MEDICINE

Review Article

Refeeding Syndrome
Diagnostic Challenges and the Potential of Clinical Decision Support Systems

Lara Heuft*, Jenny Voigt*, Lars Selig, Michael Stumvoll, Haiko Schlögl**, Thorsten Kaiser**

Summary
Background: Refeeding syndrome (RFS) can occur in malnourished patients when normal, enteral, or parenteral feeding is resumed. The syndrome
often goes unrecognized and may, in the most severe cases, result in death. The diagnosis of RFS can be crucially facilitated by the use of clinical
decision support systems (CDSS).

Methods: The literature in PubMed was searched for current treatment recommendations, randomized intervention studies, and publications on RFS
and CDSS. We also took account of insights gained from the development and implementation of our own CDSS for the diagnosis of RFS.

Results: The identification of high-risk patients and the recognition of manifest RFS is clinically challenging due to the syndrome’s unspecific
symptoms and physicians’ lack of awareness of the risk of this condition. The literature shows that compared to patients without RFS, malnourished
patients with RFS have significantly greater 6-month mortality (odds ratio 1.54, 95% confidence interval: [1.04; 2.28]) and an elevated risk of admis-
sion to intensive care (odds ratio 2.71 [1.01; 7.27]). In a prospective testing program, use of our own CDSS led to correct diagnosis in two thirds of
cases.

Conclusion: RFS is difficult to detect and represents a high risk to the patients affected. Appropriate CDSS can identify such patients and ensure
proper professional care.

Cite this as
Heuft L, Voigt J, Selig L, Stumvoll M, Schlögl H, Kaiser T:
Refeeding syndrome—diagnostic challenges and the potential of clinical decision support systems.
Dtsch Arztebl Int 2023; 120: 107–14. DOI: 10.3238/arztebl.m2022.0381

M
* Joint first authors alnourished patients in whom feeding is resumed
** Joint last authors following significantly reduced or no caloric in-
Institut für Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University of Leipzig take lasting at least 5 days are at risk for refeeding
Medical Center, Leipzig, Germany: Dr. med. Lara Heuft, Dr. rer. nat. Jenny Voigt, Prof. Dr. med. syndrome (RFS). Any form of feeding can cause RFS:
MaHM Thorsten Kaiser
oral, enteral, parenteral as well as simple normal meals.
Institute for Human Genetics, University of Leipzig Medical Center, Leipzig, Germany: Dr. med. Lara Enteral feeding poses the highest risk for RFS (1).
Heuft
The cardinal symptom is considered to be hypo-
Department of Endocrinology, Nephrology, Rheumatology, Division of Endocrinology, University of
Leipzig Medical Center, Leipzig, Germany: M. Ed. Lars Selig, Prof. Dr. med. Michael Stumvoll phosphatemia with onset within 2–5 days of resumed
Department of Endocrinology, Nephrology, Rheumatology, Special Division of Endocrinology, feeding, accompanied by hypokalemia and/or hypo-
University of Leipzig Medical Center, Leipzig, Germany Klinik und Poliklinik für Endokrinologie, magnesemia (2). This can be associated with severe
Nephrologie, Rheumatologie, Spezialbereich Ernährungsmedizin, Universitätsklinikum Leipzig: M. fluid shifts and cause central and peripheral edema. In
Ed. Lars Selig, Prof. Dr. med. Michael Stumvoll, Dr. med. Haiko Schlögl
addition, neuromuscular, cardiac, and central nervous
Helmholtz Institute for Metabolic, Obesity, and Vascular Research (HI-MAG), Helmholtz Zentrum
München, University of Leipzig, and the University Hospital Leipzig, Leipzig, Germany: Prof. Dr. med. complications such as tachycardia and impaired cog-
Michael Stumvoll, Dr. med. Haiko Schlögl nition may develop.
Institute for Laboratory Medicine, Microbiology, Clinical Pathology, and Pathobiochemistry, University RFS often goes unrecognized due to the fact that its
Hospital Lippe, Detmold, Germany: Prof. Dr. med. MaHM Thorsten Kaiser symptoms are nonspecific (3). By way of example, a
recent survey of 281 physicians clinically active in
Germany and based on sample case vignettes of RFS
cme plus found that only 14 % of physicians and nutritionist
teams correctly diagnosed RFS (4). Moreover, half of
This article has been certified by the North Rhine Academy for Continuing the respondents in audits (New Zealand and the
Medical Education. The questions on this article can be found at United Kingdom) of prescribing practices in paren-
http://daebl.de/RY95. The submission deadline is 16 February 2024. teral nutrition (PN) were unable to identify patients at
Participation is possible at cme.aerztebatt.de risk or name risk factors for RFS despite correct de-
termination of serum electrolytes (5, 6).

Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 107
MEDICINE

TABLE 1 reduced food intake over at least 5 days, meeting less

than 50% of the individual patient’s calorie require-
Criteria of the National Institute for Health Care and Excellence ments. For the identification of risk patients, appli-
for the identification of patients at risk for the development of refeeding
syndrome with main and secondary criteria (10)* cation of the British National Institute for Health and
Care Excellence (NICE) criteria (9) has become
One of the following (main criteria) Two of the following (secondary established in clinical practice (Table 1).
criteria)
There are differing definitions of RFS in the litera-
BMI < 16 kg/m2 BMI < 18.5 kg/m2
ture, and reliable findings on the incidence of RFS are
Unintentional weight reduction of > 15% Unintentional weight reduction of > 10% lacking. This is highlighted by the results of a recent
within 3–6 months within 3–6 months
meta-analysis of 35 observational studies (10). Mal-
Little or no food intake for > 10 days Little or no food intake for > 5 days nourished individuals are at risk of developing RFS.
Low to normal-low electrolyte levels Positive history of alcohol abuse, According to the 2019 Nutrition Report of the
for phosphate, potassium, and/or treatment with diuretics, chemotherapeutic German Nutrition Society (Deutsche Gesellschaft für
magnesium prior to refeeding agents, or antacid agents Ernährung, DGE), 35% of patients in German hospit-
als are affected by malnutrition (15% to a moderate
* An increased risk for the development of refeeding syndrome is present if at least one main criterion or two
secondary criteria are met. and 20% to a severe extent) (11). In the acute medical
BMI, body mass index inpatient setting, multicenter randomized studies
show an RFS incidence of 8–14.6% in the feeding of
malnourished patients (12, 13).
The clinical relevance of RFS is high: If unrecog- Diseases associated with malnutrition increase the
nized and untreated, fatal treatment outcomes are risk for RFS. These include consumptive diseases (for
possible (7, 8). Treatment includes replacement ther- example, cancer, tuberculosis, HIV), malabsorption
apy of the affected electrolytes and other vitamins (in and malassimilation syndromes (chronic inflamma-
particular, vitamin B1), as well as the reintroduction tory bowel diseases, radiation enteritis, etc.), and
of feeding in a calorie-reduced manner. Against the psychiatric disorders (alcohol addiction, anorexia
background of the diagnostic challenges posed by nervosa, etc.). Decreased food intake over several
RFS, a clinical decision support system (CDSS) for days (for example, due to multiple surgical proce-
the automated diagnosis of the syndrome was newly dures, loss of appetite) or chronic nutrient deficiency
developed and tested at the University of Leipzig following bariatric surgery can also lead to RFS (2,
Medical Center, Germany. 14, 15). The risk profile of oncology patients for RFS
is characterized by weight loss, loss of appetite,
Methods persistent inflammation, and multiple medical inter-
A selective literature search on RFS was carried out in ventions and is thus particularly unfavorable (16).
the PubMed database on the basis of current treatment The fact that chemotherapy is included in the NICE
recommendations, meta-analyses, systematic review criteria takes this risk profile into account (Table 1).
articles, randomized controlled studies, cohort studies, According to current knowledge, the metabolic
and observational studies. Publications in English and switch from catabolic to anabolic is relevant from a
German from between 1990 and 2022 were taken into pathophysiological perspective (more on pathophysi-
consideration. Particular focus was placed on publi- ology in Figure 1).
cations relating to RFS in conjunction with CDSS. The
following search terms were used: “refeeding Evidence from randomized controlled studies
syndrome” [AND] “diagnosis,” “recognition,” For over 50 years, case reports have been published
“incidence,” “management,” “prospective” “clinical describing the severe courses of RFS, such as cardiac
decision support system,” and “CDSS.” Practical arrhythmias, delirium, and death, in a number of differ-
experience with the use of our own CDSS for RFS ent patient populations (17). A large proportion of
diagnosis in inpatient care was additionally taken into scientific investigations into RFS are observational
consideration. studies. Prospective randomized controlled studies on
RFS and its treatment, especially in non-anorexic pa-
Definition and epidemiology tients, are rare (14). In German-speaking countries,
There is no standardized definition of RFS still today. there is only one secondary analysis of a multicenter
In April 2020, the American Society for Parenteral and randomized controlled study investigating the effect on
Enteral Nutrition (ASPEN) published consensus rec- malnourished medical inpatients of individualized nu-
ommendations on RFS, thereby offering for the first tritional therapy compared to standard hospital food
time standardized diagnostic criteria for future studies. (18, 19). The secondary analysis of this study on RFS
RFS is defined as falls in serum levels of phosphate, was a planned part of the study design, screened the lar-
potassium, and/or magnesium by 10–20% (mild), gest sample to date for the occurrence of RFS
20–30% (moderate), or > 30% (severe), with or without (n = 967), and analyzed its clinical outcomes. The sec-
organ damage, within 5 days of resumption of ondary analysis compares malnourished patients under
previously strongly decreased caloric intake (2). A individualized nutritional therapy that did not develop
significantly decreased caloric intake is defined as a RFS (n = 826) with malnourished patients that did

108 Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14
 MEDICINE

FIGURE 1

Anabolism Catabolism Anabolism

Food intake*1
kcal/Day

Vitamin
B1

–5 –1 0 1 5 10 15 Days
supply mechanism

→ Transport of magnesium
Principal energy

into cells by insulin*2

Glycolysis from CHO from food Glycogen Fatty acid oxidation → Ketone bodies Glycolysis from CHO from food

→ Transport of potassium
into cells by insulin
Insulin

Total amount in the organism

Phosphate

Serum concentration
Requirements

Pathophysiology of refeeding syndroms (RFS), modified from Nguyen et al (40). In catabolism, energy production takes place via glyco-
genolysis and gluconeogenesis. Liver glycogen is generally used up after 12–24 h. Thereafter, primarily ketones from fatty acid oxidation serve
as substrates of gluconeogenesis. Intracellular reserves of phosphate, potassium, and magnesium are used to maintain electrolyte homeostasis
and become depleted in the absence of food intake. Serum electrolyte concentrations within the reference range mask a growing intracellular
deficit. Upon reintroduction of feeding, insulin is secreted, thereby stimulating sodium–potassium ATPase and enabling the intracellular influx of
glucose and phosphate. Thus, carbohydrates are once again available as substrates for glycolysis. Carbohydrate metabolism requires mag-
nesium and thiamine (vitamin B1) as essential cofactors. The half-life of physiological vitamin B1 storage is 7–10 days and is quickly exhausted
in the case of low food intake. The onset of glycolysis triggers a rapidly increasing requirement for vitamin B1 and electrolytes. Acute deficiency
causes RFS complications.
*1 Balanced diet, e.g., 50% calories from carbohydrates, 30% fats, 20% protein; *2 intracellularly used for pyruvate entry into the cycle
kcal: kilocalories; CHO, carbohydrates.

develop RFS (n = 141). Patients with RFS exhibited caloric restriction (n = 163) or standard nutritional
significantly higher 180-day mortality (42/141 [29.8%] support (n = 164). The group receiving caloric restric-
versus 181/826 [21.9%], odds ratio [OR] 1.53; tion had significantly better 60-day survival (standard
p < 0.05), an increased risk for admission to intensive nutritional support: 128/164 [78%] versus caloric re-
care (6/141 [4.3%] versus 13/826 [1.6%], OR 2.71; striction 149/163 [91%]; p = 0.002) (21). In 2021,
p < 0.05), and longer hospital stay (10.5 ± 6.9 versus Olsen et al. investigated the effect of an intensive nu-
9.0 ± 6.6 days; median additional days 1.57; p = 0.01) tritional protocol of 20 kcal/kg/day, whereby energy
(12). In addition, a recent meta-analysis on mortality in intake was increased to reach the targeted caloric in-
RFS patients showed a significantly increased 6-month take within 3 days, compared to a reduced-calorie
mortality rate (OR 1.54; 95% confidence interval: protocol of 10 kcal/kg/day, whereby energy intake
[1.04; 2.28]) (20). was increased to reach targeted caloric intake within 7
The therapeutic effects of low-calorie to high- days. The 85 geriatric patients were investigated for
calorie nutritional protocols on RFS have been inves- hand grip strength, 3-month mortality, and the devel-
tigated in randomized controlled studies in intensive opment of RFS. All patients could be classified as
care patients, in the normal inpatient setting, in high-risk patients for RFS according to NICE criteria.
patients on geriatric units, as well as in patients re- Nutrition was administered enterally via nasogastric
ceiving treatment for anorexia nervosa. These studies tube feeding. Patients receiving intensive nutrition
yielded partially conflicting results. were more likely to develop RFS (group receiving
A multicenter randomized study conducted by 20 kcal/kg/day 17.1% versus group receiving 10 kcal/
Doig et al. investigated 60-day survival in intensive kg/day 9.3%; p = 0.29). The occurrence of RFS was
care patients that developed RFS 72 h following com- not statistically significant, but patients were signifi-
mencement of nutritional support and received either cantly more likely to experience respiratory distress

Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 109
MEDICINE

FIGURE 2

Before feeding Refeeding in at-risk patients High-risk patients

Initial ● 10–20 kcal/kg/Day or High-risk patients: consider 5 kcal/kg/day

caloric intake ● 100–150 g Dextrose
Increasing ● Every 1–2 days by 33% of the targeted If the level of electrolytes is low, a slower
caloric intake calories while monitoring electrolyte levels increase in calorie intake or an increase
● If electrolyte level falls, consider reducing only after electrolyte replacement may be
caloric intake by 50% necessary.
● If the level of electrolytes is low, a slower
increase in caloric intake or an increase
only after electrolyte replacement may be
necessary.
Vitamin B1 Intravenous or oral adminis- If there are signs of vitamin B1 deficiency
tration of 100 mg vitamin B1 (e.g., alcohol abuse, prolonged food
abstention), replacement for 5–7 days
Mulitvitamins ● Parenteral nutrition: daily supplement for
the duration of feeding
● Oral/enteral nutrition: daily administration
for up to 10 days or more is possible,
depending on clinical and nutrition status
Electrolytes Measure of ● Replacement according to serum level Depending on clinical status and
– K K, ● Depending on clinical status and electrolyte levels, control measurements
– Mg Mg, electrolyte levels, control measurements at least every 12 h on days 1–3
– PO4 PO4 every 12 h
Monitoring ● Vital signs every 4 h in the first 24 h Cardiovascular monitoring and
intensive care for unstable individuals

Daily: weight, fluid balance, vital signs, cardiopulmonary status

kcal/Day

–5 0 1 5 10

Recommendations on refeeding in patients at risk of refeeding syndrome and treatment recommendations in manifest refeeding syndrome according to the consensus
recommendations (2) of the American Society for Parenteral and Enteral Nutrition
P, potassium; Mg, magnesium; PO4, phosphate

(group receiving 20 kcal/kg/day 53.6% versus group included in the final data analysis (24). This resulted
receiving 10 kcal/kg/day 30.2%, p = 0.029). There in small samples in the study arms (for example, mod-
were no differences in hand grip strength or mortality erate RFS risk and high-calorie feeding with 30 kcal/
(22). Lower-calorie refeeding had a favorable effect kg/day = 10). It was unlikely that an effect would be
on morbidity and mortality in the investigations car- demonstrated with this reduced sample size.
ried out by Doig et al. and Olsen et al. (21, 23). Investigations conducted in the US on patients with
A UK interventional study conducted at two anorexia nervosa show contrasting results for caloric
centers investigated the effect of different refeeding restriction as a therapeutic intervention in RFS. A
rates with a newly prescribed PN on the occurrence of multicenter study investigated the effect of oral
electrolyte disturbances, ECG changes, infections, refeeding on the medical stability of hospitalized pa-
and hospital stay in patients at risk for RFS. Within tients receiving high-calorie (n = 60, starting at
the first 48 h, patients at risk for RFS received low- 2000 kcal/day, increasing daily by 200 kcal) com-
calorie (15 kcal/kg/day) or high-calorie (30 kcal/kg/ pared to low-calorie refeeding (n = 51, starting at
day) PN, followed by a standard PN regimen of 1400 kcal/day, increasing daily by 200 kcal). High-
30 kcal/kg/day in line with standard practice for the calorie refeeding resulted in faster medical stability
two centers. The different refeeding rates had no during hospitalization (hazard ratio 1.67 [1.10; 2.53];
effect on the endpoints (24). However, the study p = 0.01) (25). Medical stability was determined by
protocol was significantly hampered by problems heart rate (HR) > 45/min, systolic RR > 90 mmHg,
during the inclusion of patients. Instead of the body temperature > 35.6° C, an intact orthostatic
planned 225 patients, ultimately only 48 were response (maximum increase in HR ≤ 35 and

110 Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14
 MEDICINE

TABLE 2

Characteristics of patients with confirmed refeeding syndrome (RFS) in the test phase of a clinical decision support system (CDSS).
All individuals in whom the CDSS resulted in a diagnosis are shown above the line. Those in whom RFS was already known are shown
in the shaded portion

Sex, age Main diagnosis BMI Nutrition Electrolytes [mmol/L] NRS NICE RFS
[kg/m2] criteria severity*
PO4 P Mg
F, 84 Cardiogenic shock in AV block III 23.5 ON 0.41 3.21 0.52 2 Positive Severe
M, 78 NSCLC, pneumonia 20.3 PN 0.59 4.18 0.79 5 Positive Mild
F, 65 Hypopharyngeal cancer 11.9 PN 0.3 2.7 0.63 5 Positive Severe
M, 74 COVID-19 16.4 ON+LD 0.5 2.98 0.76 6 Positive Moderate
F, 79 Exsiccosis, flu 30.8 ON 0.58 3.1 n.a. 3 Negative Moderate
F, 56 Breast cancer 27.1 ON 0.55 2.93 0.76 3 Positive Moderate
F, 84 Staphylococcus aureus sepsis 20.5 ON+LD 0.53 3.06 0.74 n.a. Positive Moderate
M, 44 Good syndrome 17.4 PN 0.44 2.21 0.86 3 Positive Severe
F, 54 Anastomosis after Roux-en-Y bypass 24.7 PN 0.56 3.12 0.7 3 Positive Moderate
M, 86 Sigmoid diverticulitis 21.6 PN 0.31 5.13 0.55 4 Positive Moderate
M, 61 DLBCL, diarrhea 23.5 ON+LD 0.48 2.16 0.71 4 Negative Severe
M, 64 Liver cirrhosis, malnutrition 23.9 ON 0.15 2.7 0.7 2 Negative Moderate
M, 55 Depression 18.6 ON+LD 0.25 2.94 0.66 4 Positive Moderate
F, 68 Depression 19.5 ON+LD 0.5 2.96 0.85 5 Positive Moderate
M, 57 Poisoning (alcohol) 20.7 ON 0.43 2.33 n.a. 4 Positive Severe
F, 59 Clostridium difficile enteritis 22.6 PN 0.34 3.1 0.45 2 Negative Severe
F, 69 COVID-19, DLBCL 27.5 ON 0.5 3.08 0.77 5 Positive Moderate
M, 66 COVID-19 21.6 ON 0.53 3.09 n.a. 6 Negative Moderate
M, 47 Poisoning (alcohol) 24.7 ON 0.2 3.4 0.56 3 Positive Moderate
F, 72 Urinary tract infection 19.2 ON 0.59 3.56 n.a. 4 Negative Mild
F, 46 Hypokalemia, alcohol abuse 19.5 ON 0.36 3.08 0.48 3 Positive Severe

* Classification into severity levels is in accordance with the RFS diagnostic criteria of the American Society for Parenteral and Enteral Nutrition (ASPEN).
AV, atrioventricular; BMI, body mass index; DLBCL, diffuse large B cell lymphoma; P, potassium; M, male; Mg, magnesium; NICE, National Institute for Health and Care Excellence;
NRS, Nutritional Risk Score 2002; NSCLC, non-small-cell lung cancer; n.a., not available; ON, oral nutrition; PN; parenteral nutrition; PO4, phosphate; LD, liquid diet; F, female

maximum decrease in systolic RR 20 mmHg), as well Treatment and monitoring

as reaching at least 75% of an age- and sex-adjusted A number of different recommendations for the preven-
median BMI. RFS did not develop more frequently in tion and treatment of RFS have been published in Eu-
the high-calorie group than it did in the low-calorie rope and the US (9, 27, 28). The most recent treatment
group (four versus three individuals) (25). The stable recommendations on refeeding in patients at risk for
long-term study effects that were published in 2022 RFS are the consensus recommendations of the Ameri-
confirm the robustness of these results (26). can Society for Parenteral and Enteral Nutrition
Despite the conflicting results from interven- (ASPEN) published in 2020 (2). These recommend
tional studies, caloric restriction is the therapeutic extensive replacement of the required electrolytes and
feeding intervention recommended in RFS. The re- vitamins (in particular vitamin B1) and a stepwise
liability of the UK interventional study is reduced approach to refeeding with a reduced caloric intake
by its diminished sample size. It is unclear whether (Figure 2) (2, 9, 27–29). The primary goal of a stepwise
the results from anorexia nervosa patients can be reintroduction of nutrition is to prevent RFS, much like
extrapolated to other patients. The mean age of the prophylaxis. If manifest RFS develops, the further nu-
111 individuals included in the abovementioned trition and replacement therapy follows the recommen-
study was 16.4 ± 2.5 years (mean ± standard devi- dations on the refeeding of high-risk patients (Fig-
ation [SD]) (25). One can assume that the organism ure42). Due to the above-mentioned conflicting results
of a 16-year-old patient can better tolerate faster re- from interventional studies, the level of evidence for
feeding than can a multimorbid patient of more ad- the recommendations on caloric restriction is based on
vanced age. expert consensus. The increase in caloric intake is

Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 111
MEDICINE

determined by the risk for RFS, the patient’s clinical telephone survey of the treating physicians, 13 (62%)
status during refeeding, and the intensity of occurring cases would not have been recognized without a CDSS
electrolyte disorders, in particular phosphate. The notification (Table 2, patients above the cut-off line).
severity of electrolyte disturbances determines the fre- Determination of the necessary serum electrolytes
quency of electrolyte monitoring and simultaneously alone failed to establish the diagnosis in a relevant pro-
serves as a decision-making criterion for increases in portion of patients. Thus, an additional assessment by a
caloric intake. In the intensive care setting, monitoring CDSS can improve diagnosis, enabling affected indi-
should be adapted to the individual needs of the patient. viduals to receive nutritional medical support. Due to
The initial caloric amount per kilogram body weight is the low case numbers in the retrospective development
5–20 kcal/kg/day, depending on the specialist society, phase (100 patients in 30 months, eFigure), neither ran-
whereby the following applies: the higher the risk of domization nor comparison to a control group could be
RFS, the lower the initial caloric amount per kilogram performed in the prospective testing (6 months).
bodyweight (2, 9, 27). The ASPEN recommends an in- The greatest limitation of the CDSS is the detection
itial caloric intake of 10–20 kcal/kg/day. There is no of hypophosphatemia as the initiator of the subse-
recommendation for a restriction of salt, fluid, or pro- quent automated electrolyte analysis. Phosphate de-
tein intake. Usually, the targeted caloric intake can be terminations are not routine in clinical practice: Only
reached within 7 days. 18% of investigated individuals (n = 13,325, eFigure)
had this checked in the retrospective CDSS develop-
Diagnostic challenges and the potential ment phase. Hypophosphatemia was found in
offered by clinical decision support systems approximately a third of these (n = 4186, eFigure).
As a result of the nonspecific symptoms of RFS, phy- The CDSS is unable to identify patients with RFS
sicians often fail to recognize this clinical picture. If the without phosphate determination. Thus, it is not
risk for the development or manifestation of RFS goes possible to determine the proportion of patients
unrecognized, it is not possible to initiate clinical steps. recognized as false-negative by the CDSS. The elec-
This results in vulnerable patients being put at risk (8, trolyte disturbances in RFS show a frequency peak 72
30). Due to its characteristic electrolyte disturbances, h after nutrition therapy (12, 39). The use of the RFS-
RFS lends itself to the development of an automated CDSS in combination with a standardized deter-
diagnostic algorithm with a notification system. Auto- mination of serum phosphate 72 h following the initi-
mated diagnostic suggestions, complemented by treat- ation of feeding, at least in malnourished patients,
ment recommendations, can be represented in a CDSS. would be a cost-effective approach to screening (lab-
To date, CDSS are poorly established in clinical nutri- oratory costs according to the German Uniform
tional medicine. However, they have already been able Evaluation Standard [einheitlicher Bewertungsmaß-
to achieve improved compliance with protein and cal- stab, EBM] 32 086: 0.40 Euros) to improve the effec-
orie targets (31) and improved glycemic control in tiveness of the CDSS. It is also necessary that
clinical nutritional medicine (32–34). As part of the physicians caring for people at increased risk for the
AMPEL research project (35) (analysis and notification development of RFS educate themselves about this
system to improve patient safety through real-time inte- clinical picture. This would enable at-risk patients to
gration of laboratory findings, www.ampel.care), a be promptly identified and correspondingly treated
CDSS of this kind has been newly developed for RFS prior to the initiation of refeeding. Only then is it
at the University Medical Center Leipzig (UMCL). It possible to prevent a hazardous drop in serum electro-
was prospectively tested over a 6-month period (for in- lyte levels and the development of RFS.
clusion and exclusion criteria as well as CDSS develop-
ment, see the eMethods Section). The CDSS checks Conclusion
adult patients with hypophosphatemia (< 0.84 mmol/L) RFS receives insufficient clinical attention. It can occur
as to whether other states of electrolyte deficiency in in malnourished individuals across the entire spectrum
relation to potassium and magnesium have been present of medicine and is characterized by nonspecific symp-
in the preceding 5 days. In the case of electrolyte imbal- toms that hamper diagnosis. In severe cases, RFS can
ances suspicious for RFS, the CDSS then checks exclu- cause death. The ability to recognize this clinical pic-
sion criteria using laboratory values and coding data ture and identify at-risk patients is relevant to all those
pointing to the presence of procedures and comorbid- treating malnourished patients. The use of CDSS can
ities (for example, terminal kidney failure with dialy- assist clinically active physicians in establishing its di-
sis) that cause hypophosphatemia of other etiology agnosis. Evidence on the prevention and treatment of
(36–38). In the 6-month exploratory prospective test RFS is sparse. Current treatment recommendations
phase, the CDSS was able to identify 21 individuals based on expert consensus include replacement therapy
with suspected RFS (patient characteristics, Table 2). of the necessary electrolytes and other vitamins (in par-
Their diagnoses were confirmed in subsequent bedside ticular, vitamin B1), as well as the reintroduction of
visits. RFS severity was determined on the basis of the feeding in an initially calorie-reduced form. Prospec-
ASPEN diagnostic criteria. The distribution of tive randomized analyses are needed in the future in
principal diagnoses shows that RFS patients encounter order to systematically improve RFS treatment guide-
physicians in all specialties (Table 2). According to a lines while taking into account appropriate CDSS.

112 Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14
 MEDICINE

Acknowledgments 16. Marinella MA: Refeeding syndrome in cancer patients. Int J Clin Pract
We would like to thank Kristin Gutsmuths, Carsten Güttich, and Anna- 2008; 62: 460–5.
Sophie Junge for carrying out the study visits.
17. Boateng AA, Sriram K, Meguid MM, Crook M: Refeeding syndrome:
Funding treatment considerations based on collective analysis of literature
This initiative (eHealthSax guideline No.: 100331796) is co-financed by case reports. Nutrition 2010; 26: 156–67.
public funds based on the budget passed by members of the Saxon State
Parliament. 18. Schuetz P, Fehr R, Baechli V, et al.: Individualised nutritional support
in medical inpatients at nutritional risk: a randomised clinical trial.
Conflict of interest statement Lancet 2019; 393: 2312–21.
Dr. Heuft, PD Dr. Kaiser, and Dr. Voigt were involved in the development of
the CDSS, the clinical implementation of which is reported in this article. 19. Schuetz P, Fehr R, Baechli V, et al.: Design and rationale of the effect
of early nutritional therapy on frailty, functional outcomes and recovery
Other research partners in this project publicly funded under the eHealth- of malnourished medical inpatients trial (EFFORT): a pragmatic,
Sax guideline No.: 100331796 include the Muldental Clinics in Grimma multicenter, randomized-controlled trial. Int J Clin Trials 2018; 5: 142.
and Wurzen as well as Xantas AG. Upon completion of the project, the
option exists to transform the CDSS into a product. All research partners 20. Bioletto F, Pellegrini M, Ponzo V, et al.: Impact of refeeding syndrome
have independent and autonomous access to the knowledge gained. on short- and medium-term all-cause mortality: a systematic review
The remaining authors declare that no conflict of interest exists. and meta-analysis. Am J Med 2021; 134: 1009–18.e1.
21. Doig GS, Simpson F, Heighes PT, et al.: Restricted versus continued
Manuscript received on 31 May 2022, revised version accepted on
15 November 2022. standard caloric intake during the management of refeeding syndrome
in critically ill adults: a randomised, parallel-group, multicentre, single-
Translated from the original German by Christine Rye. blind controlled trial. Lancet Respir Med 2015; 3: 943–52.
22. Olsen SU, Hesseberg K, Aas A-M, Pripp AH, Ranhoff AH, Bye A: A
References
comparison of two different refeeding protocols and its effect on hand
1. Zeki S, Culkin A, Gabe SM, Nightingale JM: Refeeding grip strength and refeeding syndrome: a randomized controlled clinical
hypophosphataemia is more common in enteral than parenteral trial. Eur Geriatr Med 2021; 12: 1201–12.
feeding in adult in patients. Clin Nutr 2011; 30: 365–8.
23. Olthof LE, Koekkoek WACK, van Setten C, Kars JCN, van Blokland D,
2. da Silva JSV, Seres DS, Sabino K, et al.: ASPEN Consensus van Zanten ARH: Impact of caloric intake in critically ill patients with,
recommendations for refeeding syndrome. Nutr Clin Pract 2020; 35: and without, refeeding syndrome: a retrospective study. Clin Nutr
178–95. 2018; 37: 1609–17.
3. Ponzo V, Pellegrini M, Cioffi I, Scaglione L, Bo S: The refeeding 24. Ambrose T, de Silva A, Naghibi M, et al.: Refeeding risks in patients
syndrome: a neglected but potentially serious condition for inpatients. requiring intravenous nutrition support: results of a two-centre,
A narrative review. Intern Emerg Med 2021; 16: 49–60. prospective, double-blind, randomised controlled trial. Clin Nutr
4. Janssen G, Pourhassan M, Lenzen-Großimlinghaus R, et al.: The ESPEN 2021; 41: 143–52.
refeeding syndrome revisited: you can only diagnose what you know. 25. Garber AK, Cheng J, Accurso EC, et al.: Short-term outcomes of the
Eur J Clin Nutr 2019; 73: 1458–63. study of refeeding to optimize inpatient gains for patients with anorexia
5. Larsen S: The investigation of parenteral nutrition – Aotearoa (IPNA) nervosa: a multicenter randomized clinical trial. JAMA Pediatr 2021;
setting up the 1st phase of a clinical audit of the delivery of parenteral 175: 19–27.
nutrition (PN) in New Zealand (NZ). Masterarbeit. Albany, New 26. Golden NH, Cheng J, Kapphahn CJ, et al.: Higher-calorie refeeding in
Zealand 2012. anorexia nervosa: 1-year outcomes from a randomized controlled trial.
6. Dyson JK, Thompson N: Adult parenteral nutrition in the North of Pediatrics 2021; 147: e2020037135.
England: a region-wide audit. BMJ Open 2017; 7: e012663.
27. Boland K, Solanki D, O’Hanlon C (ed.): Prevention and treatment of
7. Vignaud M, Constantin JM, Ruivard M, et al.: Refeeding syndrome refeeding syndrome in the acute care setting: Irish Society for Clinical
influences outcome of anorexia nervosa patients in intensive care unit: Nutrition & Metabolism 2013. IrSPEN Guideline Document No. 1.
an observational study. Crit Care 2010; 14: R172.
28. Stanga Z, Brunner A, Leuenberger M, et al.: Nutrition in clinical
8. Koekkoek WAC, van Zanten ARH: Is refeeding syndrome relevant for practice—the refeeding syndrome: illustrative cases and guidelines for
critically ill patients? Curr Opin Clin Nutr Metab Care 2018; 21: 130–7. prevention and treatment. Eur J Clin Nutr 2008; 62: 687–94.
9. National Institute for Health and Care Excellence (NICE) (ed.): 29. Aubry E, Aeberhard C, Leuenberger M, et al.: Refeeding-Syndrom:
Nutrition support for adults: oral nutrition support, enteral tube feeding Ein konsensusbasierter Algorithmus für stationäre Patienten. Aktuel
and parenteral nutrition 2006. Ernahrungsmed 2019; 44: 33–42.
10. Cioffi I, Ponzo V, Pellegrini M, et al.: The incidence of the refeeding 30. Kagansky N, Levy S, Koren-Morag N, Berger D, Knobler H:
syndrome. A systematic review and meta-analyses of literature. Clin Hypophosphataemia in old patients is associated with the refeeding
Nutr 2021; 40: 3688–701. syndrome and reduced survival. J Intern Med 2005; 257: 461–8.
11. Volkert D, Weber J, Kiesswetter E, Sulz I, Hiesmayr M: Ernährungs- 31. Ettori F, Henin A, Zemmour C, et al.: Impact of a computer-assisted
situation in Krankenhäusern und Pflegeheimen – Auswertung der decision support system (CDSS) on nutrition management in critically
nutritionDay-Daten für Deutschland. 14 DGE Ernährungsbericht 2019; ill hematology patients: the NUTCHOCO study (nutritional care in
V2–V669. hematology oncologic patients and critical outcome). Ann Intensive
12. Friedli N, Baumann J, Hummel R, et al.: Refeeding syndrome is Care 2019; 9: 53.
associated with increased mortality in malnourished medical 32. Benyo B, Illyés A, Némedi NS, et al.: Pilot study of the SPRINT
inpatients: secondary analysis of a randomized trial. Medicine glycemic control protocol in a Hungarian medical intensive care unit. J
(Baltimore) 2020; 99: e18506. Diabetes Sci Technol 2012; 6: 1464–77.
13. Kraaijenbrink B, Lambers WM, Mathus-Vliegen E, Siegert C: 33. Pielmeier U, Rousing ML, Andreassen S, Nielsen BS, Haure P:
Incidence of refeeding syndrome in internal medicine patients. Neth J Decision support for optimized blood glucose control and nutrition in a
Med 2016; 74: 116–21. neurotrauma intensive care unit: preliminary results of clinical advice
14. Friedli N, Stanga Z, Sobotka L, et al.: Revisiting the refeeding and prediction accuracy of the Glucosafe system. J Clin Monit Comput
Syndrome: Results of a systematic review. Nutrition 2017; 35: 151–60. 2012; 26: 319–28.
15. Wirth R, Diekmann R, Janssen G, et al.: Refeeding-Syndrom. 34. Kwan JL, Lo L, Ferguson J, et al.: Computerised clinical decision
Pathophysiologie, Risikofaktoren, Prophylaxe und Therapie. Internist support systems and absolute improvements in care: meta-analysis of
2018; 59: 326–33. controlled clinical trials. BMJ 2020; 370: m3216.

Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 113
MEDICINE

35. Walter Costa MB, Wernsdorfer M, Kehrer A, et al.: The clinical Corresponding author
decision support system AMPEL for laboratory diagnostics: Dr. med. Lara Heuft
implementation and technical evaluation. JMIR Med Inform 2021; 9: Institut für Humangenetik
e20407. Universitätsklinikum Leipzig
36. Datta HK, Malik M, Neely RDG: Hepatic surgery-related Philipp-Rosenthal-Straße 55, 04103 Leipzig, Germany
hypophosphatemia. Clinica Chimica Acta 2007; 380: 13–23. [email protected]
37. Leung J, Crook M: Disorders of phosphate metabolism. J Clin Pathol
2019; 72: 741–7. Cite this as
38. Varudo R, Mota AM, Pereira E, Dias C: Impact of phosphatemia Heuft L, Voigt J, Selig L, Stumvoll M, Schlögl H, Kaiser T:
variability in neurological outcomes in patients with spontaneous Refeeding syndrome—diagnostic challenges and the potential of clinical
subarachnoid hemorrhage. Cureus 2021; 13: e18257. decision support systems. Dtsch Arztebl Int 2023; 120: 107–14.
39. Friedli N, Stanga Z, Culkin A, et al.: Management and prevention of DOI: 10.3238/arztebl.m2022.0381
refeeding syndrome in medical inpatients: an evidence-based and
consensus-supported algorithm. Nutrition 2018; 47: 13–20. ►Supplementary material
40. Nguyen P, Schlögl H, Selig L, Baerwald C: Refeeding-Syndrom. Pa-
thophysiologie, Therapie und welche rheumatologischen Patienten be- eMethods, eFigure:
sonders gefährdet sind. Z Rheumatol 2021; 80: 263–9. www.aerzteblatt-international.de/m2022.0381

CLINICAL SNAPSHOT
Acute Vocal Fold Hemorrhage While Singing
A 40-year-old female opera singer presented
to the phoniatric department due to hoarse-
ness resulting from vocal strain. The soprano
described intermittent functional impairment to
her singing voice (dysodia) since changing
from lyric to dramatic voice type (fach) a few
months previously. Transoral videolaryngo-
stroboscopy revealed a marginal edema of the
a b
right vocal fold (arrow) with supramarginal
varices (vocal cord varix; asterisk). In contrast to other varices, such as for example in the esophagus, these are caused by strain rather than con-
gestion. During the examination, subepithelial hemorrhage occurred in the region of the middle third of the vocal fold (arrowhead) while the patient
used her singing voice. After several weeks of vocal rest and hematoma resorption, phonosurgical microlaryngoscopic resection of the residual
vocal fold edema and vocal fold varices was performed. Postoperative follow-up showed functional recovery of phonatory vocal fold mobility with
complete glottic closure and restored performance. Given the low incidence of vocal fold varices, no cases of variceal hemorrhage directly during
transoral videolaryngostroboscopy have been described in the literature to date. The present case illustrates that vocal fold varices can lead to
recurrent organic voice dysfunction due to strain-related tissue hemorrhage, particularly in elite vocal performers.
Prof. Dr. med. Philipp P. Caffier, Prof. Dr. med. Tadeus Nawka, Klinik für Audiologie und Phoniatrie, Charité – Universitätsmedizin Berlin, [email protected]

Dr. med. Sebastian Weikert, Forum HNO, Gemeinschaftspraxis für HNO und Phoniatrie, Regensburg
Conflict of interest statement: The authors state that no conflict of interest exists.
Translated from the original German by Christine Rye.
Cite this as: Caffier PP, Weikert S, Nawka T: Acute vocal fold hemorrhage while singing. Dtsch Arztebl Int 2023; 120: 114. DOI: 10.3238/arztebl.m2022.0356

114 Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14
MEDICINE

Supplementary material to:

Refeeding Syndrome
Diagnostic Challenges and the Potential of Clinical Decision Support Systems
by Lara Heuft*, Jenny Voigt*, Lars Selig, Michael Stumvoll, Haiko Schlögl**, and Thorsten Kaiser**
Dtsch Arztebl Int 2023; 120: 107–14. DOI: 10.3238/arztebl.m2022.0381

eMETHODS

Development of the clinical decision support system reference range (0.84 mmol/L) and performs, in a first step, an
A diagnostic algorithm was drawn up on the basis of the diag- automated verification of exclusion criteria (eFigure). Exclusion
nostic criteria published in 2020 by the American Society for criteria are verified by the presence of laboratory values and
Parenteral and Enteral Nutrition (ASPEN). The description of the coding data in the electronic medical record suggestive of
development of the algorithm, the final algorithm, the design of hypophosphatemia of other etiology and are designed to prevent
the clinical decision support system (CDSS), and its implemen- incorrect notifications. The detection of procedure codes for
tation in clinical workflow are currently in the process of being acute dialysis treatments and surgical procedures on the liver and
published, meaning that the algorithm used is currently not pub- brain results in exclusion, as does the detection of certain labora-
licly available. A description of the content of the diagnostic tory parameters such as ketone bodies in urine (suggestive of
steps is given in the section “Process of automated patient evalu- decompensated diabetes mellitus), paracetamol in serum
ation.” The algorithm was retrospectively tested using a dataset (paracetamol poisoning), hyperphosphatemia and hyper-
from laboratory data from January 2019 to June 2021 and further magnesemia (suggestive of severe acute or chronic renal
developed for adults. The dataset contained all serum levels and dysfunction), an elevated PTH level (hyperparathyroidism), and
blood gas analyses as well as point-of-care testing of whole secretion of PTH-related peptides (lung and neuroendocrine car-
blood concentrations of the electrolytes phosphate, magnesium, cinomas). Electrolyte imbalances triggered by these diseases and
and potassium at the University Medical Center Leipzig interventions and which are unrelated to nutrition but resemble
(UMCL), Germany. The algorithm thus developed was then in- RFS have been described in the literature (35–38). For included
corporated in a CDSS together with action and treatment recom- patients, further electrolyte assessments of magnesium and
mendations according to the severity of refeeding syndrome potassium are performed for up to a total of 5 days in the past.
(RFS). The recommendations were displayed in the electronic Depending on the percentage decrease in concentration, an alarm
medical records of identified patients via warning icons. is triggered for mild, moderate, or severe RFS. Individuals
Physicians were able to view the treatment recommendations by identified by the CDSS are flagged up with a warning icon in the
clicking on the warning icon in a pop-up window. The CDSS was digital inpatient unit overview. In moderate RFS, an electrolyte
prospectively tested over a 6-month test phase, and the suspected imbalance alone results in a patient being flagged up, while in
diagnosis expressed in an automated manner could be confirmed mild RFS, decreased energy intake additionally needs to have
by a nutrition medicine specialist or nutrition therapist in a been recorded in the coding data at the time of admission to the
bedside visit of individuals detected by the system. The data unit. By clicking on the warning notification, a pop-up window
collection periods were determined by the availability of data; opens showing critical laboratory results together with an
electrolyte data were retrospectively available for the period interpretation text module for clinical use and a link to in-house
from January 2019 to June 2021. Prospective testing was scien- treatment guidelines. In total, 100 cases were retrospectively
tifically supervised for 6 months and terminated when the project confirmed as RFS in 130 notifiable patients based on a review of
period came to an end. The exploratory study was approved by electronic medical records by two independent physicians. In the
the ethics committee of the University of Leipzig Medical Center prospective test phase, 21 of 31 individuals for whom notifi-
(No. 214/18-ek). cations were triggered were confirmed as correct RFS cases. In-
correct false-positive notifications were due to liver dysfunction
Process of automated patient evaluation (n = 3), brain trauma (n = 2), dialysis with coding pending
All adult patients that had undergone blood collection and phos- (n = 3), or multiorgan failure (n = 2). These confounders in elec-
phate determination were included in the patient evaluation. The trolyte interpretation in the case of suspected RFS are already
CDSS is initiated by a phosphate determination below the lower known from the literature (36–38).

-- Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 | Supplementary material
 MEDICINE

eFIGURE 1

Adult inpatients that underwent blood collection:

retrospective n = 73 616; prospective n = 15 727

Without phosphate determination:

n = 60 291; n = 12 866

Patients with phosphate determination:

n = 13 325; n = 2861

Without hypophosphatemia:
n = 9139; n = 1986

Patients with hypophosphatemia (< 0.84 mmol/L)

n = 4186; n = 875

Exclusion of other etiologies

(laboratory parameters):
n = 1450; n = 269

Exclusion of causal procedures:

– Dialysis procedures: n = 505; n = 118
– Liver surgery: n = 19; n= 3
– Brain surgery: n = 168; n = 42

Patients included in the CDSS evaluation

n = 2044 / n = 443

Diagnostic criteria not met

n = 1914/n = 412

Patients with suspected refeeding syndrome

n = 130 (100 TP)/n = 31 (21 TP)

Flow diagram showing the data analysis of retrospective data

from January 2019 to June 2021 (left) and the prospective test phase
(right) of the clinical decision support system (CDSS) using inclusion
and exclusion criteria and the subsequent automated evaluation of
decreases in the level of electrolytes (phosphate, potassium,
magnesium) in refeeding syndrome. TP, true positive

Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 | Supplementary material --
MEDICINE

Questions on the article in issue 7/2023:

Refeeding Syndrome—Diagnostic Challenges and the Potential cme plus
of Clinical Decision Support Systems
The submission deadline is 16 February 2024. Only one answer is possible per question.
Please select the answer that is most appropriate.

Question 1 Question 6
Which of the following criteria is one of the In patients at risk for the development of refeeding syn-
main criteria for the identification of at-risk drome, the reintroduction of nutrition following a catabolic
patients according to the National Institute for phase should be started cautiously. According to the
Health and Care Excellence? American Society for Parenteral and Enteral Nutrition, which
a) Unintentional weight loss of > 5% within 3–6 months initial caloric intake should be used?
b) Little or no food intake for > 10 days a) 10–20 kcal/kg/Day
c) Little or no food intake for > 3 days b) 70–150 kcal/kg/Day
d) Unintentional weight loss of > 10% within 3–6 months c) 200–280 kcal/kg/Day
e) BMI < 19 kg/m2 d) 300–500 kcal/kg/Day
e) 700–850 kcal/kg/Day
Question 2
Which electrolyte shift is associated with Question 7
refeeding syndrome? What is the approximate time frame within which
a) Hypokalemia and hypophosphatemia refeeding can be expected to reach the targeted
b) Hyperkalemia and hypophosphatemia caloric intake?
c) Hypernatremia and hyperphosphatemia a) 2 Days
d) Hyponatremia and hyperphosphatemia b) 1 Week
e) Hyperkalemia and hyperphosphatemia c) 2 Weeks
d) 1 Month
Question 3 e) 2 Months
Which vitamin is an essential component of
replacement therapy in refeeding syndrome? Question 8
a) Vitamin D3 What does the abbreviation CDSS stand for
b) Vitamin A in the text?
c) Vitamin E a) Clinical decompensation support system
d) Vitamin B12 b) Critical decompensation support system
e) Vitamin B1 c) Clinical delirium support system
d) Clinical decision support system
Question 4 e) Clinical diagnostic support system
According to the 2019 Nutrition Report of the
German Nutrition Society, what is the overall Question 9
percentage of patients in German hospitals Which laboratory value is named in the text as a limiting step
affected by malnutrition? in the CDSS for the recognition of refeeding syndrome and
a) 5% could be used for screening purposes in at-risk patients?
b) 15% a) Lactate
c) 25% b) Serum albumin
d) 35% c) Creatinine
e) 45% d) TSH value
e) Serum phosphate
Question 5
Which one of the following elements is transported Question 10
into cells at an increased rate when adequate Which time point is particularly suited to
nutrition is reintroduced after a phase of catabolic screening for refeeding syndrome following
energy production lasting several days? the reintroduction of nutrition?
a) Calcium a) 12 h
b) Sodium b) 24 h
c) Magnesium c) 36 h
d) Selenium d) 48 h
e) Copper e) 72 h

-- Deutsches Ärzteblatt International | Dtsch Arztebl Int 2023; 120: 107–14 | Supplementary material