Background
Epidemiology
Amoebiasis is a parasitic disease caused by Entamoeba histolytica, a protozoan parasite distributed throughout the world. It is estimated that about 40 to 50 million people infected with E. histolytica develop amoebic colitis or extraintestinal abscesses, which result in up to 100,000 deaths per year (Walsh 1986; Li 1996; WHO 1997; Petri 2000). Amoebiasis is second only to malaria in terms of mortality due to protozoan parasites (WHO 1997). Prevalence rates of amoebiasis are highest in developing countries in Asia, particularly the Indian subcontinent and Indonesia, the sub-Saharan and tropical regions of Africa, and areas of Central and South America (Petri 1999). In these areas, the prevalence rates vary with the population studied, differing between countries and areas with different socioeconomic conditions, and with the diagnostic test used.
Antibodies were detected by enzyme-linked immunosorbent assay (ELISA) in 14.7% to 32.7% of asymptomatic preschool children in an urban slum in Bangladesh (Haque 1999; Haque 2001), in 19.7% of individuals in a slum community in north-eastern Brazil (Braga 1996), and in 12.7% of individuals in an urban area in Vietnam (Blessman 2002). In a rural area in Ecuador, antibodies were detected using various serologic tests in 64.6% of elementary school students (Gatti 2002). In 1998, during an outbreak of amoebiasis in Tblisi, Georgia, 9% to 14% of asymptomatic individuals were positive for antibodies (Barwick 2002). More recent studies that used an ELISA test or polymerase chain reaction (PCR) reported that the incidence of intestinal amoebiasis in highly endemic areas ranged from 13% to 67% in individuals with diarrhoea (Haque 1997; Abd-Alla 2002; Tanyuksel 2005; Rivera 2006; Samie 2006) and from 1.0% to 13.8% in asymptomatic individuals (Haque 1997; Braga 1998; Rivera 1998; Haque 2001; Ramos 2005). In a four-year prospective study, 80% of asymptomatic schoolchildren aged two to five years and living in an urban slum in Bangladesh were infected with E. histolytica at least once, as determined by stool antigen detection test (Haque 2006).
Infection is commonly acquired by ingestion of food or water contaminated with cysts of E. histolytica, but transmission also occurs through oral and anal sex, and contaminated enema apparatuses (Li 1996; Haque 2003; Stanley 2003). In developed countries, infection occurs primarily among travellers to endemic regions, recent immigrants from endemic regions, homosexuals, immunosuppressed persons, and institutionalized individuals (Reed 1992; Petri 1999). One study found that 0.3% of travellers returning from tropical regions had positive amoebic serology (Weinke 1990), while another study found that 47% had positive stool cultures identified as E. histolytica by PCR and isoenzyme typing (Walderich 1997). Despite more frequent infection with nonpathogenic E. dispar in those with acquired immune deficiency syndrome (AIDS), E. histolytica remains an important diagnostic consideration in people with human immunodeficiency virus (HIV) presenting with bloody diarrhoea (Reed 1992; Ravdin 2005).
Clinical manifestations
About 90% of people infected with E. histolytica have no symptoms of disease and spontaneously clear their infection, while the remaining 10% develop invasive disease (Walsh 1986; Gathiram 1987; Haque 2002; Stanley 2003). About 3% to 10% of untreated individuals with asymptomatic infection coming from areas endemic for amoebiasis develop symptoms of invasive amoebic disease within one year (Gathiram 1985; Haque 2001; Blessman 2003b; Haque 2002).
Intestinal amoebiasis commonly presents as ulcers and inflammation of the colon. This results in a complete spectrum of colonic signs and symptoms ranging from non-bloody diarrhoea to dysentery (acute diarrhoea with bloody stools), and to necrotizing colitis (severe inflammation of the colon) with intestinal perforation and peritonitis (infection of the abdominal cavity membranes) (Patterson 1982; Petri 1999; Ravdin 2005). Clinical symptoms of amoebic colitis include abdominal pain or tenderness, urgency to defecate, fever, weight loss, and diarrhoea or loose stools with mucus, blood, or both (WHO 1997; Haque 2003).
Amoebic colitis includes two clinical forms defined by the WHO Expert Committee on Amoebiasis as "amoebic dysentery" and "nondysenteric amoebic colitis" (WHO 1969). Amoebic dysentery is diarrhoea with visible blood and mucus in stools and the presence of haematophagous trophozoites (trophozoites with ingested red blood cells) in stools or tissues; sigmoidoscopic examination reveals inflamed mucosa with or without discrete ulcers. Nondysenteric amoebic colitis presents as recurrent bouts of diarrhoea with or without mucus but no visible blood and presence of E. histolytica cysts or nonhaematophagous trophozoite (trophozoites with no ingested red blood cells) in stools, and the results of sigmoidoscopic examination are usually normal.
The most severe complication of amoebic colitis is fulminant or necrotizing colitis. It occurs in 0.5% of cases (Petri 1999) and as many as 6% to 11% of people with symptomatic infection (Pelaez 1966; Brooks 1985). In necrotizing colitis, there is profuse bloody diarrhoea, fever, and widespread abdominal pain, frequently progressing to severe injury of the bowel wall, intestinal haemorrhage, or perforation with peritonitis (Haque 2003; Stanley 2003). Among these people, the case-fatality rate is more than 40% (Ellyson 1986; Petri 1999; Chen 2004). Young children, malnourished individuals, pregnant women, immunocompromised individuals, and those receiving corticosteroids are at higher risk for invasive disease (Adams 1977; Ellyson 1986; Li 1996; Stanley 2003). Extraintestinal complications of amoebic infection include abscesses in various organs, empyema (accumulation of pus around the lungs), and pericarditis (inflammation of membranes surrounding the heart) (Petri 1999; Ravdin 2005). In the treatment of necrotizing colitis and extraintestinal amoebiasis, surgery and additional antibiotics may be required aside from specific antiamoebic drugs (WHO 1985; Stanley 2003).
Method of diagnosis
In many countries where amoebiasis is endemic, diagnosis of amoebic colitis is commonly made by identifying cysts or motile trophozoites in a saline wet mount of a stool specimen. Finding trophozoites containing ingested red blood cells in the stool is considered by many to be diagnostic for amoebic colitis (Gonzalez-Ruiz 1994; Haque 1997; Tanyuksel 2003). The limitations of this method include its low specificity because it is incapable of differentiating E. histolytica from nonpathogenic species such as E. dispar or E. moshkovskii (Petri 2000; Haque 2003). The accuracy of microscopic methods is highly dependent on the competence of the diagnostic laboratory. Specific and sensitive means to detect E. histolytica in stools include stool antigen detection test and PCR techniques based on the amplification of the target parasite RNA and DNA (Haque 1995; Haque 1998; Petri 2000; Nesbitt 2004). Ideally, stool samples positive for E. histolytica on microscopy should be confirmed with stool antigen or PCR before treatment starts. Unfortunately, these tests are not routinely used and are not widely available for the diagnosis of amoebic colitis in many developing countries.
Public health and socioeconomic impact
In addition to being a potentially fatal disease, invasive amoebiasis has important social and economic consequences. The peak incidence of amoebic colitis is among children less than 14 years of age and a second increase is seen in adults more than 40 years old (Gathiram 1985; Wanke 1988). Amoebic colitis is a temporarily incapacitating disease that may require hospitalization in some individuals presenting with severe diarrhoea or dysentery. Amoebic colitis affecting adults in the wage-earning group may require several weeks of hospitalization and up to two to three months for full recovery (WHO 1985; Walsh 1986). Pregnant and postpartum women appear to have an increased risk of severe disease and death (Li 1996; Stanley 2003; Ravdin 2005). Persistent infection can impair physical and mental growth, and affect the nutrition and general development of children. Children with E. histolytica-associated diarrhoea during the first two years of life were 2.93 times more likely to be malnourished and 4.69 times more prone to be stunted (Mondal 2006). Another study demonstrated that malnutrition and amoebic dysentery were associated with cognitive deficiencies (Tarleton 2006).
Antiamoebic drugs for treatment
The goals of treatment for amoebic colitis are to treat the invasive disease and to eradicate intestinal carriage of the organism (Li 1996). E. histolytica may be found in the bowel lumen, in the bowel wall, and in tissues, including the liver (WHO 1969). Antiamoebic drugs vary in efficacy at the three sites where the parasites commonly exist and are generally divided into two classes based on their main site of activity. The luminal amoebicides act principally in the bowel lumen and the tissue amoebicides act principally in the bowel wall and the liver; see Table 1 for examples.
Metronidazole is considered as the drug of choice for treating invasive amoebiasis (WHO 1994; Medical Letter 2004; WHO 2005; AAP 2006). The standard regimen of metronidazole for the treatment of amoebic colitis is 500 to 750 mg given three times daily in adults and 30 to 50 mg/kg/day in children given for five to 10 days (WHO 1994; Medical Letter 2004; WHO 2005; AAP 2006). Although there are those who believe that this dose may have sufficient activity against both trophozoites and cysts (Powell 1970; WHO 1994; Li 1996), others believe that metronidazole is not reliably effective in eliminating cysts in the colonic lumen (Powell 1966; Powell 1967a; Powell 1967b; Powell 1969a; Powell 1969). Thus, the general recommendation is that patients with invasive amoebiasis should receive a luminal amoebicide after treatment with a tissue amoebicide, in order to eliminate any surviving organisms in the colon (WHO 1995; WHO 1997; Medical Letter 2004; AAP 2006). This recommendation is based on the assumption that drugs acting on different protozoal processes may enhance each other's effect. However, the evidence to support combination therapy has not been reviewed, and it is not known whether drug combinations reduce clinical symptoms or eradicate parasites more effectively compared with giving a tissue amoebicide alone. Furthermore, the increased complexity of combination regimens, additional drug costs, and possible increased adverse events, coupled with the unavailability of luminal agents in the market, act as major deterrents to compliance with this recommendation.
Adverse effects may occur even with conventional doses of metronidazole and include headaches, loss of appetite, nausea, metallic taste, and vomiting (WHO 1995; Tracy 2001). Individuals should avoid alcoholic drinks during metronidazole therapy because of vomiting, headache, flushing, and abdominal pain that may occur. Dizziness, convulsions, poor co-ordination, and numbness of the extremities are less common but more serious adverse effects that warrant discontinuation of metronidazole (Tracy 2001). Other nitroimidazole drugs with longer half lives, such as tinidazole, ornidazole, and secnidazole, allow shorter periods of treatment and appear to be better tolerated compared with metronidazole. These drugs have been used successfully when administered in shorter courses and have been recommended as alternative antiamoebic drugs to metronidazole (Haque 2003; Stanley 2003; Medical Letter 2004; WHO 2005; AAP 2006). Treatment failures have been reported with metronidazole with most failures attributed to incorrect diagnosis, unsuitable choice of drug, or failure to observe certain principles of treatment rather than drug resistance (Knight 1980; Wassman 1999). However, the induction of metronidazole-resistant E. histolytica strains in the laboratory suggests that indiscriminate use of antiamoebic drugs can result in an increased minimum inhibitory concentration against E. histolytica (Samarawickrema 1997; Wassman 1999).
A systematic review summarized the effects of different drug treatments for amoebic dysentery in endemic areas (Dans 2006). The systematic review included 12 randomized controlled trials and found that while ornidazole, secnidazole, and tinidazole were likely to be beneficial for treating amoebic dysentery, metronidazole was unlikely to be beneficial. The results of the trials were not combined, and no formal statistical methods were performed to determine summary measures of the effectiveness of the drugs.
Adequate therapy for amoebic colitis is necessary to reduce severity of illness, prevent the development of complicated disease and extraintestinal spread, and decrease infectiousness and transmission to others. In developing countries, where amoebiasis is common and most of the patients are treated in private practice or as hospital outpatients, the aim of treatment should be towards an effective, safe, and simple regimen that can be given on an outpatient basis.
A reliable summary of the evidence is needed to determine the best treatment for amoebic colitis. The occurrence of treatment failures and unpleasant adverse effects associated with metronidazole in some patients and the possibility of overt clinical resistance of E. histolytica to metronidazole make it imperative to investigate alternative treatment. The benefits of using combination regimens over monotherapy and single-dose regimens over longer regimens have to be determined. Furthermore, the effectiveness of potential new antiamoebic drugs has to be ascertained.
Objectives
To evaluate antiamoebic drugs for treating amoebic colitis. The review particularly aims to compare:
- single agent alternatives with metronidazole;
- any antiamoebic drug with placebo;
- combination regimens with monotherapy; and
- single-dose regimens with longer regimens.
Methods
Criteria for considering studies for this review
Types of studies
Randomized controlled trials. We excluded quasi-randomized trials.
Types of participants
Adults and children with clinical symptoms of amoebic colitis (as outlined in WHO 1997 and Haque 2003) and the demonstration of E. histolytica cysts or trophozoites in a stool sample, or E. histolytica trophozoites in a tissue biopsy or ulcer scraping by histopathology. We included individuals with positive E. histolytica/E.dispar on stool examination confirmed by E. histolytica antigen detection test or PCR.
We excluded trials that include only individuals with: asymptomatic infection and those requiring surgery or additional antibiotic therapy, such as fulminant or necrotizing colitis; peritonitis, intestinal perforation, or haemorrhage; or evidence of extraintestinal amoebiasis including hepatic amoebiasis.
Types of interventions
Intervention
Antiamoebic drugs, administered alone or in combination.
Control
Placebo or another antiamoebic drug.
Types of outcome measures
Primary
- Clinical failures, defined as the absence of E. histolytica in stools or scrapings but little or no relief of signs or symptoms, or with persistent rectal ulcerations on sigmoidoscopy (WHO 1969).
- Parasitological failures, defined as the persistence of E. histolytica cysts or trophozoites in stools or colonic ulcer scrapings, with or without the presence of symptoms or rectal ulcers (WHO 1969).
- Relapse, defined as reappearance of cysts or trophozoites of E. histolytica after the initial disappearance, with or without recurrence of clinical signs or symptoms of amoebic colitis after completion of treatment (Woodruff 1967).
Adverse events
- Serious adverse events (death, life-threatening event, hospitalization required or duration of hospitalization prolonged, development of a persistent or significant disability or incapacity, having offspring with a congenital anomaly or birth defect, or development of cancer (Hutchinson 1997)).
- Adverse events resulting in discontinuation of the treatment.
- Other adverse events, including gastrointestinal adverse events, systemic symptoms such as weakness or fatigue, central nervous system effects such as headache or dizziness, and dermatologic effects such as skin rashes.
Search methods for identification of studies
We searched for all publications that described randomized controlled trials on antiamoebic drugs for treating amoebic colitis, regardless of language or publication status.
Databases
We searched the following databases using the search terms and strategy described in Appendix 1: the Cochrane Infectious Diseases Group Specialized Register (September 2008); the Cochrane Central Register of Controlled Trials (CENTRAL) published in The Cochrane Library (2008, Issue 3); MEDLINE (1966 to September 2008); EMBASE (1974 to September 2008); and LILACS (1982 to September 2008). We also searched the metaRegister of Controlled Trials (mRCT) using 'amoebic' and 'amoeba' as search terms (September 2008). Additional electronic searches of PubMed were made using the format for highly sensitive search strategies for identifying reports of randomized controlled trials (Higgins 2005), on 10 September 2006 and on 11 February 2008.
Conference proceedings
We searched the electronic databases of the conference proceedings listed in Appendix 2 for relevant abstracts.
Organizations and pharmaceutical companies
To help identify unpublished and ongoing trials, we contacted researchers working in the organizations listed in Appendix 3, and the pharmaceutical companies and associated databases listed in Appendix 4.
Reference lists
We checked the reference lists and bibliographies of all studies identified by the above methods.
Data collection and analysis
Selection of studies
Two authors (MLM Gonzales and LF Dans) independently assessed the results of the literature search to determine whether the title or abstract of each trial described a randomized controlled trial. We retrieved full reports of all trials considered by one or both authors to be potentially relevant, as well as those that were unclear. We used a standard eligibility form based on the inclusion criteria to independently assess the trials. We resolved any disagreements through discussion, or where this failed, by consulting the third author (EG Martinez). If eligibility was unclear due to unclear or inadequate information, we attempted to contact the trial authors for clarification. We noted the reasons for excluding studies.
Data extraction and management
Two authors (MLM Gonzales and EG Martinez) independently extracted data from the trials using pre-tested data extraction forms. We collected details regarding the inclusion and exclusion criteria for the participants, treatment intervention given, total number randomized, number of participants in each group for all outcomes, drop outs and withdrawals, and numbers experiencing each outcome. For dichotomous data, we extracted the number of participants who experienced the event of interest and the number of participants randomized and analysed in each treatment group. We resolved any disagreements by referring to the trial report and through discussion. Where data were insufficient or missing, we made attempts to contact the trial authors. MLM Gonzales entered data for analysis using double data entry.
Assessment of risk of bias in included studies
Two authors (MLM Gonzales and LF Dans) independently assessed the risk of bias in each trial using a prepared form. We assessed the generation of allocation sequence and allocation concealment as adequate, inadequate, or unclear according to Jüni 2001. We noted who was blinded, such as the trial participants, care providers, or outcome assessors, and assessed the inclusion of randomized participants in the analysis as adequate if 90% or greater, and inadequate if not.
Assessment of reporting biases
We determined publication bias by looking for asymmetry in a funnel plot. The presence of asymmetry in the funnel plot suggests possible publication bias, although it may also indicate heterogeneity or poor methodological quality of the trials.
Data synthesis
We analysed data collected using Review Manager 5. For dichotomous outcomes, we calculated risk ratios (RR) with 95% confidence intervals (CI).
Stratification of results
The main comparisons were between any single antiamoebic drug and metronidazole (current standard therapy), any antiamoebic drug and placebo, combination regimens and monotherapy, and any single-dose regimen and longer regimens. We included but did not pool data from other trials that compared any antiamoebic drug with another antiamoebic drug and did not address any particular pharmacological or clinical question relevant to this review.
For trials reporting results at multiple or varying time points, we performed separate analyses for outcomes measured from the end of treatment to 14 days and from 15 to 60 days after the end of treatment. In trials comparing drugs with different treatment durations, we measured the time point in relation to the last day of the longest treatment period. We did not consider outcomes that were measured during treatment or before completion of treatment. Likewise, we did not include outcomes measured beyond two months because this could be a reinfection rather than true failure or relapse.
Heterogeneity
We calculated summary RR from meta-analysis using both a fixed-effect model (Mantel-Haenszel method), which assumes trial homogeneity, and a random-effects model (DerSimonian and Laird method), which accounts for trial heterogeneity.
We reported results using the random-effects model when there were differences between trials that may potentially influence the size of the treatment effect or when significant statistical heterogeneity was detected. We determined the presence of statistical heterogeneity among the same interventions by inspecting the forest plots for overlapping confidence intervals and by applying the Chi
Sensitivity analysis
We performed sensitivity analyses to assess the robustness of the overall estimates by calculating the results using all trials and then excluding trials of a lower methodological quality (ie trials with inadequate generation of allocation sequence, allocation concealment, or blinding, or trials where less than 90% of randomized participants were analysed), and excluding trials that were sponsored by pharmaceutical companies. Although pharmaceutical-sponsored trials may publish only where demonstrating positive treatment effects, it may also be possible that pharmaceutical-sponsored trials were conducted with better methodological quality because of adequate funds. We determined the effect of date of publication on the overall pooled effect in a sensitivity analysis when there were large differences in the publication dates. It was unclear if two trials (Misra 1977; Misra 1978) reported the same results and attempts to contact the authors for clarification were not successful. We entered these two trials as separate trials and carried out sensitivity analysis to determine if exclusion of the later trial would have an effect on the overall estimate.
Results
Description of studies
See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification.
Search results
We assessed 143 references included in the primary search until 23 September 2008 and excluded 112 trials (see 'Characteristics of excluded studies'). From additional searches of PubMed (September 2006 and February 2008), websites of relevant organizations and journals, reference lists and bibliographies of the retrieved trials, and contact with pharmaceutical companies, we considered 65 additional trials to be potentially relevant and retrieved the full reports; six trials were included. Of the 171 excluded trials, there were four trials that were duplicate publications of four other excluded trials (see 'Characteristics of excluded studies'). A total of 37 trials were included in the review (see 'Characteristics of included studies').
The 37 included trials enrolled a total of 4487 participants of which 1837 were adults, 1038 were children, and the rest were not identified separately as adults or children. All trial reports were in English, except for Huggins 1982 (Portuguese), Karabay 1999 (Turkish), and Donckaster 1964 (Spanish). The trials included in the review were published between 1967 and 2007.
Location
The trials were conducted in 15 different countries (see details in Appendix 5), 14 of which are considered to be highly endemic for amoebiasis. Eighteen trials were conducted in Asia: India (12), Indonesia (five), and Bangladesh (one). Six trials were conducted in Africa: Kenya (two), Egypt (two), Nigeria (one), and South Africa (one). Five trials were conducted in South America: Colombia (two), Brazil (two), and one trial in Chile (one). The other trials were conducted in the following countries: Mexico (two), Turkey (two), Iran (one), and Iraq (one). The remaining two trials were conducted in one industrialized country, Sweden.
The trials were conducted in a variety of settings (see details in Appendix 6): hospital setting (13), outpatient clinic (13), community (one), school (one). The study setting was not stated in eight trials. In one trial, most participants were treated as outpatients, but a few with severe symptoms were treated in the hospital (Toppare 1994).
Source of funding
Eighteen trials did not state the source of funding. Sixteen trials reported that a pharmaceutical company provided funding (Nnochiri 1967; Batra 1972; Naoemar 1973; Pudjiadi 1973; Panggabean 1980; Sitepu 1982; Tripathi 1986; Chunge 1989; Pamba 1990; Rossignol 2007) or supplied the study drugs (Kapadia 1968; Rubidge 1970; Misra 1974; Joshi 1975; Singh 1977; Davila 2002). At least one trial author was reported to be connected with the pharmaceutical company manufacturing the study drug in three trials (Asrani 1995; Salles 1999; Rossignol 2001), although the involvement of the company was not stated.
Participants' amoebic colitis
The trials used different inclusion criteria for the participants:
- Acute amoebic dysentery in 11 trials (Nnochiri 1967; Rubidge 1970; Batra 1972; Naoemar 1973; Pudjiadi 1973; Panggabean 1980; Sitepu 1982; Soedin 1985; Mohammed 1998; Karabay 1999; Mansour-Ghanaei 2003).
- Chronic or vague abdominal symptoms compatible with nondysenteric amoebic colitis, without bloody diarrhoea or other signs of intestinal invasion, in four trials (Huggins 1982; Pehrson 1983; Pehrson 1984; Padilla 2000).
- Clinical symptoms of intestinal amoebiasis, without distinguishing between amoebic dysentery and nondysenteric amoebic colitis (22 trials).
- Three trials stratified the participants during the analysis of outcomes into those with acute amoebic dysentery and those with nondysenteric amoebic colitis (Botero 1974; Botero 1977; Swami 1977).
- Two trials classified the participants as having invasive trophozoite forms and noninvasive cyst forms based on stool microscopy findings and analysed the two groups separately (Kapadia 1968; Pamba 1990).
- Two trials categorized the participants as having acute amoebic dysentery, subacute amoebiasis, or chronic amoebiasis based on the severity of symptoms and whether trophozoites or cysts of E. histolytica were present but analysed the participants as one group (Joshi 1975; Mathur 1976).
- Two trials classified the participants as having acute or chronic amoebiasis based on the duration of symptoms but analysed the participants as one group (Misra 1974; Tripathi 1986).
- Thirteen trials recruited and analysed participants with symptoms of intestinal amoebiasis or amoebic colitis together, regardless of whether they presented with dysentery or not.
Participant age
Participant age ranged from seven months to 80 years; see Appendix 7 for details. Adults only (ie those aged more than 15 years) were recruited in 15 trials, while 10 trials recruited only children. Both adults and children were recruited in the remaining 10 trials. Participant age was not stated in two trials (Kapadia 1968; Batra 1972).
Methods used to diagnose amoebic colitis
Stool microscopy with direct smear was used as the predominant method for determining the presence of E. histolytica cysts or trophozoites in stools (details in Appendix 8): concentration methods for better detection of cysts (16 trials); flotation technique (two trials); and polyvinyl alcohol fixative for the detection of trophozoites (one trial). One trial used stool culture for E. histolytica in addition to stool microscopy to evaluate parasitological response, but it was not used as an inclusion criterion in the trial (Batra 1972). Only one trial used stool antigen-based ELISA test (Rossignol 2007).
Concomitant infection with other intestinal parasites
Aside from E. histolytica, concomitant infection with other intestinal parasites was identified in 10 trials: giardiasis (Singh 1977; Prasad 1985; Tripathi 1986; Rossignol 2001); intestinal helminth infection (Pudjiadi 1973; Panggabean 1980; Sitepu 1982); and other intestinal protozoa and helminth infection (Pehrson 1983; Salles 1999; Davila 2002). Three trials explicitly stated that stool bacterial culture was done before enrolment and excluded those found to be positive for pathogenic bacteria (Toppare 1994; Karabay 1999; Rossignol 2007). Concomitant infection with other intestinal pathogens or bacteria was not examined or not mentioned in the remaining trials. Since clinical symptoms may not have been exclusively caused by amoebiasis in those with concomitant intestinal parasites and the effect of concomitant infection on eradication of E. histolytica by antiamoebic drugs is not known, data for E. histolytica infection alone were used in assessment of outcomes except in those trials that did not separate the data for those with single and mixed infections. Separate analysis for clinical outcomes for those with E. histolytica alone and those with concomitant infection with Giardia and E. histolytica was carried out in three trials (Prasad 1985; Rossignol 2001; Davila 2002).
Drug comparisons
The included trials contained a variety of comparisons and involved over 30 individual drugs and combinations. As shown in Appendix 9, we grouped the trials into the following categories (some trials are included in more than one category):
- single agent alternative versus metronidazole (17 trials);
- any antiamoebic drug versus placebo (four trials);
- combination regimen versus monotherapy (seven trials);
- single-dose regimens versus longer regimens (five trials); and
- other amoebic drug comparisons (nine trials).
More than two interventions were compared in six trials. Three trials compared different doses of the same drugs with standard or control groups: three dosages of quinfamide with placebo in Huggins 1982; two treatment durations of tinidazole with metronidazole in Awal 1979; and four dosages of MK-910 in Batra 1972. Ten different treatment groups were compared with placebo in Donckaster 1964, and three drugs used alone or in three different combinations were compared in Pamba 1990. Two brands of tinidazole and two brands of metronidazole were compared in one trial (Chunge 1989). For trials with more than two intervention groups, we combined multiple treatment arms as appropriate in one group and compared them collectively with the standard or control group. This is the recommended approach to avoid a unit of analysis error by not counting the placebo or control participants more than once in the same meta-analysis (Higgins 2008). For the trial comparing two brands of tinidazole and two brands of metronidazole, the two brands of tinidazole were combined as one group and compared with the two brands of metronidazole in the other group.
Duration of follow up
The follow-up period varied considerably between trials. Seventeen trials were followed up for about one month. Four trials were followed up only until the end of the treatment period (Kapadia 1968; Batra 1972; Pudjiadi 1973; Asrani 1995). Duration of follow up was less than 15 days in 10 trials (Huggins 1982; Sitepu 1982; Prasad 1985; Chunge 1989; Toppare 1994; Mohammed 1998; Padilla 2000; Rossignol 2001; Davila 2002; Rossignol 2007), while the longest period was 12 months (Nnochiri 1967).
Outcome measures
The primary outcomes in this review were clinical failure, parasitological failure, and relapse. Thirty-one trials evaluated both clinical and parasitological outcomes, and six trials evaluated parasitological outcomes only (Donckaster 1964; Nnochiri 1967; Pehrson 1983; Pehrson 1984; Padilla 2000; Davila 2002). The definition of clinical and parasitological cure or failure in the trials varied. Nine trials (Misra 1974; Joshi 1975; Mathur 1976; Misra 1977; Singh 1977; Swami 1977; Misra 1978; Awal 1979; Tripathi 1986) used the definitions set by the WHO Expert Committee on Amoebiasis (WHO 1969) where 'cure' was defined as symptom-free, ulcers healed, stools negative for E. histolytica, 'probable failure' was defined as persistent symptoms and rectal ulcerations despite disappearance of E. histolytica in stools or ulcer scrapings, and 'failure' was defined as positive E. histolytica with or without symptoms and rectal ulcers. For this review, 'probable failure' was interpreted as clinical failure and 'failure' was interpreted as parasitological failure based on the definitions given. Data for clinical and parasitological outcomes were presented as dichotomous data in most trials.
The duration of time from the start of treatment to resolution of diarrhoea and other clinical symptoms was also presented in seven trials (Batra 1972; Naoemar 1973; Pudjiadi 1973; Toppare 1994; Rossignol 2001; Mansour-Ghanaei 2003; Rossignol 2007), and the duration of time from start of treatment to disappearance of E. histolytica in stools in three trials (Naoemar 1973; Pudjiadi 1973; Karabay 1999). Two trials reported on the number of stools passed during treatment (Botero 1977; Tripathi 1986), and also on days seven and 21 after treatment (Botero 1977), while another reported average daily frequency of stools on admission and at the end of days five and 10 of treatment (Asrani 1995). One trial reported on cumulative daily clearance of E. histolytica from stools during treatment and on days 15, 30, and 60 after the start of treatment (Pamba 1990). One trial assessed clinical and parasitological outcomes jointly as 'cure' (Prasad 1985). Only dichotomous outcomes were included in the analysis because of inconsistent data in trials also reporting continuous data. Relapse or recurrence was reported in two trials (Naoemar 1973; Botero 1974) and described but not reported in another trial (Pamba 1990).
Measurements of clinical and parasitological outcomes were made at different time points. Thirteen trials reported outcomes between the end of treatment to 14 days, and 15 trials reported outcomes from 18 to 30 days after the end of treatment. Outcomes were measured repeatedly in nine trials and reported for two time points in six trials (Donckaster 1964; Nnochiri 1967; Naoemar 1973; Joshi 1975; Soedin 1985; Karabay 1999). Three trials reported results at one time point only because of high drop-out rates during the other follow-up periods (Panggabean 1980; Sitepu 1982; Pamba 1990).
Adverse events were reported in 33 trials and were not ascertained in the remaining four trials (Sitepu 1982; Chunge 1989; Karabay 1999; Mansour-Ghanaei 2003). Sixteen trials had incomplete data: 10 reported specific adverse events but not the number of participants who developed any adverse event (Nnochiri 1967; Kapadia 1968; Batra 1972; Botero 1974; Prasad 1985; Pamba 1990; Asrani 1995; Davila 2002; Padilla 2000; Rossignol 2007); two reported only the number of participants with adverse events severe enough to cause discontinuation of drug treatment (Pehrson 1983; Pehrson 1984); four did not report the actual number of participants who developed any adverse event (Rubidge 1970; Pudjiadi 1973; Soedin 1985; Toppare 1994); and one reported of adverse events only for the experimental group (Mohammed 1998).
Risk of bias in included studies
See Appendix 10 for an overview of the methodological quality assessment.
Only one trial reported using appropriate procedures to minimize or eliminate bias in allocation concealment, generation of the allocation sequence, blinding (of the care providers, participants, and outcome assessors), and inclusion of all randomized participants (Rossignol 2007).
Generation of allocation sequence
Only six trials reported adequate generation of the allocation sequence: four trials used a random-numbers table (Donckaster 1964; Awal 1979; Sitepu 1982; Mohammed 1998); one trial used computer-generated randomization (Rossignol 2007); and one trial used coin toss (Padilla 2000). The other trials did not describe the method used.
Allocation concealment
Two trials used identical coded drugs prepared independently at a site remote from the study site and had adequate allocation concealment (Pudjiadi 1973; Rossignol 2007). Two trials had inadequate allocation concealment as communicated by the primary author (Pehrson 1983; Pehrson 1984). The remaining 33 trials did not report on this.
Blinding
Only seven trials reported blinding of the participants, care providers, and outcome assessors (Nnochiri 1967; Naoemar 1973; Pudjiadi 1973; Prasad 1985; Rossignol 2001; Mansour-Ghanaei 2003; Rossignol 2007). One trial reported blinding of the participants and outcome assessors but not the care providers (Padilla 2000), one trial reported blinding only of the participants and microscopists assessing stool specimens (Chunge 1989), one trial reported blinding only of the care providers (Panggabean 1980), and another reported blinding only of the microscopists assessing stool specimens (Pamba 1990). Seven trials were reported to be 'double-blind' (Donckaster 1964; Botero 1974; Botero 1977; Huggins 1982; Sitepu 1982; Tripathi 1986; Davila 2002), but they did not describe the procedure for blinding, similarity of the appearance of drugs, or use of placebo. Five trials were also unclear regarding blinding (Kapadia 1968; Misra 1974; Misra 1977; Swami 1977; Misra 1978), and the other 14 were open label.
Inclusion of all randomized participants in the analysis
The number of participants followed up was adequate (greater or equal to 90%) in 32 trials for at least one outcome. Four trials reported losses of participants greater than 10% (Panggabean 1980; Sitepu 1982; Pamba 1990; Mohammed 1998), while in one trial, only the number analysed was reported but the actual number initially randomized, those lost to follow up, and those who were not compliant with the treatment protocol were not reported (Chunge 1989).
Effects of interventions
Details of the comparisons and interventions included in the review are shown in Appendix 9.
1. Single alternative drugs versus metronidazole
Sixteen trials compared alternative nitroimidazoles with metronidazole, and one trial compared praziquantel with metronidazole.
1.1. Tinidazole versus metronidazole
For clinical failure 15 to 60 days after the end of treatment, tinidazole reduced clinical failure by 72% compared with metronidazole (RR 0.28, 95% CI 0.15 to 0.51; 477 participants, 8 trials, Analysis 1.2 and Figure 1). A sensitivity analysis evaluating quality in relation to allocation concealment and blinding was not possible. There was no significant change in the overall result when we excluded Misra 1978, which may be a duplicate publication of an earlier trial (Misra 1977) (RR 0.31, 95% CI 0.16 to 0.61, analysis not shown). Excluding four trials funded by pharmaceutical companies (Misra 1974; Joshi 1975; Mathur 1976; Singh 1977) also did not affect the overall result (RR 0.24, 95% CI 0.11 to 0.50, analysis not shown).
 | Figure 1. Alternative drug vs metronidazole: clinical failure 15 to 60 days after end of treatment. |
Results for parasitological failure did not show that tinidazole was more effective in eradicating E. histolytica compared to metronidazole (507 participants, 9 trials, Analysis 1.4 and Figure 2), although there was significant heterogeneity in the trials. A subgroup analysis to investigate possible sources of heterogeneity showed reduced heterogeneity in those with nondysenteric amoebic colitis and unspecified amoebic colitis ( Analysis 6.1) and in those trials that used the WHO criteria ( Analysis 6.4). Age and the presence or absence of other concomitant intestinal infection did not explain heterogeneity ( Analysis 6.2 and Analysis 6.3).
 | Figure 2. Alternative drug vs metronidazole: parasitological failure 15 to 60 days after end of treatment. |
No data on relapse were reported.
Adverse events were incompletely reported. There were no serious adverse events or adverse events that necessitated drug withdrawal in the three trials that reported on this (Joshi 1975; Mathur 1976; Awal 1979). For the other adverse events, they were more common in those given metronidazole compared to those given tinidazole (RR 0.65, 95% CI 0.46 to 0.92; 477 participants, 8 trials, Analysis 1.6). The most common adverse events reported were nausea, vomiting, decreased appetite, altered taste or metallic taste, and abdominal discomfort (see Appendix 11 for other details).
1.2. Other drugs versus metronidazole
Other alternative drugs tested were ornidazole (128 participants, 3 trials), panidazole (44 participants, 1 trial), and satranidazole (40 participants, 1 trial). The number of trials was too small to detect any difference in clinical failure or parasitological failure compared with metronidazole ( Analysis 1.1, Analysis 1.2, Analysis 1.3, and Analysis 1.4).
For relapse, data were reported for two trials, and both compared ornidazole with metronidazole. There were more relapses in those given ornidazole compared with metronidazole (RR 4.74, 95% CI 1.07 to 20.99; 135 participants, 2 trials, Analysis 1.5), but there were insufficient data to draw definite conclusions.
There were no serious adverse events or withdrawals resulting from adverse events in two trials that reported on this (Pehrson 1984; Tripathi 1986). One participant given ornidazole developed a temporary numbness of the hands and tongue with difficulty of speaking that disappeared after stopping treatment (Botero 1974). In another trial (Naoemar 1973), the daily dosage of ornidazole had to be reduced from 1500 mg to 1000 mg in two participants each in the ornidazole group and metronidazole group because of dizziness or nausea. No abnormalities in laboratory tests were seen in trials in which these tests were done (see Appendix 11 for other details).
2. Any antiamoebic drug versus placebo
Any antiamoebic drug was compared with placebo in four trials: versus two different luminal agents, nitazoxanide (167 participants, 2 trials) and quinfamide (96 participants, 1 trial); and versus 10 different drugs belonging to six drug classes (367 participants, 1 trial – Donckaster 1964).
Compared with placebo, the other antiamoebic drugs reduced clinical failure one to 14 days after the end of treatment by 73% (RR 0.27, 95% CI 0.13 to 0.57; 193 participants, 3 trials, Analysis 2.1) and reduced parasitological failure by 67% (RR 0.33, 95% CI 0.23 to 0.50; 630 participants, 4 trials, Analysis 2.2). There was significant heterogeneity among the trials so only subtotals are presented. Subgroup analysis using clinical categories did not explain heterogeneity ( Analysis 7.1), but heterogeneity was reduced in trials that included adults only ( Analysis 7.2 and Analysis 7.3). Excluding the single trial that used stool antigen-based ELISA test for confirming E. histolytica (Rossignol 2007) also reduced heterogeneity in the remaining trials (RR 0.36, 95% CI 0.23 to 0.56; 93 participants, 3 trials, analysis not shown). Sensitivity analysis was not possible using concealment and blinding because only one trial was concealed (Rossignol 2007) and only two trials were blinded (Donckaster 1964; Rossignol 2007).
No data on relapse were reported.
There were no serious adverse events or withdrawals due to adverse events in the two trials that reported on this (Rossignol 2001; Rossignol 2007). For the other adverse events, there was no evidence of a difference in adverse events in those given antiamoebic drugs compared with placebo (530 participants, 3 trials, Analysis 2.3), although the results could be biased because of a great imbalance in the number of those given placebo compared to the active drugs. The most common adverse events were mild gastrointestinal symptoms such as nausea, vomiting, abdominal pain, and headache. One individual given diiodohydroxyquinoline presented with severe intestinal colic (see Appendix 12 for details).
3. Combination regimen versus monotherapy
Three trials compared various combination regimens with metronidazole alone while four trials compared other combination regimens with alternative single drugs.
3.1. Combination regimen versus metronidazole alone
Three trials that compared combination therapy with metronidazole alone showed that combination therapy reduced clinical failure one to 14 days after the end of treatment by 67% (RR 0.33, 95% CI 0.11 to 0.98; 1025 participants, 3 trials, Analysis 8.1). The heterogeneity seen in the trials could be due to the various combination regimens used: a combination of dehydroemetine, tetracycline, and diloxanide furoate (Rubidge 1970); a fixed-drug combination suspension of metronidazole and furazolidone (Prasad 1985); and a fixed-drug combination tablet of metronidazole and diiodohydroxyquinoline (Asrani 1995). Excluding one trial that was funded by a pharmaceutical company (Rubidge 1970) reduced heterogeneity, and the analysis showed that combination therapy significantly reduced clinical failure compared with metronidazole alone (RR 0.17, 95% CI 0.12 to 0.25; 986 participants, 2 trials, analysis not shown).
For parasitological failure, results showed a 64% reduction in parasitological failures one to 14 days after the end of treatment in those given the combination compared to metronidazole alone (RR 0.36, 95% CI 0.15 to 0.86; 720 participants, 3 trials, Analysis 8.2). There was no significant heterogeneity among the trials (see Figure 3). Excluding the trial funded by a pharmaceutical company (Rubidge 1970) did not significantly change the results (RR 0.25, 95% CI 0.13 to 0.46; 681 participants, 2 trials, analysis not shown).
 | Figure 3. Combination regimen vs monotherapy: parasitological failure 1 to 14 days after end of treatment, subgrouped by intervention. |
For both clinical and parasitological failure, sensitivity analysis was not possible because none of the trials were adequately concealed and only Prasad 1985 was double-blind. The overall results for both outcomes were driven by one trial (Asrani 1995) that analysed a large number of participants (896 participants analysed for clinical failure; 591 participants analysed for parasitological failure) compared to the other two trials. This trial was an open label trial with unclear allocation concealment and method of randomization, hence the high possibility of bias. A sensitivity analysis performed by excluding this trial significantly changed the overall results for both outcomes such that no difference between combination therapy and metronidazole alone was seen (RR 0.58, 95% CI 0.20 to 1.73; 129 participants, 2 trials, analysis not shown). The change in the overall result weakens the strength of the evidence for the advantage of combination therapy over metronidazole alone.
No data on relapse were reported.
Only one trial reported details for adverse events (Asrani 1995). One participant given a fixed-drug combination tablet of metronidazole and diiodohydroxyquinoline developed an unspecified allergic reaction on the first day necessitating withdrawal from the trial (see Appendix 13 for details).
3.2. Combination regimen versus other single drug regimens
Four trials made this comparison. Two trials compared combination regimen with other nitroimidazoles: a combination of tetracycline and clioquinol with secnidazole alone (80 participants, 1 trial); and a combination of tinidazole and diloxanide furoate with tinidazole alone (41 participants, 1 trial). The third trial compared three different combinations (nimorazole and aminosidine, nimorazole and etophamide, and etophamide and aminosidine) with the same drugs given as monotherapy (400 participants, 1 trial). The fourth trial compared quinfamide and mebendazole with nitazoxanide (80 participants, 1 trial).
The trials could not be pooled because of different drug comparisons, but the data for clinical failure ( Analysis 3.1) and parasitological failure ( Analysis 3.2) are presented. The trials did not show any difference in clinical or parasitological failure rates between combination regimens and single drug regimens, except for two comparisons. One trial showed that secnidazole alone resulted in greater resolution of clinical symptoms and greater eradication of E. histolytica compared to the combination of tetracycline and clioquinol (80 participants, Soedin 1985, Analysis 3.1. and Analysis 3.2). Another trial showed that the combination of tinidazole and diloxanide furoate resulted in greater eradication of E. histolytica compared with tinidazole alone (41 participants, Pehrson 1984).
No data on relapse were reported.
None of the trials reported on serious adverse events. Recruitment of participants in the combination etophamide-aminosidine group was discontinued because of the high incidence of severe diarrhoea. No further details for adverse events were reported in the other trials.
4. Single-dose regimen versus longer regimen
Five trials compared a single-dose regimen with a multiple-dose regimen or longer durations of therapy.
4.1. Single-dose secnidazole versus longer duration of other drugs
Single-dose secnidazole was compared with tinidazole for two days (303 participants, Salles 1999), metronidazole for 10 days (44 participants, Karabay 1999), and a combination of tetracycline and clioquinol given for five days (80 participants, Soedin 1985).
Soedin 1985 showed that single-dose secnidazole resulted in greater resolution of clinical symptoms compared with five days of tetracycline and clioquinol (80 participants, 1 trial, Analysis 4.1). Salles 1999 did not show any difference between single-dose secnidazole and tinidazole given for two days (275 participants, 1 trial). Results for clinical failure could not be pooled because of difference in time of evaluation of clinical outcomes in the two trials (see Analysis 4.1).
For parasitological failure, single-dose secnidazole resulted in lower parasitological failure compared with longer durations of therapy when the results from Soedin 1985 and Karabay 1999 were combined (RR 0.14, 95% CI 0.06 to 0.35; 124 participants, 2 trials, Analysis 4.2 and Analysis 9.1). Although there was no heterogeneity no conclusions can be made because both trials were small and did not use blinding.
No data on relapse were reported.
Only Salles 1999 reported on adverse events. There were no serious adverse events or withdrawals from adverse events. The most common adverse events in those given single-dose secnidazole and longer durations of tinidazole for two days were bitter taste, nausea, vomiting, and abdominal pain, with no difference in frequency in the two groups (see Appendix 13 for details).
4.2. Single-dose quinfamide versus multiple doses of quinfamide or longer duration of another drug
Single-dose quinfamide was compared with two or three doses of quinfamide (96 participants, Huggins 1982) and with nitazoxanide for three days (105 participants, Davila 2002) (see Analysis 4.1 and Analysis 4.2).
Huggins 1982 showed no difference in clinical failure in those given one dose compared to two or three doses of quinfamide (27 participants, Analysis 4.1).
For parasitological failure, there was a trend favouring longer duration compared to single-dose quinfamide in eradicating E. histolytica (RR 2.13, 95% CI 1.02 to 4.46; 97 participants, 2 trials, Analysis 9.1). Both trials were unclear regarding generation of the allocation sequence, concealment, and blinding. Results were not heterogenous, but the number of trials and participants were also too small to draw any definite conclusions.
No data on relapse were reported.
Only Huggins 1982 reported on adverse events, which were based on only two symptoms, nausea and headache. There was no evidence of a difference in adverse events in those given single-dose quinfamide compared with two or three doses of quinfamide (see Appendix 13 for details).
5. Other antiamoebic drug comparisons
Nine trials studied different drug comparisons (see Appendix 9 for details) and data were not pooled. None of the trials were blinded or adequately concealed, and drop-out rates were high in two trials, with only 62.5% of those initially randomized analysed in one trial (Panggabean 1980) and 82% analysed in the other trial (Sitepu 1982).
Five trials assessed clinical failure one to 14 days after the end of treatment (Kapadia 1968; Batra 1972; Panggabean 1980; Sitepu 1982; Toppare 1994). Only chlorhydroxyquinoline was shown to be more effective than diiodohydroxyquinoline in reducing clinical failure (RR 0.24, 95% 0.11 to 0.53; 100 participants, 1 trial, Analysis 5.1). There was no evidence of a difference in the other antiamoebic drugs compared: ornidazole versus tinidazole (66 participants, 2 trials); and ornidazole versus secnidazole (102 participants, 1 trial). There were no clinical failures in any of the four dosage regimens of MK-910 studied (40 participants, 1 trial).
Parasitological failure was assessed one to 14 days after the end of treatment in all nine trials. One trial also assessed parasitological failure 15 to 60 days after treatment (Nnochiri 1967). Only chlorhydroxyquinoline was more effective than diiodohydroxyquinoline in reducing parasitological failure (100 participants, 1 trial, Analysis 5.2). There was no evidence of any difference between groups in the other trials: quinfamide versus secnidazole (239 participants, 1 trial); ornidazole versus secnidazole (102 participants, 1 trial); and quinfamide versus nitazoxanide (24 participants, 1 trial). There were no parasitological failures in any of the four dosage regimens of MK- 910 studied (40 participants, 1 trial). One trial compared metronidazole and iodoquinol with the same combination plus Saccharomyces boulardii, and no difference was seen (54 participants, 1 trial). Another small trial that compared a fixed-drug combination of diloxanide furoate, tetracycline, and chloroquine with the fixed-drug combination without chloroquine showed no evidence of a difference between groups at the end of treatment (59 participants, Nnochiri 1967, Analysis 5.2), but there was a significant advantage for the combination containing chloroquine on follow up seven weeks after the end of treatment (RR 0.43, 95% CI 0.20 to 0.93; 59 participants, analysis not shown).
No data on relapse were reported.
None of the trials reported on serious adverse events. One trial reported that in two participants the higher dosage regimen of MK- 910 resulted in nausea, vomiting, and abdominal pain severe enough to require withdrawal from treatment (Batra 1972). One participant given ornidazole developed vomiting (Panggabean 1980). Three trials reported only on the specific adverse events but not the number of participants with adverse events (Nnochiri 1967; Kapadia 1968; Batra 1972; see Appendix 13 for details).
Funnel plot
We constructed a funnel plot for the one outcome measure with a sufficient number of trials and examined it visually for possible bias or heterogeneity: any antiamoebic drug versus metronidazole and measuring parasitological failure 15 to 60 days after the end of treatment (13 trials, Figure 4). This included nine trials that compared tinidazole with metronidazole. The funnel plot is asymmetrical due to the absence of smaller trials at the base and fewer trials to the right of the overall estimate. Although this may indicate the presence of publication bias, asymmetry in the funnel plot may also indicate inadequate methodological quality of the trials. Heterogeneity in the trials as a result of differences in study population, interventions, outcome measurements, trial design, and quality also have to be considered.
 | Figure 4. Funnel plot. Alternative drug vs metronidazole: parasitological failure 15 to 60 days after end of treatment. |
Discussion
Effectiveness of antiamoebic drugs
This systematic review examined the effectiveness of various antiamoebic drugs for treating amebic colitis by measuring improvements in both clinical and parasitological outcomes. The rapid relief of diarrhoea and other gastrointestinal symptoms associated with intestinal amoebiasis is an important concern of the individual with disease, while eradication of the parasite is important to prevent further invasion with damage to the intestinal mucosa and possible extraintestinal spread. The purpose of the review was to determine the best drug and treatment regimen required for effective treatment of amoebic colitis.
Tinidazole versus metronidazole
Metronidazole, the current recommended first-line drug for treating amoebic colitis, is a tissue amoebicide predominantly but with some activity against cysts in the bowel lumen. Among the other nitroimidazole drugs, only tinidazole had a sufficient number of trials to be able to draw any meaningful conclusions in the pooled results. The pooled results show that tinidazole demonstrated greater resolution of clinical symptoms, but there was inconclusive evidence of its advantage over metronidazole in eradication of E. histolytica in the stools. The comparable parasitological response could be attributed to similar activities of drugs belonging to the nitroimidazole drug class. In one trial, the poor eradication of parasites with tinidazole was attributed to its greater absorption and higher blood and tissue concentrations, possibly resulting in inadequate levels in the bowel lumen (Pehrson 1984). Lower concentrations of tinidazole could also be explained by the suboptimal dosing interval used in some trials (Misra 1974; Joshi 1975; Mathur 1976; Pehrson 1984). Tinidazole is better given as a single dose than divided doses because of its longer half life of approximately 12 to 14 hours resulting in longer concentrations in the body (Monro 1974; Looke 1987). In contrast, metronidazole has a shorter half life of about six to eight hours and is better if given in divided doses. Also, longer courses may lead to its re-excretion through the bile resulting in higher concentrations within the bowel lumen (Knight 1980; Tracy 2001). This is supported by the summary report of nine trials conducted in India where tinidazole given once daily was more effective compared with divided doses, and was also more effective than metronidazole given once daily (Bakshi 1978).
The incidence of mild to moderate gastrointestinal complaints was noted to be greater in those given metronidazole compared with tinidazole, particularly when it was given as a single dose. Again, this is similar to the summary report of Bakshi 1978, where the number of gastrointestinal adverse effects were greater in those given single-dose metronidazole compared with divided doses and compared with single daily administration of tinidazole. In both groups, the most frequently reported adverse effects were mild gastrointestinal side effects such as nausea, decreased appetite, vomiting, and metallic or bitter taste.
The calculation of the risk difference for clinical failure in those given tinidazole and those given metronidazole would give an absolute change in risk of 0.16 that is attributable to tinidazole. The inverse of the risk difference for clinical failure gave a number needed to treat to benefit (NNTB) of 6.25. Thus, we will need to treat seven people with tinidazole to reduce clinical failure in one more individual. However, this could not be applied to the parasitological failure because there appears to be no significant difference in eradication of E. histolytica in those given tinidazole compared with metronidazole.
Relapse can only be evaluated if stools are examined several times after the end of treatment. Data for relapse were only reported in two trials comparing ornidazole with metronidazole, with an increased risk in the ornidazole group compared with metronidazole; however, the available data were too few to draw any conclusions regarding the result.
The funnel plot constructed from the trials comparing any antiamoebic drug versus metronidazole and measuring parasitological failure demonstrated asymmetry. Selection bias may arise when small studies reporting greater treatment benefit for the experimental drugs are more likely to be published (publication bias) and found (language and citation). Asymmetry could also result from poor methodological quality of the trials since greater treatment effect of the experimental drugs was seen in the smaller studies. Heterogeneity in treatment regimens used could also explain the asymmetry in the funnel plot. There was a trend favouring the experimental drugs in three trials (Singh 1977; Swami 1977; Awal 1979) where metronidazole was given at a dose of 2 g once daily for two or three days; a much lower dose compared to the recommended dose of 500 to 750 mg thrice daily in adults or 30 to 50 mg/kg/day in children for at least five days. This could account for the higher parasitological failure in those on metronidazole. When metronidazole was given at a dose of 2.4 g for five days, it was significantly more effective compared to tinidazole (Pehrson 1984). The presence of asymmetry in the funnel plot could lessen confidence in the results.
Antiamoebic drugs versus placebo
Antiamoebic drugs were shown to be more effective in reducing clinical symptoms of amoebic colitis and in eradicating E. histolytica in the stools compared with placebo. The significant heterogeneity in the pooled results could be attributed to factors such as differences in participant characteristics or the varied antiamoebic drugs used. Twelve different antiamoebic drugs were used and all, except for dehydroemetine, are more active for parasites found in the bowel lumen with no activity against invasive tissue forms. Since none of the participants were diagnosed as having amoebic dysentery, the significant response in those given these luminal drugs may result from the activity of these drugs against cysts in those with nondysenteric amoebic colitis or unspecified intestinal amoebiasis.
The disappearance of parasites in 50 out of 133 (38%) individuals while on placebo may result from spontaneous eradication of E. histolytica or infection with nonpathogenic amoeba. Several studies have shown that up to 90% of individuals with untreated E. histolytica infection spontaneously clear their infection within one year (Gathiram 1985; Haque 2001; Blessman 2003b; Haque 2002). It may have been possible that E. histolytica was misidentified in the trials utilizing stool microscopy as the only means for diagnosis.
Use of placebo as a comparison drug for individuals with symptomatic amoebic colitis may draw objections. Even in those with asymptomatic infection, approximately 3% to 10% of untreated individuals may still develop symptoms of invasive amoebiasis (Gathiram 1985; Haque 2001; Blessman 2003b; Haque 2002). For this reason, it is generally recommended that antiamoebic treatment be given to all individuals with definite E. histolytica, even those who are asymptomatic (WHO 1997; Medical Letter 2004; AAP 2006). Therefore, unless the diagnosis of E. histolytica infection is uncertain in an asymptomatic individual, the use of placebo as a comparison drug, particularly in cases with symptoms of invasive disease, may not be justifiable nor ethical.
Combination regimen versus metronidazole alone
For all forms of invasive disease, including amoebic colitis, the standard recommendation is to give a tissue amoebicide followed by a luminal amoebicide to eliminate surviving cysts in the bowel lumen (WHO 1995; WHO 1997; Medical Letter 2004; AAP 2006). The risk for parasitological failure was reduced by about one-third in those given combination therapy compared with metronidazole alone. The greater risk for failure with metronidazole when administered alone may result from its inconsistent effect in eliminating cysts in the bowel lumen and its failure to reach adequate therapeutic concentrations in the large intestines (Reed 1992; Petri 1999; Haque 2003; Stanley 2003). The advantage of combination therapy is attributed to the distinct activities of the different drugs against the cysts and trophozoites found at the different sites (WHO 1995; Tracy 2001; Medical Letter 2004).
Adverse events were incompletely reported, and it is not known whether combination therapy would lead to increased adverse events. Uncertainty also remains over which luminal agent is preferred in combination with metronidazole or another nitroimidazole to achieve eradication of cysts in the stools and to prevent relapse, since none of the trials were of sufficient size to determine this.
Comparison with other reviews
An earlier systematic review on amoebic dysentery concluded that metronidazole was "unlikely to be beneficial" since some ineffectiveness or associated harm was demonstrated in some trials and that ornidazole, secnidazole, and tinidazole were "likely to be beneficial", since effectiveness for these drugs, with no increased harms, was demonstrated in other trials (Dans 2006). In this current review, we concluded that although tinidazole may be more effective than metronidazole in reducing clinical failure, and was probably associated with fewer adverse effects, it did not show any significant advantage over metronidazole in reducing parasitological failure. For ornidazole and secnidazole, there was insufficient data to be able to draw any definite conclusions. The difference in conclusions could be due to important differences in methodology. The systematic review on amoebic dysentery used the Clinical Evidence search strategy (Dans 2006) and included 12 randomized controlled trials, defined therapeutic failure as persistence of symptoms or persistence of parasites or both, analysed outcomes reported at different time points together, and did not pool data to generate an overall summary measure.
Applicability of results
This review was limited only to symptomatic individuals with uncomplicated amoebic colitis. We did not study the effect of antiamoebic drugs on those with severe amoebic colitis, complicated disease, or extraintestinal amoebiasis. The potential effect of malnutrition, immune suppression, or AIDS on treatment is not known. Although asymptomatic infection with E. histolytica is more common than symptomatic disease, treatment of these individuals remains controversial because most will clear their infection within one year and only about 3% to 10% will manifest invasive disease (Gathiram 1987; Haque 2001; Blessman 2003b; Haque 2002).
The limited availability of many antiamoebic drugs must be addressed in the light of reports that newer nitroimidazole drugs may be as effective as, and better tolerated than, metronidazole and that there may be fewer clinical and parasitological failures when luminal agents are given in conjunction with tissue amoebicides. Metronidazole is widely used and may be the only available antiamoebic drug in many countries. Tinidazole and the other nitroimidazole drugs, such as ornidazole and secnidazole, and luminal agents such as diloxanide furoate, iodoquinol, and paromomycin, are not widely available and may only be purchased from certain pharmaceutical companies or requested from government agencies. Although tinidazole was shown to be probably more effective and better tolerated than metronidazole in this review, the limitations of the currently available evidence, and the limited availability of tinidazole in many regions, would make a widespread recommendation for its use impractical. Similarly, the evidence to recommend combination therapy is inadequate, and the limited availability of luminal agents in the market is a major deterrent to compliance with the general recommendation for combination therapy.
Limitations of the review
Limitations in study quality, the imprecise or sparse data in some outcomes, important inconsistencies across the trials, and a high probability of reporting or publication bias decrease the quality of evidence. Therefore the conclusions of the review should be interpreted with caution. Inaccurate diagnosis of E. histolytica infection by stool microscopy, absence of standardized classification of the various categories of amoebic colitis (particularly nondysenteric amoebic colitis), and variable timing and definition of outcome measurements would all lead to inaccuracy in assessing treatment effects. In areas highly endemic for amoebiasis, true treatment failures or relapse would be difficult to differentiate from reinfection without the benefit of finger typing or genotyping. Incomplete reporting of adverse events may lead to an inaccurate assessment of adverse events.
Authors' conclusions
Implications for practice
|
Implications for research
|
Acknowledgements
This document is an output from a project under the Effective Health Care Research Programme Consortium (RPC) – Philippines, which is funded by the UK Department for International Development (DFID) for the benefit of developing countries. The views expressed are not necessarily those of DFID. The editorial base for the Cochrane Infectious Diseases Group is funded by DFID.
Data and analyses
- Top of page
- Background
- Objectives
- Methods
- Results
- Discussion
- Authors' conclusions
- Acknowledgements
- Data and analyses
- Appendices
- History
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||
Appendices
Appendix 1. Search methods: detailed search strategies
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 2. Search methods: conference proceedings searched
| ||||||||||||||||||||||||||||||||||
Appendix 3. Search methods: organizations or institutions contacted for trials on amoebic colitis
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 4. Search methods: pharmaceutical companies
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 5. Region and country of trial
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 6. Trial setting
| ||||||||||||||||||||||||||
Appendix 7. Participant age in included trials
| |||||||||||||||||||||||||||
Appendix 8. Methods used to diagnose amoebic colitis
| ||||||||||||||||||||||||||||||||||||||||
Appendix 9. Interventions and comparisons included in the trials
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 10. Risk of bias assessmenta
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 11. Adverse events: alternative drug vs metronidazole
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 12. Adverse events: any antiamoebic vs placebo
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Appendix 13. Adverse events: other comparisons
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
History
Protocol first published: Issue 3, 2006
Review first published: Issue 2, 2009
Contributions of authors
Dr MLM Gonzales is primarily responsible for conceiving and designing the review, and co-ordinating its development. Dr LF Dans advised Dr MLM Gonzales about the design and co-ordination of the review and, together with Dr MLM Gonzales, assessed the results of the literature search to determine trial eligibility and the risk of bias in each included trial. Dr EG Martinez extracted data from the included trials and resolved differences in the assessment of papers between the other two authors. Dr EG Martinez and Dr L Dans conducted the systematic review on drug therapy for amoebic dysentery (Dans 2006) that set the groundwork for this Cochrane Review.
Declarations of interest
Potential conflict of interest: Dr Elizabeth G Martinez is currently employed as Medical Manager in United Laboratories, Inc., a major pharmaceutical company in the Philippines. It manufactures and distributes an intravenous preparation of metronidazole. None of the trials included in the review were conducted or sponsored by United Laboratories, Inc., and Dr Elizabeth G Martinez was not involved in any trial involving antiamoebic drugs.
Sources of support
Internal sources
- No sources of support supplied
External sources
- Department for International Development (DFID), UK.
Differences between protocol and review
Objectives
Four specific objectives were added to give a more focused direction to the review.
Search methods
Aside from searching the databases mentioned in the protocol as described in Appendix 2, we conducted additional electronic searches of PubMed using the format for highly sensitive search strategies for identifying reports of randomized controlled trials (Higgins 2005), with the latest search conducted on 11 February 2008. We also conducted additional searches of the databases of ongoing or unpublished trials by pharmaceutical companies and other conference proceedings not originally mentioned in the protocol.
Data collection and analyses
Data synthesis
For trials with more than two intervention groups (eg two or more experimental interventions or different doses or preparations of the same drug), multiple treatment arms were combined as appropriate and compared collectively with the standard or control group in the review. This would avoid a unit of analysis error by not counting the placebo or control participants more than once in the same meta-analysis.
Stratification of results
In the protocol, separate analyses were planned for outcomes reported at the end of treatment until seven days after treatment and for outcomes reported eight to 21 days after the end of treatment. However, since many trials reported outcomes 28 days or one month after treatment, we decided to stratify the outcomes from the end of treatment to 14 days and from 15 to 60 days after the end of treatment instead.
Subgroup analyses
Clinical categories (amoebic dysentery or nondysenteric amoebic colitis) and participant age (adults and children) were determined to be important subgroups even before data collection, although we failed to specify this in the protocol. Severity of disease and response to treatment differ for those with amoebic dysentery compared to those with nondysenteric amoebic colitis. Participant age is also an important confounder that may influence treatment effect and would be valid to explore as a potential source of heterogeneity. In those comparisons with enough trials, age was included as a subgroup, with the participants categorized as adults (age 15 years or older) or children (age less than 15 years). Other sources of heterogeneity that were not mentioned in the protocol but included in the post hoc subgroup analysis were type of intestinal infection (E. histolytica infection alone or mixed intestinal infection), criteria for determining outcome (based on WHO 1969 criteria or another criteria), and regimens used.
Sensitivity analyses
We intended to explore validity by using the tests for diagnosing E. histolytica as a surrogate for trial quality. However, we could not carry out sensitivity analysis based on the diagnostic tests because only one trial used stool antigen-based ELISA test for confirming E. histolytica. In the review, additional sensitivity analyses were done to determine the possible effect of pharmaceutical-sponsored trials since this could have an effect on the quality of the trials.
Index terms
Medical Subject Headings (MeSH)
Amebicides [adverse effects; *therapeutic use]; Drug Therapy, Combination; Dysentery, Amebic [*drug therapy; parasitology]; Entamoeba histolytica; Metronidazole [adverse effects; therapeutic use]; Randomized Controlled Trials as Topic; Tinidazole [adverse effects; therapeutic use]
MeSH check words
Animals; Humans
* Indicates the major publication for the study