Description of the condition
Enteric fever is a common term to encompass two similar clinical illnesses, caused by different serotypes of the bacterium Salmonella enterica. Typhoid fever (due to Salmonella typhi) is generally more common, and more severe, but recent reports suggest that the relative frequency of paratyphoid fever (due to Salmonella paratyphi) may be increasing (Chandel 2000; Ahmad 2002; Butt 2005; Ochiai 2005; Jesudason 2005; Woods 2006; Maskey 2008). In the year 2000, there were an estimated 21.6 million cases of typhoid fever, with 210,000 deaths, and 5.4 million cases of paratyphoid fever (Ochiai 2008; Crump 2004).
The symptoms of enteric fever are generally non-specific and can vary among different populations (Parry 2002). Common symptoms include fever, headache, and gastrointestinal complaints; such as diarrhoea, constipation, abdominal pain, nausea and vomiting, or loss of appetite (Lee 2000; Richens 2000). Severe disease occurs in 10 to15% of people, and complications such as; intestinal perforation, intestinal bleeding, shock, pancreatitis (inflammation of the pancreas), pneumonia, myocarditis (inflammation of the heart muscles), meningitis (inflammation of the covering of the brain), or psychosis (altered mental state) can occur, typically after the illness has lasted more than two weeks (Parry 2002).
The bacteria may be shed in the faeces during the acute illness, during convalescence, and occasionally for prolonged periods when the person is labelled a 'chronic carrier' (defined as excretion of the bacterium in the stool or urine for more than one year (Bhan 2005)). Infection occurs when food or water contaminated with faeces harbouring the bacteria are ingested. The organisms then penetrate the intestinal lining, multiply in lymphoid tissues, and are released into the blood stream from where they spread throughout the body to various organs; most commonly the liver, spleen, bone marrow, and gall bladder (Lesser 2001).
The enteric fevers remain a major health problem in low- and middle-income countries where water and sanitation services may be inadequate. They are endemic throughout Asia (with the highest incidence in South and Southeast Asia), the Middle East, Africa, and South and Central America (Ivanoff 1995; Crump 2004). In high-income countries, most cases occur in travellers returning from these endemic areas (McNabb 2008). The highest incidence has been reported in children between five and 10 years of age (Lin 2000; Siddiqui 2006; Sur 2006), and in those under five years of age (Sinha 1999; Saha 2001; Saha 2003; Brooks 2005).
Diagnosis and treatment
The diagnosis of enteric fever can be difficult due to the non specific nature of the symptoms. A definitive diagnosis is possible when the organisms are isolated from blood, bone marrow or other body fluids. Blood cultures are typically positive in 60 to 80% of cases, while bone marrow cultures are more sensitive with 80 to 95% positive, even after prior antibiotic therapy (Parry 2002). Serological tests, such as the Widal reaction, have been widely used but these are non-specific, giving false positive results, and can be difficult to interpret. More recently, there has been interest in the use of DNA probes and polymerase chain reaction (PCR) testing , but these are not widely available in enteric fever endemic areas (Parry 2002).
Untreated the disease last 3 to 4 weeks with fever, septicaemia, and a 10-30% mortality. Treatment is with antibiotics and most patients are managed as outpatients.
Resistance of S. typhi and S. paratyphi to commonly used antibiotics has become problematic. Resistance to the highly effective chloramphenicol in the 1970's was associated with simultaneous resistance to sulfonamides, tetracycline, and streptomycin; this led to the use of alternative agents such as co-trimoxazole and amoxicillin (Parry 2002). Subsequently, multi-drug resistant (MDR) strains (resistant to chloramphenicol, ampicillin, co-trimoxazole and streptomycin) emerged and are now prevalent in many parts of the world.
In the Indian subcontinent and China, the frequency of MDR strains ranges from 50% to 80% of all S. typhi isolates and has reached 100% during outbreaks (Lee 2000). In sub-Saharan Africa, MDR S. typhi has been found in 61% and 82.4% of isolates in Nigeria and Kenya, respectively (Akinyemi 2005;Kariuki 2004). Surveillance studies can show considerable geographic differences in the proportion of MDR isolates within the same region; MDR S. typhi is far more common in India, Pakistan and Vietnam than in areas of China and Indonesia (Ochiai 2008). Longitudinal studies have also shown that the proportion of MDR strains can decrease over time following changes in antibiotic use (Lakshmi 2006; Maskey 2008). Indeed several areas have reported a re-emergence of strains susceptible to first-line antibiotics such as chloramphenicol (Takkar 1995; Sood 1999; Wasfy 2002; Rodrigues 2003; Butt 2005; Walia 2005; Mohanty 2006; Gupta 2009).
Infection with resistant strains can lead to higher treatment failure rates, an increased risk of complications, and an increased potential for transmission due to prolonged faecal carriage (Coovadia 1992; Bhutta 1996; Mermin 1999; Rupali 2004; Walia 2005; Crump 2008).
The isolates that respond less well clinically to fluoroquinolones are usually nalidixic acid resistant (NaR) by disc susceptibility testing and have high minimum inhibitory concentrations (MICs) although their breakpoints remain within the range set by the Clinical and Laboratory Standard Institute (CLSI). Using current CLSI disk breakpoints (CLSI 2007) means that fluoroquinolones may continue to be used inappropriately thereby increasing the risks of treatment failure. A key consideration now is the suggested need to redefine breakpoints for isolates with reduced susceptibility to fluoroquinolones in order to identify these strains, offer appropriate therapy and stem the emergence of more resistant organisms (Crump 2003, Parry 2010, BSAC 2011).
Description of the intervention
The fluoroquinolones are a large family of anti-infective drugs, synthesized around a quinolone core, that possess a broad spectrum of antibacterial activity (Congeni 2002).
Nalidixic acid, the prototype quinolone, was first introduced into clinical use in 1962. Four generations of fluoroquinolones have subsequently been developed, classified according to their spectrum of antibacterial activity, and used to treat a range of urinary tract, respiratory, gastrointestinal, and sexually transmitted infections (Oliphant 2002):
- second generation; eg ciprofloxacin, ofloxacin, pefloxacin, norfloxacin; broad gram-negative cover but limited activity against gram-positive bacteria;
- third generation; eg levofloxacin, sparfloxacin, gatifloxacin, moxifloxacin; improved activity against gram-positive bacteria;
- fourth generation; eg trovafloxacin, gemifloxacin; improved activity against anaerobic bacteria.
Subsequently, several of these products have been withdrawn from clinical use (Committee 2006), and norfloxacin is not generally recommended for the treatment of enteric fever due to its poor bioavailability (Miller 2000; Hooper 2000).
Fluoroquinolones generally have few adverse effects. The most common are mild and self-limiting symptoms affecting either the gastrointestinal system (nausea, vomiting or diarrhoea), or the central nervous systems (headaches and dizziness) (Bertino 2000; Oliphant 2002). Rare and serious adverse effects have been linked to specific fluoroquinolone compounds and several have subsequently been withdrawn from clinical use: prolongation of the corrected QT (QTc) interval with grepafloxacin, liver toxicity with trovafloxacin, and anaphylaxis, haemolytic anaemia and renal failure with temafloxacin.(Bertino 2000; Fish 2001)
How the intervention might work
In the past, enteric fevers responded extremely well to the fluoroquinolones, but quinolone resistant strains of S. typhi, especially in Asia, have become a major public health problem (Chuang 2009; Parry 2010; Smith 2010; Parry 2010).The susceptibility of S. typhi to the fluoroquinolones can be divided into three categories:
- fully susceptible; susceptible to both nalidixic acid and ciprofloxacin;
However, not all strains with reduced susceptibility to fluoroquinolones are NaR suggesting the likelihood of a new mechanism of resistance unrelated to the principal mechanisms of resistance already known (Threlfall 2003; Cooke 2006). There are emerging reports of isolates with absolute fluoroquinolone resistance (Harish 2004; Adachi 2005; Renuka 2005; Ahmed 2006; Mohanty 2006; Walia 2006; Joshi 2007). To date, fluoroquinolone resistance has been reported in several countries including India (Renuka 2005,Gaind 2006, Kownhar 2007), Vietnam (Ahmed 2006), Kuwait (Dimitrov 2009), South Africa (Keddy 2010), UK (Cooke 2007) and the USA (Medalla 2011). Most of those reported in the UK and the USA have been imported from India, Vietnam and Bangladesh.
There is current interest in gatifloxacin, which has been found to be active against NaR strains. The alteration in its structure is such that it may hypothetically make the drug less susceptible to the mutations that caused resistance to the older fluoroquinolones (Fukuda 2001). Studies of gatifloxacin suggest that there may be fewer cardiac adverse effects than seen with older generation fluoroquinolones, but with a higher incidence of dysglycaemia (high or low blood sugar) (Frothingham 2005; Park-Wyllie 2006), although some authorities state this may be confined to the elderly, and those with non-insulin dependent diabetes (Ambrose 2003).
Why it is important to do this review
This review aims to summarise trials comparing fluoroquinolones and other antibiotics in treating enteric fever. Interpreting trial data needs to take into account other factors, in particular the year and location of the study, as antibiotic resistance (and therefore efficacy), is dynamic and changes with time.
In the earlier version of this review, different generations of fluoroquinolones were combined in the analysis with sub groups according to the prevalence of NaR strains (Thaver 2008). It was clear that there were important differences between the fluoroquinolones, and this update therefore seeks to group studies by each fluoroquinolone individually. As norfloxacin has poor bioavailability and is no longer a credible treatment option, studies evaluating this drug were excluded.
To evaluate the fluoroquinolone antibiotics in the treatment of enteric fever in children and adults.
Criteria for considering studies for this review
Types of studies
Randomized controlled trials.
Types of participants
People diagnosed with typhoid or paratyphoid fever based on microbiological confirmation from blood, stool or bone marrow.
Types of interventions
Different fluoroquinolone antibiotics, excluding norfloxacin or other fluoroquinolones not currently in use
Any non-fluoroquinolone antibiotic used to treat enteric fever; chloramphenicol, ampicillin, amoxicillin, cotrimoxazole, azithromycin or cephalosporins.
An alternative fluoroquinolone, or a different treatment duration of the same fluoroquinolone.
Types of outcome measures
- Clinical failure; defined as development of complications, requiring a change of antibiotic therapy, or remaining symptomatic beyond a time period specified by trial authors.
- Microbiological failure; defined as a positive culture from blood, bone marrow, or any sterile anatomic site, beyond a time period specified by trial authors.
- Relapse; defined as the recurrence of symptoms with a positive culture from blood or bone marrow or any sterile anatomic site, beyond a time period defined by trial authors.
- Fever clearance time; defined as the time in hours/days taken to clear fever from the start of the intervention or control drug with the definition of fever clearance as specified by trial authors.
- Length of hospital stay; defined as the time in days from entry into trial until discharge.
- Convalescent faecal carriage; defined as a positive faecal culture detected at any time after the end of treatment up to one year of follow up.
Adverse events (as defined by trial authors)
- Serious adverse events; defined as adverse events leading to death, inpatient hospitalization, prolonged hospitalization, or life threatening, resulting in persistent or significant disability or incapacity, such as joint disease, tendonitis and tendon rupture, prolongation of QTc interval, seizures, nephrotoxicity, haematological reactions, or severe dermatologic reactions.
- Other adverse events, such as nausea, diarrhoea, headache, dizziness, mild photosensitivity, hepatic enzyme elevations, and hypersensitivity reactions.
Search methods for identification of studies
Emmanuel Effa worked with Vittoria Lutje (Information Retrieval Specialist, Cochrane Infectious Diseases Group) to identify all relevant trials regardless of language or publication status.
We searched the following databases using the search terms and strategy described in Appendix 1: Cochrane Infectious Diseases Group Specialized Register (February 2011); Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library (2011, Issue 2); MEDLINE (1966 to February 2011); EMBASE (1974 to Febrary 2011); and LILACS (1982 to February 2011). We also searched the metaRegister of Controlled Trials (mRCT) in February 2011 using the search term "(typhoid fever) NOT vaccine".
Searching other resources
We contacted Christiane Dolecek (in October, 2010) who provided information on unpublished and ongoing trials.
Reference lists and review authors' personal collections
We also checked the reference lists of all retrieved trials and searched the review authors' personal literature collections for relevant trials.
Data collection and analysis
Selection of studies
Two review authors, Emmanuel E Effa (EEE) and Zohra S Lassi (ZSL), independently assessed all the potential studies identified by the search strategy and applied the inclusion criteria. Any disagreements were resolved through discussion. The excluded studies, and the reason for their exclusion are included in the 'Characteristics of excluded studies'.
Data extraction and management
For eligible studies, two review authors (EEE and ZSL) extracted the data using a pre-tested data extraction form. For dichotomous outcomes, such as clinical failure, we extracted the total number of participants and number of participants that experienced the event. For continuous outcomes, such as fever clearance time, we extracted the total number of participants, arithmetic means, and standard deviations. If the standard deviation was not reported, we attempted to use the confidence interval or P value to derive it. The extracted data were entered data into Review Manager 5.1. and cross-checked by a second author for accuracy.
Assessment of risk of bias in included studies
Two review authors (EEE and ZSL) independently assessed the risk of bias for each included trial using the Cochrane collaboration's 'Risk of bias' tool as described in the Cochrane Handbook of Systematic Reviews of Intervention (Higgins 2011).
We followed the guidance to assess whether adequate steps were taken to reduce the risk of bias across six domains: sequence generation, allocation concealment, blinding (of participants, personnel and outcome assessors), incomplete outcome data, selective outcome reporting and other sources of bias. We have categorized our judgements as 'yes' (low risk of bias), 'no' (high risk of bias) or 'unclear'. We compared our entries and resolved disagreements by discussion.
|Figure 1. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 2. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
Measures of treatment effect
Dichotomous data are presented and compared using risk ratios (RR), and continuous data using a mean difference (MD). All results are presented with the corresponding 95% confidence interval (CI).
Unit of analysis issues
Trials including more than two comparison groups have been split and analysed as individual pair-wise comparisons. When conducting meta-analysis we have ensured that participants and cases in the placebo group are not counted more than once, by dividing the placebo cases and participants evenly between the intervention groups.
Dealing with missing data
We were unable to conduct an intention-to-treat analysis on culture-positive cases since no further information was available for culture-positive participants who were lost to follow up.
Assessment of heterogeneity
We assessed for heterogeneity by visually inspecting the forest plots and by using the Chi
Assessment of reporting biases
We planned to assess for the presence of publication bias by looking for funnel plot asymmetry but this was not possible due to the low number of trials.
We analysed data using Review Manager 5.1.
We analysed data using pair-wise comparisons. we compared the fluoroquinolones with each alternative antibiotic and subgrouped by the specific fluoroquinolone. The data are organised into four sections:
- fluoroquinolones versus first-line antibiotics (chloramphenicol, co-trimoxazole, and ampicillin or amoxicillin);
- fluoroquinolones versus second-line antibiotics (cefixime, ceftriaxone, azithromycin);
- comparison of different fluoroquinolones and different durations of fluoroquinolones;
- a summary of the evidence for gatifloxacin
Where there is no statistical heterogeneity we have used the fixed-effect model. Where statistical heterogeneity was detected, and we still considered it appropriate to pool the data, we used the random-effects model.
Subgroup analysis and investigation of heterogeneity
We planned to investigate heterogeneity by conducting subgroup analyses according to; drug dose; severe or complicated enteric fever (as defined by trialists) versus uncomplicated enteric fever; and different time points for outcome measurements. This was not possible due to the limited number of trials in each comparison. We have instead commented on these factors within the text where appropriate.
We planned to assess the robustness of the data by performing a sensitivity analysis for each of the risk of bias assessment factors, but were again unable to do this due to the low number of trials.
Description of studies
Results of the search
We assessed 72 trials for eligibility. Twenty-six were included and 36 excluded. Seven studies are awaiting classification and one trial is ongoing.
Among the seven trials awaiting classification; we were unable to retrieve full text copies for two (Flores 1991; Soewandojo 1992), and four did not provide adequate information on the methodology for inclusion (Quintero 1988; Weng 1996, Xiao 1991, Yu 1998). (See the Characteristics of studies awaiting classification table).
The 26 trials included 3033 participants. Most trials were small and lacked statistical power to detect differences between the treatment regimens. The smallest trial had 23 participants and the largest had 352 participants.
Nine trials were conducted in Vietnam, two trials in each of Italy, Nepal and Pakistan, and one trial in each of Albania, Bahrain, Bangladesh, Egypt, Guatemala, Indonesia, Laos, Morocco and Turkey. We could not determine the location of one trial (Gottuzzo 1992 N/A).
Twenty-two of the 26 trials were conducted on inpatients. Alam 1995 BGD was conducted on both inpatients and outpatients. Tran 1995 VNM was a community-based outpatient trial, while Pandit 2007 NPL and Arjyal 2011 recruited outpatients presenting to the outpatient or emergency department of the study hospital.
Twenty-two trial reports included data on the prevalence of MDR strains among trial participants, and 13 trial reports included data on the prevalence of NaR strains.
Of the 13 trials comparing fluoroquinolones with first-line antibiotics, MDR strains were only present in two trials (Phongmany 2005 LAO; Arjyal 2011), they were absent in seven trials, and four trials did not report it (Gottuzzo 1992 N/A; Yousaf 1992 PAK; Flores 1994 MEX; Rizvi 2007 PAK).
See Appendix 2 for further details on microbiological results and sensitivity.
Three trials were exclusively in children (Vinh 1996 VNM; Phuong 1999 VNM; Vinh 2005 VNM). Seven trials included both children and adults (Alam 1995 BGD; Tran 1995 VNM; Pandit 2007 NPL; Parry 2007 VNM;Rizvi 2007 PAK; Dolecek 2008 VNM; Arjyal 2011), 15 trials were exclusively in adults (Hajji 1988 MAR; Limson 1989 PHL; Gottuzzo 1992 N/A; Morelli 1992 ITA; Yousaf 1992 PAK; Wallace 1993 BHR; Smith 1994 VNM; Cristiano 1995 ITA; Unal 1996 TUR; Chinh 1997 VNM; Kalo 1997 ALB; Girgis 1999 EGY; Chinh 2000 VNM; Gasem 2003 IDN; Phongmany 2005 LAO), and five trial reports did not mention the participants' age of which one used the keyword "adult" (Flores 1994 MEX).
Eighteen trials were conducted specifically on participants with uncomplicated enteric fever or participants without major complications of enteric fever (Limson 1989 PHL; Gottuzzo 1992 N/A; Wallace 1993 BHR; Flores 1994 MEX; Tran 1995 VNM; Unal 1996 TUR; Vinh 1996 VNM; Vinh 2005 VNM; Phuong 1999 VNM; Chinh 1997 VNM; Girgis 1999 EGY; Chinh 2000 VNM; Gasem 2003 IDN; Phongmany 2005 LAO; Pandit 2007 NPL; Parry 2007 VNM; Dolecek 2008 VNM; Arjyal 2011 ), and one included only participants with severe enteric fever (Cristiano 1995 ITA). The remaining trials did not provide this information.
Most trials used blood cultures, bone marrow cultures, or both, to confirm cases of enteric fever. In Rizvi 2007 PAK, a rapid diagnostic test - Dot Enzyme immunosorbent Assay - was also used although all but one participant was culture positive.
Trials that included patients diagnosed clinically tended to report outcomes only for culture-confirmed cases of enteric fever and excluded culture-negative cases from their analysis, even if initially enrolled in the study. Only Arjyal 2011 detailed analyses were done both reporting culture positive cases only and intention to treat which included patients randomized but who were culture negative..
Nineteen trials compared fluoroquinolones with alternative antibiotics: chloramphenicol (eight trials), amoxicillin or ampicillin (two trials), co-trimoxazole (three trials), azithromycin (four trials), ceftriaxone (two trials), and cefixime (three trials). Seven trials compared different fluoroquinolone treatment durations: two days versus three days (three trials); three days versus five days (one trial), five days versus seven days (one trial); seven days versus 10 days (one trial); and 10 days versus 14 days (one trial).
Most trials comparing fluoroquinolones with a non-fluoroquinolone antibiotic treated the participants with the fluoroquinolone for seven (eight trials) or 10 days (six trials) (range: three to 15 days).
There were considerable variations regarding the time points at which outcomes were measured, particularly microbiological failure (such as day two, the end of treatment, and some days after treatment) and relapse (such as during therapy or up to two months after treatment completion). The precise descriptions also varied considerably; for example, some trialists defined "relapse" as the recurrence of similar signs and symptoms with confirmation by blood and/or bone marrow culture (sterile site, as defined in protocol), and others as confirmed by positive stool cultures (non-sterile site) only. Some trialists did not explicitly state how they confirmed relapse in their trial (see Appendix 3 'Definitions of outcomes'). A full summary of adverse events as stated in the papers is summarized in Appendix 4 and Appendix 5.
Further details are presented in the Characteristics of included studies tables.
Of the excluded studies, five were excluded as they used norfloxacin (Nalin 1987; Sarma 1991; Huai 2000; Bai 1995; ZhongYang 1997), and three involved fluoroquinolones no longer in clinical use (Abejar 1993; Arnold 1993; Tran 1994). (For further details see the Characteristics of excluded studies table).
Risk of bias in included studies
See summary of 'risk of bias' assessment in Figure 2
The method used to generate the allocation sequence was at low risk of bias in sixteen trials, and unclear in ten.
Fourteen trials used an adequate method (sealed envelopes) to conceal allocation. The method used in the remaining 12 trials was unclear.
Three trials were described as "double blinded" and 22 trials were open; one trial did not mention use of placebo, but we assumed it was open (Flores 1994 MEX). In one trial, blinding was unclear (Rizvi 2007 PAK).
Incomplete outcome data
Most trials reported both efficacy and safety data. In one trial (Wallace 1993 BHR), there were no reports of adverse events while in another, the report was incomplete as only mortality and associated data for one participant were reported (Phuong 1999 VNM).
Other potential sources of bias
Two trials were stopped early. One because of a significant difference in the primary outcome (Phongmany 2005 LAO), and one due to apparent lower efficacy in the control group, the cost of control drug and inconvenience of intravenous administration (Wallace 1993 BHR). Two trials were funded by pharmaceutical companies (Alam 1995 BGD; Girgis 1999 EGY)
Effects of interventions
Fluoroquinolones versus first-line antibiotics (chloramphenicol, co-trimoxazole, and ampicillin or amoxicillin)
Comparison 1. Fluoroquinolones versus chloramphenicol
Overall, a seven-day course of any fluoroquinolone appears at least as effective as a 14-day course of chloramphenicol at reducing clinical and microbiological treatment failures (eight trials, 916 participants). In the most recent study, from Pakistan in 2003-04, high failure rates were seen with chloramphenicol, and the fluoroquinolones used (ciprofloxacin and ofloxacin) were superior.
Eight trials have compared four different fluoroquinolones with chloramphenicol: Four trials used ciprofloxacin as the comparator drug (Gottuzzo 1992 N/A; Morelli 1992 ITA; Gasem 2003 IDN; Rizvi 2007 PAK), four trials used ofloxacin (Morelli 1992 ITA; Yousaf 1992 PAK; Phongmany 2005 LAO; Rizvi 2007 PAK), two used pefloxacin (Morelli 1992 ITA; Cristiano 1995 ITA), and one trial assessed gatifloxacin (Arjyal 2011).
Two studies did not clarify the proportion of participants with MDR strains (Gottuzzo 1992 N/A; Yousaf 1992 PAK), and seven did not report NaR data. The dosing of fluoroquinolones varied from trial to trial and is included in the forest plots as footnotes to aid interpretation.
Clinical and microbiological response
Only one three-arm study, from Pakistan in 2003-04, demonstrated a statistically significant benefit with fluoroquinolones compared to chloramphenicol (Rizvi 2007 PAK). The incidence of clinical and microbiological failure with chloramphenicol was high in this trial (9/44) suggesting significant resistance, although this was not confirmed microbiologically. Clinical failures were lower with both ciprofloxacin (RR 0.05, 95% CI 0.00 to 0.81, 92 participants, one trial, Analysis 1.1; Analysis 1.2), and ofloxacin (RR 0.05, 95% CI 0.00 to 0.86, 89 participants, one trial, Analysis 1.1; Analysis 1.2).
Conversely, the largest trial to date found no significant difference between gatifloxacin and chloramphenicol in Nepal in 2006-08 (352 participants, one trial, Analysis 1.1; Analysis 1.2). The remaining older trials were too small to detect clinically important differences between the treatment regimens should they exist.
Relapse and convalescent faecal carriage
The current trials have not shown a statistically significant difference in post-treatment relapses or fecal carriage with any fluoroquinolone compared to chloramphenicol ( participants, six trials, Analysis 1.5; Analysis 1.6). The follow-up in the included trials varied from two weeks to six months.
Fever clearance time
Fever clearance time was significantly longer with chloramphenicol compared with ciprofloxacin (MD -62.46, 95% CI -75.52 to -49.39, 147 participants, two trials, Analysis 1.5) and ofloxacin (MD -75.85, 95% CI -88.52 to -63.17, 140 participants, two trials, Analysis 1.5).
Duration of hospitalisation
Participants who had chloramphenicol in Phongmany 2005 LAO, stayed a significantly longer number of days in hospital compared with ofloxacin (MD -9.90, 95% CI -11.42 to -8.38, 60 participants, one trial, Analysis 1.6). However, we note that ofloxacin was given only for three days compared to the 14 days of chloramphenicol, so this is perhaps unsurprising.
No difference has been shown between ciprofloxacin and chloramphenicol (173 participants, two trials, Analysis 1.7), or ofloxacin and chloramphenicol where no serious adverse events were recorded (50 participants, one trial, Analysis 1.7).
Non-serious adverse events were significantly lower following treatment with gatifloxacin than with chloramphenicol (RR 0.58, 95% CI 0.44 to 0.78, 844 participants, one trial, Analysis 1.8). This data included all randomized participants including those who were culture negative. The events were mainly gastrointestinal in nature and the common ones included abdominal pains, diarrhoea, nausea and vomiting. Elevated blood sugar was more common in the gatifloxacin group between the second and seventh days of the study. There was no difference in the number of participants with low blood sugar for both groups.
The differences between the other fluoroquinolones and chloramphenicol did not reach statistical significance ( Analysis 1.8).
Comparison 2. Fluoroquinolones versus cotrimoxazole
In one study, from an area of Pakistan in 2003-04, the fluoroquinolones used (ciprofloxacin and ofloxacin) were superior to co-trimoxazole. Two small trials done in the 1980s, in the absence of MDR strains, failed to show a difference with both drugs performing well.
Three trials have compared three different fluoroquinolones with cotrimoxazole: two trials used ciprofloxacin; Limson 1989 PHL; Rizvi 2007 PAK, and one trial each assessed ofloxacin; Rizvi 2007 PAK, and pefloxacin; Hajji 1988 MAR.
Hajji 1988 MAR and Limson 1989 PHL both report the absence of MDR strains and Hajji 1988 MAR also records that there were no participants with NaR strains. Limson 1989 PHL and Rizvi 2007 PAK do not report the presence or absence of NaR strains.
Clinical and microbiological response
Of the three trials, only Rizvi 2007 PAK reports any clinical failures at all. In this trial, from Pakistan in 2003-04, there was a high incidence of clinical and microbiological failure following treatment with co-trimoxazole (13/44) suggesting significant resistance, compared with no clinical failures following ciprofloxacin (RR 0.03, 95% CI 0.00 to 0.56, 92 participants, one trial, Analysis 2.1) or ofloxacin (RR 0.04, 95% CI 0.00 to 0.59, 89 participants, one trial, Analysis 2.1).
Relapse and convalescent faecal carriage
Fever clearance and duration of hospitalisation
Serious adverse events were not reported.
No statistically significant difference in non-serious events has been shown between any individual fluoroquinolone and co-trimoxazole (219 participants, three trials, Analysis 2.6). The events were mainly gastrointestinal in nature and were self limiting.
Comparison 3. Fluoroquinolones versus amoxicillin or ampicillin
Two small studies conducted in the 1990s, found that ofloxacin given for 10 to 14 days reduced clinical and microbiological failures compared to a 10 to 14 day course of amoxicillin or ampicillin. The prevalence of antibiotic resistance was not reported.
There was no indication as to the presence or not of MDR or NaR strains.
Clinical and microbiological response
The risk of clinical or microbiological failure was significantly lower in the ofloxacin group compared to ampicillin or amoxicillin (RR 0.11, 95% CI 0.02 to 0.57, 90 participants, two trials, Analysis 3.1; RR 0.13; 95% CI 0.03 to 0.68, 90 participants, two trials, Analysis 3.2 respectively). It should be noted that these two trials are almost 20 years old and may not be relevant today.
Relapse and convalescent faecal carriage
Fever clearance and duration of hospitalization
No serious adverse events occurred in either of the studies. Non-serious events were significantly more following treatment with ofloxacin compared to amoxicillin (RR 0.27; 95% CI 0.09 to 0.86, 50 participants, one trial, Analysis 3.3). The reported events were mostly diarrhoea and skin rashes.
Fluoroquinolones versus second-line antibiotics (cefixime, ceftriaxone, azithromycin)
Comparison 4. Fluoroquinolones versus cefixime
In one study from Pakistan in 2003-04 no clinical or microbiological failures were seen with either ciprofloxacin, ofloxacin or cefixime. In Nepal in 2005, gatifloxacin reduced clinical failure and relapse compared to cefixime, despite a high prevalence of NaR in the study population.
Three trials have compared a fluoroquinolone with cefixime. One trial used ciprofloxacin as the comparator drug (Rizvi 2007 PAK), two trials used ofloxacin (Phuong 1999 VNM; Rizvi 2007 PAK) and one gatifloxacin (Pandit 2007 NPL).
In one trial, participants were mostly adults (Pandit 2007 NPL) while one trial had only child participants (Phuong 1999 VNM). The third trial included both adult and child participants (Rizvi 2007 PAK). One trial (Pandit 2007 NPL) had a high proportion of NaR strains, but the other two trials did not report the presence of these strains ( Phuong 1999 VNM; Rizvi 2007 PAK). In Pandit 2007 NPL, because of its wholly out patient status, community medical auxiliaries conducted twice daily home-based assessments and provided directly observed treatment with study drugs. All participants were then compulsorily seen at the hospital on Day 10.
Clinical and microbiological response
Of the three tested fluoroquinolones, only gatifloxacin has demonstrated a statistically significant reduction in clinical failure compared to cefixime (RR 0.04, 95% CI 0.01 to 0.31, 158 participants, one trial, Analysis 4.1). Microbiological failures were too low across all three trials to demonstrate any significant differences for any of the comparisons (379 participants, three trials, Analysis 4.2).
Relapse and convalescent faecal carriage
Only gatifloxacin has demonstrated a statistically significant reduction in relapse (RR 0.20, 95% CI 0.04 to 0.93, 138 participants, one trial, Analysis 4.3). There were no reported incidents of faecal carriage.
Fever Clearance and duration of hospital stay
Fever clearance time was significantly shorter for ofloxacin (MD -24.00, 95% CI -41.46 to -6.54, one trial, 91 participants, Analysis 4.5). There was a statistically significant difference in the median time to fever clearance for gatifloxacin (92 hours vs 138 hours, P <0.0001) and ofloxacin (105 hours vs 201 hours, P <0.0001). This reductions for ofloxacin were also reflected in the shorter length of hospital stay in that group (MD -3.00, 95% CI -4.53 to -1.47, 80 participants, one trial, Analysis 4.6).
Serious adverse events were low in two trials comparing ofloxacin and gatifloxacin with cefixime but there was no significant difference between the comparisons (251 participants, two trials, Analysis 4.7).
Non-serious adverse events appear to be higher with gatifloxacin than with cefixime (RR 20.92, 95% CI 2.9 to 150.90, 169 participants, one trial, Analysis 4.8). However it is not clear whether adverse events were completely reported in this trial. No difference has been shown between ciprofloxacin or ofloxacin and cefixime.
Comparison 5. Fluoroquinolones versus ceftriaxone
Two studies, conducted almost 20 years ago, compared five to seven days of an oral fluoroquinolone with three days of intravenous ceftriaxone, and were too small to demonstrate important differences if they exist. The prevalence of NaR strains was either absent or unreported.
In both trials, over half of participants had MDR strains. There were no participants with NaR strains in Smith 1994 VNM whereas the proportion was not stated in Wallace 1993 BHR. Sample sizes for these studies were quite small resulting in very wide confidence intervals.
Clinical and microbiological response
The proportion of clinical failures was lower with fluoroquinolones but the result was not statistically significant (89 participants, two trials, Analysis 5.1). Only one microbiological failure is reported in either group ( Analysis 5.2).
Relapse and convalescent faecal carriage
Fever clearance and duration of hospitalization
Only Smith 1994 VNM reported adequate data for the fever clearance time which was significantly shorter with ofloxacin (MD -115.0; 95% CI -150.67 to -79.33, 47 participants, 1 trial, Analysis 5.5). In Wallace 1993 BHR, mean fever clearance times for the ciprofloxacin and ceftriaxone groups were 4 and 5.2 days respectively. No standard deviation was reported but the P value was given as 0.04.
Similarly, only Smith 1994 VNM reported the duration of hospitalization which averaged nine days (range: 6 to 13 days) in the ofloxacin group and 12 days (range: 7 to 23 days) in the ceftriaxone group. No values for standard deviation were reported but a P value was given as 0.01.
No serious adverse events were reported. Non-serious events were few, mild and self limiting in both groups in the only trial that recorded them (47 participants, 1 trial, Analysis 5.6).
Comparison 6. Fluoroquinolones versus azithromycin
Azithromycin was superior to ofloxacin in reducing clinical failures and convalescent faecal carriage in populations with both MDR and NaR enteric fever in Vietnam. The most recent study, also from Vietnam, found no difference between gatifloxacin and azithromycin with both drugs performing well.
Four trials involving 564 participants made this comparison. One trial each compared ciprofloxacin with azithromycin (Girgis 1999 EGY, 64 participants) and gatifloxacin with azithromycin (Dolecek 2008 VNM, 287 participants). Two trials compared ofloxacin with azithromycin (Chinh 2000 VNM, 88 participants and Parry 2007 VNM, 125 participants).
In Girgis 1999 EGY, a third of participants were infected with MDR strains. The proportion of NaR strains was not reported. The other trials had varying proportions of participants with MDR and NaR strains.
Clinical and microbiological response
Treatment with azithromycin resulted in a statistically significant decrease in clinical failures compared to ofloxacin (RR 2.20, 95% CI 1.23 to 3.94, 213 participants, two trials, Analysis 6.1), but no difference has been shown between ciprofloxacin (64 participants, one trial, Analysis 6.1), or gatifloxacin (287 participants, one trial, Analysis 6.1)
No statistically significant difference in microbiological failure has been seen in any of the trials comparing fluoroquinolones with azithromycin (564 participants, four trials, Analysis 6.2).
Relapse and convalescent faecal carriage
No statistically significant difference in relapse rate has been seen in any of the trials comparing fluoroquinolones with azithromycin (479 participants, 4 trials, Analysis 6.3).
Convalescent faecal carriage was lower in the azithromycin group compared with ofloxacin (RR 13.52, 95% CI 2.64 to 69.36, 193 participants, 2 trials, Analysis 6.4), but no difference has been shown between ciprofloxacin (64 participants, 1 trial, Analysis 6.4), or gatifloxacin (268 participants, 1 trial, Analysis 6.4)
Fever clearance time
No consistent statistically significant difference in fever clearance has been shown between any of the fluoroquinolones and azithromycin (564 participants, four trials, Analysis 6.5).
Duration of hospitalization
There was a statistically significant reduction in the duration of hospital stay in the ofloxacin group (RR 1.01, 95% CI 0.19 to 1.83, 213 participants, two trials, Analysis 6.6)
No significant difference in serious events has been seen between the ofloxacin with azithromycin groups (88 participants, 1 trial, Analysis 6.7). Overall, non-serious adverse events were similar across all the trials (564 participants, four trials, Analysis 6.8).
Head to head comparisons of different fluoroquinolones or different durations of treatment
Differences in efficacy between the different fluoroquinolones has not been demonstrated in head to head clinical trials.
The different fluoroquinolones have only been compared as part of multiple arm studies (Morelli 1992 ITA; Rizvi 2007 PAK). In these studies no clinical or microbiological failures, or relapses were seen in the fluoroquinolone treatment arms (see Analysis 1.1; Analysis 1.2; Analysis 1.3).
None of the comparisons demonstrated one duration was superior to another for failure or relapse, even in the presence of MDR and NaR strains. Studies were generally too small to detect what might be important differences.
Comparison 7. Fluoroquinolones for two days versus three days
Three trials made this comparison: one in adults (Chinh 1997 VNM) and two in children (Vinh 1996 VNM; Vinh 2005 VNM). They were all ofloxacin trials. All three trials reported the percentage of participants with NaR and MDR strains. These were 2.5.% and 90% (Vinh 2005 VNM), 5% and 79% (Chinh 1997 VNM), and 13% and 84% (Vinh 1996 VNM), respectively.
There were no statistically significant differences for all the outcomes in either groups of the trials. There were no serious adverse events.
Comparison 8. Fluoroquinolones three days versus five days
Only one trial (Tran 1995 VNM) with over 70% children compared three days with five days of ofloxacin. The majority of S. typhi isolates (91%) were MDR. Some participants had NaR strains, although the precise number of these participants was not available.
Fever clearance time was significantly shorter in the three day group compared with the five day group (MD -12.0; 95% CI -18.07 to -5.93,195 participants, one trial, Analysis 8.2). There were no differences in the risk of relapse and adverse events in either groups.
Comparison 9. Fluoroquinolone five days versus seven days
One trial made this comparison (Unal 1996 TUR) with pefloxacin. Thirteen percent of the strains were MDR. There was no report of the proportion with NaR strains.
There were no clinical failures in either arm, and we did not detect a statistically significant difference in microbiological failure, relapse and fever clearance time. Adverse events were not serious and they were similar in both groups.
Comparison 10. Fluoroquinolone seven days versus 10 or 14 days
One trial compared pefloxacin for seven days with 10 days (Kalo 1997 ALB) in a population wholly infected with ampicillin resistant S. typhi some of whom were MDR. The proportion of participants with NaR strains was not reported.
There was no statistically significant difference in microbiological failure. There were no clinical failures or convalescent faecal carriers. Adverse events were mild and self limiting.
Comparison 11. Fluoroquinolone 10 days versus 14 days
One trial, with 7% of the participants infected with NaR strains, made this comparison (Alam 1995 BGD). There was no statistically significant difference in relapse or fever clearance time. There were no clinical or microbiological failures, or convalescent faecal carriers. Adverse events (gastrointestinal symptoms, headache and rashes in both arms, and one case of joint pain in the 14-day arm) were mild and self limiting.
Summary of gatifloxacin comparisons
In the light of the emerging interest in gatifloxacin, we have summarised the results in this section for this one drug.
One trial each compared gatifloxacin with chloramphenicol (Arjyal 2011), cefixime (Pandit 2007 NPL) and azithromycin (Dolecek 2008 VNM). All the trials had a majority of participants infected with NaR strains equally distributed between groups. In Pandit 2007 NPL and Arjyal 2011, MDR strains were negligible (0.58% and 0% respectively). All the trials were conducted in areas previously known to have a high prevalence of MDR and NaR salmonella isolates.
Comparison 13. Gatifloxacin versus chloramphenicol
No statistically significant difference has been shown in the risk of clinical or microbiological failure, or relapse, between a 7-day course of gatifloxacin and 14 days of chloramphenicol in Nepal (352 participants, one trial, Analysis 11.1). Treatment with gatifloxacin may however be associated with fewer adverse events (RR 0.58, 95% CI 0.44 to 0.78, 844 participants, one trial, Analysis 11.1).
Comparison 14. Gatifloxacin versus cefixime
Compared to 7-days of cefixime, a 7-day course of gatifloxacin was shown to produce a statistically significant reduction in clinical failure (RR 0.04; 95% CI 0.01 to 0.31, 158 participants, one trial, Analysis 13.1), and relapse (RR 0.2; 95% CI 0.04 to 0.93, 138 participants, one trial, Analysis 13.1) in Nepal. There was however no difference in microbiological failure assessed at day 10 (158 participants, one trial, Analysis 13.1). Gatifloxacin was associated with a statistically significant increase in adverse events (RR 19.25, 95% CI 2.66 to 139.30, 169 participants, one trial, Analysis 13.1). The events were mainly vomiting and in two cases, this was severe enough to require intravenous fluids.
Comparison 15. Gatifloxacin versus azithromycin
No statistically significant difference has been shown in the risk of clinical or microbiological failure, or relapse, between a 7-day course of gatifloxacin and 7 days of azithromycin in Vietnam (287 participants, one trial, Analysis 14.1). There is also no evidence of a difference in the incidence of adverse events (285 participants, one trial, Analysis 14.1).
Summary of main results
Fluoroquinolones versus older antibiotics
In one study from Pakistan in 2003-04, high clinical failure rates were seen with both chloramphenicol and co-trimoxazole, although resistance was not confirmed microbiologically. A seven day course of either ciprofloxacin or ofloxacin was found to be superior. Older studies of these comparisons failed to show a difference.
In two small studies conducted almost two decades ago the fluoroquinolones were demonstrated to be more effective than ampicillin and amoxicillin.
Fluoroquinolones versus current second-line options
The two studies comparing a seven day course of oral fluoroquinolones with three days of intravenous ceftriaxone were too small to detect important differences between antibiotics should they exist.
In Pakistan in 2003-04 no clinical or microbiological failures were seen with seven days of either ciprofloxacin, ofloxacin or cefixime. In Nepal in 2005, gatifloxacin reduced clinical failure and relapse compared to cefixime, despite a high prevalence of NaR in the study population.
Compared to a seven day course of azithromycin, a seven day course of ofloxacin had a higher rate of clinical failures in populations with both MDR and NaR enteric fever in Vietnam in 1998-2002. However, the most recent study, also from Vietnam in 2004-05, found no difference between gatifloxacin and azithromycin, with both drugs performing well.
Fluoroquinolones versus alternative fluoroquinolones
Differences in efficacy between the available fluoroquinolones, or between different durations of treatment with an individual fluoroquinolone, have not been demonstrated in head to head clinical trials.
Overall, the adverse event profiles were similar for the fluoroquinolone and non-fluoroquinolone antibiotics. They were mostly mild and self limiting.The risk of dysglycaemia with gatifloxacin has been reported in several studies (Frothingham 2005, Park-Wyllie 2006). However, in the three studies included in this review which report on dysglycaemia ( Pandit 2007 NPL; Dolecek 2008 VNM; Arjyal 2011), no difference was detected in the risk of hypoglycaemia or hyperglycaemia among those studied
Overall completeness and applicability of evidence
Most of the included trials were conducted on inpatients and may not be representative of the majority of settings where most enteric fever is managed as outpatients. The data are likely to represent a subset of patients with more severe illness who may respond less favourably to conventional therapy.
The changing epidemiology of resistance patterns across various regions precludes any generalization of the results of the included studies. Indeed, some included studies are nearly two decades old and thus may not be useful in informing practice.
In addition, overall, there are too few studies in each comparison, and the studies themselves are too small, to make any firm conclusions on the prescience or absence of important differences between the different treatment options.
Potential biases in the review process
Although we found several trials from China, and published in the Chinese language, we were unable to extract adequate details on the trial methodology to allow inclusion. These studies are listed in the 'Studies awaiting assessment' table. A recent study of over 30,000 apparent RCTs in China showed that only 6.8% were authentic RCTs (Wu 2009).
Agreements and disagreements with other studies or reviews
In our previous update (Thaver 2008), different types of fluoroquinolone were combined in the meta analyses in spite of their dissimilarity. In this revision, we have analysed them separately with the intention of highlighting the effectiveness of different fluoroquinolones. We also considered the changing pattern of resistance across various regions over different times.
Implications for practice
Generally, fluoroquinolones performed well in the treating typhoid. Generally, fluoroquinolones performed well in the treating typhoid, and maybe superior to alternatives in some settings. However, we were unable to draw firm general conclusions on comparative contemporary effectiveness given that resistance changes over time, and many studies were small. In choosing any fluoroquinolone, clinicians need to take into account current, local resistance patterns.
There is some evidence that the newest fluoroquinolone, gatifloxacin, remains effective in some regions where resistance to older fluoroquinolones has developed. However, the different fluoroquinolones have not been compared directly in head to head trials.
Implications for research
The re-emergence of chloramphenicol sensitive strains in some regions may suggest a similar trend for other first line drugs which had been abandoned following prevalent MDR. Trials may therefore focus on re-examining these relatively inexpensive alternatives in robust comparisons with fluoroquinolones in appropriate endemic populations.
Most of the studies were small . Given the importance of the study question, we would recommend multi-centred, adequately powered trials, with robust methods and analytical design. Given the nature of the disease and the importance of accurate diagnoses, we would recommend the development of robust diagnostic tests and gold standards for defining disease and resistance patterns, based on molecular methods if possible. Rapid diagnostic tests for diagnosing enteric fever should be made more widely available in endemic areas; this will ensure more efficient participant recruitment in trials and avoid the problem of syndromic treatment. In addition, it will reduce the widespread abuse of antibiotics, especially the use of fluoroquinolones for suspected typhoid fever.
Definitions of outcomes and their measurement should also be standardized to make more effective comparisons and adaptability across regions.
Emmanuel Effa's fellowship to Liverpool School of Tropical Medicine, and the update of this review was supported by a grant from the World Health Organization. The Cochrane Infectious Diseases Group is funded by the UK Department for International Development (DFID) for the benefit of low- and middle-countries.
We acknowledge Vittoria Lutje, Trials Search Co-ordinator for the CIDG for assistance with the searches as well as staff of the CIDG for assistance during the process of updating the review.
The Contact Editor for this review was Dr Mical Paul.
Thanks to Durrane Thaver, who wrote the original and first update of this review. She is fondly remembered by all of us that knew her.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. Detailed Search Strategy
Appendix 2. Description of drug resistance by study
MDR: multiple-drug-resistant strain; MIC: minimum inhibitory concentration; NaR: nalidixic acid resistant strain.
Appendix 3. Definitions of outcomes
S. Typhi/S. Paratyphi: Salmonella enterica serovar Typhi/Paratyphi.
Appendix 4. Serious adverse events
Appendix 5. Non-serious adverse events
Last assessed as up-to-date: 1 February 2011.
Protocol first published: Issue 4, 2003
Review first published: Issue 2, 2005
Contributions of authors
Emmanuel Effa and Zohra Lassi , considered the new search, extracted and enter data, updated the risk of bias assessment and Dave Sinclair co-extracted data, assisted with restructuring and writing up of the review. Julia Critchley provided technical inputs and assisted with the restructuring of the review. Prof Zulfiquar Bhutta, Prof Paul Garner, and Piero Olliaro guided the restructuring, examined the data, provided technical direction and edited the manuscript. All authors contributed to the final manuscript.
Declarations of interest
None known. Professor ZA Bhutta has been part of trials of treatment for typhoid therapy in children, none of which involved fluoroquinolones.
Sources of support
- University of Calabar Teaching Hospital, Calabar, Nigeria.
- Nigeria branch of South African Cochrane centre, Nigeria.
- No sources of support supplied
Differences between protocol and review
We changed the intervention from 'Fluoroquinolone antibiotic' to 'Different fluoroquinolone antibiotic excluding norfloxacin or other fluoroquinolones not currently in use'
The Contact Editor for this review was Dr Mical Paul.
Medical Subject Headings (MeSH)
Anti-Bacterial Agents [adverse effects; *therapeutic use]; Fluoroquinolones [adverse effects; *therapeutic use]; Norfloxacin [therapeutic use]; Paratyphoid Fever [*drug therapy]; Randomized Controlled Trials as Topic; Treatment Outcome; Typhoid Fever [*drug therapy]
MeSH check words
Adult; Child; Humans