Clinical Outcomes of Refeeding Syndrome: A Systematic Review of High vs. Low-Calorie Diets for the Treatment of Anorexia Nervosa and Related Eating Disorders in Children and Adolescents

Over the years, the standard of care for re-alimentation of patients admitted for the treatment of anorexia nervosa (AN) has been a conservative or cautious approach described as "start low and go slow." These traditional refeeding protocols advocate for a low-calorie diet that restricts carbohydrates, with the primary goal of hypothetically lowering the risk of refeeding syndrome (RFS) and its complication. However, no consensus exists for the optimal inpatient approach to refeeding children and adolescents with AN. There is still some disagreement about what constitutes an ideal pace for nutritional rehabilitation. Varying treatment protocols have emerged across the globe, often reflecting the preferences and biases of individual practitioners and contributing to the lack of a universally accepted protocol for refeeding in AN. Although it is widely accepted that low-caloric refeeding (LCR) is safe for inpatient treatment of AN, this strategy has been shown to have several significant drawbacks, leading to increased criticism of the LCR method. Research from the last decade has led to calls for a more aggressive refeeding protocol, one that suggests a higher caloric intake from the offset. As a result, this research aimed to conduct a systematic review of the existing literature on strategies for refeeding hospitalized pediatric/adolescent patients with AN and related eating disorders. We aimed to compare high-caloric refeeding (HCR) and LCR in terms of weight gain, length of stay, and risk of RFS. We conducted a thorough search of medical databases for abstracts published in English, including Google Scholar, PubMed, and MEDLINE, to find relevant studies published between 2010 and February 2023. Our focus was on articles that evaluated high versus low refeeding protocols in children and adolescents hospitalized for treating AN and related eating disorders. Only articles that reported on at least one of the outcome variables of interest, such as hypophosphatemia, weight gain, RFS, or length of hospital stay, were considered. This review included 20 full-text articles published in the last decade on the HCR protocol in children and adolescents, with a total sample size of 2191 participants. In only one of the 20 studies did researchers find evidence of a true clinical case of RFS. We, therefore, found no evidence that HCR increased the risk of RFS in adolescents, even in those with a very low body mass index (BMI). However, evidence suggests a lower BMI at the time of hospital admission is a better predictor of hypophosphatemia than total caloric intake. In conclusion, based on the evidence from this review, a high-caloric diet or rapid refeeding in children/adolescents suffering from AN may be both safe and effective, with serial laboratory investigations and phosphate supplementation as needed. Hence, more research, particularly, randomized controlled trials, is required to help shape an evidence-based refeeding guideline outlining target calorie intakes and rates of advancement to assist clinicians in the treatment of adolescents with AN and related eating disorders.


Introduction And Background
Anorexia nervosa (AN) is a type of eating disorder that, even with treatment, has the potential to be fatal [1]. It has the highest death rate of all mental disorders [2,3] and is characterized by severe dietary restrictions and extreme anxiety about gaining weight [4,5]. The most common age range for the onset of AN is between the ages of 15 and 19 years [6,7], with the majority of affected patients being female adolescents [8][9][10]. It is most severe in adolescents who, due to caloric restriction, excessive exercise, and other behaviors, can quickly become malnourished during a crucial period of growth and development [11]. This emphasizes the significance of hospital admission for medically unstable children and adolescents suffering from severe malnutrition due to AN. The mainstay of treatment includes nutritional support, psychological therapy, and inpatient care [12]. To stabilize the patient's health and undo the effects of malnutrition, nutritional rehabilitation is crucial in the treatment of AN. It has also been reported that a patient's weight gain during hospitalization for AN is an important factor in predicting weight recovery a year later, as well as in facilitating initial stabilization and reducing the risk of rehospitalization [13,14].
The first and most important step toward recovery from AN is reintroducing feeding or re-nutrition [15,16]. However, no consensus guidelines exist for the optimal inpatient approach to refeeding children and adolescents with AN. There is still some disagreement about what constitutes an ideal pace for nutritional rehabilitation. Due to the lack of a universally accepted protocol for refeeding in AN, varying treatment protocols have emerged across the globe, often reflecting the preferences and biases of individual practitioners. The consensus is that the need for adequate nutrition to regain weight and achieve medical stability must be weighed against the risks of refeeding syndrome (RFS) and its potentially fatal complications such as cardiac arrest, arrhythmia, coma, seizures, and sudden death [17,18]. Over the years, the standard of care for re-alimentation in AN has been a more conservative approach popularly described as "start low and go slow." These traditional refeeding guidelines advocate for a low-calorie diet (LCD) that restricted carbohydrates, with the primary goal of hypothetically lowering the risk of RFS and its complication [17]. For instance, the standard of care for in-patient refeeding in the United States has been an LCD beginning at approximately 1200 kcal/day and increasing by 200 additional calories every alternate day [19,20]. In other parts of the world, such as Australia, the United Kingdom, and Europe, daily calorie intakes as low as 200-600 kcal have been advised [20][21][22].
Varying treatment protocols have emerged across the globe, often reflecting the preferences and biases of individual practitioners and contributing to the lack of a universally accepted protocol for refeeding in AN. Although it is widely accepted that low-caloric refeeding (LCR) is safe for inpatient treatment of AN, this strategy has been shown to have several significant drawbacks, leading to increased criticism of the LCR method. Recently, a growing body of evidence suggests that it may be associated with poor initial weight gain, early weight loss (first one to two weeks), and extended hospital stays, which add to the financial and emotional burdens on families and the healthcare system [23,24]. In addition, some argue that LCR has not eliminated the incidence of RFS. Moreover, studies have shown that the severity of pretreatment malnutrition is a more significant risk factor for refeeding hypophosphatemia (RH) than either calorie intake or rate of weight gain [25][26][27]. Moreover, newer research from the last decade has led to calls for a more aggressive refeeding protocol, one that suggests a higher caloric intake from the offset, typically above 1400 kcal/day, and a rapid increase in that amount [28][29][30][31][32][33].
Unfortunately, there is a shortage of research addressing whether or not refeeding at higher calorie levels is risky for severely undernourished adolescents who have AN and associated feeding disorders. Garber et al. made two critical observations in their systematic review of the different approaches to refeeding in hospitalized patients with AN in 2016 [16]. They concluded that higher calorie refeeding is both feasible and safe with close medical supervision and with the correction of electrolyte abnormalities. Furthermore, a systematic review that was conducted in 2012 by O'Connor et al. demonstrated that the severity of malnutrition at the beginning of the re-nutrition process was a more significant factor in determining RH than total energy intake [34]. On the other hand, a randomized controlled trial conducted by Golden et al. in 2021, comparing LCD and high-calorie diet (HCD) with initial calorie intakes of 1400 kcal/day and 2000 kcal/day, respectively, found no significant differences in weight gain, length of hospital stay, or rate of occurrence of RFS [35].
Consequently, the goal of this research was to carry out a systematic review of the existing literature on strategies for refeeding hospitalized pediatric/adolescent patients with AN and related eating disorders. We aimed to compare high-caloric refeeding (HCR) and LCR in terms of weight gain, length of stay, and risk of RFS.

Protocol
This systematic review followed the latest Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [36].

Search Strategy
We conducted a thorough search of medical databases for abstracts published in English, including Google Scholar, PubMed, and MEDLINE, to find relevant studies published between 2010 and February 2023. The following keywords were included in the search strategy: refeeding syndrome, anorexia nervosa, eating disorder, re-nutrition, high caloric diet, low caloric diet, and hypophosphatemia. After locating and retrieving all relevant studies, the reference lists of each of those studies were reviewed to identify any additional relevant studies. The references were imported into Mendeley Web Importer (Elsevier, Amsterdam, Netherlands), where they were compared by author, year, title, and reference type. This enabled the removal of any discovered duplicates.

Eligibility Criteria
We focused on articles that evaluated high versus low refeeding protocols in children and adolescents hospitalized for the treatment of AN and related eating disorders. Articles were included if they were published within the last decade. Furthermore, articles were only included if they outlined a clear refeeding protocol that was reproducible, both in terms of caloric intake and rate of administration. Lastly, articles were included if they reported on at least one of the outcome variables of interest, such as hypophosphatemia, weight gain, RFS, or length of hospital stay (LOS).
We excluded studies solely conducted in non-pediatric/adolescent populations, case studies, and review articles. Furthermore, articles with participants who had comorbidities, such as cancer, or other severe medical disorders, such as renal, cardiac, or hepatic failure, were excluded because these could affect our outcome variables.

Study Selection
Two reviewers (EM and AM) conducted an initial screening, with any discrepancies resolved by our supervisor (RV). Following an initial screening, full-text publications that met our inclusion criteria were added to the database. Figure 1 is a flowchart that depicts the process of selecting appropriate articles for this study. Initial searches of electronic databases yielded 4433 references. A review of the bibliography and reference lists of related studies yielded 77 additional articles. After duplicates were removed, 2889 abstracts were included in the screening process. Initial screening based on inclusion and exclusion criteria resulted in the exclusion of 1915 abstracts. We excluded studies conducted solely in non-pediatric/adolescent populations, articles published more than 12 years ago, case studies, and review articles. We also excluded studies involving malnourished patients who did not have an underlying eating disorder. Following this preliminary selection, we were left with 51 abstracts, the full texts of which were retrieved and re-examined against the inclusion and exclusion criteria. Thirty-one additional studies were excluded because the refeeding method and outcomes were not clearly defined.  Table 1 depicts the varied characteristics of the 20 included studies, more than half of which were done in the United States and Australia. Almost 95% of the papers that satisfied our inclusion criteria were published during the last decade, and the majority (80%) were either randomized controlled trials (RCTs) or retrospective studies (cohort and chart reviews). Ten studies compared low-calorie diets to high-calorie diets [26,35,[37][38][39][40][41][42][43][44][45], whereas the remaining studies only looked at high-calorie diets. There was no standard protocol used. We observed a wide range of refeeding strategies, with varying initiation caloric rates and daily rates of caloric advancement. The average initial refeeding intake ranged from 740 to 1400 for LCR and 1400 to 3000 for HCR. Furthermore, most studies included serial laboratory investigations and phosphate monitoring. In five studies [40,[46][47][48][49], phosphate supplementation was given routinely to all patients as a measure of reducing the risk of RS; however, in other studies, phosphate supplementation was given as needed depending on laboratory readings during admission [26,44,50,51]. In terms of bias assessment, there was a high risk of bias due to the observational/retrospective study designs utilized in the majority of studies, as well as a high risk of attrition bias due to incomplete outcome reporting.

Discussion
The conservative approach to refeeding severely malnourished patients, including those with AN, has traditionally been used in hospitals. This cautious approach of starting low and gradually increasing caloric intake has been thought to be critical in preventing or reducing the risk of RFS. The central idea behind this conservative refeeding protocol, which is not based on any empirical evidence, stems from the pathophysiology for refeeding malnourished patients, specifically the idea that reducing total caloric intake prevents insulin surge, which in turn suppresses the rapid intracellular movement of electrolytes, water, and glucose, thus preventing RFS. While there is consensus that a low-caloric approach is a safe option for the inpatient treatment of AN, this approach has been seen to have several significant drawbacks. Recently there has been mounting evidence that it may be associated with poor weight gain and extended hospital stays. Therefore, in recent years, there has been growing support for a more robust refeeding protocol.
In this systematic review, we compared the results of LCR to those of HCR, specifically looking at the rates of weight gain, hospital stays, and occurrences of RFS or hypophosphatemia.

Initial Caloric Intake and Occurrence of Refeeding Syndrome or Hypophosphatemia
Concerning the potential for hypophosphatemia and RFS, there is currently no consensus regarding the safety of a high-caloric diet re-alimentation protocol for adolescents with AN. Based on the findings of this systematic review, we believe that the concerns about HCR causing RFS have been exaggerated over time.
We found no evidence that HCR increased the risk of RFS in adolescents, even in those with a very low BMI. Our review comprised a total of 20 articles published in the last decade on the HCR protocol in children and adolescents, with a total sample size of 2191 participants. In only one of the 20 studies did researchers find evidence of a true clinical case of RFS [45]. Schlapfer et al. reported a total of 10 cases of overt RFS in their 2022 study comparing HCR and LCR. Although not statistically significant, it was interesting to see that more of these cases happened in the LCR group (4.4%; 6/137) than in the HCR group (2.6%; 4/154) [45].
Furthermore, findings from our research, as illustrated in Table 1, showed that the occurrence of hypophosphatemia was highly variable with total caloric intake, making its prediction challenging. The highest incidence of hypophosphatemia was 45%, as reported by Garber et al., at an initial caloric intake of 800-1200 kcal for LCR and 1400-2400 kcal for HCR [26]. In their study, there was a slightly increased tendency to receive phosphate supplementation in the HCR group than in the LCR. Though the incidence of mild hypophosphatemia was more prevalent in the HCR group, it also occurred in the LCR group. The seemingly inconsistent presence of RH in adolescents with AN who were initiated with both high and low refeeding rates adds to the complexity of this physiological phenomenon and suggests that RH may not be completely related to energy intake. Given the inconsistency in the presentation of RH occurring at varying energy intakes (HCR and LCR), our study suggests that several factors may play a part in the incidence of RH in an adolescent with AN.
Furthermore, in some studies, patients in the HCR group were given routine phosphate supplements at the start of refeeding, resulting in no or few cases of mild hypophosphatemia [40,[47][48][49]53]. In other studies, participants who were deemed to be at risk of developing severe hypophosphatemia based on serial labs performed during admission were given prophylactic phosphate supplementation [26,44,50,51]. This supports the notion that a rapid high-caloric feeding protocol can be safely implemented in hospitalized adolescents with AN, with close monitoring and electrolyte supplementation as needed.

Initial Mean BMI and Occurrence of Hypophosphatemia
Five studies found an association between BMI at the time of admission and the incidence of hypophosphatemia; specifically, they reported that a lower BMI at admission was a better predictor of hypophosphatemia than the initial caloric intake [30,35,37,42,44]. This was similar to the conclusion of O'Connor and Nicholls in their systematic review in 2013 on RH in adolescents with AN [34]. They included 17 publications in their article including 10 case reports, five chart reviews, and an observation study. Like our study, they found that the severity of hypophosphatemia was not influenced by total caloric intake but was directly associated with decreasing BMI on admission. Whitelaw et al.'s 2010 study was one of the very first to examine the HCR refeeding protocol. They observed an increased incidence of hypophosphatemia in HCR participants with an ideal body weight of less than 68% at the time of admission [30].

Rate of Weight Gain and Total Caloric Intake
It has been reported that the amount of weight a patient with AN gains while hospitalized is a key determinant both for early stabilization and in predicting weight recovery a year later. The findings of this study, along with other studies published in the last decade, cast doubt on the conservative "start low, go slow" feeding model. In this review, 15 studies looked at weight gain as an outcome; seven compared weight gained in the high-caloric group to that of LCR protocols [26,35,40,41,[43][44][45], and eight looked only at the HCR protocol [27,39,47,[49][50][51][52][53]. In 11 of these 15 studies, the HCR group's average weekly weight gain was significantly higher than that of the LCR group, ranging from 0.82 kg to 2.7 kg [26,27,40,41,44,46,47,[49][50][51][52]. The most significant weight gain was reported in Garber et al.'s prospective observational study, which used a quasi-experimental design to compare LCR versus HCR refeeding methods. They observed that after one week of admission, the rate of weight gain in the HCR group was nearly double than that of the LCR group [26]. These studies demonstrated the feasibility of faster weight gain in HCR participants without an increased risk of RFS. This is an important finding that could help doctors ensure that AN patients get the nutrition they need to recover from the disease thereby preventing problems like inadequate refeeding, suboptimal weight gain, or even weight loss in the early phase of treatment. Furthermore, the sooner these patients gain weight, the better their cognitive function, which in turn leads to better engagement in additional therapy such as psychotherapy, which is required for eventual recovery from AN.

Total Caloric Intake and Length of Hospital Stay
Nine studies examined the LOS as an outcome variable, with seven finding a correlation between HCR refeeding protocol and shorter hospital stay [26,35,39,40,45,50,51], while the other two found no significant difference in length of stay between the two groups [44]. The most notable was reported by Agostino et al., who reported a statistically and clinically significant reduction in hospital stays of 17 days [40]. Given the scarcity of inpatient AN treatment facilities, which results in a lack of available beds and prohibitive costs, reducing the LOS not only saves money but also allows for the treatment of more patients. This review's findings are consistent with those of Staab et al., who conducted a prospective cohort study on 325 adult patients with a BMI less than 18 who were hospitalized for the treatment of AN in Canada [53]. They found an average LOS reduction of 21 days when using the quick refeeding approach. According to their study, each patient could expect to save $7,748 for every 13 days of LOS reduction [53].

Strengths and limitations
The greatest strength of our review is the inclusion of 20 articles, including five RCTs, as opposed to the previous two systematic reviews on this topic, which had limited RCTs and included case reports.
The main limitation of this review is that the definition of hypophosphatemia was not consistent across studies. It is possible that RH is either under or over-reported. In addition, there was no consistent definition of HCR or LCR across the articles. The final estimates of outcome variables may be impacted by the fact that what was considered HCR in some studies was LCR in others.

Conclusions
The result of this study provides strong evidence challenging the traditional refeeding approach to managing adolescent patients hospitalized for AN. Our findings suggest that the severity of hypophosphatemia or RFS may not correlate with total caloric intake but may be influenced by the severity of malnutrition before admission. With serial laboratory investigations and phosphate supplementation, when necessary, a highcaloric diet or rapid refeeding in children/adolescents suffering from AN may be both safe and effective. In addition, HCR may shorten the LOS, which minimizes cost and improves weight gain, which is beneficial in long-term recovery. In conclusion, more research, particularly, RCTs, is required to help shape an evidencebased refeeding guideline outlining target calorie intakes and rates of advancement to assist clinicians in the treatment of adolescents with AN and related eating disorders.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.