An open-label, interventional, randomized, comparator-controlled, and multicenter study was conducted across 15 sites in the United Arab Emirates, Kuwait, Kingdom of Saudi Arabia, Oman, and Lebanon.

Between December 2016 and October 2017, infants were recruited from outpatient clinics at 15 participating hospitals. Infants aged six months or less, fully formula-fed, and meeting modified Rome IV criteria (at least one of the following: two or fewer defecations per week; a history of excessive stool retention; painful/hard bowel movements) were eligible for inclusion. Patients were excluded if they were exclusively breastfed or mixed-fed, diagnosed with infant dyschezia, had an organic cause of constipation, or on laxative treatment. Infants with a significant developmental delay associated with a musculoskeletal or neurologic condition affecting the gastrointestinal tract were excluded as well. Infants with constipation secondary to endocrine, metabolic, neurologic, organic, autoimmune disorders, surgery, or drugs as well as those with clinically significant cardiac, vascular, liver, pulmonary, or psychiatric disorders, severe renal insufficiency, human immunodeficiency virus or acquired immunodeficiency syndrome, hepatitis B or hepatitis C were also excluded. Finally, infants with clinically significant abnormalities of hematology, urinalysis, or blood biochemistry at the time of screening, known cow milk protein allergy, allergy to investigational formulae, and anemic infants under oral iron supplementation were excluded.

The protocol and consent form were reviewed and approved by the Institutional Review Board/Independent Ethics Committee of each participating site prior to study initiation. The study was not registered in a central database. Written informed consent was obtained from the parents or guardians of infants meeting the inclusion criteria prior to study initiation.

Consenting patients were randomized in a 1:1 ratio to receive either the Mg-rich formula (Novalac IT 1; United Pharmaceuticals SAS, Paris, France) or the control starter formula (Similac Comfort 1; Abbott Nutrition, Columbus, OH, USA) treatment for 30 days, as per a computer-generated random sequence. A summary of formula compositions is provided in Table 1. Subjects were fed exclusively with the study formulae, according to each formula's standard feeding schedule provided by the manufacturer. Dosing as well as preparation and storage of formula were in accordance with the manufacturer's recommendations. If constipation continued for three days after starting either of the formula treatment, rescue medication was permitted. Rescue medication was the administration of 2.5 mL lactulose (Duphalac^®) or if constipation persisted for further two days despite lactulose treatment, infants were given a micro-enema (Microlax^®), PEG 3350 (Movicol^®), or other laxatives, at the physician's discretion. Use of rescue medication, including dose, route, frequency, indication, and duration of use were recorded in the patient case report form (CRF). In the 30-day study period, the use of medications influencing intestinal motility was not allowed, unless necessary for the infants' wellbeing; in which case, information regarding the type of medication, dose, route, frequency, indication, and duration of use was recorded in the CRF.

Three physician visits per patient were scheduled at day 0 (baseline), 8, and 30. At the baseline visit, a CRF was completed by the physician and the information relating to patient demographics (age, gender, and ethnicity), birth weight and length, head circumference, gestational age at birth, mode of delivery, number of siblings, number of infants born in the same pregnancy, medical history, and feeding history was obtained. At the 2 follow-up visits, on days 8 and 30, further CRFs were completed, which recorded details of the clinical examination of the patient, concurrent medications, assessment of constipation symptoms, adherence to formula, and occurrence of adverse events.

Parents completed the patient diaries on days 1 to 7 and 23 to 29, to describe the patient's gastrointestinal symptoms. Information regarding the number of stools, stool consistency, crying during defecation, pain during defecation (using Faces Pain Scale), withholding stools, use of rescue medication, food intake, medication use, number and duration of crying episodes, adverse events, and parent satisfaction with treatment were reported at each follow-up visit.

Primary endpoints were improved stool consistency and increased stool frequency. Stool consistency was reported as per a pre-specified stool scale of three groups wherein group 1 consisted of "hard stools" and was defined as pellets/separate hard lumps like nuts or sticky dough-like consistency, group 2 consisted of “normal stool” which was defined as soft stool with smooth consistency, and group 3 consisted of watery stool. An improvement consisted of a change from a constipated stool to stool with normal consistency. Stool frequency was defined as the maximum number of stools reported per day over the observed period as described in the patient's diary. Secondary endpoints were changes in pain during defecation, need for rescue medication, parents' satisfaction with treatment, and parents' willingness to continue formula beyond seven days. At day 7, complete responders were infants achieving both primary endpoints, i.e. improved stool consistency and increased stool frequency; partial responders were infants achieving either of the primary endpoints; and non-responders were infants not achieving either of the primary endpoints. At day 29, complete responders were infants sustaining both the primary endpoints (having achieved both endpoints at day 7); partial responders were infants achieving either of the primary endpoints (irrespective of having achieved either of the endpoints at day 7); and non-responders were infants not achieving either of the primary endpoints.

Descriptive analysis using counts and percentages was performed for primary and secondary study endpoints for the full analysis set population, which included all randomized patients receiving at least one dose of formula. For percentage calculations, patients with missing information relating to the study endpoints were not included in the denominator. Missing data were treated as missing, no imputations were performed. Differences between the treatment groups were analyzed using χ² or Fisher's exact tests for categorical variables and 2-sample t-test for continuous variables. Statistical analysis was performed using SAS software (version 9.4; SAS Institute, Cary, NC, USA).

To detect a difference of 20% between the 2 treatment groups for both the primary endpoints with 80% power and 5% 2-sided significance level, a sample size of 128 patients were required in each group, assuming an efficacy rate of 40% in the control group. We considered a dropout rate of 10% and concluded that 142 patients needed to be included in each treatment group.

A total of 286 patients were recruited between December 2016 and October 2017; one patient was withdrawn from the study prior to receiving treatment due to possible lactose intolerance and was excluded from the analysis. The reasons why a small number of infants (n:7) were lost to follow-up are unknown. No significant differences with respect to baseline characteristics were found between the 2 groups (Table 2). Seven patients were lost to follow-up at the day 8 visit (Fig. 1). Final analysis of the 285 randomized patients receiving at least one dose of formula was undertaken: 143 patients (53.9% male; median [standard deviation; SD], age 6.1 [3.43] months) in the Mg-rich group and 142 (47.2% male; median [SD], age 5.2 [2.95] months) in the control group.

Normal stool consistency was significantly more common in the Mg-rich group (n=117, 81.8%) compared to the control group (n=58, 41.1%; p<0.001) at day 7 after initiation of treatment (Table 3). Similarly, at day 29, normal stool consistency was more common in the Mg-rich group (n=135, 95.7%) compared to the control group (n=89, 65.0%; p<0.001).

Stool frequency was also greater in the Mg-rich compared to the control group (p<0.001) at day 7 after initiation of treatment, with 124 (86.7%) infants fed Mg-rich formula having passed one or more stools per day compared to 96 (68.1%) infants fed the control formula (Table 3). Similarly, at day 29, stool frequency was greater in the Mg-rich group compared to the control group (p<0.001), with 140 (99.3%) infants fed the Mg-rich formula having a stool frequency of one or more stools per day compared to 128 (93.4%) infants fed the control formula. None of the children presented with large diameter stools.

Greater numbers of complete responders to treatment (i.e. infants with improved stool consistency and increased stool frequency) were observed in the intervention group at day 7 (n=38, 26.6%; control n=6, 4.3%; p<0.001) and day 29 (n=36, 25.5%; control n=5, 3.7%; p<0.001) following treatment (Fig. 2). Numbers of complete or partial responders to treatment were greater in the Mg-rich formula group (n=116, 81.1% vs. n=89, 63.1%) at day 7.

The enhanced decrease in mean pain score was observed over the first 7 days of treatment in the Mg-rich group (day 0: 6.97; day 7: 1.86; mean change: –4.85) compared to the control group (day 0: 6.17; day 7: 4.65; mean change: –2.30). Significant differences between the groups were observed from day 3 of treatment (Fig. 3).

Rescue medication was required in the first 7 days following initiation of the different formula for 5 infants in both the groups (3.50% intervention group; 3.52% control group). Rescue medications received were specified in nine of the ten infants: lactulose (n=1) and glycerin suppositories (n=8).

Overall satisfaction, as shown in Table 4, was significantly higher in the intervention group (32.2% very satisfied) compared to the control group (4.9% very satisfied) after 7 days. Satisfaction in terms of stool consistency, stool frequency, and crying time were all significantly higher in the intervention (35.7%, 34.3%, and 36.4% very satisfied, respectively vs. 3.52%, 8.45%, and 4.93% very satisfied, respectively). Similarly, at day 29, overall satisfaction was significantly higher in the Mg-rich formula group (80.9% very satisfied vs. 10.2%). Satisfaction in terms of stool consistency, stool frequency, and crying time were all significantly higher in the treatment group (80.1%, 72.3%, and 72.3% very satisfied, respectively vs. 18.3%, 6.57%, and 11.0%, respectively). There was no difference in intake between both the groups, resulting in a comparable, normal weight gain. In both groups, adverse effects were not observed.

After 7 days of treatment, 99.3% (n=142) of parents in the intervention group were willing to continue formula treatment compared to 95.8% of parents in the control group (p=0.07). However, after 29 days of treatment, almost all the parents were willing to continue with the Mg-rich formula (n=138, 97.9% vs. n=72, 52.6%; p<0.001) in the control group.

Adherence to formula, determined at day 7 after starting the treatment, was high in both the groups, with a single patient using another infant formula (0.7%). There were no reports of the formula being stopped for more than 48 hours during the study period.