European Journal of Internal Medicine 13 (2002) 369–375

www.elsevier.com / locate / ejim

Original article

Hypertension is frequently present in patients with reflux esophagitis or

Barrett’s esophagus but not in those with non-ulcer dyspepsia
a,b c b d a,
Sunna Gudlaugsdottir , W.M. Monique Verschuren , Jan Dees , Theo Stijnen , J.H. Paul Wilson *
a
Department of Internal Medicine, University Hospital Rotterdam, Dr Molewaterplein 40, 3015 GD Rotterdam, The Netherlands
b
Department of Gastroenterology and Hepatology, University Hospital Rotterdam, Rotterdam, The Netherlands
c
National Institute of Public Health and the Environment ( RIVM), Rotterdam, The Netherlands
d
Institute of Epidemiology and Biostatistics, Erasmus University Rotterdam, Rotterdam, The Netherlands

Received 28 June 2001; received in revised form 3 December 2001; accepted 20 December 2001

Abstract

Background: Elevated mortality due to cardiovascular disease has been reported for patients with Barrett’s esophagus (BE). We
compared the prevalence of risk factors for cardiovascular disease in patients with BE, reflux esophagitis (RE), and non-ulcer dyspepsia
(NUD) with that of the general population. Methods: Patients with upper gastrointestinal complaints and BE, RE, or NUD were
compared with a matched cohort from the general population using a questionnaire and blood pressure and cholesterol measurements.
Results: Hypertension occurred more frequently in patients with BE (odds ratio 5.1, P,0.0001) and RE (odds ratio 3.8, P,0.001), but
not in those with NUD. Serum total cholesterol was higher in BE (P50.02) and borderline in RE (P50.06) but not in NUD. Mean HDL
cholesterol levels, body mass index, and smoking did not differ. Conclusions: This study suggests that BE and RE found at diagnostic
endoscopy are associated with an increased prevalence of hypertension and a higher total cholesterol level than in the general population.
If so, this would explain the increased mortality during the follow-up of BE patients, and it should be taken into account when designing
or evaluating follow-up studies of BE.
 2002 Elsevier Science B.V. All rights reserved.

Keywords: Barrett’s esophagus; Reflux esophagitis; Non-ulcer dyspepsia; Esophageal cancer; Coronary heart disease; Hypertension

1. Introduction have cast doubt on the effectiveness of endoscopic biopsy

surveillance since few patients diagnosed as having BE
The prevalence of endoscopically confirmed esophagitis actually die from adenocarcinoma of the esophagus [7–
in the community is thought to be as much as 2% [1]. 10].
Esophagitis can progress to complications such as deep In a follow-up study of 166 patients in whom the
ulceration, stricture formation, and the development of diagnosis BE had been established between 1973 and
Barrett’s esophagus (BE) [2]. Because BE has a pre- 1986, we found an elevated mortality (50%) compared to
malignant potential [3] and because of the dramatically what was expected in an age and sex-matched control
rising incidence of adenocarcinoma in the esophagus in the population [10]. During a mean follow-up of 9.3 years
past decade [4], several groups have proposed endoscopic (amounting to 1440 patient years), eight patients developed
biopsy surveillance for BE patients to detect malignancy at esophageal cancer at random intervals, giving one case in
an early and curable stage [5,6]. However, earlier studies 180 patient years. Seventy-nine patients died, one-third due
to cardiovascular disease (CVD), and in only two cases
*Corresponding author. Tel.: 131-10-463-5940; fax: 131-10-463- was esophageal cancer the cause of death [10]. These
3268. results raised the hypothesis that patients with gas-
E-mail address: [email protected] (J.H.P. Wilson). troesophageal reflux disease (GERD), or at least a sub-

0953-6205 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved.
PII: S0953-6205( 02 )00090-0
370 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375

group with BE, have an increased risk of CVD with an tions used for comparison in the present study: (1) Do you
increased prevalence of one or more CVD risk factors. currently use medication for hypertension? (2) Have you
This study was undertaken to investigate the prevalence of ever smoked? If yes → (3) Do you still smoke? Answering
the main risk factors for CVD (blood pressure, cholesterol, categories: yes (→ current smokers); If no (→ ex-smokers)
smoking) in patients with BE, in patients with reflux (4) Have you ever smoked? Answering categories: no,
esophagitis (RE), and in patients without endoscopic never (→ never smokers). Body mass index (BMI) was
abnormalities and to compare these with the normal calculated as weight (kg) / height (m)2 . Blood pressure was
population. The study was part of an epidemiological measured once at rest at the end of the outpatient visit with
investigation into the prevalence of Helicobacter pylori in a mercury sphygmomanometer. Systolic pressure was
RE, BE, and non-ulcer dyspepsia (NUD) patients. recorded at the first Korotkoff phase and diastolic pressure
at the fifth Korotkoff phase.
The Medical Ethics Committee of the University Hospi-
2. Materials and methods tal Rotterdam approved the study. Informed consent was
obtained from all participating patients before the endo-
The patients we studied were 20–59.9 years of age and scopy was performed.
had RE, BE, or NUD. Two sources of data were used: One A reference population was derived from the Monitoring
allowed us to review computerized endoscopic and medi- Project on Risk Factors for Chronic Diseases (MORGEN
cal data of all patients with BE found at routine endoscopy project). This project was carried out in the Netherlands
in the University Hospital Rotterdam in the years 1992– from 1993 to 1997. The general purpose was to determine
1996. Information regarding the length of BE and the both the prevalence of risk factors for chronic disease (e.g.
histological criterion of a biopsy sample from the BE was plasma cholesterol, blood pressure, smoking habits) and
noted. Patients were included if the BE met the criterion of the prevalence of some specific, chronic conditions in a
columnar mucosa of at least 3 cm in the tubular esophagus random sample of the general population [13]. Information
found at endoscopy or histological proof of intestinal on current health status, medical history, current medica-
metaplasia (IM) when the columnar mucosa was less than tion use, smoking behavior, and family history was
3 cm [11]. The patients studied were offered repeat obtained using a self-administered questionnaire at home.
esophagogastroduodenoscopy (EGD) with a biopsy to We asked the following questions for comparison with the
confirm the diagnosis of BE. If the patient refused EGD, study groups: (1) Do you currently use medication for
we used the older data about the length of the Barrett’s hypertension? (2) Do you smoke? Answering categories:
segment and histology. Secondly, all consecutive patients Yes (→ current smokers), No, not anymore (→ ex-smok-
with GERD or with no endoscopic abnormalities who were ers), No, and have never smoked (→ never smokers).
referred to the University Hospital Rotterdam–Dijkzigt for Height and weight were measured with participants wear-
diagnostic upper gastrointestinal endoscopy during the ing indoor clothes, without shoes, and with empty pockets.
period January 1999 to July 2000 were asked to participate Body mass index was calculated as weight (kg) / height
in the study. GERD was defined as the presence of (m)2 . At the research center, the systolic and diastolic
endoscopic signs of RE, BE, or both. RE was scored blood pressures were measured twice on the right arm in
according to the Savary–Miller system [12] as follows: all subjects in a sitting position with a mercury manometer
grade 0, normal esophageal mucosa with no abnormalities; (random zero sphygmomanometer). Systolic pressure was
grade 1, mucosal erythema or diffusely red mucosa, with recorded at the first Korotkoff phase and diastolic pressure
or without friability; grade 2, linear erosions extending at the fifth Korotkoff phase.The average of the two
from the gastroesophageal junction upward in relation to measurements was used in the original analysis, but in
the folds; grade 3, confluent erosions extending around the present study we used the first measurement for reasons of
entire circumference; grade 4, frank ulcer, stricture, or BE. comparison.
If the patient had no malignancies or active peptic ulcer In both studies the participants were asked to write
disease, he or she was asked to fill in a standard ques- down all medication used and blood was taken for
tionnaire about the presence of risk factors of CVD, determination of serum total and high-density lipoprotein
current medication, and smoking behavior. One patient (HDL) cholesterol concentrations with an automated en-
with BE was excluded because he was unable to travel. zymatic procedure.
With a simple checklist we ascertained the following In both studies all cholesterol determinations were
demographic characteristics of all three study cohorts: age, performed at the Clinical Chemistry Laboratory of the
sex, weight, height, smoking (current, past, never), date of University Hospital in Rotterdam, which is the coordinator
first diagnosis of BE, and date of last endoscopy. of the Dutch National Cholesterol Standardization Pro-
A blood sample for laboratory testing was taken at the gram. All respondents sat down for 5 min before the blood
outpatient clinic for most of the BE patients and on the day pressure measurement was taken. The criterion for hy-
of endoscopy for the other participants. The study popula- pertension used in all groups (both studies) was at least
tions (RE, BE, and NUD) answered the following ques- one of the following factors: systolic blood pressure $160
 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375 371

mmHg, diastolic blood pressure $95 mmHg, or use of Table 2

Endoscopic features
antihypertensive medication. According to the Sixth Re-
port of the Joint National Committee on Prevention, Reflux esophagitis Barrett’s esophagitis
Detection, Evaluation, and Treatment of High Blood Grade I 5
Pressure (JNC VI) and the World Health Organization, Grade II 11
which defines and classifies hypertension in adults, this Grade III –
Grade IV 15
blood pressure range represents mild to moderate hyperten-
Length BE (cm) mean (range) 4.7 (0.8–8.5)
sion [14,15]. In the present study, a systolic blood pressure Intestinal metaplasia (%) 30 (88)a
of 160 mmHg or more and a diastolic blood pressure of 95 Dysplasia (%) 6 (17.6)a
mmHg or more was chosen since we only have single Number of subjects per grade according to the Savary–Miller endo-
measurements in these cohorts and since these values have scopic scoring system.
previously been used as cut-off points for the definition of a
Information was not available for two subjects.
hypertension in a population study [16]. The current
systolic blood pressure threshold for hypertension treat-
ment is 140 mmHg for all adults. The WHO and the homogeneity test was performed to investigate whether the
International Society of Hypertension have proposed that odds ratio differed between strata.
normal pressure be lower than 130 mmHg, with an
optimum pressure of less than 120 mmHg [14,15]. The
cut-off limit of our study could exclude individuals with 3. Results
mild hypertension, but is unlikely to misclassify normoten-
sive persons. Eighty-six patients aged 20–59.9 years with upper
gastrointestinal complaints in whom BE (n534), RE (n5
2.1. Statistical analysis 31), or NUD (n521) was diagnosed at routine endoscopy
took part in the study and were evaluated. We found 212
For the continuous, potential risk factors (BMI, SBP, patients (137 men and 75 women) diagnosed with BE who
DBP, total and HDL cholesterol), the study group was were alive and without adenocarcinoma in the age range
compared with the reference population as follows. The 18–80 years. One hundred and nine responded to the
reference population was stratified by sex and 5-year age invitation to participate. Of these, 88 were found to have
categories. The mean was computed per stratum. From the BE at repeat endoscopy, 34 of whom were in the age
observed value of each patient in the study group, the category 20–59.9 years. The other participating individuals
mean of the corresponding stratum of the reference group were all consecutive patients with GERD or no endoscopic
was subtracted, and the average difference was determined. abnormalities who were referred to the University Hospital
A 95% confidence interval was computed and the null Rotterdam–Dijkzigt for diagnostic upper gastrointestinal
hypothesis of no mean difference was tested with Student’s endoscopy during the period January 1999 to July 2000.
one-sample test. Since the stratum sizes of the reference Demographic characteristics are given in Table 1 and
group were large (.1000), there was no need to account endoscopic features in Table 2. Twenty-seven patients
for the uncertainty in the stratum-specific means. To (79%) in the BE group were using proton pump inhibitors
investigate whether the difference between the study and (mostly omeprazole). This was less frequent in the RE
reference population was related to age and / or sex, group (17 patients, 55%) and in the NUD group (only four
multiple linear regression analysis was used with (continu- patients,19%; P,0.0001).
ous) age and sex as independent variables. For the cate- The prevalence of risk factors for CVD are shown in
gorical potential risk factors (smoking behavior and hy- Tables 3 and 4. Body mass index was not significantly
pertension), both the study and the reference groups were different between groups. Mean systolic blood pressure
stratified by sex and 5-year age categories, followed by was significantly higher in the BE and RE groups than in
computation of the Mantel–Haenszel odds ratio, the 95% controls, as can be seen in Table 3, which shows the mean
confidence interval, and the Mantel–Haenszel test. A blood pressures of all patients within a group, whether or

Table 1
Demographic characteristics of the patients in the three subgroups
Reflux esophagitis Barrett’s esophagus Non-ulcer dyspepsia
Age 44610 4868.4 44611
Number of subjects 31 34 21
Male 21 26 11
Weight (kg)a 81618 77615 76611
Height (cm)a 175610 175610 175610
a
Means6S.D.
372 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375

Table 3
Mean level of cardiovascular risk factors
Variable Mean (S.D.) P value 95% CI
Study group Controls
Body-mass index
Barrett’s esophagitis group 25.5 (3.6) 26.0 (0.9) 0.5 21.7–0.8
Reflux esophagitis group 26.1 (6)a 25.5 (0.6) 0.6 21.7–2.9
Non-ulcer dyspepsia group 24.6 (4.4) 25.5 (0.3) 0.3 22.8–1
Systolic blood pressure, mmHg (S.D.)
Barrett’s esophagus group 136 (17) 127 (5) 0.004 3.4–16
Reflux esophagitis group 134 (16) 124 (6) 0.002 4.2–16.5
Non-ulcer dyspepsia group 129 (14) 124 (6) 0.07 20.5–11
Diastolic blood pressure, mmHg (S.D.)
Barrett’s esophagus group 82 (11) 80 (3) 0.4 22.4–5.3
Reflux esophagitis group 83 (14) 79 (3) 0.05 5.8–10
Non-ulcer dyspepsia group 80 (9) 78 (3) 0.2 21.6–5.7
Serum total cholesterol, mmol / l (S.D.)
Barrett’s esophagus group 6.0 (1.2) 5.5 (0.3) 0.02 6.8–0.9
Reflux esophagitis group 5.7 (1.0) 5.4 (0.4) 0.06 21.5–0.7
Non-ulcer dyspepsia group 5.8 (1.5) 5.4 (0.5) 0.3 20.3–1
Serum HDL cholesterol (S.D.)
Barrett’s esophagus group 1.38 (0.4) 1.26 (0.4) 0.08 22.3–0.3
Reflux esophagitis group 1.27 (0.5) 1.29 (0.1) 0.3 20.2–0.2
Non-ulcer dyspepsia group 1.35 (0.3) 1.34 (0.2) 0.9 20.2–0.2
95% CI595% confidence interval.
Body-mass index was calculated as the weight in kilograms divided by the square of the height in meters.
P values were calculated using the one-sample Student’s t-test.
a
One subject was not available for analysis.

not they were on antihypertensive treatment. The mean NUD patients, followed by ACE inhibitors: in four (12%),
systolic blood pressure in the NUD group was higher than one (3.3%), and one (5%), respectively. In the third and
in the reference population, although the difference was fourth places came calcium antagonists: two (6%), two
not statistically significant. The mean diastolic blood (6.4%), and one (5%), and hydrochloorthiazide: two (6%),
pressure was significantly higher in the RE group, but it two (6.4%), and three (14%), respectively. Combination
did not differ in the BE and NUD groups. The frequencies treatment with beta blockers and calcium antagonists was
of hypertension in study groups and controls are given in being used by two BE patients, one RE patient, and one
Table 4. Hypertension occurred significantly more fre- NUD patient. Two BE patients were being given the
quently among the BE (P,0.0001; odds ratio 5.1) and RE combination of an ACE inhibitor and a beta blocker, as
(P,0.001; odds ratio 3.8) patients than among controls, was one RE patient. Five of the patients with GERD and
but not among patients in the NUD group (P50.34; odds one with NUD, diagnosed as having hypertension, had a
ratio 1.7). Beta blockers were the most frequently given history of myocardial infarction. An additional three
treatment for hypertension in the three study groups: in ten normotensive patients had a history of myocardial infarc-
(29%) BE patients, two (6%) RE patients, and two (9.5%) tion. Five patients were known to have diabetes mellitus;

Table 4
Prevalence of hypertension
Hypertension (%) P value Odds ratio 95% CI
(total group)
Barrett’s esophagus 47 ,0.0001 5.61 2.5–10.0
32
19
8
Reflux esophagus ,0.001 3.76 1.7–8.3
Non-ulcer dyspepsia 0.34 1.7 0.5–5.0
Reference group – – –
Values for hypertension are shown as percent of hypertension in the three study and in the reference population.
P values were calculated using the Mantel–Haenszel test stratified on age and sex. 95% CI595% confidence interval.
 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375 373

no new cases were identified by blood sugar measure- factors. The individuals participating in the MORGEN
ments. Two of the diabetics were also hypertensive. study were considered a valid control group because
Two known hypertensives had a serum creatinine slight- participants were derived from the general population
ly above the normal limit, as did one normotensive patient. living in the same area and examined in the same time
One normotensive patient had a serum creatinine of 310 period. Comparability of the two sets of data was achieved
mmol / l. Serum creatinine values were normal in all other by using the same items from the standard questionnaire of
cases. the reference group and the three study populations with
Serum total cholesterol levels (Table 3) were signifi- regard to the presence of CVD risk factors. Therefore, we
cantly higher in the BE group (P50.02), with a trend in could not use questions about known family history of
the RE group (P50.06), but not in the NUD group. HDL CVD for comparison because of differences in the phras-
cholesterol levels and the prevalence of smoking habits did ing of the questions used.
not differ between the study groups and controls. The The criterion for hypertension was based on both blood
prevalence of both current smoking and no smoking tended pressure measurements and treatment for hypertension.
to be highest in the youngest age categories (almost 50% in Therefore, it could be argued that the difference in
both categories) but decreased with age to the lowest levels prevalence of hypertension was due to selection bias, as
of 30 and 20%, respectively, in the 55–59.9 years cate- the study patients, many being hospital outpatients, might
gory. Past smoking was lowest in the youngest age conceivably be a group requiring more frequent medical
category—around 10%—and highest in the 55–59.9 cate- attention, which would increase the chance of detection
gory—almost 50%. Mean blood glucose level was lower in and treatment of hypertension. However, the invitation also
the RE and NUD groups. included those who had only come for diagnostic endo-
scopy but were not known as outpatients at our hospital. It
should, therefore, represent the great majority of patients
4. Discussion currently diagnosed with RE, BE, or NUD. The difference
is difficult to explain purely in terms of selection bias. In
This is the first study to consider the issue of the all participants blood pressure was measured once with a
prevalence of risk factors for CVD in patients with all sphygmomanometer in a sitting position after 5 min of
grades of GERD. It was undertaken in patients with rest. We cannot exclude the possibility that the higher
endoscopically confirmed and clearly defined subgroups of levels of blood pressure in the RE and BE groups,
GERD and in patients without endoscopic abnormalities, compared to those in the control group, were due to
classified as NUD. We observed a higher blood pressure in circumstances surrounding the measurements (the ‘white
the RE and BE groups than in the reference population, but coat effect’ of a physician in the study groups but a
not in the NUD group. Whether hypertension occurred research assistant in the MORGEN project) [17]. The
before or after the start of gastrointestinal complaints could phenomenon of ‘white coat hypertension’ refers to hy-
not be assessed in this study (the questions asked did not pertension in a health care institute or doctor’s office; it
specify the chronology of complaints). We also observed a involves multiple factors and does not correlate with the
significantly higher serum total cholesterol in the BE pressor response to the doctor [15]. Thus, it should imply
group, with a trend in the RE patients, but not in the NUD that isolated office hypertension, which might also be the
group. The mean HDL cholesterol level was borderline earliest manifestation of real hypertension [18], is equally
significantly higher in the BE group. However, this did not important in both cases and controls. The same remarkable
differ between the two other study groups and the refer- outcome was not seen in the NUD group compared to
ence group. Although BMI was higher in BE patients than controls, which makes a white coat effect unlikely as a
in controls, the difference did not reach statistical signifi- reason for higher blood pressure measurements in the other
cance. groups. A single measurement does not represent a per-
To appreciate the findings of this study, certain meth- son’s average or usual blood pressure [19], leading to
odological aspects should be considered. The diagnoses in overestimation of the prevalence of hypertension in both
the study groups were originally made with diagnostic the cases and controls. We believe that by depending only
endoscopy in patients with upper gastrointestinal com- on the first measurement in both the patient and the
plaints. All patients meeting our criteria for RE, BE, or reference groups, the comparison we made is reliable.
NUD and diagnosed in the abovementioned periods were In the RE, BE, and NUD groups, weight and height
asked to participate in a study to look for the prevalence of were often self-reported, in contrast to the MORGEN
Helicobacter pylori infection. They were not asked to study. Self-reporting tends to give 5–7% lower BMIs than
participate in a study to detect risk factors for CVD. A measured BMIs, which tend to be more biased in older and
notice was included in the invitation that we would like to overweight groups [20]. However, self-reporting on BMI is
know a few things about their general health status. more reliable when used on a younger population [21]. Our
Therefore, it is not likely that the responses of the three three study populations were relatively young (mean age
study groups were influenced by the presence of CVD risk 44 years in the RE and NUD groups and 48 years in the
374 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375

BE group), but due to the broad age range (20–59.9 years) prospective and include multiple measurements of blood
it could mean that BMI may have been underestimated in pressure, examination for organ damage, such as re-
the present study, although this is less likely in the tinopathy, and an evaluation of factors that may contribute
youngest age group. to elevated blood pressure, such as salt intake.
While our findings may be true, the underlying mecha- In conclusion, BE and RE found at diagnostic endo-
nisms are not known. Although high cholesterol may scopy are associated with an increased prevalence of
contribute to an increased risk of CVD, differences in the hypertension and possibly a higher total cholesterol level
prevalence of hypertension in the present study were than in the general population. If our results can be
statistically more striking. Because we matched for age confirmed, they will provide an explanation for the in-
and gender, age does not play a role in the comparison. creased mortality during the follow-up of BE patients.
Since BMI did not differ between groups, obesity as a They should also be taken into account when designing or
hypertensive factor in the RE and BE groups seems evaluating follow-up studies or selecting patients with BE
unlikely. We did not include any additional investigations for surveillance.
in this study to determine whether the patients had
secondary hypertension or other factors known to increase
blood pressure, such as insulin resistance, high alcohol and References
salt intake; furthermore, sedentary lifestyle, stress, and low
potassium and calcium intake were not examined. One [1] Spechler SJ. Epidemiology and natural history of gastro-esophageal
could hypothesize that genetic factors play a role. The reflux disease. Digestion 1992;51(Suppl 1):24–9.
influence of genes on blood pressure has been suggested [2] McDougall NI, Johnston BT, Kee F, Collins JS, McFarland RJ,
by family studies, and mutations in at least ten genes have Love AH. Natural history of reflux oesophagitis: a 10 year follow up
been shown to raise or lower blood pressure [18]. Recent- of its effect on patient symptomatology and quality of life. Gut
1996;38:481–6.
ly, on the basis of clustering of GERD in families, it has
[3] Naef AP, Savary M, Ozzello L. Columnar-lined lower esophagus:
also been suggested that there may be a GERD gene An acquired lesion with malignant predisposition. Report on 140
[22,23]. The same causative factors—genetic or environ- cases of Barrett’s esophagus with 12 adenocarcinomas. J Thorac
mental or both—may be operative in GERD diseases and Cardiovasc Surg 1975;70:826–35.
hypertension. However, further studies need to be per- [4] Blot WJ, Devesa SS, Kneller RW, Fraumeni JF. Rising incidence of
adenocarcinoma of the esophagus and gastric cardia. J Am Med
formed to establish this possibility.
Assoc 1991;265:1287–9.
In an earlier follow-up study at our university hospital, [5] van Sandick JW, van Lanschot JJ, Kuiken BW, Tytgat GN,
we found surprisingly high mortality rates from CVD in a Offerhaus GJ, Obertop H. Impact of endoscopic biopsy surveillance
cohort of 166 patients in whom the diagnosis BE had been of Barrett’s oesophagus on pathological stage and clinical outcome
established between 1973 and 1986 [10]. In retrospect, of Barrett’s carcinoma. Gut 1998;43:216–22.
[6] Cameron AJ, Carpenter HA. Barrett’s esophagus, high-grade dys-
those patients would not have benefited from endoscopic
plasia, and early adenocarcinoma: a pathological study. Am J
surveillance because their elevated mortality was due to Gastroenterol 1997;92:586–91.
CVD in one-third of all cases. This is in agreement with a [7] Spechler SJ, Robbins AH, Rubins HB et al. Adenocarcinoma and
recently published long-term observational study from Barrett’s esophagus. An overrated risk? Gastroenterology
MacDonald et al. [24] on BE patients at the University 1984;87:927–33.
[8] Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma
Hospital in the United Kingdom. This high mortality from
in columnar-lined (Barrett’s) esophagus. New Engl J Med
cardiovascular diseases rather than cancer was recently 1985;313:857–9.
reported as a surprising finding in a long-term follow-up [9] Van der Veen AH, Dees J, Blankensteijn JD, van Blankenstein M.
study of GERD patients treated surgically or medically Adenocarcinoma in Barrett’s oesophagus: an overrated risk. Gut
[25]. Recently, in an elderly population with BE, we found 1989;30:14–8.
[10] van der Burgh A, Dees J, Hop WCJ, van Blankenstein M.
a higher prevalence of myocardial infarction with a notably
Oesophageal cancer is an uncommon cause of death in patients with
higher prevalence of hypertension in the same group [26]. Barrett’s oesophagus. Gut 1996;39:5–8.
The present study was performed on a younger category of [11] Sharma P, Morales TG, Sampliner RE. Short segment Barrett’s
the whole spectrum of GERD patients and was compared esophagus: the need for standardization of the definition and of
with another age and sex-matched reference population. endoscopic criteria. Am J Gastroenterol 1998;93:1033–6.
[12] Little AG, DeMeester TR, Kirchner PT, O’Sullivan GC, Skinner
Again, we noticed a noteworthy high prevalence of
DB. Pathogenesis of esophagitis in patients with gastroesophageal
hypertension in the GERD population, particularly within reflux. Surgery 1980;88:101–7.
the BE group. [13] Houterman S, Verschuren WMM, Oomen CM, Boersma-Cobbaert
Cardiovascular and overall mortality are known to CM, Kromhout D. Trends in total and HDL cholesterol and their
increase continuously with increasing blood pressure [19]. determinants in The Netherlands between 1993 and 1997. Int J
Epidemiol 2001;30:1063–10670.
If our findings are true, they could explain part of the
[14] JNC. The sixth report of the Joint National Committee on preven-
observed higher mortality in the BE population in the tion, detection, evaluation, and treatment of high blood pressure.
follow-up study performed in our hospital performed some Arch Intern Med 1997;157:2413–46.
years ago [10]. Further studies are needed. These should be [15] WHO. 1999 World Health Organization—International Society of
 S. Gudlaugsdottir et al. / European Journal of Internal Medicine 13 (2002) 369–375 375

Hypertension Guidelines for the Management of Hypertension. [22] Trudgill NJ, Kapur KC, Riley SA. Familial clustering of reflux
Guidelines Subcommittee. J Hypertens 1999;17:151–83. symptoms. Am J Gastroenterol 1999;94:1172–8.
[16] Hofman A, Grobbee DE, de Jong PT, van den Ouweland VM. [23] Romero Y, Cameron AJ, Locke III GR et al. Familial aggregation of
Determinants of disease and disability in the elderly: the Rotterdam gastroesophageal reflux in patients with Barrett’s esophagus and
Elderly Study. Eur J Epidemiol 1991;7:403–22. esophageal adenocarcinoma. Gastroenterology 1997;113:1449–56.
[17] Pickering TG. White coat hypertension: time for action. Circulation [24] Macdonald CE, Wicks AC, Playford RJ. Final results from 10 year
1998;98:1834–6. cohort of patients undergoing surveillance for Barrett’s oesophagus:
[18] Bidlingmeyer I, Burnier M, Bidlingmeyer M, Waeber B, Brunner observational study. Br Med J 2000;321:1252–5.
HR. Isolated office hypertension: a prehypertensive state? J Hy- [25] Spechler JS, Lee E, Ahnen D et al. Long-term outcome of medical
pertens 1996;14:327–32. and surgical therapies for gastroesophageal reflux disease. J Am
[19] Carretero OA, Oparil S. Essential hypertension. Part I. Definition Med Assoc 2001;285:2331–8.
and etiology. Circulation 2000;101:329–35. [26] Gudlaugsdottir S, Romme AM, Dees J, Wilson JHP. The prevalence
[20] Roberts RJ. Can self-reported data accurately describe the preval- of coronary heart disease and its risk factors in patients with
ence of overweight? Public Health 1995;109:275–84. Barrett’s esophagus [abstract]. Eur J Gastroenterol Hepatol
[21] Alvarez-Torices JC, Franch-Nadal J, Alvarez-Guisasola F, Hernan- 2000;12:A4.
dez-Mejia R, Cueto-Espinar A. Self-reported height and weight and
prevalence of obesity. Study in a Spanish population. Int J Obes
Relat Metab Disord 1993;17:663–7.