Misconceptions about Protein Requirements for Wound Healing: Results of a Prospective Study
The importance of nutrition for wound prevention and healing is well established.1-5 Formulas such as the Harris-Benedict equation6 currently used to determine initial amounts of protein necessary utilize weight, height, age, and gender with an added “stress” factor that takes into account additional requirements for wound healing. Current approaches to address protein malnutrition involve daily intake of 1.5 to 1.8 g/Kg of protein without wound size considerations.7 The author’s experience treating patients with massive wounds has shown that these recommendations may not be sufficient to normalize protein stores.
Agreement is widespread in the literature that nutrition – especially protein level – is critical for wound healing.2,3 Protein malnutrition has been implicated as a factor in infections, length of hospitalization, and mortality.8-10 Lee’s2 recent prospective, randomized, controlled, double-blind study examined wound healing in two groups of wound patients (n = 89) – those who did or did not receive supplemental protein. The Pressure Ulcer Scale for Healing (PUSH) tool, an instrument that measures wound severity based on size, amount of drainage, and tissue type, was used to determine healing. The supplemented and non-supplemented groups received an average protein intake of 1.40g/Kg/day and 0.64g/Kg/day, respectively. Although differences in wound closure were not statistically significant, the group receiving more protein had better healing rates.
A common theme in much of the literature concerning protein repletion for wound patients is that levels of protein provision above 1.5 to 2.0 g/Kg/day are not indicated.10-13 In nearly all instances, articles and chapters subscribing to this assertion did not examine populations of patients with wounds and/or studied provision of nutrients via parenteral (ie, total parenteral nutrition [TPN]) not enteral routes. For instance, Shizgal11 studied the body composition of 75 patients diagnosed with malnutrition using multiple isotope dilution with the simultaneous intravenous injection of isotope-tagged albumin and red cells. The majority of patients had postoperative complications and all had been referred for a course of TPN. Summaries of Shizgal’s work11,12 state “there are no apparent advantages to increasing the daily protein intake to levels above 1.5 to 2.0 g/Kg body weight per day.” Similarly, Posthauer12 suggested that levels above 1.5 g/Kg may not promote further protein synthesis and may cause dehydration but again the article (by Long13) cited for this observation is a study involving TPN, not enteral feeds.12,13
Reviews of the research14-16 of the past 15 years, including a meta-analysis by Moore,14 strongly favor enteral nutrition over TPN. Kudsk et al15 studied 98 acute care trauma patients who had required laparotomy, all with surgically placed jejunostomy tubes. Patients were randomized to receive either TPN or enteral feeding. Although no significant differences in nitrogen balance were found between the groups, enterally fed patients developed significantly fewer infections. The authors postulated that enteral feeding improves gut architecture and microflora and helps the mucosa withstand challenges.
Peterson et al16 prospectively randomized 59 major abdominal trauma patients to TPN versus enteral feeds. They found that although the TPN patients received slightly higher amounts of calories and protein during the 10-day study period, serum levels of total protein, albumin, transferrin, and retinol-binding protein rose in the enterally fed patients but fell in the TPN patients. They concluded that enteral feeding improves hepatic protein synthesis more effectively than TPN after major torso injury.
In Fong et al’s17 study, six volunteers were fed enteral and six were fed parenteral formulas for 7 days. Participants were given endotoxin challenges. The authors found that the TPN patients had more occurrence of fever, higher heart rate, higher levels of circulating epinephrine, and higher C-reactive protein (CRP) and tumor necrosis factor (TNF) levels and concluded that bowel rest alters host resistance to injury independent of any previous malnutrition. In summary, enteral nutrition delivers nutrients more effectively, is less expensive, and has fewer serious complications. A review by Zaloga18 notes that bacterial translocation, a common source of sepsis, can be reduced with enteral feeds.
Fear of overzealous attempts at feeding has been fostered by descriptions of the “refeeding syndrome,” which can precipitate fatal hypophosphatemia and cause cardiac, pulmonary, and neuromuscular dysfunction. This occurrence, described by Weisner19 and others, occurs when large amounts of hypertonic solutions of dextrose and amino acids are infused as TPN. These reports, however, have involved TPN, not enteral feeds.20 Overfeeding patients with pulmonary insufficiency via enteral feeding may increase CO2 production but this is more commonly reported in patients on TPN.21,22 In fact, McClave21 found that respiratory function was maximized and the hypermetabolic response minimized by carefully matching caloric provision to metabolic requirements via indirect calorimetry and providing nutrients via enteral feeding.
Study Purpose and Design
The primary purpose of this prospective, descriptive study was to evaluate the hypothesis that many wound patients require higher levels of protein than is commonly recommended and that wound size and severity affect protein requirement. Because many tube-fed patients have wounds and estimates of actual intake are easier to obtain in this patient population, tube-fed patients with and without wounds participated in the study. The secondary purposes were to examine if more accurate, wound-based guidelines and feeding techniques/amounts could be developed for protein repletion in wound patients and to characterize reasons why attempts to normalize protein stores in tube-fed patients fail.
Because this was/is the standard feeding protocol for both groups of patients at the author’s facility (ie, no active treatment arm was created), informed consent or IRB approval was not required. Patient confidentiality was strictly maintained.
Methods and Procedures
Patient inclusion criteria. All tube-fed patients admitted to a long-term acute care (LTAC) facility from March of 2002 to May of 2003 who stayed at least 2 weeks were included for evaluation. In order to determine if the presence and characteristics of wounds increase protein requirements, consecutively admitted patients with wounds and without wounds were eligible to participate.
Patient care. Both patient groups were followed closely by a registered dietitian who monitored all nutritional study parameters prospectively and promoted a dynamic regimen designed to replete protein stores. Wound patients and non-wound patients were started at 1.25 g/Kg/day protein and 1.0 g/Kg/day protein, respectively. The standard protocol recommended by the dietitian for both groups emphasized weekly pre-albumin level checks and rapid escalation of protein provision if the pre-albumin was not normal or increasing (see Figure 1). Changes were made weekly based on weight and pre-albumin trends until patients were discharged.
Data collection. A Nutrition Flow sheet was maintained for each patient (see Figure 2) according to standard protocol for wound patients at this facility. An additional study data sheet that recorded calories and protein provided, weights, pre-albumin trends, comorbidities, and causes of feeding failure such as feeding tube complications and diarrhea was maintained for each patient. The dietitian also recorded the actual intake of protein and calories each patient received based on feeding formulation, feeding rate, and protein supplements.
To evaluate the effects of wound characteristics on protein repletion, a Certified Wound Specialist (CWS) recorded wound information, including wound size (measured in cm with a disposable paper ruler) and the PUSH score obtained on admission. For patients with more than one wound, individual wound PUSH scores (range 0 for healed wounds to 17 for most severe wounds) were summed to obtain a total PUSH score per patient as an indicator for patient wound burden as previously described.23 Patients with Stage II, Stage III, and Stage IV wounds were included in the wound group. To assess possible factors within the wound group that could best predict protein needs for the wound population, statistical analysis of the Stage III and Stage IV areas were combined and compared to the total PUSH scores to assess ability to predict protein requirements and to determine which was superior in its predictive value.
Patient confidentiality was maintained by assigning study numbers to individual patients for all analyses. Data were recorded on a three-page collection form for each patient and entered into a computerized spreadsheet (Excel, Microsoft, Seattle, Wash).
Outcomes. To evaluate protein stores and assess feeding failure or improvement, maximum levels of protein were recorded in g/Kg/day for all patients and pre-albumin levels were measured. Improvement was defined as achieving normal range pre-albumin levels or an increase of 8 units (the latter represents an approximate doubling of pre-albumin in patients whose levels improved but did not normalize). Direct complications from the feeding and other factors that interfered with the ability to deliver adequate nutrients were recorded and tabulated.
Data analysis. An independent professional statistician (Professional Statistical Services, Pasadena, Calif) completed the data analysis. The dependent variables used in this study’s statistical tests included pre-albumin levels and maximum level of protein provided per day. Both variables were measured on ratio-level scales. The study’s primary independent variables were status (wound or a non-wound patient) and patient normalization of protein level. These classifications constituted nominal-level measurement. All group differences were analyzed using t-tests. Levene’s24 test for homogeneity of variances was applied in each case; where the variance homogeneity assumption was found to have been violated, appropriate corrections were applied to the degrees of freedom in computing the t statistic and in determining P values. In several cases, the strength of association was assessed between maximum protein intake and measures of wound severity. In these cases, the product-moment correlation was used to assess the relationships.
Baseline population characteristics. The 150 patients enrolled in the study included 93 wound patients (37 men, 56 women; mean age 72.5 years) and 57 non-wound patients (26 men, 31 women; mean age 70.3 years). Women comprised the majority of patients in both groups (60% of the wound group, 54% of the non-wound group). The most common comorbidity was diabetes (47% in the wound and 32% in the non-wound group), followed by ventilator dependency, cancer, and dialysis (see Table 1). Wound patients were slightly older than non-wound patients and the proportion of wound patients who had diabetes or were on dialysis was higher in the wound than in the non-wound patients group but the differences were not statistically significant.
Nutrition and follow-up data. On admission, 12% of wound and 21% of non-wound patients had normal pre-albumin levels (the normal pre-albumin range for the author’s facility is 18 to 45). All patients remained in the study until death or discharge. Although average length of stay (LOS) was longer for wound patients (36.9 versus 29.7 days), this difference was not statistically significant (see Table 2). Pre-albumin normalization or an increase by 8 units or more was achieved in 42% and 46% of wound and non-wound patients, respectively. Four patients (4.3%) in the wound group and none in the non-wound group expired. None of the deaths was wound-related.
Feeding failure and complications. Slightly more than half the patients in both groups had at least one cause for feeding failure (see Table 3). The wound group had slightly more identified reasons for feeding failure per patient than did the non-wound patients (63% versus 54%). The most prevalent cause for feeding failure in both groups was “caloric goals met but provision or assimilation of protein inadequate.” This was a cause of feeding failure in 23 (38.9%) wound and eight (25.8%) non-wound patients.
The second most common feeding complication in the wound patient group was abdominal pain, distension, or high residuals (11, 18.6%) followed by non-gastro-intestinal (GI) complications such as sepsis (16.9%), diarrhea (10.2%), and mechanical tube failure (blockage or removal; 3.3%). In non-wound patients, the second most prevalent cause for feeding failure was an inadequate length of stay (six patients). Their pre-albumin levels were increasing but had not normalized or increased by 8 units at the time of discharge. Other causes for feeding failure (in order of prevalence) were abdominal pain, distension, high residuals, non-GI complications, diarrhea, mechanical tube failure, and hyperglycemia (see Table 3). No instances of hypophosphotemia or respiratory complications were attributed to the feeding regimen at any time in either group.
Data analysis. Nutritional outcomes data from the two groups were divided according to ability to improve pre-albumin levels. Improvement of protein levels was defined as achieving a level in the normal range for the author’s lab or increasing by 8 units or more during the study time period. According to the highest level of protein/day given for each patient for the wound and non-wound patients who were and were not able to improve their pre-albumin, the results from this study show that wound patients require significantly more protein than non-wound patients to normalize their protein stores.
Patients who improved their protein received 1.85 g/Kg/day and 1.47 g/Kg/day for the wound and non-wound patients, respectively (t = 3.229, df = 63, one-tailed P = .001). In the wound group, the patients who normalized their protein received more protein than those who did not (1.85 versus 1.69 g/Kg/day) but this difference did not reach statistical significance (see Table 4). For patients who did not improve their protein, the difference between the wound and non-wound groups (1.69 versus 1.56 g/Kg/day) did not reach significance (t = 1.507, df = 83, one-tailed P = .068.)
In the wound patients, the admit PUSH score and the total area (length x width in cm) of Stage III and Stage IV wounds were analyzed to determine which was a better predictor, if at all, of ability to improve protein levels (see Table 5). The admit PUSH score is the sum of the individual wound PUSH scores at admission.21 The wound group members who did not improve their protein had lower PUSH scores on admission and total surface area of Stage III and Stage IV wounds than the group that improved, suggesting that higher severity of wounds was not the responsible factor for their failure to normalize their protein. The correlation between total area of Stage III and Stage IV wounds on admission and the maximum protein given to wound group members who improved their protein stores was .305, which barely missed reaching significance (P = .059) (see Figure 3).
Protein requirements suggested by total admission PUSH scores demonstrated a higher correlation/statistical significance (0.463, P = .003) (see Figure 4). It should be noted that many patients received more than 2g/Kg/day, especially those with higher total PUSH scores. The correlation between protein intake levels and PUSH scores for wound patients who did not improve their protein stores was 0.081, which was non-significant (P =.559).
The prospective design of this study, sample size, and inclusion of all tube-fed patients admitted to one facility yielded reliable, real world data. The inclusion of tube-fed patients increased the accuracy of providing and recording protein intake. However, in some patients, the exact cause of protein repletion failure remained unknown and could have been due to inadequate provision of protein or to patient inability to assimilate protein based on biologic factors. The majority of patients whose protein repletion failed had adequate caloric intake as identified by weight gain and meeting caloric goals but their pre-albumin levels did not improve. In the non-wound patients, the group with improved protein stores received less total protein than the group that did not normalize protein. Because five of the eight patients (62.5%) in the non-wound/non-assimilation group whose pre-albumin levels did not improve had cancer, the results suggest that biologic factors related to the disease prevented protein assimilation in some patients. The results of this study confirm the hypothesis that wound patients require higher levels of protein than is commonly recommended in order to improve pre-albumin levels. The average maximum amount of protein provided to patients whose pre-albumin improved was significantly higher in the wound than in the non-wound group (1.85 g/Kg/day compared to 1.47 g/Kg/day). Moreover, on admission to the author’s facility, only 11 of the 93 wound patients had normal pre-albumin levels.
Wound patients can lose as much as 100 g of protein per day through wound exudate.25 The amount of wound drainage is related to many factors, including the size and depth of the wound(s). When examining wound area of Stage III and Stage IV wounds in patients who increased protein, a slight trend toward requiring more protein is noted. However, because many patients with Stage III and Stage IV wounds of small total areas required large amounts of protein, the usefulness of this information is limited. The low values for total area of Stage III and Stage IV wounds often represent patients with small but deep Stage III or Stage IV wounds and/or multiple Stage II wounds; neither of these characteristics are reflected in the total area of Stage III and Stage IV wounds in this study.
Other factors related to wound drainage and need for protein repletion include the presence of infection or heavy colonization in or beneath the wound, the presence of deep tracts or undermining, wound location, patient fluid volume, and osmotic protein levels. This list includes local factors related to the wound as well as systemic factors. For example, cardiac or renal dysfunction can lead to hypervolemia while malnutrition or liver disease may lead to low protein levels that decrease intravascular osmotic forces and allow more fluid to escape into tissues and wounds.
Not surprisingly, the PUSH score is a better tool for calculating the protein requirement than size of Stage III and Stage IV wounds because it considers additional factors, including the amount of drainage and the presence of non-viable tissue. However, while the PUSH score currently may be the most accurate tool in predicting protein requirements, many local and systemic factors are dynamic and likely to be different with each passing week. Figure 4 shows a wide variance from the regression line, underscoring the need for frequent patient reassessment. Thus, given the reality of ever-changing protein losses through wound fluid, as well as the massive catabolic demands wounds and other comorbidities impose, a static approach to protein provision will fail in many instances.
Also, a substantial difference in the strength of the relationship between wound severity and protein intake between patients who did and did not normalize protein stores suggests that patients who normalized protein stores received a more dynamic protein intake regimen than those who did not normalize protein stores. The term dynamic in this context means responsive to the lack of improvement of the protein parameters (ie, the likelihood that the patients kept getting increases in protein provision if their pre-albumin was not increasing). Thus, it is reasonable to infer that a more dynamic or responsive regimen is associated with a higher likelihood of improving protein stores.
The LTAC hospital represents a unique environment in American healthcare, offering optimal resources and the time to improve patients with complex medical issues such as malnutrition and wounds. Only the LTAC combines the expertise and resources necessary to treat the wounds and the longer time frame needed in many cases to improve nutrition adequately. Acute care hospitals’ diagnosis-related group (DRG)-based system does not allow patients to stay long enough to normalize their nutrition if they are severely malnourished. Skilled nursing facilities offer a lengthier time frame for care but lower reimbursement rates create challenges to having the specialized protocols, equipment, and highly trained personnel necessary to treat complex wounds. Rehabilitation facilities may provide wound care but because patients qualify for rehab by acknowledgment of participation and progress, rehab, not wound care, is the primary focus.
The results of this study show that tube-fed patients often take 3 to 4 weeks or more to normalize protein stores – the length of stay Medicare covers for the typical wound DRG in the LTAC environment. If pre-albumin is measured weekly, the patients with the most severe malnutrition on admission (pre-albumin <10 in the author’s facility) will need to increase their pre-albumin by 3 to 4 units per week to achieve a normal pre-albumin (18 or greater) by their likely time of discharge. Because many patients with wounds and other complex problems do not have feeding tubes and depend on oral intake of protein to optimize their protein stores, study data indicate that the challenge to achieve an optimal nutritional status in these patients is significant. The corollary to this observation is that because time is precious for these patients, conservative approaches to protein provision are not indicated. Study data also reinforce the unique opportunity the LTAC environment provides for patients, enabling them to stay the 3 or more weeks necessary to address nutrition which, in turn, is critical for healing. Thus, legislative efforts or policies that encourage or mandate shorter lengths of stay for patients with severe wounds (most of whom are malnourished) are misguided.
The lack of “overfeeding” complications seen in study patients, even with an aggressive regimen, reinforces McClave’s21 assertion that the gut acts as a natural barrier, making complications seen with TPN such as hypophosphotemia or respiratory complications less likely. The results also call into question the common theme that protein provision above 1.5 to 2.0 g/Kg/day is never indicated.
In order to develop more accurate, wound-based guidelines for protein supplementation, the effect of wound burden (total wound size/PUSH score/stage) on protein requirements was examined. In the wound group with improved pre-albumin levels, patients with a larger total wound area generally were found to have received higher amounts of protein; this trend was more consistent when total PUSH score was used as a wound burden variable. Neither the PUSH score, nor wound area, include wound depth but the PUSH score includes other wound burden variables such as drainage and the presence of non-viable tissue. Although the PUSH score currently may be the most accurate tool in predicting protein requirements, a wide variance from the regression line of Figure 4 indicates frequent patient reassessment is necessary.
A range of protein values versus PUSH scores is noted among wound patients who did not improve their protein stores (see Figure 5). The slope of the line represents the range of protein given and the aggressiveness of the regimen (ie, the likelihood that the patients kept getting increases in protein if their pre-albumin was not increasing). Comparing these results to the wound patients who normalized protein stores suggests that a more aggressive regimen is associated with a higher likelihood of improving protein stores.
Based on study data, the following algorithm, using the patient’s total admission PUSH score as a basis for the amount of starting protein to provide to wound patients, has been developed for use in the author’s facility:
PUSH score 0 to 15 = 1.4 to 1.6 g/Kg/day
PUSH score 16 to 30 = 1.6 to 2.0 g/Kg/day
PUSH score >30 = 2.0 to 2.4 g/Kg/day
Because estimated protein requirements may vary greatly, the most effective practice is to monitor pre-albumin regularly and adjust protein provision accordingly. Pre-albumin should be measured weekly until protein is normalized; protein intake should be increased if pre-albumin does not rise by 3 or 4 units. Wound patients require a dynamic, rather than a static approach.
In addition to answering remaining questions about the cause of feeding failure, studies to examine the external validity of this study’s findings and their applicability to other patient care environments are needed.
The results of a prospective study of wounded and non-wounded LTAC patients indicate that persons with wounds generally require more protein than their non-wounded counterparts. The probability of normalizing protein stores with current methods in tube-fed patients is low (42% and 46% in wound and non-wound patients, respectively). These results and reports (eg, McClave21) suggest that the risk of complications due to under-provision of protein via enteral feeding is higher than the risk of giving too much protein.
Use of a protein provision algorithm in the author’s facility has been greatly enhanced by a dietitian-maintained Nutrition Flow Sheet that tracks patient weight, feeding regimens, and pre-albumin and albumin levels. The Nutrition Flow Sheet can be a valuable tool for healthcare providers addressing wound issues by tracking important information in an easy-to-use format, enabling providers to make informed decisions about the need for pre-albumin testing and subsequent feeding adjustment. Additional research into optimal levels of protein provided to patients with wounds is warranted.
The author thanks Jody Abels, RN, CWS, and Laura Ledkins, RD/LD – without them this study would not exist. Also special thanks to LifeCare Hospital Dallas for supporting this research. The author also gratefully acknowledges the statistical analysis assistance furnished by Dr. Jeffrey S. Kane of Professional Statistical Services (www.ProStatServices.com).
1. Breslow RA, Hallfrisch J, Guy DG, Crawley B, Goldberg AP. The importance of dietary protein in healing pressure ulcers. J Am Geriatr Soc. 1993;41(4):357-362.
2. Lee KS, Posthauer ME, Dorner B, Redovian V, Maloney MJ. Pressure ulcer healing with a concentrated, fortified collagen protein hydro/sate supplement: a randomized controlled trial. Adv Skin Wound Care. 2006;19(2):92,94-96.
3. Chernoff R, Milton KY, Lipschitz DA. The effect of a very high protein liquid formula on decubitus ulcer healing in long-term tube fed institutionalized patients. J Am Diet Assoc. 1990;90(9):A130.
4. Bourdel-Marchasson I, Barateau M, Rondeau V, et al. A multi-center trial of the effects of oral nutritional supplementation in critically ill older inpatients. GAGE Group. Groupe Aquitain Geriatrique d’Evaluation. Nutrition. 2000;16(1):1-5.
5. Pinchcofsky-Devin GD, Kaminski MV. Correlation of pressure sores and nutritional status. J Am Geriatr Soc. 1986;34(6):435-440.
6. Pinchcofsky-Devin GD. Nutritional assessment and intervention. In: Krasner D, Rodeheaver GT, Sibbald RG, eds. Chronic Wound Care: A Clinical Source Book for Healthcare Professionals. Wayne, Pa: Health Management Publications;1997:77.
7. Gersovitz M, Motil K, Munro HN, Scrimshaw NS, Young VR. Human protein requirements: assessment of the adequacy of the current recommended dietary allowance for dietary protein in elderly men and women. Am J Clin Nutri. 1982;35(1):6-14.
8. Sullivan DH, Patch GA, Walls RC, Lipschitz DA. Impact of nutrition status on morbidity and mortality in a select population of geriatric rehabilitation patients. Am J Clin Nutri. 1990;51(5):749-758.
9. Sullivan DH. Risk factors for early hospital readmission in a select population of geriatric rehabilitation patients: the significance of nutritional status. J Am Geriatr Soc. 1992;40(8):792-798.
10. Rudman D, Feller AG. Protein-calorie malnutrition in the nursing home. J Am Geriatr Soc. 1989;37(2):173-183.
11. Shizgal HM. Body composition and nutritional support. Surg Clin North Am. 1981;61(3):739-741.
12. Posthauer ME. Nutrition and wound care. In: Baronsky S, Ayello E, eds. Wound Care Essentials. New York, NY: Lippincott Williams & Wilkins;2003;157-186.
13. Long CL, Nelson KM, Akin JM, Geiger JW, Merrick HW, Blakemore WS. A physiologic basis for the provision of fuel mixtures in normal and stressed patients. J Trauma. 1990; 30(9):1077-1086.
14. Moore FA, Feliciano DV, Andrassy RJ, et al. Early enteral feeding, comparing with parenteral, reduces postoperative septic complications. Ann Surg. 1992;216(2):172-183.
15. Kudsk KA, Croce MA, Fabian TC, et al. Enteral versus parenteral feeding: effects on septic morbidity after blunt and penetrating abdominal trauma. Ann Surg. 1992;215(5):503-513.
16. Peterson VM, Moore EE, Jones TN, et al. Total enteral nutrition versus total parenteral nutrition after major torso surgery: attenuation of hepatic protein reprioritization. Surgery. 1988;104(2):199-207.
17. Fong YM, Marano MA, Barber A, et al. Total parenteral nutrition and bowel rest modify the metabolic response to endotoxins in humans. Ann Surg. 1989;210(4):449-457.
18. Zaloga GP, MacGregor DA. What to consider when choosing enteral or parenteral nutrition. J Crit Illness. 1990;5(11):1180-1200.
19. Weisner RL. Death resulting from overzealous total parenteral nutrition: the refeeding syndrome revisited. Am J Clin Nutri. 1980;34(3):393-399.
20. Silvis SE, Paragas PD. Paresthesias, weakness, seizures and hypophosphatemia in patients receiving hyperalimentation. Gastroenterology. 1972;62(4):513-520.
21. McClave SA, Kleber MJ, Lowen CC. The consequences of overfeeding and underfeeding. J Respir Care Pract. 1997;April/May:57-64.
22. Covelli HD, Black JW, Olsen MS, Beekman JF. Respiratory failure precipitated by high carbohydrate loads. Ann Intern Med. 1981;95(5):575-581.
23. Pompeo M. Implementing the PUSH tool in clinical practice: revisions and results. Ostomy Wound Manage. 2003;49(8):32-46.
24. Levene H. Robust test for the equality of variances. In: Olkin I, ed. Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling. Palo Alto, Calif: Stanford University Press;1960:278-292.
25. Thomas B. Catabolic states. In: Thomas B, ed. British Dietetic Association Manual of Dietetic Practice. Oxford, UK: Blackwell Scientific;1994:537-549.