Body Composition and Dietary Intake in Relation to Drug Abuse in a Cohort of HIV-Positive Persons

We examined the relationships between drug abuse, weight, body composition, and dietary intake in persons infected with HIV in a cross-sectional analysis of baseline data from a longitudinal study of nutritional status and HIV. Body composition was measured by bioelectrical impedance analysis. Dietary data were collected by 3-day food records or 24-hour recalls. We analyzed data from 39 current intravenous drug users (IVDU), 103 past intravenous drug users (past-IVDU), 239 users of nonintravenous drugs (users-NIVD), and 61 nonusers (reference category). In the men, there were no differences in weight, body mass index (BMI), or body composition among the drug-use groups. In the women, there was a trend to lower weight and BMI across the drug use categories: IVDU women had lower average weight (-13.7 kg;p = .006), BMI (-5.6 units;p = .003) and less fat mass than non-users (-9.8 kg;p = .0001). In women, drug users had higher weight-adjusted energy intakes than nonusers, whereas in the men both drug using groups, NIVD and IVDU, had higher energy intakes than nonusers. These data suggest that intravenous drug-abuse is associated with lower weight and fat mass in women with HIV infection despite adequate self-reported energy intake.

Weight loss is a common problem in people infected with HIV. According the U.S. Centers for Disease Control and Prevention (CDC), wasting was the second most frequently reported AIDS-defining condition in the United States in 1997 ⁽¹⁾. Drug abuse may contribute to HIV-associated weight loss. Smit et al. ⁽²⁾ examined injection drug abusers with and without HIV infection. Despite reported dietary intakes in excess of 1000 kilocalories > U.S. National Health And Nutrition Examination Survey (NHANES) estimates, 5% of the HIV-negative drug abusers reported involuntary weight loss of > than 10 lb. The frequency of involuntary weight loss was even greater among HIV-positive drug abusers (16%). Morabia et al. ⁽³⁾, in Geneva, Switzerland, reported lower body mass index (BMI) in male heroin addicts compared with findings in nonaddicts.

Although these studies suggest that low weight is a common problem among HIV-infected drug abusers, we could not identify studies that have examined body composition. The loss of lean body mass is of importance in HIV infection since it is believed to be a better predictor of mortality than weight loss ^(4,5). Therefore, we examined weight, BMI, and body composition in relation to drug abuse in a cohort of HIV-positive persons. We also report data on the dietary intake of macronutrients and micronutrients in relation to drug abuse.

METHODS

Participants

Data were taken from the baseline visits of the first 588 participants to enroll in the Nutrition for Healthy Living Study, a longitudinal study of the role of nutrition in HIV disease. Recruitment of participants was done through advertisements on radio, in local newspapers, health clinics, and physician networks. Persons 18 years of age and older who were HIV positive and consented to participate were considered eligible. People who met any one of the following criteria were excluded: pregnant at recruitment, diagnosed with diabetes mellitus, thyroid disease, or malignancies (other than HIV-associated malignancies), inadequate fluency in English, or refusal to sign a consent form for release of medical records. Clinic visits occurred at three sites: New England Medical Center and Fenway Community Health Center in Boston, Massachusetts and the Miriam Hospital in Providence, Rhode Island. Participants were seen twice yearly at one of our participating clinics: the New England Medical Center or the Fenway Community Health Center in Boston, or the Miriam Hospital in Providence, Rhode Island.

Assessments: Height (±0.1 cm), weight (±0.1 kg) and body composition were measured using standardized techniques. Body mass index (BMI) was calculated by weight (kg)/height² (m²). Body composition was measured by bioelectrical impedance analysis (BIA) following the manufacturer's recommended procedures (RJL Systems, Clinton, MI, U.S.A.), and using the equations of Lukaski et al. ⁽⁶⁾ to derive measures of fat-free mass (principally lean body mass and bone). Participants were asked to keep a 3-day food record during the week before each visit. The food items were converted to nutrients using the Nutrition Data System for Research (NDS-R) software, versions 2.7 through 4.0 ⁽⁷⁾. If a participant was unable to keep a 3-day food record, the study nutritionist collected dietary information by 24-hour recall. The drug-use status of each participant was determined by standardized interview. Each participant was placed into one of four categories of drug abuse: Those who had never used drugs (nonusers), users of drugs other than intravenous drugs (users-NIVD), past intravenous drug users (past-IVDUs), and current intravenous drug users (IVDUs). For the tests of trend, the drug-use data were treated as ordinal with the rank of drug abuse from lowest to highest being: nonusers, users-NIVD, past-IVDUs, and IVDUs. The rank of drug-use categories was established before the analyses and was based on our perception of the probable effect of the grade of drug abuse on the outcomes examined. Past IVDU was considered to be a higher level of drug use than current use of non-IV drugs, given that the latter category included occasional use of club drugs such as ecstasy, inhalants, and abuse of prescription drugs. On the assumption that there would be under-reporting of drug abuse at the first clinic visit and that drug abuse habits are stable over time, we used data from all available interviews (baseline and follow-up visits) to classify drug abuse, and each participant was assigned to the highest rank of drug abuse ever reported. For example, if the participant was classified as a past-IVDU at the baseline interview and a user-NIVD in an interview during a follow-up visit, the participant was assigned to the past-IVDU category. Marijuana-only users were classified with the nonusers because marijuana is frequently used as an appetite stimulant among persons infected with HIV. Information collected at follow-up visits was used only for the categorization of drug use. All other data were taken from the baseline visit only.

Statistical Analyses

Cross-sectional analyses of the data were done using linear regression models in SAS, version 6.12 (SAS Institute, Cary, NC, U.S.A.) with robust variance estimators to allow for valid inferences even when regression residuals are nonnormal ^(8,9). Analyses of between-group differences used reference-cell coding, with the nonuser group as the reference category. Separate models were run for each of the outcomes. Trends in weight, BMI, fat mass, and fat-free mass across the drug use categories were tested by assigning each subject a score as follows: one (nonusers), two (users-NIVD), three (past-IVDUs) or four (IVDUs), and each outcome was regressed on the score. We controlled for age, HIV disease stage using CD4 cell counts, and the current use (yes/no) of antiretroviral therapy, protease inhibitors, and anabolic steroids in all analyses. Because micronutrient intake and percentage of calories consumed from alcohol were highly skewed, we reported the median intake, first, and third quartiles. Between-group differences in micronutrient intakes were tested using the natural log transformation of the micronutrient values and robust variance estimators. For this reason, we did not report confidence intervals (CIs) for the between-group differences in micronutrient intake. We checked all models for violations of assumptions by examination of residuals. Influential observations were checked by Cook's distance (SAS generalized linear models program), but none was found. All CIs are reported at the 95% level. The two-tailed α was set at 0.05.

RESULTS

Of the first 588 participants recruited to the study, we were able to classify drug abuse for 502 participants. Of these 502 participants, data on HIV-medications (used as covariates) were missing for 54 participants. Most (79%) of these data were missing because the participants were concurrently enrolled in blinded clinical trials. In addition, 6 women were missing information on body composition. Thus, there were complete data on weight and body composition, including covariates, for 442 (75%) participants (312 men and 130 women). Three additional women were missing information on dietary intake.

Tables 1 and 2 show the numbers of men and women in each drug use category. Drug users tended to have lower education and lower income levels than nonusers. The rate of completion of the 3-day food record was lowest among IVDU men and user-NIVD women.

Table 1

Table 2

In the men, no significant between-group differences in weight, BMI, or measures of body composition were found (Table 1). In contrast, among women (Table 2), there were trends toward lower weight (p value for test for trend = .04) and lower BMI (p value for tests for trend = .04) with increasing drug use category. Average weight of IVDU women was 13.7 kg < that of nonusers (95% CI, -23.4 to -4.1;p = .006) and average BMI was 5.6 units less (95% CI, -9.2 to -1.9;p = .003) than the BMI of nonusers. The differences in weight were accounted for by differences in body fat. Body fat decreased with increasing drug use (p value for test for trend = .01). IVDU women had an average of 9.8 kg less body fat than nonusers (95% CI, -14.7 to -4.9;p = .0001). There was a nonsignificant trend (p = .2) to lower fat-free mass with increasing degree of drug use. Race (black versus other) was evaluated as a confounder, but not retained in the models.

Male drug users, both NIVD and IVDU, had greater weight-adjusted energy intakes than nonusers (difference from nonusers respectively: 3.8 kcal/kg; 95% CI, 0.13-7.4;p = .04; 9.8 kcal/kg, 95% CI, 2.5-17.0;p = .009) (Table 3). The IVDU group had a 30% higher intake of carbohydrates (p = .004) and 35% higher fat intake (p = .008) than nonusers. When these differences in macronutrient intake were adjusted for energy, the past-IVDU group had a lower average intake of carbohydrate (-25.5 g; 95% CI, -50.1 to -0.8];p = .04) and higher average intake of fat (10.0 g; 95% CI, 1.6-18.3;p = .02) than nonusers (Table 3).

Table 3

In the women, weight-adjusted energy intakes were greater in all the drug-user groups compared with those found in nonusers (nonusers versus users-NIVD, 7.3 kcal/kg; 95% CI, 0.3-14.5];p = .04; versus past IVDU: 6.9 kcal/kg; 95% CI, 0.7-13.0;p = .03; versus IVDUs: 14.7 kcal/kg; 95% CI, 4.5-24.9];p = .006). These differences in energy intake were accounted for by greater intakes of protein among IVDUs (0.46 g/kg, 95% CI, 0.087-0.83;p = .02), and fat among the user-NIVDs (23.6 g; 95% CI, 2.4-44.9];p = .03) and past-IVDUs (difference, 21.0 g; 95% CI, 5.2-36.8;p = .004). Although the estimated fat intake in IVDUs was greater in magnitude than that estimated in past-IVDUs, this difference did not reach statistical significance at the 0.05 level (p = .12). Similar to findings among the respective groups of men, user-NIVDs and IVDUs appeared to have higher levels of carbohydrate intake compared with nonusers, but these differences were not statistically significant in the women (Table 4).

Table 4

We examined differences between the nonusers and three drug user categories in dietary intake (including supplements) of the micronutrients, thiamine, niacin, riboflavin, vitamins A, B₆, B₁₂, C, E, folate, iron, and zinc. In the men, compared with nonusers, users-NIVD had significantly higher non-energy-adjusted intakes of thiamine (p = .004), niacin (p = .001), riboflavin (p = .01), B₆ (p = .01), and folate (p = .048). When the between-group differences were adjusted for energy, the differences between the nonuser men and user-NIVDs remained significant for thiamine (p = .01), niacin (p = .006), riboflavin (p = .03) and vitamin B₆ (p = .04). IVDUs had significantly lower intakes of vitamin C (p = .004), vitamin E (p = .002), folate (p = .03), iron (p = .01) and zinc (p = .01) than the nonusers (Table 5).

Table 5

Among the women, IVDUs had significantly greater (nonenergy-adjusted) intakes of vitamin A than nonusers (p = .03). After adjustment for energy, intakes of thiamine (p = .037), niacin (p = .048) and B₆ (p = .043) were significantly lower in the past-IVDU group compared with nonusers (Table 6). We found no evidence of confounding by education and income level.

Table 6

DISCUSSION

Among the women with HIV infection in this study, those who use drugs were more likely to have lower weight and BMI than women who have never used drugs. The differences in weight between the women who use drugs and the nonusers were due, largely, to differences in fat mass rather than differences in fat-free mass. The women who reported using drugs also reported higher energy intakes than non-user women despite evidence of lower weights. These findings are similar to those of Smit et al. ⁽²⁾. These authors found self-reported involuntary weight loss among HIV-positive IVDUs, whereas reported energy intakes were in excess of estimated needs. Frequent drug abuse may alter energy requirements, although, to our knowledge, no mechanism by which this could occur has been proposed. It is also possible that the dietary intakes reported in this study do not reflect average intake, and that drug users have days during which energy intake is less than energy expenditure. Dietary intake on days when drug use is minimal might be higher than the intake of nonusers to compensate for days when energy intake is inadequate. Lower energy intakes were reported to be associated with heavy drug use among IVDUs (82% heroin users) in Spain ⁽¹⁰⁾. In contrast, Morabia et al. ⁽³⁾ reported nutrient deficits without energy deficits in male heroin addicts compared with nonaddicts. Heroin use was associated with replacement of protein and fat for carbohydrates, principally sucrose and alcohol, with deficiencies in vitamins and minerals. Although we found evidence of increased intake of carbohydrates and fat among men who use drugs or who reported past IVDU, these data were not as clear for female IVDUs, in whom intakes of all macronutrients tended to be higher. There were no clear differences in the percentage calories derived from alcohol ingestion among these groups. However, we did not specifically ask about alcohol intake, so these data likely reflect under-reporting across all groups.

It is not clear why we did not find between-group differences in weight and body composition among the men. It is possible that the economic consequences of drug abuse are greater for women than men. Because women are more likely to have the care of children, personal income may not reflect disposable income equally across gender. It is also possible that there are gender-based differences in the type and frequency of drugs used. For example, non-IV drugs in our analyses included drugs such as poppers (nitrite inhalants) and ecstasy (methylene doxymethamphetamine, MDMA) that are commonly used in the gay-male culture of the United States, but may be less commonly used among women. Cocaine abuse can lead to food deprivation and nutrient imbalance ⁽¹¹⁾, and cocaine abusers are reported to have a high prevalence of eating disorders ⁽¹²⁾. The design of the questions relating to drug abuse used in this study did not allow us to isolate, in detail, the types and frequencies of drugs abused.

The men had median intakes of micronutrients above the recommended dietary allowance (RDA) for all micronutrients except zinc, for which male IVDU and past-IVDU had average intakes just below the RDA of 15 mg. The nonuser, user-NIVD, and past-IVDU women also had zinc intakes below the RDA (12 mg). As with the macronutrients, these intakes may not reflect average intakes. The significantly lower levels of (energy-adjusted) intakes of vitamins C and E, folate, iron, and zinc in IVDU versus nonuser men suggests that the diet of male IVDUs, although apparently higher in energy, is lower in nutrient density than the diet of nonusers. This observation is similar to that of Morabia et al. ⁽³⁾ in their study of heroin users.

The estimated dietary intakes in this study are self-reported, based on 3-day food records or 24-hour recalls. Dietary data collected by 24-hour recall are considered of a lower quality than food record data and may not reflect usual intake. Studies comparing food record data with estimates of energy requirements suggest that underreporting of intake in food records is common ⁽¹³⁾ and may be biased by characteristics of the reporter. For example, obese people are believed to underreport their true intake more than people of normal weight ⁽¹⁴⁾. Although this could explain the differences in between-group energy intakes in the women, it cannot explain the high absolute energy intake reported by current IVDUs.

Differences in the use of marijuana might explain the differences in energy intake between the groups, but marijuana use is unlikely to explain differences in weight and body composition. Dronabinol, a marijuana derivative, can increase appetite and stabilize weight in HIV patients, but its use has not been shown to lead to weight gain ^(15,16).

In summary, these data provide evidence that in women with HIV infection, those whose use intravenous drugs have lower weight and fat mass than women who do not use drugs, despite reported adequacy in energy and micronutrient intakes. Thus, HIV-positive women who use intravenous drugs may be at higher risk of developing the AIDS-wasting syndrome.

More studies of the contribution of drug abuse to weight loss and the progression of HIV disease are needed. Better methods to measure usual nutrient intake in drug users need to be designed. Studies done in populations that use drugs will require additional resources.

Acknowledgments:

This study was financially supported by NIDDK grant DK4 5734-03. The General Clinical Research Center of the New England Medical Center, Boston, Massachusetts is financially supported by the Division of Research Resources of the NIH, grant number MO1-RR00054. The authors thank the participants for their time and enthusiasm, and thank the data collection team for their diligence.