点击显示 收起
1 From the Departments of Community Medicine (AB, G, and HJ) and Child Health (AB, PM, PR, and CK), Christian Medical College, Vellore, India; the Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD (CLC, REB, WAB, and MS); and the Centre for Health and Population Research, International Centre for Diarrhoeal Disease, Research, Dhaka, Bangladesh (WAB)
See corresponding editorial on page 991.
ABSTRACT
Background:Severe pneumonia remains a leading cause of morbidity and mortality in undernourished young children in developing countries.
Objective:This study evaluated the effect of adjuvant zinc therapy on recovery from severe pneumonia by hospitalized children in southern India who were receiving standard antibiotic therapy.
Design:This randomized, double-blind, placebo-controlled clinical trial was conducted at the Christian Medical College Hospital, an 1800-bed teaching hospital in Tamilnadu, India. Enrollment and follow-up occurred between September 2003 and August 2004. Children aged 2–23 mo (n = 299) who were hospitalized with severe pneumonia were randomly assigned to receive 10-mg tablets of zinc sulfate or placebo twice a day during hospitalization, along with standard therapy for severe pneumonia. All clinical signs and symptoms of pneumonia were assessed and recorded at 8-h intervals.
Results:There were no clinical or statistically significant differences in the duration of tachypnea, hypoxia, chest indrawing, inability to feed, lethargy, severe illness, or hospitalization. Zinc supplementation was associated with a significantly longer duration of pneumonia in the hot season (P = 0.015).
Conclusions:Zinc supplementation had no overall effect on the duration of hospitalization or of clinical signs associated with severe infection in young children hospitalized for severe pneumonia in southern India. This finding differs from the results of 2 previously reported trials wherein zinc supplementation was associated with a shorter period of recovery from severe pneumonia. Given the conflicting results, further research in representative settings is required to help clarify the role of zinc in the treatment of severe pneumonia.
Key Words: Pneumonia • severe pneumonia • zinc • supplementation • treatment • adjuvant therapy • India
INTRODUCTION
Worldwide, pneumonia is the leading cause of pediatric morbidity and mortality. It is estimated that pneumonia is responsible for >2 million deaths each year in children <5 y old, which represents 19% of the annual deaths in this age group (1). Approximately 95% of the pneumonia-related deaths occur in developing countries, and the youngest age groups have the highest risk of death (2). India has the largest number of deaths of children <5 y old in the world—an estimated 2402000, or >3 times the number in China (3). Pneumonia case management, which relies on early diagnosis and prompt empiric antibiotic therapy, has been effective, reducing pneumonia-related deaths by 47% (4). However, the efficacy of this strategy may be diminished by poor nutritional status (5, 6). Undernutrition is known to be associated with greater severity of pneumonia, a higher frequency of complications, longer episodes of infection, and greater case fatality rates (7, 8). In light of these issues, the evaluation of interventions that may improve the effectiveness of case management in children with poor nutritional status is warranted.
In southern Asia, macronutrient malnutrition and micronutrient deficiencies, especially deficiencies of iron, zinc, and vitamin A, are common in young children. This problem is attributed to dietary insufficiency, limited nutrient bioavailability from local diets, and excretion of nutrients during recurrent episodes of infection. Of the micronutrients, zinc plays a critical role in immunomodulation and in maintaining the integrity of the immune system (9, 10). Results of preventive trials in children living in areas of endemic deficiency show that zinc supplementation significantly reduced the incidence of pneumonia and persistent diarrhea (11–13). Significant reductions in acute and persistent diarrhea morbidity were also observed in zinc-supplemented children in therapeutic trials (14).
Consensus as to whether zinc supplementation provides a similar therapeutic benefit to children with severe pneumonia has not yet been established. Published results from a recent clinical trial conducted in Bangladesh suggested that zinc supplementation given with empiric antimicrobial therapy can significantly shorten the duration of severe pneumonia, tachypnea, hypoxia, and chest indrawing and can also shorten the hospital stay for young children with pneumonia (15). A therapeutic trial conducted in Kolkata, India, evaluated the effects of zinc and vitamin A supplementation on recovery from acute lower respiratory infection (16). These investigators reported significant reductions in the duration of severe illness and fever among male children supplemented with zinc. However, for reasons that are unknown, zinc supplementation had no effect on clinical outcomes in females, nor did it have any effect on any other acute lower respiratory infection–related symptoms. More research is needed to clarify the role of zinc in the treatment of children who are hospitalized for severe pneumonia. We evaluated the efficacy of zinc supplementation in the treatment of severe pneumonia in hospitalized Indian children aged 2–23 mo who received standard antimicrobial therapy.
SUBJECTS AND METHODS
Study design and setting
This study was a randomized, double-blind, placebo-controlled clinical trial conducted between September 2003 and August 2004 at the Christian Medical College (CMC) Hospital in Vellore, India. CMC Hospital is a large teaching hospital and medical center that is accessed directly by the population of Vellore and adjoining districts and by referrals from local medical practitioners.
Subject eligibility and randomization
Children between 2 and 23 mo of age admitted to the pediatric wards of CMC Hospital were assessed by the study physicians and considered eligible for enrollment in the trial if they fulfilled the criteria for a diagnosis of severe pneumonia. Severe pneumonia was defined as a respiratory rate > 50/min, which was accompanied with crepitations on auscultation and the presence of 1 of the following danger signs: lethargy, inability to feed, chest indrawing, or central cyanosis. Children were excluded if they had history of chronic cardiac or renal disease, illness severe enough to require ventilation, malnutrition severe enough to require immediate nutritional rehabilitation therapy [ie, weight-for-age <60% of the reference for Indian children (17)], illness requiring hospitalization in the previous 21 d, or current zinc supplementation.
Written consent was obtained after the parents or guardians read a study information pamphlet and reviewed its contents with a study physician. If the parents or guardians were illiterate, the content of the pamphlet was read to them, and consent was documented by a thumbprint impression of one of the parents or guardians in the presence of an unrelated witness. The study was approved by the institutional review boards of CMC Hospital and The Johns Hopkins University Bloomberg School of Public Health.
At the time of enrollment, study physicians collected information on demographics, current illness, and history of respiratory diseases for each subject. All physical findings, anthropometric data, chest X-ray findings, complete blood count results, and baseline plasma zinc concentrations were recorded.
Randomization was conducted at the individual level and in blocks of 8 and 10. The use of large blocks was part of an effort to maintain blinding of the treatment assignments and to increase the likelihood that the treatment groups would be well balanced. The randomization codes were generated by a scientist at the headquarters of the World Health Organization (WHO), which was not associated with the study. Treatment codes were kept in an envelope in a locked cabinet at WHO headquarters in Geneva, Switzerland.
Intervention
Children with severe pneumonia were randomly assigned to receive supplementation with either elemental zinc or placebo tablets. Each tablet contained 10 mg zinc sulfate or 10 mg placebo (ie, crystalline cellulose, cornstarch, and vanilla flavoring). The zinc and placebo tablets were indistinguishable in appearance, consistency, and taste. The supplements were formulated and manufactured as dispersible tablets by Nutriset (Malunay, France) in collaboration with the WHO Department of Child and Adolescent Health and Development and packaged in blister packs of 14 tablets. The foil on the back of blister pack was labeled with the patient identification number. It was assumed that the study children would recover and be discharged within 14 d; therefore, each child was assigned 2 blister packs with the same treatment code. Supplementation ceased once children were discharged. The blister packs were stored on the pediatric wards in cool, locked cabinets that could be accessed only by study physicians who were responsible for dosing the participants at enrollment and at 0800 and 2000 every day during hospitalization.
Children received 20 mg (2 tablets) of either zinc sulfate or placebo by mouth at the time of enrollment. From day 2, they received 10 mg of their assigned treatment by mouth twice a day throughout the course of their hospitalization. For young infants, the tablets were dissolved in a teaspoon of distilled water before being administered.
All enrolled children were treated according to the standard protocol for treatment of infants and children with pneumonia. They were treated parenterally with a combination of benzyl penicillin and gentamicin. If a consultant pediatrician was of the opinion that the chest radiograph or clinical presentation was consistent with staphylococcal pneumonia, the patient was treated with a combination of cloxacillin and gentamicin. Patients who failed to improve after 48 h on this regimen of first-line antimicrobial therapy were switched to a regimen of parenteral cefotaxime.
During hospitalization, the child’s condition was assessed at the beginning of each 8-h shift by a study physician or a study nurse, and the assessment was recorded in the patient’s chart. Respiratory rate was measured for a full minute, after the child’s upper-torso clothing was removed. If the respiratory rate was >50 breaths/min, it was measured again, and if the count was different by >5 breaths/min, a third reading was taken, and the 2 closer readings were averaged. The count was done at a time when the child was not crying. Pulse oximetry was measured with the use of a probe (Nellcor Inc, Hayward, CA) placed on a finger or toe. For intercurrent readings in patients who required oxygen supplementation, the oxygen supply was discontinued for 5 min before pulse oximetry. Axillary temperature was measured by using a standard mercury thermometer. The presence of cough, crepitations, wheezing, chest indrawing, cyanosis, inability to feed, and lethargy was also noted in the patient’s chart.
Enteral feeds were started once the child’s respiratory rate fell to <60 breaths/min, oxygen saturation was 93%, and the baby was able to tolerate small sips of feeds. Feeds were started as early as possible to optimally balance caloric intake with the need for caution in initiating feeds in a tachypneic child. Oral antibiotics were started when the child was feeding well and when oxygen saturation and respiratory rate were stabilized. Once the patient was on oral antibiotics, he or she remained under observation on the ward for a further 24 h before discharge. Study subjects were discharged when they were being fed entirely with oral feeds, the respiratory rate was <50 breaths/min, oxygen saturation was 93%, and the attending pediatrician decided that the patient’s clinical condition had resolved and did not require further hospital care. A blood sample for zinc measurement was taken at the time of discharge.
Measurement of plasma zinc concentration
Blood samples for the measurement of plasma zinc concentrations were collected at enrollment and at discharge by using a disposable syringe and needle; the samples were transferred to a zinc-free polypropylene tube containing 100 IU lithium heparin of negligible zinc content. The blood was centrifuged at 3500 rpm for 10 min (Heraeus Multifuge 3-S; Kendro Laboratory Products, Langensebold, Germany), and the plasma was separated immediately by using a disposable Pasteur pipette and stored in a zinc-free polypropylene tube at –20°C until analysis. All errors due to glassware, sample collection, processing, and analysis were eliminated according to previously established methods (18, 19). Measurement of the zinc concentrations in plasma was carried out by using an Atomic Absorption Spectrophotometer (model Analyst 100; Perkin Elmer, Boston, MA) with air-acetylene as the oxidant. Certified zinc standard (Cat no. Z-2750; Sigma Chemical Co, St Louis, MO) was used in the assay. Plasma was diluted 1:5 (0.2 mL plasma + 0.8 mL distilled deionized water) in disposable zinc-free polypropylene tubes. The assay was done in duplicate. The diluted samples were directly aspirated into the Atomic Absorption Spectrophotometer, and the reading obtained was compared with a standard containing 25 µg zinc/100 mL. Glyceryol [5% (by vol)] was used as the blank solution because the working standard was prepared in 5% glycerol. The result was calculated by using a formula: plasma zinc (in µg/100 mL) = reading of test in µg/100 mL x 5. Internal quality control of bovine origin stabilized with 15% (by vol) Ethanediol was used during every assay to monitor the precision, and a level 2 control (BioRad, Hercules, CA) was used every 2 wk to assess the accuracy. The results obtained for the above controls were well within the WHO criteria of <8% for internal quality control and within a mean (±2 SD) for quality control of accuracy. The normal value obtained for plasma zinc by using this method was 70–125 µg/100 mL (10.7–19.1 µmol/L).
Definition and measurement of outcomes
The primary clinical outcomes of the study were time to resolution (no longer meeting the definitions as outlined below) of severe pneumonia and the duration of hospitalization. Because there is no accepted standard definition of cessation of severe pneumonia, we used 3 definitions of severe pneumonia to evaluate recovery: 1) respiratory rate > 50 breaths/min and oxygen saturation < 93%; 2) inability to drink, respiratory rate > 50 breaths/min, and oxygen saturation < 93%; and 3) chest indrawing, respiratory rate > 50 breaths/min, and oxygen saturation < 93%.
The duration of hospitalization was defined as the time (in h) between study enrollment and discharge. In addition, we also examined the time to resolution of the clinical symptoms of chest indrawing, tachypnea (ie, respiratory rate > 50/min), inability to feed orally, hypoxemia (ie, arterial O2 saturation < 93%), fever (ie, axillary temperature > 37.5°C), and cough. Time to resolution from wheezing was evaluated as a secondary outcome.
The research team was composed of 3 study physicians and 2 research nurses. The medical officers and nurses received 2 wk of training in the assessment of clinical findings and study outcomes to ensure the reproducibility of results.
Sample size
We aimed to enroll 136 children with severe pneumonia in each treatment group, for a total of 272 children. The sample size was calculated to detect a minimum reduction of 30% in the duration of hospitalization between treatment groups. The figure assumes 80% power, a 2-sided type 1 error of 5%, loss to follow-up of 10%, and a mean duration of hospitalization for pneumonia of 5.7 d in children in this age group. The estimate was also valid for detecting similar reductions in duration of clinical signs and symptoms associated with severity in the 2 treatment groups. To account for exclusions due to possible readmissions and on the recommendation of the Data and Safety Monitoring Board, 6 mo after the start of enrollment, the original sample size was increased from 272 to 300 (150 each in the zinc and placebo groups).
Statistical analysis
The effects of zinc supplementation on outcomes of interest were analyzed on an intention-to-treat basis. The data were analyzed by using STATA software (version 8.0; Stata Corp, College Station, TX). The t test for continuous variables and the 2-tailed chi-square analysis or Fisher’s exact test for contingency data were used, as appropriate, to assess treatment group differences at baseline. Kaplan-Meier survival functions were used to measure the median duration of outcomes and to assess the effect of treatment on the study outcomes. Cox proportional hazard regression models were constructed to adjust the treatment effects for potential confounding factors and to evaluate effect modification. The risk ratios (RRs) compare the recovery rates from the indicators in the zinc-supplemented group with those in the placebo group. RRs > 1 are associated with the probability that the duration of the outcome will be shorter in the zinc group. Analyses were based on clinical outcomes, and time to recovery was defined as the end of the last consecutive 16-h period in which the case did not meet the definition of severe pneumonia. All analyses were by intention-to- treat.
RESULTS
Between 15 September 2003 and 31 August 2004, we approached the parents of 307 children who were eligible for participation in the trial (Figure 1); the parents of 7 children refused. The remaining 300 children were enrolled, and half were randomly assigned to each of the treatment groups. One child in the zinc group withdrew from the study before discharge (331 h after enrollment). Two children in the placebo group left against medical advice: one child left after 2 h, and that child was excluded from the analysis; another left after 69 h. There was one death in the placebo group, which occurred 8 h after enrollment.
FIGURE 1.. Trial profile.
Baseline characteristics of the children were well balanced between treatment groups (Table 1). Approximately 70% of the children were male, and 70% were 2–11 mo old. In addition, 202 (67%) of the 299 children followed were admitted during the rainy season (September–February). Enrollment by season significantly differed by age group. Of those children enrolled during the rainy season, only 44 (21.8%) were 12 mo old. In contrast, the distribution during the hot season (March–August) was more equitable: 46 (47.6%) of the 97 children admitted during that season were 12 mo old. Approximately 78% of our patients were anemic; ie, their hemoglobin concentrations were < 11 g/dL. Mean plasma zinc concentrations on admission did not differ significantly between groups (Table 2). At discharge, mean (±SD) plasma zinc concentrations were significantly higher in the zinc group than in the placebo group (13.0 ± 2.5 and 12.0 ± 4.1 µmol/L, respectively; P = 0.013).
View this table:
TABLE 1. Baseline characteristics (as categorical variables) of the zinc and placebo treatment groups1
View this table:
TABLE 2. Baseline characteristics (as continuous variables) of the zinc and placebo treatment groups1
There was no significant difference between the zinc and placebo groups in the time to recovery from severe pneumonia, as defined by inability to feed, arterial O2 saturation < 93%, and respiratory rate > 50 breaths/min (Table 3). The lack of association between the recovery experience and treatment group was also apparent when survival curves were used (Figure 2). RRs for recovery from severe pneumonia did not differ significantly, irrespective of the definition of severe pneumonia that was used. Moreover, the median length of hospitalizations did not differ significantly between the zinc and placebo groups (Table 3). No significant differences were observed when recovery from other clinical features was assessed.
View this table:
TABLE 3. Effect of treatment on recovery time by clinical indicator1
FIGURE 2.. Kaplan-Meier survival curves of recovery from severe pneumonia in the placebo group (solid line) and the zinc-supplemented group (broken line).
In our subgroup analysis, we discovered that there was a significant negative treatment effect during the hot season but not during the rainy season (Table 4). During the hot season, the median duration of severe illness was significantly longer in the zinc-supplemented group than in the placebo group (97 and 72.3 h, respectively; RR = 0.60, P = 0.015), as was the length of hospitalization (95.2 and 73.0 h, respectively; RR = 0.57, P = 0.008). Significant treatment effects favoring the placebo group were observed with all clinical indicators except cough. There was, however, no difference in treatment effect by sex, age, age group, nutritional status, number of days before illness, wheeze, and antibiotic use before admission. None of the 3-way interaction models was statistically significant.
View this table:
TABLE 4. Effect of treatment on median number of hours to recovery by clinical indicator and by season of enrollment1
There was one death in the placebo group, which occurred 8 h after enrollment. There were no other adverse events. Thirty percent of the children in the zinc group and 23% of those in the placebo group had to be switched from a first-line antibiotic to a second-line antibiotic. The difference in the proportion of treatment failures between treatment groups was not statistically significant.
DISCUSSION
The current study did not show a clinically or statistically significant reduction in duration or severity of pneumonia or a reduction in overall hospital stay for children <2 y old given daily zinc supplementation along with standard antimicrobial therapy. We could identify no subgroups that benefited from zinc supplementation. We noted that the season modified the association between treatment and outcome. Specifically, zinc-treated children had a slower recovery than did those receiving placebo in the hot season but not during the rainy season. We are unable to explain this association. It may be possible that the effect is related to differences in the etiology of pneumonia by season or differences in the pattern in micronutrient deficiencies that are due to seasonal variations in diet. In the Vellore area, viral, bacterial, and mixed respiratory infections predominate in the rainy season, whereas there is a low incidence of viral respiratory infections in the hot season (20, 21). Consistent with this observation is the high prevalence of wheeze in children in the rainy season and a marked decrease in wheeze in the hot season. Thus, our data may imply that zinc can prolong the duration of severe pneumonia caused by bacterial infection. Given that the results were obtained from subgroup analyses, implications are speculative, but they may have some importance for the generation of new hypotheses. We did not find evidence of any other treatment effects.
In contrast to the main findings of the current study, results from 2 clinical trials showed significant benefit associated with zinc supplementation in children with severe pneumonia. In a clinical trial conducted in Bangladesh (15), zinc supplementation resulted in a reduction by 12 h in the duration of severe pneumonia and a reduction by 16 h in the hospital stay, in addition to reductions in tachypnea, chest indrawing, and hypoxia. Likewise, investigators who conducted a similar trial in Kolkata, India reported a reduction by half in the duration of severe illness and a reduction by two-thirds in the duration of fever (16). The endpoints in the Kolkata study were different from those in the Bangladesh study, although both studies evaluated cessation of signs of severe pneumonia. In Kolkata, there were no effects on tachypnea and feeding status, and the effects of treatment on chest indrawing, hypoxia, and wheeze were not reported. This study reported that zinc supplementation benefited only male children; it did not have any effect on females. The finding remains unexplained.
There were several differences in the population characteristics between our study and the Kolkata and Bangladesh trials. Mean plasma zinc concentrations in Vellore study subjects at enrollment (11.0 µmol/L) were significantly higher than those in subjects in Bangladesh (10.1 µmol/L) and Kolkata (9.6 µmol/L). Although the differences are statistically significant (P < 0.001), it is unclear whether this difference is of clinical importance. Unlike the other studies, the current study evaluated the effect of plasma zinc status (<9.18 µmol compared with 9.18 µmol Zn/L) on treatment and found no evidence of effect modification. In addition, the time of blood collection is known to cause variation in plasma zinc concentrations (22). In all 3 randomized trials, blood samples for plasma zinc assessment were collected at enrollment and at discharge, which occurred at any hour of the day or night throughout the study. Therefore, the time of blood collection is unlikely to be a confounder. The average number of days during which the children reported being ill before admission in the current study was double that in the Bangladesh study (5.0 and 2.5 days, respectively; P < 0.001). Although this difference may suggest that the children in our study were more likely to have a milder illness or to be in recovery at admission than were the Bangladeshi children, our data do not support this idea. In our subgroup analysis, days of illness before admission (<3 compared with 3 d) did not alter the lack of treatment effect.
Moreover, the proportion of children with wheeze at admission was significantly higher in our study population than in the Bangladeshi study (62.5% and 37%, respectively; P < 0.001). This may suggest a difference in the etiology of pneumonia between populations, because wheeze is frequently associated with viral infection. Brooks et al (15) noted that the association between zinc and recovery from severe pneumonia and the other clinical indicators of illness was strongest in children who presented without wheeze at admission. Because wheeze is commonly associated with viral infection, the authors hypothesize that zinc supplementation may have a greater effect on bacterial infections. However, this possibility remains controversial. Evidence from laboratory investigations suggests that the zinc deficiency impairs cell-mediated immunity in addition to reducing antibody production (9, 23). If this is correct, then supplementation is likely to aid in the restoration of immunity to both viral and bacterial infections. Moreover, as was mentioned previously, zinc supplementation was significantly associated with a prolonged recovery in the hot season, a time when the incidence of viral infections is thought to be low (20, 21). Finally, the results of our stratified analysis show no modification of treatment effect by wheeze status.
All 3 trials mentioned above used comparable study designs, enrollment criteria, and supplement dosages and standardized antibiotic protocols. However, there were also several differences in method between the current trial and the other 2 trials, which may have affected the association between zinc and recovery from pneumonia. The Bangladesh trial was conducted in a research ward of a small hospital in a rural district, where all clinical decisions and assessments of recovery status were made by study physicians using objective criteria. In contrast, our study was conducted on 3 pediatric wards in a large general hospital. Although study physicians at the CMC Hospital were responsible for recording observations according to well-defined criteria, all clinical decisions and determinations of recovery status were based on the clinical judgment of the attending pediatricians, as they were in the Kolkata trial. The implication of this is that some potential exists for nondifferential misclassification of recovery status, which would shift the association between zinc and recovery from pneumonia toward the null hypothesis. The consistency between the objective observations recorded by the study physicians and the subjective assessments of recovery status by the attending pediatricians suggests that this is unlikely to have masked the effect of treatment.
In the current study, emphasis was given to infant feeding, reduction of hypoxia, and a reduction in respiratory rate to <60 breaths/min for initiating oral feeds and to <50 breaths/min for decision to discharge. This protocol was in keeping with that for the treatment of infants with severe pneumonia and is dictated by both the high prevalence of malnutrition, which necessitates earlier enteral feeding, and a high demand for beds, which necessitates discharge even when the respiratory rate is >40 breaths/min.
The zinc supplements were well tolerated and safe, and no vomiting was noted. The plasma zinc concentrations at the time of admission were lower than those at discharge in both the zinc and placebo groups. The change in plasma zinc status is consistent with zinc’s negative acute phase response to infection.
The current study found that zinc supplementation had no effect on the duration of hospitalization or of clinical signs associated with severe infection in young children hospitalized for severe pneumonia in southern India. Differences in the study populations and methods between the current study and the 2 studies that found zinc to be beneficial may help to explain differences in the outcomes. The possibilities also exist that zinc supplementation for the treatment of severe pneumonia may benefit only an as-yet-undefined subpopulation of pediatric pneumonia patients and that the etiology of the pneumonia is a deciding factor. Alternatively, it may be that zinc is not efficacious in the treatment of acute pneumonia. In light of the conflicting study results, additional large trials in representative settings are required to help clarify the role of zinc in the treatment of severe pneumonia.
ACKNOWLEDGMENTS
We thank Jayaprakash Muliyil (Department of Community Medicine, Christian Medical College, Vellore, India) for his sage advice on data analysis and S Swaminathan (Department of Biochemistry, Christian Medical College, Vellore, India) for his excellent technical assistance.
WAB, CLC, MS, RB, AB, PM, PR, and CK conceived and designed the study; AB, HJ, G, CLC, PM, PR, and CK implemented the study; AB, HJ, and G were responsible for data collection, patient enrollment, and monitoring; AB, HJ, G, and CLC supervised the data management; CLC, AB, HJ, and G analyzed the data; AB and CLC wrote the manuscript; and PR, REB, MS, and WAB contributed to the writing and editing of the manuscript. None of the authors had any personal or financial conflicts of interest.
REFERENCES
Related articles in AJCN: