Investigating heterogeneity in studies of resting energy expenditure in persons with HIV/AIDS: a meta-analysis

¹ From the Smart Foods Centre, University of Wollongong, Australia.

² Supported by the Australian Research Council.

³ Reprints not available. Address correspondence to M Batterham, Smart Foods Centre, University of Wollongong, Northfields Avenue, Wollongong, NSW 2522, Australia. E-mail: marijka{at}uow.edu.au.

ABSTRACT  
Background: There is conflict in the literature about the extent of alterations of resting energy expenditure (REE) in persons with HIV.

Objective: The study was conducted to ascertain the mean difference in REE (in kJ) per kilogram of fat-free mass (FFM; REE/FFM) between HIV-positive subjects and control subjects and to investigate heterogeneity in the literature.

Design: A meta-analysis comparing classical and Bayesian methods was conducted. Heterogeneity was investigated by using subgroup analysis, metaregression, and a mixed indirect comparison.

Results: Of 58 studies meeting the inclusion criteria, 32 included both HIV-positive and control groups; 24 of these 32 were included. Thirty-seven studies were used in the mixed indirect comparison, and 30 were used in the subgroup comparisons of the HIV-symptomatic, lipodystrophy, weight-losing, and weight-stable subgroups and the healthy (HIV-negative) control group. Mean REE/FFM was significantly higher in 732 HIV-positive subjects than in 340 control subjects [11.93 kJ/kg (95% CI: 8.44,15.43 kJ/kg) and 12.47 kJ/kg (95% CI: 8.19,16.57 kJ/kg), classical and Bayesian random effects, respectively]; the test for heterogeneity was significant (P < 0.001). Both the mixed indirect comparison and the subgroup analysis indicated that REE/FFM was highest in the symptomatic subgroup; however, the small number of studies investigating symptomatic subjects limited statistical comparisons. The presence of lipodystrophy, use of highly active antiretroviral therapy, subject age, and method of body-composition measurement could not explain the heterogeneity in the data with the use of metaregression.

Conclusions: REE/FFM (kJ/kg) is significantly higher in HIV-positive subjects than in healthy control subjects. Symptomatic HIV infection may contribute to the variations reported in the literature.

Key Words: HIV • resting energy expenditure • resting metabolic rate • meta-analysis • metaregression • Bayesian methods • body composition • lipodystrophy

INTRODUCTION  
Controversy surrounds the role of resting energy expenditure (REE) in HIV-related metabolic abnormalities such as wasting and lipodystrophy syndrome. Early in the HIV pandemic, it was proposed that REE may be elevated in persons with HIV, because it was thought that neither malabsorption or decreased energy intake alone nor both together could explain the weight loss that was characteristic of untreated infection (1). In noninfective malabsorption, malnutrition, or underfeeding, an appropriate metabolic response is a compensatory drop in REE to preserve the fat-free mass (FFM) (2). The first 2 published reports were conflicting; Kotler et al (3) reported significantly lower REE/kg FFM in HIV-positive subjects than in healthy control subjects, whereas Hommes et al (4) reported significantly higher REE adjusted for FFM in HIV-positive subjects than in healthy control subjects. The literature published since those initial reports is difficult to summarize informally, because different authors have presented the results by using different summary statistics and different population subgroups, and because methods of measuring body composition vary in the studies. Because FFM is the primary determinant of REE, accounting for 70–80% of the variation in REE in healthy subjects (5), comparisons between studies must investigate whether the use of reference or field methods of body-composition measurement explains some of the discrepancies in the results.

Since the widespread introduction of highly active antiretroviral therapy (HAART) in westernized countries in 1996, a new metabolic abnormality, lipodystrophy, has been described (6); its etiology remains unknown, but the syndrome is usually associated with the use of HAART, particularly the protease inhibitor class of drugs but also the nucleoside analogue reverse transcriptase inhibitors (7). More recently, there have been conflicting reports about the differences in REE between HIV-positive subjects with and without lipodystrophy syndrome (8, 9).

Clinically, it is of value to establish whether REE is indeed elevated in HIV-positive persons (or in subsets of the HIV-positive population) so that appropriate nutritional advice on the necessary energy intake to achieve the desirable weight can be provided. An integration of the currently available literature is necessary to provide this clinical information.

The primary aim of this study was, by conducting a meta-analysis combining previous research, to ascertain an overall mean difference in REE between HIV-positive subjects and healthy control subjects. Two secondary aims of this study were to investigate variations in this mean difference between various clinical subgroups (eg, persons with lipodystrophy, those who are losing weight, those who are symptomatic, and those who are weight stable) and to investigate the effect of potential confounding covariates on that mean difference. A priori, these covariates were the presence or absence of lipodystrophy in the HIV-positive group, the choice of method of body-composition measurement, the use of HAART in the HIV-positive group, and the mean age of the subjects. In addition, because of the rapidly increasing popularity of the Bayesian framework over classical analysis, which results from the Bayesian framework's allowing the incorporation of external evidence and accounting for all sources of variability (10), both types of analysis were reported.

MATERIALS AND METHODS  
Search strategy and identification of reports
A database search of Medline, CINAHL, Current Contents Connect, HIV/AIDS database, and Expanded Academic Index (from 1981, when AIDS was first described, to September 2004) was conducted by using the search terms "resting energy expenditure and HIV," "resting metabolic rate and HIV," "indirect calorimetry and HIV," and "fat oxidation and HIV." Identified articles were then searched to ascertain whether they met the following inclusion criteria: 1) the studies were conducted in humans; 2) the subjects were adults; 3) the subjects were measured after an overnight fast or under a strict postprandial protocol (because diet-induced thermogenesis has been shown to be prolonged past 300 min in HIV-positive subjects) (11); and 4) data on body composition were collected.

Only studies published in English were considered. The references of the articles identified were also searched, and authors were contacted to identify additional publications.

Data analysis
Crude, classical, and Bayesian methods of meta-analaysis were performed for comparison and to maximize the use of the available data. To use all of the available data comparing HIV-positive subjects and healthy control subjects when estimates of variance were not provided, crude meta-analysis based on the method of Gotzsche (12) was performed. This technique, in which the mean REE of each group is divided by the mean FFM of each group, and that value is then used to calculate an overall mean difference and 95% CIs, has been used previously in a similar meta-analysis (13).

A traditional meta-analysis, in the classical framework, comparing REE divided by FFM (kJ/kg) in the HIV-positive and control groups and within the HIV subgroups, was performed. The random-effects model based on the method-of-moments estimator as proposed by DerSimonian and Laird (14) was used, with the inverse-variance fixed-effects model shown for comparison; algebraic descriptions of these models are available elsewhere (15, 16). These analyses were performed by using the METAN (17) command in STATA software (version 7.0; Stata Corporation, College Station, TX).

A Bayesian random-effects method was also used to estimate the common effect and the between-study variance (²) and to compare this estimate with the estimates obtained by using the method of DerSimonian and Laird (18). Here, too, the Bayesian fixed model is shown for comparison. Classical (or frequentist) methods of statistics assume that each study (in the current case, this meta-analysis) is one in a long-running series of experiments in which the current study estimate (the overall mean difference) is likely to lie within the stated CIs 95% of the time. This differs from the Bayesian approach, in which the study estimate (the overall mean difference) has a probability distribution that expresses our prior belief (prior distribution) about the mean combined with the available data (likelihood) (19). Two prior distributions were used: a noninformative prior distribution and an informative prior distribution calculated by using the data provided from the single-arm HIV studies. Single-arm studies can be incorporated into a Bayesian analysis in the form of a prior distribution, which allows for the use of these data (15). The ability to incorporate external information, such as previous research, by using an informed prior distribution (10) is an advantage of the Bayesian technique. To ascertain the effect size and variance for these single-arm estimates, a bootstrap resampling procedure was used to generate control samples with replacement from the pooled data of the available individual studies. In all Bayesian models, the likelihood was initially assumed to be from a normal distribution, the prior distribution for the mean difference was also assumed to be normally distributed, and the prior distribution for the heterogeneity term in the random-effects models was assumed to be a conjugate gamma distribution, because this is computationally convenient and because it provides the effect estimate (posterior distribution) in the same form as the likelihood (normal distribution) (18). The choice of gamma prior distributions for the heterogeneity term was tested by comparing the results with those obtained by using a uniform distribution on (20). Random-effects models with the likelihood that the mean difference between the studies would be coming from a Student's t distribution with the noninformative prior distributions were also evaluated for comparison and to test the assumption of normality of the mean difference estimate. A model with 4 df and a model with the df modeled as an additional unknown parameter are presented (21, 22).

A mixed indirect comparison was performed in the Bayesian framework on the basis of the method proposed by Spiegelhalter et al (20). This model allows the incorporation of all available information regardless of whether there are data for all comparisons. The model is set up to compare the HIV subgroups (ie, lipodystrophy, weight-losing, weight-stable, and symptomatic) with healthy control subjects, and within-group comparisons are also modeled.

All Bayesian models were performed by using WINBUGS software (version 1.4; Imperial College & Medical Research Council, Cambridge UK, 2003). A burn-in period of 5000 iterations was used for all models, and the subsequent 5000 iterations were used to estimate the variables of interest. Diagnostic procedures including inspection of autocorrelation plots and trace histories were performed to verify the results. WINBUGS employs a simulation-based Gibbs sampling technique, which is a type of Markov chain Monte Carlo method (10).

An ancillary analysis was performed on the available individual data to obtain an estimate of REE adjusted for FFM (23). This analysis was conducted by using the general linear model (analysis of covariance) in SPSS for WINDOWS software (version 11.5.0; SPSS Inc, Chicago, IL). Two models were employed: an analysis with HIV-positive or healthy control subjects as the factor and an analysis considering the study site in addition to HIV status of the subject.

Assessment and exploration of heterogeneity
Heterogeneity was detected by visual inspection and by using the chi-square test and was quantified by calculating the I² statistic by the equation

RESULTS  
Seventy-three studies were identified in the initial search. Seven studies were conducted in children and were excluded on that basis (30–36). Two studies were excluded because the measures were not fasting or standardized (37, 38), and 6 studies were excluded because body composition was either not assessed or not reported (39–44). Thus, 58 studies remained for consideration in the analysis (3, 4, 8, 9, 11, 45–97). Of these 58 studies, 32 included an HIV-negative control group, and they were considered for the primary analysis. Twenty-six of these 32 studies provided enough information for calculation of the crude effect estimate, and the inclusion of both an estimate and SD or SE allowed formal statistical analysis of 24 studies. When data for the HIV-positive group were presented only in subsets, either the asymptomatic, the weight-stable, or the HIV as opposed to the AIDS group was used for the primary analysis, and the additional subsets were used in the mixed indirect comparison estimate for which 37 studies were included and in the subgroup comparisons for which 30 studies were available. Authors of the 58 studies initially considered were contacted to ascertain whether they would provide either their original data for calculations of summary statistics or the summary statistics if the raw data were not provided in their reports. Individual data were available for 21 studies, 9 of which included a control group; it is important that individual data or summary estimates were provided for all studies that included subjects with lipodystrophy. Relevant details of all the studies conducted in adults are summarized elsewhere (see Appendix A under "Supplemental data" in the current issue at www.ajcn.org).

The mean effect size and 95% CIs for the difference between 785 HIV-positive subjects and 403 healthy control subjects—11.02 kJ/kg FFM (95% CI: 8.03, 14.01)—was obtained by using the method of Gotzsche (12), and those findings indicated that overall REE was significantly higher in HIV-positive subjects than in healthy control subjects. The means and sample sizes for the studies included in this analysis are shown in Table 1.

View this table:
TABLE 1. Resting energy expenditure (REE) and sample sizes for the study comparing HIV-positive subjects with healthy control subjects¹

 
The forest plot in Figure 1 shows each study's effect estimate and 95% CI, as well as the overall estimate for the DerSimonian and Laird random-effects primary analysis comparing the HIV-positive subjects and the control subjects. Means, SDs, and sample sizes are shown in Table 1. The mean difference estimates and 95% CIs (or credible intervals for the Bayesian analyses) for the fixed- and random-effects models for the traditional and Bayesian analyses are shown in Table 2. All analyses indicated that REE/FFM was significantly higher in HIV-positive subjects than in healthy control subjects.

View larger version (35K):
FIGURE 1.. Forest plot showing the overall comparison of resting energy expenditure (REE; in kJ) per kg fat-free mass (FFM) in HIV-positive subjects and healthy control subjects.

 

View this table:
TABLE 2. Summary of mean difference for the fixed and random-effects models and the heterogeneity parameter (²) for the primary analysis comparing HIV-positive and healthy control subjects¹

 
For the random-effects models, the between-study variance (²) is shown along with the overall mean difference estimates. The ² is inflated in the Bayesian models on the basis of the normal distribution and, to a lesser extent, by using the t distribution with the df modeled as an unknown parameter. Therefore, the estimate based on the t distribution with 4 df is used for comparison.

In this analysis, the informed prior distribution was similar to the data set incorporated in the likelihood and did not alter the posterior distribution (Table 2). In addition, the use of uniform prior distributions on , instead of the gamma prior distribution, resulted in similar estimates for the random-effects model, which suggests that the estimates were not sensitive to the distribution of the prior distribution (20) (Table 2).

The use of the random-effects classical estimate of the overall mean difference (11.93 kJ/kg) and multiplication of that number by the mean FFM in the HIV-positive groups (53 kg) gives a daily difference in energy expenditure between HIV-positive and control subjects of 630 kJ (classical estimate) or 661 kJ [Bayesian random-effects estimate based on the t distribution (12.47 kJ/kg)]. This difference reflects an elevation in REE of 9% when the mean REE/FFM for HIV-positive subjects is divided by the REE/FFM for control subjects.

By using the pooled individual-subject dataset, REE in HIV-positive subjects (n = 587) was significantly higher than that in control subjects (n = 189) after adjustment for FFM (7229 and 6729 kJ/kg, respectively; P < 0.001) by analysis of covariance (using the general linear model). Similarly, after adjustment for the study site, the elevation remained significantly (P < 0.001) different between HIV-positive subjects (n = 420; 7036 kJ/kg) and control subjects (6452 kJ/kg). These differences represent elevations of 7% and 9%, respectively, and are consistent with the meta-analysis values.

Assessment and exploration of heterogeneity
The test for heterogeneity was significant (² = 88.73, df = 23; P < 0.001), and the proportion of variability in the mean difference due to heterogeneity rather than to sampling error (I²) was 74%. Forest plots of the various subgroupings are shown in Figure 2. Within the HIV-positive subgroups, there was a trend toward higher REE/FFM in the symptomatic subjects than in the weight-stable subjects (P = 0.079). Differences in REE/FFM between the weight-losing (P = 0.590) and lipodystrophy (P = 0.976) groups and the healthy control subjects were not significant. All HIV subgroups showed significantly higher REE/FFM than was seen in the control subjects: symptomatic subjects, P = 0.012; weight-losing subjects, P = 0.020l; subjects with lipodystrophy, P < 0.001; and weight-stable subjects, P < 0.001). Heterogeneity is evident in all subgroup comparisons (P < 0.001) except that between weight-losing and weight-stable subjects (P = 0.212). The mixed indirect comparison model (Table 3) shows that REE/FFM is significantly higher in all HIV subgroups than in the healthy control subjects and, in addition, that symptomatic subjects have significantly higher REE/FFM than do both the healthy control subjects and the other HIV subgroups.

View larger version (33K):
FIGURE 2.. Subgroup analyses: forest plots showing the comparison of resting energy expenditure (REE; in kJ) per kg fat-free mass (FFM) in healthy control subjects and subgroups of HIV-positive subjects. A, weight-stable group compared with symptomatic group; B, control group compared with symptomatic group; C, weight-stable group compared with weight-losing group; D, control group compared with weight-losing group; E, weight-stable group compared with lipodystrophy group; F, control group compared with lipodystrophy group; G, control group compared with weight-stable group.

 

View this table:
TABLE 3. Bayesian mixed indirect comparison of resting energy expenditure (REE; in kJ) per kg fat-free mass (FFM)¹

 
The slope coefficients and intercepts of the metaregression analyses are shown in Table 4. Lipodystrophy, method of body-composition measurement, subject age, and HAART usage in the HIV-positive group were unable to explain the heterogeneity by using metaregression in either a classical or Bayesian framework, as indicated by the 95% CIs (as confidence intervals and credible intervals, respectively) and P values of the slope coefficients.

View this table:
TABLE 4. Univariate metaregression (classical and Bayesian) investigating contribution of covariates to heterogeneity in resting energy expenditure (in kJ) per kg fat-free mass¹

 
Data were available from only 2 studies that included groups of HIV-positive subjects without lipodystrophy who were receiving and not receiving HAART. Our previous research (87) found REE/FFM in 32 subjects taking HAART to be 9.5 kJ/kg (95% CI: –21.51, 2.51) lower than that in 8 subjects not taking HAART, whereas Kosmiski et al (8) found that REE/FFM in 13 subjects taking HAART was 8.79 kJ/kg (95% CI: –2.21, 19.78) higher than that in 5 subjects not taking HAART, which led to an overall mean difference, combining these 2 studies, of –0.19 kJ/kg (95% CI: –18.11, 17.73). As a preliminary analysis, the 24 studies for the main analysis were split into 2 separate groups (5 studies in which all subjects were using HAART and 19 studies in which HIV-positive subjects were not using HAART); the studies in which subjects were using HAART had an overall mean difference in REE of 15.10 kJ/kg FFM compared with healthy controls, whereas the studies in which the HIV-positive subjects were not using HAART showed a mean difference of 11.02 kJ/kg FFM, which equates to an estimated difference of 216 kJ/d.

Publication bias
The funnel plot (Figure 3) shows both an overestimation and underestimation of the mean effect size, with the results in 10 studies clearly greater than the mean and those in another 10 studies below the mean; 4 studies fall virtually on the mean. There is some evidence of a "tunnel" or gap separating the outlying study by Kotler et al (3). The 2 studies around the line of no difference between the groups both have a relatively small sample size, which indicates that nonsignificant studies with small sample sizes are published. The formal tests for publication bias—the tests of Egger et al (bias: = –0.45, t = –0.41; P = 0.686) and Begg and Mazumdar (z = 0.32, P = 0.747)—were not significant. Visually, this finding is supported by presentation of the funnel plot of Begg and Mazumdar (Figure 3), which is symmetrical except for the outlying study by Kotler et al (3). The trim and fill method suggested there were no missing studies and produced an unadjusted estimate, and that, again, suggests that there was no publication bias in this analysis.

View larger version (11K):
FIGURE 3.. Publication bias plots of the mean difference in resting energy expenditure (in kJ) per kg fat-free mass in HIV-positive subjects and healthy control subjects: the funnel plot shows sample size (expressed as the inverse of the SE) compared with treatment effect, the vertical line at 0 is the line of no difference, and the line at 12.47 kJ/kg is the mean difference. The plot of Begg and Mazumdar (with pseudo-95% CIs) shows symmetry around the overall mean.

 
Sensitivity analysis
The fixed-effect estimates provided in Table 2 were similar to the random-effects estimates obtained by using both the classical and Bayesian methods. The mean difference between studies was similar with the use of all models. The Bayesian random-effects analysis using the t distribution on 4 df produced the lowest value for the heterogeneity parameter ² and is therefore the preferred estimate.

The calculated pooled SD was 15.15 kJ/kg FFM. This value was substituted for the SD for the 2 studies included in the crude analysis for which a variance estimate was not available so that the study could be included in the primary analysis (Table 1). The random-effects meta-analysis was redone with these additional studies included. The combined mean difference was 12.77 kJ/kg FFM (95% CI: 9.45, 16.10 kJ/kg FFM; z = 7.53, P < 0.001). This difference would represent a daily difference in energy expenditure of only 16 kJ (based on an average FFM in the HIV studies of 53 kg) when compared with the random-effect Bayesian estimate (Student's t distribution on 4 df).

The Bayesian analysis of the influence of individual studies is shown in Table 5. The removal of the data from the 2002 study by Korach et al (82) produced the highest mean difference, and the removal of the data from the 2004 study by Crenn et al (93) produced the lowest estimate; this difference translated into a small change in daily energy expenditure (61 kJ/d based on the average FFM of 53 kg). The most substantial effect on the between-study variation (²), which decreased by 74% of the combined estimate value, was seen when the data from the study of Kotler et al (3) were removed. The classical framework produced similar results, although the omission of the studies of Kotler et al (mean 12.72 kJ/kg; 95% CI: 9.49, 15.95) and Korach et al (mean 12.72 kJ/kg; 95% CI: 9.33, 16.10) tied to give the highest mean difference estimate, and the omission of the study by Melchior et al (47; mean 11.17 kJ/kg; 95% CI: 7.82, 14.52) gave a slightly lower mean estimate than did that of the study of Crenn et al (mean 11.40 kJ/kg; 95% CI: 7.90, 14.90). The estimated difference in energy expenditure between the highest and lowest studies was small (82 kJ/d), and, once again, the data of Kotler et al contributed substantially to the between-study variation (²), with the ² value dropping to 39.21 (78% of the combined estimate) on removal of this study.

View this table:
TABLE 5. Influence analysis, showing the effect on the overall mean and ² as each study in turn is removed and replaced¹

 

DISCUSSION  
The primary aim of this research was to identify an overall mean difference in REE between HIV-positive and healthy control subjects. Although there is substantial variation in the literature, the analyses presented show that overall REE/FFM is higher in HIV-positive subjects than in healthy controls by an estimated 630-661 kJ/d, or 9%. Secondary analyses suggest that subjects symptomatic for AIDS may contribute to the heterogeneity, but the number of studies including these subjects was too small to investigate in formal analyses using metaregression. The presence of lipodystrophy in the HIV-positive group, the method of body-composition measurement used, the use of HAART in the HIV-positive group, and the average age of the subjects were unable to explain the variation by metaregression.

Before the introduction of HAART, elevations of REE in HIV-positive subjects were postulated to be the result of increased cytokine activity, particularly during symptomatic disease (76) or weight loss (38) or the result of increased viral activity as indicated by HIV RNA viral loads (37, 63). Research was conflicting, however, with some studies finding no relation between REE and cytokines (54) or viral load (65, 87); difficulties in measuring cytokine activity (38, 55) and inconsistencies in additional confounding variables in the viral load models may account for these discrepancies. Since the introduction of HAART, 2 main factors have been causally implicated in increasing REE in HIV—lipodystrophy and HAART use per se.

It it important that this study shows that REE/FFM is higher in HIV-positive subjects with lipodystrophy than in healthy control subjects, but there is no significant difference between HIV-positive subjects with lipodystrophy and weight-stable HIV-positive subjects without lipodystrophy. Differences in antiretroviral regimens, mean subject ages, and other patient factors that are related to lipodystrophy (6, 98) could not be considered in this subanalysis because of the small sample size. Definitions of lipodystrophy in the various studies differed, and the recent development of a case definition of lipodystrophy (99) may provide more homogeneous results in future research. The validity of bioelectrical impedance analysis in subjects with lipodystrophy has been challenged (100), and, again, the sample size did not allow a metaregression on the effect of body-composition measurement methods in the lipodystrophy studies alone.

Shevitz et al (37) reported that REE adjusted for FFM, age, CD4 cell count, and HIV RNA viral load was 339 kJ/d higher in HIV-positive subjects using HAART than in those not using HAART; this difference was significant. On the basis of the present meta-analysis, this difference was estimated to be 216 kJ/d between studies in which subjects were using HAART and those in which subjects were not using HAART. This difference was not significant in the metaregression, but only 5 studies included subjects with HAART, and further research may allow an update to this analysis and may show a significant effect. The results of the study by Shevitz et al (37) must be interpreted cautiously because the study protocol did not require subjects to be measured after an overnight fast. The 2 longitudinal studies that have provided data on changes in REE with the introduction of protease inhibitors found no significant changes in REE over a median of 62 d (67) or 4.8 mo (82); longer-term data with adjustment for body-composition measurement method may be required to ascertain whether antiretroviral therapy per se affects REE.

Both classical and Bayesian methods of analysis were included in this meta-analysis. Bayesian approaches overcome the 2 main criticisms of classical random-effects analysis, because heavier-tailed t distributions can be used to avoid the normality assumption (10), and all uncertainty is modeled in the estimate of the mean difference (18, 101). The use of the t distribution in this analysis suggests that the between-study variation, ², was sensitive to the normality assumption and that it supported the use of the t distribution estimates in this case. Formal tests of normality in meta-analysis rely on large sample sizes and hence were not conducted in this analysis (102).

There is some evidence that the initial study of Kotler et al (3) is an outlier, because the extreme effect size of that study accounted for 25% of the between-study variation (²), although the effect on the mean difference is not clinically significant. Their study was 1 of 2 included in which REE was not measured after an overnight fast but instead after a standardized protocol; however, the available evidence suggests this would increase rather than decrease REE/FFM (11) and thus is unlikely to explain the extreme nature of this result. Although the tests for publication bias were not significant, these tests have low power (15), and the extreme nature of the effect size of the study by Kotler et al may indicate there is some bias in publishing studies showing that REE is lower in HIV-positive subjects. The visual inspection should be interpreted with caution because the sample size is small (15), and further data are necessary to ascertain whether the data from the study of Kotler et al provided evidence of publication bias.

Analysis of covariance–adjusted estimates are the correct way to express the relation between REE and FFM because they avoid the mathematical artifacts associated with non-zero intercepts when REE is divided by FFM (23). Although several authors have acknowledged this error (53, 66, 85, 87), the unadjusted estimate (REE/FFM) was more common and was therefore used for this analysis. An ancillary analysis of covariance was conducted on the available individual data, and it supported the findings of this meta-analysis. Fat mass (87), waist-to-hip ratio (8), HIV RNA viral load (37, 63), CD4 cell count (37), malabsorption (66), and the use of antiretrovirals (37) have also been shown to be associated with REE in HIV-positive subjects; none of these factors could be incorporated into the current analysis because of inadequate data. Age (103), sex (104), dietary intakes (2), smoking status (105), and physical activity level (106) have all been shown to affect REE in healthy subjects, but of these only age could be considered in the present analysis.

The subgroup analyses must be interpreted with caution because multiple comparisons were not adjusted for in the Forest plots, and the incorporation of single-arm studies in a mixed indirect comparison can provide a guide only to the results that could be expected from a direct comparison. Nevertheless, these analyses provide hypotheses for future research and a meaningful way to use the data when data from only one group of interest are available or when sample sizes are too small for formal analysis using metaregression, or both.

The decision to exclude studies with nonstandardized postprandial protocols resulted in the exclusion of 2 of the largest published studies (37, 38), but the available literature shows that diet-induced thermogenesis is altered in HIV infection (11), and estimates from these studies would have been influenced by this alteration. Moreover, these studies did not have a healthy control group and would not have been included in the primary analysis.

In conclusion, this meta-analysis shows that REE/FFM is significantly higher in HIV-positive subjects than in healthy control subjects. Within the HIV-positive population, there is evidence that persons with symptomatic infection have significantly higher REE/FFM than do other HIV-positive persons; however, further studies are required to confirm these subgroup analyses. Additional studies may also provide further data from which to investigate the effect of HAART on REE in HIV. The research published to date suggests that REE/FFM in persons with lipodystrophy does not differ significantly from that in other weight-stable HIV-positive persons.

ACKNOWLEDGMENTS  
I acknowledge the authors who responded to the requests for data or sample estimates for this meta-analysis.

MB conceptualized the study, obtained and analyzed the data, and wrote the manuscript. The author had no financial or personal conflicts of interest.

REFERENCES  

1. Chlebowski RT. Significance of altered nutritional status in acquired immune deficiency syndrome (AIDS). Nutr Cancer 1985;7:85–91.
2. Newsholme EA, Leech AR. Biochemistry for the medical sciences. Chichester, United Kingdom: Wiley & Sons, 1988:536–61.
3. Kotler DP, Tierney AR, Brenner SK, Couture S, Wang J, Pierson-RN J. Preservation of short-term energy balance in clinically stable patients with AIDS. Am J Clin Nutr 1990;51:7–13.
4. Hommes MJ, Romijn JA, Godfried MH, et al. Increased resting energy expenditure in human immunodeficiency virus-infected men. Metabolism 1990;39:1186–90.
5. Cunningham JJ. Body composition as a determinant of energy expenditure: a synthetic review and a proposed general prediction equation. Am J Clin Nutr 1991;54:963–9.
6. Carr A, Samaras K, Burton S, et al. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS 1998;12:F51–8.
7. Battegay M, Hirsch HH. Lipodystrophy syndrome by HAART in HIV-infected patients: manifestation, mechanisms and management. Infection 2002;30:293–8.
8. Kosmiski LA, Kuritzkes DR, Lichtenstein KA, et al. Fat distribution and metabolic changes are strongly correlated and energy expenditure is increased in the HIV lipodystrophy syndrome. AIDS 2001;15:1993–2000.
9. van der Valk M, Reiss P, van Leth FC, et al. Highly active antiretroviral therapy-induced lipodystrophy has minor effects on human immunodeficiency virus-induced changes in lipolysis, but normalizes resting energy expenditure. J Clin Endocrinol Metab 2002;87:5066–71.
10. Sutton AJ, Abrams KR. Bayesian methods in meta-analysis and evidence synthesis. Stat Methods Med Res 2001;10:277–303.
11. Poizot-Martin I, Benourine K, Philibert P, et al. Diet-induced thermogenesis in HIV infection. AIDS 1994;8:501–4.
12. Gotzsche PC. Meta-analysis of grip strength: most common, but superfluous variable in comparitive NSAID trials. Dan Med Bull 1989;36:493–5.
13. Astrup A, Gotzsche PC, van de Werken K, et al. Meta-analysis of resting metabolic rate in formerly obese subjects. Am J Clin Nutr 1999;69:1117–22.
14. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–86.
15. Sutton AJ, Abrams KR, Jones DR, Sheldon TA, Song F. Methods for meta-analysis in medical research. Chichester, United Kingdom: Wiley & Sons Ltd, 2000.
16. Normand S-LT. Meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321–59.
17. Bradburn MJ, Deeks JJ, Altman DG. Metan—an alternative meta-analysis command. Stata Tech Bull 1999;STB-44:sbe24.
18. Whitehead A. Meta-analysis of controlled clinical trials. Chichester, United Kingdom: Wiley & Sons, Ltd, 2002.
19. Bland JM, Altman DG. Statistics notes: Bayesians and frequentists. BMJ 1998;317:1151–60.
20. Spiegelhalter DJ, Abrams KR, Myles JP. Bayesian approaches to clinical trials and health-care evaluation. Chichester, United Kingdom: Wiley & Sons Ltd, 2004.
21. Congdon P. Bayesian statistical modelling. Chichester, United Kingdom: Wiley & Sons Ltd, 2001.
22. Spiegelhalter DJ, Thomas A, Best N, Gilks W. BUGS 0.5*Examples Volume 1 (version i), P33-36 Blocker: a random effects meta-analysis of clinical trials. Cambridge, United Kingdom: MRC Biostatistics Unit, Institute of Public Health, 1996.
23. Ravussin E, Bogardus C. Relationships of genetics, age, and physical fitness to daily energy expenditure and fuel utilization. Am J Clin Nutr 1989;49:968–75.
24. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58.
25. Sharp S. Meta-analysis regression. Stata Tech Bull 1998;STB-42:sbe23.
26. Light RJ, Pillemer DB. Summing up: the science of reviewing research. Boston: Havard University Press, 1984.
27. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088–101.
28. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–34.
29. Duval S, Tweedie R. A nonparametric "trim and fill" method of accounting for publication bias in meta-analysis. J Am Stat Assoc 2000;95:89–98.
30. Arpadi SM, Cuff PA, Kotler DP, et al. Growth velocity, fat-free mass and energy intake are inversely related to viral load in HIV-infected children. J Nutr 2000;130:2498–502.
31. Johann-Liang R, O'Neill L, Cervia J, et al. Energy balance, viral burden, insulin-like growth factor-1, interleukin-6 and growth impairment in children infected with human immunodeficiency virus. AIDS 2000;14:683–90.
32. Fox-Wheeler S, Heller L, Salata CM, et al. Evaluation of the effects of oxandrolone on malnourished HIV-positive pediatric patients. Pediatrics 1999;104:e73.
33. Cole CR, Rising R, Hakim A, et al. Comprehensive assessment of the components of energy expenditure in infants using a new infant respiratory chamber. J Am Coll Nutr 1999;18:233–41.
34. Henderson RA, Talusan K, Hutton N, Yolken RH, Caballero B. Resting energy expenditure and body composition in children with HIV infection. J Acquir Immune Defic Syndr Hum Retrovirol 1998;19:150–7.
35. Henderson RA, Talusan K, Hutton N, Yolken RH, Caballero B. Whole body protein turnover in children with human immunodeficiency virus (HIV) infection. Nutrition 1999;15:189–94.
36. Alfaro MP, Siegel RM, Baker RC, Heubi JE. Resting energy expenditure and body composition in pediatric HIV infection. Pediatr AIDS HIV Infect 1995;6:276–80.
37. Shevitz AH, Knox TA, Spiegelman D, Roubenoff R, Gorbach SL, Skolnik PR. Elevated resting energy expenditure among HIV-seropositive persons receiving highly active antiretroviral therapy. AIDS 1999;13:1351–7.
38. Roubenoff R, Grinspoon S, Skolnik PR, et al. Role of cytokines and testosterone in regulating lean body mass and resting energy expenditure in HIV-infected men. Am J Physiol Endocrinol Metab 2002;283:E138–45.
39. Grunfeld C, Pang M, Shimizu L, Shigenaga JK, Jensen P, Feingold KR. Resting energy expenditure, caloric intake, and short-term weight change in human immunodeficiency virus infection and the acquired immunodeficiency syndrome. Am J Clin Nutr 1992;55:455–60.
40. Anderson R, Grady C, Ropka M. A comparison of calculated energy requirements to measured resting energy expenditure in HIV-1-infected subjects. J Assoc Nurses AIDS Care 1994;5:30–4.
41. Schwarz JM, Mulligan K, Lee J, et al. Effects of recombinant human growth hormone on hepatic lipid and carbohydrate metabolism in HIV-infected patients with fat accumulation. J Clin Endocrinol Metab 2002;87:942.
42. Sheehan LA, Macallan DC. Determinants of energy intake and energy expenditure in HIV and AIDS. Nutrition 2000;16:101–6.
43. Slusarczyk R. The influence of the human immunodeficiency virus on resting energy expenditure. J Acquir Immune Defic Syndr 1994;7:1025–7.
44. Bowers JM, Scott RW, Ampel NM. Comparison of resting energy expenditure measured by indirect calorimetry and the Harris-Benedict equation in HIV-infected men of normal body weight. Respir Care 1997;42:1018–21.
45. Sharpstone DR, Murray CP, Ross HM, et al. Energy balance in asymptomatic HIV infection. AIDS 1996;10:1377–84.
46. Sharpstone D, Ross H, Hancock M, Phelan M, Crane R, Gazzard B. Indirect calorimetry, body composition and small bowel function in asymptomatic HIV-seropositive women. Int J STD AIDS 1997;8:700–3.
47. Melchior JC, Salmon D, Rigaud D, et al. Resting energy expenditure is increased in stable, malnourished HIV-infected patients. Am J Clin Nutr 1991;53:437–41.
48. Suttmann U, Muller MJ, Ockenga J, et al. Malnutrition and immune dysfunction in patients infected with human immunodeficiency virus. Klin Wochenschr 1991;69:156–62.
49. Hommes MJ, Romijn JA, Endert E, Sauerwein HP. Resting energy expenditure and substrate oxidation in human immunodeficiency virus (HIV)-infected asymptomatic men: HIV affects host metabolism in the early asymptomatic stage. Am J Clin Nutr 1991;54:311–5.
50. Melchior JC, Raguin G, Boulier A, et al. Resting energy expenditure in human immunodeficiency virus-infected patients: comparison between patients with and without secondary infections. Am J Clin Nutr 1993;57:614–9.
51. Suttmann U, Ockenga J, Hoogestraat L, et al. Resting energy expenditure and weight loss in human immunodeficiency virus-infected patients. Metabolism 1993;42:1173–79.
52. Mulligan K, Grunfeld C, Hellerstein MK, Neese RA, Schambelan M. Anabolic effects of recombinant human growth hormone in patients with wasting associated with human immunodeficiency virus infection. J Clin Endocrinol Metab 1993;77:956–62.
53. Macallan DC, Noble C, Baldwin C, et al. Energy expenditure and wasting in human immunodeficiency virus infection. N Engl J Med 1995;333:83–8.
54. Godfried MH, Romijn JA, van der Poll T, et al. Soluble receptors for tumor necrosis factor are markers for clinical course but not for major metabolic changes in human immunodeficiency virus infection. Metabolism 1995;44:1564–9.
55. Hellerstein MK, Wu K, McGrath M, et al. Effects of dietary n–3 fatty acid supplementation in men with weight loss associated with the acquired immune deficiency syndrome: relation to indices of cytokine production. J Acquir Immune Defic Syndr Hum Retrovirol 1996;11:258–70.
56. Suttmann U, Ockenga J, Schneider H, et al. Weight gain and increased concentrations of receptor proteins for tumor necrosis factor after patients with symptomatic HIV infection received fortified nutrition support. J Am Diet Assoc 1996;96:565–9.
57. Paton NI, Elia M, Jebb SA, Jennings G, Macallan DC, Griffin GE. Total energy expenditure and physical activity measured with the bicarbonate-urea method in patients with human immunodeficiency virus infection. Clin Sci Colch 1996;91:241–5.
58. Schwenk A, Hoffer-Belitz E, Jung B, et al. Resting energy expenditure, weight loss, and altered body composition in HIV infection. Nutrition 1996;12:595–601.
59. Haslett P, Hempstead M, Seidman C, et al. The metabolic and immunologic effects of short-term thalidomide treatment of patients infected with the human immunodeficiency virus. AIDS Res Hum Retroviruses 1997;13:1047–54.
60. Heijligenberg R, Romijn JA, Westerterp KR, Jonkers CF, Prins JM, Sauerwein HP. Total energy expenditure in human immunodeficiency virus-infected men and healthy controls. Metabolism 1997;46:1324–6.
61. Sharpstone DR, Ross HM, Gazzard BG. The metabolic response to opportunistic infections in AIDS. AIDS 1996;10:1529–33.
62. Hoh R, Pelfini A, Neese RA, et al. De novo lipogenesis predicts short-term body-composition response by bioelectrical impedance analysis to oral nutritional supplements in HIV-associated wasting. Am J Clin Nutr 1998;68:154–63.
63. Mulligan K, Tai VW, Schambelan M. Energy expenditure in human immunodeficiency virus infection. N Engl J Med 1997;336:70–1 (letter).
64. Mulligan K, Tai VW, Schambelan M. Effects of chronic growth hormone treatment on energy intake and resting energy metabolism in patients with human immunodeficiency virus-associated wasting—a clinical research center study. J Clin Endocrinol Metab 1998;83:1542–7.
65. Grinspoon S, Corcoran C, Miller K, et al. Determinants of increased energy expenditure in HIV-infected women. Am J Clin Nutr 1998;68:720–5.
66. Jimenez-Exposito MJ, Garcia-Lorda P, Alonso-Villaverde C, et al. Effect of malabsorption on nutritional status and resting energy expenditure in HIV-infected patients. AIDS 1998;12:1965–72.
67. Carbonnel F, Maslo C, Beaugerie L, et al. Effect of indinavir on HIV-related wasting. AIDS 1998;12:1777–84.
68. Strawford A, Barbieri T, Neese RA, et al. Effects of nandrolone decanoate therapy in borderline hypogonadal men with HIV-associated weight loss. J Acquir Immune Defic Syndr Hum Retrovirol 1999;20:137–46.
69. Strawford A, Barbieri T, Van Loan M, et al. Resistance exercise and supraphysiologic androgen therapy in eugonadal men with HIV-related weight loss. JAMA 1999;281:1282–90.
70. Sharpstone D, Murray C, Ross H, et al. The influence of nutritional and metabolic status on progression from asymptomatic HIV infection to AIDS-defining diagnosis. AIDS 1999;12:1221–6.
71. Renard E, Fabre J, Paris F, Reynes J, Bringer J. A syndrome of body fat redistribution in HIV-1-infected patients: relationships to cortisol and catecholamines. Clin Endocrinol 1999;51:223–30.
72. Hardin DS, LeBlanc A, Young D, Johnson P. Increased leucine turnover and insulin resistance in men with advanced HIV infection. J Investig Med 1999;47:405–13.
73. Melchior JC, Niyongabo T, Henzel D, Durack-Bown I, Henri SC, Boulier A. Malnutrition and wasting, immunodepression, and chronic inflammation as independent predictors of survival in HIV-infected patients. Nutrition 1999;15:865–9.
74. Pernerstorfer-Schoen H, Schindler K, Parschalk B, et al. Beneficial effects of protease inhibitors on body composition and energy expenditure: a comparison between HIV-infected and AIDS patients. AIDS 1999;13:2389–96.
75. Lane BJ, Provost-Craig MA. Resting energy expenditure in asymptomatic HIV-infected females. J Womens Health Gend Based Med 2000;9:321–7.
76. Suttmann U, Holtmannspotter M, Ockenga J, Gallati H, Deicher H, Selberg O. Tumor necrosis factor, interleukin-6, and epinephrine are associated with hypermetabolism in AIDS patients with acute opportunistic infections. Ann Nutr Metab 2000;44:43–53.
77. Sharpstone D, Phelan M, Gazzard B. Differential metabolic response in AIDS-related chronic protozoal diarrhoea. HIV Med 2000;1:102–6.
78. Berneis K, Battegay M, Bassetti S, et al. Nutritional supplements combined with dietary counselling diminish whole body protein catabolism in HIV-infected patients. Eur J Clin Invest 2000;30:87–94.
79. Coors M, Suttmann U, Trimborn P, Ockenga J, Muller MJ, Selberg O. Acute phase response and energy balance in stable human immunodeficiency virus-infected patients: a doubly labeled water study. J Lab Clin Med 2001;138:94–100.
80. Sattler FR, Schroeder ET, Dube MP, et al. Metabolic effects of nandrolone decanoate and resistance training in men with HIV. Am J Physiol Endocrinol Metab 2002;283:E1214–22.
81. Hadigan C, Borgonha S, Rabe J, Young V, Grinspoon S. Increased rates of lipolysis among human immunodeficiency virus-infected men receiving highly active antiretroviral therapy. Metabolism 2002;51:1143–7.
82. Korach M, Leclercq P, Peronnet F, Leverve X. Metabolic response to a C-glucose load in human immunodeficiency virus patients before and after antiprotease therapy. Metabolism 2002;51:307–13.
83. Sekhar RV, Jahoor F, White AC, et al. Metabolic basis of HIV-lipodystrophy syndrome. Am J Physiol Endocrinol Metab 2002;283:E332–7.
84. Behrens GMN, Boerner A-R, Weber K, et al. Impaired glucose phosphorylation and transport in skeletal muscle cause insulin resistance in HIV-1-infected patients with lipodystrophy. J Clin Invest 2002;110:1319–27.
85. Kosmiski L, Kuritzkes D, Lichtenstein KA, et al. Total energy expenditure and carbohydrate oxidation are increased in the human immunodeficiency virus lipodystrophy syndrome. Metabolism 2003;52:620–5.
86. Luzi L, Perseghin G, Tambussi G, et al. Intramyocellular lipid accumulation and reduced whole-body lipid oxidation in HIV-infected patients with lipodystrophy. Am J Physiol 2003;284:E274–80.
87. Batterham MJ, Morgan-Jones J, Greenop P, Garsia R, Gold J, Caterson I. Calculating energy requirements in men with HIV/AIDS in the era of highly active antiretroviral therapy. Eur J Clin Nutr 2003;57:209–17.
88. McNurlan MA, Garlick PJ, Steigbigel RT, et al. Responsiveness of muscle protein synthesis to growth hormone administration in HIV-infected individuals declines with severity of disease. J Clin Invest 1997;100:2125–32.
89. Andersen O, Haugaard SB, Anderson UB, et al. Lipodystrophy in human immunodeficiency virus patients impairs insulin action and induces defects in B-cell function. Metabolism 2003;52:1343–53.
90. Yarasheski KE, Zachwieja JJ, Gischler J, Crowley J, Horgan MM, Powderly WG. Increased plasma Gln and Leu R_a and inappropriately low muscle protein synthesis rate in AIDS wasting. Am J Physiol Endocrinol Metab 1998;275:E577–83.
91. Charlin V, Carrasco F, Sepulveda C, Torres M, Kehr J. Nutritional supplementation according to energy and protein requirements in malnourished HIV-infected patients. Arch Latinoam Nutr 2002;52:267–73.
92. Salehian B, Niyongabo T, Salmon D, et al. Tumor necrosis factor and resting energy expenditure during the acquired immunodeficiency syndrome. Am J Clin Nutr 1993;58:715–6.
93. Crenn P, Rakotoanbinina B, Raynaud J-J, Thuillier F, Messing B, Melchior JC. Hyperphagia contributes to the normal body composition and protein-energy balance in HIV-infected asymptomatic men. J Nutr 2004;134:2301–6.
94. Garcia-Lorda P, Serrano P, Jimenez-Exposito MJ, et al. Cytokine-driven inflammatory response is associated with the hypermetabolism of AIDS patients with opportunistic infections. JPEN J Parenter Enteral Nutr 2000;24:317–22.
95. Garcia Luna P, Aguayo P, Jimenez Exposito M, Florit A, Garcia Lorda P, Salas Salvado J. Hypermetabolism and progression of HIV infection. Am J Clin Nutr 1999;70:299–303.
96. Hommes MJ, Romijn JA, Endert E, Eeftinck SJ, Sauerwein HP. Basal fuel homoeostasis in symptomatic human immunodeficiency virus infection. Clin Sci Colch 1991;80:359–65.
97. Schindler K, Pernerstorfer-Schoen H, Schneider B, Rieger A, Elmadfa I. Positive impact of protease inhibitors on body composition and energy expenditure in HIV-infected and AIDS patients. Ann N Y Acad Sci 2000;904:603–6.
98. Carr A, Samaras K, Thorisdottir A, Kaufmann GR, Chisholm DJ, Cooper DA. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet 1999;353:2093–9.
99. Carr A, Law M, Group HLCDS. An objective lipodystrophy severity grading scale derived from the lipodystrophy case definition score. J Acquir Immune Defic Syndr Hum Retrovirol 2003;33:571–6.
100. Schwenk A, Breuer P, Kremer G, Ward L. Clinical assessment of HIV-associated lipodystrophy syndrome: bioelectrical impedance analysis, anthropometry and clinical scores. Clin Nutr 2001;20:243–9.
101. Brockwell SE, Gordon IR. A comparison of statistical methods for meta-analysis. Stat Med 2001;20:825–40.
102. Hardy R, Thompson S. Detecting and describing heterogeneity in meta-analysis. Stat Med 1998;17:841–56.
103. Das SK, Moriguti JC, McCrory MA, et al. An underfeeding study in healthy men and women provides further evidence of impaired regulation of energy expenditure in old age. J Nutr 2001;131:1833–8.
104. Buchholz AC, Rafii M, Pencharz PB. Is resting metabolic rate different between men and women? Br J Nutr 2001;86:641–6.
105. Dallosso HM, James WP. The role of smoking in the regulation of energy balance. Int J Obes 1984;8:368–75.
106. Wadden TA, Vogt RA, Andersen RE, et al. Exercise in the treatment of obesity: effects of four interventions on body composition, resting energy expenditure, appetite, and mood. J Consult Clin Psychol 1997;65:269–77.