Content uploaded by Jens Søndergaard
Author content
All content in this area was uploaded by Jens Søndergaard on Nov 19, 2015
Content may be subject to copyright.
© The Author 2012. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com.
The Author 2012. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com.
Family Practice 2011; 00:1–8
doi:10.1093/fampra/cmr130
Socio-economic status influences blood pressure
control despite equal access to care
M S Paulsen
a,
*, M Andersen
a,b
, A P Munck
c
, P V Larsen
a
,
D G Hansen
a
, I A Jacobsen
d
, M L Larsen
e
, B Christensen
f
and
J Sondergaard
a
a
Research Unit of General Practice, Institute of Public Health, University of Southern Denmark, Odense, Denmark,
b
Centre for
Pharmacoepidemiology, Department of Medicine Solna, Karolinska Institutet, Stockholm, Sweden,
c
Audit Project Odense,
Research Unit of General Practice, University of Southern Denmark, Odense, Denmark,
d
The Clinic of Hypertension, Depart-
ment of Endocrinology, Odense University Hospital, Odense, Denmark,
e
Department of Cardiology, Odense University Hospital,
Odense, Denmark and
f
Department of General Practice, School of Public Health, University of Aarhus, Aarhus, Denmark.
*Correspondence to M. S. Paulsen, Research Unit of General Practice, Institute of Public Health, University of Southern
Denmark, J.B. Winsløws vej 9A 2nd floor, 5000 Odense C, Denmark; E-mail: mspaulsen@health.sdu.dk
Received 18 August 2011; Revised 27 October 2011; Accepted 20 December 2011.
Objective. Denmark has a health care system with free and equal access to care irrespective of
age and socio-economic status (SES). We conducted a cross-sectional study to investigate a pos-
sible association between SES and blood pressure (BP) control of hypertensive patients treated
in general practice.
Methods. We enrolled 184 general practices and 5260 hypertensive patients. The general practi-
tioners reported information about BP and diagnosis of diabetes. Information about education, in-
come, antihypertensive drug treatment and other co-morbidity was retrieved from relevant
registers from Statistics Denmark. The outcome measure was BP control defined as BP <140/90
mmHg in general and <130/80 mmHg in diabetics.
Results. Patients <65 years and with an educational level of 10–12 years had increased odds ratio
(OR) of BP control compared to patients with an educational level <10 years. Patients >65 years
had increased OR of BP control if they were married/cohabiting as compared to being single,
whereas education and income had no impact in this age group. Diabetics had significantly re-
duced odds of BP control irrespective of age, educational or income level.
Conclusions. Despite equalaccess to care for all patients, SES had significant impact on BP control
in this survey. Diabetes and cardiovascular disease also had a substantial influence irrespective of
age, educational and income level.
Keywords. Socio-economic status, hypertension, primary care, education, access to care.
Introduction
How socio-economic status (SES) influences blood
pressure (BP) control is not clear.
1,2
Studies of hospi-
tal populations have shown an association between
low SES and higher prevalence of hypertension, poor-
er BP control and higher mortality rate.
3–7
However,
the association between SES and BP control in pa-
tients treated in general practice has not been studied.
Investigating SES inequalities in patients often origi-
nates from health care systems based on a health in-
surance policy, limiting access to care for patients
with lower SES.
3–5,7–9
Denmark has a health care sys-
tem with free and equal access to care. Studying and
finding socio-economic inequalities in this health care
system with equal access to care could eliminate ‘ac-
cess to care’ as an important cause of social inequality.
In Denmark, diagnosis and treatment of hypertension
are primarily carried out in general practice and hy-
pertension accounts for 15% of all contacts.
10
Only
29.1–54% of treated hypertensive patients in Denmark
achieve BP control.
11,12
Based on the hypothesis that
SES influences BP control despite free access to care,
we conducted a cross-sectional study with the aim of
examining the association between SES and optimal
BP control in hypertensive patients treated in general
practice in Denmark.
Materials and methods
Study design
From November 2007 to January 2008, we enrolled
184 general practices. The GPs included 5413 patients
Page 1 of 8
Family Practice 2012; 29:503–510
doi:10.1093/fampra/cmr130
Advance Access published on 10 January 2012
503
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
with hypertension and registered information about
each patient’s BP and diagnosis of diabetes when rele-
vant.
12
A patient questionnaire included information
about lifestyle factors. Information about antihyperten-
sive drug treatment, other co-morbidity [cardiovascular
diseases (CVDs) and chronic kidney diseases] and socio-
economic factors (education, cohabitation status and
income) were retrieved from relevant registers provided
by Statistics Denmark.
Danish health care
The health care system in Denmark is tax funded, pro-
viding free access to general practice, outpatient clin-
ics and hospital care for all inhabitants irrespective
of age, SES and geographical residence. Reimburse-
ment increases with patient’s expenses for prescription
medication.
13
Audit Project Odense
Patients were enrolled during an audit on hyperten-
sion carried out by Audit Project Odense (APO).
APO is a quality development concept for general
practice which is widely used.
14,15
During consulta-
tions, the GPs registered information about patient
characteristics and treatment on an APO registration
chart. Data were collected consecutively over a prede-
fined period and provided a basis for evaluation of
treatment and work patterns within the GPs’ practices.
Guidelines for hypertension
The GPs were provided with national guidelines for
BP measurement and instructed to follow them. Pa-
tients’ office BP was measured at least twice with pa-
tients sitting down after 5 minutes of rest and using
a cuff properly adapted to the arm size.
16
A maximum
of 5-mmHg difference between measurements was ac-
cepted. The mean of the last two BP measurements
was registered. Treatment goals were BP <140/90
mmHg in general (including patients with CVD) and
<130/80 mmHg for diabetics and patients with chronic
kidney disease.
17
Data sources and measurements
Patient questionnaire. The patient questionnaire in-
cluded information about smoking status (never, cur-
rent or former smoker, with an indication of the year
when stopped), alcohol intake (0, 1–8, 9–20, 20–50 or
>50 U/week), weight (kilogram) and height (metre).
12
Statistics Denmark. All inhabitants in Denmark are
identified by a unique civil registration number, which
allows individual linkage across national registers. So-
cio-economic information was retrieved for 2007. The
Danish Register of Medical Product Statistics contains
all prescriptions since 1995, with patient identifier,
date and drug (ATC-code, name, package size, formu-
lation and quantity). We retrieved all information on
prescriptions for each patient from 1995 to 2008. From
the Danish National Patient Registry, we obtained in-
formation on admissions, outpatient services and
emergency room contacts with Danish hospitals classi-
fied according to the International Classification of
Diseases (ICD-10) for the period 1994–2007.
Participants. A total of 7111 hypertensive patients
were invited to participate during a BP follow-up with
their GP and 5413 patients accepted participation.
Some 153 patients only had a home BP measurement
registered and were excluded. The study population
therefore comprises 5260 patients with an office BP.
Variables. Outcome measure was optimal BP control
defined as BP <140/90 mmHg in general and <130/80
mmHg in diabetics. Body mass index (BMI) was clas-
sified as normal (<25 kg/m
2
), excess weight (25 <
BMI < 30) and obesity (>30 kg/m
2
).
Socio-economic variables. Education was categorized
according to the length of highest attained educational
level: <10 years (primary and lower secondary school),
10–12 years (vocational education and upper second-
ary school), >12 years (short, medium and long-term
higher education).
18–20
We used the variable equiva-
lent disposable income for a single family member,
which comprises all income (wages, salaries, all types
of benefits and pensions) after taxation for the entire
household and then adjusted for number of persons in
the household. Disposable income was categorized as
low (first quartile), medium (second and third quar-
tile) and high (fourth quartile).
21
Labour market status
was categorized as working, not on the workforce,
early retirement pension and retirement pension.
22
Cohabitant status was categorized as married/cohabi-
tating or single (divorced, widowed and never mar-
ried).
Co-morbidity. As co-morbidity, we included CVDs,
diabetes or chronic kidney disease since there are spe-
cial recommendations for hypertension treatment of
these three groups. From the Danish National Patient
Registry, we extracted primary and secondary diagno-
ses from the last 13 years within the following diseases:
ischaemic heart diseases, cerebral diseases, peripheral
vascular diseases and chronic kidney diseases. CVD
included ischaemic heart diseases, cerebral and
peripheral vascular diseases.
Antihypertensive medication. Antihypertensive drug
treatment included the following: diuretics, angiotensin-
converting enzymes inhibitors (ACE-Is), angiotensin
receptor blockers, calcium channel blockers and beta-
blockers. A patient was defined as being treated with
one or more of the antihypertensive drug classes, if
he/she had tablets available on the day they consulted
Family Practice—The International Journal for Research in Primary CarePage 2 of 8
their GP for BP follow-up and thereby were included
in the study. An example: redemption of 100 tablets
was assumed to cover 120 days because we allowed for
minor non-adherence. Combination drugs were split
into each drug class.
Statistical analysis
Due to age-specific differences in income and educa-
tion, we stratified all analyses into two age groups:
<65 years and >65 years. The age 65 years was chosen
as it is the normal retirement age in Denmark.
18,19
Differences in baseline characteristics stratified on
educational and income levels were investigated using
linear, logistic or multinomial logistic regression
adjusted for age and sex. A P-value <0.05 was consid-
ered significant. Educational level <10 years and low-
income level were used as baseline, respectively. All
regression analyses were adjusted for cluster effect at
practice level using robust cluster estimation. Odds ra-
tio (OR) is presented with 95% confidence intervals
(CIs). We analysed the association between optimal
BP control and various covariates in three different lo-
gistic regression models. Model 1 analysed the associa-
tion between optimal BP control and each covariate,
adjusted for age and sex. Model 2 analysed the associ-
ation between optimal BP control and educational
level or disposable income adjusted for age, sex, co-
habitation status, co-morbidity and antihypertensive
drug treatment. Model 3 further adjusted for lifestyles
factors like BMI, smoking and alcohol habits. The
models were compared using Akaike information cri-
terion (AIC) to identify the preferred model, i.e. the
model explaining most of the variation in the outcome
variable. We assumed that labour market status and
income were both on the causal pathway from educa-
tion to optimal BP control (Fig. 1) and thus did not in-
clude education and income in the same model.
Labour market status was not analysed separately due
to the age stratification below and above 65 years,
which separates the working population from the pa-
tients on retirement pension. Rank order correlation
(Kendall’s tau) was used to measure correlation be-
tween the following covariates: (i) education and
income, (ii) education and lifestyle factors and (iii) in-
come and lifestyle factors. Interaction terms between
education and lifestyle factors (BMI, smoking, alco-
hol) and between income level and lifestyle factors
were tested, as was the interaction between diabetes
and BMI. Missing values were considered missing at
random. STATA release 11.0 (StataCorp, College
Station, TX) was used for all statistical analyses.
Ethics
Patients gave consent to participate in the study by an-
swering the questionnaire. The study was approved by
the Danish Data Protection Agency. The study did
not need approval by the Regional Ethics Committee.
Results
Baseline characteristics stratified on education and
income
Table 1 presents the baseline characteristics. Of the
5260 patients included, 44.3% were males. Overall
mean age was 65.9 years (CI: 65.6–66.2). For patients
<65 years and with an educational level <10 years,
a lower proportion were working and a lower propor-
tion were married/cohabiting compared to patients
with an educational level of 10–12 years. When strati-
fying on income level, a higher proportion of patients
<65 years with a low-income level were diabetics,
a higher proportion were on early retirement pension
and a higher proportion were single.
There were far fewer patients with low educational
and low-income level in the age group <65 years
(10.2% with low income and 29.3% with education
<10 years) compared to patients <65 years (37.0%
with low income and 46.0% with education <10 years).
Optimal BP control
The overall proportion of patients achieving optimal
BP control was 29.1% (CI: 27.9–30.3) (Table 2). The
proportion of patients <65 years with optimal con-
trolled BP was 31.7% (CI: 29.8–33.6) with no differ-
ence between sex (P= 0.080, age adjusted), while
27.1% (CI: 25.5–28.7) of patients >65 years achieved
optimal BP control with no difference between sex (P
= 0.446, age adjusted) (Table 2). Since there were no
sex differences, the results in the following are not
presented separately for each sex.
Education and income
Patients <65 years of age had significantly higher
odds (OR: 1.23, CI: 1.01–1.52) of achieving optimal
BP control if their educational level was 10–12 years
compared with a level <10 years adjusted for all cova-
riates (Table 3). However, the association between
educational level and optimal BP control is attenuated
from Model 1 to 3. This is due to the confounding
FIGURE 1A simplified model for possible pathways between
socio-economic factors and optimal BP control
Page 3 of 8SES and BP control
504 Family Practice—The International Journal for Research in Primary Care
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
with hypertension and registered information about
each patient’s BP and diagnosis of diabetes when rele-
vant.
12
A patient questionnaire included information
about lifestyle factors. Information about antihyperten-
sive drug treatment, other co-morbidity [cardiovascular
diseases (CVDs) and chronic kidney diseases] and socio-
economic factors (education, cohabitation status and
income) were retrieved from relevant registers provided
by Statistics Denmark.
Danish health care
The health care system in Denmark is tax funded, pro-
viding free access to general practice, outpatient clin-
ics and hospital care for all inhabitants irrespective
of age, SES and geographical residence. Reimburse-
ment increases with patient’s expenses for prescription
medication.
13
Audit Project Odense
Patients were enrolled during an audit on hyperten-
sion carried out by Audit Project Odense (APO).
APO is a quality development concept for general
practice which is widely used.
14,15
During consulta-
tions, the GPs registered information about patient
characteristics and treatment on an APO registration
chart. Data were collected consecutively over a prede-
fined period and provided a basis for evaluation of
treatment and work patterns within the GPs’ practices.
Guidelines for hypertension
The GPs were provided with national guidelines for
BP measurement and instructed to follow them. Pa-
tients’ office BP was measured at least twice with pa-
tients sitting down after 5 minutes of rest and using
a cuff properly adapted to the arm size.
16
A maximum
of 5-mmHg difference between measurements was ac-
cepted. The mean of the last two BP measurements
was registered. Treatment goals were BP <140/90
mmHg in general (including patients with CVD) and
<130/80 mmHg for diabetics and patients with chronic
kidney disease.
17
Data sources and measurements
Patient questionnaire. The patient questionnaire in-
cluded information about smoking status (never, cur-
rent or former smoker, with an indication of the year
when stopped), alcohol intake (0, 1–8, 9–20, 20–50 or
>50 U/week), weight (kilogram) and height (metre).
12
Statistics Denmark. All inhabitants in Denmark are
identified by a unique civil registration number, which
allows individual linkage across national registers. So-
cio-economic information was retrieved for 2007. The
Danish Register of Medical Product Statistics contains
all prescriptions since 1995, with patient identifier,
date and drug (ATC-code, name, package size, formu-
lation and quantity). We retrieved all information on
prescriptions for each patient from 1995 to 2008. From
the Danish National Patient Registry, we obtained in-
formation on admissions, outpatient services and
emergency room contacts with Danish hospitals classi-
fied according to the International Classification of
Diseases (ICD-10) for the period 1994–2007.
Participants. A total of 7111 hypertensive patients
were invited to participate during a BP follow-up with
their GP and 5413 patients accepted participation.
Some 153 patients only had a home BP measurement
registered and were excluded. The study population
therefore comprises 5260 patients with an office BP.
Variables. Outcome measure was optimal BP control
defined as BP <140/90 mmHg in general and <130/80
mmHg in diabetics. Body mass index (BMI) was clas-
sified as normal (<25 kg/m
2
), excess weight (25 <
BMI < 30) and obesity (>30 kg/m
2
).
Socio-economic variables. Education was categorized
according to the length of highest attained educational
level: <10 years (primary and lower secondary school),
10–12 years (vocational education and upper second-
ary school), >12 years (short, medium and long-term
higher education).
18–20
We used the variable equiva-
lent disposable income for a single family member,
which comprises all income (wages, salaries, all types
of benefits and pensions) after taxation for the entire
household and then adjusted for number of persons in
the household. Disposable income was categorized as
low (first quartile), medium (second and third quar-
tile) and high (fourth quartile).
21
Labour market status
was categorized as working, not on the workforce,
early retirement pension and retirement pension.
22
Cohabitant status was categorized as married/cohabi-
tating or single (divorced, widowed and never mar-
ried).
Co-morbidity. As co-morbidity, we included CVDs,
diabetes or chronic kidney disease since there are spe-
cial recommendations for hypertension treatment of
these three groups. From the Danish National Patient
Registry, we extracted primary and secondary diagno-
ses from the last 13 years within the following diseases:
ischaemic heart diseases, cerebral diseases, peripheral
vascular diseases and chronic kidney diseases. CVD
included ischaemic heart diseases, cerebral and
peripheral vascular diseases.
Antihypertensive medication. Antihypertensive drug
treatment included the following: diuretics, angiotensin-
converting enzymes inhibitors (ACE-Is), angiotensin
receptor blockers, calcium channel blockers and beta-
blockers. A patient was defined as being treated with
one or more of the antihypertensive drug classes, if
he/she had tablets available on the day they consulted
Family Practice—The International Journal for Research in Primary CarePage 2 of 8
their GP for BP follow-up and thereby were included
in the study. An example: redemption of 100 tablets
was assumed to cover 120 days because we allowed for
minor non-adherence. Combination drugs were split
into each drug class.
Statistical analysis
Due to age-specific differences in income and educa-
tion, we stratified all analyses into two age groups:
<65 years and >65 years. The age 65 years was chosen
as it is the normal retirement age in Denmark.
18,19
Differences in baseline characteristics stratified on
educational and income levels were investigated using
linear, logistic or multinomial logistic regression
adjusted for age and sex. A P-value <0.05 was consid-
ered significant. Educational level <10 years and low-
income level were used as baseline, respectively. All
regression analyses were adjusted for cluster effect at
practice level using robust cluster estimation. Odds ra-
tio (OR) is presented with 95% confidence intervals
(CIs). We analysed the association between optimal
BP control and various covariates in three different lo-
gistic regression models. Model 1 analysed the associa-
tion between optimal BP control and each covariate,
adjusted for age and sex. Model 2 analysed the associ-
ation between optimal BP control and educational
level or disposable income adjusted for age, sex, co-
habitation status, co-morbidity and antihypertensive
drug treatment. Model 3 further adjusted for lifestyles
factors like BMI, smoking and alcohol habits. The
models were compared using Akaike information cri-
terion (AIC) to identify the preferred model, i.e. the
model explaining most of the variation in the outcome
variable. We assumed that labour market status and
income were both on the causal pathway from educa-
tion to optimal BP control (Fig. 1) and thus did not in-
clude education and income in the same model.
Labour market status was not analysed separately due
to the age stratification below and above 65 years,
which separates the working population from the pa-
tients on retirement pension. Rank order correlation
(Kendall’s tau) was used to measure correlation be-
tween the following covariates: (i) education and
income, (ii) education and lifestyle factors and (iii) in-
come and lifestyle factors. Interaction terms between
education and lifestyle factors (BMI, smoking, alco-
hol) and between income level and lifestyle factors
were tested, as was the interaction between diabetes
and BMI. Missing values were considered missing at
random. STATA release 11.0 (StataCorp, College
Station, TX) was used for all statistical analyses.
Ethics
Patients gave consent to participate in the study by an-
swering the questionnaire. The study was approved by
the Danish Data Protection Agency. The study did
not need approval by the Regional Ethics Committee.
Results
Baseline characteristics stratified on education and
income
Table 1 presents the baseline characteristics. Of the
5260 patients included, 44.3% were males. Overall
mean age was 65.9 years (CI: 65.6–66.2). For patients
<65 years and with an educational level <10 years,
a lower proportion were working and a lower propor-
tion were married/cohabiting compared to patients
with an educational level of 10–12 years. When strati-
fying on income level, a higher proportion of patients
<65 years with a low-income level were diabetics,
a higher proportion were on early retirement pension
and a higher proportion were single.
There were far fewer patients with low educational
and low-income level in the age group <65 years
(10.2% with low income and 29.3% with education
<10 years) compared to patients <65 years (37.0%
with low income and 46.0% with education <10 years).
Optimal BP control
The overall proportion of patients achieving optimal
BP control was 29.1% (CI: 27.9–30.3) (Table 2). The
proportion of patients <65 years with optimal con-
trolled BP was 31.7% (CI: 29.8–33.6) with no differ-
ence between sex (P= 0.080, age adjusted), while
27.1% (CI: 25.5–28.7) of patients >65 years achieved
optimal BP control with no difference between sex (P
= 0.446, age adjusted) (Table 2). Since there were no
sex differences, the results in the following are not
presented separately for each sex.
Education and income
Patients <65 years of age had significantly higher
odds (OR: 1.23, CI: 1.01–1.52) of achieving optimal
BP control if their educational level was 10–12 years
compared with a level <10 years adjusted for all cova-
riates (Table 3). However, the association between
educational level and optimal BP control is attenuated
from Model 1 to 3. This is due to the confounding
FIGURE 1A simplified model for possible pathways between
socio-economic factors and optimal BP control
Page 3 of 8SES and BP control
SES and BP control 505
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
effect of antihypertensive drug treatment and lifestyles
factors on the association between optimal BP control
and educational level (Table 3). The OR of optimal
BP control increased among patients with high in-
come, when adjusting only for age and sex. However,
the association between income and BP control was
attenuated, when adjusting for further covariates.
For patients >65 years of age, education and income
showed no significant association with BP control, but
being married/cohabiting increased the OR of control
to 1.21 (CI: 1.00–1.46, Model 3) for models adjusted for
education and an OR of 1.22 (CI: 1.00–1.48, Model 3)
for models adjusted for income.
Co-morbidity
Irrespective of age, diagnosis of diabetes strongly re-
duced the odds of optimal BP control in all analyses
(Table 3). CVD increased the odds of optimal BP
control irrespective of age, education and income
(Table 3). Chronic kidney disease was not included
in the analysis since only 12 patients were registered
with the disease.
Correlation and interactions terms
The rank order correlation (r) between educational
level and income was r= 0.34 (low to moderate), be-
tween educational level and BMI r= –0.10 and
between income and BMI r= –0.04. No interaction
terms were significant at a 5% level for either age
group. For patients <65 years, the AIC values from
comparing the educational models decreased from
Model 1 (AIC = 2851) to Model 2 (AIC = 2658) and
Model 3 (AIC = 2564). For income models, the same
was seen (AIC = 2885, 2691, 2596, respectively). This
indicated that a logistic model adjusting for all covari-
ates (civil status, co-morbidity, antihypertensive drug
treatment and lifestyles factors) should be the preferred
and best fitted model. A similar model comparison
applies for patients >65 years.
Discussion
We found that higher SES was associated with optimal
BP control for hypertensive patients treated in pri-
mary care. The socio-economic gradient differed with
age. Firstly, for the age group <65 years, educational
level was positively associated with optimal BP con-
trol, whereas cohabiting status had little influence.
Secondly, for patients >65 years being married/cohab-
iting, the odds increased for optimal BP control com-
pared to patients being single, while education and
income had little impact for this age group. Thirdly,
CVD was strongly associated with optimal BP control,
TABLE 1Baseline characteristic of 5260 hypertensive patients stratified by age, income and educational level
Age <65 years Age >65 years
All Educational level Income All Educational level Income
n= 2313 <10 10–12 >12 Low Medium High n= 2947 <10 10–12 >12 Low Medium High
Number, % 29.3 47.7 22.9 10.2 51.8 38.0 – 46.0 36.5 17.6 37.0 48.5 14.5
Male, % 46.3 39.5 50.2
a
46.3
a
43.6 43.6 50.6 42.8 33.5 52.3
a
51.1
a
34.8 45.5
a
54.1
a
Age, mean 55.7 56.3 55.5 55.6 55.9 55.4 56.2 73.8 74.0 72.6 73.0 75.6 73.1
a
71.7
a
BP
Systolic BP, mean 140.9 140.6 141.3 140.6 142.7 140.7 140.8 144.9 144.9 144.8 144.4 145.1 144.8 144.7
Diastolic BP, mean 85.4 84.7 85.4 86.2
a
84.1 85.1 86.1
a
80.7 80.0 81.4
a
81.5 79.8 81.0 82.0
Labour market status, %
Working 70.5 57.0 72.3
a
84.7
a
30.6 64.1
a
88.6
a
4.7 2.7 5.1 10.3
a
– 3.6
a
20.3
a
Retirement pension – – – – – – – 92.0 94.8 89.8 87.0
a
98.8 91.7
a
75.6
a
Early retirement pension 15.0 19.0 15.4 8.9
a
28.2 17.7
a
8.0
a
2.7 1.7 4.4
a
2.2 0.06 4.2
a
2.6
Not on the workforce 14.5 24.0 12.3
a
6.4
a
41.2 18.2
a
3.4
a
0.6 0.8 0.7 0.0 0.05 0.6
a
1.4
Cohabitation status, %
Single 24.3 27.1 22.7
a
23.9 53.8 27.2
a
12.5
a
39.5 42.8 33.4 35.1 51.6 33.9
a
27.4
a
Married/cohabiting 75.7 72.9 77.3
a
76.1 46.2 72.8
a
87.5
a
60.5 57.2 66.6 64.9 48.4 66.1
a
72.6
a
Co-morbidity
No diabetes or CVD 76.3 71.1 76.4
a
83.4
a
70.1 73.8 81.2
a
64.0 62.0 63.7
a
68.4
a
61.7 64.6 67.7
a
Diabetes
b
, % 13.6 17.1 13.2
a
9.0
a
20.1 15.4 9.5
a
17.0 18.0 17.4 15.8
a
18.6 16.5
a
14.8
a
CVD
c
, % 10.1 11.8 10.4 7.7
a
9.8 10.8 9.3
a
19.0 19.9 19.0 15.8
a
19.6 18.9 17.4
Kidney disease, % 0.04 – – – 0 0 0.11 0.4 0.2 0.8 0.2 0.3 0.5 0.5
All numbers presented in columns are percent (%), unless otherwise indicated as mean values in column 1.
a
Indicates a P-values <0.05 using linear, logistic or multinomial logistic regression for the differences in baseline characteristics with, respectively,
educational level <10 years or low income level as baseline.
b
Diagnosis of diabetes with or without CVD (202 diabetics had a CVD).
c
CVD without diagnosis of diabetes.
Family Practice—The International Journal for Research in Primary CarePage 4 of 8
whereas diabetes was negatively associated, irrespec-
tive of age, education and income.
Strengths and weaknesses
We used information from Statistics Denmark, which
is a valid data source on socio-economic factors and
diagnoses from hospitalizations.
21,23
Data on a large
Danish population of hypertensive patients from gen-
eral practice were sampled through a data collection
tool, which GPs were accustomed to use.
12,14
This op-
timized the consistency in the registration data and
BP measurements. We were furthermore able to in-
clude information on lifestyle factors like BMI, smok-
ing status and alcohol consumption with register
information. Another strength of our study was the
careful design of analytical models. We tested for rele-
vant correlations and decided not to include education
and income in the same analytical models because we
believe that income level is on the causal pathway
between education and optimal BP control (Fig. 1).
The sampling methods with enrolment of patients
during 4–8 weeks may have given rise to some sam-
pling bias. Patients seeing their doctor often had
a higher chance of being included in the study. Since
patients with low SES visit their doctor more seldom
than patients with higher SES,
24
patients with high
SES could be over-represented in this study popula-
tion. Sampling bias might therefore have caused the
TABLE 2Optimal BP control according to socio-economic factors, co-morbidity, antihypertensive drug treatment and lifestyle factors
Proportion of patients with optimal BP control
a
OR for optimal BP control, adjusted for age and sex
All subjects (%) <65 years (%) >65 years (%) <65 years, OR (95% CI) >65 years, OR (95% CI)
Optimal BP control 29.1 31.7 27.1 ––
Highest attained education (years)
<10 27.9 28.6 27.5 1 1
10–12 30.2 33.7 26.4 1.28 (1.04–1.58) 0.96 (0.80–1.15)
>12 30.0 32.3 27.6 1.20 (0.94–1.52) 1.02 (0.82–1.26)
Income
Low (1st quartile) 27.7 26.9 27.9 1 1
Medium (2nd + 3rd quartile) 28.4 30.9 26.4 1.20 (0.88–1.65) 0.94 (0.79–1.11)
High (4th quartile) 31.9 34.1 27.2 1.43 (1.04–1.96) 0.99 (0.76–1.29)
Labour market status
Working 31.6 32.4 21.6 1 1
Retirement from 67 years 27.4 27.4 – 1.35 (0.91–2.01)
Early retirement from 60 years 26.8 27.6 23.1 0.82 (0.62–1.10) 1.07 (0.49–2.34)
Not on the workforce 33.3 33.1 36.8 1.03 (0.80–1.32) 2.08 (0.75–5.80)
Cohabitation status
Single 30.1 30.3 25.6 1 1
Married/cohabiting 27.1 32.2 28.1 1.11 (0.90–1.36) 1.20 (1.01–1.43)
Co-morbidity
No diabetes or CVD 31.3 33.8 28.9 1 1
Diabetes
b
, % 10.9 10.8 11.0 0.24 (0.17–0.36) 0.30 (0.21–0.45)
CVD
c
, % 37.8 43.6 35.4 1.57 (1.14–2.17) 1.36 (1.10–1.68)
Antihypertensive treatment, %
0 Drug 23.5 26.6 20.1 0.80 (0.58–1.12) 0.64 (0.43–0.94)
1 Drug 29.4 30.6 28.3 1 1
2 Drugs 29.9 34.2 26.6 1.20 (0.99–1.46) 0.91 (0.75–1.10)
3 Drugs 30.3 33.0 28.8 1.18 (0.92–1.51) 1.02 (0.82–1.27)
>4 Drugs 25.1 24.1 25.6 0.76 (0.45–1.27) 0.87 (0.60–1.26)
BMI, %
BMI <25 30.8 34.3 28.8 1 1
BMI 25–30 30.0 31.4 28.8 0.92 (0.73–1.15) 1.01 (0.84–1.21)
BMI >30 24.7 29.1 19.6 0.81 (0.65–1.02) 0.60 (0.47–0.77)
Smoking, %
Non-smoker 29.7 32.2 27.5 1 1
Former smoker 28.3 32.1 26.1 1.05 (0.86–1.28) 0.95 (0.79–1.14)
Smoker 29.7 30.4 28.9 0.94 (0.76–1.16) 1.09 (0.86–1.39)
Alcohol, %
0 U/week 28.4 32.1 25.6 1 1
1–7 U/week 30.7 33.6 28.5 1.13 (0.89–1.43) 1.17 (0.94–1.45)
8–20 U/week 26.9 28.7 25.3 0.93 (0.70–1.22) 1.01 (0.80–1.28)
>20 U/week 28.0 29.4 26.1 1.00 (0.69–1.47) 1.07 (0.69–1.65)
a
Optimal BP control: BP <140/90 mmHg and BP <130/80 mmHg for diabetics.
b
Diagnosis of diabetes with or without CVD (202 diabetics had a CVD).
c
CVD without diagnosis of diabetes.
Page 5 of 8SES and BP control
506 Family Practice—The International Journal for Research in Primary Care
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
effect of antihypertensive drug treatment and lifestyles
factors on the association between optimal BP control
and educational level (Table 3). The OR of optimal
BP control increased among patients with high in-
come, when adjusting only for age and sex. However,
the association between income and BP control was
attenuated, when adjusting for further covariates.
For patients >65 years of age, education and income
showed no significant association with BP control, but
being married/cohabiting increased the OR of control
to 1.21 (CI: 1.00–1.46, Model 3) for models adjusted for
education and an OR of 1.22 (CI: 1.00–1.48, Model 3)
for models adjusted for income.
Co-morbidity
Irrespective of age, diagnosis of diabetes strongly re-
duced the odds of optimal BP control in all analyses
(Table 3). CVD increased the odds of optimal BP
control irrespective of age, education and income
(Table 3). Chronic kidney disease was not included
in the analysis since only 12 patients were registered
with the disease.
Correlation and interactions terms
The rank order correlation (r) between educational
level and income was r= 0.34 (low to moderate), be-
tween educational level and BMI r= –0.10 and
between income and BMI r= –0.04. No interaction
terms were significant at a 5% level for either age
group. For patients <65 years, the AIC values from
comparing the educational models decreased from
Model 1 (AIC = 2851) to Model 2 (AIC = 2658) and
Model 3 (AIC = 2564). For income models, the same
was seen (AIC = 2885, 2691, 2596, respectively). This
indicated that a logistic model adjusting for all covari-
ates (civil status, co-morbidity, antihypertensive drug
treatment and lifestyles factors) should be the preferred
and best fitted model. A similar model comparison
applies for patients >65 years.
Discussion
We found that higher SES was associated with optimal
BP control for hypertensive patients treated in pri-
mary care. The socio-economic gradient differed with
age. Firstly, for the age group <65 years, educational
level was positively associated with optimal BP con-
trol, whereas cohabiting status had little influence.
Secondly, for patients >65 years being married/cohab-
iting, the odds increased for optimal BP control com-
pared to patients being single, while education and
income had little impact for this age group. Thirdly,
CVD was strongly associated with optimal BP control,
TABLE 1Baseline characteristic of 5260 hypertensive patients stratified by age, income and educational level
Age <65 years Age >65 years
All Educational level Income All Educational level Income
n= 2313 <10 10–12 >12 Low Medium High n= 2947 <10 10–12 >12 Low Medium High
Number, % 29.3 47.7 22.9 10.2 51.8 38.0 – 46.0 36.5 17.6 37.0 48.5 14.5
Male, % 46.3 39.5 50.2
a
46.3
a
43.6 43.6 50.6 42.8 33.5 52.3
a
51.1
a
34.8 45.5
a
54.1
a
Age, mean 55.7 56.3 55.5 55.6 55.9 55.4 56.2 73.8 74.0 72.6 73.0 75.6 73.1
a
71.7
a
BP
Systolic BP, mean 140.9 140.6 141.3 140.6 142.7 140.7 140.8 144.9 144.9 144.8 144.4 145.1 144.8 144.7
Diastolic BP, mean 85.4 84.7 85.4 86.2
a
84.1 85.1 86.1
a
80.7 80.0 81.4
a
81.5 79.8 81.0 82.0
Labour market status, %
Working 70.5 57.0 72.3
a
84.7
a
30.6 64.1
a
88.6
a
4.7 2.7 5.1 10.3
a
– 3.6
a
20.3
a
Retirement pension – – – – – – – 92.0 94.8 89.8 87.0
a
98.8 91.7
a
75.6
a
Early retirement pension 15.0 19.0 15.4 8.9
a
28.2 17.7
a
8.0
a
2.7 1.7 4.4
a
2.2 0.06 4.2
a
2.6
Not on the workforce 14.5 24.0 12.3
a
6.4
a
41.2 18.2
a
3.4
a
0.6 0.8 0.7 0.0 0.05 0.6
a
1.4
Cohabitation status, %
Single 24.3 27.1 22.7
a
23.9 53.8 27.2
a
12.5
a
39.5 42.8 33.4 35.1 51.6 33.9
a
27.4
a
Married/cohabiting 75.7 72.9 77.3
a
76.1 46.2 72.8
a
87.5
a
60.5 57.2 66.6 64.9 48.4 66.1
a
72.6
a
Co-morbidity
No diabetes or CVD 76.3 71.1 76.4
a
83.4
a
70.1 73.8 81.2
a
64.0 62.0 63.7
a
68.4
a
61.7 64.6 67.7
a
Diabetes
b
, % 13.6 17.1 13.2
a
9.0
a
20.1 15.4 9.5
a
17.0 18.0 17.4 15.8
a
18.6 16.5
a
14.8
a
CVD
c
, % 10.1 11.8 10.4 7.7
a
9.8 10.8 9.3
a
19.0 19.9 19.0 15.8
a
19.6 18.9 17.4
Kidney disease, % 0.04 – – – 0 0 0.11 0.4 0.2 0.8 0.2 0.3 0.5 0.5
All numbers presented in columns are percent (%), unless otherwise indicated as mean values in column 1.
a
Indicates a P-values <0.05 using linear, logistic or multinomial logistic regression for the differences in baseline characteristics with, respectively,
educational level <10 years or low income level as baseline.
b
Diagnosis of diabetes with or without CVD (202 diabetics had a CVD).
c
CVD without diagnosis of diabetes.
Family Practice—The International Journal for Research in Primary CarePage 4 of 8
whereas diabetes was negatively associated, irrespec-
tive of age, education and income.
Strengths and weaknesses
We used information from Statistics Denmark, which
is a valid data source on socio-economic factors and
diagnoses from hospitalizations.
21,23
Data on a large
Danish population of hypertensive patients from gen-
eral practice were sampled through a data collection
tool, which GPs were accustomed to use.
12,14
This op-
timized the consistency in the registration data and
BP measurements. We were furthermore able to in-
clude information on lifestyle factors like BMI, smok-
ing status and alcohol consumption with register
information. Another strength of our study was the
careful design of analytical models. We tested for rele-
vant correlations and decided not to include education
and income in the same analytical models because we
believe that income level is on the causal pathway
between education and optimal BP control (Fig. 1).
The sampling methods with enrolment of patients
during 4–8 weeks may have given rise to some sam-
pling bias. Patients seeing their doctor often had
a higher chance of being included in the study. Since
patients with low SES visit their doctor more seldom
than patients with higher SES,
24
patients with high
SES could be over-represented in this study popula-
tion. Sampling bias might therefore have caused the
TABLE 2Optimal BP control according to socio-economic factors, co-morbidity, antihypertensive drug treatment and lifestyle factors
Proportion of patients with optimal BP control
a
OR for optimal BP control, adjusted for age and sex
All subjects (%) <65 years (%) >65 years (%) <65 years, OR (95% CI) >65 years, OR (95% CI)
Optimal BP control 29.1 31.7 27.1 ––
Highest attained education (years)
<10 27.9 28.6 27.5 1 1
10–12 30.2 33.7 26.4 1.28 (1.04–1.58) 0.96 (0.80–1.15)
>12 30.0 32.3 27.6 1.20 (0.94–1.52) 1.02 (0.82–1.26)
Income
Low (1st quartile) 27.7 26.9 27.9 1 1
Medium (2nd + 3rd quartile) 28.4 30.9 26.4 1.20 (0.88–1.65) 0.94 (0.79–1.11)
High (4th quartile) 31.9 34.1 27.2 1.43 (1.04–1.96) 0.99 (0.76–1.29)
Labour market status
Working 31.6 32.4 21.6 1 1
Retirement from 67 years 27.4 27.4 – 1.35 (0.91–2.01)
Early retirement from 60 years 26.8 27.6 23.1 0.82 (0.62–1.10) 1.07 (0.49–2.34)
Not on the workforce 33.3 33.1 36.8 1.03 (0.80–1.32) 2.08 (0.75–5.80)
Cohabitation status
Single 30.1 30.3 25.6 1 1
Married/cohabiting 27.1 32.2 28.1 1.11 (0.90–1.36) 1.20 (1.01–1.43)
Co-morbidity
No diabetes or CVD 31.3 33.8 28.9 1 1
Diabetes
b
, % 10.9 10.8 11.0 0.24 (0.17–0.36) 0.30 (0.21–0.45)
CVD
c
, % 37.8 43.6 35.4 1.57 (1.14–2.17) 1.36 (1.10–1.68)
Antihypertensive treatment, %
0 Drug 23.5 26.6 20.1 0.80 (0.58–1.12) 0.64 (0.43–0.94)
1 Drug 29.4 30.6 28.3 1 1
2 Drugs 29.9 34.2 26.6 1.20 (0.99–1.46) 0.91 (0.75–1.10)
3 Drugs 30.3 33.0 28.8 1.18 (0.92–1.51) 1.02 (0.82–1.27)
>4 Drugs 25.1 24.1 25.6 0.76 (0.45–1.27) 0.87 (0.60–1.26)
BMI, %
BMI <25 30.8 34.3 28.8 1 1
BMI 25–30 30.0 31.4 28.8 0.92 (0.73–1.15) 1.01 (0.84–1.21)
BMI >30 24.7 29.1 19.6 0.81 (0.65–1.02) 0.60 (0.47–0.77)
Smoking, %
Non-smoker 29.7 32.2 27.5 1 1
Former smoker 28.3 32.1 26.1 1.05 (0.86–1.28) 0.95 (0.79–1.14)
Smoker 29.7 30.4 28.9 0.94 (0.76–1.16) 1.09 (0.86–1.39)
Alcohol, %
0 U/week 28.4 32.1 25.6 1 1
1–7 U/week 30.7 33.6 28.5 1.13 (0.89–1.43) 1.17 (0.94–1.45)
8–20 U/week 26.9 28.7 25.3 0.93 (0.70–1.22) 1.01 (0.80–1.28)
>20 U/week 28.0 29.4 26.1 1.00 (0.69–1.47) 1.07 (0.69–1.65)
a
Optimal BP control: BP <140/90 mmHg and BP <130/80 mmHg for diabetics.
b
Diagnosis of diabetes with or without CVD (202 diabetics had a CVD).
c
CVD without diagnosis of diabetes.
Page 5 of 8SES and BP control
SES and BP control 507
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
proportion of patients with low income or low edu-
cational level to be much smaller in the age group
<65 years than in the age group >65 years as seen in
Table 1. And if sampling bias is the main cause of this
difference, the association between education and BP
control for the lower educated and the low-income
group may have been underestimated in our study.
However, the difference in educational and income
levels between age groups observed in Table 1 is most
likely due to the demographic development in the
Danish population, where the younger patients have
longer educations than the elderly, and the younger
working population has higher income levels than the
elderly on retirement pension.
However, the effect of educational level on BP con-
trol remained after adjustment for a number of fac-
tors. It is not clearly defined through which pathways
educational level affects BP control. Higher education
could lead to a better understanding of risks and the
importance of treatment, but the impact of education
might also be mediated through other socio-economic
factors (Fig. 1) or through lifestyle.
25–27
Methodologi-
cal studies have recommended that socio-economic
factors like education, occupation and income are not
used interchangeably.
21,26,28
In our analysis, these fac-
tors were studied independently and we found differ-
ent effects of education and income on BP control
and a pronounced difference in the patterns according
to age.
It is worrying that low educational level is associ-
ated with lower OR of BP control in hypertensive pa-
tients <65 years, especially in a health care system
with equal access to care. In patients >65 years, edu-
cational level had no impact on BP control. Other fac-
tors such as physiology, co-morbidity, polypharmacy
and performance status can be more important for BP
control than educational level in the elderly. Further-
more, the age group >65 years had a high mean age
TABLE 3OR for optimal BP control according to socio-economic factors
Model 1
a
Education Income
Model 2
b
Model 3
c
Model 2
b
Model 3
c
Patients <65 years
Highest attained educational level (years)
<10 1 1 1
10–12 1.28 (1.04–1.58) 1.24 (1.01–1.53) 1.23 (1.01–1.52)
>12 1.20 (0.94–1.52) 1.13 (0.89–1.45) 1.18 (0.91–1.52)
Income
Low (1st quartile) 1 1 1
Medium (2nd + 3rd quartile) 1.20 (0.88–1.65) 1.13 (0.81–1.58) 1.09 (0.77–1.53)
High (4th quartile) 1.43 (1.04–1.96) 1.28 (0.92–1.79) 1.29 (0.92–1.81)
Cohabitation status
Single 1 1 1 1 1
Married/cohabiting 1.11 (0.90–1.36) 1.06 (0.86–1.32) 1.01 (0.81–1.25) 1.03 (0.82–1.29) 0.98 (0.78–1.22)
Co-morbidity
No co-morbidity 1 1 1 1 1
Diabetes 0.24 (0.17–0.36) 0.22 (0.15–0.34) 0.24 (0.15–0.36) 0.25 (0.17–0.37) 0.26 (0.17–0.39)
CVD 1.57 (1.14–2.17) 1.59 (1.15–2.20) 1.57 (1.13–2.18) 1.59 (1.14–2.21) 1.57 (1.13–2.19)
Patients >65 years
Highest attained educational level (years)
<10 1 1 1
10–12 0.96 (0.80–1.15) 0.93 (0.78–1.12) 0.98 (0.81–1.18)
>12 1.02 (0.82–1.26) 0.99 (0.80–1.23) 1.03 (0.82–1.29)
Income
Low (1st quartile) 1 1 1
Medium (2nd + 3rd quartile) 0.94 (0.79–1.11) 0.89 (0.74–1.06) 0.91 (0.75–1.10)
High (4th quartile) 0.99 (0.76–1.29) 0.91 (0.69–1.19) 0.91 (0.67–1.25)
Cohabitation status
Single 1 1 1 1 1
Married/cohabiting 1.20 (1.01–1.43) 1.14 (0.95–1.37) 1.21 (1.00–1.47) 1.17 (0.98–1.40) 1.22 (1.01–1.48)
Co-morbidity
No co-morbidity 1 1 1 1 1
Diabetes 0.30 (0.21–0.45) 0.31 (0.21–0.45) 0.33 (0.22–0.50) 0.30 (0.20–0.44) 0.32 (0.21–0.48)
CVD 1.36 (1.10–1.68) 1.38 (1.11–1.73) 1.38 (1.10–1.74) 1.34 (1.08–1.66) 1.34 (1.08–1.68)
a
Model 1 is OR (95% CI) analysed separately, adjusted for age and sex.
b
Model 2 is OR (95% CI) with educational level or income, adjusted for age, sex, cohabitation status, co-morbidity and number of antihypertensive
drug used as baseline characteristics.
c
Model 3 is OR (95% CI) for optimal BP control with educational level or income adjusted for age, sex, cohabitation status, co-morbidity, number
of antihypertensive drug used, BMI, smoking and alcohol.
Family Practice—The International Journal for Research in Primary CarePage 6 of 8
of 73.8 years. At this age, one would expect that the
lower educated patients included in our study are the
healthiest patients, who had survived long enough to
be included. This could have eliminated a possible ed-
ucational difference in BP control for this age group.
GPs should pay particular attention to patients’ edu-
cational level and, in case of the elderly, the cohabit-
ing status. This simple approach could strengthen the
attention towards patients without optimal BP control
and together with intensive BP follow-up and careful
information to this vulnerable group of hypertensive
patients, an increased focus would be on the patients
needed it. An important aspect to consider in this con-
text is the communication between the GP and the pa-
tient.
29
In the elderly, education and income level had
no impact on the ability to achieve BP control, while
being married or cohabiting increased the odds, which
could be due to the fact that the married/cohabiting el-
derly patients often visit their GP together with their
partner that might improve the understanding of the
information given by the GP.
We found that only 29.1% of all patients achieved
optimal BP control. Previous studies from Denmark
have shown that the proportion of hypertensive pa-
tients with controlled BP increased from 21% to 43%
from 1998 to 2004.
11,30
Our observation of only 29.1%
having achieved controlled BP may to some extent be
due to a higher proportion of hypertensive diabetics
in our study (Table 1).
11
BP targets for diabetics are
<130/80 mmHg, which is more difficult to achieve.
Table 2 illustrates this since only 10.9% of all diabetics
had controlled BP. Nevertheless, it is in accordance
with other studies that hypertensive diabetics had sig-
nificantly reduced OR of BP control and this reduced
OR persists in our study regardless of using a model
with education or income level (Tables 2 and 3).
9,30–35
It is noteworthy that in the age group <65 years,
a much lower proportion of diabetics were seen
among those with a high educational and high income
level and that this difference persisted in the age
group >65 years, although attenuated (Table 1). Rea-
sons for lack of optimal BP control in diabetics have
in other studies been explained by poor access to regu-
lar medical care or lack of health insurance,
33,34
but in
our study, other reasons should be considered. The
BP target <130/80 mmHg is more difficult to reach
than BP limits <140/90 mmHg in general, and in a re-
cent study, we found that not all diabetics in our study
population were treated with an ACE-I as recommen-
ded by guidelines, which could cause a lower degree
of BP control.
12
As opposed to diabetics, patients with a CVD had
much higher odds of optimal BP control (Table 3). Pa-
tients with CVD are often more intensively treated
with combinations of drugs, which could give a better
BP control. Having had a cardiovascular event, e.g.
myocardial infarction or apoplexy, could also increase
a patient’s compliance to treatment because of their
awareness and experiences with a life-threatening
disease closely related to hypertension.
Implication for practice and further research
Despite equal access to care, SES has a significant im-
pact on BP control in hypertensive patients treated in
general practice in Denmark. Future development of
tools to incorporate SES in the daily clinical work in
primary care should be given priority, when trying to
improve care of hypertensive patients. It should also
be emphasized to the GPs and the health care system
in general that SES should not be ignored when set-
ting strategies in the management of hypertension.
Diabetics are still far from achieving optimal BP con-
trol and there is an urgent challenge in improving the
inadequate treatment of diabetics.
Conclusions
SES has an impact on BP control for hypertensive pa-
tients treated in the Danish primary health care sys-
tem. Free access to care does not eliminate this
problem. Co-morbidities such as diabetes and CVD
also have a substantial influence on BP control,
irrespective of educational or income level.
Acknowledgements
We thank all the GPs, their staff and patients for par-
ticipating in the study, the secretary of APO for help
with data collection and support and administrator
Lise Stark for proofreading the manuscript.
Declaration
Funding: Danish Health Insurance Foundation
(2009B169) and Danish Heart Association (09-04-
R71-A2429_09-S2-22525F).
Ethical approval: Danish Data Protection Agency.
Conflict of interest: none.
References
1
Lang T, de GR, Chatellier G, Hamici L, Diene E. Prevalence and
therapeutic control of hypertension in 30,000 subjects in the
workplace. Hypertension 2001; 38: 449–54.
2
de Gaudemaris R, Lang T, Chatellier G et al. Socioeconomic in-
equalities in hypertension prevalence and care: the IHPAF
Study. Hypertension 2002; 39: 1119–25.
3
Mensah GA, Mokdad AH, Ford ES, Greenlund KJ, Croft JB. State
of disparities in cardiovascular health in the United States.
Circulation 2005; 111: 1233–41.
4
Avendano M, Glymour MM, Banks J, Mackenbach JP. Health dis-
advantage in US adults aged 50 to 74 years: a comparison of the
Page 7 of 8SES and BP control
508 Family Practice—The International Journal for Research in Primary Care
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
proportion of patients with low income or low edu-
cational level to be much smaller in the age group
<65 years than in the age group >65 years as seen in
Table 1. And if sampling bias is the main cause of this
difference, the association between education and BP
control for the lower educated and the low-income
group may have been underestimated in our study.
However, the difference in educational and income
levels between age groups observed in Table 1 is most
likely due to the demographic development in the
Danish population, where the younger patients have
longer educations than the elderly, and the younger
working population has higher income levels than the
elderly on retirement pension.
However, the effect of educational level on BP con-
trol remained after adjustment for a number of fac-
tors. It is not clearly defined through which pathways
educational level affects BP control. Higher education
could lead to a better understanding of risks and the
importance of treatment, but the impact of education
might also be mediated through other socio-economic
factors (Fig. 1) or through lifestyle.
25–27
Methodologi-
cal studies have recommended that socio-economic
factors like education, occupation and income are not
used interchangeably.
21,26,28
In our analysis, these fac-
tors were studied independently and we found differ-
ent effects of education and income on BP control
and a pronounced difference in the patterns according
to age.
It is worrying that low educational level is associ-
ated with lower OR of BP control in hypertensive pa-
tients <65 years, especially in a health care system
with equal access to care. In patients >65 years, edu-
cational level had no impact on BP control. Other fac-
tors such as physiology, co-morbidity, polypharmacy
and performance status can be more important for BP
control than educational level in the elderly. Further-
more, the age group >65 years had a high mean age
TABLE 3OR for optimal BP control according to socio-economic factors
Model 1
a
Education Income
Model 2
b
Model 3
c
Model 2
b
Model 3
c
Patients <65 years
Highest attained educational level (years)
<10 1 1 1
10–12 1.28 (1.04–1.58) 1.24 (1.01–1.53) 1.23 (1.01–1.52)
>12 1.20 (0.94–1.52) 1.13 (0.89–1.45) 1.18 (0.91–1.52)
Income
Low (1st quartile) 1 1 1
Medium (2nd + 3rd quartile) 1.20 (0.88–1.65) 1.13 (0.81–1.58) 1.09 (0.77–1.53)
High (4th quartile) 1.43 (1.04–1.96) 1.28 (0.92–1.79) 1.29 (0.92–1.81)
Cohabitation status
Single 1 1 1 1 1
Married/cohabiting 1.11 (0.90–1.36) 1.06 (0.86–1.32) 1.01 (0.81–1.25) 1.03 (0.82–1.29) 0.98 (0.78–1.22)
Co-morbidity
No co-morbidity 1 1 1 1 1
Diabetes 0.24 (0.17–0.36) 0.22 (0.15–0.34) 0.24 (0.15–0.36) 0.25 (0.17–0.37) 0.26 (0.17–0.39)
CVD 1.57 (1.14–2.17) 1.59 (1.15–2.20) 1.57 (1.13–2.18) 1.59 (1.14–2.21) 1.57 (1.13–2.19)
Patients >65 years
Highest attained educational level (years)
<10 1 1 1
10–12 0.96 (0.80–1.15) 0.93 (0.78–1.12) 0.98 (0.81–1.18)
>12 1.02 (0.82–1.26) 0.99 (0.80–1.23) 1.03 (0.82–1.29)
Income
Low (1st quartile) 1 1 1
Medium (2nd + 3rd quartile) 0.94 (0.79–1.11) 0.89 (0.74–1.06) 0.91 (0.75–1.10)
High (4th quartile) 0.99 (0.76–1.29) 0.91 (0.69–1.19) 0.91 (0.67–1.25)
Cohabitation status
Single 1 1 1 1 1
Married/cohabiting 1.20 (1.01–1.43) 1.14 (0.95–1.37) 1.21 (1.00–1.47) 1.17 (0.98–1.40) 1.22 (1.01–1.48)
Co-morbidity
No co-morbidity 1 1 1 1 1
Diabetes 0.30 (0.21–0.45) 0.31 (0.21–0.45) 0.33 (0.22–0.50) 0.30 (0.20–0.44) 0.32 (0.21–0.48)
CVD 1.36 (1.10–1.68) 1.38 (1.11–1.73) 1.38 (1.10–1.74) 1.34 (1.08–1.66) 1.34 (1.08–1.68)
a
Model 1 is OR (95% CI) analysed separately, adjusted for age and sex.
b
Model 2 is OR (95% CI) with educational level or income, adjusted for age, sex, cohabitation status, co-morbidity and number of antihypertensive
drug used as baseline characteristics.
c
Model 3 is OR (95% CI) for optimal BP control with educational level or income adjusted for age, sex, cohabitation status, co-morbidity, number
of antihypertensive drug used, BMI, smoking and alcohol.
Family Practice—The International Journal for Research in Primary CarePage 6 of 8
of 73.8 years. At this age, one would expect that the
lower educated patients included in our study are the
healthiest patients, who had survived long enough to
be included. This could have eliminated a possible ed-
ucational difference in BP control for this age group.
GPs should pay particular attention to patients’ edu-
cational level and, in case of the elderly, the cohabit-
ing status. This simple approach could strengthen the
attention towards patients without optimal BP control
and together with intensive BP follow-up and careful
information to this vulnerable group of hypertensive
patients, an increased focus would be on the patients
needed it. An important aspect to consider in this con-
text is the communication between the GP and the pa-
tient.
29
In the elderly, education and income level had
no impact on the ability to achieve BP control, while
being married or cohabiting increased the odds, which
could be due to the fact that the married/cohabiting el-
derly patients often visit their GP together with their
partner that might improve the understanding of the
information given by the GP.
We found that only 29.1% of all patients achieved
optimal BP control. Previous studies from Denmark
have shown that the proportion of hypertensive pa-
tients with controlled BP increased from 21% to 43%
from 1998 to 2004.
11,30
Our observation of only 29.1%
having achieved controlled BP may to some extent be
due to a higher proportion of hypertensive diabetics
in our study (Table 1).
11
BP targets for diabetics are
<130/80 mmHg, which is more difficult to achieve.
Table 2 illustrates this since only 10.9% of all diabetics
had controlled BP. Nevertheless, it is in accordance
with other studies that hypertensive diabetics had sig-
nificantly reduced OR of BP control and this reduced
OR persists in our study regardless of using a model
with education or income level (Tables 2 and 3).
9,30–35
It is noteworthy that in the age group <65 years,
a much lower proportion of diabetics were seen
among those with a high educational and high income
level and that this difference persisted in the age
group >65 years, although attenuated (Table 1). Rea-
sons for lack of optimal BP control in diabetics have
in other studies been explained by poor access to regu-
lar medical care or lack of health insurance,
33,34
but in
our study, other reasons should be considered. The
BP target <130/80 mmHg is more difficult to reach
than BP limits <140/90 mmHg in general, and in a re-
cent study, we found that not all diabetics in our study
population were treated with an ACE-I as recommen-
ded by guidelines, which could cause a lower degree
of BP control.
12
As opposed to diabetics, patients with a CVD had
much higher odds of optimal BP control (Table 3). Pa-
tients with CVD are often more intensively treated
with combinations of drugs, which could give a better
BP control. Having had a cardiovascular event, e.g.
myocardial infarction or apoplexy, could also increase
a patient’s compliance to treatment because of their
awareness and experiences with a life-threatening
disease closely related to hypertension.
Implication for practice and further research
Despite equal access to care, SES has a significant im-
pact on BP control in hypertensive patients treated in
general practice in Denmark. Future development of
tools to incorporate SES in the daily clinical work in
primary care should be given priority, when trying to
improve care of hypertensive patients. It should also
be emphasized to the GPs and the health care system
in general that SES should not be ignored when set-
ting strategies in the management of hypertension.
Diabetics are still far from achieving optimal BP con-
trol and there is an urgent challenge in improving the
inadequate treatment of diabetics.
Conclusions
SES has an impact on BP control for hypertensive pa-
tients treated in the Danish primary health care sys-
tem. Free access to care does not eliminate this
problem. Co-morbidities such as diabetes and CVD
also have a substantial influence on BP control,
irrespective of educational or income level.
Acknowledgements
We thank all the GPs, their staff and patients for par-
ticipating in the study, the secretary of APO for help
with data collection and support and administrator
Lise Stark for proofreading the manuscript.
Declaration
Funding: Danish Health Insurance Foundation
(2009B169) and Danish Heart Association (09-04-
R71-A2429_09-S2-22525F).
Ethical approval: Danish Data Protection Agency.
Conflict of interest: none.
References
1
Lang T, de GR, Chatellier G, Hamici L, Diene E. Prevalence and
therapeutic control of hypertension in 30,000 subjects in the
workplace. Hypertension 2001; 38: 449–54.
2
de Gaudemaris R, Lang T, Chatellier G et al. Socioeconomic in-
equalities in hypertension prevalence and care: the IHPAF
Study. Hypertension 2002; 39: 1119–25.
3
Mensah GA, Mokdad AH, Ford ES, Greenlund KJ, Croft JB. State
of disparities in cardiovascular health in the United States.
Circulation 2005; 111: 1233–41.
4
Avendano M, Glymour MM, Banks J, Mackenbach JP. Health dis-
advantage in US adults aged 50 to 74 years: a comparison of the
Page 7 of 8SES and BP control
SES and BP control 509
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
health of rich and poor Americans with that of Europeans. Am
J Public Health 2009; 99: 540–8.
5
Avendano M, Kunst AE, Huisman M et al. Socioeconomic status
and ischaemic heart disease mortality in 10 western European
populations during the 1990s. Heart 2006; 92: 461–7.
6
Grotto I, Huerta M, Sharabi Y. Hypertension and socioeconomic
status. Curr Opin Cardiol 2008; 23: 335–9.
7
Kaplan MS, Huguet N, Feeny DH, McFarland BH. Self-reported
hypertension prevalence and income among older adults in
Canada and the United States. Soc Sci Med 2010; 70: 844–9.
8
McWilliams JM. Health consequences of uninsurance among
adults in the United States: recent evidence and implications.
Milbank Q 2009; 87: 443–94.
9
Ong KL, Cheung BM, Man YB, Lau CP, Lam KS. Prevalence,
awareness, treatment, and control of hypertension among
United States adults 1999-2004. Hypertension 2007; 49: 69–75.
10
Munck A. [Audit on ‘‘Prevention in General Practice’’, Report.
Total Results from 394 General Practitioners.] Audit Project
Odense (APO).Odense, Denmark: Research Unit for General
Practice in Odense, 2005.
11
Kronborg CN, Hallas J, Jacobsen IA. Prevalence, awareness, and
control of arterial hypertension in Denmark. J Am Soc Hyper-
tens 2009; 3: 19–24.
12
Paulsen MS, Sondergaard J, Reuther L et al. Treatment of hyper-
tension in general practice: a cross-sectional study of 5413 hy-
pertensive patients. Fam Pract 2011; 28: 599–607.
13
Moller PK. Pricing and reimbursement of drugs in Denmark. Eur J
Health Econ 2003; 4: 60–5.
14
Bjerrum L, Munck A, Gahrn-Hansen B et al. Health Alliance for
Prudent Prescribing, Yield and use of Antimicrobial drugs in
the treatment of Respiratory Tract Infections (HAPPY AU-
DIT). BMC Fam Pract 2010; 11: 29.
15
Sondergaard J, Hansen DG, Aarslev P, Munck AP. A multifaceted
intervention according to the Audit Project Odense method
improved secondary prevention of ischemic heart disease:
a randomised controlled trial. Fam Pract 2006; 23: 198–202.
16
Bang LE. [Diagnostic Blood Pressure Measurement—Diurnal,
Home and Office Measurements].Copenhagen, Denmark:
The Danish Hypertension Society, 2006.
17
The Danish Society of General Practitioners, Christensen B. [Pre-
vention of Ischemic Cardiovascular Disease in General Practice,
2007 Disease in General Practice]. 3rd edn. Copenhagen,
Denmark: The Danish Society of General Practitioners, 2007.
18
Rasmussen JN, Rasmussen S, Gislason GH et al. Mortality after
acute myocardial infarction according to income and educa-
tion. J Epidemiol Community Health 2006; 60: 351–6.
19
Rasmussen JN, Rasmussen S, Gislason GH et al. Persistent socio-
economic differences in revascularization after acute myocar-
dial infarction despite a universal health care system—a Danish
study. Cardiovasc Drugs Ther 2007; 21: 449–57.
20
Rasmussen JN, Gislason GH, Rasmussen S et al. Use of statins and
beta-blockers after acute myocardial infarction according to in-
come and education. J Epidemiol Community Health 2007; 61:
1091–7.
21
Chaix B, Bean K, Leal C et al. Individual/neighborhood social factors
and blood pressure in the RECORD Cohort Study: which risk
factors explain the associations? Hypertension 2010; 55: 769–75.
22
Madsen M, Davidsen M, Rasmussen S, Abildstrom SZ, Osler M.
The validity of the diagnosis of acute myocardial infarction in
routine statistics: a comparison of mortality and hospital dis-
charge data with the Danish MONICA registry. J Clin Epide-
miol 2003; 56: 124–30.
23
Dalton SO, Steding-Jessen M, Gislum M et al. Social inequality and
incidence of and survival from cancer in a population-based
study in Denmark, 1994-2003: background, aims, material
and methods. Eur J Cancer 2008; 44: 1938–49.
24
Arendt JN, Jensen BT, Nexoe J, Muntzberg G, Sorensen J. [Low
income and number of contacts with general practice]. Ugeskr
Laeger 2010; 172: 1591–7.
25
Chaix B, Ducimetiere P, Lang T et al. Residential environment and
blood pressure in the PRIME Study: is the association medi-
ated by body mass index and waist circumference? J Hypertens
2008; 26: 1078–84.
26
Lahelma E, Martikainen P, Laaksonen M, Aittomaki A. Pathways
between socioeconomic determinants of health. J Epidemiol
Community Health 2004; 58: 327–32.
27
Andersen I, Gamborg M, Osler M, Prescott E, Diderichsen F. In-
come as mediator of the effect of occupation on the risk of
myocardial infarction: does the income measurement matter?
J Epidemiol Community Health 2005; 59: 1080–5.
28
Geyer S, Hemstrom O, Peter R, Vagero D. Education, income, and
occupational class cannot be used interchangeably in social ep-
idemiology. Empirical evidence against a common practice.
J Epidemiol Community Health 2006; 60: 804–10.
29
Naik AD, Kallen MA, Walder A, Street RL Jr. Improving hyper-
tension control in diabetes mellitus: the effects of collaborative
and proactive health communication. Circulation 2008; 117:
1361–8.
30
Sehestedt T, Ibsen H, Jorgensen T. Awareness, treatment and con-
trol of hypertension in Denmark. The inter99 study. Blood
Press 2007; 16: 312–19.
31
Bebb C, Coupland C, Stewart J et al. Practice and patient charac-
teristics related to blood pressure in patients with type 2 diabe-
tes in primary care: a cross-sectional study. Fam Pract 2007; 24:
547–54.
32
Hedblad B, Nerbrand C, Ekesbo R et al. High blood pressure despite
treatment: results from a cross-sectional primary healthcare-
based study in southern Sweden. Scand J Prim Health Care
2006; 24: 224–30.
33
Nelson SA, Dresser GK, Vandervoort MK et al. Barriers to blood
pressure control: a STITCH substudy. J Clin Hypertens (Green-
wich) 2011; 13: 73–80.
34
Borzecki AM, Berlowitz DR. Management of hypertension and di-
abetes: treatment goals, drug choices, current practice, and strat-
egies for improving care. Curr Hypertens Rep 2005; 7: 439–49.
35
Kearney PM, Whelton M, Reynolds K, Whelton PK, He J. World-
wide prevalence of hypertension: a systematic review. J Hyper-
tens 2004; 22: 11–9.
Family Practice—The International Journal for Research in Primary CarePage 8 of 8
510 Family Practice—The International Journal for Research in Primary Care
by guest on November 7, 2015http://fampra.oxfordjournals.org/Downloaded from
