Introduction

Early, aggressive nutritional intervention among hospitalized patients with restrictive eating disorders (RED) is safe, shortens hospitalization, and improves treatment response [1,2,3,4,5,6,7,8,9]. Renourishment is key to addressing medical and emotional complications of RED, a process critical to recovery [10,11,12].

Refeeding syndrome (RS)—a potentially life-threatening metabolic response to abrupt nutritional advance involving hypophosphatemia, hypokalemia, and hypomagnesemia—is one risk of aggressive nutritional rehabilitation [13]. RS is relatively uncommon in hospital settings due to preventive practices (e.g., gradually advancing nutrition while monitoring serial electrolytes/exam findings) [11, 14], most often occurring among individuals < 70% of treatment goal weight (TGW) [10, 15] in the first week of nutritional rehabilitation [2, 13, 16]. Most studies examining RS focus only on hypophosphatemia without considering associated hypokalemia and hypomagnesemia [15], or appreciating normal physiologic variability and other potential causes of electrolyte disturbance in malnourished individuals [13]. The current study examined changes in multiple electrolytes and need for supplementation in hospitalized youths < 70% TGW undergoing refeeding; we expected low rates of RS and minimal need for supplementation.

Methods

Participants and procedures

We studied inpatients < 75% TGW (ages 11–26y) medically hospitalized with RED between May, 2015–February, 2021. The target weight range was selected because it aligns with the criteria for medical hospitalization from the Society for Adolescent Health and Medicine and allows for comparison of severely and moderately malnourished inpatients [10, 11, 15]. Participants were identified through electronic medical record (EMR) reports of inpatients admitted to the Adolescent Medicine eating disorder service at < 75% TGW on admission. Participants were managed on a standard refeeding protocol including a daily multivitamin but no targeted prophylactic electrolyte supplementation for RS; hospital formulary multivitamins used during the study period did not contain any phosphorus, potassium, or magnesium. Twelve participants were admitted more than once during the study period; each admission was considered a separate encounter. Participants were categorized as moderately (< 75% TGW) or severely (< 70% TGW) malnourished on admission. Inpatient dietitians used a standard approach to estimating an individual’s TGW, including consideration of both percent of the mean BMI for age and change from pre-morbid BMI growth trajectory [11]. Hospitalization criteria aligned with established clinical guidelines [11].

Phosphorus, potassium, and magnesium were drawn every morning for 5 days following initiation of the protocol and on alternating days thereafter; additional laboratory testing was obtained as clinically indicated. Potassium ≤ 3.5 mEq/L; phosphorous ≤ 3.0 mg/L; and/or magnesium ≤ 1.4 mEq/L were considered abnormally low. Participants typically started a 1750 kcals/day diet and advanced by 500 kcal every other day until target intake was achieved; exceptions were directed by the team’s clinical judgement (e.g., younger patients with lower caloric goals might start below 1750 kcal/day). Since electrolyte disturbance in malnutrition can have various etiologies which inform management, the decision to replete electrolytes was made based on clinical assessment of laboratory findings (results, trend over time, and single vs multiple electrolyte disturbance), vitals, and physical exam. The study was approved by the hospital’s Institutional Review Board.

Measures

Retrospective chart review identified demographics; anthropometrics; reason for hospitalization; illness severity; nutritional plan; laboratory findings; and electrolyte interventions to address RS (Table 1). Two researchers cross-checked 20% of the data.

Table 1 Demographics and clinical data
Full size table

Statistical analyses

Descriptive statistics were applied to demographic and nutrition-related variables. Independent t- and chi-square tests compared participants < 70% and 70–75% TGW on rates of hypophosphatemia, hypokalemia, and/or hypomagnesemia, and supplementation receipt.

Results

Among 81 encounters, %TGW on admission ranged from 55 to 74% (69.8% ± 4.0%), with 35.8% admitted at < 70% TGW. Of those < 70% TGW, 97% were > 60% TGW. Demographics were similar between < 70% and 70–75% TGW groups (Table 1). Comorbid psychiatric conditions were present in 79.3% of participants (anxiety: 94%; mood: 56%; developmental disorders: 12%). Medical co-morbidities included gastrointestinal (19.0%), pulmonary (9.5%), neurologic (7.8%), endocrine (6.3%), hematologic (4.8%), and cardiac (3.2%) concerns.

Though ranges of caloric start and advance were wide, nearly ~ 80% of participants both started between 1500–2500 kcal/day and increased by 250–350 kcal/day. There were no differences between those 70–75% versus < 70% TGW in pre-admission caloric intake, initial inpatient caloric requirements, average daily increases, or discharge caloric requirements (Table 2). The two groups differed on length of stay and change in %TGW during admission (p < 0.05).

Table 2 Analysis of subgroups
Full size table

Hypophosphatemia, hypokalemia, and/or hypomagnesemia were present in 26.9% of participants 70–75% TGW and 34.5% of participants < 70% TGW (29.6% of the entire sample). Supplementation was prescribed in 22.5% of cases where levels were low (8.6% of the entire sample): 5 received phosphorous and potassium; 1 potassium; 1 phosphorous; none received magnesium. Nadirs triggering supplementation ranged from 2.4–3.3 mEq/L for potassium and 2.0–2.8 mg/L for phosphorus. Average nadirs among participants receiving the intervention were: phosphorous 2.4, magnesium 1.4, and potassium 3.0. Participants who did not receive supplementation were monitored closely and experienced physiologic correction. Among participants who did not receive supplementation, average nadirs were as follows: phosphorous 2.8, potassium 3.4, and there were no magnesium abnormalities in this group. Rates of electrolyte abnormalities and supplementation did not differ between the two groups (Chi-squared 0.51, p = 0.475; 0.17, p = 0.684 respectively).

Seven variables were tested for association with electrolyte disturbance and/or need for supplementation (Table 2). Percent TGW was associated with both outcomes only in participants < 70% TGW (3.046, 3.784, p < 0.01).

Discussion

This study examined electrolytes associated with RS and found that rates of hypophosphatemia, hypokalemia, and/or hypomagnesemia were low despite the absence of prophylactic electrolyte supplementation. Though supplementation was more common among participants < 70% TGW, findings suggest that even in these lowest-weight patients, assertive nutritional advancement may be safely undertaken in the inpatient setting with vigilant watchful waiting and informed decision-making regarding electrolyte monitoring and management.

Standard practice for electrolyte management in early refeeding of hospitalized patients with RED has not been established [15]. Concern for RS typically arises based on review of an electrolyte panel alongside physical assessment [13, 15]. Universal supplementation prevents clear assessment of an individual’s physiologic function in a time of metabolic transition. Indeed, the majority of study patients with electrolyte derangements corrected without intervention, which suggests that watchful waiting—as opposed to reflexive or prophylactic supplementation—may be appropriate for hospitalized patients with severe malnutrition.

Our study had several limitations. First, the sample size was small, presenting a challenge to data analysis by smaller sub-groups (e.g. age), though our focus on severely malnourished inpatients addresses a notable gap in the existing literature. Second, although dietary management was generally consistent with an established protocol, data were collected over several years during which the protocol’s caloric starting point and advance were increased to reflect adjustments in standards of care [1]; this prevented application of universal caloric metrics. Similarly, supplementation was not implemented systematically or in a randomized fashion due to pragmatics in the clinical context (i.e., multiple factors needing to be considered/further assessed to determine supplementation), thus outcomes could not be definitively linked to supplementation status. Third, sociodemographic descriptors were limited by data in the EMR, including binary gender terms and broad characterization of race/ethnicity. Fourth, several participants were admitted multiple times, which may have introduced confounds related to severity or duration of illness; however, repeat admissions are part of the reality of inpatient eating disorder care and therefore important to consider in examination of RS in this setting. Finally, hypophosphatemia, hypokalemia, and/or hypomagnesemia in RED cannot always be attributed to RS; other potential reasons for electrolyte disturbance (e.g., purging) might present confounding factors.

Inpatient providers routinely contend with refeeding risks without a clear definition of RS [17] or standard electrolyte management guidelines. By comprehensively examining electrolytes during refeeding without use of prophylactic supplementation, this study might contribute to future guidelines on how to best manage RS risk in vulnerable individuals.