Treatment with histamine-type 2 receptor antagonists and omeprazole increase the risk of diarrhoea in neonatal foals treated in intensive care units
Reasons for performing study: The use of anti-ulcer medication in the neonatal intensive care unit (ICU) is common due to the concern for development of catastrophic gastric ulcer disease. In man, however, the use of acid-suppressive medication has been shown in some studies to be a substantial risk factor for the development of Clostridium difficile-associated diarrhoea (CDAD), bacteraemia and neonatal sepsis.
Objective: The purpose of the study reported herein is to evaluate the influence of anti-ulcer medications on the development of diarrhoea in the neonatal foal.
Hypothesis: The use of anti-ulcer medication does not alter the incidence of diarrhoea in foals treated in an ICU.
Methods: The records of 1710 foals from 6 different equine hospitals were examined and the use of anti-ulcer drugs was recorded. The presence of in-hospital acquired diarrhoea, CDAD, Clostridium perfringens-associated diarrhoea, neonatal sepsis and salmonellosis were documented. In addition, the presence of gastric ulceration, duration of hospital stay and short-term outcome were examined.
Results: The use of anti-ulcer medications increased the odds of in-hospital diarrhoea by 2.0 (95% CI 1.4–2.9; P<0.0001), relative to the use of no anti-ulcer medication. There was no significant association of anti-ulcer medication with CDAD (P = 0.3189) (OR 2.0; 95% CI 0.4–9.5). Further, results indicated that decreased prevalence of gastric ulceration was not associated with use of anti-ulcer drugs among foals in the study for which these data were known (P = 0.5522).
Conclusions: Use of anti-ulcer drugs increases the odds of developing diarrhoea, and may not reduce the incidence of gastric ulceration in hospitalised equine neonates.
Potential relevance: The use of anti-ulcer drugs in neonatal foals being treated in a hospital setting should be carefully evaluated on an individual basis to determine if such use is warranted.
Gastric ulceration of neonatal foals is common and can have serious consequences. In one study the incidence of gastric ulcers in foals with an average age of 26 days was 51% (37/72), with the squamous lining of the stomach the most frequent site affected (Murray et al. 1987). In another endoscopy-based study, the incidence of glandular mucosal lesions was 9% of foals aged <33 days, suggesting an association between clinical illness and gastric glandular mucosal ulceration (Murrary et al. 1990). This was further supported by the finding that gastric glandular mucosal lesions were noted in 40% of foals affected by concurrent illness and hospitalisation (Furr et al. 1992) compared with a 3% occurrence in previous reports (Murray 1989). In a post mortem study of hospitalised, critically ill foals aged <30 days, 12.9% (11/85) had a gastric ulcer; however, the lesion site was not described (Barr et al. 2000). Gastric ulceration has been reported to cause death due to perforation and subsequent peritonitis, or due to exsanguination (Rebhun et al. 1982; Traub-Dagartz et al. 1985; Becht and Byers 1986).
The pathophysiology of gastric ulceration is considered to include an imbalance of aggressive and protective factors, with gastric acid being considered the most ulcerogenic. Gastric pH measurements as low as 1.85 have been described in horses as young as age 2 days (Murray and Grodinsky 1989). In another study, mean pH over an 8 h period was 4.1 for foals aged one day, and 3.4 for foals aged 2 days; by day 7 of life, the average 8 h pH was 2.0 (Baker and Gerring 1993). Values as low as pH 1 were occasionally observed in normal foals during the first day of life (Baker and Gerring 1993). Additional factors involved in gastric ulcer development in critically ill neonates may also include alteration in gastrointestinal blood flow, ileus and hypoxia, as have been demonstrated in man (Stollman and Metz 2005). The use of nonsteroidal anti-inflammatory drugs may also impact gastric mucosal blood flow and could have a role. Feeding frequency and volume may also have an impact upon the development of gastric ulceration in hospitalised foals.
Due to the reported high incidence of gastric lesions in neonatal foals, the concern for potential fatal consequences and the demonstration of an acid gastric environment in very young foals, the treatment of neonatal foals in an intensive care unit (ICU) often includes the use of acid-suppressive medication, consistent with recommendations for human patients in a recent concordance statement (Dellinger et al. 2008). It has been questioned, however, whether this practice is warranted or even effective. It has been demonstrated, for example, that critically ill neonatal foals often have intragastric pH profiles that are substantially different from normal or less severely affected foals, with a continuously alkaline intragastric pH demonstrated in roughly 50% of neonates, thus making acid-suppressive medication ineffective in many foals (Sanchez et al. 2001).
Additional concerns about the indiscriminant use of acid-suppressive medications in the neonatal ICU include the observation that decreased gastric acidity is known to increase the risk for development of cholera, salmonellosis and travellers' diarrhoea in man (Williams 2001). In addition, survival of Clostridium organisms is facilitated by an increased pH (Gurion et al. 1982). Treatment with proton pump inhibitors (PPIs) has been associated with colonisation of the upper gastrointestinal tract, which is usually sterile (Thorens et al. 1996) and the use of PPIs in human adults and infants has been shown to increase the incidence of Clostridium difficile-associated diarrhoea (CDAD) (Dial et al. 2004). The risk for development of community-acquired pneumonia was increased in patients being treated with acid-suppressive medications (Laheji et al. 2004), and the risk for bacterial sepsis in human neonates was increased with the use of either PPIs or histamine type 2 receptor antagonists (H2RAs) (Graham et al. 2006).
It is unclear whether a similar relationship exists in equine neonates treated in a hospital setting. The following study was performed to determine whether the use of anti-ulcer medication results in increased odds of undifferentiated diarrhoea or diarrhoea caused by C. difficile, Clostridium perfringens or Salmonella spp., or increases the odds of neonatal sepsis and the systemic inflammatory response syndrome (SIRS). The hypothesis tested was that use of such drugs does not alter the incidence of these conditions.
Materials and methods
Records of foals aged ≤14 days admitted to an equine ICU were examined retrospectively and data were entered into a spreadsheet. Duration of data collection varied by centre, ranging from 1–10 years (median 3 years). Data collected included the foal age at presentation (in days), gender, breed, presenting clinical complaint, hospital use of anti-ulcer medications and type, duration of use of the medication, presence of diarrhoea during hospitalisation, short-term outcome (i.e. survival to discharge), and sepsis score at presentation. Foals that demonstrated diarrhoea prior to treatment with anti-ulcer medications (typically observed at admission) were excluded from the analysis. Diarrhoea was defined as an increased volume, frequency or fluidity of faeces as assessed by the treating clinician. Diagnostic testing was at the investigators' discretion, and if a test was not performed, then no result was included in the data set. The diagnostic method for confirmation of C. difficile or C. perfringens diarrhoea was recorded. Salmonellosis was defined as culture of any salmonella organism from faeces or blood.
Following data entry, the data were analysed using commercial statistical software (S-Plus, version 8.2)a using a predetermined significance value of P<0.05. Data were analysed using descriptive and inferential methods; not all data were available for all foals. For descriptive purposes, continuous data were summarised using medians and ranges, and categorical data as proportions or using contingency tables. For inferential analysis, the primary clinical endpoint was the dichotomous outcome of development of undifferentiated diarrhoea during hospitalisation (yes or no); secondary endpoints were specific aetiological diagnoses for diarrhoea (e.g. clostridial diarrhoea) or sepsis/SIRS. The exposure of primary interest was use of anti-ulcer medication during hospitalisation; exposures of secondary interest included specific anti-ulcer agents used (i.e. omeprazole, H2RAs or sucralfate). Because the frequency of both the primary outcome (diarrhoea) and primary exposure varied significantly among sites (see Results), data were analysed using random-effects logistic regression (RELR) for the dichotomous outcome of diarrhoea (or specific type of diarrhoea or sepsis). Each hospital site was modelled as a random effect, while other exposure variables were modelled as fixed effects. Some continuous data were recoded as categorical data for analysis, using the median of the value for the population of foals that developed diarrhoea. Bivariate RELR analysis was performed for associations of the various outcomes of interest with individual exposure variables (e.g. age, breed, gender, acid-suppressive medications etc.). All variables associated with diarrhoea (or other outcomes) at a level of P<0.15 were included in purposeful multivariable RELR, including bivariate or trivariate interaction terms. Associations were expressed as the odds ratios (ORs), obtained by exponentiating coefficients of the RELR models; 95% confidence intervals (95% CIs) for ORs were obtained using maximum likelihood methods, with standard errors reflecting the correlation among observations obtained from the same study site. In some instances (e.g. to describe centre effects), standard logistic regression analysis was performed.
The records of 1710 foals from 6 different equine neonatal ICUs were reviewed for inclusion in the study. Participating hospitals included 3 university neonatal ICUs and 3 private practices; 4 from North America, one from the UK, and one from Australia. Of these foals, data regarding whether or not the foal had developed diarrhoea during hospitalisation were missing for 216 foals. Of the remaining 1494 foals, 392 admitted with a chief complaint of diarrhoea were excluded, such that the study population comprised 1102 foals admitted to the participating centres for problems other than diarrhoea and for which presence or absence of diarrhoea during hospitalisation was recorded. Summary information regarding the study population is presented in Table 1.
Table 1. Clinical summary of 1102 foals treated at 6 different equine hospitals and examined for the relationship of diarrhoea and use of anti-ulcer medication
| N || ||102||68||426||101||371||34||1102|
| Age (days) ||Median||1.0||1.0||1.0||1.3||0.5||0.3|| |
|Interquartile range||1.0–2.0||0.2–5.0||0.5–3.0||0.6–3.0||0.1–1.0||0.02–1.0|| |
| Anti-ulcer usage ||N||47a||58a||252a||89a||23a||33a||502|
| Occurrence of diarrhoea ||N||25a||10||153a||36||28a||6||258|
| Outcome (survived) ||%||74.5||86.0||79.9||86.8||87.2||83.3||81.2|
| Sepsis score ||Median||10.0||6.0||9.0||11.0||8.0||7.5|| |
|Interquartile range||5.0–13.0||3.0–8.7||5.5–14.0||6.5–15.0||4.0–12.0||4.2–12.0|| |
The number of foals contributed by the participating centres varied from 34–426, with the median being 102 foals. The occurrence of diarrhoea during hospitalisation varied significantly among centres, ranging from 8% (28/371) to 36% (153/426 and 36/101) (Table 1). Using standard logistic regression, the cumulative incidence of diarrhoea relative to that at Hospital A was significantly (P<0.0001) lower at Hospital E and significantly greater at Hospital C (P = 0.0302). There was no difference in cumulative incidence of diarrhoea relative to Hospital A at Hospital B (P = 0.1263), and Hospital D (P = 0.0857). Use of anti-ulcer medication during hospitalisation also varied significantly among centres (Table 1). Use of anti-ulcer medication during hospitalisation was reported for 1091 foals. Frequency of administration varied from 6–97% (33/34) of hospitalised foals. Using standard logistic regression, the use of anti-ulcer medication was significantly greater at Hospitals B, C, D and F (P<0.0001, P = 0.0078, P<0.0001, and P = 0.0004, respectively) relative to Hospital A, and significantly lower at Hospital E (P<0.0001); use at Hospital E was significantly (P<0.0001) lower than all other centres. These data supported the use of RELR (or other methods to account for centre effects) for data analysis.
Factors associated with diarrhoea
Of the 1102 study foals, 258 (23.4%) developed diarrhoea during hospitalisation. Foals that developed diarrhoea were significantly (P<0.0001) older than foals that did not. The median age at presentation of foals that developed diarrhoea was 1.5 days compared with 0.75 days for foals that did not. Categorising age as ≥1.5 days or <1.5 days, the odds of foals developing diarrhoea were 2.2-fold greater (Table 2) among foals ≥1.5 days than those <1.5 days.
Table 2. Factors significantly associated with development of diarrhoea during hospitalisation among 1102 foals admitted to 6 equine hospitals for problems other than diarrhoea
|Age at admission|| || || || |
| <1.5 days||126(49%)||606(72%)||1||NA|
| ≥1.5 days||132(51%)||237(28%)||2.2(1.6–2.9)||<0.0001|
|Chief complaint|| || || || |
|Acid-suppressive medication during hospitalisation|| || || || |
|Anti-ulcer medication during hospitalisation|| || || || |
|Hospitalisation >7 days|| || || || |
|Outcome|| || || || |
| Did not survive||47(18%)||160(19%)||0.9(0.6–1.3)||0.4970|
Almost 50 breeds of foals were represented, and breed was reported for 1093 foals. For analysis, breed was recoded as Thoroughbred (n = 771; 70.5%), Quarter Horse (QH)-type (which included QH, Appaloosa, Paint Horse and QH-crossbred; n = 121; 11.0%), European Warmblood (n = 64; 5.8%), or other breeds (n = 137; 12.5%). The proportion of foals with diarrhoea did not differ among breeds, and was 22.9% (177/771) for Thoroughbreds, 25.6% (31/121) for QH-type horses, 20.3% (13/64) for Warmblood horses and 24.8% (34/137) for other breeds. Gender was reported for 1084 foals, and was not significantly (P = 0.2883) associated with diarrhoea. The proportion of males was 52.9% (135/255) and 57.6% (478/829) among foals with and without diarrhoea, respectively.
Chief complaint at admission
Over 60 unique primary complaints were specified for 1091 foals. For purposes of analysis, these clinical complaints were collapsed into categories of ‘sick’, ‘colic’ or ‘other’. The most frequently reported complaint was foals that were classified as ‘sick’ (including foals that were described as weak, not nursing, recumbent, dull, ‘not right’ or unthrifty) (n = 613; 56.2%). Colic was reported as the chief complaint in 143 foals (13.1%). Remaining foals were classified as ‘other’ (n = 335; 30.7%), which included foals born following caesarean section or dystocia in the mare (n = 65), neurological disease (n = 61), musculoskeletal disorders (n = 40), respiratory disease (n = 32), or other complaints (n = 137), which included trauma, wounds, congenital defects, ocular injuries, twins, rejected foals etc. Chief complaint at presentation was significantly (P<0.0001) associated with the development of diarrhoea (see Table 2).
Use of anti-ulcer medication
Data regarding use of anti-ulcer medications were reported for 1091 foals, of which 502 (46.0%) received anti-ulcer medication during hospitalisation. Use of anti-ulcer medication during hospitalisation was significantly (P<0.0001) associated with increased odds of development of diarrhoea (Table 2). Among the 502 foals that received anti-ulcer medication, the number of types of medications was recorded for 501; most (89.2%; 447/501) received only one drug, and the remainder (10.8%; 54/501) received 2. The proportion of foals that developed diarrhoea was similar among foals that received one (34.4%; 154/447) or 2 medications (33.3%; 18/54). The most common combination of medications was omeprazole with sucralfate (n = 24 foals) followed by the combination of ranitidine with either sucralfate (n = 8 foals) or omeprazole (n = 8 foals). The first (primary) acid-suppressive medication was recorded for 501 of the 502 foals receiving these drugs. The most common medication used alone was omeprazole (n = 216 foals), followed by ranitidine (n = 204 foals), cimetidine (n = 63 foals) and sucralfate (n = 18 foals). Indicator variables were created for each anti-ulcer drug, and also for the combination of ranitidine and cimetidine as H2RAs. Each of the drugs was significantly associated with diarrhoea (Table 2), and the odds of diarrhoea did not differ significantly among the individual medications. Of the 1091 foals given anti-ulcer medication, 483 received acid-suppressive drugs only (e.g. omeprazole or H2RAs). The use of these acid-suppressive drugs during hospitalisation was significantly (P = 0.0003) associated with increased odds of developing diarrhoea (Table 2).
The number of days after treatment with acid-suppressive medication was initiated when the diarrhoea began was reported for only 127 foals (11.5%), with a median of one and range 0–24 days.
Duration of hospitalisation and outcome
The number of days of hospitalisation was recorded for only 510 foals (46.2%). The duration of hospitalisation was significantly longer for foals with diarrhoea (median 6 days; range 0–67 days; n = 155 foals) compared with those without diarrhoea (median 4 days; range 0–59 days; n = 355 foals). For purposes of analysis, hospitalisation >7 days was considered as a dichotomous variable (<7 days, or ≥7 days), and hospitalisation for a week or more was significantly (P = 0.0007) associated with diarrhoea (Table 2).
Outcome was reported for 1099 foals, of which 892 foals survived to discharge (81.2%), 147 foals (13.4%) were subjected to euthanasia, 11 specifically because of financial constraints, and 49 foals (4.4%) died. For purposes of analysis, all foals subjected to euthanasia were considered as a single group. Considering outcome as either survived to discharge or did not (i.e. died or subjected to euthanasia), there was no significant association of diarrhoea with outcome.
Because both age and duration of hospitalisation were significantly associated with diarrhoea, multivariable modelling was performed to adjust for the effects of these potential confounders on the association of diarrhoea during hospitalisation and administration of acid-suppressive medication (H2RAs and omeprazole). Three different multivariable models fitted because the duration of hospitalisation was only reported for 510 (46.2%) of the foals included in the study; data regarding acid-suppressive medication administration were available for 499 foals. In the first multivariable model, administration of acid-suppressive medication remained significantly associated with diarrhoea even after adjusting for effects of age at admission (considered as a dichotomous categorical variable; Table 3); there was no significant interaction of effects of age category and acid-suppressive medication. The estimated ORs were similar to those observed for bivariate analyses (Table 2), indicating that age was not a confounding variable for the association of acid-suppressive medication administration with development of diarrhoea. Duration of hospitalisation was considered as a dichotomous categorical variable, and remained significantly associated with diarrhoea after adjusting for effects of acid-suppressive administration (Table 3); there was no significant interaction between duration of hospitalisation and acid-suppressive medication administration. The ORs were consistent with those seen for each variable using bivariate analysis (Table 2). Finally, a model was fitted to the data from the 499 foals for which there were data available regarding use of acid-suppressive medication and duration of hospitalisation (Table 3c). Results were similar to those for the model in Table 3a and b, indicating that age was not an important confounder for the other variables (use of acid-suppressive medications and duration of hospitalisation); the trivariate and all bivariate interaction combinations were not significant for this latter model. Neither age nor hospitalisation (data not shown) altered the magnitude of the OR by >20% or the statistical significance of associations with individual anti-ulcer medications (omeprazole, H2RAs or sucralfate).
Table 3. Results of multivariable random-effects logistic regression model examining the association of acid-suppressive medication with diarrhoea among foals, adjusted for effects of age, duration of hospitalisation, or both
|Age and acid-suppressive medication (n = 1090)|| || || |
|Acid-suppressive medication|| || || |
|Age at admission|| || || |
| <1.5 days||1||NA||NA|
| ≥1.5 days||2.1||1.6–2.9||<0.0001|
|Duration of hospitalisation and acid-suppressive medication (n = 499 foals)|
|Acid-suppressive medication|| || || |
|Hospitalisation ≥7 days|| || || |
|Age, duration of hospitalisation and acid-suppressive medication (n = 499 foals)|
|Acid-suppressive medication|| || || |
|Age at admission|| || || |
| <1.5 days||1||NA||NA|
| ≥1.5 days||2.4||1.6–3.5||<0.0001|
|Hospitalisation ≥7 days|| || || |
Presence or absence of gastric ulceration was recorded for 166 foals, and method of diagnosis for 103 foals. Gastric ulceration was detected in the majority of foals by necropsy examination (95.9%; 95/99), with the 4 remaining foals detected by endoscopic examination. The proportion of foals with gastric ulceration detected did not differ significantly (P = 0.5522) between foals that were treated with anti-ulcer medications (15.4%; 19/123) or not treated (20.9%; 9/43).
Association of anti-ulcer medication with other clinical endpoints
Of the foals that developed diarrhoea, 78.6% (203/258) were tested for either C. difficile or C. perfringens and of these, only 101/203 (49.7%) foals had a confirmed diagnosis of diarrhoea due to a Clostridium species. The method of diagnosis was reported for only 70 foals, and was most frequently based on an enzme-linked immunosorbent assay ELISA for clostridial exotoxin or other immunoassay (85.7%; 60/70). Of 95 foals with diarrhoea tested for C. perfringens, a positive diagnosis was found in 19 (20.0%). The use of anti-ulcer medication among foals reported to have C. perfringens diarrhoea was 57.8% (11/19) and was 73.6% (56/76) among foals not having C. perfringens diarrhoea; the difference was not significant (P = 0.3105). Of 108 foals with diarrhoea tested for C. difficile, 82 (75.9%) were affected. Use of anti-ulcer medication was 13.4% (11/82) among foals reported to have C. difficile diarrhoea and 7.6% (2/26) among foals not having C. difficile diarrhoea. The OR for the association of anti-ulcer medication with C. difficile diarrhoea was 2.0 (95% CI 0.4–9.5; P = 0.3189).
Whether or not foals were diagnosed with salmonellosis was recorded for 565 foals, of which 13 (2.3%) were reported as being positive. Positive cases were distributed among 3 study sites, none of which had a recognised outbreak of salmonellosis during the study period. Because of the small number of cases of salmonellosis, and marked differences in reporting patterns among sites, results of RELR modelling the association of salmonellosis with the use of anti-ulcer therapy were unreliable and fit the data poorly. Therefore, these data were analysed ignoring centre effects, using both Chi-squared analysis and standard logistic regression. The proportion of foals diagnosed with salmonellosis that received anti-ulcer medication (76.9%; 10/13) was significantly (P = 0.0012) greater than that among foals not diagnosed with salmonellosis (30.6%; 169/552); the odds of salmonellosis were 7.6-fold (95% CI 2.1–27.6) greater among foals treated with anti-ulcer medication than among foals not treated with anti-ulcer medication.
The sepsis score (Brewer and Koterba 1988) was reported for 708 foals (155 foals with diarrhoea and 553 without). Using RELR, the sepsis scores of diarrhoeic foals (median 10; range 0–24) were significantly (P = 0.0340) greater than those of nondiarrhoeic foals (median 8; range 0–30). The odds of having a sepsis score ≥11 were not significantly (P = 0.2588) greater for foals that were diarrhoeic relative to those that were not (OR 1.2; 95% CI 0.8–1.8).
There were 506 foals with sepsis, ‘presumed sepsis’ or SIRS, including 220 culture positive and 286 without a confirmatory bacterial culture; 519 foals were considered to be nonseptic. Considering this broad definition of sepsis, anti-ulcer medication was significantly associated with the presence of sepsis (OR 1.5; 95% CI 1.1–2.1; P = 0.0097). Neither age nor total duration of hospitalisation was significantly associated with a diagnosis of sepsis, and neither variable was significantly associated with sepsis in multivariable RELR models when included with anti-ulcer medication administration as dependent variables.
The results of this study demonstrate that the use of anti-ulcer medication in neonatal foals in an ICU setting increased the odds of developing undifferentiated diarrhoea. This risk was not different for different classes of acid-suppressive therapy (i.e. H2RAs or the PPI omeprazole). The risk for developing CDAD associated with anti-ulcer therapy did not achieve statistical significance in this study. This is similar to the reported OR in human adults and infants, in which the use of acid-suppressive medication increases the risk of developing diarrhoea from a variety of infectious agents (Williams 2001). According to one human study, the adjusted risk ratios for CDAD associated with usage of PPIs and H2RAs are 2.9 (95% CI 2.4–3.4) and 2.0 (95% CI 1.6–2.7), respectively (Dial et al. 2005). However, another study on the relationship between CDAD development and exposure to acid-suppressive therapy in hospitalised patients revealed an association with PPIs (OR 4.5; 95% CI 1.4–14.4) but not with H2RAs (OR 3.6; 95% CI 0.7–18.9) (Turco et al. 2010). In the current study, there was no difference in the effect on development of diarrhoea between H2RAs (ranitidine and cimetidine) and the PPI omeprazole.
An increased risk for the development of CDAD from the use of acid-suppressive medication is not universally found in human studies, however (Kyne et al. 2000; Shah et al. 2000; Feja et al. 2005). This variability in the effects of acid-suppressive therapy on CDAD probably arises due to differing influence of the numerous risk factors that have been found for CDAD in the various clinical settings in which the studies were conducted. One major factor is exposure to the organism, which varies for different populations, and which was reflected by variation in CDAD among sites in our study data.
Surprisingly, sucralfate was also found to increase the odds of developing diarrhoea. This has not been reported previously and, as sucralfate is classified as a cytoprotectant with no effect upon acidity, is not consistent with the presumed mechanism of action to increase risk of infectious diarrhoea. In a large randomised study of 1200 human ICU patients, however, the risk of developing ventilator-associated pneumonia was the same for patients given sucralfate as for those given i.v. ranitidine 3 times per day (Cook et al. 1998). This was surprising, as it has been well documented that the risk of developing ventilator-associated pneumonia is greater in patients treated with acid-suppressive drugs. In our study, only 18 foals were given sucralfate alone, so the results should be interpreted with caution. In addition, it is possible that there was selective use of sucralfate in foals with diarrhoea, or suspected impending diarrhoea. This observation, however, warrants further investigation.
Use of anti-ulcer medications was found to be a strong risk factor for the diagnosis of salmonellosis. This is consistent with previous reports in man; however, the frequency of reported salmonellosis was very low in the foals of the current study, and there was substantial variability in microbiological testing for Salmonella among centres. Hence, the results for salmonellosis should be interpreted with caution and require further validation.
An increased risk of blood culture-positive status associated with the presence of diarrhoea has been reported in neonatal foals (Hollis et al. 2008). In very low birthweight human infants, the use of acid-suppressive medication resulted in an increased risk of bloodstream infections (OR 3.1; 95% CI 0.96–10.2; P = 0.059) (Graham et al. 2006). In another study, the association was even stronger (OR 7.6; 95% CI 2.76–21.2; P = <0.001) (Smith et al. 2010). The increased incidence of bloodstream infections was presumed to occur due to damage to the intestinal mucosa associated with enterocolitis and loss of the mucosal ‘barrier’ function, the major component of which is acidity. The importance of gastric acidity in protecting against bacterial translocation is demonstrated by the observation in mice that acidification of the diet resulted in decreased bacterial translocation and bacterial counts in lungs and lymph node (Boneti et al. 2009). Factors other than acidity are associated with intestinal barrier function, however, as it has been demonstrated that intestinal hypoperfusion in serious illness and SIRS contributes to loss of intestinal barrier function (Deitch 1990; Rahman et al. 2003). Although anti-ulcer medication was found to be a risk factor for a clinical diagnosis of sepsis or SIRS, we were unable to investigate the effects of treatment on bacteraemia as the temporal association of performing the blood culture and beginning treatment with anti-ulcer medications could not be established from the data.
A possible contributing factor to the development of sepsis may be the influence of H2RAs on leucocyte function. Histamine directly stimulates the production of IL-1 and IL-6, and blocking histamine receptors results in diminished inflammatory responses (Vannier and Dinarello 1994; Aybay et al. 1995; Zedtwitz-Liebenstein et al. 2002). Furthermore, experimental evidence suggests that acid-suppressive drugs inhibit neutrophil functions and cytotoxic T lymphocyte and natural killer cell activity, which might add to the increased susceptibility to infection associated with these drugs (Aybay et al. 1995; Mikawa et al. 1999; Zedtwitz-Liebenstein et al. 2002). The clinical outcome of such effects was demonstrated in a report by O'Keefe et al. (1998), in which it was reported that the use of H2RAs resulted in a greater number of infectious complications, higher mortality and longer duration of stay compared with patients that did not receive H2RAs. In the current study, there was no effect of acid-suppressive drugs upon outcome; however, the duration of hospitalisation was longer in foals with diarrhoea. The number or type of complications in the foals was not reported in the current study. Results of the current study reflect at best a weak association with sepsis in this cohort of foals and suggest that the findings should be interpreted and extrapolated to clinical cases with caution.
An interesting finding of the study was the failure to demonstrate a negative association between acid-suppressive therapy and occurrence of gastric ulceration. The development of gastric ulceration associated with critical disease (‘stress ulceration’) is of concern to equine neonatologists and is one of the primary driving forces for widespread (although not universal) use of acid-suppressive therapy in the ICU. In an endoscopic survey of 20 critically ill foals aged <30 days in an ICU, the incidence of gastric glandular ulcers was 40% (Furr et al. 1992). However, in a necropsy study of 85 foals treated in an ICU, the incidence of gastric ulcers was reported to be only 12.9% (11/85), and the incidence of ulceration did not differ between foals treated or not treated with acid-suppressive therapy (Barr et al. 2000). The evidence from these previous studies and the current study suggests that the mechanism of stress ulcer formation in foals does not necessarily involve acid injury alone. Research in animal models and experience in human patients has shown that increased catecholamine concentrations, decreased cardiac output and proinflammatory cytokines all contribute to gastrointestinal hypoperfusion and ischaemia. The consequences of this include reduced mucosal blood flow, which delays cellular restitution and reduces bicarbonate secretion. Ileus may also be present, and if so contributes to reflux and retention of bile (Stollman and Metz 2005). Animal studies have demonstrated that increased gastric acidity, bile salts and ischaemia must all be present for stress ulcers to form (Ritchie 1975).
It is important to recognise that the study by Barr et al. (2000) utilised foals that died or were subjected to euthanasia and did not examine any other foals, so the results may be distorted due to case selection bias. Similarly, presence or absence of gastric ulceration was not well reported in the present study, and was confined almost exclusively to post mortem cases, which might induce selection bias as less severely affected foals were not examined and because data regarding the presence of gastric ulceration were only available for a small fraction of the study population. These results, however, suggest that acid-suppressive medication has poor efficacy at best for prevention of gastric ulceration in neonatal foals undergoing treatment in an ICU. This is supported by the findings of a study of 23 hospitalised foals aged <5 days, in which the effects of ranitidine could not be assessed because the gastric content was primarily alkaline (Sanchez et al. 2001).
Certain limitations exist in the current study. The primary outcome variable was the development of diarrhoea; however, it is likely that some foals presented to the hospital with impending diarrhoea that had not yet become evident at the time of presentation. If such a case were treated with acid-suppressive drugs, the calculated risk of using such drugs on the development of diarrhoea would be artificially increased. We attempted to address this by excluding cases that developed diarrhoea prior to beginning treatment with anti-ulcer drugs; this was usually, but not always, at the time of admission. This was not always possible to confirm, however, because of the retrospective nature of the study and lack of documentation of all medical histories. Misclassification is always a possibility in such retrospective studies, but we are unable to recognise any systematic bias that could have occurred to result in more foals getting diarrhoea after admission in one group over the other. An additional potential limitation was variability in the definition of diarrhoea among clinicians and centres. Identification of diarrhoea in foals, particularly from review of records, is sometimes challenging. Of particular concern in this study is the use of antimicrobials that is a risk factor for the development of diarrhoea in horses and foals. It is considered unlikely that antimicrobial use biased the results, however, because the distribution of case classification and severity was similar in both anti-ulcer treated and nontreated groups, and it is reasonable to assume that antimicrobial treatment would therefore be similar between the 2 groups. Further, the authors consider that antimicrobial usage among hospitalised foals is very high and few foals would not have been treated with antimicrobials. Other risk factors identified in human patients such as the presence of renal failure and hypoproteinaemia, to name but a few, were not examined in this study and warrant further review.
The diagnosis of clostridial diarrhoea in the equine neonate can be problematic, and there is no universally accepted method of confirmation. In the current study, the methods of diagnosis was not standardised among study sites; however, the most commonly used diagnostic method (exotoxin ELISA assay) was used in >85% of cases. While the diagnostic accuracy of this method may not be optimal, it is reasonable to assume that it performed equivalently among treated and nontreated foals, resulting in minimal effect upon the final conclusions.
We believe that the large study size, and the fact that the total number of foals treated or not treated was roughly equal, as well as the assumption that the above clinical effects would be randomly distributed minimised any potential overall influence upon data analysis.
In conclusion, the results of this study demonstrate that the use of acid-suppressive drugs increased the odds of developing undifferentiated diarrhoea during hospitalisation by a factor of 2.0. In addition, results suggested no effect upon reducing gastric ulceration in neonatal ICU patients in a small subset of the study foals. These data indicate that acid-suppressive medication should not be used indiscriminately in the critically ill equine neonate. Prospectively designed studies to address some of the limitations of our study design are indicated to more validly and reliably assess the magnitude of the association between acid-suppressive treatment and diarrhoea in neonatal foals treated in ICUs.
Authors' declaration of interests
No conflicts of interest have been declared.
The authors thank Drs Morgan Kraus and Jennifer Carpernter for help with data collection.
a TIBCO Inc., Seattle, Washington, USA.