This study was performed retrospectively on a population of 142 consecutive patients who had been admitted to the pediatric ward at Dongsan Medical Center, Keimyung University School of Medicine between March 2003 and February 2006. The inclusion criteria were as follows: age 15 to 45 days, chief complaints of vomiting and/or diarrhea, and cow's milk formula or cow's milk and breast milk mixed feeding. During this period, 402 patients, aged 15 to 45 days, were admitted to our pediatric ward, and 210 patients had chief complaints of vomiting and/or diarrhea. Among these patients, 66 patients who had been exclusively breast-fed were excluded because CMPIE has not been known to occur in exclusively breast fed babies (2). Two patients were finally diagnosed as having CMPIE, but were excluded in the study population because they had been referred under the impression of CMPIE from the outside hospital. Hence, we analyzed 142 consecutive patients. Study procedures were approved by the Keimyung University Institutional Review Board.

CMPIE was diagnosed using an open standardized OCC based on the diagnostic guidelines suggested by Powell (3, 4), Burks et al. (5), and Sicherer et al. (2). On admission, we regarded 26 patients as being at risk of having CMPIE based on associated descriptions (2-9, 12) and authors' personal experiences. When feeding was possible, a protein hydrolysate formula (HA®, Maeil Dairy Industry Co., Seoul, Korea) or exclusively breast milk was fed until the final diagnosis was made and laboratory work ups for electrolytes and acid-base imbalance, anemia, underlying infection, surgical conditions, metabolic disease, and immune disorder or other underlying diseases were performed and treated. When patients showed a stable general condition with good weight gain, and in the absence of any other identifiable cause, we performed a single and open standardized OCC (12) with informed parental consent. Total or partial parenteral nutrition was not adopted in any patient suspected of having CMPIE. Before OCC testing, baseline studies were performed, i.e., a stool smear test for occult blood, RBC and leukocyte counts, and blood sampling for absolute neutrophil count (ANC). An open challenge was performed using 0.15 g of cow's milk protein per kilogram of body weight in one feed, and patients were observed for 4 hr in a nothing-per-oral state. After OCC, clinical symptoms and laboratory findings were observed and evaluated, and in particular, the following five criteria were recorded: projectile vomiting >2 or more (at 1-4 hr post-OCC); lethargy with relative insensitivity to stimulation (at 1-4 hr); a bloody and/or a pus-like diarrhea (at 6-10 hr), all three of which were absent at baseline; a rise of peripheral blood ANC >3,500 cells/µL from baseline (at 6 hr) (2-5); and a stool smear test for occult blood and/or WBC, RBC >5 cells/HPF, not present by baseline stool smear testing (at 6-10 hr). Diaper rash and anal fissure were eliminated as possible blood sources. A positive challenge was defined as the presence of two or more positive results of these five criteria. According to these diagnostic criteria, was diagnosed in 16 (61.5%) patients among the 26 suspicious cases. Only one CMPIE patient was fed cow's milk and breast milk mixed on admission. After diagnosis, she was fed protein hydrolysate formula and breast milk mixed owing to the lack of exclusive breast milk feeding to induce adequate growth.

The 142 study subjects were divided into three groups based on underlying etiologies, i.e., the CMPIE, infection, and non-infection groups. All the patients allocated to the infection and non-infection groups were successfully re-fed cow's milk formula, and experienced no clinical symptoms or complications during one week of subsequent observation or after discharge from hospital. Subjects were allocated to the infection group when they failed to meet the diagnosis criteria for CMPIE and were deemed to have an infectious condition based on clinical findings, e.g., febrile illness, and the following laboratory results, e.g., positive culture, pyuria, positive stool rotavirus test, and abnormal rise of serum C-reactine protein (CRP) level. Subjects were allocated to the non-infection group when they were not allocable to the CMPIE or infection groups and had no clinical evidences of infection.

Seven index parameters for CMPIE were considered based on clinical findings that were reported to be characteristic of CMPIE (2-9, 12) and based on the authors' clinical experiences (12); daily weight gain between birth and admission (body weight on admission [g]-birth weight [g]/age [day]), serum albumin level on and during admission, peripheral blood leukocyte count, metabolic acidosis, peripheral blood eosinophil count, peripheral blood platelet count, and abnormal stool smear test. Failure to gain weight was defined as a daily weight gain <10 g. A low serum albumin level was regarded as <3.5 g/dL. Peripheral blood leukocytosis was regarded as >19,500 cells/µL, and metabolic acidosis as a serum bicarbonate level <22 mEq/L. Eosinophilia was defined as >500 cells/µL, and peripheral blood thrombocytosis as >750,000 cells/µL. An abnormal stool smear test was defined as positivity for occult blood and/or WBC, RBC >5/HPF. The results corresponding to the above seven parameters were compared in the three groups. The presence of methemoglobinemia (≥15%) was also included.

analysis of variance (ANOVA) with Turkey's test for all pairwise comparisons was used to compare the above-mentioned six indexes, except a stool smear test, and symptom durations between the groups. An abnormal stool smear test was analyzed by Fisher's exact test. The paired t-test was used to compare changes in serum albumin levels, and Fisher's exact test was used to compare the methemoglobinemia presence in the three groups. Multiple logistic regression analysis was performed on the CMPIE and infection group; CMPIE presents clinically as a sepsis-like illness or gastroenteritis in neonates or in early infancy and resembles clinical findings in the infection group, which may not be difficult to differentiate from the clinical findings of the non -infection group. Also, multiple logistic regression analysis was performed in terms of the three ANOVA-identified independent risk factors; metabolic acidosis was not determined because blood gas analysis was performed only in 62 patients (60.8%) in the infection group. Statistical significance was accepted for p values of <0.05. Data were presented as mean±SD.

Of the 142 study subjects, CMPIE was diagnosed in 16 patients (11.3%), and the infection and non-infection group were composed of 102 patients (71.8%) and 24 patients (16.9%), respectively. Mean ages of patients at admission of the CMPIE, infection, and non-infection groups were 27.0 ±8.8 (18-42), 24.7±8.7 (15-44), and 27.9±9.5 (16-44) days, respectively. Mean symptom durations before admission were 4.5±2.4 (1-9), 3.7±2.5 (1-11), and 4.7±2.7 (1-9) days, respectively, and no significant differences were observed between groups. Family history of atopy was observed in 2 (12.5%), 9 (8.8%), and 4 (16.7%) patients, respectively (1).

On admission, daily weight gain (p=0.003), serum albumin level on admission (p=0.035), peripheral blood leukocyte count (p=0.012), and metabolic acidosis (p=0.015) were more significant in the CMPIE group than in the other two groups. Peripheral blood eosinophil count, platelet count, and an abnormal stool test results were not significantly associated with CMPIE group compared with the other two groups (Table 1). Methemoglobinemia was observed in 3 patients (18.8%) in the CMPIE group, but not in the other two groups (p=0.001).

Daily weight gain in the CMPIE group was -3.7±12.2 (-32.5-13.3) g/day, and 81.2% of CMPIE group patients had a daily weight gain of <10 g. Daily weight gains in the infection and non-infection groups were 17.2±21.5 (-48.2-52.6) g (p<0.05, vs. CMPIE) and 21.2±12.7 (-7.5-36.6) g (p< 0.05, vs. CMPIE), respectively, and 32.4% and 29.2%, respectively, were <10 g per day.

The mean serum albumin level at admission in the CMPIE group was 3.3±0.9 (2.2-4.4) g/dL, and 56.3% of CMPIE patients had a level of <3.5 g/dL. The mean serum albumin levels in the infection and non-infection groups were 3.8±0.5 (3.1-4.8) g (p<0.05, vs. CMPIE) and 3.7±0.4 (3.3-4.5) g, respectively, and 16.7% and 29.2% of the patients in these groups were <3.5 g/dL. In the CMPIE group, the mean serum albumin level decreased to 2.6±0.3 (2.2-3.0) g/dL during admission (at 6.1±1.5 [4-11] days after admission) (p<0.05). In fact, at some stage after admission, all patients in the CMPIE group had an albumin levels of <3.0 g/dL.

The mean peripheral blood leukocyte count on admission in the CMPIE group was 18,311±4,901 (10,530-29,920) cells/µL, and 43.8% of CMPIE patients had a leukocyte count >19,500/µL, whereas mean peripheral blood leukocyte counts in the infection and non-infection groups were 12,010±5,984 (2,520-44,100) cells/µL (p<0.05, vs. CMPIE) and 10,125±2,930 (5,790-18,590) cells/µL (p<0.05, vs. CMPIE), respectively, and 15.7% and none of patients in these groups had a mean peripheral blood leukocyte count of >19,500/µL.

Serum bicarbonate levels were measured in all CMPIE group patients, in 62 patients (60.8%) in the infection group and in 14 patients (58.3%) in the non-infection group. The mean serum bicarbonate level on admission in the CMPIE group was 15.2±6.1 (8.2-28.5) mEq/L, and 82.1% of these patients were <22.0 mEq/L, whereas mean serum bicarbonate levels in the infection and non-infection groups were 18.3±5.5 (7.0-24.8) mEq/L and 23.4±4.6 (15.0-30.8) mEq/L (p<0.05, vs. the CMPIE group and p<0.05, vs. the infection group), respectively, and 79.4% and 33.3%, respectively, had a mean serum bicarbonate level of <22.0 mEq/L.

Multiple logistic regression analysis was performed on the CMPIE and infection group in terms of the three ANOVA-identified independent risk factors, i.e., daily weight gain, serum albumin level on admission, and peripheral blood leukocyte count. A failure to gain weight of <10 g/day (OR, 10.75 [95% CI, 1.53-66.12]) (p=0.014) and serum hypoalbuminemia <3.5 g/dL (OR, 9.53 [95% CI, 1.62-49.01]) (p=0.010) were identified as significant indexes of CMPIE.