1. Objectives

To determine the effects of oral nutritional supplements in people with active tuberculosis being treated with anti-tuberculosis drug therapy

2. How studies were identified

The following databases were searched in February 2016:

• Cochrane Infectious Diseases Group Specialized Register
• CENTRAL (The Cochrane Library 2016, Issue 1)
• MEDLINE
• EMBASE
• LILACS
• mRCT
• WHO International Clinical Trials Registry Platform

Reference lists and the Indian Journal of Tuberculosis were also hand-searched

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials

3.2 Study participants

Children or adults receiving treatment for active tuberculosis, regardless of HIV or nutritional status

3.3 Interventions

Oral macro- or micronutrient supplements given for a minimum of four weeks, compared with no intervention, placebo, or dietary advice alone

3.4 Primary outcomes

• All-cause mortality
• Cure (completed treatment and sputum-smear or sputum-culture negative) at six and 12 months

Secondary outcomes included completion of treatment, sputum positive at follow-up, self-reported recovery from illness or resolution of symptoms, change in lean or total mass (weight, skinfold thickness), growth in children, physical functioning, quality of life, ability to return to work, calorie intake, and micronutrient concentrations before and after supplementation

4. Main results

4.1 Included studies

Thirty-five randomized controlled trials, enrolling 8285 participants, were included in this review

• Thirty-one trials were in adults; 28 of these studies recruited adults with pulmonary tuberculosis and three trials also included adults with extrapulmonary tuberculosis
• Four trials were in children; one trial was restricted to children with pulmonary tuberculosis and three trials also included children with extrapulmonary tuberculosis
• Eleven trials specifically recruited participants with HIV infection and 11 trials excluded those with HIV infection. Most participants with HIV infection were not receiving anti-retroviral therapy
• Twenty-one trials used single or dual micronutrient supplements, seven trials used macronutrient supplements (daily meals, high-energy supplements, monthly rations), six trials used multi-micronutrient supplements providing one to 10 times the dietary reference intake of each micronutrient, and one trial used capsules containing 6 g of fresh-water fish (Channa striata)

4.2 Study settings

• Bangladesh, Egypt, Ethiopia (2 trials), Georgia, Guinea-Bissau, India (7 trials), Indonesia (5 trials), Iran (2 trials), Malawi, Mexico (2 trials), Nigeria, Singapore, South Africa (2 trials), Timor-Leste, the United Republic of Tanzania (6 trials), and the United Kingdom of Great Britain and Northern Ireland
• Studies were conducted in both inpatient and outpatient settings

4.3 Study settings

How the data were analysed
Nine comparisons were made: i) increased energy supplementation versus standard care/nutritional advice/no supplement, ii) high-cholesterol diet versus low cholesterol diet, iii) multivitamin and trace element supplements versus placebo, iv) vitamin A versus placebo, v) zinc versus placebo, vi) zinc plus vitamin A versus placebo, vii) vitamin D versus placebo or no supplement, viii) arginine versus placebo, and xi) vitamin E plus selenium versus placebo. Risk ratios (RR) were generated for dichotomous data and mean differences (MD) were produced for continuous data, and corresponding 95% confidence intervals (CI) were presented with all results. Where substantial heterogeneity was detected (I²>50%), random effects meta-analysis was used. Sensitivity analyses were not conducted due to insufficient data, and subgroup analyses were conducted by HIV status.

Results
Increased energy supplementation versus standard care/nutritional advice/no intervention
All-cause mortality and cure (completed treatment and sputum-smear or sputum-culture negative)
Pooled analysis of four trials involving 567 participants found no statistically significant effect of macronutrient supplementation in comparison to no macronutrient supplementation on the risk of all-cause mortality, although few events (n=10) occurred (RR 0.34, 95% CI [0.10 to 1.20]). In one trial involving 102 participants, no treatment effect was found on tuberculosis cure at six months (RR 0.91, 95% CI [0.59 to 1.41]). Results were not altered meaningfully for either primary outcome when subgrouped by HIV status (death, HIV negative: RR 0.18, 95% CI [0.02 to 1.48], 3 trials/230 individuals; death, HIV positive: RR 0.60, 95% CI [0.12 to 3.01], 2 trials/72 individuals; cure, HIV negative: RR 0.85, 95% CI [0.53 to 1.35], 1 trial/80 participants; cure, HIV positive: RR 1.38, 95% CI [0.46 to 4.14], 1 trial/22 individuals).

Additional outcomes
No statistically significant difference was found between intervention and control groups in pooled analysis of the outcomes treatment completion and sputum negative at eight weeks. Mean weight gain was improved with macronutrient supplementation after six weeks (MD 1.73 kg, 95% CI [0.81 to 2.65], p=0.00022; 1 trial/34 participants), 12 weeks (MD 2.60 kg, 95% CI [1.74 to 3.46], p<0.00001; 1 trial/100 participants), and 32 weeks (MD 2.60 kg, 95% CI [0.52 to 4.68], p=0.014; 1 trial/265 individuals); but not after eight weeks, 20 weeks, or 24 weeks. Lean body mass was 1.13 kg greater (95% CI [1.74 to 3.46], 100 participants) at six weeks follow-up in the treatment group in one trial assessing high-energy supplements. Maximum grip strength was improved with macronutrient treatment at six weeks (MD 3.44 kg, 95% CI [0.78 to 6.10], p=0.011; 1 trial/34 participants) and at 12 weeks (MD 1.50 kg, 95% CI [1.08 to 1.92], p<0.00001; 1 trial/100 participants), but not at eight, 20, or 24 weeks. Change in quality of life was assessed in two trials involving 60 participants; benefits of macronutrient supplementation were observed in certain scores, but not overall, and results could not be pooled due to the skewed nature of the data.

High-cholesterol versus low-cholesterol diet
One trial involving 21 participants compared high-cholesterol (800 mg/day) and low-cholesterol (250 mg/day) diets with a similar nutrient profile (2500 kcal/day). The primary outcomes all-cause mortality and cure were not reported on in this trial. At two weeks follow-up, fewer participants in the high-cholesterol group were sputum-culture positive (2 versus 10 individuals; RR 0.22, 95% CI [0.06 to 0.77]).

Multivitamin and trace element tablets versus placebo
All cause mortality
In pooled analysis of five trials, no effect of multivitamin supplementation in comparison to placebo was found on all-cause mortality (RR 0.96, 95% CI [0.79 to 1.18], 2902 individuals). Results were not altered meaningfully when subgrouped by HIV status (HIV negative: RR 0.86, 95% CI [0.46 to 1.60], 4 trials/1218 individuals; HIV positive: RR 0.92, 95% CI [0.69 to 1.23], 3 trials/1429 individuals).

Additional outcomes
No evidence of an effect of multivitamin supplementation was found for the outcomes treatment completion; smear or sputum-culture positive at one month or two months; clearance of chest X-ray at six months; weight at two, five or six months follow-up; weight-for-age Z-score change at two months; weight change at two months or six months; height-for-age Z-score change at two or six months; and body mass index (BMI) Z-score at two or six months. Weight after seven months was improved in one trial comparing multivitamins with and without zinc to placebo (with zinc: MD 2.37 kg, 95% CI [2.21 to 2.53], 192 participants; without zinc: MD 0.30 kg, 95% CI [0.17 to 0.43], 198 participants). Change in weight-for-age Z-score at six months was improved with micronutrients plus zinc in one trial of 198 participants (MD 0.52 kg, 95% CI [0.09 to 0.95]). Change in handgrip strength at two months was improved with multivitamin supplementation in one trial (MD 1.22 kg, 95% CI [0.49 to 1.95], 771 participants), but at five months follow-up, no statistically significant effect was evident (MD 0.91 kg, 95% CI [-0.56 to 2.36], 709 individuals).

Vitamin A versus placebo
All cause mortality
Vitamin A had no effect on all-cause mortality in comparison to placebo when given alone (RR 1.79, 95% CI [0.19 to 16.69], 1 trial/115 individuals), with zinc (RR 2.38, 95% CI [0.87 to 6.53], 4 trials/535 individuals), or as part of a multi-micronutrient supplement (RR 0.94, 95% CI [0.81 to 1.09], 4 trials/2658 individuals). Results were not subgrouped by HIV status.

Additional outcomes
Mean serum retinol concentrations in one trial of 85 children were not statistically significantly different between groups at six weeks or three months follow-up, and were not different in pooled analysis of three trials involving 242 adults at two months and six months follow-up. The outcomes treatment completion, symptomatic at six weeks, sputum-smear or -culture positive during follow-up, BMI, and percentage body fat were not different between treatment and control groups.

Zinc versus placebo
All cause mortality
Zinc had no effect on all-cause mortality at six to eight months when given alone (RR 1.25, 95% CI [0.70 to 2.22], 4 trials/714 individuals), with vitamin A (RR 2.08, 95% CI [0.65 to 6.64], 4 trials/477 individuals), or as part of a multi-micronutrient supplement (RR 0.91, 95% CI [0.77 to 1.07], 3 trials/1671 individuals). Results were not altered meaningfully when analyses were subgrouped by HIV status (HIV negative: RR 1.92, 95% CI [0.53 to 6.98], 3 trials/442 individuals; HIV positive: RR 1.06, 95% CI [0.59 to 1.91], 2 trials/211 individuals).

Additional outcomes
Serum zinc concentrations were improved with zinc supplementation at two months (MD 0.59 μmol/L, 95% CI [0.24 to 0.93], p=0.00088; 4 trials/452 individuals) and at six months (MD 0.60 μmol/L, 95% CI [0.31 to 0.88], p=0.000056; 4 trials/440 individuals). No effect of zinc supplementation was found for the outcomes treatment completion, clearance of chest X-ray at six months, BMI, percentage body fat, weight-for-age Z-score, BMI-for-age Z-score, or height-for-age Z-score. The risk of being sputum-smear or sputum-culture positive at two weeks was increased with zinc supplementation (RR 1.09, 95% CI [1.02 to 1.17], p=0.014; 3 trials/806 individuals), but no difference between treatment groups was found at four or eight weeks follow-up. Likewise, mean weight at seven months was decreased with zinc supplementation in one trial of 193 participants (MD -0.21 kg, 95% CI [-0.36 to -0.06]), but no difference between treatment groups was found at two months or at six months follow-up.

Zinc plus vitamin A versus placebo
All cause mortality
Overall, zinc plus vitamin A versus placebo did not statistically significantly affect the risk of mortality by six months (RR 2.33, 95% CI [0.90 to 6.07], 4 trials/578 individuals). However, in subgroup analysis by HIV status, the risk of death was increased with zinc plus vitamin A supplementation among those with HIV infection (RR 5.94, 95% CI [1.07 to 32.96], p=0.042; 2 trials/136 individuals).

Additional outcomes
Weight was not different between treatment groups at two months, but was improved at six months in one trial including 80 participants (MD 3.10 kg, 95% CI [0.74 to 5.46]). The Karnofsky score, measuring quality of life, was improved at six months in one trial of 80 participants (MD 2.50 points, 95% CI [0.91 to 4.09]). No evidence of an effect of zinc plus vitamin A treatment was found for the outcomes treatment completion at six months; sputum-smear or sputum-culture positive during follow-up; BMI; mid-upper arm circumference; biceps, triceps, subscapular, or suprailiac skinfold thickness; percentage body fat; or fat mass.

Vitamin D versus placebo or no supplement
All-cause mortality and cure (completed treatment and sputum-smear or sputum-culture negative)
Vitamin D versus placebo or no supplement did not statistically significantly affect the risk of mortality between two and 12 months follow-up (RR 0.96, 95% CI [0.81 to 1.12], 7 trials/2649 individuals). The results were did not differ meaningfully in subgroup analyses of trials providing vitamin D alone, with arginine, or as part of a multi-micronutrient supplement, or in subgroup analyses by HIV status. Cure at six months follow-up was assessed in one trial involving 151 individuals, in which no difference between treatment and control groups was found (RR 0.99, 95% CI [0.75 to 1.31]).

Additional outcomes
At both eight weeks and six to eight months follow-up, vitamin D supplementation improved serum vitamin D concentrations (MD 33.68 nmol/L, 95% CI [3.99 to 63.37], p<0.00001; 3 trials/460 individuals, and MD 7.90 nmol/L, 95% CI [0.52 to 15.28], p=0.036; 2 trials/376 individuals, respectively). No evidence of an effect of vitamin D treatment was found for the outcomes tuberculosis score (a measure of recovery based on symptoms and anthropometry), sputum-smear or sputum-culture positive during follow-up, BMI, body weight, or Karnofsky score.

Arginine versus placebo
All-cause mortality and cure (completed treatment and sputum-smear or sputum-culture negative)
Arginine versus placebo did not statistically significantly affect the risk of mortality during treatment (RR 0.67, 95% CI [0.21 to 2.09], 3 trials/394 individuals), and the results were did not differ meaningfully in subgroup analyses by HIV status. Cure at six to eight months follow-up was assessed in two trials involving 279 individuals, in which no difference between treatment and control groups was found (RR 1.05, 95% CI [0.90 to 1.22]).

Additional outcomes
The risk of cough was significantly reduced with arginine treatment in comparison to placebo in two trials involving 404 individuals (RR 0.76, 95% CI [0.62 to 0.93], p=0.0076). No statistically significant difference between treatment and control groups was found for the outcomes sputum-smear or sputum-culture positive during follow-up or weight gain >10%. Arginine reduced the risk of having a BMI <18.5 kg/m² in one trial after one and two months anti-tuberculosis treatment in comparison to placebo (both p≤0.04, data not shown).

Vitamin E plus selenium versus placebo
Plasma concentrations of vitamin E and selenium were reportedly higher in the treatment group in both trials investigating the effect of these micronutrients in individuals with tuberculosis (data not provided). In one trial involving 35 participants, vitamin E plus selenium had no statistically significant effect on the risk of being sputum-spear positive between 15 and 60 days follow-up. No other pre-specified outcomes were reported on.

C. striata capsules versus placebo
The primary outcomes all-cause death and cure were not reported in this trial. While the rate of sputum smear conversion was increased among those receiving the fish capsules, no statistically significant difference was found.

5. Additional author observations*

Most trials were small and lacked the power to detect statistically significant effects for the primary outcomes. Although studies were predominantly conducted in low- and middle-income countries, the settings may not have been food insecure, where nutritional interventions would have had the greatest potential for benefit. Using GRADE quality of evidence criteria, for the intervention increased energy supplementation, the outcomes death, cure, treatment completion, sputum negative and quality of life were rated as being of very low quality, while mean weight gain was rated as being of moderate quality. For the intervention multi-micronutrient supplementation, the outcomes death (HIV-negative individuals) and weight gain were rated as low quality, and the outcomes treatment completion and remaining sputum positive were rated as very low quality; for individuals with HIV, the evidence for the outcome all-cause death was rated as being of moderate quality.

The provision of additional meals or high-energy supplements likely increases weight gain in individuals with tuberculosis, and very low quality evidence suggests no benefit of macronutrient supplementation on mortality, cure or treatment completion. For tuberculosis patients with concurrent HIV infection who are not receiving antiretroviral therapy, multi-micronutrients have no effect on mortality. Multi-micronutrients are likely to have no effect on weight gain, or on mortality in HIV-negative individuals with tuberculosis, and very low quality evidence suggests no effect of multi-micronutrients on treatment completion or the proportion of participants remaining sputum positive. Supplementation with a micronutrient generally improves the corresponding blood concentrations of the micronutrient.

While no clear evidence of a treatment effect was found for most outcomes, no firm conclusions on the effect of nutritional supplementation in individuals with tuberculosis can be made due insufficient data from large, high-quality trials in food-insecure populations. As food support is currently provided in many tuberculosis programmes, further food provision trials are needed to provide evidence to support the continuation of such expenditure.