INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic obstructive pulmonary disease (COPD) is an important cause of morbidity, hospitalization, and mortality worldwide. Patients with COPD often lose weight and, depending on the population studied and the indicator used to determine the nutritional status, between 19 and 60% of patients are classified as malnourished (1). The clinical deterioration associated with weight loss leads to a deterioration in the quality of life in many patients with COPD (5). Although the reasons for the weight loss cannot be fully explained, increased resting energy expenditure (REE) in weight-losing patients with COPD (6), and higher serum tumor necrosis factor-alpha levels in malnourished patients with stable COPD have been implicated as possible reasons for weight loss in some patients with COPD (7, 8).

Patients with COPD are frequently hospitalized with an acute exacerbation. This acute phase may be triggered by an infection or environmental stimuli (dust, pollution, cigarette smoke), and the body's response to infection will ultimately result in elevated energy requirements, which are difficult to meet in acutely stressed patients. Consequently a deterioration in nutritional status (loss of lean body mass) is a likely repercussion (9). The importance of nutritional support for patients with COPD is widely accepted, and under well-controlled conditions refeeding trials in malnourished patients with stable COPD have been successful in improving the nutritional status and respiratory muscle strength (9). Attempts at refeeding malnourished patients with stable COPD in the community have had mixed results (12). Providing appropriate nutritional support to patients with COPD during an acute exacerbation of their disease to prevent the consequences of nutritional deterioration has not been investigated to date.

The principal objective of our study was to evaluate the impact of aggressive oral nutritional support on respiratory and peripheral muscle strength, pulmonary function, length of stay in hospital, distance walked in 6 min, and quality of life compared with traditional nutritional care during an acute exacerbation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study methodology was a randomized clinical trial comparing supplementary oral nutritional support with traditional hospital feeding procedures provided to patients with COPD hospitalized with an exacerbation of their condition. Outcome measures, including weight, FEV₁, FVC, PImax, PEmax, handgrip strength, degree of breathlessness, and general well-being were measured as changes over the course of hospitalization at 2 wk postadmission.

Study Population

Between November 1993 and May 1996, consecutive patients 40 to 85 yr of age admitted to the Montreal Chest Institute with a diagnosis of COPD, and a FEV₁ that was equal or less than 60% of the predicted value were evaluated. Patients who required mechanical ventilation, had a gastrointestinal tract disorder, had active cancer or other conditions predisposing to weight loss, were terminally ill, were unable to communicate in English or French, suffered from mental confusion or followed a special diet were excluded. The purpose and the required adherence to our study protocol were explained to eligible patients, and informed, signed consent was obtained from participants. Participants were randomly assigned to either a supplemental (treatment) or usual (control) feeding group. The study was approved by the ethics committee of the Montreal Chest Institute.

Measurements

Height and weight were measured using a beam balance scale (Detecto Scales Inc., Brooklyn, NY) and body mass index (BMI = kg/m²) was obtained from these measurements. Patients were dressed in hospital gowns and not wearing shoes. Handgrip strength, a well-established measure of upper body strength and activities of daily living in older adults (16), was measured using a handgrip dynamometer (Jamar Dynamometer; Alimed Inc., Dedham, MA). The measurement of isometric grasp was done in the dominant hand with the patient sitting, the shoulder adducted and neutrally rotated, elbow flexed to 90 degrees, forearm and wrist in neutral positions. The mean of three measurements was recorded.

The FVC and FEV₁ were measured using the 570 Wedge Spirometer (Med-Science Electronics, St. Louis, MO). The best of three measurements was recorded. Measurements were carried out according to standards of the American Thoracic Society (17).

  Using a handheld manometer (Boehringer Laboratories, Inc., Norristown, PA), inspiratory muscle strength (PImax) was measured as the maximal static inspiratory pressure at the mouth at the end of a maximal expiration. Expiratory muscle strength (PEmax) was measured as the maximal expiratory pressure at the mouth after a maximal inspiration. The best of three measurements was recorded. All strength measurements were done by laboratory personnel who were blind to patients' assignment.

The degree of breathlessness was assessed using the oxygen-cost diagram. The instrument lists everyday activities, which are placed proportional to their oxygen cost along a 100-mm vertical line. The distance from the bottom of the scale to the patient's mark is measured, and it represents an index of the subject's dyspnea. A general well-being questionnaire (18) of broad-ranging indicators of subjective feelings of psychologic well-being and distress was administered. The self-administered questionnaire assesses how a person feels about his or her "inner personal state," rather than conditions such as income, work, environment, etc. The questionnaire covers both positive and negative feelings. Scores between 0 and 60 reflect "severe distress," scores between 61 and 71 "moderate distress," and scores between 73 and 110 represent "positive well-being." Length of stay in hospital was recorded for each patient and the range of length of stay was reported.

Urinary urea nitrogen was measured using 24-h urine collections for a subset of the study population. Total urinary nitrogen was estimated by adding a correction factor of 2 g nitrogen for nonurea nitrogen compounds such as creatinine nitrogen, ammonia nitrogen, uric acid nitrogen, and other minor nitrogenous compounds that are thought to remain stable on a general diet (19). Other miscellaneous nitrogen losses, which are seldom directly measured (stool, skin, expired air, saliva, blood) and which may vary depending on the disease state, are accounted for by the addition of another correction factor of 2 g. Using the relationship of protein intake (g)/6.25  [(urinary urea nitrogen (g) + 4)], nitrogen balance data were obtained.

Mean daily glucocorticosteroid intake was abstracted from the hospital charts and was calculated from admission up to and including the study period of 14 d as well as from admission to day of the nitrogen balance study. Because of the difference in activity of the glucocorticosteroids, all medications (hydrocortisone, prednisone, and methylprednisolone) were converted to an equivalent anti-inflammatory dose of methylprednisolone (20).

The 6-min walk test measures the distance covered in 6 min and was administered once at the end of the study because a number of subjects would have been unable to do the test upon admission to hospital.

Patients were randomized to either a supplemental (treatment) or usual feeding (control) group during hospitalization after written consent was obtained. The Harris and Benedict equation was used to establish resting energy expenditure (REE). Subjects in both groups ordered their food and beverages from the hospital menu. In addition to the hospital tray, patients in the supplemental feeding group received oral supplements (Ensure, Ensure Plus, or a variety of puddings) or extra snacks to assure a caloric intake of at least 1.5 × REE if their BMI was normal (20 to 27) and at least 1.7 × REE if their BMI was below 20. The type and amount of food and beverages consumed by subjects while in hospital were recorded by hospital personnel using a calorie count, and these results were verified by the research dietitian with 24-h recalls every other day. If a subject was discharged prior to 14 d, 24-h recalls were administered by telephone. Nutrient analysis of food intake was completed with the Food Processor Plus program (Version 5.0; ESHA Research, Salem, OR). Mean energy and macronutrient intakes were recorded.

Statistical Analysis

The analysis of the equivalency of the trial groups was evaluated by comparing the similarity of the baseline characteristics of the two study groups using Student's independent t test. Because of the loss of subjects either to death or to follow-up, complete data versus incomplete data were compared using independent t tests. To determine changes within each group during the study period, paired t tests were used. Changes in variables of interest (final-baseline) between the two groups were analyzed using Student's independent t test. In a further observational evaluation of changes in nutritional status, Pearson's correlation coefficient was used to describe the relationship between glucocorticosteroid use and muscle strength and nitrogen balance. The statistical analyses were performed using SAS statistical software (SAS Institute Inc., Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Randomization of the 33 consenting subjects resulted in 16 subjects in the control group and 17 subjects in the treatment group. There was one error whereby one patient in the control group with a baseline FEV₁ (% predicted) of > 60% was initially included but data could not be used in the analysis. Four patients (three control, one treatment) refused to come back for final outcome measures or could not be located; their data were not included in the analysis. After randomization, one patient in the treatment group refused to continue, and no data were available for analyses. Three patients (two control, one treatment) became too ill to undergo outcome measures, two of these patients (one control, one treatment) died and one patient remained in hospital for 93 d. Their data were not used in the analysis. The final analysis consisted of complete data for 10 subjects in the control group and 14 subjects in the treatment group.

Patient Characteristics

The baseline characteristics of the study population are summarized in Table 1. All variables, except age, were normally distributed. Baseline characteristics were similar between control and treatment subjects. The study group consisted of elderly patients with severe airflow obstruction (FEV₁ < 35% predicted). Subjects experienced breathlessness while standing or walking on a flat surface, and on admission to hospital patients were moderately distressed and anxious. Respiratory muscle strength, measured as PImax and PEmax, was normal for this elderly population as was their grip strength, which was comparable to values of healthy elderly men and women (16). More men (n = 15) than women (n = 9) were studied, reflecting the greater prevalence of this disease among men.

Changes during the Course of Hospitalization

No change in FEV₁ (% predicted) in either treatment group occurred during the 14 d of the study (Table 2), and no significant difference in the change in FEV₁ (% predicted) during the study period between the two study groups was observed (p = 0.099). This sample size had adequate power (80%) to detect a difference as small as 10%. In contrast, FVC (% predicted) improved during the trial in the treatment group (+8.7%, p = 0.030) and did not change in the control group (3.5%, p = 0.138). A significant difference in the change in FVC (% predicted) between the treatment group and the control group (p = 0.015) was observed.

No significant change in either PImax or PEmax occurred within each group nor were there any differences in the changes of PImax and PEmax between the treatment groups during the study period. Unfortunately, because of the high variability in the change in measures of respiratory muscle strength in this acutely ill group, a sample size of 103 subjects would have been required to ensure adequate statistical power for the detection of a change of at least 10 cm H₂O for PImax. Similarly, despite a sufficiently large sample size to detect a difference of 4 kg, there were no significant changes in grip strength.

The change in the degree of breathlessness was not sufficiently different between the treatment and control groups (p = 0.410). Subjects in both groups tended towards less breathlessness as they recovered (control, +7.7 mm, p = 0.073; treatment, +13.7 mm, p = 0.033).

The total general well-being score in the control group was unchanged over the study period (10 points, p = 0.485), whereas that of the treatment group improved significantly (+12 points, p = 0.020). The difference between these changes was of borderline statistical significance (p = 0.066) on a two-tailed t test.

At the end of the study, the treatment group could not walk farther than the control group (252 versus 200 m, respectively). The sample size of the study provided adequate statistical power to detect a change of 70 m.

The energy intakes of the two groups are summarized in Table 3. The treatment group consumed significantly more energy/kg body weight (39 kcal/kg/d) than did the control group (29 kcal/kg/d, p = 0.004) without reporting any side effects such as increased breathlessness or gastrointestinal discomfort. The difference between the total energy intakes for the two groups was of marginal significance (p = 0.052) and most likely because of the nonsignificant tendency for the control subjects to weigh more than the treatment subjects. Protein intake/kg body weight was also significantly higher in the treatment group, 1.5 versus 1.2 g/kg/d for the control group (p = 0.025), but carbohydrate and fat intake were not different between the groups. The ratio of energy intake/Harris Benedict equation resulted in a ratio of 1.89 × REE for the treatment group, significantly higher than the control group's ratio of 1.47 × REE (p = 0.004).

Nitrogen Balance Studies

Nitrogen balance studies were completed for a subset of five subjects in the control group and nine subjects in the treatment group (Table 4). The mean nitrogen loss was 5.10 ± 1.60 g nitrogen/d for the control group and similar for the treatment group 6.46 ± 1.99 (p = 0.654), indicating that despite higher protein and energy intakes, nitrogen balance was very negative in virtually all subjects. Nitrogen balance studies took place, on average, on the sixth day of hospitalization. The well-known catabolic effect of glucocorticosteroids is a possible explanation of the very negative nitrogen balance found among even the well-fed subjects. The mean intake of methylprednisolone from admission to the end of the study period was similar between the control group (67 mg/d) and the treatment group (66 mg/d) (p = 0.949). In addition, methylprednisolone intake from admission to the day of each subject's nitrogen balance study was similar between the groups, with an intake of 115 mg/d for the control group and 101 mg/d for the treatment group, p = 0.625. 

The relationship between methylprednisolone intake from admission to the date of the nitrogen balance study and nitrogen balance indicated a negative correlation (r = 0.73, p = 0.0048, n = 13). In addition, methylprednisolone intake during the study period and change in grip strength were negatively correlated (r = 0.58, p = 0.007, n = 20), indicating declines in muscle function among those receiving the highest doses of corticosteroids.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The rationale for our study was based on the fact that well-designed, randomized controlled trials providing nutritional support to already malnourished patients with COPD have been successful and have resulted in significant improvements in weight and concomitant improvements in respiratory and peripheral muscle function (10, 11, 14). However, the prevention of loss of lean body mass in hospitalized patients with an exacerbation of their disease has never been investigated, and it is important to prevent loss of lean body mass as this may be more effective in terms of therapeutic benefits and may also result in reduced health care costs. Refeeding already malnourished patients is costly and difficult to maintain; also, regaining lean body mass in older adults requires very high energy intakes (21).

This randomized clinical trial indicated that oral intake can be substantially improved. At baseline the treatment groups were very similar. Over the course of the study period no changes in FEV₁ (% predicted) were observed. This is in keeping with successful nutritional intervention studies carried out in malnourished patients with stable COPD (10, 14), where despite weight gain and improvements in respiratory muscle function, pulmonary function measures did not improve. FEV₁ is a measure of functional impairment (narrowed airways, decreased elastic recoil), and as such it may not respond to nutritional therapy. Preadmission values for FEV₁ (% predicted) were not available in our study, and it is unknown whether pulmonary function measurements had returned to preillness values. Contrary to findings in successful refeeding studies (10, 11, 14), however, FVC improved significantly within the treatment group (8.7%, p = 0.030), whereas no such improvement was seen within the control group (3.5%, p = 0.138).

No changes in PImax and PEmax were noted over the study period in either group. Although respiratory muscle strength evaluated as PImax and PEmax is reduced in malnourished persons (22) and can be strengthened with short-term nutritional therapy in stable patients (9), the highly variable changes in PImax and PEmax observed in this study of acutely ill subjects make it difficult to detect the magnitude observed among stable subjects. In addition, the respiratory muscle strength measurements found in our patients who were not malnourished were normal for those of their age, so that improvement would not be expected and only a prevention of a decline could be anticipated.

Change in body weight as a measure of change in nutritional status in acute exacerbations of COPD is not very useful. As fluid balance is often disturbed, small changes in weight may be the result of either fluid retention or fluid loss complicated by steroid treatment. Our results indicated that weight remained unchanged during the study period, which does not necessarily indicate that nutritional depletion or loss of lean body mass did not occur. According to Schols and colleagues (23), a substantial number of normal weight patients with COPD with abundant fat mass have a depletion of fat-free mass.

In contrast to the difficulties of evaluating changes in body weight in this population, short-term changes in body protein can be estimated using nitrogen balance (24). This method requires that nitrogen intake and output are accurately determined. Good agreement between urinary urea nitrogen and total urea nitrogen in critically ill patients with a variety of clinical conditions (25, 26) has been reported. If urinary area nitrogen does not adequately reflect total urea nitrogen (27) one might expect an underestimation of the degree of negative nitrogen balance, making the effect even stronger. In our study, complete 24-h urine collections were available for 14 subjects, 11 were in negative balance and three in balance. As only one measure was done per subject we are left with the question of whether or not the state of negative nitrogen balance continued throughout the study period. Our results revealed a negative correlation between methylprednisolone intake up to the day of nitrogen balance studies and nitrogen balance (r = 0.73, p = 0.0048). Although the negative nitrogen balances were not totally unexpected, this is the first study to show such an effect in patients with COPD. To put our findings of the nitrogen balance studies in perspective, it is useful to underline that a nitrogen loss of ~ 6 g/d will result in a loss of ~ 37.5 g protein/d. Using the relationship that lean body tissue is 20% protein, an estimated loss of 187.5 g of lean tissue/d or 1.3 kg/wk can develop if negative nitrogen balances are not corrected.

The adverse effects of steroids on nitrogen balance have also been observed in other diseases. Patients with rheumatoid arthritis (28), whose nitrogen balances were negative during therapy with methylprednisolone (1,000 mg/d for 3 d) (5.77 ± 1.30 g nitrogen/d), continued to be negative after therapy had stopped (4 d post-treatment) (3.54 ± 1.38 g nitrogen/d) (mean ± SEM) despite high energy (46.8 ± 6.2 kcal/kg/d) and protein (1.6 ± 0.6 g/kg/d) intakes. Whether this nitrogen loss can be prevented in patients with COPD while in hospital with an exacerbation has not been established. Stable, malnourished patients with emphysema were able to achieve nitrogen equilibrium with energy intakes of 46.8 ± 1.9 kcal/kg/d and 1.8 ± 0.08 g protein/kg/d (29), which were slightly higher than the energy and protein intakes in the supplemented group. The catabolic effects of corticosteroids or high circulating levels of TNF-alpha that have been observed in weight-losing patients with COPD may override the positive effects of increased dietary intake.

Changes in body protein need to be investigated because of the consequences that the loss of lean body mass has on respiratory muscle strength (22), exercise capacity (2), and mortality (5, 30). In animals, chronically active muscles such as the respiratory muscles are thought to be less susceptible to steroid-induced myopathy than less active peripheral muscles such as would be involved in handgrip (31). Our study supports these arguments, as no correlation between methylprednisolone intake and PImax and PEmax was noted; however, a negative correlation between methylprednisolone intake during the study period and change in grip strength was observed. In clinical studies, the relationship between corticosteroid intake and muscle strength has recently been explored (32) and no correlation between methylprednisolone intake during a short-term hospital stay and respiratory muscle strength was found, but the average daily dose calculated for the previous 6 mo significantly (p < 0.05) influenced PImax but not PEmax. Not all investigators have found associations between corticosteroid treatment and decreased muscle strength (33, 34). Our observation of the correlation between methylprednisolone intake and change in grip strength may have been confounded by severity of illness. Without an experimental study it is difficult to establish whether corticosteroid treatment causes loss of muscle strength or is a marker of disease severity.

In conclusion, our study clearly demonstrated that an important increase in oral intake in patients hospitalized with an acute exacerbation is possible, but the effects of any improvements as a result of better intake are difficult to measure. The FVC was increased significantly in the supplemented group, and a strong trend for improved general well-being was observed. Measurements of muscle strength were not improved. The implications of negative nitrogen balances, particularly if prolonged, are serious as loss of lean body mass is associated with poorer prognosis. The catabolic process may have resulted from high energy or protein requirements that were not met or had been induced by high doses of methylprednisolone or may have been the consequences of other catabolic processes.