Professional Documents
Culture Documents
Objective
Multiple studies have confirmed associations between acid suppression
medication and Clostridium difficile infections (CDIs) in adults.
Therefore, we sought to evaluate an association between acid
suppression medications and CDI in children.
Study design
A retrospective self-controlled case series was performed utilizing
billing records from the TRICARE Management Activity military health
system database. Children ages 2-18 years from October 1, 2001 to
July 31, 2013, who had an outpatient or inpatient record of CDI
diagnosis were included. The relative incidences (RIs) of CDI or
recurrent CDI were calculated comparing time periods prescribed and
not prescribed proton pump inhibitors (PPIs) or histamine-2 receptor
antagonists (H2RAs).
Results
There were 2531 cases of CDI among 2437 patients, and 1190 (48.8%)
were prescribed acid suppression medications. CDI were more likely to
occur during periods when patients were prescribed a PPI (RI 2.36; 95%
CI 2.22-2.52), H2RA (RI 1.95; 95% CI 1.63-2.34), or during periods while
prescribed both simultaneously (RI 2.40; 95% CI 1.90-3.04). There were
265 (10.4%) cases that were classified as recurrent among 217 (8.9%)
patients. Recurrent CDI also was found to be more likely during
prescription periods of PPI (RI 1.74; 95% CI 1.51-2.00) and H2RA (RI
Conclusions
Acid suppression medications are associated with an increased risk of
CDI and recurrent CDI. Judicious use of acid suppression medication
should be considered, especially among those at highest risk for CDI.
CDI (Clostridium difficile infection), H2RA (Histamine-2 receptor
antagonist), ICD-9-CM (International Classification of Diseases, Ninth
Revision, Clinical Modification), MHS (Military health system), PPI
(Proton pump inhibitor), RI (Relative incidence), SCCS (Self-controlled
case series)
Clostridium difficile is a gram-positive, spore forming, anaerobic
bacillus that can colonize the gastrointestinal lumen and lead to
Clostridium difficile infection (CDI). CDI can vary in its presentation
including diarrhea, pseudomembranous colitis, and toxic megacolon.
CDI is associated with an increased risk for both colectomy and death
in children.1 There have been several epidemiologic changes in CDI
over the last decades. These include increasing rates of CDI in
hospitalized children, increasing rates of community-acquired CDI, and
emergence of hypervirulent strains.1, 2, 3, 4, 5 These epidemiologic
changes of CDI may be due to other factors including changing host
factors such as exposure to medications.
Gastric acid suppression medications are used for a variety of
conditions in children including gastroesophageal reflux disease,
erosive esophagitis, gastric and duodenal ulcers, Helicobacter pylori
gastritis, and eosinophilic esophagitis.6, 7, 8 Indications for acid
suppression
without
evidence
of
esophagogastrointestinal
inflammation are expanding. These include optimization of
pancreolipase therapy, stress ulcer prophylaxis, treatment of
respiratory symptoms, and treatment of sleep disorders. 9, 10, 11, 12
Many clinicians feel that acid suppression medications also are
prescribed inappropriately as in cases of overmedicalized physiologic
infant reflux or functional gastrointestinal disorders.13
There are 3 main classes of acid suppression medications: neutralizing
antacids, histamine-2 receptor antagonists (H2RAs), and proton pump
inhibitors (PPIs). There has been a rapid increase over the last decade
in prescriptions for both PPIs and H2RAs. 14 For example, prescriptions
of PPI in children increased 3-fold from 2002-2009, with the majority of
these prescriptions in children aged 1-12 years old.15 Although acid
suppression medications generally are perceived to be relatively
Methods
A self-controlled case series (SCCS) method was chosen to study the
association of a diagnosis of CDI with the prescription of PPI or H2RA.
SCCS can be used to study associations between an acute event and
an exposure without requiring a control group. 22 This method
investigates the incidence of events within well-defined risk periods
relative to the incidence of events during control periods. The control
periods can include periods before or after the cases experienced
exposure. The major benefit of the SCCS method is that all time
independent known or unknown confounders are controlled for
implicitly. Applied to this study, the incidence of CDI during periods
when subjects were prescribed PPI or H2RA was compared with periods
when acid suppression medications were not prescribed.
Data were obtained from the TRICARE Management Activity military
health system (MHS) database covering the period of October 1, 2002
to July 31, 2013. The TRICARE Management Activity oversees health
care delivery for US uniformed services members and their families in
the US and abroad. The MHS database includes all outpatient and
inpatient billing records, and outpatient pharmacy utilization, for all
eligible military dependents in both military and civilian facilities.
TRICARE provides medication coverage for either no cost/copay at
military pharmacies or for very low copays at civilian pharmacies; this
includes coverage for both PPIs and H2RAs. Data were selected for all
children ages 2-18 years who were dependent beneficiaries of US
uniformed service members. Children less than 2 years old were
excluded because of the uncertainty about the true morbidity of CDI in
young children and infants.23
Variable Definition
Case selection was performed by searching the MHS database for the
International Classification of Diseases, Ninth Revision, Clinical
Modification (ICD-9-CM) diagnostic code 008.45. This is the sole
diagnostic code dedicated to CDI. All outpatient and inpatient billing
records for both military and civilian medical facilities were
investigated over the time period. Visits or hospitalization with a code
for CDI must have been greater than 60 days apart to be counted as
unique. The typical treatment course for CDI is 10 days, and repeat
testing for recurrent CDI is not recommended prior to 4 weeks after
initial testing.24 We selected this 60-day time period between events
as a conservative measure to distinguish independent cases of CDI.
Outpatient pharmacy records for the identified cases were queried
utilizing American Hospital Formulary Service therapeutic class codes
for PPI and H2RA. Data for the prescriptions included the number of
days supplied. Cases were considered within the exposed medication
time frame if they occurred during periods after a prescription was
billed by a pharmacy (through the number of days supplied). For cases
with multiple prescriptions, the exposure time periods were combined
if the next prescription was filled within 14 days of the anticipated
exhausted supply of the previous prescription, with the assumption of
continued use. This grace period was to account for partial
noncompliance and accumulation of medication. The exposure time
periods were divided into 3 groups: PPI, H2RA, or both. Dose or patient
weight information was not available. Inpatient pharmacy records were
not available. If an inpatient CDI case occurred during an exposure
period based on outpatient pharmacy records, it was considered to be
within the at-risk exposure period. In addition, if an inpatient CDI case
occurred during a nonexposure period based on outpatient pharmacy
records, the case was assumed to not be exposed to antacid
medications. Any subject within the greater SCCS who had 2 or more
cases of CDI 60 days or greater apart was considered to have recurrent
CDI. Only the CDI cases after the subject's first case were utilized for
analysis for recurrent CDI.
Statistical Analyses
Univariate analyses were conducted to investigate normality of the
data. We summarized normally distributed data with mean and SD and
skewed data with median and IQR. The 2 test was used to test
differences of categorical variables. The Kruskal-Wallis Test was used
to compare age across groups and the Mantel-Haenszel 2 test was
used to compare sex distribution across risk groups. Relative incidence
(RI) was calculated utilizing a conditional Poisson regression. The
Results
There were a total of 2531 cases of CDI identified over the study period
among 2437 patients. These patients represented a total of 16971
person-years in the study. Of CDI cases, 1708 (59.5%) were outpatient.
The median (IQR) age at time of CDI was 8 years (4-15) and 52.1% of
cases were male. Overall, there was a significant difference in age
across the acid suppression risk groups (Table I). Comorbid conditions
previously shown to be associated with CDI were associated with
higher rates of prescribed acid suppression (Table II). There were 265
(10.4%) cases of CDI that were classified as recurrent among 217
(8.9%) patients. Among those who had multiple events, the median
(IQR) time between cases was 127 days (79-211). The recurrent CDI
patients represented a total of 136 person-years. Among those with
recurrent CDI, the median number of recurrent cases of CDI per patient
was 1 (range 1-8).
Table I
Characteristics of children in the CDI case series by exposure group
Cases
Nonmedicat
ed
PPI
s
H2
RA
s
2087
308
119
Simultaneous
PPIs and
H2RAs
Total
17
2531
Person-years
15743
864
318
46
16971
Incidence density
per 1000 personyears
132.6
365.
5
374.
2
369.6
149.1
Median age at
diagnosis (IQR)
years
8 (4-15)
10
(415)
8
(314)
8 (6-13)
8 (4-15)
Female (%)
1028 (49.6)
136
(44.2
)
49
(41.
2)
7 (41.2)
1220
(48.2)
test, P = .19.
Differences of sex across risk groups was compared by MantelHaenszel 2, P = .02.
Table II
Comorbid diagnoses and acid suppression among the CDI case series
No acid
suppression
,
N = 1247
Acid
suppression
,
N = 1190
P
value
Transplantation/complications of
graft
750
825
<.001
117
176
<.001
Lymphoma/leukemia
69
233
<.001
Rheumatoid arthritis/systemic
Lupus erythematosus
29
69
<.001
Cystic fibrosis
35
<.001
HIV
12
<.001
Of the 2437 individuals who were in the case series, 1190 (48.8%)
were prescribed acid suppression medications. There were 10488 total
prescriptions of PPI administered to 813 individuals, representing 910
person-years while on PPI (including overlap with H2RA). The mean
supply for each individual PPI prescription was 36 days (SD 21). After
linking the successive PPI prescriptions into combined periods, the
median (IQR) day's supply of a prescription was 45 days (30-94). The
median (IQR) age at time of prescription was 9.6 years (4.7-14.4) and
prescriptions were evenly distributed between males and females
(5261 [50.1%] male). The prescribed frequencies of individual PPI
medications are presented in Table III (available at www.jpeds.com).
There were fewer H2RAs prescribed to patients in the CDI case series,
with 4117 total prescriptions administered to 776 individuals
representing 364 person-years (including overlap with PPI). The mean
prescription supply for H2RAs was 34 days (SD 17). After consolidating
multiple prescriptions, the median (IQR) H2RA prescription period was
30 days (20-71 days). The median (IQR) age at time of prescription of a
H2RA was 7.3 years (3.1-12.6). Among the CDI case series, males were
more commonly prescribed H2RA (2390 [58.1%] male and 1727
[41.9%] female [P < .001]). The frequencies of H2RA prescribed are
presented in Table IV (available at www.jpeds.com). Among those
prescribed acid suppression medications, there were 191 individuals to
whom both PPIs and H2RAs were simultaneously prescribed. This
represented 45.0 person-years. The median (IQR) linked prescription
periods during which patients were prescribed both a PPI and H2RA
was 30 days (15-54).
There were 2087 (82.5%) cases of CDI during a period not receiving
PPIs or H2RAs therapy, with an incidence density of 132.6 per thousand
person-years. There were 308 (12.2%) cases of CDI during a period
while on a PPI, with a rate of 365.5 per thousand person-years. There
were 119 (4.7%) cases of CDI during a period while subjects were
taking an H2RA, with an incidence density of 374.2 per thousand
person-years. Lastly, there were 17 (0.6%) cases of CDI during periods
when subjects were prescribed both an H2RA and PPI; the incidence
density for this group was 369.6 per thousand person-years. PPI, H2RA,
and combination therapy all demonstrated a significant increase in risk
for CDI. The unadjusted and adjusted RI for CDI during periods on PPI,
H2RA, or both medications, as well as sex, age, and calendar year are
summarized in Table IV. In the sensitivity analysis, similar RIs for CDI
were seen using 90, 120, and 180 days between CDI events defined as
independent (data not shown).
There were 187 (68.2%) cases of recurrent CDI during a period while
not taking PPI or H2RA. The incidence density was 166.6 per thousand
Table V
RI of acid suppression medications for CDI and recurrent CDI
CDI
Recurrent CDI
Unadjusted
Adjusted
Unadjusted
Adjusted
PPI
2.58 (2.422.75)
2.36 (2.222.52)
1.81 (1.572.10)
1.74 (1.512.00)
H2RA
2.59 (2.312.90)
1.95 (1.632.34)
1.52 (1.142.03)
2.63 (1.893.66)
2.49 (1.973.15)
2.40 (1.903.04)
1.17 (0.672.08)
1.12 (0.622.05)
Male vs female
1.04 (1.011.07)
1.00 (0.971.03)
1.14 (1.031.26)
1.05 (0.951.15)
Age
1.00 (0.981.01)
0.99 (0.980.99)
1.01 (1.001.03)
1.00 (0.991.01)
Calendar year
1.12 (1.111.13)
1.11 (1.101.12)
1.19 (1.161.22)
1.19 (1.161.21)
Figure
Trend of CDI and recurrent CDI. Combined outpatient and inpatient
cases of CDI. A diagnosis of CDI must have been greater than 60 days
from the previous event to be counted. P value for trend was obtained
from conditional Poisson regression. For purposes of the graph, cases
are presented only for years in which the entire year of data was
available.
Discussion
Our findings indicate that acid suppression medication, both PPIs and
H2RAs, are risk factors for the development of CDI and recurrent CDI.
We demonstrate that the combination of PPI and H2RA has
approximately the same degree of risk than either medication alone.
There is 1 published pediatric study that previously found an
association between PPI and CDI; Turco et al performed a case-control
study with 68 children with CDI and reported an increase use of PPI in
the CDI group vs controls, OR of 4.5; 95% CI 1.4-14.4. 26 Our large case
series study of 2437 subjects confirms these findings.
Although not the focus of our study, we confirmed an increasing trend
of CDI in a mixed inpatient and outpatient pediatric case series. 1, 27 Of
note within the case series, there was a significantly increasing
proportion of time subjects were prescribed PPIs in later years. A rising
trend of PPI prescription also is well documented in the pediatric
population as a whole.14, 15 Our finding suggests that the rising trend
of CDI may be at least in part due to a rising trend of PPI prescriptions
in children.
The association of acid suppression medication with CDI is biologically
plausible with several possible explanations. Hypochlorhydria likely
contributes to the identified increased risk. The primary form of C
difficile excreted in the stool is the vegetative form which, if ingested,
is normally destroyed in stomach acid. In addition, hyochlorhydria
likely establishes an environment more favorable to C difficile spore
germination in the small bowel. 28 Besides affecting the gastric acid
barrier, antacids may also affect immune function. PPIs have been
shown to suppress immune surveillance, immune cell migration,
cytokine production, lysosomal enzyme function, and neutrophil
function.29, 30, 31, 32, 33 H2RA may also affect the intestinal immune
system by decreasing cell proliferation, inhibiting inflammationmediated nitric oxide concentrations, detrimentally affecting colonic
healing, and altering cytokine production.34, 35 In addition, the balance
of endogenous bacteria of the colon provides protection from enteric
infection, and gastric acid reduction has been shown to alter the lower
intestinal flora.36, 37 Alteration in colonic bacteria likely contributes to
the increased risk of CDI. This is further evidenced by the effectiveness
of microflora restorative therapies such fecal bacteriotherapy and
probiotics in the management of CDI.38
Our findings have several clinical applications. Providers should
prescribe acid suppression medication judiciously and prescribe the
minimum duration of medication required to either treat the necessary
condition or control symptoms. In the context of treating CDI, it may be
prudent to discontinue acid suppression medication to optimize
response to therapy, although this has yet to be studied. To prevent
recurrent CDI, subsequent prescription of acid suppression could again
be minimized and if possible delayed, balancing appropriate treatment
of clinical symptoms with the increased risk of recurrent CDI. Although
not specifically evaluated in this study, it is possible that minimizing
the dose of acid suppression medication could be beneficial in
preventing CDI or CDI recurrence.
The strengths of our study include a large, representative,
geographically, and demographically diverse pediatric population
across a variety of both military and civilian facilities. TRICARE is a
universal health care system minimizing bias because of access to
care. The SCCS method allowed for implicit controlling for known and
unknown patient level comorbid conditions that would place patients at
risk for CDI. There are several limitations of our study. We were not
able to obtain records of inpatient administration of acid suppression
medications. Also, despite no or low medication co-pays, it is possible
that patients obtained over-the-counter PPIs or H2RAs. Both the lack of
inpatient records or unaccounted over-the-counter sources would likely
lead to an underestimation of the association of PPI and H2RA with CDI.
Poor compliance with the prescribed medications could lead to
overestimation of the effect of PPI or H2RA. Our study was dependent
on ICD-9-CM diagnostic codes for our definition of CDI, which could
lead to misclassification bias. The ICD-9-CM code for CDI, however, has
been validated previously in pediatric billing records. 39 We did not
evaluate additional medications that likely also contribute to CDI, such
as antibiotics and immunosuppressive medications. We found that acid
Appendix.
Table III
Frequencies of PPI prescriptions in the CDI case series
PPI
Frequency (n = 10
488)
Lansoprazole
5191 (49.5)
Omeprazole
3220 (30.7)
Esomeprazole
1502 (14.3)
Pantoprazole
340 (3.3)
Rabeprazole
175 (1.7)
Omeprazole/sodium
bicarbonate
46 (0.4)
Dexlansoprazole
14 (0.1)
Table IV
Frequencies of H2RA prescriptions in the CDI case series
H2RA
Frequency
(n = 4117)
Ranitidine
3349 (81.3)
Famotidine
612 (14.9)
Nizatidine
103 (2.5)
Cimetidine
53 (1.3)
References
1. Nylund, C.M., Goudie, A., Garza, J.M., Fairbrother, G., and Cohen,
2.
3.
4.
5.
KESIMPULAN DISKUSI
Priskila Marlen Yoltuwu