Articulo publicado en el PEDIATRICS por Marek Lukacik, MDa, Ronald L. Thomas, PhDb, Jacob V. Aranda, MD, PhDb. sobre el uso de zinc en la diarrea aguda y persistente, donde se verifica que acorta la duracion de la enfermedad.
A Meta-analysis of the Effects of Oral Zinc in theTreatment of Acute and Persistent DiarrheaMarek Lukacik, MDa, Ronald L. Thomas, PhDb, Jacob V. Aranda, MD, PhDb
aDepartment of Pediatrics, Children’s Medical Center, Medical College of Georgia, Augusta, Georgia; bDepartment of Pediatrics, Wayne State University School ofMedicine, and Children’s Hospital of Michigan, Detroit, Michigan, and National Institute of Child Health and Human Development, Pediatric PharmacologyResearch Unit Network, Wayne State University, Detroit, Michigan
The authors have indicated they have no financial relationships relevant to this article to disclose.
ABSTRACT
OBJECTIVE.Children in developing countries are at a high risk for zinc deficiency.Supplemental zinc has previously been shown to provide therapeutic benefits indiarrhea. The objective of this study was to examine the efficacy and safety ofsupplemental oral zinc therapy during recovery from acute or persistent diarrhea.
METHODS.We conducted a meta-analysis of randomized, controlled trials to comparethe efficacy and safety of supplementary oral zinc with placebo in children with acuteand persistent diarrhea. Results were reported using a pooled relative risk or aweighted mean difference. A total of 22 studies were identified for inclusion: 16examined acute diarrhea (n � 15 231), and 6 examined persistent diarrhea (n �2968).
RESULTS.Mean duration of acute diarrhea and persistent diarrhea was significantlylower for zinc compared with placebo. Presence of diarrhea between zinc andplacebo at day 1 was not significantly different in acute diarrhea or persistentdiarrhea trials. At day 3, presence was significantly lower for zinc in persistentdiarrhea trials (n � 221) but not in acute diarrhea trials. Vomiting after therapy wassignificantly higher for zinc in 11 acute diarrhea trials (n � 4438) and 4 persistentdiarrhea trials (n � 2969). Those who received zinc gluconate in comparison withzinc sulfate/acetate vomited more frequently. Overall, children who received zincreported an 18.8% and 12.5% reduction in average stool frequency, 15.0% and15.5% shortening of diarrhea duration, and a 17.9% and 18.0% probability ofreducing diarrhea over placebo in acute and persistent trials, respectively.
CONCLUSIONS. Zinc supplementation reduces the duration and severity of acute and persistent diarrhea; however, themechanisms by which zinc exerts its antidiarrheal effect have not been fully elucidated.
DIARRHEAL DISEASES POSE a significant public health problem on a global scale and especially in developingcountries. It is estimated that there are �1.5 billion episodes of diarrhea per year and that diarrheal disease
accounted for 21% of all deaths in children who were younger than 5 years. This is equivalent to 2.5 million deathsin the same age group.1,2
This compares more favorably with the results of a previous study from 1982 in which on the basis of a reviewof active surveillance data from studies conducted in the 1950s, 1960s, and 1970s, it was estimated that 4.6 millionchildren died annually from diarrhea.3 Newer data from the World Health Organization (WHO) show that diarrhealdisease accounts for 18% of the 10.6 million deaths in children who were younger than 5 years.4
One of the major advances in the reduction of mortality from diarrhea was the introduction of WHO oralrehydration solution (ORS)5; however, WHO ORS does not significantly decrease stool output and duration ofdiarrhea, and therefore other approaches to add to or to enhance the available ORS have been sought. Several newerapproaches have included the addition of zinc to the treatment regimen. Zinc is an essential micronutrient andprotects cell membranes from oxidative damage. Zinc is not stored in the body, so the level of zinc is determined bythe balance of dietary intake, absorption, and losses. A zinc deficiency state may exist in children with acute diarrhea
www.pediatrics.org/cgi/doi/10.1542/peds.2007-0921
doi:10.1542/peds.2007-0921
KeyWordsdiarrhea, zinc
AbbreviationsWHO—World Health OrganizationORS—oral rehydration solutionRR—relative riskWMD—weighted mean differenceCI—confidence intervalcAMP—3�,5�-cyclic monophosphateK—potassiumCa—calcium
Accepted for publication Jul 24, 2007
Address correspondence to Marek Lukacik,MD, Children’s Medical Center Department ofPediatrics, Medical College of Georgia, 112015th St, Augusta, GA 30912. E-mail: [email protected]
as a result of intestinal loss. A comprehensive review onthis subject was recently published.6 An alternative viewis that zinc may be working as a pharmacologic agent atthe level of gene expression.7 The efficacy of zinc in thetreatment of diarrhea is supported by several random-ized, controlled trials that showed reduction of diarrheaduration, stool output, and stool frequency. Meta-anal-yses on the therapeutic effects8 of zinc in acute andpersistent diarrhea as well as prevention9 of diarrheawith zinc supplementation have been previously pub-lished. The published data so far have shown the efficacyof zinc in the treatment of acute and chronic diarrhea.Our meta-analysis was performed to include new studiespublished since the last meta-analysis and to examinethe efficacy and safety of zinc therapy during recoveryfrom acute or persistent diarrhea.
METHODS
Inclusion CriteriaStudies that were selected for inclusion tested the sameprimary hypotheses (average duration of diarrhea andpresence of diarrhea at days 1, 3, and 5) using similarpatient characteristics (primarily children aged between1 and 60 months), with either acute or persistent diar-rhea, including dysentery. Acute diarrhea was defined aslasting up to 14 days, with persistent diarrhea lasting�14 days. Random allocation to treatment groups andconcealment of allocation had to be met to satisfy inclu-sion because inadequate allocation concealment, despitethe use of randomization, allows a risk for selection bias.Intervention with oral zinc salt supplementation, allow-ing for any zinc salt type or formulation (sulfate, glu-conate, or acetate) if applied at �5 mg/day for anylength of duration, was examined against a control usinga placebo. All comparisons between treatment groupshad to be free of confounding by additional agents orco-interventions. Study groups who, after randomiza-tion, received zinc supplementation and ORS or zincsupplemented with vitamin A were excluded.
Identification of TrialsThe search strategy used computerized bibliographicsearches of Medline (1966–2006); the Cochrane CentralRegister of Controlled Trials (2006); Embase (1974–2006);Lilacs (1982–2006); CINAHL (1982–2006); Current Con-
trolled Trials (2006); and abstracts published in PediatricResearch (1991–2006) and the First (Boston, 2000) andSecond (Paris, France, 2004) World Congress of PediatricGastroenterology, Hepatology and Nutrition. Both pub-lished and unpublished trials were included in an effortto control for publication bias. Citations of appropriatestudies were verified by reviewing the bibliographies andreference lists of identified trials. Identified titles of ab-stracts with potential relevance were downloaded, andfull manuscripts were then obtained for all abstracts thatwere deemed relevant on the basis of the inclusioncriteria. Twenty-two trials met inclusion criteria: 16 pub-lished studies relative to the definition of acute diarrheaand 6 relative to persistent diarrhea.
Primary and Secondary OutcomesData on 8 clinically relevant outcome measures werecollected. We held average duration of diarrhea andpresence of diarrhea episodes at days 1, 3, and 5 as ourprimary outcomes. Data on vomiting frequency, vom-iting frequency by therapy type, stool frequency re-duction, and probability of diarrhea continuationwere extracted as secondary outcomes. All 3 authorsindependently extracted data from the same articles us-ing a data extraction sheet and subsequently comparedresults for agreement. The data thus obtained werechecked for consistency among authors, integrity of ran-domization, and concealment of allocation. Questionsregarding the interpretability of certain data values wereresolved by all 3 authors. The final database entries wereverified by the statistician (Dr Thomas). Few studiessatisfied criteria for inclusion on every datum variable.When necessary, authors of selected studies were con-tacted to verify extracted data values derived fromgraphs and/or to provide additional information in ascaling form that could be combined with other studies.Where those instances occurred, they are noted in Tables1 and 2.
DefinitionsDefinitions of diarrhea varied somewhat in all includedstudies. In acute trials, generally, the definitions statedfor diarrhea were the passage of �3 loose, watery stoolsor 1 loose, watery stool with blood within 24 hours forbetween 3 and 7 days in duration. In persistent diarrheatrials, the definitions were similar, with the exceptionthat they persisted up to 14 days in duration.
Definitions for duration of diarrhea varied as well butwas defined, generally, from the time of enrollment intothe study until the first formed stool. Duration wasmeasured in either days or hours. For the purpose of thismeta-analysis, hours were converted to days. After en-rollment/randomization, either the zinc treatment or theplacebo was assigned within 24 hours.
TABLE 1 Average Duration of Diarrhea (Days)
Reference Zinc Placebo
Patel et al20 (2005) 4.34� 2.28 4.41� 1.98Valery et al19 (2005) 3.26� 3.31 3.30� 5.21Fischer Walker et al16 (2006) 4.93� 3.90 4.49� 3.17
Data are means � SD. Data previously obtained during the course of the study.
TABLE 2 Number of ChildrenWith Diarrhea at Days 1, 3, and 5
Reference Zinc Day 1 Placebo Day 1 Zinc Day 3 Placebo Day 3 Zinc Day 5 Placebo Day 5
Statistical AnalysesComprehensive Meta-Analysis,10 a stand-alone program,was used to synthesize data that were obtained from the22 trials identified for inclusion: 16 acute and 6 persis-tent diarrhea trials. Briefly, the analysis software pro-duces a Forrest plot as a schematic description of themeta-analysis results. The program is augmented usingaccepted computational algorithms. Where appropriate,results were reported using a pooled relative risk (RR).For continuous outcomes, the weighted mean difference(WMD) was calculated. The 95% confidence intervals(CIs) were reported around the weighted effect size.
HeterogeneityGiven that studies that are selected for inclusion in ameta-analysis will differ, the types of variability (clinical,methodologic, and/or statistical) that may occur amongstudies must be investigated. These various types of vari-ability are termed heterogeneity. Meta-analysis shouldbe considered only when a group of trials is sufficientlyhomogeneous (as indicated in the inclusion criteria) interms of participants, interventions, and outcomes toprovide a meaningful summary. Strict adherence to theinclusion criteria listed, such as blinding and conceal-ment of allocation, help to control for clinical/method-ologic heterogeneity. Still, statistical heterogeneity canalso occur when variability in the treatment effects beingevaluated in the different trials exists. This results whenthe observed treatment effects are more different fromeach other than would be expected as a result of randomerror (chance) alone. Following convention, statisticalheterogeneity in the results of this meta-analysis arereferred to simply as heterogeneity.
Different approaches for identification and measure-ment of heterogeneity were therefore undertaken toexamine the extent to which the results of the studiesincluded were consistent. CIs for the results of individualstudies (depicted graphically using horizontal lines) wereexamined for poor overlap, a general indication of pres-ence of statistical heterogeneity. Variability (heterogene-ity) among the obtained effects sizes was formally op-erationalized using a �2 test of significance. The formulafor heterogeneity assesses the dispersion of individualoutcomes, vis-a-vis the combined effect, and denotesthis value using a Q statistic.11 A low P value (or a large�2 statistic relative to its degree of freedom) providesevidence of heterogeneity of treatment effects (variationin effect estimates beyond chance).
Because some degree of clinical and methodologicdiversity always occurs in a meta-analysis, some statis-tical heterogeneity is inevitable; therefore, the test forheterogeneity is irrelevant to the choice of analysis: het-erogeneity will always exist regardless of whether it canbe detected using a statistical test. Still, methods havebeen developed for quantifying inconsistency acrossstudies that move the focus away from testing whetherheterogeneity is present to assessing its impact on themeta-analysis. A useful statistic for quantifying inconsis-tency is I2, the percentage of the variability in effect sizeestimates that is attributable to heterogeneity ratherthan sampling error (chance).12 A value �50% may be
considered substantial heterogeneity, and that percent-age cutoff was adopted and examined also in our anal-yses.
GravityAnother more recent approach13 proposed jackknife re-sampling to measure a concept termed “gravity.” In anymeta-analysis, arguments have focused on the inclusionor exclusion of some studies, with debate on which onesshould be included or excluded because studies are com-monly weighted according to their sample size and/orinternal variability. Gee13 proposed that jackknife re-sampling could be used to examine study influence anddetect outlier studies. The technique recomputes themeta-analysis once for each of k studies, where eachstudy is individually excluded. K results are then ob-tained. The difference between the average of these kresults and each study’s individual result (when omit-ted) is taken as an index of “raw gravity.” This differ-ence, divided by the SD of the k differences, is taken asa z score, or “standardized gravity,” which can be used toestablish which studies might be unusually influential.SPSS 15.014 was used to calculate standardized gravityvalues.
Fixed- or Random-Effects ModelChoice of whether to interpret a fixed-effects or ran-dom-effects model was considered thoroughly. Fixed-effect meta-analyses ignore heterogeneity. The fixed-effect estimate and its CI address the question, “What isthe best estimate of the treatment effect?” The random-effects estimate and its CI address the question, “What isthe average treatment effect?” The answers to thesequestions are analogous when no heterogeneity ispresent or when the distribution of the treatment effectsis roughly symmetrical. If they are not, then the ran-dom-effects estimate may not reflect the actual effect inany population being studied. In a fixed-effects meta-analysis, a pooled-effect estimate is termed, generally, asthe best estimate of the treatment effect. It is for thesereasons that we chose a fixed-effects model for ourmeta-analysis, along with the various stated approachesto examine heterogeneity if found.
RESULTSThe author, year, country, amount of zinc supplemen-tation and type, sample size, and age for each of the 22studies selected for inclusion in the meta-analysis arelisted in Tables 3 and 4. Although all 22 studies wererandomly assigned clinical trials, it seemed that 515–19
were not double-blinded. Sixteen of these publishedstudies met the definition for acute diarrhea and 6 forpersistent diarrhea.
Overall, 56.3% (9 of 16) of acute diarrhea trials wereconducted in inpatient hospital settings, and 43.7% (7 of16) were conducted in outpatient homes and commu-nities. Of the 6 persistent diarrhea trials, 66.7% (4 of 6)were inpatient and 33.3% (2 of 6) were outpatient.
328 LUKACIK et al by on August 1, 2009 www.pediatrics.orgDownloaded from
MortalityMortality was originally a primary outcome in this meta-analysis; however, of both acute and persistent trials,only 315,20,21 reported mortality outcome, making it diffi-cult to compare across all included trials. Two of thesewere acute diarrhea trials,15,20 and 1 was a persistentdiarrhea trial.21 In the largest acute diarrhea outpatienttrial15 (n � 8070), 33 children (0.008%; 33 of 3974) diedin the zinc-treated group and 37 (0.009%; 37 of 4096)died in the placebo group. Thirty deaths were attributedto drowning, and the remaining were not injury related(ie, not attributable to zinc intervention). When re-stricted to noninjury deaths, there were 13 in the zinc-treated group and 27 in the placebo group. The investi-gators attributed the lower noninjury death rate in theintervention group almost entirely to fewer deaths fromdiarrhea and acute lower respiratory infection. Diarrheaand acute lower respiratory infection together accountedfor 10 deaths in the zinc intervention group and 20deaths in the placebo group. In the other acute diarrheatrial,20 2 children in the placebo group died of septicemia.In the persistent diarrhea trial,21 the causes of death weresepticemia with diarrhea in 3 children, septicemia in 1child, bronchopneumonia in 1 child, and continued di-arrhea in 1 child. Because acute and persistent diarrheaare, most likely, distinct disease entities, the outcomes
obtained are presented initially for acute diarrhea (last-ing up to 14 days) and followed by persistent diarrhea(lasting �14 days).
Results for Acute Diarrhea Trials
Duration of Acute DiarrheaIn 16 trials that examined the primary measure of aver-age duration of acute diarrhea15–17,19,20,22–32 (n � 15 231),those who received zinc experienced a significantlylower average duration of diarrhea than those who re-ceived a placebo (WMD: 0.24; SE: 0.02; 95% CI: 0.21–0.27; P � .001; Table 5, Fig 1) but also with the presenceof statistically significant heterogeneity (Q � 95.58, de-grees of freedom [df]Q � 15, P � .001, I2 � 84.3%).Figure 1 depicts a Forrest plot for these results, in whichevery study is displayed as a point estimate with CIs.
Examination of significant heterogeneity in the acutediarrhea trials revealed 5 trials17,19,20,25,30 with insignifi-cant differences between zinc and placebo groups inaverage duration of diarrhea. P values ranged from .478to nonsignificant in sample sizes that ranged from 50 to215. Although those who received zinc had a shorteraverage duration of diarrhea, the difference in 4 tri-als17,19,20,30 was very small, with an average difference of0.18 � 0.18 days ranging from 0.04 to 0.40 days. One
TABLE 3 Characteristics of Acute Diarrhea Trials
Reference Country Zinc Supplement Zinc Dosage Zinc/Control Group, N Age, mo
Sachdev et al17 (1988) India Sulfate 20mg 25/25 6–18Sazawal et al31 (1995) India Gluconate 20mg 456/481 6–35Roy et al30 (1997) Bangladesh Acetate 20mg 57/54 3–24Faruque et al27 (1999) Bangladesh Acetate 14/40mg 343/341 6–23Hidayat et al28 (1998) Indonesia Acetate 4/5mg/kg 739/659 3–25Dutta et al26 (2000) India Sulfate 40mg 44/36 3–24Strand et al32 (2002) Nepal Gluconate 15/30mg 445/449 6–35Bahl et al23 (2002)a India Gluconate 15/30mg 404/401 6–35Al-Sonboli et al22 (2003) Brazil Sulfate 22.5/45mg 37/37 3–60Polat et al29 (2003)b Turkey Sulfate 20mg 92/90 2–29Bhatnagar et al24 (2004) India Sulfate 15/30mg 143/144 3–36Valery et al19 (2005)c Australia Sulfate 20/40mg 107/108 0–11, 12–23, �24Patel et al20 (2005) India Sulfate/copper sulfate 40mg/5mg 102/98 6–59Brooks et al25 (2005)d Bangladesh Acetate 20mg 86/89 1–6Baqui et al15 (2002) Bangladesh Acetate 20mg 3974/4096 3–59Fischer Walker et al16 (2006) Pakistan, Ethiopia, India Sulfate 10mg 554/556 1–5a Three study groups were examined (control, zinc syrup, and zinc/ORS). We included only those who received zinc syrup or a control.b Four study groups were examined: low/normal zinc in 2 intervention groups and low/normal zinc in 2 control groups. We combined the groups into either intervention or control, withoutexcluding those with low zinc levels.c Children up to 11 years of age were included; however, 45.1% (97 of 215) were 0 to 11 months of age; 38.1% (82 of 215) were 12 to 23 months; and only 16.8% (36 of 215) were�24 months. Allstudy participants were included in our analyses.d Three groupswere used (control, 5mgof zinc acetate, and 20mgof zinc acetate).We examined only thosewho used 20mgof zinc versus control subjects. Brooks et al enrolled onlymale children.
TABLE 4 Characteristics of Persistent Diarrhea Trials
Reference Country Zinc Supplement Zinc Dosage Zinc/Control Group, N Age, mo
Sachdev et al18 (1990) India Sulfate 20mg 20/20 6–18Roy et al21 (1998) Bangladesh Acetate 20mg 95/95 3–24Khatun et al34 (2001) Bangladesh Acetate 20mg 24/24 6–24Bhutta et al33 (1999) Pakistan Sulfate 3mg/kg 43/44 6–36Penny et al35 (1999) Peru Gluconate 20mg 139/136 6–35Bhandari et al36 (2002) India Gluconate 10/20mg 1228/1236 6–30
PEDIATRICS Volume 121, Number 2, February 2008 329 by on August 1, 2009 www.pediatrics.orgDownloaded from
trial25 found no difference at all between treatmentgroups. Participants in all 5 trials had been admitted fordehydration secondary to diarrhea, although the sever-ity of dehydration ranged. Four of the trials17,20,25,30 ad-ministered an ORS before treatment assignment. Threetrials received zinc sulfate and 2 received acetate. Incontrast, all acute diarrhea trials23,31,32 that provided zincgluconate and not zinc sulfate had a shorter duration ofdiarrhea than placebo (P � .08). Two trials17,20 originatedfrom India, 225,30 from Bangladesh, and 119 from Austra-lia. One trial15 in which average duration was signifi-cantly lower (1.2 days lower) with zinc use also had a
tremendously higher sample size (n � 8070) than all ofthe others.
Table 6 shows the effect sizes, calculated raw gravityvalues, standardized gravity values, and sample sizes foreach study when removed. It is clear that 1 study15 hada great deal of impact on the strength and direction ofthe estimated effect size value found for average dura-tion of acute diarrhea among all studies. When removed,the reaveraged effect size obtained (0.187) and plottedstandardized gravity value (3.531; Fig 2) were consid-ered outlying values in comparisons with all others. Thisis largely attributed to the enormous sample size (n �8070) used in the trial, because even very small differ-ences in mean duration of diarrhea would be statisticallysignificant.
Occurrence of Diarrhea at Day 1Five acute diarrhea trials16,19,20,27,32 reported the occur-rence of diarrhea at day 1 (n � 3100). No statisticallysignificant difference in the occurrence of acute diarrheaat day 1 was found (RR: 1.01; 95% CI: 0.99–1.03; P �0.30). Although the variability in effect sizes rangedfrom a low of 0.968 to 1.695, significant heterogeneitydid occur (Q � 10.60, dfQ � 4, P � .03, I2 � 62.3%).
Occurrence of Diarrhea at Day 3Six acute diarrhea trials16,19,20,23,27,32 collected data for oc-currence of diarrhea at day 3. No statistically significantdifferences occurred between treatment groups in occur-rence of diarrhea at day 3 (RR: 0.97; 95% CI: 0.91–1.03;P � .36); however, the occurrence of statistically signif-icant heterogeneity was found (Q � 10.880, dfQ � 5, P �0.05, I2 � 54.0%). Only 1 trial30 found a significantly(P � .01) lower occurrence of diarrhea at day 3 with zinc(27.4%) than placebo (35.4%; effect size: 0.774); how-ever, the occurrence of statistically significant heteroge-neity was found (Q � 10.880, dfQ � 5, P � .05, I2 �54.0%).
FIGURE 1Mean difference in duration of acute diarrhea. The effect size index in this plot is thestandard mean difference, so a point estimate of 0.0 indicates no effect. Values �0.0reflect a better outcome for the placebo group, and values �0.0 indicate a better out-come for the zinc group. If the point estimate and CI fell above 0.0, then the study wouldmeet the criterion for statistical significance (� � .05). If the CI overlapped 0.0, then the Pvalue would exceed .05 and the study would not be statistically significant.
N1 indicates sample size for zinc group; N2, sample size for the placebo group; Lower, lower limit of the 95% CI for the standard difference;Upper, upper limit of the 95% CI for the standard difference; Effect, standard difference; NS, nonsignificant.
330 LUKACIK et al by on August 1, 2009 www.pediatrics.orgDownloaded from
Occurrence of Diarrhea at Day 5Similarly, in the same 6 acute diarrhea trials,16,19,20,23,27,32
no statistically significant differences occurred between
treatment groups in occurrence of diarrhea at day 5 (RR:0.94; 95% CI: 0.84–1.05; P � .26). Similar to day 3results, the occurrence of statistically significant hetero-
FIGURE 2Standardized gravity results.
TABLE 6 Acute Diarrhea: Gravity Values for Duration of Diarrhea
Reference Effect Size Raw Gravity Standardized Gravity Sample Size
Valery et al19 (2005) 0.243 �0.00481 �0.332 215Strand et al32 (2002) 0.243 �0.00481 �0.332 894Sazawal et al31 (1995) 0.239 �0.00081 �0.056 937Sachdev et al17 (1988) 0.240 �0.00181 �0.125 50Roy et al30 (1997) 0.240 �0.00181 �0.125 74Polat et al29 (2003) 0.234 0.00419 0.289 182Patel et al20 (2005) 0.243 �0.00481 �0.332 200Hidayat et al28 (1998) 0.252 �0.01381 �0.953 1397Fischer Walker et al16 (2006) 0.249 �0.01081 �0.746 1110Faruque et al27 (1999) 0.242 �0.00381 �0.263 681Dutta et al26 (2000) 0.233 0.00519 0.358 80Brooks et al25 (2005) 0.243 �0.00481 �0.332 175Bhatnagar et al24 (2004) 0.240 �0.00181 �0.125 287Baqui et al15 (2002) 0.187 0.05119 3.531 8070Bahl et al23 (2002) 0.246 �0.00781 �0.539 805Al-Sonboli et al22 (2003) 0.237 0.00119 0.082 74
PEDIATRICS Volume 121, Number 2, February 2008 331 by on August 1, 2009 www.pediatrics.orgDownloaded from
geneity was found (Q � 18.957, dfQ � 5, P � .002, I2 �73.6%).
VomitingIn 11 acute diarrhea trials16,17,19,22–25,29–32 (n � 4438), theproportion of participants who vomited after the initialdose was significantly higher with zinc (278 [12.7%] of2196) use than with placebo (171 [7.6%] of 2242; RR:1.55; 95% CI: 1.30–1.84; P � 0.001%; Q � 25.54, P �.004).
Vomiting After Administration of Zinc Sulfate or GluconateIn 3 acute diarrhea trials,23,31,32 a significantly higherproportion of patients who received zinc gluconate vom-ited (160 [14.6%] of 1095) than zinc sulfate/acetatetherapy16,17,19,22,24,25,29,30 (118 [10.7%] of 1101; RR: 1.18;95% CI: 1.05–1.31; P � .006).
Shortening of Diarrhea DurationEight trials of acute diarrhea15–17,20,25,26,29,31 found an av-erage shortening of diarrhea duration of 15.0% for thosewho received zinc in comparison with placebo (Table 7).
Reduction in Stool FrequencySeven trials of acute diarrhea17,22,23,25,29,31,32 found an av-erage reduction in stool frequency of 22.1% with zinctherapy in comparison with placebo. One single trial16
found a 5.0% higher stool frequency using zinc thanplacebo.
Stool OutputThree trials of acute diarrhea24,26,30 found an averagelowering of stool output of 30.3%.
Probability of Diarrhea ReductionEight acute diarrhea trials20,23–25,27,28,30,32 measured theprobability of diarrhea reduction and found a 17.9%reduction using zinc compared with placebo.
Results for Persistent Diarrhea Trials
Duration of Persistent DiarrheaIn 5 persistent diarrhea trials18,21,33–35 (n � 489), thosewho received zinc also experienced a significantly loweraverage duration of diarrhea than the placebo group(WMD: 0.30; SE: 0.09; 95% CI: 0.12–0.48; P � .001;Table 8) but without significant heterogeneity (Q � 3.08,
TABLE 7 Effects of Zinc Therapy of Acute Diarrhea
Reference Country Stool Frequency Reduction Probability of Diarrhea Continuation
Sachdev et al17 (1988) India 18% lower frequency 9% shorter durationSazawal et al31 (1995) India 39% lower frequency 19% shorter durationRoy et al30 (1997) Bangladesh 28% lower stool output 14% reduction in probabilityFaruque et al27 (1999) Bangladesh Not reported 20% reduction in probabilityHidayat et al28 (1998) Indonesia Not reported 11% reduction in probabilityDutta et al26 (2000) India 38% lower stool output 32% shorter durationStrand et al32 (2002) Nepal 8% lower frequency 26% reduction in probabilityBahl et al23 (2002) India 17% lower frequency 11% reduction in probabilityAl-Sonboli et al22 (2003) Brazil 59% lower frequency Not reportedPolat et al29 (2003) Turkey 14% lower frequency 20% shorter durationBhatnagar et al24 (2004) India 25% lower stool output 30% reduction in probabilityValery et al19 (2005) Australia Not reported Not reportedBrooks et al25 (2005) India Not reported 19% reduction in probability, 7% shorter durationBrooks et al25 (2005) Bangladesh 0% lower frequency 12% reduction in probability, 0% shorter durationBaqui et al15 (2002) Bangladesh Not reported 24% shorter durationFischer Walker et al16 (2006) Pakistan, Ethiopia, India 5% higher frequency 9% shorter duration
Average stool frequency reduction� 18.8%; average lowering of stool output� 30.3%; average shortening of duration� 15.0%; average probability of diarrhea reduction� 17.9%. Variances indata reporting of outcomemeasures: For this meta-analysis, shortening of diarrhea durationwas defined as the percentage ratio of themean number of days of diarrhea in each study group. It wasthen reported as a shorter percentage of time with diarrhea for one group or the other. Probability of diarrhea duration was calculated by authors using various statistical approaches, such as theodds ratio, risk ratio, or hazards ratio. This difference in statistic negated a comparison in the meta-analysis. Stool frequency reduction was calculated by taking a ratio of the average diarrheafrequency in some studies per 24 hours or by the risk ratio of the mean number of stools in the first 4 days of another study. Lower stool output was calculated, in 2 studies, by taking a ratio of thetotal stoolweight per kilogramof bodyweight and reporting themedian. The ratio of themedianwas then taken. The resultingpercentagewas interpreted as a loweringof stool output in onegroupor the other. In another study, it was reported as the total stool output until the last first formed stool, measured in grams per kilogram for each group. The geometric mean was then taken and aratio between groups obtained. The group with the lower percentage was interpreted as a lowering of stool output in one group or another.
dfQ � 4, P � .544, I2 � 29.9%). Figure 3 depicts theForrest plot for these results.
Occurrence of Diarrhea at Day 1In 2 trials of persistent diarrhea34,35 (n � 221), no statis-tically significant differences occurred between treat-ment groups in occurrence of diarrhea at day 1 (RR:1.00; 95% CI: 0.93–1.08; P � .98), and no statisticallysignificant variability occurred among the effect sizes(Q � 0.01, dfQ � 1, P � .93).
Occurrence of Diarrhea at Day 3In 2 trials of persistent diarrhea34,35 (n � 221), a signifi-cantly lower occurrence of diarrhea at day 3 occurred inthose who were treated with zinc in comparison withplacebo (RR: 0.70; 95% CI: 0.51–0.94; P � .02). Nostatistically significant variability occurred among theeffect sizes (Q � 0.33, dfQ � 1, P � .56).
Occurrence of Diarrhea at Day 5This was not examined; fewer than 2 studies reported.
VomitingIn 4 persistent diarrhea trials18,21,35,36 (n � 2969), a sig-nificantly higher proportion vomited on zinc (41 [2.8%]of 1482) than with placebo (2 [0.001%] of 1487; RR:3.64; 95% CI: 1.02–13.02; P � .047; Q � 5.91, P � .116).
Vomiting After Zinc Sulfate or GluconateIn 4 persistent diarrhea trials,18,21,35,36 those who receivedzinc gluconate35,36 vomited more frequently (41 [3%] of1367) than did those who received zinc sulfate/acetate(0 [0%] of 115; RR: 1.09; 95% CI: 0.94–1.09; P � .07).
Shortening of Diarrhea DurationIn 4 persistent diarrhea trials,18,21,34,35 those who receivedzinc experienced a 15.5% average shortening of diarrheaduration than those who got a placebo (Table 9).
Reduction in Stool FrequencyFour trials of persistent diarrhea found that those whoreceived zinc also experienced an average of 9.8% re-duction in frequency.
Stool OutputStool output was not measured in the persistent trials.
Probability of Diarrhea ReductionTwo persistent diarrhea trials33,36 that measured theprobability of diarrhea reduction found an 18.0% reduc-tion when zinc was used over placebo.
DISCUSSIONOn the basis of these findings, which now add to thelarge body of previously published clinical data and up-date previous meta-analyses and systematic reviews,8,37
zinc therapy is useful for treating both acute and persis-tent diarrhea and for their prophylaxis. Still, as exten-sively addressed in a recent systematic review,6 muchinformation is lacking relative to the mechanisms bywhich zinc physiologically exerts its antidiarrheal effect.In this meta-analysis, 5 (31.3%) of 16 acute diarrheastudies17,19,20,25,30 found no statistically significant differ-ences between zinc and placebo on the average durationof diarrhea (at least a P � .48). Similarly, 2 (40.0%) of 5persistent diarrhea studies21,33 also found no statisticallysignificant differences in average duration of diarrheabetween treatments (at least a P � .43). Still, the averagestool frequency reductions, shortening of diarrhea dura-tions, and probabilities of a shortening of diarrhea dura-tion reported were higher in studies with zinc therapy incomparison with placebo.
To the majority of individuals, diarrhea means anincreased frequency or decreased consistency of bowelmovements. In many developed countries, the averagenumber of bowel movements is 3 per day; however,diarrhea is associated with an increase in stool weight,mainly as a result of excess water, which normallymakes up a large percentage of fecal matter. Given this,diarrhea is distinguished from diseases that cause onlyan increase in the number of bowel movements or fecalincontinence.
Determining the exact causes of diarrhea can be dif-ficult because there are many different diarrheal agents,with such a variety of infectious agents, including bac-teria, parasites, and viruses. Identification of specific di-arrheal agents is complicated by the lack of access tolaboratory tests in many developing countries. Viral gas-troenteritis caused by rotavirus is the primary cause ofdiarrhea among infants worldwide. Other causes includebacterial pathogens such as Vibrio cholerae, Shigella, andSalmonella. Protozoa such as Cryptosporidium parvum andGiardia lamblia are 2 of the most common protozoandiarrheal agents. The primary symptoms of rotavirusinfection are fever and vomiting for several days, fol-lowed by nonbloody diarrhea. Although not normallyfatal, the diarrhea caused by the virus can be quitesevere, leading to potentially life-threatening dehydra-tion. Although easily treated with intravenous fluids indeveloped nations, these supplies are often unavailablein the developing world, and the dehydration that iscaused by rotavirus is a significant cause of mortality.
In fact, conclusions from these randomized trials for
FIGURE 3Mean difference in duration of persistent diarrhea. The effect size index in this plot is thestandard mean difference, so a point estimate of 0.0 indicates no effect. Values �0.0reflect a better outcome for the placebo group, and values �0.0 indicate a better out-come for the zinc group. If the point estimate and CI fell above 0.0, then the study wouldmeet the criterion for statistical significance (� � .05). If the CI overlapped 0.0, then the Pvalue would exceed .05 and the study would not be statistically significant.
PEDIATRICS Volume 121, Number 2, February 2008 333 by on August 1, 2009 www.pediatrics.orgDownloaded from
the efficacy of zinc treatment on diarrhea duration in-cluded an improved absorption of water and electrolytesby the intestine and quicker regeneration of gut epithe-lium.38 Increased levels of brush border (apical) enzymessuggesting a zinc transporter for enterocytes39 and astronger immune response that increased clearance ofpathogens from the intestine40 were also described.
Efficacy of oral rehydration therapy in correcting de-hydration and reducing mortality led to treatment mod-ifications of ORS with zinc therapy. Success with zinctherapy has generally been attributed to a decrease inthe volume of small intestinal fluid and sodium absorp-tion triggered by zinc delivery. Still, the mechanisms bywhich zinc improves fluid and electrolyte transportationhave not been elucidated fully. This includes the effect ofzinc on intestinal ion transport, whether zinc initiates orincreases cation absorption and/or suppresses anion se-cretion, and whether deficiency enhances the likelihoodof secretory diarrhea.
Most likely, the location of the effect of zinc is in thesmall intestine, given its inhibition of adenosine 3�,5�-cyclic monophosphate (cAMP)-induced chloride-depen-dent fluid secretion. Treatment with ORS would have itsgreatest effect on reducing fluid loss by increasing smallintestine absorption. Thus, zinc therapy after pretreat-ment with ORS may not have shown a beneficial effect(reduced average duration of diarrhea) over placebo in 5trials17,19,20,25,30 of this meta-analysis simply because pre-treatment with ORS had already maximized the smallintestine absorption rate.
Zinc inhibits cAMP-induced chloride secretion by spe-cifically inhibiting basolateral potassium (K) channelswith no blockage effect on calcium (Ca)-mediated Kchannels in in vitro studies with the rat ileum.41 Zinc alsoinhibits cholera toxin–induced but not Escherichia coliheat-stable enterotoxin-induced ion secretion in cul-tured Caco-2 cells. One study42 showed that cAMP actedas the intracellular effector of heat-labile enterotoxin-induced fluid secretion. Guanosine 3�,5�-cyclic mono-phosphate mediates heat-stable–induced fluid secretion.If substantiated, then the effectiveness of zinc would be
limited to heat-labile–induced diarrhea or to diarrheamediated by cAMP but not either 3�,5�-cyclic mono-phosphate or intracellular Ca. It has been reported also43
that a zinc-sensing receptor triggers the release of intra-cellular Ca2� and regulates ion transport. A micromolarconcentration of extracellular zinc set off a massive re-lease of calcium from intracellular pools in the colono-cytic cell line. A sustained increase in intracellular Calevel may augment K efflux and a hyperpolarization ofcell membrane potential, leading to an advantageouselectrical gradient for chloride secretion.
Although the alternative treatment of oral rehydra-tion therapy is more available, there are still significantsetbacks in distributing the therapy. An antisecretorydrug vaccine would be a much more cost-effective solu-tion. An antisecretory drug vaccine could induce immu-nity without the children’s needing to go through mul-tiple infections and the risks associated with infections.By preventing children from acquiring infection, a drugvaccine could greatly reduce the number of deaths as aresult of diarrheal diseases and greatly reduce the bur-den on the health system.
The model for an antisecretory drug should performby inhibiting intestinal chloride and HCO3 secretion6 incontrast to focusing on decreasing gastrointestinal mo-tility and regeneration and/or restoration of gut epithe-lium. Accelerated research directed to achieving aclearer understanding of the biology, chemistry, andpathobiology of zinc in the gastrointestinal system isnecessary. Does zinc maintain intestinal defense sys-tems? What is the relationship of zinc to intestinalfluid balance? Definitively what are the linkages ofintestinal zinc transporters to body zinc status? Isthere a brush border (apical) membrane zinc trans-porter for enterocytes? Answers to these and otherquestions will hopefully drive the creation of a treat-ment drug that collectively induces cation absorption;inhibits anion secretion; reduces stool frequency andoutput; reduces diarrhea duration; and is safe, tolera-ble, and inexpensive.
TABLE 9 Effects of Zinc Therapy of Persistent Diarrhea
Reference Country Stool Frequency Reduction Probability of Diarrhea Continuation
Sachdev et al18 (1990) India 22% lower frequency 19% shorter durationRoy et al21 (1998) Bangladesh Not reported 7% shorter durationKhatun et al34 (2001) Bangladesh 7% lower frequency 17% shorter durationBhutta et al33 (1999) Pakistan 9% lower frequency 14% reduction in probabilityPenny et al35 (1999) Peru Not reported 19% shorter durationBhandari et al36 (2002) Nepal 12% lower frequency 22% reduction in probability
Average stool frequency reduction � 12.5%; average shortening of duration � 15.5%; average probability of diarrhea reduction � 18.0%.Variances in data reporting of outcomemeasures: For this meta-analysis, shortening of diarrhea duration was defined as the percentage ratio ofthemean number of days of diarrhea in each study group. It was then reported as a shorter percentage of timewith diarrhea for one group or theother. Probability of diarrhea duration was calculated by authors using various statistical approaches, such as the odds ratio, risk ratio, or hazardsratio. This difference in statistic negated a comparison in the meta-analysis. Stool frequency reduction was calculated by taking a ratio of theaveragediarrhea frequency in some studies per 24hours or by the risk ratio of themeannumber of stools in thefirst 4 days of another study. Lowerstool output was calculated, in 2 studies, by taking a ratio of the total stool weight per kilogram of body weight and reporting the median. Theratio of themedian was then taken. The resulting percentage was interpreted as a lowering of stool output in one group or the other. In anotherstudy, it was reported as the total stool output until the last first formed stool, measured in grams per kilogram for each group. The geometricmean was then taken and a ratio between groups obtained. The group with the lower percentage was interpreted as a lowering of stool outputin one group or another.
334 LUKACIK et al by on August 1, 2009 www.pediatrics.orgDownloaded from
ACKNOWLEDGMENTWe thank William D. Lyman, PhD, for help and sugges-tions in writing this article.
REFERENCES1. Black RE, Morris SS, Bryce J. Where and why are 10 million
children dying every year? Lancet. 2003;361(9376):2226–22342. Kosek M, Bern C, Guerrant RL. The global burden of diarrhoeal
disease, as estimated from studies published between 1992 and2000. Bull World Health Organ. 2003;81(3):197–204
3. Snyder JD, Merson MH. The magnitude of the global problemof acute diarrhoeal disease: a review of active surveillance data.Bull World Health Organ. 1982;60(4):605–613
4. Bryce J, Boschi-Pinto C, Shibuya K, Black RE. WHO estimatesof the causes of death in children. Lancet. 2005;365(9465):1147–1152
5. Claeson M, Merson MH. Global progress in the control ofdiarrheal diseases. Pediatr Infect Dis J. 1990;9(5):345–355
6. Hoque KM, Binder HJ. Zinc in the treatment of acute diarrhea:current status and assessment. Gastroenterology. 2006;130(7):2201–2205
7. Blanchard RK, Cousins RJ. Regulation of intestinal gene ex-pression by dietary zinc: induction of uroguanylin mRNA byzinc deficiency. J Nutr. 2000;130(5S suppl):1393S–1398S
8. Bhutta ZA, Bird SM, Black RE, et al. Therapeutic effects of oralzinc in acute and persistent diarrhea in children in developingcountries: pooled analysis of randomized controlled trials. Am JClin Nutr. 2000;72(6):1516–1522
9. Bhutta ZA, Black RE, Brown KH, et al. Prevention of diarrheaand pneumonia by zinc supplementation in children in devel-oping countries: pooled analysis of randomized controlled tri-als. Zinc Investigators’ Collaborative Group. J Pediatr. 1999;135(6):689–697
10. Comprehensive Meta-Analysis: A Computer Program for ResearchSynthesis [computer program]. Englewood, NJ: Biostat Inc;2003
11. Cohen J. The earth is round (p �. 05). Am Psychol. 1994;49(12):997–1003
13. Gee T. Capturing study influence: the concept of ‘gravity’ inmeta-analysis. Counsel Psychother Health J. 2005;1:52–75
14. SPSS 15.0 for Windows [computer program]. Version 15.0. Chi-cago, IL: SPSS Inc; 2006
15. Baqui AH, Black RE, El Arifeen S, et al. Effect of zinc supple-mentation started during diarrhoea on morbidity and mortalityin Bangladeshi children: community randomised trial. BMJ.2002;325(7372):1059
16. Fischer Walker CL, Bhutta ZA, Bhandari N, et al. Zinc supple-mentation for the treatment of diarrhea in infants in Pakistan,India and Ethiopia. J Pediatr Gastroenterol Nutr. 2006;43(3):357–363
17. Sachdev HP, Mittal NK, Mittal SK, Yadav HS. A controlled trialon utility of oral zinc supplementation in acute dehydratingdiarrhea in infants. J Pediatr Gastroenterol Nutr. 1988;7(6):877–881
18. Sachdev HP, Mittal NK, Yadav HS. Oral zinc supplementationin persistent diarrhoea in infants. Ann Trop Paediatr. 1990;10(1):63–69
19. Valery PC, Torzillo PJ, Boyce NC, et al. Zinc and vitamin Asupplementation in Australian Indigenous children with acutediarrhoea: a randomised controlled trial. Med J Aust. 2005;182(10):530–535
20. Patel AB, Dhande LA, Rawat MS. Therapeutic evaluation ofzinc and copper supplementation in acute diarrhea in children:
double blind randomized trial. Indian Pediatr. 2005;42(5):433–442
21. Roy SK, Tomkins AM, Mahalanabis D, et al. Impact of zincsupplementation on persistent diarrhoea in malnourishedBangladeshi children. Acta Paediatr. 1998;87(12):1235–1239
22. Al-Sonboli N, Gurgel RQ, Shenkin A, Hart CA, Cuevas LE. Zincsupplementation in Brazilian children with acute diarrhoea.Ann Trop Paediatr. 2003;23(1):3–8
23. Bahl R, Bhandari N, Saksena M, et al. Efficacy of zinc-fortifiedoral rehydration solution in 6- to 35-month-old children withacute diarrhea. J Pediatr. 2002;141(5):677–682
24. Bhatnagar S, Bahl R, Sharma PK, Kumar GT, Saxena SK, BhanMK. Zinc with oral rehydration therapy reduces stool outputand duration of diarrhea in hospitalized children: a randomizedcontrolled trial. J Pediatr Gastroenterol Nutr. 2004;38(1):34–40
25. Brooks WA, Santosham M, Roy SK, et al. Efficacy of zinc inyoung infants with acute watery diarrhea. Am J Clin Nutr.2005;82(3):605–610
26. Dutta P, Mitra U, Datta A, et al. Impact of zinc supplementationin malnourished children with acute watery diarrhoea. J TropPediatr. 2000;46(5):259–263
27. Faruque AS, Mahalanabis D, Haque SS, Fuchs GJ, Habte D.Double-blind, randomized, controlled trial of zinc or vitamin Asupplementation in young children with acute diarrhoea. ActaPaediatr. 1999;88(2):154–160
28. Hidayat A, Achadi A, Sunoto, Soedarmo SP. The effect of zincsupplementation in children under three years of age with acutediarrhea in Indonesia. Med J Indonesia. 1998;7(4):237–241
29. Polat TB, Uysalol M, Cetinkaya F. Efficacy of zinc supplemen-tation on the severity and duration of diarrhea in malnour-ished Turkish children. Pediatr Int. 2003;45(5):555–559
30. Roy SK, Tomkins AM, Akramuzzaman SM, et al. Randomisedcontrolled trial of zinc supplementation in malnourished Ban-gladeshi children with acute diarrhoea. Arch Dis Child. 1997;77(3):196–200
31. Sazawal S, Black RE, Bhan MK, Bhandari N, Sinha A, Jalla S.Zinc supplementation in young children with acute diarrhea inIndia. N Engl J Med. 1995;333(13):839–844
32. Strand TA, Chandyo RK, Bahl R, et al. Effectiveness and effi-cacy of zinc for the treatment of acute diarrhea in youngchildren. Pediatrics. 2002;109(5):898–903
33. Bhutta ZA, Nizami SQ, Isani Z. Zinc supplementation in mal-nourished children with persistent diarrhea in Pakistan. Pedi-atrics. 1999;103(4). Available at: www.pediatrics.org/cgi/content/full/103/4/e42
34. Khatun UH, Malek MA, Black RE, et al. A randomized con-trolled clinical trial of zinc, vitamin A or both in undernour-ished children with persistent diarrhea in Bangladesh. ActaPaediatr. 2001;90(4):376–380
35. Penny ME, Peerson JM, Marin RM, et al. Randomized, commu-nity-based trial of the effect of zinc supplementation, with andwithout other micronutrients, on the duration of persistent child-hood diarrhea in Lima, Peru. J Pediatr. 1999;135(2 Pt 1):208–217
36. Bhandari N, Bahl R, Taneja S, et al. Substantial reduction insevere diarrheal morbidity by daily zinc supplementation inyoung north Indian children. Pediatrics. 2002;109(6). Availableat: www.pediatrics.org/cgi/content/full/109/6/e86
37. Black RE. Zinc deficiency, infectious disease and mortality in thedeveloping world. J Nutr. 2003;133(5 suppl 1):1485S–1489S
38. Bettger WJ, O’Dell BL. A critical physiological role of zinc inthe structure and function of biomembranes. Life Sci. 1981;28(13):1425–1438
39. Gebhard RL, Karouani R, Prigge WF, McClain CJ. The effect ofsevere zinc deficiency on activity of intestinal disaccharidasesand 3-hydroxy-3-methylglutaryl coenzyme A reductase in therat. J Nutr. 1983;113(4):855–859
PEDIATRICS Volume 121, Number 2, February 2008 335 by on August 1, 2009 www.pediatrics.orgDownloaded from
40. Fenwick PK, Aggett PJ, Macdonald DC, Huber C, Wakelin D.Zinc deprivation and zinc repletion: effect on the response ofrats to infection with Strongyloides ratti. Am J Clin Nutr. 1990;52(1):173–177
41. Hoque KM, Rajendran VM, Binder HJ. Zinc inhibits cAMP-stimulated Cl secretion via basolateral K-channel blockade inrat ileum. Am J Physiol Gastrointest Liver Physiol. 2005;288(5):G956–G963
42. Canani RB, Cirillo P, Buccigrossi V, et al. Zinc inhibits choleratoxin-induced, but not Escherichia coli heat-stable entero-toxin-induced, ion secretion in human enterocytes. J Infect Dis.2005;191(7):1072–1077
43. Hershfinkel M, Moran A, Grossman N, Sekler I. A zinc-sensing receptor triggers the release of intracellular Ca2�and regulates ion transport. Proc Natl Acad Sci USA. 2001;98(20):11749 –11754
HIGH-STAKES FLIMFLAM
“It’s time to rein in the test zealots who have gotten such a stranglehold onthe public schools in the US. Politicians and others have promoted high-stakes testing as a panacea that would bring accountability to teaching andsubstantially boost the classroom performance of students. ‘Measuring,’ saidPresident Bush, in a discussion of his No Child Left Behind law, ‘is thegateway to success.’ Not only has high-stakes testing largely failed to magi-cally swing open the gates to successful learning, it is questionable in manycases whether the tests themselves are anything more than a shell game.Daniel Koretz, a professor at Harvard’s Graduate School of Education, told mein a recent interview that it’s important to ask ‘whether you can trustimprovements in test scores when you are holding people accountable for thetests.’ The short answer, he said, is no. If teachers, administrators, politiciansand others have a stake in raising the test scores of students—as opposed toimproving student learning, which is not the same thing—there are all kindsof incentives to raise those scores by any means necessary. ‘We’ve now hadfour or five different waves of educational reform,’ said Dr. Koretz, ‘that werebased on the idea that if we can just get a good test in place and beat peopleup to raise scores, kids will learn more. That’s really what No Child LeftBehind is.’ The problem is that you can raise scores the hard way by teachingmore effectively and getting the students to work harder, or you can takeshortcuts and start figuring out ways, as Dr. Koretz put it, to ‘game’ thesystem. Guess what’s been happening? ‘We’ve had high-stakes testing, really,since the 1970s in some states,’ said Dr. Koretz. ‘We’ve had maybe six goodstudies that ask: “If the scores go up, can we believe them? Or are peopletaking shortcuts?” And all of those studies found really substantial inflation oftest scores.’”
Herbert B. New York Times. October 9, 2007Noted by JFL, MD
336 LUKACIK et al by on August 1, 2009 www.pediatrics.orgDownloaded from
following collection(s): This article, along with others on similar topics, appears in the
Permissions & Licensing
http://www.pediatrics.org/misc/Permissions.shtmltables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,
