Discussion
The main findings of this study are that the prevalence of patients with AN in ICUs is very low and the crude mortality is about 10%. Particularly, inappropriate nutritional support was associated with a high prevalence of refeeding syndrome. On average, patients received a total calorie intake of 22 ± 13 kcal/kg/24 h.
The recent UK NICE (National Institute for Health and Clinical Excellence) guidelines suggested that calorie repletion in AN patients should be slow, and should depend on the assessed severity of refeeding syndrome risk [
10]. For patients at high risk, the initial nutritional level should be approximately 10 kcal/kg/d, falling to as low as 5 kcal/kg/d in patients considered to be at extreme risk. A gradual increase in calorie intake, particularly during the first week of refeeding, in combination with regular biochemical and fluid balance monitoring, is important until a patient becomes metabolically stable.
Unsurprisingly, refeeding induced metabolic disorders and hepatic cytolysis in 10 to 20% of AN patients. The mean risk factors are AN
per se, the classic forms of slump, and malnutrition related to chronic disease. Only a few studies have analyzed the incidence of refeeding syndrome in the ICU. In a prospective study, serum prealbumin concentration was the only biomarker predictive of the development of refeeding syndrome [
11]. In the present study, prealbumin levels were recorded only in a third of the cases. In our retrospective study, full calorie intake was initiated on the first day of refeeding in about half of AN patients. In patients for whom calorie intake was gradually increased, physicians did not adequately appreciate the evidence of refeeding syndrome, as shown by biological abnormalities, in seven patients. In five patients, refeeding resulted in multi-organ failure and death, although nutrition was stopped. Refeeding syndrome can be defined as a potentially fatal shift in fluid and electrolyte levels that may occur in malnourished patients receiving artificial nutrition (whether enteral or parenteral) [
12]. All of oral, enteral, and parenteral feeding routes were used in our study. Most experts agree that oral refeeding is the best approach to weight restoration. In situations in which patients refuse to eat, or in patients with extreme malnutrition, feeding via a nasogastric tube may be required [
13]. If the digestive tract is functional, the enteral route is preferable to the parenteral even though parenteral nutrition can be safe and efficient [
14,
15].
As previously described [
16], the observed prevalence of pneumothorax after central venous catheterization was six percent, approximately twice that usually observed in ICU patients [
17]. To reduce the risk of this condition, we propose that an internal jugular site, and not a subclavian site, be used, with ultrasound guidance [
18]. This proposal should be tempered by the infectious complications rate reported with that site [
19].
The current recommendations for diagnostic investigation and monitoring in AN patients admitted to psychiatric and medical units may be inappropriate for ICU patients [
20]. In our study, the high incidence of cardio vascular complications, particulary hypotension and repolarization problems, suggest that electrocardiography and echocardiography should be routinely performed at the admission of AN patients. In fact, in many publications a high incidence of occult left ventricular failure and pericardial effusion was reported in such patients [
21]. In addition, improvement in cardiac function upon renutrition may be a good index of the quality of nutritional support. Metabolic disorders were the main reason for ICU admission. These disorders are the best-known metabolic complications in AN patients, and are caused by starvation or purgative practices. Profound hypoglycemia usually recurred after glucose administration, as a consequence of pathologic hyperinsulinism, and was associated with poor prognosis [
22]. Hypokalemia, hyponatremia, hypomagnesemia, and metabolic alkalosis are associated with purgative practices or diuretic abuse. Hypophosphatemia was less often reported, although this is the most common sign of refeeding syndrome. As suggested, detection and correction of hypophosphatemia should be systematic at ICU admission of AN patients and before refeeding [
23]. The second most common reason for ICU admission was nutritional support. When the body mass index is less than 12 kg/m², resting energy expenditure is only 60 to 65% of normal levels [
13]. During refeeding, this expenditure increases significantly. Thus, it is a challenge for physicians to find a compromise between low nutritional input, with the risk of insufficient weight gain, and higher nutritional input, causing refeeding syndrome. Hemodynamic and electrocardiographic disorders were also common reasons for ICU referral. Hepatic cytolysis in AN patients was reported by 20% of physicians. Several studies and case reports have highlighted increases in serum liver enzymes in patients with AN or extreme malnutrition, whether or not associated with liver failure [
24,
25]. AN, and malnutrition in general, can be linked to neurological disorders such as psychomotor slowing, memory difficulties, and disorientation, that are generally reversible after renutrition [
26]. Hematological disorders include leukoneutropenia, associated with bone marrow gelatinous degeneration macrocytic anemia, secondary to intra-erythrocytic ATP deficiency and thrombocytopenia [
27,
28]. Moreover, in patients with AN, a reduction in the contractile force of the diaphragm, and alteration in the regulation of respiratory centers, may induce respiratory failure.
Nineteen percent of patients had pneumonia and nine percent had acute respiratory distress syndrome.
In vitro studies have suggested that starvation may be associated with altered cellular and humoral immunity [
29,
30]. Immune suppression during AN may also involve abnormal responses of the complement system and hypercorticism.
AnorexieRea study group
Sophie Cayot Constantin, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Renaud Guerin, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Matthieu Jabaudon, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Christian Chartier, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Sebastien Perbet, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Antoine Petit, General ICU, Estaing Hospital, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Samir Jaber, SAR B, Saint Eloi Hospital, university Hospital of Montpellier, Montpellier, France.
Gerald Chanques, SAR B, Saint Eloi Hospital, university Hospital of Montpellier, Montpellier, France.
Philippe Verdier, General ICU, Montlucon Hospital, Montlucon, France.
Robert Chausset, General ICU, Montlucon Hospital, Montlucon, France.
Dominique Guelon, RMC, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France.
Claude Guerin, Medical ICU, La croix rousse, Lyon university Hospital, Lyon, France
Laurent Papazian, Medical ICU, APHM, Marseille, France.
Jean Paul Mira, Medical ICU, Cochin, APHP, Paris V University, France.
Bernard Blettery, Medical ICU, Dijon university Hospital, Dijon, France.
Bernard Claud, General ICU, Le Puy en velay Hospital, Le Puy en velay, France.
Jean Yves Lefrant, General ICU, Nimes University Hospital, Nimes, France.
Jean Michel Arnal, Medical ICU, Toulon Hospital, Toulon, France.
Carole Ichai, Surgical ICU, Nice University Hospital, Nice, France.
Olivier Leroy, Genera ICU, Tourcoing Hospital, Tourcoing, France.
Benoît Valet, General ICU, University hospital of Lille, Lille, France.
Olivier Pajot, General ICU, Argenteuil Hospital, Argenteuil, France.
Bernard Garrigues, General ICU, Aix en provence Hospital, Aix-en-provence Hospital, France.
Authors' contributions
MV and JMC participated in the design of the study, carried out the study and drafted the manuscript. MR, MVP, EF and JEB participated in the design of the study and data analysis. DA participated in the design of the study and helped to draft the manuscript. All authors read and approved the final manuscript.