Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance(review)

Transcription

1 Cochrane Database of Systematic Reviews Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance(review) Dickson RC, Awasthi S, Demellweek C, Williamson PR Dickson RC, Awasthi S, Demellweek C, Williamson PR. Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance. Cochrane Database of Systematic Reviews 2000, Issue 2. Art. No.: CD DOI: / CD Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance(review) Copyright 2007 The Cochrane Collaboration. Published by John Wiley& Sons, Ltd.

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY BACKGROUND OBJECTIVES METHODS RESULTS DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES FEEDBACK WHAT S NEW HISTORY DECLARATIONS OF INTEREST SOURCES OF SUPPORT NOTES INDEX TERMS i

3 [Intervention Review] Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance Rumona C Dickson 1, Shally Awasthi 2, Colin Demellweek 3, Paula R Williamson 4 1 Liverpool Reviews and Implementation Group, University of Liverpool, Liverpool, UK. 2 Department of Paediatrics, King George s Medical University, Lucknow (UP), India. 3 Other. 4 Centre for Medical Statistics and Health Evaluation, The University of Liverpool, Liverpool, UK Contact address: Rumona C Dickson, Liverpool Reviews and Implementation Group, University of Liverpool, 2nd Floor, Sherrington Building, Liverpool, Merseyside, L69 3GE, UK. rdickson@liv.ac.uk. Editorial group: Cochrane Infectious Diseases Group. Publication status and date: Unchanged, comment added to review, published in Issue 1, Citation: Dickson RC, Awasthi S, Demellweek C, Williamson PR. Anthelmintic drugs for treating worms in children: effects on growth and cognitive performance. Cochrane Database of Systematic Reviews 2000, Issue 2. Art. No.: CD DOI: / CD Background A B S T R A C T In communities where helminth (worm) infections are common, they may contribute to poor nutritional status, anaemia, and impaired growth and learning in children. The World Health Organization, the World Bank and others recommend that children are routinely given deworming drugs in developing countries. This requires resources to deliver. Objectives To summarise the effects of anthelmintic drug treatment in children in relation to their growth and cognitive performance. Search methods We searched the Cochrane Infectious Diseases Group Specialized Register (August 2004), CENTRAL) The Cochrane Library Issue 3, 2004), MEDLINE (1966 to August 2004), EMBASE (1980 to August 2004), and LILACS (August 2004). We contacted experts in the field. Selection criteria Randomised and quasi-randomised trials of drug treatment compared with placebo or no drug treatment for intestinal helminths in children. Data collection and analysis Trial quality was assessed and data were extracted independently by two reviewers. Study authors were contacted for additional information. Main results Thirty trials involving more than 1500 children were included. There was potential for bias from inadequate concealment of allocation. Studies varied in relation to target groups, drugs administered and treatment regimens. Compared to placebo or no drug treatment, drug treatment for helminths was associated with some positive effects on change in weight, height and skinfold thickness. However there was significant heterogeneity between the results of the trials. 1

4 There were some positive effects on mean weight change in the trials reporting this outcome; after a single dose (any anthelminth) the pooled estimate was 0.17 kg, (95% CI 0.10 to 0.25; fixed effects model assumed); and 0.38 kg (95% CI 0.00 to 0.77; random effects model assumed). Results from trials giving multiple doses showed mean weight change under one year of follow up of 0.10 kg (95% CI 0.04 to 0.17; fixed effects assumed); or 0.15 (95% CI 0.00 to 0.30; random effects assumed). At more than one year of follow up, mean weight change was 0.12 kg (95% CI to 0.26; fixed effects assumed) and 0.43 (95% CI to 1.47; random effects model assumed). Results from studies of cognitive performance were mixed and inconclusive. Authors conclusions There is some limited evidence that routine treatment of children in areas where helminths are common has small effects on weight gain, but this is not consistent between trials. There is insufficient evidence to know whether this intervention improves cognitive performance. Our interpretation of these results is that the current public health programme investments in this intervention, based on the expectation that there will be an improvement in growth and learning, are not based on consistent or reliable evidence. P L A I N L A N G U A G E S U M M A R Y Synopsis pending B A C K G R O U N D One third of the world s population is infected with one or more species of intestinal helminth (Warren 1993). While these infections can have acute and severe clinical effects, policy makers and health professionals are concerned with the impact of these infections on long term health status. Helminth infections have been associated with undernutrition, iron deficiency anaemia, growth stunting, and impairment of learning (Evans 1995, Savioli 1992). The relation between malnutrition, anaemia, and helminth infection is consistent with our understanding of worm infections (Warren 1984). The infections can interfere with appetite, decrease food absorption through vomiting and diarrhoea, and cause mild blood loss from the gut (Stephenson 1987). There is often a correlation between helminth infections in children and school attendance, school performance or performance on other tests of cognitive performance (Simeon 1994, Nokes 1994). The link could be related to levels of nutrition or to the fact that the parasites have specific and direct influences on cognitive function (Baddeley 1992). One problem in assessing changes in cognitive performance is the identification of appropriate assessment tools (Connolly 1993, Baddeley 1992). The majority of tests used to assess cognitive performance have been standardised for children in developed countries and there are questions about their applicability across cultures (Grantham , Baddeley 1995). Malnutrition, anaemia, and learning impairment have also been closely linked to poverty (Warren 1993). This association could be important in assessing the effectiveness of anthelmintic interventions for two reasons: (1) eradication of worms would have little effect on malnutrition, anaemia or cognitive performance if these were a direct result of poverty; and (2) environmental and economic circumstances associated with endemic worm infection in a community could mean that the population would remain vulnerable to helminth infection, even if drug programmes decreased infection rates for a time. Public health programmes designed to decrease the prevalence of intestinal helminths have focused on two areas: (1) decreasing transmission through improved sanitation and handling of human waste; and (2) reducing human infections through drug treatment. The potential advantage of drug treatment is a rapid reduction in infection. The problem with this approach is that these effects may be temporary as the population is rapidly reinfected (WHO 1996). Early sanitation and drug treatment programmes attempted to eradicate intestinal helminths from whole populations. Results of programmes implemented as early as 1913 indicated that eradication was an unrealistic goal and that such programmes should aim for infection control (Warren 1993, Cort 1922). These early attempts were impaired, in part, by ineffective and/or toxic drugs. Research has since provided effective, broad spectrum, single dose drugs that are relatively safe (de Silva 1997, Warren 1993). A sec- 2

5 ond limiting factor was the difficulty in improving living conditions and sanitation. This remains a barrier to helminth control programmes in developing countries today. In Japan, following the Second World War, an intensive drug treatment programme was implemented at the same time that sanitation conditions were being improved. Although this strategy proved successful in controlling helminth infections (Yokogawa 1985), the programme was both time and resource intensive and spanned a number of decades. Current international anthelmintic programmes promote a chemotherapeutic approach (Hall 1997, World Bank 1993, WHO 1990, 1996, 1997). These programmes are based on infection and reinfection data from prevalence and treatment studies in developing countries (Anderson 1991). Warren et al (1993) proposed an outline of chemotherapeutic strategies to control helminth prevalence. The treatment populations are divided into two groups: the total population, or children only. The authors describe three possible strategies for each of these groups. 1. No screening: treatment of all members of the population (population mass treatment); or treatment of children only (targeted mass treatment) 2. Screening for infection: treatment of all infected members of the population (population screening for prevalence); or treatment of infected children (targeted screening for prevalence) 3. Screening for intensity of infection: treatment of all members of the population with heavy infections (population screening for intensity); or treatment of children with heavy infections (targeted screening for intensity). Given the high intensity of roundworm and whipworm in children, and the cost of screening, the authors recommend that targeted mass treatment of children is likely to be the most cost effective method of decreasing helminth infections. On the basis of this information and the availability of broad spectrum, effective/safe drugs, children are the target of chemotherapeutic control programmes advocated by the World Health Organization (WHO 1990, 1996). The aim of these programmes is to treat all children in an attempt to reduce their worm burdens and to improve their nutritional status and cognitive performance (Nokes 1992, Stephenson 1993). Although studies have shown the effectiveness of available drugs to decrease parasite infection rates, not all researchers agree that these treatment programmes are effective in improving the nutritional status and cognitive performance of children (Greenberg 1981, Lai 1995). The complexity of helminth infections is likely to be responsible, at least in part, for the discrepancy of study findings and scientific opinion. It is worthwhile examining some of these issues. One way to do this is to divide them into three categories: the intensity of infection; the importance of multiple infections; and the effectiveness of current treatments. The intensity of helminth infections is age related. Heavy worm burdens tend to be found in specific age groups with young children often harbouring the largest number of parasites. In addition, certain individuals are known to carry heavier/more intense infections. Not only do these individuals harbour more parasites, but following treatment infection intensity is likely to return to these high pretreatment levels. Specific helminth infections do not occur in isolation. The majority of people harbouring helminth infections have multiple infections, with Ascaris, Trichuris and hookworm being the most common. The consequences of independent infections may be different, as might the consequences of a combination of infections. Available drugs are not equally effective against these three species of helminth. Although these drugs have not been assessed in a systematic review, it is often said that the broad spectrum drugs (albendazole and mebendazole) are most effective against Ascaris and less effective against hookworm and Trichuris (de Silva 1997). Public health policies that impact on the long term health of children in terms of nutritional status or cognitive performance are not easy to evaluate. The time taken for the effects to become apparent is often long and outcomes may be confounded by other policy measures. On the other hand, public health policies often have considerable cost implications and it is important to know if they are effective and if the effect(s) they produce can be sustained. Routine anthelmintic programmes for children (with or without a screening component) are expensive and require governments to make long term organisational and financial commitments to maintain them. The effectiveness of such programmes should therefore be established before they are recommended for widespread implementation. In this review we examine systematically the available evidence of effectiveness of anthelmintic chemotherapeutic programmes in improving nutritional status and cognitive performance in children. O B J E C T I V E S To summarise the effects of anthelmintic drug treatment in children in relation to their growth and cognitive performance. M E T H O D S Criteria for considering studies for this review 3

6 Types of studies All randomised trials in which children were treated with drug interventions for intestinal helminths. Trials which used pseudorandomisation techniques or alternate allocation to treatment or control groups were also included. Types of participants Children aged 1-16 years. Types of interventions anthelmintic drug treatment of children, administered at any location (such as health facilities, schools or communities). anthelmintic drugs listed in the British National Formulary (1997) for use in treatment of the selected common helminth infections (see search strategy for list of infections) were considered in the review. These included mebendazole, piperazine, albendazole, levamisole, pyrantel, and thiabendazole. In addition, drugs previously used, or currently being evaluated, in the treatment of these infections were considered. These included bephenium, tetrachlorethylene, metronidazole and ivermectin. Types of outcome measures Nutritional status as measured by growth or measures of cognition. Studies that reported only parasite prevalence or intensity as outcome measures were excluded from the review. Search methods for identification of studies The topic terms used for searching trial registers and databases were: Ancylostoma duodenale/necator americanus, Ascaris lumbricoides, Enterobius vermicularis, Trichuris trichiura, Strongyloides stercoralis, hookworm, roundworm, pinworm, whipworm, and strongyloidiasis; albendazole, mebendazole, piperazine, levamisole, pyrantel, thiabendazole. The Cochrane Infectious Diseases Group Specialized Register was searched for relevant trials up to August Full details of the Cochrane Infectious Diseases Group methods and the journals handsearched are published in The Cochrane Library in the section on Collaborative Review Groups. The Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library (Issue 3, 2004), was searched. Full details of the sources and methods used are published in The Cochrane Library. The following databases were searched using the topic search terms in combination with the Cochrane highly sensitive search strategy for identifying trials and detailed in the Cochrane Reviewers Handbook (Clarke 2002): MEDLINE (1966 to August 2004) EMBASE (1980 to August 2004) LILACS (La Literatura Latinoamericana y del Caribe de Informacion en Ciencias de Salud) ( accessed August 2004). Organisations and individuals working in the field were contacted. A panel of individuals supporting this review and the external referees have checked the completeness of the search strategy and the efforts made to identify unpublished, on-going, and planned trials. The reviewers also drew on existing reviews of the topic and checked the citations of all the trials identified by the above methods. Data collection and analysis All trials identified by the search strategy were entered in a trials register. Using a two-stage unblinded selection process two reviewers (SA, RD) independently applied the inclusion criteria. Disagreements were referred to the review editor and a consensus reached. The methodological quality of each trial was assessed independently by two reviewers (RD, PW) using the guidelines of the Cochrane Infectious Diseases Group. Trials were classified according to allocation of concealment, generation of allocation sequence, and inclusion of all randomised participants. Data were cross checked and discrepancies were settled through discussion. Data on growth outcomes were extracted independently by two reviewers (SA, RD) using previously designed data abstraction tables. Data was double checked and discrepancies settled through discussion. Data related to cognitive performance were extracted by one reviewer (CD) and checked by a second reviewer (RD). Attempts were made to contact authors of trials where data were thought to be incomplete, not in a form suitable for inclusion in the meta-analysis or where further information was required to assess the quality of the study. For each outcome specified, a pooled estimate of treatment effect across the studies was calculated (odds ratio for dichotomous outcomes, weighted mean difference (WMD) for continuous outcomes). A chi-square test for homogeneity was applied. A priori factors thought to influence treatment effects included: site of intervention (school, community or health facility); prevalence and intensity of infection; and age of the children. Two additional heterogeneity factors were later identified in discussions with researchers in the field. These were the pre-study nutritional status of the children, and the manufacturer of the drug administered. Further evaluation of these factors is currently being carried out. Analysis included comparison of single or multiple dose trials. Within these categories analysis was carried out comparing each drug with placebo independently and overall. Where available, post-randomisation data were abstracted for the following: Anthropometric measures: weight; height; body mass index; mid upper arm circumference; skin fold thickness (triceps and sub- 4

7 scapular); fitness/activity measures. Cognition/learning measures: any test of cognition in children; academic performance; school absenteeism. Laboratory markers: iron status including haemoglobin, serum ferritin, and erythrocyte protoporphyrin; vitamin A status Side effects R E S U L T S Description of studies A total of 30 trials reported in 36 articles met the inclusion criteria. Of these, 27 trials are in published reports, one is an unpublished report, one is a draft report, and one is unpublished preliminary data from an on-going trial. The results of all included trials were available in English. Details of each included trial are provided in the table of included studies. The trials were carried out in 17 different countries. These included: Bangladesh and Kenya - three studies in each; Ethiopia, Guatemala, Haiti, India, Indonesia, Jamaica, South Africa, and Tanzania - two studies in each; Botswana, Brazil, Malaysia, Myanmar, Sierra Leone, Sri Lanka, and Zaire - one study in each. Children included in the trials were recruited from school populations in 17 trials, rural communities in nine trials, and through health clinics or health workers in four trials. Twenty-four of the included trials used what Warren et al (1993) refer to as targeted mass treatment. That is, all children were treated regardless of infection status. The other six trials used a targeted screening for prevalence model where children were screened for infection and then infected children were included in the trial. Of these six trials, two measured only growth, three only cognitive performance, and one measured both. Most trials compared drug treatment with placebo except for one trial that included a nutritional supplement, one that included the use of iron treatment, and one that compared the use of two different drugs. Drugs assessed in these trials included: albendazole - five studies single dose trials and five multiple dose trials (one included a nutritional supplement); piperazine - three single dose trials and one multiple dose trial; mebendazole - one single dose trial and three multiple dose trials; levamisole - one single dose trial, one multiple dose trial, and one single dose trial where it was used in combination with iron therapy; pyrantel - one single dose trial and one multiple dose trial; tetrachlorethylene - one single dose trial; piperazine and metronidazole combined in one trial; pyrantel versus albendazole in one trial. Follow-up periods for single dose trials extended from 4 weeks to 1 year, while multiple dose trial follow-up periods ranged from 26 weeks to 6 years. Nineteen of the included trials reported the evaluation of the impact of treatment on Ascaris, Trichuris or hookworm. These evaluations were reported as prevalence rate of specific helminth or in actual egg counts. These 19 trials provided pre- and post-treatment measures of infection. Prevalence of at least one of these three helminths was high in all of these trials. Except where noted in the table of included studies, helminth prevalence was decreased in the treatment groups of each trial. Nutritional outcomes reported included mean weight, height, mid upper arm circumference, skin fold thickness (triceps and subscapular), and haemoglobin concentration as well as changes in each of these measures. In addition, two single dose trials provided data on physical performance (Harvard step test), while one study provided a comparison of body mass index. Three trials reported measures of appetite. Cognitive performance Six trials attempted to evaluate the effects of treatment on some measure of cognition. Three of these trials evaluated albendazole. One trial evaluated levamisole combined with iron while two examined mebendazole. A number of cognition tests that evaluate a range of mental functions were used in the trials (see data tables for the listing of tests used in each trial). Follow-up periods ranged from 2 weeks to 26 weeks. Two studies attempted to measure school attendance, while one examined educational attainment. in included studies Of the 30 included studies three appear to have adequately concealed allocation, while 24 make no mention of concealment of allocation. A further three trials mention that participants were listed according to variables such as age, sex or intensity of infection and then randomly allocated to treatment or control groups. Although it may be that these trials had adequate concealment of allocation it is not possible to make a definitive judgement based on the information currently available. Twenty-eight trials stated that participants had been randomised. Two trials stated they were double-blind trials but make no mention of randomisation or methods of blinding. Four trials mentioned the use of random number tables, one the use of random numbers, one the use of odd or even numbers, and one used the toss of a coin in allocation. The remainder make no mention of the method of randomisation. Five trials used cluster randomisation. The authors of one of these trials (Awasthi 1997) have indicated that adjustment for clustering was carried out during the data analysis. Assessment of data in a second trial (Stoltzfus 1997) indicates that adjustment for the effects of cluster randomisation has taken place. The reports of the remaining trials give no indication that such considerations were included in their analysis. If this was not done, these trials may have underestimated the variability of the treatment effect and the 5

8 confidence intervals may be too narrow. This in turn would affect the meta-analysis by giving too much weight to the trial. Two trials (Awasthi 1997 and Stoltzfus 1997) provided data in a format that could be included in the meta-analysis. Twenty-eight trials appeared to use an intention to treat analysis in so far as they reported children being analysed in the groups to which they had been randomised. One trial re-grouped children post-randomisation according to the number of drug treatments actually received (Evans 1986). One trial (Michaelsen 1985) excluded tall children from their analysis since the results from these children exceeded the upper limits of the comparison reference values being used. Analysis of data according to different variables and at different times made the estimate of loss to follow-up in studies assessing growth outcomes problematic. As a general rule loss to follow-up was calculated on the basis of height or weight at the final testing time. In five studies the loss to follow-up is unclear. Eight studies had a loss to follow-up of less than 10%, while 15 had a loss to follow-up of between 10 and 35%. One study had loss to followup >40% with one other study (Thein 1991) reporting a loss to follow-up of over 60%. Trials varied in reporting the effects of treatment with regard to growth outcomes. Some trials reported actual weight or height while others reported these variables as combinations of weight for height or weight for age. In addition some authors reported only a selected sample of the data and data analysis (for example, Fernando 1983 performed three analyses of effect on weight but reported only one result). This apparent selective reporting is a concern as it is unclear whether a variety of nutritional outcomes were calculated but only positive findings reported. One study acknowledged financial support for the trial from a drug company, while four stated they received the drugs used in the trial from a drug company. Effects of interventions Growth outcomes Results from the trials were grouped for analysis into: single dose trials; multiple dose trials with outcomes at one year or less; and multiple dose trials with follow-up of more than one year. Readers are cautioned to use care in their interpretation of the summary statistics. In all of the analyses where there are statistically significant findings there is also significant heterogeneity in the results between studies. Further investigation of this heterogeneity is being examined by the reviewers. Single dose placebo controlled trials (follow-up 4 weeks to 11 months) Eleven trials evaluated a single dose treatment and reported growth and development outcomes. An additional four multiple dose trials recorded data following their first treatment dose and these data were included in the single dose analysis. A total of nine trials provided data in a way that could be used in the meta-analysis. Weight post-treatment Mean weight post-treatment was reported in eight trials. Drugs used were albendazole (five trials), mebendazole (one trial), piperazine (one trial), and pyrantel (one trial). No pattern emerged and there was no significant difference between treatment or control groups in the comparison by type of drug or overall (WMD kg, 95% confidence interval (CI) to 0.18). Mean change in weight post-treatment was reported in seven trials, six using albendazole and one mebendazole. Study findings showed a considerable range from small negative effect (WMD , 95% CI to -0.04) to demonstrating a major effect of 1.3 kg (95% CI 1.04 to 1.56). The WMD in the analysis shows a small statistically significant gain of 0.17 kg (95% CI 0.10 to 0.25) in treated children. Height post-treatment Mean height post-treatment was reported in six trials, five using albendazole and one mebendazole. No pattern emerged and overall there was no statistically significant difference between treatment and control groups (WMD 0.09 kg, 95% CI to 0.88). Mean change in height post-treatment was reported in seven trials, six using albendazole and one mebendazole. Effects ranged from no change to 0.6 cm and the combined results showed a statistically significant WMD in favour of treatment of 0.14 cm (95% CI 0.04 to 0.23). Mid upper arm circumference post-treatment Mean mid upper arm circumference post-treatment was reported in seven trials. Four trials tested albendazole and two tested piperazine and one mebendazole. The overall analysis showed no significant difference between the groups (WMD 0.08 cm, 95% CI to 0.27). Mean change in mid upper arm circumference post-treatment was reported in four trials using albendazole. The overall WMD showed a significant treatment effect of 0.28 cm ( 95% CI 0.21 to 0.35). Triceps skin fold thickness post-treatment Triceps skin fold thickness post-treatment was measured in four trials. Three trials used albendazole and one piperazine. Results varied across the trials with only one albendazole trial showing a significant improvement in the treatment group. The overall WMD favoured the treatment group (WMD 0.76 mm, 95% CI 0.32 to 1.19). Mean change in triceps skinfold thickness post-treatment were measured in three trials that used albendazole. Combined results indicate a statistically significant improvement in skinfold thickness for the treatment group (WMD 1.30 mm, 95% CI 1.15 to 1.45). Subscapular skin fold thickness post-treatment Subscapular skin fold post-treatment was measured in three albendazole trials. The overall WMD favoured the treatment group (0.65 mm, 95% CI 0.35 to 0.94). 6

9 Change in subscapular skin fold thickness post-treatment in the same three albendazole trials showed a significant overall WMD of 1.22 mm (95% CI 1.08 to 1.37). Mean haemoglobin concentrations post-treatment Mean haemoglobin concentrations post-treatment were reported in two trials using albendazole. There was no significant treatment effect (WMD 3.74 g, 95% CI to 7.96). Mean change in haemoglobin post treatment was reported in one trial using albendazole. The results showed no difference between treatment and placebo groups (WMD g, 95% CI to 0.16). Body mass index One albendazole trial reported body mass index and found no difference between treatment and control groups (WMD -0.20, 95% CI to 0.06). Harvard step test post-treatment Two trials using albendazole reported significant differences in test outcomes with treatment groups showing significantly better results with a WMD of 6.0 points (95% CI 4.31 to 7.69). Appetite Appetite (amount of food consumed for a snack) was measured in two trials, both testing albendazole (Adams 1994, Stephenson 1993). Both reported improvement in measures of appetite posttreatment. Data were not comparable so were not included in the meta-analysis. Five single dose studies did not provide information that could be directly compared with other studies. Greenberg (1981) - piperazine: showed no effect in 11 months on change in weight or height, weight-for-age, weight-for-height, height-for-age, triceps skinfold or mid upper arm circumference. Kloetzel (1982) - mebendazole: reported nutritional status as improvement, deterioration or no change over 10 months based on percent reference weight-for-age in children up to 60 months and percent reference weight-for-length in older children. They found no significant difference between the treatment and control groups. Koroma (1996) - albendazole: showed significant improvement in weight-for-height, weight-for-age, and height-for-age in children treated while untreated children showed no significant improvement in these measures, at six months. Michaelsen (1985) - tetrachlorethylene: showed no significant difference between treatment and control groups at five months post-intervention when haemoglobin concentrations and percent weight-for-height were compared. Beach (1999) - Albendazole: showed no significant differences between treatment and control groups. Subsequent stratification of participants by infection showed small positive effects in treatment groups. Overall, in single dose trials, treatment groups showed small but statistically significant improvements in measures of mean change in weight, height, and mid upper arm circumference. These changes were 0.17 kg, 0.14 cm, and 0.28 cm, respectively. There were also small, statistically significant improvements in mean and mean change in triceps and subscapular skinfold thickness measures, mean haemoglobin concentrations, and Harvard fitness tests. All but one of these comparisons are from studies that compared albendazole with placebo. Multiple dose placebo controlled trials Fourteen trials reported on the effectiveness of more than one dose of a drug. Drug treatment intervals varied from 2 to 6 months. Analysis was carried out comparing outcomes up to 1 year and those over 1 year. Two trials that assessed albendazole (Awasthi 1995, Awasthi 1997) and one that assessed piperazine (Fernando 1983) fell into the latter category. Results from trials measured up to 1 year: There was no difference between treatment and placebo groups for the measures of mean weight or height, mean change in height, mean mid upper arm circumference or change in haemoglobin concentration. Mean change in weight One trial examined the use of piperazine, metronidazole or these two drugs in combination and showed no difference between groups. A trial of mebendazole in various doses showed no significant effect. Results from the four trials that evaluated albendazole showed the treatment groups gained an additional 0.10 kg (95% CI 0.02 to 0.19). Overall the treatment effect was a statistically significant WMD of 0.10 kg (95% CI 0.04 to 0.17). Mean change in mid upper arm circumference Two trials that tested albendazole found a positive difference between the treatment and placebo groups with a WMD of 0.25 cm (95% CI 0.16 to 0.35). One trial using mebendazole showed negative effects (WMD cm, 95% CI to 0.05). Pooled effect size was 0.20 cm (95% CI 0.11 to 0.29). One albendazole study (Stephenson 1993) reported a statistically significant increase in mean and mean change in triceps skinfold thickness, and mean change in subscapular skinfold thickness measures in the treatment group. Seven of the multiple dose trials did not provide data that could be used in the meta-analysis. Evans (1986) - pyrantel: The authors did not conduct an intention to treat analysis because too many children in the control group had received anthelmintic treatment. The study reported an increase in weight and height in children who received treatment. They also reported on morbidity outcomes and found that treated children were less likely to develop measles. This trial, with its biweekly morbidity survey, points out the difficulty in maintaining treatment and control groups in research of this type. Fernando (1983) - piperazine: carried out a number of statistical assessments but report only data from selective groups and no conclusions can be drawn. Hadju (1997) - compared pyrantel, albendazole, and placebo: The authors found no significant changes in growth patterns in the treated groups versus the placebo group. However, this study was 7

10 confounded by decreases in infection prevalence rates in the control group and limited effectiveness of the drugs in the treatment groups. Lai (1995) - mebendazole: showed no difference in growth measures for treated children. Rousham (1994) - mebendazole: found no significant differences in growth measures in treated children. Thein (1991) - Levamisole: reported increases in height at 6 months and increases in weight at 24 months in the treatment group. Willett (1979) - Levamisole: reported an increase in weight in treated children at one year. In multiple dose trials, with a duration of less than 1 year, small but statistically significant improvements were shown in mean change in weight and mean change in triceps and subscapular skin fold measures. The changes in triceps and subscapular skin fold thickness measurements come from one trial of albendazole versus placebo. Results from trials measured at more than 1 year: Results from two trials of more than 1 year duration using albendazole were available (Awasthi 1995, 1997). One of these trials, the smaller trial of around 1000 children (Awasthi 1997), was randomised by child while the second larger trial (Awasthi 1995) was cluster randomised by slum area with 25 slums in each cluster. This larger trial reported data on 3999 children (aged 1 to 5 years) following 2.5 years of treatment. It showed a statistically significant difference in mean weight change (0.97 kg) between children in the control group treated with vitamin A alone and those in the treatment group who received vitamin A and albendazole. Mean (SE) weight change in the control group was 3.63 (0.12) kg while in the treatment group it was 4.60 (0.11) kg. The point prevalence of helminth infections in this group of children was low at only 25%. One other trial, which assessed the effectiveness of piperazine (Fernando 1983), reported only selective data that indicated increases in height in some children in the treatment group. Learning and cognition outcomes (trials=6) Details of the cognition tests used in each study are included in the analysis tables for other data. Two studies using albendazole attempted to assess the effects of anthelmintic treatment on academic progress. Simeon 1995 (407 children) examined performance on reading, spelling and arithmetic tests, as well as school attendance, and found no difference between treated and untreated children positive for Trichuris. Watkins 1996 (228 children) similarly examined reading vocabulary and attendance records of students treated with albendazole or placebo and found no difference in the groups at six months. Nokes (1992) found some beneficial effect on some aspects of cognitive functioning while one other study (Simeon 1995) failed to find this benefit, although multiple regression analysis indicated that children with heavy infections showed more improvement than those with lighter infections. Boivin (1993) found that levamisole combined with iron supplementation had beneficial effects on cognitive function. However, the authors could not separate the effects of the two treatments. In one other study (Kvalsvig 1991), where mebendazole was not effective in significantly decreasing helminth prevalence, there were no consistent changes in cognitive performance post-treatment. The authors attributed the lack of effect to the fact that the drug treatments had been ineffective in treating the helminth infections. They consequently attempted to replicate their study using helminth-specific treatment. However, a natural disaster (flood) precluded the collection of the post-treatment data in the second study. Hadidjaja (1998) compared the use of mebendazole alone or in combination with health education. Their groups however, were not comparable at inception and loss to follow-up was over 30%, with losses in some groups exceeding 50%. All studies that examined effects of treatment on cognitive performance also reported measuring growth factors such as height and weight. However, only one study (Watkins 1996) provided data in a form that could be used in the meta-analysis. Side effects Only one study of tetrachlorethylene (Michaelsen 1985) provided information about side effects of the administration of the drug. In this trial 17% of children suffered side effects that began one and a half hours after treatment. All symptoms disappeared within four hours. This drug is not in current use as an anthelmintic. D I S C U S S I O N There has been a progression, over time, towards conducting better quality trials to assess the effect of anthelmintic drug treatment on growth and cognitive performance in children. However, with a few exceptions, trials remain small and there has been a limited attempt to ensure concealment of allocation. Some trials have used cluster randomisation but have not consistently indicated whether the analysis of the resulting data has taken this into consideration. It is also possible that there has been selective reporting of trial results. Some authors report outcomes from selected groups of participants that benefited from treatment. In other studies authors have carried out a number of different analyses and report only those with significant findings. Some of these potential biases may have been addressed by the researchers, but not included in the published results. The authors of this review are attempting to contact researchers to obtain further information. The majority of studies have used a targeted mass treatment model, the first strategy as outlined by Warren et al (1993). That is, researchers have treated children, regardless of their individual infection status. 8

11 Growth outcomes Effects on growth were measured using standard anthropometric measures such as weight, height, mid upper arm circumference, and skin fold thickness (triceps and subscapular). These measures have the advantage of being simple, safe, and inexpensive and are easily and accurately obtained (Gibson 1990). However, researchers have reported these growth measures in a variety of ways, including: absolute values, changes in values, indices of values (such as weight-for-height, height-for-age), and comparison of values with accepted international reference data. These variations make it difficult to synthesise the results. It also raises questions about selective reporting. The analysis in this review identified significant heterogeneity between the results of the included trials. Some a priori factors of heterogeneity had been established. These included: site of the intervention (school, community or health facility); prevalence and intensity of infection; and age of the children. Discussions during the review identified that pre-study nutritional status of the children and the manufacturer of the drug administered might affect study findings. These areas of heterogeneity require further investigation and methods to do this are currently being discussed. Some researchers have argued that the mass treatment of children can have a worthwhile health impact even though a large randomised trial might show no significant effect on mean growth variables. The basis of this argument is that mass treatment will produce a significant health benefit in a subgroup of children who have high intensity worm infections. This was addressed in one study (Stephenson 1989) where multiple regression analysis was carried out and indicated a larger increase in growth rate in children who had either a higher egg count at exam 1 (for which they were treated with albendazole) or a larger decrease in egg count between exams. One way to address this issue of effect of infection intensity would be through a randomised trial designed to examine specifically the effects on growth in relation to intensity of infection. However, discussions during the recent review panel meeting indicate that, given the effectiveness of drug treatment to decrease worm burdens, it is unlikely that further treatment versus placebo trials will receive ethical approval. It could be argued that such trials are ethical since the long term effects (both positive and negative) are not known. A second way to address the issue is to obtain individual patient data from researchers involved in trials included in this review and to analyse the results, taking into consideration the intensity of infection. Examination of the included trials indicates that researchers involved in as many as 13 trials (including over 5600 children) potentially have this data. We are considering the potential value of pursuing this option. Another factor that needs to be considered in assessment of studies is that children in the placebo groups may have received anthelmintic treatment during the trial. In the study by Evans 1986, where fortnightly assessments were carried out, the researchers found that a proportion of children in the placebo group had in fact received intermittent anthelmintic treatment. The reported crossover rate in the Awasthi 1997 trial further indicates the difficulty in maintaining treatment groups within a study population. Nevertheless, some trials included in the review did report significant changes in growth in children in the treatment groups. Analysis shows a positive and significant mean change in weight in treatment groups of 0.17 kg (95% CI 0.10 to 0.25) in single dose trials. Of interest is that the multiple dose trial that showed the greatest impact on change in weight (0.97 kg) was carried out in a population where the point prevalence of infection in the population was only 25%. A question that needs to be addressed is whether these statistically significant differences in growth have an effect on the long term health status of children. Identifying other relevant health outcomes is important in future studies. Two studies in this review examined the effect of treatment on activity levels and appetite. These showed positive results in treated children. One trial attempted to examine morbidity measures such as episodes of diarrhoea, respiratory infection or measles (Evans 1986). The results of this study indicated that episodes of measles were less in treated children. Cognition Cognition studies conducted to date do not provide consistent evidence of an effect of anthelmintic treatment on performance in cognition tests. The reasons for discrepancies in study results are unclear. Since there is no universally accepted test battery to assess cognitive performance the studies have used different combinations of tests. However, even where the same tests have been included in different batteries the results vary. Thus anthelmintic treatment has been found to have beneficial effects on digit span (forwards) and verbal fluency in one study (Nokes 1992) but not in another (Simeon 1995). It has been suggested that anthelmintic treatment may be beneficial in children who are heavily infected (Simeon 1995). Whether or not this can account for the different findings remains to be determined. Cognitive performance tests are based on theories of cognition. In worm infection, it is not clear whether any putative effect on cognition is due to a direct effect on specific cognitive centres or an indirect effect of malaise and lethargy that accompany the infection. This makes it more difficult to choose cognitive outcomes (and thus the relevant test). On the other hand, it is not always 9

12 clear how a change in test scoring actually reflects in a person s ability to learn or to think. The effects of helminth infections in children is a multifaceted problem that includes the local effects of sanitation and the environment, and the global effects of poverty. To expect a single approach (e.g. drug intervention) to counter balance these other factors is simplistic. However, the treatment of helminth infections may prove to be one of the links in the chain of necessary interventions to improve nutritional status and cognitive performance in children. A U T H O R S C O N C L U S I O N S Implications for practice Significant heterogeneity exists between the results of the studies designed to measure the effectiveness of anthelmintic treatment on growth and cognitive performance in children in areas where helminths are endemic. Single or repeated chemotherapeutic treatment of helminth infections in children has been shown to have small effects in weight gain and improvement in other anthropometric measures, but the long term impact of this effect on the health status of these children is not known. There is no evidence of improvement in cognitive performance in children treated with chemotherapeutic anthelmintic agents. That is not to say that there is no effect, but that further refinement of testing and outcomes is required. A number of people engaged in global public health programmes for routine anthelminth control disagreed with our interpretation of the data. For people interested in some of the views in this field, we recommend you read this correspondence (Bundy 2000; Savioli 2000) and our response (Garner 2000). Implications for research To make informed future policy decisions related to the effectiveness of anthelmintic treatment on growth and cognitive performance, key issues need to be addressed in future research. Such trials need to: - implement simple policies of mass intermittent treatment in children taking into consideration worm endemicity, infection intensity, and effectiveness of chemotherapeutic agents; - provide consistent reporting of measurements of growth; - use practical measures of cognitive performance and learning, such as school attendance or performance in school examinations; - assess the effects of anthelmintic treatment on morbidity and/or mortality and effect on health status over longer periods of time. A C K N O W L E D G E M E N T S The authors wish to thank all of those people who gave of their time and expertise to comment on the protocol and drafts of this review, including Dr Marco Albonico, Dr Guy Barnish, Dr Don Bundy, and Dr Lorenzo Savioli. However, the data presented and inferences drawn remain entirely those of the authors. Their contribution to the review process does not imply they support the findings or the reviewers interpretation. This review was being conducted as part of the Effective Health Care Alliance Programme supported by the Department for International Development (UK). R E F E R E N C E S References to studies included in this review Adams 1994 {published data only} Adams EJ, Stephenson LS, Latham MC, Kinoti SN. Physical activity and growth of Kenyan school children with hookworm, Trichuris trichiura and Ascaris lumbricoides infections are improved after treatment with albendazole. J Nutr 1994;124(8): Awasthi 1995 {published data only} Controlling parasitic infection in children under five years of age: giving albendazole in conjunction with an Indian government Vitamin A supplement program. Philadelphia: International Clinical Epidemiology Network (INCLEN), 1995 Monograph No. 3. Awasthi 1997 {unpublished data only} Awasthi S, Pande V, Fletcher R. Efficacy and costeffectiveness of albendazole in improving nutritional status of pre-school children in the urban slums of Lucknow, North India. Unpublished manuscript, Beach 1999 {published data only} Beach, M.J. Streit, T.G. Addiss, D.G. Prospere, R. Lafontant, J.G. Lammie, P.J. Assessment of combined invermectin and albendazole for treatment of intestinal helminth and Wuchereria Bancrofti Infections in Haitian school children. American Journal of Tropical Medicine and Hygiene 1999;60(3): Boivin 1993 {published data only} Boivin MJ, Giordani B. Improvements in cognitive performance for schoolchildren in Zaire, Africa, following an iron supplement and treatment for intestinal parasites. J Pediatr Psychol 1993;18(2):