Complexities and Perplexities: A Critical Appraisal of the Evidence for Soil-Transmitted Helminth Infection-Related Morbidity

Background: Soil-transmitted helminths (STH) have acute and chronic manifestations, and can result in lifetime morbidity. Disease burden is difficult to quantify, yet quantitative evidence is required to justify large-scale deworming programmes. A recent Cochrane systematic review, which influences Global Burden of Disease (GBD) estimates for STH, has again called into question the evidence for deworming benefit on morbidity due to STH. In this narrative review, we investigate in detail what the shortfalls in evidence are. Methodology/Principal Findings: We systematically reviewed recent literature that used direct measures to investigate morbidity from STH and we critically appraised systematic reviews, particularly the most recent Cochrane systematic review investigating deworming impact on morbidity. We included six systematic reviews and meta-analyses, 36 literature reviews, 44 experimental or observational studies, and five case series. We highlight where evidence is insufficient and where research needs to be directed to strengthen morbidity evidence, ideally to prove benefits of deworming. Conclusions/Significance: Overall, the Cochrane systematic review and recent studies indicate major shortfalls in evidence for direct morbidity. However, it is questionable whether the systematic review methodology should be applied to STH due to heterogeneity of the prevalence of different species in each setting. Urgent investment in studies powered to detect direct morbidity effects due to STH is required.


Introduction
Soil-transmitted helminth (STH) infections are among the most prevalent of the neglected tropical diseases (NTDs), characterised by chronic and subtle impacts on human health and development. They rarely cause direct mortality; instead they are major contributors to morbidity. Morbidity effects of STH infection are difficult to quantify given the long duration of infection, often over many years, presence of other concurrent diseases, and factors such as poverty and malnutrition, to which they are strongly linked, and which can confound measures of STH-associated morbidity. Accurate quantification of STH-associated morbidity and disease burden is critical to rationalise large-scale deworming programmes.
Debate exists around the health benefits and cost-effectiveness of STH intervention strategies, rising to prominence with Cochrane systematic reviews reporting equivocal evidence of health benefits [1][2][3][4] and a statistical re-analysis of a major deworming trial from 1998-1999 that found differing results to original author conclusions [5,6]. In response, concerns have been raised about methodological bias in systematic reviews [7], the importance of not confining anthelmintic treatment to infected children [7,8], the economic importance of deworming [9], and the conclusions drawn in the replication analyses [9,10]. The Cochrane systematic review, the replication analyses, and the resultant discussions are positive insofar as they will further progress an evidence-enhancing research agenda, but there is a strong underlying imperative not to adversely influence international policy for mass deworming.
Global burden of disease (GBD) studies quantifying STH burden, with the most recent estimates published in 2012 [11], have also been subject to debate, in part due to the exclusion of certain morbidities because available evidence is deemed insufficient to justify inclusion. The findings from Cochrane systematic reviews influence what morbidities are included in GBD estimates, which, also, can be used to influence policy; therefore, it is important to understand what the underlying shortfalls are.
It is timely to revisit the pool of knowledge for STH impact on health, with a particular focus on how recent evidence contributes to knowledge gaps, and to critically appraise the approaches being used in systematic reviews, particularly the Cochrane systematic review, to disentangle why results are consistently inconclusive.
In this narrative review, we critically appraise narrative and systematic reviews and consider the recent observational and experimental literature, covering both STH morbidity and treatment associations with health outcomes. We explore the reasons for the reported lack of effect in the most recent Cochrane systematic review and investigate why conclusions regarding STH impact on haemoglobin do not concur with other systematic reviews [12,13]. This paper is not a meta-analytic paper (which would replicate previous reviews). By considering both wellknown and more recent evidence, we provide an updated perspective of where evidence is insufficient to enable conclusions on STH morbidity to be drawn and highlight where research needs to be directed in future.

Methods
We searched scientific literature in the MEDLINE database (January 2000 to January 2016) for evidence of any morbidity or mortality outcomes associated with Ascaris lumbricoides, Trichuris trichiura, and the hookworms Necator americanus, Ancylostoma duodenale, and Ancylostoma ceylanicum. The nematode Strongyloides stercoralis was not included in this review. Specifically, the following combination of text and Medical Subject Headings (MeSH) terms was used: ("Necator americanus" or "Ancylostoma duodenale" or "Ascaris lumbricoides" or "Trichuris trichiura" or "Ancylostoma ceylanicum" or hookworm or "soil-transmitted helminth") and (morbidity or mortality or anaemia or stunting or retardation or wasting or malnutrition or cognitive or cognition or impair). The search was limited by English language. This search aimed to (1) identify narrative and systematic reviews of STH morbidity and (2) identify recent research papers on STH-associated morbidity. Article abstracts were reviewed and literature was retrieved if there was specific reference to a morbidity or mortality outcome from STH or if they could not be excluded (i.e., if the abstract did not clearly indicate morbidity outcomes). Reference lists of identified articles were cross-checked. We further cross-checked peer-reviewed literature with information from United Nations, World Health Organization (WHO), and several other non-government organisation websites.
Once identified, narrative and systematic reviews were analysed to determine current knowledge and evidence gaps (Tables 1 and 2). Critical appraisal checklists were used to analyse systematic reviews [14]. For the systematic reviews, we focussed on the research question being investigated (null hypothesis), search and selection criteria, trials selected, inclusion or exclusion of factors such as concurrent diseases or interventions, definitions given by authors to "quasi" randomised controlled trials (RCTs), evidence rankings from authors, definition of trial participants, baseline measures, classes of infection intensity, intervention and outcome measures, consideration of absolute versus relative outcome measures, length of follow-up, pooling of results and sub-group analyses undertaken, and the conclusions drawn within manuscripts. We did not analyse bias or adjusting of intra-class correlations in cluster RCTs, but we did check that the systematic reviews had investigated these. We did not investigate the specific meta-regression methods of the models. Given the requirement for regular updates of Cochrane systematic reviews, we only included the most recent of these. We report our findings in relation to the most recent Cochrane systematic review.
We further investigated recent evidence to determine whether knowledge gaps identified in earlier reviews were being addressed. Informed by the evidence summaries (Tables 1 and 2), we repeated the literature search (Fig 1, Table 3), applying the following criteria: (i). If studies related to hookworm impact on haemoglobin or anaemia, we included only experimental evidence since the 2010 systematic review was done; (ii). If studies related to impact on haemoglobin or anaemia of other STH, or any STH impact on physical (e.g., stunting or wasting) or cognitive measures of morbidity, we included studies from 2006 (the date of the last major STH review) and considered all observational and experimental evidence, with the exception of case series and case reports; and (iii). In the absence of any identified epidemiological investigations into A. lumbricoides or T. trichiura acute complications, we included five case series, but not individual case reports.
We looked at coinfections among STH species. For STH and non-STH parasite interactions, STH and vaccine interactions, and associations with allergies, atopy, and asthma, we note that effects on morbidity are likely to be synergistic and that research is under way to investigate these effects. However, discussion of these is beyond the scope of the current review.
After applying the above search criteria, studies were further excluded from Table 3 if they had been analysed in one of the systematic reviews, if they did not report results for STH species separate from other parasites, or results on haemoglobin, anaemia, or a physical or cognitive measure (with the exception of the case series for A. lumbricoides and T. trichiura). For this reason, studies that reported prevalence or intensity of infection only (indirect morbidity measures), or changes in these following interventions, were not reported. Furthermore, this means that for some studies, not all study outcomes were included in our evidence tables. Critical appraisal checklists were again followed to assess these studies [14].  [19]; Tchuem Tchuenté, 2011 [35] Strong hookworm associations with intellectual growth retardation, cognitive performance in school, reduced school attendance, and educational deficits; however, evidence is not yet considered causal. Most of these effects may be attributable to iron deficiency. There are agreed strong associations between iron deficiency and cognitive performance. TDS also associated with a marked decrease in cognitive score tests. Regular deworming of schoolchildren has shown improvements in some measures of cognitive performance, educational achievement, and school attendance. Helminth species, including Schistosoma mansoni, are often comorbid in humans. Individuals harbouring multiple helminth species often harbour higher-intensity infections from each species than those with single-species infections; this may lead to additive impacts on morbidity as infection intensity is closely linked to severity of disease. STH are often co-endemic with malaria and Human Immunodeficiency Virus (HIV) and may increase host susceptibility to these and possibly other infectious diseases such as tuberculosis. STH and malaria in particular are known aetiological contributors to anaemia; the extent to which they may interact and potentially worsen anaemia is unknown. A. lumbricoides may reduce post-vaccination immune responses to tetanus vaccine and STH infection has been suspected to lower the efficacy of Bacille Calmette-Guerin (BCG) vaccine. Some studies have found no effect of anthelmintics on response to immunisation. Intestinal obstruction due to A. lumbricoides is regularly encountered in children from developing countries. Children tend to harbour more A. lumbricoides than adults. Complications of the biliary system occur in adults. Both of these conditions can be fatal. Numerous ectopic sites for A. lumbricoides migration have been encountered, often with risks, e.g., airway obstruction. Mortality rates from ascariasis are difficult to estimate. Deworming has been clearly shown to reduce the frequency of complications. However, intestinal blockage arising from anthelmintic purging of A. lumbricoides has been reported. TDS associated with heavy T. trichiura infection mainly occurs in children and leads to severe growth stunting and cognitive deficits that might not be reversible. Other TDS symptoms include chronic dysentery, rectal prolapse, anaemia, and clubbing of fingers. TDS can be fatal. Lighter T. trichiura infections are also problematic, especially in children.
Topic area: A. ceylanicum

Citation Reported evidence
Traub, 2013 [50] A. ceylanicum may have important morbidity, but sparse evidence is available. In experimental infections, A. ceylanicum can cause "ground itch" and moderate to severe abdominal pain, diarrhoea, and occult blood in the faeces accompanied by peripheral eosinophilia. Patent A. ceylanicum can produce chronic infections that may occur in high enough burdens to produce anaemia.

Findings from Observational and Experimental Evidence
Tables 1 and 2 provide a summary of evidence for STH morbidity derived from previous narrative and systematic reviews. These have been used to determine current knowledge and evidence gaps. Table 3 summarises the information for recent observational and experimental evidence, including 44 studies by key topic area. Whilst some general statistical and epidemiological comments have been provided on study design, processes, analyses, and limitations, we have not applied formal scoring criteria to these studies.

STH entry and establishment within the human host
No recent epidemiological studies addressing STH entry and establishment in the human host were identified. Sequelae (Table 1) associated with STH entry into the human host tend to be regarded as transient events, often reported as features of the STH life cycle rather than in terms of quantifiable morbidity on the host. Narrative reviews have reported a broad range of symptoms following larval migration to, and establishment within, the gastrointestinal tract ( Table 1). The only epidemiological studies likely to be ethically acceptable to quantify these symptoms would be cohort studies being conducted on people moving to endemic areas for More RCTs of pregnant women should be undertaken to strengthen evidence. Gulani et al., 2007 [13] RCTs and quasi-RCTs assessing routine deworming impact on Hb, in any population.
Ten of the studies used albendazole as the anthelmintic drug, three used mebendazole, and one used bephenium. Routine administration of anthelmintics results in a marginal increase in Hb.
This review did not distinguish between different STH species or account for intensity of infection, which may have underestimated true treatment effect [12,54]. Some of the conclusions differ from those of Taylor Moderate-to-heavy ascariasis was risk factor for anaemia. Stunting associated with moderate-to-heavy ascariasis and trichuriasis.
Post-treatment assessment showed no difference in the mean increments in growth indices between negatively-to-lightly infected and moderately-to-heavily infected groups. No effect of albendazole on mean birth weight or on proportion of low birth weight. Anthelmintic use during pregnancy showed no effect on perinatal mortality or congenital anomalies.
Sample size calculations based on different trial outcomes. Study powered to detect 0.3g/L difference in maternal Hb and 70g difference in infant birthweight for either intervention [84]. Adequate power to detect whole group effects but not perinatal mortality effects in subgroup analyses [84]. Excluded women with clinical requirements to treat low Hb; may have excluded those most likely to benefit from interventions [84]. Loss to follow-up did not vary between trial arms. Analyses adjusted for albendazole or praziquantel (according to treatment being investigated in the model TDS has not previously been described in adults. (Continued) Table 3. Cases selected on basis of high T. trichiura egg counts. Descriptive analysis; some morbidity data (e.g., persistence of dysentery) self-reported. There is a need to conduct detailed community studies in trichuriasis morbidity, effective treatment assessment, and clinical response in severely malnourished parasitised children [102].
* RCTs are focus given prior systematic reviews, however one cross-sectional study for T. trichiura is included.
# Two replication analyses [5,6] are not included in this table as they were analysed in [4]. doi:10.1371/journal.pntd.0004566.t003 the first time, or on very young children being exposed to STH for the first time. Quantifying infection-related or gastrointestinal-related morbidity from STH does not seem to be a current research focus.

Blood loss, haemoglobin, and anaemia
The evidence for negative impact of STH infections is strongest for morbidity due to blood loss and anaemia. There is clear evidence that hookworms in particular cause blood loss from feeding on mucosal tissue in the small intestine [23], thereby causing iron-deficiency anaemia. However, anaemia is of multifactorial origin [103], and it is difficult to disentangle the effect from all other potential confounders. Furthermore, there are likely to be different impacts of anaemia in different age groups. With the exception of two systematic reviews investigating impact in pregnant women [52,53] and one that investigated observational evidence separately for school-aged children and adults [12], systematic reviews have not investigated age-dependent effects. Whilst this is likely due to lack of evidence, this is particularly important for anaemia indicators. Evidence from systematic reviews and recent experimental studies is mixed; not all studies have found an impact of hookworm (or STH more generally) on either haemoglobin levels or anaemia. Three systematic reviews (one in non-pregnant populations, one in pregnant populations, and one across populations), confirm previous observational direct correlations between intensity of hookworm infection and reduced haemoglobin, or improvements in haemoglobin levels following deworming [12,13,53]. Two other systematic reviews (one in children aged 0 to 16 years, one in pregnant women) [4,52] found no evidence for reduced anaemia following deworming. Two RCTs found changes in STH intensity or haemoglobin levels, but not in levels of anaemia, following deworming [55,58]. This is possibly due to differing levels of hookworm burden [104] and hookworm species, different anthelmintics used, underlying nutritional status, or other confounding factors, as well as specific differences in systematic reviews as discussed below.
It is only since 2002 that deworming has been advocated during pregnancy (after the first trimester), with the result that experimental data on maternal and child outcomes are relatively recent and more evidence is needed. Whilst data are somewhat dated, iron-deficiency anaemia has been suggested to be responsible for 20% of maternal deaths worldwide [23], with estimates that hookworm causes at least 30% of moderate or severe cases of anaemia among this population group [105]. The WHO estimates that more than half of pregnant women in developing countries have morbidity related to iron-deficiency anaemia [27]. Given this high prevalence, quantitative investigations into the role of hookworm in anaemia-related maternal mortality are required. No estimates of hookworm-associated mortality are currently included in GBD calculations.
Studies vary with regards the involvement of T. trichiura in blood loss and anaemia. T. trichiura-associated blood loss has been previously inferred to only become significant in heavy infections [48]; newer evidence [59] is in agreement with this. Interestingly, in one recent study undertaken in a high T. trichiura and A. lumbricoides (but low hookworm) endemic environment, moderate to heavy A. lumbricoides infection was a risk factor for anaemia in school-aged children [82]. Further experimental evidence investigating the impact of STH on haemoglobin and anaemia, particularly in preschool-aged children and pregnant women, is required.

Physical development, fitness, and worker productivity
Whilst this is the focus area for which we found the greatest quantity of recent studies, results for impact of STH on measures of height, weight, and head circumference are mainly observational and vary widely. Some studies found impacts of T. trichiura, A. lumbricoides, and/or hookworm on height but not weight, weight but not height, both height and weight, height and head circumference, and some studies found no impacts at all (Table 3). Underlying prevalence and infection intensity varied considerably by geographic location and, although most studies adjusted for some potential confounders, it is difficult to control for the role of malnutrition within populations, which could have had a major influence on results. Generally, more studies reported associations with stunting and wasting than studies that did not. The Cochrane systematic review found that deworming may increase weight gain [4], but was inconclusive on other physical health measures.
There are inherent difficulties in detecting growth changes in school-aged children [106], with, amongst other things, an appropriate length of follow-up time required to adequately assess these. However RCTs need to balance sufficient follow-up to detect effects with the potential for STH re-infection. Given rapid growth and potentially greater STH morbidity (evidenced by intensity of infection data) in preschool-aged children and the fact that, in some areas, preschool-aged children are now included in deworming programmes, greater emphasis on investigating morbidity in this cohort is required. It could be that this age group is where the strongest evidence of effect may lie.
It is biologically plausible that young girls who grow poorly become stunted women with a greater chance of giving birth to low birth weight infants who are likely to be stunted in adulthood [37]. Whilst longitudinal investigations into the impact of chronic STH infection in girls, following their general health status as they become mothers, have not been conducted, studies investigating pregnancy outcomes are being conducted. In terms of albendazole impact on pregnant women and neonates, an RCT not included in the systematic reviews provides further evidence of the lack of clear benefit of albendazole on maternal or neonatal health [84]. Some observational studies (but again, not all) [88] reported associations between hookworm infection and low birth weight.
We found one recent attempt in the published literature to investigate effects of STH on longer-term schooling and worker productivity; this analysis found improved economic outcomes for a cohort of dewormed individuals followed over approximately 10 years [89], similar to a previous economic analysis [107]. There is negligible other direct evidence that STH infections reduce adult productivity. However, the health consequences, such as anaemia, are known to affect productivity, and hookworm could be a major contributor to this. More well-designed longitudinal analyses are needed.

STH impact on cognitive development, school performance, and absenteeism
The impact of STH on cognitive development is the area that has come under greatest scrutiny over the years. It is extremely complex to measure accurately. Cognitive psychology is a dynamic field, encompassing different theories of the interplay of a broad range of psychological and environmental factors. Impaired cognition is rarely from a single cause, with an array of cognitive tests needed to assess impacts such as STH on a range of cognitive functions [106]. It is perhaps not surprising that despite much research undertaken to investigate whether STH contribute to cognitive impairment in children, few conclusions have been able to be drawn. The Cochrane systematic review investigated three RCTs undertaken in children of known high-intensity infection status, specifically designed to measure cognitive outcomes, but did not meta-analyse these due to different outcome definitions and therefore drew no new conclusions. Recent experimental and observational studies further illustrate this. Firm evidence continues to be elusive. Recent evidence investigating school absenteeism is sparse.

Additional clinical morbidity from STH species
Severe clinical complications such as trichuris dysentery syndrome (TDS), or intestinal obstruction and hepatopancreatitis as a result of A. lumbricoides infections, are relatively rare and represent only a small portion of the disease burden [108], although they are sufficiently serious to warrant attention. Apart from very few case series, there is an almost complete lack of epidemiological investigation into quantifying these acute complications over recent years. This is important to highlight, as ascariasis is the most common of the STH infections and causes the majority of STH mortality [109]. In the absence of recent information, it is unclear how mortality estimates have been derived [110]. It is additionally unclear whether any recent estimates exist for one of the most serious presentations of ascariasis, hepatopancreatic ascariasis (HPA), or a disease known as recurrent pyogenic cholangitis (RPC) (caused by stone formation, usually around dead A. lumbricoides in the bile duct). This has been epidemiologically linked to recurrent biliary invasion by A. lumbricoides in endemic areas [111]. Similarly, there appear to be no detailed quantitative investigations of trichuriasis, particularly TDS, in recent years. TDS can cause major, acute disease that is sometimes life-threatening [28]. There are no recent empirical data on either global incidence of TDS or any attributable mortality.
Current STH prevalence and burden estimates do not include Ancylostoma ceylanicum, which recent data show to be the second most prevalent hookworm species after N. americanus, in some Asian countries [50]. Whilst evidence is scant, A. ceylanicum may have more severe morbidity than A. duodenale [112,113]. Studies of A. ceylanicum-associated morbidity are somewhat dated and highlight the evidence gap arising from a failure to investigate morbidity associations. Diagnosis of A. ceylanicum requires coprodiagnostic molecular biology techniques not readily available in developing countries; however, in the current era of large-scale deworming programmes, there is growing impetus for utilisation of contemporary diagnostic methods, hence renewed focus on this hookworm species is needed.
It is clear that many morbidities associated with STH cannot be investigated in an experimental design and, further, that many associations (such as maternal mortality from hookworm) are not feasible to investigate at all. In assessing evidence for deworming on STHassociated morbidities, the Cochrane Collaboration has exclusively considered RCT and quasi-RCT evidence in its systematic reviews of deworming. As has been raised elsewhere [7,8], this major limitation results in lack of consideration of a vast quantity of broader evidence of association. The Cochrane systematic review is assessed in detail below.

A Critical Appraisal of a Cochrane Systematic Review
Cochrane systematic reviews are regarded as the benchmark for high-quality evidence, utilising rigorous methodologies undertaken in accordance with a set protocol. In the most recent Cochrane review [4] on the health benefits of deworming, the authors considered length of trial follow-up and different assessment points. They undertook analyses by classes of STH infection intensity. They further investigated dose number as either single dose or multiple dose treatments. They were not able to undertake analyses by age group due to insufficient data. The authors considered only absolute measures of heights and weights. As is appropriate, trials were not pooled where the outcome definitions varied, such as cognitive tests.
Under the rules of the Cochrane Collaboration, the systematic review protocol must be produced prior to undertaking the review [114]. It is clear from the report that a protocol was developed. However, the protocol is not publicly available on the Cochrane website and we could not verify whether key points raised below were written in the protocol, or whether changes were made during the review process, possibly due to lack of data. The research objective of the Cochrane systematic review was "to summarise the effects of giving deworming drugs to children to treat soil-transmitted intestinal worms, in weight, haemoglobin, and cognition; and the evidence of impact on physical well-being, school attendance, school performance, and mortality" [4]. We believe this is a situation in which the null hypothesis and the intended subgroup analyses need to be clearly stated and be verifiable with the protocol. One interpretation of the null hypothesis from the stated objective is that "deworming does not improve the listed health outcomes." This is supported by the description of the participants as "infected children identified by screening in community trials. All children must have lived in endemic areas" [4]. This does not clearly imply consideration of unscreened children. Yet, for data synthesis, the participants are separately analysed according to these two groups: infected children and all children living in an endemic area. Thirty-seven of the 45 included RCTs were based on mass drug distribution of an unscreened population [4]. Therefore, the alternative interpretation of the null hypothesis also follows: "mass drug distribution as delivered to all children in endemic areas does not improve health outcomes." The objective, participants, and intended subgroup analyses need to be more clearly explained, as the two hypotheses require a major conceptual shift in interpretation and raise different questions in terms of included studies and how outcomes are determined. The issue of considering unscreened children has already been strongly criticised, primarily because international policy promotes treating all children in endemic communities, and a systematic review is not required to establish that treating uninfected children will have no health benefit on these children [7,8].
With the analysis of unscreened children, trials have been pooled irrespective of STH species, treatment types, and drug distribution strategies. Conceptually, pooling trials in systematic reviews is appropriate; however, for STH there is a very high level of heterogeneity, and this approach is prone to methodological flaws. By considering RCTs conducted in different locations in which screening has not been done, there is no baseline assessment of STH prevalence and/or intensity. The underlying assumption is that baseline prevalence is the same between intervention and control groups, which enables post-intervention assessment attributable to the intervention (provided randomisation adequately enabled control for confounding and that there was no systematic differential bias between groups). Whilst justifiable for RCTs, this causes a difficulty for systematic review methods, as there is marked STH heterogeneity in different endemic areas and, if baseline testing is not done, there must be another estimate to determine the STH of greatest prevalence in the population (e.g., from other epidemiological surveys) to ensure that heterogeneity is addressed when pooling RCTs. In the systematic review, no baseline of STH prevalences are reported. Whilst evidence is limited, there is sufficient prior knowledge of differential impacts of STH on morbidity outcomes, e.g., the role of hookworms, but not A. lumbricoides, in blood loss, to indicate that pooling of STH is not an accurate way of assessing morbidity. Similarly, it is also not accurate to pool different anthelmintic treatments of known and well reported [115,116] very different efficacies according to STH. Lastly, drug distribution strategies, particularly targeted delivery to school-aged children versus mass drug administration to all community members [117], is currently a major area of research investigation; differential impacts between school and non-school child cohorts are likely according to delivery strategies of targeted school programmes versus broader community treatments [118], or different frequency treatments as are recommended according to endemicity [117]. Such pooling of studies would cause dilution of effects due to different STH responsiveness to treatments with known differential efficacies, potentially delivered according to different strategies. Thus, the Cochrane authors appear to have not pooled like with like. It comes as no surprise that few conclusions can be drawn from the Cochrane systematic review.
The rationale for pooling studies may have been data-driven or aimed at replicating the deworming programme context. A more robust approach would be to apply the method of Smith and Brooker [12], who used known baseline prevalence of hookworm infection, analysed by hookworms only, and by anthelmintic drug classes separately. It is very probable indeed that there is insufficient evidence to undertake meta-analyses for many morbidity outcomes by different STH and different anthelmintics. However, this is not a valid rationale for pooling them. If such meta-analyses cannot currently be done, there is insufficient evidence to say that deworming does not contribute to health outcomes. As has been indicated by others [8], this is not the same as a lack of effect.
The lack of evidence is correctly and clearly pointed out by the Cochrane authors with regards treatment of children with known STH infection. Participant and trial numbers for many outcomes were very low and many of the results were not sufficient to meta-analyse. Lack of sufficient data was also the reason why subgroup analysis by age was not conducted. This is, however, an extremely important evidence gap given differing age prevalence profiles of STH and the likely greater morbidity in preschool-aged children (with the consequence that these effects, too, could be diluted across age groups). The authors conclude that most evidence is of very low, low, or moderate quality. The most useful and important conclusion of the systematic review is that there is a major shortfall in evidence for most morbidities to feed into meta-analyses in the first place. Until such time as evidence is generated, meta-analyses will not be able to appropriately assess the health benefits of STH interventions.
If the protocol was publicly available, we could specifically ascertain whether the included trials met the protocol, whether excluded trials did not meet the protocol, and whether other trials should have been identified under the terms of the published search strategy. Other authors have noted omission of RCTs that ideally should have been considered [8]. Further evidence that has not been provided in the most recent Cochrane systematic review include GBD estimates since 2003 and a 2010 systematic review that found statistically significant effects for some anthelmintics on haemoglobin [12]. Finally, the authors have raised the issue of young children choking on deworming tablets in their discussion, referring to an unreferenced WHO newsletter. Their viewpoint is not derived from their systematic analyses.

Conclusion
In this manuscript, we have found that there is a paucity of recently collected data to inform our knowledge of STH morbidity. In particular, relatively little quantifiable evidence of STH morbidity has been forthcoming in recent years. Of the evidence that has been provided, few firm conclusions can be drawn. Perhaps this, too, is a reflection of the insidiousness of STH. Alternatively, this could be partly a result of trials that are powered to measure different primary outcomes; secondary morbidity outcomes may therefore be inadequately powered for effects to be detectable. Furthermore, this may be because intervention trials are impossible to conduct over sufficient time periods to assess deworming impacts on morbidity in the manner that such programmes are delivered in real-world settings (i.e., repeated rounds administered throughout childhood). There is also a possibility that our review has applied selection criteria that could have excluded key evidence. We considered that applying these criteria was the most accurate way to differentiate between direct and indirect STH morbidity measures. The main discrepancy that the Cochrane systematic review highlights is insufficient and heterogeneous underlying evidence. This is reinforced by our own findings.
We do have an evidence problem regarding STH morbidity and health effects of deworming. The use of prevalence and, to a lesser extent, intensity of infection as indicators for intervention planning, monitoring, and evaluation may have reduced the impetus to investigate more direct morbidity measures. As a consequence, we might not currently be able to prove the benefits of deworming. Furthermore, evidence of morbidity may become increasingly hard to detect over time if prevalence and intensity continue to reduce in populations. This is obviously a good outcome, but a poor basis upon which to make any assessments. Our main conclusion is that further investments in appropriately designed studies that are powered to measure changes in direct STH morbidity indicators are urgently required.

Key Learning Points
• There is a paucity of quantifiable evidence of STH morbidity in recent years when assessed by direct morbidity measures such as changes in height, weight, haemoglobin, and cognition.
• The most recent Cochrane systematic review has assessed possible benefits of deworming on morbidity outcomes by pooling RCTs of deworming regardless of individual infection status, STH species, type of anthelmintic drug, and distribution strategy. In our opinion, this is methodologically inaccurate given current knowledge of STH heterogeneity. There may be insufficient evidence to prove benefits of deworming, but this is not the same as the authors' conclusion of lack of an effect.
• Careful consideration needs to be given to use of systematic reviews of RCTs for measuring improvements in morbidity from deworming. Furthermore, there needs to be clarification of the role of observational evidence for assessing STH-associated morbidity given that not all morbidity investigations are feasible within an RCT design.
• Studies designed to detect direct morbidity from STH are urgently required to strengthen evidence for deworming.
Top Five Papers