaths; in
addition, deaths are reported in exit interviews. In the SHARE data, deaths are reported by survivors in exit
interviews.
6
and that these countries represent well the entire set of SHARE countries in terms of life
expectancy (see appendix). Mortality above age 80 estimated from the survey is somewhat lower
than that in the life tables.4 For this reason, we limit our analysis to the age-range 50-79.
Figure 1 shows 2-year mortality rates in the U.S. and in the pooled 5 European countries
(weighted by population size) from both the surveys and the HMD. The survey estimates for the
U.S. are close to, but slightly higher, than the life table death rates at all ages while the European
survey estimates are slightly below the life table estimate. There are multiple reasons why this is
the case.
SHARE is a new survey; while the HRS is a long-running longitudinal survey.
Respondents die on average a year after the initial survey date, so people who are already very ill
may not enter the initial wave of a survey but if the survey is long running, they may be more
likely to be included.5 Because few people are in nursing homes prior to age 80, and our analysis
is limited to those 50-79, this difference in population coverage is not likely to explain the higher
mortality risk in the U.S.6
[Insert Figure 1 around here]
Non-response is an issue in all longitudinal surveys and, as indicated above, is higher in
SHARE than in HRS (about 5.6% lost to follow-up in HRS versus 34% in the 5 SHARE
countries). The fact that mortality tracks life-tables relatively well shows that follow-up
differences are unlikely to explain the mortality differences we find. However, differential
attrition could still have an effect on incidence rates. We have analyzed the determinants of nonresponse in SHARE and developed weights for non-response, which we compared to the weights
4
This discrepancy at older ages is likely due to the exclusion of nursing home residents.
Adams et al. (2003) show that the match to life-table mortality for older HRS respondents (age 70+ in 1993)
improves in later waves after the first follow-up.
6
In 2004, the fraction of the population in a nursing home between age 50 and 80 in the U.S. is 0.8%. The mortality
rate between interviews from 2004 to 2006 interview was 37% for nursing home residents compared to 3.2% for the
non-nursing home population. The actual aggregate mortality rate is 3.5% so the inclusion of the nursing home
population would add 0.3% to the aggregate mortality rate for this age group.
5
7
provided with the survey data. We conclude that differential non-response is accounted for in the
SHARE weights and unlikely to bias the inferences we make on the two populations (for more
details, see appendix).
Measuring Incidence of Disease
Disease presence in both SHARE and HRS is based on self-reports of major chronic conditions:
heart disease, stroke, lung disease, diabetes, hypertension, and cancer. The surveys were
specifically designed to harmonize their instruments so that in most cases, questions were worded
identically. Information on the presence of six diseases is reported in response to the question:
“Has a doctor ever told you that you had any of the following conditions? Each condition is then
asked about separately. It is hard to determine how diagnostic and reporting differences could
affect national differences, but Banks et al. (2006) showed that differences in self-reports of
disease between England and the U.S are not driven by differences in reporting. They show that
the differences are generally confirmed by biomarkers when self-reports and relevant biomarkers
can be compared. For cancer, more aggressive diagnostic procedures in the U.S. might explain
some of the higher prevalence observed in the U.S. (Crimmins et al. 2010b).
While the initial assessment of disease presence was identical in HRS and SHARE,
follow-up questions and procedures differed somewhat. In the 2006 HRS, responses provided at
earlier waves about disease presence are pre-loaded. Those who did not have a condition in the
previous wave were asked has a “doctor ever told you” you have a disease; a positive response is
used to construct our measure of incidence. Those who previously said that they had a condition
were reminded of this at the 2006 interview. Sometimes people deny that they have reported the
8
presence of disease. People who denied having a condition in the past were then asked about their
present state and they were retroactively recoded as not previously having the disease at the
earlier wave. This could lower U.S. incidence rates relative to European ones if these false
positives are corrected as in 2006 SHARE did not allow respondents to dispute a condition
reported in 2004.
Respondents in SHARE who did not have a condition in 2004 were asked in 2006 “has a
doctor ever told you that you had/do you currently have any of these conditions? With this we
mean that a doctor has told you that you have this condition and that you are either currently
being treated for or bothered by this condition”. On one hand, this additional condition of being
either treated or bothered could deflate the number of incident cases as respondents who had a
condition which does not bother them and for which they are not being treated will be counted as
undiagnosed. However, since new conditions are likely to be treated and bother the respondent, it
is unlikely to affect the numerator of incidence rates very significantly. However, this difference
in wording would tend to lower SHARE incidence rates relative to HRS incidence rates. Overall,
denials in HRS and the additional requirement of being bothered/treated in SHARE have opposite
effects on the gap between SHARE and HRS incidence rates.
Other Variables
Because selection may affect the age patterns of survival and incidence we uncover, we control
for socio-demographic characteristics and health behaviors in most of the analysis. Sociodemographic controls for age, gender and education are included. The education variable is
constructed as: less than some college (13 or less years of schooling) and some college or more
9
(more than 13 years of schooling). We also include measures of overweight and smoking in our
regression analyses. Smoking is self-reported as being a former smoker, current smoker, or never
smoker. Weight classes are determined using self-reports of height and weight provided in both
surveys. These are converted into body mass index (BMI) and classified into two groups: class II
obesity and higher (>=35), and those with lower BMI. While weight is notorious for being
underreported and height is overestimated by older respondents (Burkhauser and Cawley 2008),
Michaud et al. (2007) note there is no indication that biases are different across countries
meaning that even if there is an underestimation of obesity rates, the relative ranking of countries
is likely to remain unaffected. We use this high level of obesity in or regressions as mortality risk
is higher at this level of obesity (Alley et al. 2010).
Methods
We first examine descriptive data on prevalence and incidence of diseases and mortality by age
groups in Europe and the U.S. Then we examine two-year disease incidence rates among those
who did not have each condition at the initial interview. We estimate incidence using hazard
models for each of the six diseases using a set of pooled samples for the U.S. and Europe for
those without specified diseases at the initial interview. Model 1 includes an indication of being
in the European sample and controls for age and gender; health behaviors and education are then
added to the model (model 2). The Weibull distribution is used for age in these analyses to
address the non-monotonic pattern of onset with age observed in the descriptive data. This
approach allows us to determine whether there are differences between Europe and the US in
10
incidence rates and whether these are eliminated or reduced when health behaviors and education
are controlled.
To study differences in mortality rates by disease across countries, we estimate survival
models pooling U.S. and European data for those with specified diseases at the initial interview.
Four models are presented: Model 1 includes an indicator of being in the European sample and
controls for age and gender; Model 2 adds controls for college education; Model 3 adds controls
for health behaviors; and model 4 adds controls for co-morbidities from each of the other
conditions. In estimating the survival models, we take account of right-censoring as well as
differential exposure time between interviews. Although average time between interviews does
not vary much between the U.S. and Europe there is considerable heterogeneity within countries.
Weights are used in estimating the models. All analyses are conducted using STATA statistical
software version 12.0.
Results
Prevalence and Incidence
We document differences in the prevalence of diseases and also obesity for the entire sample
aged 50-79 in Figure 2. As expected, the prevalences of disease and obesity are significantly
higher in the U.S. than in Europe for each of these conditions. Close to one fifth (18.3%) of
Americans have heart disease in this age range compared to one-tenth (10.4%) of the Europeans.
Prevalence of most conditions is between 40% and 100% higher in the U.S. compared to Europe.
[Insert Figure 2 here]
11
In Table 2, we show the prevalence of disease in the age range 50-54 as well as incidence
rates in five age groups from 50 to 79 for individual diseases and obesity in the U.S. and Europe.
Focusing first on the prevalence at the initial age group in our sample, we observe that levels are
consistently higher for the U.S. and that the differences are large for most conditions at age 50.
The differences are smaller for stroke, lung conditions and cancer. This clarifies that part of the
difference in disease prevalence at older ages begins at ages well before age 50.
[Insert Table 2 here]
A review of age specific incidence rates in Table 2 and Figure 3 shows that incidence rates for
many conditions post age 50 are worse in Europe. In fact, overall incidence levels for the diseases
are higher in Europe. For example, average incidence of heart disease in Europe post age 50 is
4.47% compared to 3.91% in the U.S. Stratifying by age shows that incidence rates are
significantly higher in Europe after age 60 for many causes.
For several conditions, the
differences increase with age. This is the case of lung disease, diabetes, hypertension and to some
extent for stroke. This provides evidence of an initial delay in the age at onset of these conditions
in Europe; but then acceleration in onset at later ages.
On the other hand, incidence rates for cancer are higher in the U.S. than in Europe post age
50. We have reason to believe that this is related to more cancer screening in the U.S.,
particularly at older ages (Crimmins et al. 2010a). As expected, the incidence rate for obesity is
higher in the U.S. than in Europe at all ages.
[Insert Figure 3 here]
Part of the story for these differences in age patterns of disease may have to do with
unobserved frailty in the population at risk. As the prevalence of a condition increases, the
undiagnosed population is composed of more people with less inherent risk for the condition, as
those most susceptible have had earlier incidence. Since, the U.S. prevalence is higher at the
12
earlier ages, the undiagnosed population in the U.S. at older ages may be less at risk than the
undiagnosed population in Europe. This may lead to a negative relationship between prevalence
and incidence.
Because aggregate age patterns can be misleading due to selection, we analyze survival
models which control for socio-demographic and health behavior differences. We assume the
dependence on age is captured by a Weibull form7. Adjusting for demographic status, the onset of
lung disease and hypertension is higher for Europeans compared to the U.S; the onset of diabetes
and cancer is lower in Europe than in the U.S.; and there are no significant differences in the
onset of heart disease and stroke between Europe and the U.S. (model 1). There is a positive
relationship with age and the onset of all conditions. Males are also more likely to experience the
onset of most conditions.
[Insert Table 3 here]
When having some college education and health behaviors are added to the model (model
2), Europeans only have significantly higher onset of lung disease; they have lower onset of heart
disease and cancer; and there are no significant differences between Europe and the U.S. in the
onset of stroke, diabetes, and hypertension. This means that the lower incidence of diabetes and
the higher incidence of hypertension among Europeans without these controls is linked to the
educational and behavioral compositions of the populations. Significant higher disease onset for
males persists for heart disease, stroke, diabetes and cancer; and appears for hypertension.
Having some college education is negatively related to the onset of lung disease, diabetes and
hypertension.
7 An age parameter greater than 2 indicates a rising age profile; a parameter between 1 and 2 indicates a positive
relationship with age; and if the parameter less than one there is a negative relationship with age.
13
Former and current smokers have a significantly higher likelihood of lung disease onset.
Being a current smoker is related to a higher likelihood of cancer. The effect of obesity on the
risk of onset for heart disease, lung disease, diabetes, hypertension, and cancer is strong.
Mortality
We now focus on mortality among those with diseases. Indeed, if individuals with a condition
survive longer in one country compared to another, prevalence may be higher without higher
incidence of disease. We report the proportion dying by age among those with each of the six
diseases at the initial interview in Table 4: heart disease, stroke, lung disease, diabetes,
hypertension, and cancer. We examine the significance of the difference in mortality by age
between Europe and the U.S. We then compute the ratio of the mortality rate for those with a
condition in the U.S. divided by the mortality rate for those with a condition in Europe. For all
diseases, we find higher mortality risks in the U.S. than Europe. For stroke, lung disease,
diabetes, and hypertension the ratio is highest at age 50-59 indicating that the mortality
differences are greater in younger old age.
[Insert Table 4 here]
The mortality differences may be at least partly explained by the fact that Americans are more
likely than Europeans to have multiple chronic diseases (Crimmins et al. 2010b); that Americans
are more obese; and have a different smoking history. In order to test the possibility that these
factors explain the differences in mortality from individual diseases, we estimate survival models
with a Gompertz specification among those with each disease using a pooled sample of
Americans and Europeans. After the initial model which includes age, sex, and being in the
14
European sample, sequential models control for education, current and past smoking experience,
obesity, and the prevalence of each of the other five health conditions (Table 5).
[Insert Table 5 here]
The relative risk of mortality for respondents with four diseases - heart disease, lung disease,
diabetes, and hypertension - is significantly lower in Europe than the U.S. controlling for age and
gender (Table 5, Model 1). For stroke and cancer there is no significant difference in the relative
risk of mortality between Europe and the U.S. Controls for higher education reduce the size of
the difference between Europe and the U.S. in mortality risk among those with hypertension, but
the difference remains significant. The size of the mortality differences for those with other
conditions does not change much with education controls (Table 5, Model 2). When health
behaviors are introduced in model 3, the size of the coefficients indicating the effect of being in
the European sample are increased substantially meaning that if Americans and Europeans were
alike in smoking and obesity, Europe would have even lower relative mortality from all diseases,
except cancer.
Finally, our last model adds controls for co-morbidity from other health
conditions. With these controls, the significance of the mortality differences between Europe and
the U.S. vanishes for stroke, lung disease, and diabetes. For heart disease and hypertension the
risk of mortality remains significantly lower in Europe than in the U.S., but the size of the
difference is halved. This indicates that much of the higher mortality from disease in the U.S.
comes from the complications of co-morbidity.
Discussion
15
Our results reaffirm that disease prevalence is significantly higher in the U.S. than in Europe, and
that this difference begins well before age 50. Higher American prevalence in late middle age is
particularly large for heart disease, diabetes, hypertension and obesity. Overall incidence rates
for the diseases are higher in Europe; however, when age-stratified incidence is examined,
significantly higher incidences in Europe are found for many conditions after age 60. Age
specific incidence of diabetes, lung disease and hypertension are generally higher in Europe, with
the exception of diabetes at younger old ages. This may indicate that Americans have a different
age patterning of chronic conditions with onset of chronic conditions earlier in the life cycle. This
may be related to the high levels of obesity and past smoking. It is possible that after the younger
old ages those most susceptible to the condition have been eliminated from the risk group and the
undiagnosed population in the U.S. at older ages may be at less risk than the undiagnosed
population in Europe. On the other hand, the higher prevalence of cancer and obesity in the U.S.
is accompanied by higher incidence at all ages. As we have indicated, cancer screening is very
different in the U.S. and Europe, especially at older ages (Garcia and Crimmins 2013).
Documenting higher incidence of obesity across the ages for the U.S. is an important addition to
our knowledge of the process leading the U.S. to rank as one of the most obese countries
internationally (Alley et al. 2010). Where there are significant differences in incidence of heart
disease and stroke, Europeans have higher incidence. While the incidence of these conditions is
much more similar between Europe and the U.S. than for the other conditions; the pattern of
slightly higher incidence for the U.S. at younger ages and somewhat higher incidence for
Europeans at older ages is also seen in these conditions. When the determinants of disease
incidence are estimated using survival models, there are some changes in the differences
observed for the U.S. and Europe. We should note that our findings differ from Banks et al.
(2010) who found higher disease incidence to be higher among Americans compared to persons
16
in England; however, other examinations of differences across countries have found health
among the English to be more similar to that of Americans than that of continental Europeans
(Crimmins et al. 2010a).
We found higher mortality rates in the U.S. for stroke, lung disease, diabetes, and
hypertension at younger old ages (50-59). Americans with most diseases have higher mortality;
however, the higher mortality from a number of conditions disappears when the higher comorbidity of American is controlled. What remains is higher American mortality among those
who have hypertension and heart disease. Heart disease mortality is the major cause of the
European-American differential in life expectancy at age 50, and we are not able to explain the
higher mortality of Americans from heart disease by differences in co-morbidity, education, or
health behaviors.
One motivation for studying disease incidence was to examine the contributions to greater
prevalence of disease at older ages in the U.S.: initial conditions (disease at age 50), incidence
between age 50 and those older ages, and survival conditional on having the disease. However, at
least qualitatively, the relationships among initial conditions, incidence, survival and subsequent
prevalence are not coherent. Table 6 summarizes data from Tables 2 and 4 and Figure 2 for six
diseases. Consider, for example, heart disease. At age ages 50-54 prevalence in the U.S. and
Europe is about nine and four percent respectively. Incidence in the U.S. is lower in the age
range 50-79 which should lead to a reduction in the prevalence gap. Mortality conditional on
having heart disease is higher in the U.S. which should also reduce the gap. Yet, prevalence age
50-79 in the U.S. is about twice as high as prevalence in Europe. The relationship between
prevalence, incidence and mortality is approximately the same for the other five diseases.8 One
possible explanation is that the processes are not stationary: there may be cohort effects in
8
We have verified this lack of coherence quantitatively via simulation.
17
prevalence at age 50 and variation over time in both incidence and survival conditional on
disease. If this is the case, it has important implications for studying the effect of policy on
health: past policy and past population characteristics influenced past incidence in a way that is
not observed in current incidence. Thus the relationship between current prevalence and current
policy is mediated in a complex manner by historical variation in incidence.
[Insert Table 6 here]
We can use the observed prevalence rates for people in their 50s and simulate what would
happen to the population prevalence of disease in Europe and the U.S. by age 75 assuming the
observed incidence and the observed mortality. The implication of the calculations is that the
U.S. would continue to have a higher prevalence of heart disease and cancer at older ages;
however, Europe would exceed the U.S. by age 75 in diabetes, stroke, and hypertension. Europe
would exceed the U.S. in the prevalence of lung disease by age 65. The difference in lung
disease is not surprising given the different cohort patterns of smoking; and it is possible that the
other implied changes result from the timing of both smoking changes and other cohort changes
such as obesity.
There are some limitations in the present analysis that could affect our findings. SHARE
under-estimates population mortality, whereas HRS mortality is very close to life table mortality.
SHARE is a new survey: people who are already very ill, and have an elevated likelihood of
dying between the first two waves, may not enter the initial wave of a survey. The HRS is a
long-running longitudinal survey so the very ill are more likely to be included. Adams et al.
(2003) show that the match to life-table mortality for older HRS respondents (age 70+ in 1993)
improves in later waves after the first follow-up. Because people who are already very ill may
not enter the initial wave of a survey, but will be there if the survey is long running, the level of
disease and mortality in SHARE might be lower than would be true with the unhealthy fully
18
represented. If survey mortality rates were higher and closer to the life table mortality rates for
the European population, survival rates conditional on having a disease would be higher and
closer to those in the U.S. Additionally, because SHARE respondents who were neither bothered
nor treated for a condition may have failed to report it, incidence in SHARE might have been
even higher than we observed.
Because we do not consider the nursing home population, our results do not reflect the
entire population. The nursing home population in both the United States and in Europe would
contain persons with relatively high risk of mortality and morbidity. Nursing home residence is
generally relatively low in these countries at ages less than 80. While it is often hard to find data
on nursing home population by age, in the HRS the nursing home population is 0.8% of the
sample at ages 50-80. In Europe estimates for this age range are not readily available. Estimates
among those in similar age ranges indicate relatively low nursing home usage for two of our
countries: the Netherlands (2% of those 65-80) and Denmark (1% of persons 60-79).
Usage
appears somewhat higher for France (6% for those 65-80). In the entire 65+ population nursing
home usage is low in Italy (2%) and higher in Spain (5.8%). In the U.S. the percentage in nursing
homes
in
the
total
65+
population
is
5%
(http://www.cdc.gov/HomeandRecreationalSafety/Falls/nursing.html). We believe that because
the percentages in nursing homes are relatively similar in Europe and the U.S. differential use of
nursing homes is unlikely to influence the differences we observe. The combination of effects
from these data issues might result in a slight increase of the prevalence rates for the European
population; however, it is unlikely that this would be sufficient to change the direction of our
findings. In addition, we used data for a relatively short time interval from both surveys. Using
data for more waves over a longer time interval would add stability to some estimates, and would
verify the direction of our estimates.
19
While we recognize these limitations, the analysis presented here is an important first step in
understanding the complex health and survival differences between Europe and the U.S. Overall,
the higher prevalence of health conditions in the U.S. does not result from a higher incidence of
diseases across the age range; and where there are differences in incidence, they are sometimes
explained by compositional differences. Higher American prevalence does not result from higher
survival with disease as mortality among those with disease is higher in the U.S. than Europe.
The need to study mortality within individual persons rather than by disease is emphasized by the
fact that higher American mortality among those with disease is partly explained by the higher
comorbidities of Americans.
20
Acknowledgements
The authors acknowledge funding from NIA R01 AG040176-02, funding provided from the
Spanish Ministry of Science and Innovation (ECO2010-21787-C03-01), and the Beatriu de Pinós
grant 2010-2012. An earlier version of this paper was presented at the annual meeting of the
Population Association of America, held in San Francisco, May 3-5, 2012. The SHARE data
collection has been primarily funded by the European Commission through the 5th framework
program (project QLK6-CT-2001-00360 in the thematic program Quality of Life). Further
support by the European Commission through the 6th framework program (projects SHARE-I3,
RII-CT-2006-062193, as an Integrated Infrastructure Initiative, COMPARE, CIT5-CT-2005028857, as a project in Priority 7, Citizens and Governance in a Knowledge Based Society, and
SHARE-LIFE (CIT4-CT-2006-028812)) and through the 7th framework program (SHAREPREP (No 211909) and SHARE-LEAP (No 227822)) is gratefully acknowledged. Substantial cofunding for add-ons such as the intensive training program for SHARE interviewers came from
the U.S. National Institute on Aging (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30
AG12815, R21 AG025169, Y1-AG-4553-01, IAG BSR06-11 and OGHA 04-064). Support for
the HRS data collection was primarily provided by the National Institute on Aging (U01
AG009740).
21
References
Adams, P., Hurd, M. D., McFadden, D., Merrill, A., & Ribeiro, T. (2003). Healthy, wealthy, and
wise? Tests for direct causal paths between health and socioeconomic status. Journal of
Econometrics, 112(1), 3-56.
Alley, D., Lloyd, J., & Shardell, M. (2010). Can Obesity Account for cross-national differences in
Life Expectancy trends? In E. M. Crimmins, S. H. Preston, & B. Cohen (Eds.),
International Differences in Mortality at Older Ages: Dimensions and Sources. : Panel on
Understanding Divergent Trends in Longevity in High-Income Countries National
Research Counci. The National Academies Press.
Banks, J., Marmot, M., Oldfield, Z., & Smith, J. P. (2006). Disease and disadvantage in the
united states and in england. JAMA: The Journal of the American Medical Association,
295(17), 2037-2045, doi:10.1001/jama.295.17.2037.
Banks, J., Muriel, A., & Smith, J. P. (2010). Disease prevalence, disease incidence, and mortality
in the United States and in England. Demography, 47, 211-231.
Burkhauser, R. V., & Cawley, J. (2008). Beyond BMI: The value of more accurate measures of
fatness and obesity in social science research. Journal of Health Economics, 27(2), 519529, doi:10.1016/j.jhealeco.2007.05.005.
Crimmins, E. M., Garcia, K., & Kim, J. K. (2010a). Are International Differences in Health
Similar to International Differences in Life-Expectancy?,. In E. M. Crimmins, S. H.
Preston, & B. Cohen (Eds.), International differences in mortality at older ages:
Dimensions and Sources. . Washington, DC: Panel on Understanding Divergent Trends in
Longevity in High-Income Countries. The National Academies Press.
Crimmins, E. M., Preston, S. H., & Cohen, B. (2010b). International differences in mortality at
older ages: dimensions and sources. Washington, DC: National Academies Press.
Glei, D. A., Meslé, F., & Vallin, J. (2010). Diverging Trends in Life Expectancy at Age 50: A
Look at Causes of Death. In E. M. Crimmins, S. H. Preston, & B. Cohen (Eds.): In
International Differences in Mortality at Older Ages:Dimensions and Sources. Panel on
Understanding Divergent Trends in Longevity in High-Income Countries National
Research Council. The National Academies Press.
Lynch, S. M., Brown, J. S., & Harmsen, K. G. (2003). Black-White Differences in Mortality
Compression and Deceleration and the Mortality Crossover Reconsidered. Res Aging,
25(5), 456-483.
Manton, K. G., Stallard, E., & Vaupel, J. W. (1986). Alternative Models for the Heterogeneity of
Mortality Risks Among the Aged. Journal of the American Statistical Association,
81(395), 635-644.
Manton, K. G., & Vaupel, J. W. (1995). Survival after the Age of 80 in the United States,
Sweden, France, England, and Japan. New England Journal of Medicine, 333(18), 12321235, doi:10.1056/nejm199511023331824.
Michaud, P.-C., Kapteyn, A., Smith, J. P., & van Soest, A. (2011). Temporary and permanent
unit non-response in follow-up interviews of the Health and Retirement Study.
Longitudinal and Life Course Studies vol.2 (2011) nr.2 p.145-169.
Michaud, P.-C., Soest, A. v., & Andreyeva, T. (2007). Cross-country variation in obesity patterns
among older Americans and Europeans. Program for Research on, Social Economic
Dimensions of an Aging, Population.
22
Thorpe, K. E., Howard, D. H., & Galactionova, K. (2007). Differences in disease prevalence as a
source of the U.S.-European health care spending gap. Health affairs (Project Hope),
26(6).
23
Tables
Table 1: Sample sizes (2004) and Respondent Status (2006) in 5 European
Countries (SHARE) and the U.S. (HRS); ages 50-79.
Country
N (2004)
Denmark
France
Italy
Netherlands
Spain
Total European Countries
United States
Total
1,428
2,697
2,323
2,619
2,059
11,126
15,890
27,016
% Responded
(2006)
74.8
63.2
68.1
63.1
57.7
64.0
91.5
79.6
Died
% Died
% Lost
36
52
38
45
48
219
529
748
2.4
1.9
1.8
1.7
2.4
2.0
2.9
2.5
22.8
34.9
30.2
35.2
39.9
34.0
5.6
17.9
Source: SHARE and HRS 2004, 2006.
24
Table 2: Prevalence and Incidence by Age in the U.S. (HRS) and 5 European Countries (SHARE).
Individuals aged 50-79
Panel A: U.S. Prevalence Age 50-54 and Incidence rates by age
Heart disease Stroke Lung Disease
Diabetes Hypertension Cancer
AGE
9.02
2.21
4.98
11.04
32.71
4.85
Prevalence 50-54
2,812
2,812
2,812
2,812
2,808
2,811
N 50-54
Incidence
50-54
2.18
0.65
1.04
3.07
8.11
1.24
55-59
3.30
0.93
1.59
3.99
9.65
1.99
60-64
3.89
0.80
1.73
4.27
11.71
2.63
65-69
5.03
1.22
1.57
3.60
11.49
3.20
70-74
5.33
2.56
2.06
3.19
12.28
3.37
75-79
7.24
3.16
2.71
3.08
11.44
4.20
Total Incidence
3.91
1.28
1.65
3.59
10.18
2.45
(50-79) HRS
N 50-79
11,559
13,604
13,263
11,939
7,053
12,801
Panel B: 5 Country European Prevalence Age 50-54 and Incidence rates by age
Heart disease Stroke Lung Disease
Diabetes Hypertension Cancer
AGE
Prevalence 50-54
4.31
1.19
3.31
5.11
18.00
3.34
1,848
1,848
1,848
1,848
1,848
1,848
N
Incidence
50-54
1.81
0.61
1.55
3.46
9.61
0.64
55-59
2.09
0.81
2.51
3.33
13.32
1.54
60-64
5.06
0.93
3.90
4.16
17.00
1.48
65-69
5.80
1.66
3.46
4.94
17.67
1.84
70-74
9.0
3.35
6.40
4.83
24.12
1.85
75-79
6.43
4.36
5.81
5.48
19.42
2.64
Total Incidence
(50-79) SHARE
4.47
1.65
3.57
4.16
15.51
1.55
N 50-79
6,417
6,929
6,702
6,456
4,911
6,783
Obese >=35
14.06
2,808
2.78
3.23
2.85
3.18
1.57
0.84
2.59
12,675
Obese >=35
6.02
1,878
1.70
1.37
1.39
1.32
1.31
0.94
1.38
6,723
Source: HRS 2004 - 2006. HRS weights are used in this table. SHARE, 2004 – 2006. SHARE weights are used in this table. Europe
includes: Denmark, France, Italy, the Netherlands, and Spain. The significance of differences is shown in Figure 3.
25
Table 3: Disease Incidence Model Estimates. Individuals aged 50-79.
Heart disease
Model 1
Model 2
Socio-demographic
Europe
Age
Male
College
Health behaviors
Former smoker
Current smoker
BMI>35
Constant
Socio-demographic
Europe
Age
Male
College
Health behaviors
Former smoker
Current smoker
BMI>35
Constant
-0.09
1.57 ***
0.22 ***
-0.17 *
1.66 ***
0.17 **
-0.10
Stroke
Model 1
Model 2
-0.05
1.69 ***
0.36 ***
-0.10
1.67 ***
0.27 **
-0.08
-0.06
0.20
0.48 ***
-5.85 ***
0.02 ***
Diabetes
Model 1
Model 2
-0.12
0.44 **
0.26
-7.37 ***
-7.27 ***
Hypertension
Model 1
Model 2
-0.15 *
1.17 ***
0.14 *
0.14 ***
1.30 ***
-0.02
-4.58 ***
-0.12
1.22 ***
0.20 **
-0.29 ***
-0.07
-0.02
1.11 ***
0.04 ***
-3.81 ***
0.09
1.29 ***
0.10 *
-0.22 ***
-0.16
-0.15
0.58
0.14
**
*
***
***
Lung disease
Model 1
Model 2
0.53 ***
1.35 ***
0.05
0.39 ***
1.51 ***
-0.04
-0.39 ***
0.34
1.09
0.84
-6.61
**
***
***
***
-5.94 ***
Cancer
Model 1
Model 2
-0.51 ***
1.49 ***
0.38 ***
-6.16 ***
-0.41 ***
1.58 ***
0.20 *
0.15
0.22
0.44 **
0.30 *
-6.67 ***
Source: Weibull model estimates obtained by maximum likelihood. *** p<0.01, ** p<0.05, * p<0.1. Model 1: Includes
controls for age, and gender. Model 2: adds controls for college education and health behaviors. Europe includes: Netherlands,
Italy, France, Denmark and Spain. Spain is the reference country in Europe. Sample age 50-79. Sample weights used.
26
Table 4: Proportion dying in 2 years among
those with a disease
Heart disease
50-59
60-69
70-79
Total
Stroke
50-59
60-69
70-79
Total
Lung disease
50-59
60-69
70-79
Total
Diabetes
50-59
60-69
70-79
Total
Hypertension
50-59
60-69
70-79
Total
Cancer
50-59
60-69
70-79
Total
U.S.
Europe
(N= 3,273)
(N= 1,128)
0.036
0.054**
0.102***
0.067***
U.S.
(N= 1,026)
0.047
0.087**
0.122
0.092**
U.S.
(N= 1,405)
0.046
0.079**
0.146*
0.090***
U.S.
(N= 2,852)
0.038*
0.061**
0.100
0.064***
U.S.
(N= 8,158)
0.020
0.033***
0.073***
0.039***
U.S.
(N= 1,905)
0.031
0.079*
0.110
0.078
0.025
0.037
0.054
0.041
Europe
(N= 353)
0.012
0.053
0.121
0.073
Europe
(N= 668)
0.018
0.048
0.079
0.052
Europe
(N= 1,124)
0.008
0.042
0.079
0.047
Europe
(N=3,334)
0.005
0.027
0.045
0.027
Europe
(N= 588)
0.047
0.033
0.104
0.018
Ratio
1.44
1.46
1.89
1.63
Ratio
3.92
1.64
1.01
1.26
Ratio
2.56
1.65
1.85
1.73
Ratio
4.75
1.45
1.27
1.36
Ratio
4.00
1.22
1.62
1.44
Ratio
0.66
2.39
1.06
4.33
Source: HRS and SHARE, age 50-79, sample weights
used. Test of significance with condition (U.S. over
Europe): * p<0.01, ** p<0.05, * p<0.1. The first two
columns for both regions report the 2-year mortality rate
for those diagnosed with a condition. The ratio of the two
(U.S. over Europe) gives the relative mortality risk of those
with a condition.
27
Table 5: Coefficients showing the effect on mortality of living in Europe
relative to the U.S. for those with specified diseases (N in parentheses)
Heart disease (4,497)
Stroke (1,449)
Lung disease (2,107)
Diabetes (4,064)
Hypertension (11,676)
Cancer (2,535)
Model 1
-0.51 ***
-0.29
-0.44 **
-0.35 **
-0.44 ***
0.03
Model 2
-0.56 ***
-0.33
-0.42 **
-0.36 **
-0.23 ***
0.02
Model 3
-0.65 ***
-0.47 **
-0.55 ***
-0.54 ***
-0.65 ***
-0.19
Model 4
-0.32 *
-0.08
-0.30
-0.13
-0.28 **
-0.06
Source: HRS and SHARE, age 50-79, sample weights used. Gompertz models fitted by maximum
likelihood. * p<0.01, ** p<0.05, * p<0.1. Each model is run for those with one disease at a time. Model 1:
Includes controls for age, and gender. Model 2: adds controls for college education. Model 3: adds controls
for health behaviors. Model 4: adds controls for comorbidities.
28
Table 6: Comparison of disease prevalence and incidence
Heart disease
Stroke
U.S.
Europe
9.0
4.3
2.2
1.2
U.S.
Europe
3.9
4.5
1.3
1.7
U.S.
Europe
6.7
4.1
9.2
7.3
U.S.
Europe
18.3
10.1
5.5
2.8
Lung Disease
Diabetes
Prevalence 50-54
5.0
11.0
3.3
5.1
Incidence 50-79
1.7
3.6
3.6
4.2
Two-year mortality 50-79
9.0
6.4
5.2
4.7
Prevalence 50-79
8.2
15.7
6.0
10.8
Hypertension
Cancer
32.7
18.0
4.9
3.3
10.2
15.5
2.5
1.6
3.9
2.7
7.8
1.8
46.8
31.4
10.7
4.9
Source: HRS and SHARE, age 50-79, sample weights used.
29
Figures
Fig. 1
0
.02
.04
.06
2-year mortality rate
.08
.1
Comparison of Survey Mortality rates and HMD life table mortality rates 2004 for U.S. and
5 European Countries
50
60
HRS
age
SHARE
70
US-HMD
80
EU-5-HMD
Note: Curve of survey mortality rates smoothed using lowess filter and weighted using sampling weights
SHARE includes: Denmark, France, Italy, the Netherlands, and Spain
30
Fig. 2
Prevalence of disease in Europe and the U.S. in 2004, Age 50-79
US Europe
.5
.3
.2
.1
0
US Europe
.5
Obese >35
0
.1
.2
.3
.4
.5
0
.1
.2
.3
.4
.5
.4
.3
.2
.1
0
US Europe
Cancer
Diabetes
.4
.5
.4
.3
.2
.1
US Europe
Hypertension
US Europe
Lung Disease
0
.1
.2
.3
.4
.5
Stroke
0
0
.1
.2
.3
.4
.5
Heart Disease
US Europe
US Europe
Notes: Sample weights used. Europe includes: Denmark, France, Italy, the Netherlands, and Spain.
Age 50-79. All significantly different.
31
Fig. 3
U.S. and European Incidence Rates by Age
Heart disease
Stroke
55-59
60-64
65-69
.01 .02 .03 .04
70-74
55-59
60-64
65-69
70-74
75-79
Diabetes
*
*
*
*
*
0
0
.02
.02
.04
*
*
50-54
.04
.06
Lung disease
75-79
.06
50-54
55-59
60-64
65-69
70-74
0 .05 .1 .15 .2 .25
Hypertension
*
*
60-64
65-69
50-54
55-59
60-64
65-69
70-74
75-79
Cancer
*
*
*
65-69
70-74
0
*
*
75-79
.01 .02 .03 .04
50-54
50-54
*
55-59
*
70-74
75-79
50-54
55-59
60-64
75-79
Class II Obesity
*
*
0
.01
.02
.03
*
0
0 .02 .04 .06 .08 .1
*
50-54
55-59
60-64
65-69
70-74
75-79
Notes: Dark bars refer to United States while grey bars to Europe. Europe includes: Denmark, Netherlands,
France, Spain and Italy. Sample weights used to compute two-year incidence rates.
32
Appendix
Evaluation of Mortality Reports and Effect of Differential Response Rates
In Appendix Figure A.1 we show age-specific mortality estimated from surveys and from the
Human Mortality Data Base (HMD) for 9 European countries in SHARE and the U.S. Mortality
for the 5 countries we use in our analysis is quite close to those from HMD, with the exception of
some ages for the Netherlands and Italy where survey-based estimates are slightly below the life
table estimates. Average mortality rates (mx) at age 50 in 2004 in the HMD for the set of 9
European countries is 0.0035; while the average mortality rates for our subset 5 European
countries is 0.0029. The mortality rate for the U.S. is 0.0044. In 2004, average life expectancy at
age 50 in the HMD for the set of 9 European countries is 31.7 years; while life expectancy in our
5-country subsample is 31.8. At the same date, life expectancy in the U.S. is a year lower or 30.8
years.
Because of high levels of non-reponse in SHARE, we analyzed the determinants of nonresponse for each country using multinomial logit models for responded, died and status
unknown, with baseline demographic and health covariates (Table A.1 in the appendix).
Overall, we found no consistent pattern across countries in the association of the observed
characteristics with being in the unknown status, although some health conditions are negatively
associated with non-response which would suggest that sicker individuals are more likely to have
responded rather than be lost from SHARE. To measure how much these associations matter for
our calculations, we constructed inverse probability weights to correct for selection on those
observables following the strategy outlined by Michaud et al. (2011). We multiplied the SHARE
baseline weight, which corrects for differential baseline non-response based age and socio33
demographics based using census information, by this inverse probability weight, which now
corrects for differential non-response in the follow-up, to obtain a new weight.
When we
compared the distribution of this new weight to the weight provided by SHARE, we concluded
that the distributions were similar and that our approach was similar to the SHARE approach in
developing the weights. We also compared the incidence rates reported in our analysis based on
SHARE weights with incidence rates based on our version of the weights and found them to be
very similar. Therefore, we conclude that differential non-response is accounted for in the
SHARE weights and unlikely to bias the inferences we make on the two populations.
34
Figure A.1 – Mortality from Surveys and the Human Mortality Database by countries
35
Table A.1 – Multinomial Logit Results: Charactertics associated with Death and Unknown Status Relative to Alive Status
Variables
Socio-demographic
Age
Male
Foreign
Marriage
College
Health behaviors
Smoking
Obese
Health Conditions
Cancer
Hypertension
Heart disease
Stroke
Diabetes
Lung disease
Any missing
Constant
Austria
0.08
0.50
-0.50
0.61
-0.11
***
*
0.31
0.29
1.72
-0.29
-0.55
0.60
-0.11
-0.56
0.81
-8.85
Belgium
0.11
0.75
-0.88
0.87
-0.04
1.08
-0.05
***
Austria
Socio-demographic
Age
0.01
*
Male
6.99
Foreign
0.54
**
Marriage
-0.09
College
0.00
Health behaviors
Smoking
0.17
Obese
-0.60
***
Health Conditions
Cancer
0.21
Hypertension
-0.44
***
Heart disease
-0.34
Stroke
-0.04
Diabetes
0.06
Lung disease
-0.53
Any missing
13.96
Constant
-8.39
Source: SHARE (2004) and HRS (2006)
0.67
-0.13
0.12
0.45
-0.84
0.57
1.45
-12.32
Denmark
***
*
*
***
*
0.58
-0.85
***
Belgium
0.01
0.29
0.34
0.08
-0.01
0.08
-6.71
-0.68
-0.89
0.02
France
1.56
-0.36
-0.72
0.98
0.64
0.48
-14.01
-2.09
***
*
0.94
0.03
0.45
0.56
-0.32
0.85
-9.96
-4.08
0.00
0.41
0.35
0.03
-0.02
0.00
-0.10
0.03
-0.12
-0.03
-0.39
-0.14
-0.01
0.03
-0.37
0.00
0.33
0.57
-1.86
-0.26
-0.45
-0.07
0.16
0.40
-0.03
1.23
-1.97
*
*
***
***
**
***
0.11
-0.00
-0.52
-0.36
-0.06
Greece
***
0.38
-0.15
-0.13
-0.16
-0.17
-0.08
-0.03
-0.03
0.75
-0.99
*
**
0.01
0.34
0.27
-0.33
0.00
***
0.18
-0.18
**
-0.11
-0.03
0.02
0.18
0.05
-0.04
0.70
-0.87
*
***
0.09
-5.05
1.17
-0.78
0.00
Italy
***
*
*
-0.53
0.29
1.77
***
0.24
0.86
*
0.12
0.21
*
-0.14
0.00
92.05
Panel B: Status Unknown
France
Germany
0.01
0.58
-0.08
0.02
-0.01
*
***
0.24
-0.15
Denmark
*
0.08
-4.60
0.03
0.09
-0.13
Germany
Panel A: Died
**
***
*
1.83
0.88
0.45
0.28
-0.43
-0.41
-11.28
-4.53
0.11
0.81
-4.70
-0.84
0.01
Netherlands
***
**
*
0.65
-0.33
**
**
Greece
1.31
0.25
0.24
0.91
0.96
0.35
3.44
-12.94
0.01
-5.58
-2.40
-0.48
-0.03
-0.12
0.33
**
*
**
**
***
Italy
1.26
-0.54
0.91
1.19
0.12
0.24
0.20
-14.55
0.01
0.11
0.44
-0.33
0.02
**
*
0.01
0.86
0.27
0.26
0.00
0.17
-0.08
-0.21
-0.36
*
**
-0.27
-0.04
-0.34
0.03
-0.28
-0.61
0.38
-2.43
-0.14
0.07
-0.03
-0.17
-0.19
-0.47
0.01
-1.18
***
*
*
0.01
0.31
-0.10
-0.74
0.01
0.17
-0.51
**
*
**
***
Netherlands
0.01
0.21
0.26
0.11
0.01
*
***
Spain
***
-0.03
0.18
0.09
0.67
0.67
0.15
-0.16
-9.98
Sweden
***
**
*
*
**
Spain
0.13
0.47
-0.14
0.22
0.01
***
0.76
-1.31
*
*
0.70
0.23
0.05
0.39
0.50
1.41
0.20
-14.55
***
Sweden
0.00
-0.01
-0.08
-0.38
0.01
***
*
0.01
0.17
0.04
0.14
0.00
0.01
0.00
-0.10
-0.24
-0.17
-0.14
*
**
0.02
-0.11
0.00
-0.10
0.12
-0.19
1.71
-2.48
0.09
-0.28
0.11
-0.27
0.32
-0.33
0.00
-0.25
0.32
-0.08
0.06
0.01
-0.05
-0.15
0.41
-1.29
*
**
*
***
**
**
36
**
**