Open Journal of Gastroenterology, 2013, 3, 231-236 OJGas
doi:10.4236/ojgas.2013.34039 Published Online August 2013 (
Predictors of H. pylori infection in a safety-net hospital in
Sailaja Boddu, David Drachman, Jyotsna Ravi, Abdul Nadir
Maricopa Medical Center, Phoenix, USA
Received 22 June 2013; revised 25 July 2013; accepted 4 August 2013
Copyright © 2013 Sailaja Boddu et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Objective: The purpose of this study was to deter-
mine the risk factors associated with having H. pylori
infection as proven by endoscopic biopsy at Maricopa
Medical Center (MMC), a safety-net hospital in phoe-
nix, Arizona which serves primarily patients with
limited financial and insurance resources. Methods:
A total of 1116 biopsies were identified in a Depart-
ment of Pathology database searched from Novem-
ber 2004 to March 2013. To be included, the subjects
had to have an endoscopy with gastric biopsy. After
the inclusion criteria were applied, 282 control sub-
jects without histological evidence of H. pylori infec-
tion and 256 cases with histological evidence of H.
pylori infection were identified. Patient charts were
reviewed to extract information on variables collected
for this study. Results: The mean age of cases and
controls was 50.5 and 52.3 years respectively. The
BMI of the cases and controls was 28.1 and 28.0 res-
pectively. The mean number of upper endoscopic
exams performed was 1.3 in cases and 1.4 in controls
respectively. Potential predictors examined were gen-
der, history of drug abuse, history of alcohol abuse,
chronic pain medication use, smoking, employment
status, outpatient vs. inpatient upper endoscopy exam,
language spoken (English, Spanish, or bilingual), race/
ethnicity, type of insurance, heart burn, dysphagia,
abdominal pain, gastroduodenal ulcers, intestinal me-
taplasia, and having vs. not having a primary care
physician. Based on univariate analyses, having a
gastroduodenal ulcer, having a history of abdominal
pain, Hispanic race, government insurance status,
self-pay insurance status, and speaking Spanish only
were found to be significantly related to having H.
pylori infection. These variables were next entered
into a multivariate analysis. The multivariate analysis
revealed that gastroduodenal ulcer, abdominal pain,
Hispanic race and self-pay insurance status remained
significant predictors of H. pylori infection. For the
last 119 subjects, information regarding the country
of their birth was available. These 119 subjects were
separately analyzed. Country of birth, age, sex, His-
panic race, self-pay insurance status, Spanish speaker,
and having government insurance were included in a
multivariate analysis. Twenty seven percent of pa-
tients without H. pylori were born in the US, com-
pared to fifty eight percent H. pylori positive patients
who were born outside of the US. Conclusion: His-
panics are at high risk for H. pylori infection. In a
multivariate model, pa tients with H. pylori were found
to have a higher risk of ulcers, Hispanic race, abdo-
minal pain and self-insurance status. When self-insur-
ance and Hispanic race were included in another ana-
lysis with place of birth, only birth outside US re-
mained significant.
Keywords: H. pylori; Insurance; Foreign Born
Despite the decreasing prevalence of H. pylori within the
indigenous United States population [1], migration from
Mexico, particularly in Texas and California [2], has re-
sulted in a higher burden of H. pylori infection. Data
from El Paso, Texas, and Ciudad Juarez, Mexico utiliz-
ing stool antigen test for H. pylori among asymptomatic
patients revealed a prevalence 38.2% of H. pylori [3]. In
the San Francisco Bay area, the prevalence of H. pylori
in immigrants, first degree and second generation US-
born Hispanics were reported at 31.4%, 9.1% and 3.1%
respectively [4]. Recently, 37.9% prevalence of H. pylori
infection based on gastric biopsies was reported from
northwestern Ontario [5]. The Canadian Helicobacter Study
group has recommended focusing on treating Canadians
*This research was presented as a poster at Digestive Disease Week
(DDW) 2013 in Orland o, Florida.
Published Online August 2013 in SciRes.
S. Boddu et al. / Open Journal of Gastroenterology 3 (2013) 231-236
with the highest prevalence of H. pylori infection [6].
Based on 2010 US census data, 26.9% of the popula-
tion in Maricopa County Arizona comprised individuals
of Hispanic or Latino origin, third after Texas and Cali-
fornia, where 37.6% of the population is Hispanic [2].
We did not find any data on the characteristics of pa-
tients with H. pylori infection in Arizona which shares
approximately 370 miles of porous border with Mexico.
The purpose of this study was to de termine the predicto rs
of H. pylori infection in patients seen at a safety-net hos-
pital located in Maricopa County, Pho enix, Arizona.
Maricopa Medical Center (MMC) is located in Phoenix.
The hospital is supported by the government to provide
care to individuals who are either on Medicaid (gover-
nment support) or are uninsured. Less than 5% of pa-
tients with commercial insurance are also seen in this
hospital. The hospital is staffed by full-time employed
gastroenterologists who see patients in the clinics as well
as when they are admitted to the hospital. Typically pa-
tients admitted to the hospital undergo upper endoscopy
examinations (EGD) for GI bleeding or unexplained up-
per abdominal pain, while outpatients undergo EGDs for
dyspepsia, abdominal pain, occult GI bleeding or screen-
ing for varices.
We identified 1116 biopsies obtained during EGD in
the department of pathology database from November
2004 to March 2013. Patients having only esophageal
or/and duodenal biopsies were excluded, and only pa-
tients with EGD and gastric biopsies with or without any
other biopsies were included. A total of 256 patient s were
identified to have histological evidence of H. pylori in-
fection. Another 282 patients without histological evi-
dence of H. pylori were selected as controls. Gastric bi-
opsies were read by four pathologists at MMC. Modi-
fied Giemsa stain was used for confirmation of H. pylori
infection. All endoscopists obtained at least two biop-
sies from the gastric antrum.
Data were retrospectively collected on age, gender, sex,
ethnicity, BMI, history of alcohol or drug abuse, narcotic
prescription for chronic pain, dysphagia, heart burn, ab-
dominal pain, gastroduodenal ulcer on EGD, patients
having a follow up with a physician, EGD performed on
in patients or out patients, presence or absence of intes-
tinal metaplasia on gastric biopsies, having government
insurance, self-pay status, language spoken (Spanish,
English, or bilingual) retired status, employed, and num-
ber of EGDs performed. The data were obtained from a
computerized database. Self-reported information was col-
lected on drug or alcohol abuse, ethnicity, languages
spoken, and employment status. A research assistant
went through each electronic medical chart to obtain
relevant data. For the last 119 patients seen from Sep-
tember 2012 to March 201 3, infor mation r egarding coun -
try of birth was prospectively identified besides other
information collected on all patients. The information
available was analyzed for all subjects and separately
analyzed also for subjects with information available on
the country of birth (Tables 1-5).
The data were analyzed using SPSS version 20. Uni-
variate as well as multivariate analyses were conducted
to identify significant predictors of positive H. pylori
infection. For insurance status, language status, and em-
ployment status, dummy variables were created with
commercial insurance, English only, and employed used
as the reference variables, respectively.
Table 1. Univariate analyses—dichotomous variables (general characteristics).
Percentage with this characteristic
Characteristic Controls Cases
p value*
Male 48% (135/282) 49% (125/256) 0.821
African-America n 13% (33/257) 10% (22/228) 0.336
Hispanic 48% (123/257) 72% (164/228) <0.0005
Spanish-Speaking 15% (41/279) 31% (75/245) <0.0005
Bilingual 7% (19/279) 7% (17/245) 1.000
Unemployed 68% (191/282) 70% (179/256) 0.714
Retired 19% (53/282) 16% (40/256) 0.336
Government Insurance 86% (234/273) 72% (164/228) <0.0005
Self-Pay 5% (13/273) 23% (54/236) <0.0005
*Continuity-correcte d chi-square test with 1 degree of freedom.
Copyright © 2013 SciRes. OJGas
S. Boddu et al. / Open Journal of Gastroenterology 3 (2013) 231-236 233
Table 2. (a) Univariate analyses—dichotomous variables (clinical characteristics); (b) Univariate analyses—continuous variables.
Percentage with this characteristic
Controls Cases p value*
Has Primary Care Provider 78% (159/204) 74% (150/204) 0.352
Inpatient 31% (64/205) 25% (52/210) 0.143
Hx of Alcohol Abuse 37% (37/273) 38% (93/245) 0.900
Smoker 48% (131/273) 40% (98/245) 0.098
Chronic Pain Meds 34% (93/276) 28% (69/247) 0.157
Dysphagia 13% (36/279) 11% (27 /25 3) 0.619
Abdominal Pain 46% (128/279) 72% (181/252) <0.0 005
Heart Burn 37% (102/278) 45% (113/253) 0.060
Intestinal Metaplasia 11% (30/278) 9% (22/249) 0.562
Gastric Duodenal Ulcer 2% (5/252) 14% (34/249) <0.0005
Characteristic Controls (N) Cases (N) p value**
Age 52.3 (282) 50. 5 (256) 0.516
Number of Endoscop ies 1.4 (282) 1.3 (255) 0.207
BMI 28.0 (258) 28.1 (239) 0.959
*Continuity-correcte d chi-square test with 1 degree of freedom; **Independent groups t-test.
Table 3. Multivariate logistic regression analysis.
95% CI for odds ratio
p value Odds ratio Lower Upper
Gastric Duodenal Ulcer <0.0005 8.140 3.100 21.372
Abdominal Pain <0.0005 3.377 2.143 5.322
Hispanic 0.001 2.343 1.399 3.925
Government Insurance 0.384 1.479 0.613 3.565
Self-Pay 0.002 6.236 2.001 19.433
Spanish-Speaker 0.709 0.888 0.475 1.660
All predictors with a p value of 0.05 or less wer e entered into a logistic regression model. A total of 409 cases were analyzed. Hosmer-Lemeshow test: p = 0.911;
Nagelkerke R-squar ed = 0.292.
The univariate analyses revealed that Hispanic race,
speaking Spanish only, reporting abdominal pain, having
either a duodenal or gastric ulcer, having government
insurance, and self-pay status were significantly related
to a positive finding for H. pylori. The multivariate ana-
lysis revealed that abdominal pain, presence if gastro-
duodenal ulcer, self-pay status and Hispanic race re-
mained significant when the impact of all of the predic-
tors together was assessed. For the last 119 subjects,
complete data on age, gender, Hispanic race, self-pay
status, having government insurance, speaking Spanish
only and foreign born status were available on 103 sub-
jects. Sub-analysis on this data in a multivariate regres-
sion model showed that foreign born status remained
significant Table 5.
The prevalence of H. pylori has been reported to be de-
creasing in industrialized nations such as the US be-
Copyright © 2013 SciRes. OJGas
S. Boddu et al. / Open Journal of Gastroenterology 3 (2013) 231-236
Table 4. Analysis of US born vs. foreign born variable (N = 118).
Control Case
Percentage foreign born Percentage foreign born
27% (20/75) 58% (24/43)
Table 5. Gender, age, hispanicity, self-pay, government insurance, Spanish speaker, and foreign born vs. US born were entered into a
logistic regression model. A total of 103 cases were available for this analysis.
95% CI. for EXP ( B)
Sig. Exp (B)
Lower Upper
Age 0.374 0.985 0.952 1.019
Male_Gender 0.300 0.610 0.240 1.554
Hisp 0.365 1.686 0.544 5.225
Selfpay 0.304 4.255 0.269 67.289
Govt 0.819 1.147 0.356 3.690
Spa 0.904 1.085 0.290 4.056
Foreign Born 0.008 4.513 1.481 13.757
Foreign Born vs. US Born remained significant, even after adjustment for the other variables. Hosmer-Lemeshow test: p = 0.585. Nagelkerke R-squared =
cause of better hygiene, a higher standard of living, and
use of antibiotics [7]. The Data reported in this study
show that patients undergoing endoscopy at a safety net
hospital who have a gastric duodenal ulcer are more than
eight times more likely to be H. pylori positive than
those who do not have an ulcer; and patients undergoing
endoscopy who report abdominal pain are more than
three times more likely to be H. pylori positive compared
to those who do not report abdominal pain.
Results of this data also showed significantly higher
odds of H. pylori positivity among patients who spoke
Spanish, were Hispanic in origin and had no insurance
(Table 3). On a separate analysis where place of birth
(foreign born versus US born) was included in addition
to race, language spoken and uninsured status, only place
of birth remained significant (Table 5 ). Taken together,
these data reveal H. pylori is more common in our hos-
pital among those Hispanics who are born in Mexico.
Nonetheless, many patients are infected with H. pylori
who are not migrants from a foreign country and the
source of H. pylori within that sub-set is not clear. It is
possible that within the family members of H. pylori
infected individuals, there is clustering of H. pylori in-
fection [4].
Lower socioeconomic status has been consistently
shown to be associated with H. pylori infection [8]. Ex-
posure to vomitus or feces or residence in an over-
crowded household is likely the cause of spread of H.
pylori infection [9]. Approximately 2 million people of
Hispanic or Latino originally live in Arizona; of them
524,480 people were born in Latina America [10,11].
The percentage of Hispanics without health coverage has
been reported to be 30.1%; higher than the overall rate
reported as 15.7%. The rates of non-insurance for White
non-Hispanics, Blacks and Asians have been reported to
be 11.1%, 19.5% and 16.8% respectively [12]. Lack of
insurance is also a surrogate marker of poverty and the
non-coverage rates range from 25.4% in households with
income less than $25,000 to 7.8% when household in-
come was $75,000 or more [13]. This information sug-
gests that not having insurance and Hispanicity are both
associated with poverty and might be related to factors
that predispose such individuals to H. pylori infection.
Since data were collected from a safety net hospital,
many undocumented subjects were included in our ana-
lyses although exact information on illegal individuals
was not available. Nonetheless, non-citizens (legal and
undocumented) are about three times more likely to be
uninsured than citizens [14].
Intestinal metaplasia, alcohol/drug abuse and use of
narcotics did not have any predictive value for H. pylori
infection in our database. Intestinal metaplasia, atrophic
gastritis and gastric cancer have been widely reported to
be associated with H. pylori infection. 2.9% percent of
patients with H. pylori infection can develop gastric
cancer [15]. Intestinal metaplasia associated with H. py-
lori can be seen in gastric body as well as antrum. In our
study, two biopsies were obtained on each patient; how-
Copyright © 2013 SciRes. OJGas
S. Boddu et al. / Open Journal of Gastroenterology 3 (2013) 231-236 235
ever, they were uniformly obtained from antrum and
gastric body. Because of the retrospective nature of our
study and lack of follow up endoscopic biopsies, we did
not see any difference in prevalence of H. pylori in cases
and controls in reported subjects of this study [16].
Conflicting data have been reported regarding asso-
ciation of H. pylori and alcohol consumption. Negative
effect on H. pylori has been reported with moderate al-
cohol consumption [17]. However, among patients with
functional dyspepsia, H. pylori has been shown to be
associated with alcohol consumption [18]. Data reported
in this database did not show any difference in alcohol
use between patients with and without H. pylori, al-
though specific amount of alcohol consumption is not
available. Similarly, data regarding H. pylori and smok-
ing are mixed. One study shows no significant relation-
ship between H. pylori serology and smoking [19], while
the other one reports a higher prevalence of H. pylori in
smokers compared to non-smoker [20]. Our data did not
show any relationship between smoking and H. pylori
(Table 2(a)).
Findings reported in this study are important. First,
data reported about predictors of H. pylori have never
been reported from Arizona which is bordering Mexico
and has one of the highest numbers of Hispanic
population who have either emigrated from Mexico or
born in US; several Hispanics live undocumented in
Arizona and seek medical care at our safety net hospital.
Second, this study shows that H. pylori remains a signi-
ficant cause of peptic ulcer disease that requires hospita-
lization straining an already over stretched health care
system. Finally and most importantly, the data reported
herein identify a sub-group of patients including those
who are uninsured or born outside US, who should be
properly screened with either H. pylori stool antigen test
or endoscopy, depending on an individual’s circums-
tances and treating them accordingly with appropriate
medications, particularly because H. pylori besides being
a significant cause of peptic ulcer disease has also been
classified as a class 1 carcinogen by WHO [6].
[1] Everhart, J.E., Kruszon-Moran, D., Perez-Perez, G.I.,
Tralka, T.S. and McQuillan, G. (2000) Seroprevalence
and ethnic differences in Helicobacter pylori infection
among adults in the United States. The Journal of Infec-
tious Diseases, 181, 1359-1363. doi:10.1086/315384
[2] US Census Bureau (2012) Resident population by his-
panic origin and state: 2010.
[3] Cardenas, V.M., Mena, K.D., Ortiz, M., et al. (2010) Hy-
perendemic H. pylori and tapeworm infections in a U.S.-
Mexico border population. Public Health Reports, 125,
[4] Tsai, C.J., Perry, S., Sanchez, L. and Parsonnet, J. (2005)
Helicobacter pylori infection in different generations of
Hispanics in the San Francisco Bay Area. American
Journal of Epidemiology, 162, 351-357.
[5] Sethi, A., Chaudhuri, M., Kelly, L. and Hopman, W.
(2013) Prevalence of Helicobacter pylori in a first nations
population in northwestern Ontario. Canadian Family
Physician, 59, e182-e187.
[6] Jones, N., Chiba, N., Fallone, C., et al. (2012) Helico-
bacter pylori in first nations and recent immigrant popu-
lations in Canada. Canadian Journal of Gastroenterology,
26, 97-103.
[7] Grad, Y.H., Lipsitch, M. and Aiello, A.E. (2012) Secular
trends in Helicobacter pylori seroprevalence in adults in
the United States: Evidence for sustained race/ethnic dis-
parities. American Journal of Epidemiology, 175, 54-59.
[8] Leclerc, H. (2006) Epidemiological aspects of Helico-
bacter pylori infection. Bulletin de lAcademie Nationale
de Medecine, 190, 949-962.
[9] Brown, L.M. (2000) Helicobacter pylori: Epidemiology
and routes of transmission. Epidemiologic Reviews, 22,
283-297. doi:10.1093/oxfordjournals.epirev.a018040
[10] Pew Hispanic Center (2010) Statistical portrait of the
foreign-born population in the United States.
[11] US Census Bureau (2010) Hispanics or latino population
for the United States, region, states, and for Puerto Rico.
[12] DeNavas-Walt, C., Proctor, B.D. and Smith, J.C. (2012)
Income, poverty, and health insurance coverage in the
United States: 2011. US Census Bureau, Current Popula-
tion Reports, P60-243, US Government Printing Office,
Washington DC.
[13] US Census Bureau (2011) Uninsured rate by real house-
hold income: 1999 to 2011.
[14] (2010) Statistical portrait of the foreign-born population
in the United States Pew Hispanic Center.
[15] Uemura, N., Okamoto, S., Yamamoto, S., et al. (2001)
Helicobacter pylori infection and the development of gas-
tric cancer. The New England Journal of Medicine, 345,
784-789. doi:10.1056/NEJMoa001999
[16] Kuipers, E.J., Uyterlinde, A.M., Pena, A.S., et al. (1995)
Long-term sequelae of Helicobacter pylori gastritis. Lan-
cet, 345, 1525-1528.
[17] Brenner, H., Bode, G., Adler, G., Hoffmeister, A., Koe-
nig, W. and Rothenbacher, D. (2001) Alcohol as a gastric
disinfectant? The complex relationship between alcohol
Copyright © 2013 SciRes. OJGas
S. Boddu et al. / Open Journal of Gastroenterology 3 (2013) 231-236
Copyright © 2013 SciRes.
consumption and current Helicobacter pylori infection.
Epidemiology, 12, 209-214.
[18] Zhang, L., Eslick, G.D., Xia, H.H., Wu, C., Phung, N.
and Talley, N.J. (2010) Relationship between alcohol con-
sumption and active Helicobacter pylori infection. Alco-
hol and Alcoholism, 45, 89-94. doi:10.1093/alcalc/agp068
[19] Hishida, A., Matsuo, K., Got o, Y., et al. (2010) Smoking
behavior and risk of Helicobacter pylori infection, gastric
atrophy and gastric cancer in Japanese. Asian Pacific
Journal of Cancer Prevention, 11, 669-673.
[20] Konturek, S.J., Bielanski, W., Plonka, M., et al. (2003)
Helicobacter pylori, non-steroidal anti-inflammato ry drugs
and smoking in risk pattern of gastroduodenal ulcers.
Scandinavian Journal of Gastroenterology, 38, 923-930.