Scrub typhus is a bacterial disease caused by Rickettsia tsutsugamushi (Orientia tsutsugamushi). It is transmitted by larval Leptotrombidium mites, which are commonly called chiggers. Infection can cause widespread inflammation of the blood vessels in many organs, especially the lungs, kidneys, and central nervous system (Silpapojakul 1997).
People with scrub typhus develop fever, headache, and a cough. Uncommonly patients develop a characteristic skin lesion, called an eschar (Brown 1976; Sirisanthana 1989), which makes the diagnosis much more likely. The eschar occurs at the site of the bite, starts as a large pimple ('papule'), and then the centre part of the skin dies and turns black, giving it the appearance of a cigarette burn. Laboratory tests are not widely available: the Weil-Felix test (with Proteus OX-K) is insensitive (Sirisanthana 1989), and serodiagnostic tests are only available in reference centres.
Scrub typhus is common in the western Pacific region and many parts of Asia. It is listed as one of the differential diagnoses of fever of unknown origin in people in endemic areas. In Thailand, some estimates suggest that 10% of people with fever have scrub typhus (Silpapojakul 1997). A seroprevalence study among 200 febrile people attending some malaria clinics in western Thailand has indicated that 59.5% had serology (immunoglobulin M (IgM) and/or immunoglobulin G (IgG)) positive for scrub typhus infection (Chanyasanha 1998). A study conducted in five hospitals in Thailand among patients with undifferentiated febrile illness without focal infection showed that 19.9% had serology positive for scrub typhus (Suttinont 2006). A study among 207 hospitalized patients with febrile illness of 5 to 30 days duration in Southern India indicated that 24% had elevated IgM antibodies of scrub typhus (Varghese 2006).
The risk of scrub typhus is closely related to occupation. Most cases in Asia are acquired through agricultural exposure to oil palm and rubber plantations in Malaysia, or rice fields in Thailand. Scrub typhus is also associated with travel activities such as camping, rafting, or trekking in endemic areas (Silpapojakul 1997). A number of cases have been reported in 'eco-travellers' from Europe and America (McDonald 1988; Watt 1994). Recently, three cases from Scandinavia were reported from infections acquired in Thailand, Laos, and Sri Lanka respectively (Jensenius 2006). In addition, scrub typhus is an important disease in military personnel undertaking field duties in endemic areas (Deller 1967; Berman 1973). Severe, life threatening scrub typhus has also been reported in neonates as a result of the infection being transmitted from their mothers (Wang 1992; Suntharasaj 1997).
Prognosis and treatment
The severity of the disease is thought to depend on the virulence of the Rickettsia strain, the patient's age and genetic factors, and whether the person has previously been infected. The outbreak of scrub typhus in India reported 17.2%(19/113) mortality (Kumar 2004). Antibiotic treatment is thought to shorten the illness and reduce mortality. It is usually presumptive, being given to febrile individuals where the disease is endemic. Chloramphenicol was the first drug described in a series of studies to reduce the morbidity and mortality associated with the disease (Smadel 1950). Tetracycline and doxycycline have also been used.
Data from case series
To evaluate information about treatment from case series, we systematically searched for case series of people treated for scrub typhus. The search was carried out in 2000, and the results summarised in Table 1. In one study involving 42 adults treated with a single dose of 200 mg doxycycline, 88% became afebrile and all clinical symptoms disappeared within 72 hours, and no relapses occurred in the month following treatment (Supparatpinyo 1990). In children, a good response to tetracycline and chloramphenicol treatment was described in 25 cases, who all became afebrile in 48 hours. Relapse occurred in two children and fever subsided spontaneously in one of them within four days (Sirisanthana 1989).
In a case series from Chiangrai, northern Thailand, a poor response to a 7-day course of doxycycline was reported; only 5/12 (40%) patients were afebrile at 72 hours. Moreover, the result of doxycycline susceptibility testing in mouse fibroblast cell culture showed that only 39% of the patients had strains that were fully susceptible to doxycycline (Watt 1996) ( Table 1).
The new macrolide antibiotic, azithromycin, was recently evaluated in vitro against R. tsutsugamushi. Azithromycin was effective against some strains of R. tsutsugamushi, particularly doxycycline-resistant strains. Azithromycin is considered safer than other antibiotics for use in young children and pregnant women. It has advantages over chloramphenicol, which occasionally causes bone marrow suppression, and tetracycline, which affects the growing bones and teeth of children and foetuses (Strickman 1995).
Tetracycline or chloramphenicol are generally recommended (CDC 2007). However, the potential advantage of alternative drugs (azithromycin, telithromycin, clarithromycin, and ciprofloxacin) has not been directly established; and length of treatment to prevent recrudescence is not clear. This review aims to summarize the information about the effects of various classes of antibiotics on scrub typhus.
To evaluate antibiotic regimens for treating scrub typhus.
Criteria for considering studies for this review
Types of studies
Randomized and quasi-randomized controlled trials.
Types of participants
People diagnosed with scrub typhus, as defined by the trial authors.
Types of interventions
Any antibiotic treatment that aims to treat scrub typhus, compared with another antibiotic regimens.
Types of outcome measures
Fever still present 48 hours after treatment started.
Relapse within three months (return of fever or other symptoms during follow up).
Treatment failures (persistence of symptoms, fever, and laboratory abnormalities at end of treatment).
Duration of illness.
Duration of fever.
Number of adverse events.
Search methods for identification of studies
We attempted to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and in progress).
We searched the following databases using the search terms and strategy described in Table 2 : Cochrane Infectious Diseases Group Specialized Register (2 January 2010); Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library (Issue 4, 2009); MEDLINE (1966 to 2 January 2010); EMBASE (1980 to 2 January 2010); and LILACS (1982 to 2 January 2010). We also searched the metaRegister of Controlled Trials (mRCT) on 2 January 2010 using 'scrub typhus' or 'orientia tsutsugamushi' as search terms.
We also checked the reference lists of all studies identified by the above methods.
Data collection and analysis
Selection of studies
We screened the results of the search strategy to identify potentially relevant trials, obtained the full reports of these trials and applied the inclusion criteria to assess their eligibility for inclusion in the review. We resolved disagreements by discussion and have given the reason for excluding trials in the 'Characteristics of excluded studies'.
Data extraction and management
We extracted data using a standardized data extraction form and the second author checked data extraction.; we checked the data sources to avoid multiple publication based on the same data. We extracted data to allow for an intention-to-treat analysis where possible. We resolved discrepancies by discussion.
Assessment of risk of bias in included studies
We classed generation of the allocation sequence and allocation concealment as adequate, inadequate, or unclear according to Jüni 2001. We recorded which people were blinded, such as the participants, care providers, or assessors. We considered the inclusion of all randomized participants in the analysis to be adequate if it was greater than or equal to 90% and inadequate if less than 90%. We displayed this information in a table and described it in the text. We resolved inconsistencies through discussion. Both authors participated in the assessment.
We analysed the data using Review Manager (Review Manager 5). For binary data, we calculated relative risk (RR) and 95% confidence intervals (CIs) using the fixed-effect model; for continuous data, we calculated weighted mean difference and 95% CIs.
We assessed heterogeneity by visually examining the forest plot and through the Chi
Description of studies
Seven trials met the inclusion criteria (see Characteristics of included studies). Both excluded studies evaluated prophylaxis (see Characteristics of excluded studies). Two ongoing trials compare rifampin with doxycycline, and azithromycin with doxycycline respectively (see Characteristics of ongoing studies).
Sheehy 1973 reported on military servicemen who acquired scrub typhus in Vietnam, and were evacuated "at random" to one of two military hospitals. Participants were recruited in 1966 although the study was not published until 1973. It is not entirely clear whether it was the evacuation that was random or the hospital allocation; pending clarification, we have included the study. A total of 63 participants were enrolled, and three with malaria were then excluded. Participants evacuated to hospital A were given chloramphenicol (n = 30), and those to hospital B were given tetracycline (n = 30); both groups were followed up for three weeks. The outcomes measured by the researchers were participants afebrile by 48 hours, and duration of fever.
Brown 1978 enrolled 149 adult participants in Malaysia presumed to have scrub typhus. They were randomly assigned to receive either doxycycline or tetracycline. In their analysis, the authors then excluded participants with no serological evidence of infection, or participants with serological changes but a co-infection. This left 55 participants (doxycycline n = 31; tetracycline n = 24) with clinical and serological evidence of scrub typhus infection. The outcomes measured by the researchers were disappearance of symptoms, participants afebrile by 48 hours, and any adverse drug effects.
Song 1995conducted a multicentre study in Korea, and enrolled 129 adult participants. Of these, 116 met their serological diagnostic criteria. They were randomly allocated to either doxycycline (n = 66) or tetracycline (n = 50) groups, and followed up for 4 weeks. The outcomes measured by the researchers were failure of treatment and relapse.
Watt 2000 enrolled 126 adult participants in Thailand who were diagnosed with mild scrub typhus. They were randomly assigned to receive either doxycycline, rifampicin, or combined regimen of rifampicin and doxycycline. During the first year of the study there were a number of participants in one arm of the study who had protracted fever more frequently than the others. This led the researchers to perform an interim analysis. The fever clearance time was significantly different between the three treatment groups: 88 hours (n = 8); 35 hours (n = 5); and 24 hours (n = 9). They decided to break the code of the group with the 88 hours clearance time and found that it was a combination group. This regimen was discontinued and participants subsequently recruited to this group were assigned high dose rifampicin (900 mg/day). During the study, the authors excluded participants with adverse treatment effects and those who had illness other than scrub typhus. In their analysis, the participants who had no serological evidence of infection were excluded. This left 78 participants (doxycycline n = 28; standard rifampicin dose n = 26 and high rifampicin dose n = 24). The outcomes measured were duration of fever, participants febrile at 48 hours, relapse, and side effects.
Kim 2004 enrolled 99 adult participants in Korea diagnosed mild scrub typhus on the basis of clinical criteria. They were randomly assigned to receive either azithromycin or doxycycline. The authors excluded four participants confirmed to have combined infection with other disease, one participant who developed severe vomiting and one participant who was given the wrong dose of medication. This left 93 participants (azithromycin n = 47; Doxycycline n = 46) with clinical evidence of scrub typhus infection. Both groups were followed up for 30 days.The outcomes measured by the researchers were time to defervescence, cure, failure, relapse and adverse drug effects.
Kim 2007 conducted a multicenter study in Korea, and enrolled 95 adult patients with possible scrub typhus. Of these, three had concurrent diseases and was excluded. The left 92 participants were randomly allocated to either telithromycin (n = 47) or doxycycline (n = 45) group, and followed up for four weeks. Seventy-six out of the 92 participants met diagnostic laboratory criteria of scrub typhus. The outcomes measured by the researchers were fever clearance time, cure, failure, relapse, toxicity and adverse events.
Phimda 2007 conducted a multicenter study in Thailand, and enrolled 296 adults with suspected leptospirosis or scrub typhus. They were randomly allocated to receive either doxycycline or azithromycin, and the median duration of follow-up was 15 days. The cause of acute fever were obtained for 151 out of 296 patients. Of these, 57 patients (doxycycline n = 27; azithromycin n = 30) met diagnostic laboratory criteria of scrub typhus. The outcomes measured by the researchers were cure, failure, defervescence and adverse events.
Risk of bias in included studies
Our assessment of risk of bias is summarised in Table 3 with individual trial details provided in the 'Characteristics of included studies'. Generation of allocation sequence was adequate in three trials, and one trial had adequate allocation concealment; follow up was classed as adequate in four trials; no trial had blinding.
The method of random allocation was not clear in Sheehy 1973. The diagnosis of scrub typhus was based on clinical criteria, and only 19/60 participants were tested using the Weil-Felix test with Proteus OX-K agglutinins. This test was the only one available at this time, but is not specific for the disease. While all participants received drugs for at least three days, the total length of treatment was decided by clinicians and not reported. All participants were followed up for three weeks.
The random allocation was not clear in Brown 1978, but entry criteria were more rigorously applied. The authors excluded a large proportion of participants who did not have serological evidence of scrub typhus. Clinicians also had discretion to alter treatment after 48 hours if there was no clinical improvement.
In Song 1995, 129 adult participants were enrolled. They were allocated at random by computer generated numbers, but allocation was not concealed. In total, 76/129 met the diagnostic laboratory criteria. Both groups were similar in relation to severity of disease and clinical manifestations. All participants were followed up for four weeks.
The method of random allocation and concealment was not clear in Watt 2000. The diagnosis of scrub typhus was based on clinical criteria and positive serological dipstick test. After enrolment, a large number of participants were excluded. The protocol was changed during the first year of the study. One intervention group was discontinued and replaced with the other new treatment group. These factors meant the number of participants completing the study in each treatment group was small.
In Kim 2004, computer-generated random sequences was mentioned but no description of concealment or blinding. The diagnosis of scrub typhus was based on clinical and laboratory criteria. After enrolment, a number of participants were excluded. An intention-to-treat analysis was based on 93 participants who completed the treatment. All participants were followed up for 30 days.
Kim 2007 used a quasi-randomized controlled trial design and allocated by last digit of a resident registration number. Concealment of allocation was not clearly documented. Seventy-six out of 92 included patients met the diagnostic laboratory criteria. Results were presented based on the intention-to-treat analysis. All participants were followed up for four weeks.
In Phimda 2007, 296 participants were allocated at random by computer generated numbers, and the random allocation sequences were sealed in an opaque envelope and numbered. A large number of participants were lost to follow-up after discharge from the hospital. In total, 57/296 met the diagnostic laboratory criteria of scrub typhus. Results were presented based on the intention-to-treat analysis.
Effects of interventions
In Sheehy 1973, presence of fever at 48 hours was not different after tetracycline or chloramphenicol (60 participants, Figure 1, Analysis 1.1); mean duration of fever was 28 hours (range 14 to 68 hours) in the tetracycline group, and 35 hours (range 16 to 94 hours) in the chloramphenicol group. The authors did not report the standard deviation or any statistical tests performed on the duration of fever data. Relapse was reported in two participants in the tetracycline group and in five in the chloramphenicol group (60 participants, Analysis 1.2).
|Figure 1. Forest plot of comparison: 1 Tetracycline vs chloramphenicol, outcome: 1.1 Febrile after 48 hours.|
Two trials compared doxycycline with tetracycline (N = 171; Brown 1978, Song 1995). No difference in fever at 48 hours was detected in the one study (Brown 1978) measuring this (55 participants, Figure 2, Analysis 2.1) and no relapses were reported in either group. Treatment failure was reported by Song 1995: 4/66 participants in the doxycycline group and 0/50 in the tetracycline group, but the difference was not statistically significant (116 participants, Analysis 2.4). The length of fever between the two treatment groups was similar 34.0 (standard deviation 26.5) hours compared to 37.0 (standard deviation 26.6) hours, respectively, and not statistically significantly different.
|Figure 2. Forest plot of comparison: 2 Doxycycline vs tetracycline, outcome: 2.1 Febrile after 48 hours.|
One trial compared doxycycline with rifampicin (Watt 2000). Fewer patients in the rifampicin group remained febrile at 48 hours (RR 0.41, 95% CI 0.22 to 0.77; 78 participants, Figure 3, Analysis 3.1). This result was calculated by combining the standard and high-dose rifampicin groups). The median fever clearance time in the doxycycline group was 52 hours (range 4 to 108) compared to 27.5 hours (range 4 to 84) in the standard rifampicin group, and 22.5 hours (3 to 76) in high-dose rifampicin group (P = 0.01). Relapse was reported only in the doxycycline group (2/28) over a one-month period of follow-up ( Analysis 3.2). Mild gastrointestinal symptoms were common in all groups (10 (50%) in doxycycline group; 8 (31%) in standard rifampicin group and 14 (43%) in high-dose rifampicin group).
|Figure 3. Forest plot of comparison: 3 Rifampicin vs doxycycline, outcome: 3.1 Febrile after 48 hours.|
In the comparison between high and low rifampicin dose in the same trial, there was no statistically significant difference of effect on participants being febrile at 48 hours (50 participants, Analysis 4.1).
Two trials compared azithromycin with doxycycline (Kim 2004, Phimda 2007). No difference was detected between azithromycin and doxycycline for fever at 48 hours (150 participants, 2 trials, Figure 4, Analysis 6.1). No difference in median fever clearance time was found between azithromycin group (21 hours, range 1 to 120 in Kim 2004; 60 hours, range 12 to 128 in Phimda 2007) and doxycycline group (29 hours, range 4 to 176 in Kim 2004; 48 hours, range 16 to 120 in Phimda 2007). No relapses were reported in either group. Gastrointestinal side effects were reported by Kim 2004 in the azithromycin group (7/47) and in the doxycycline group (12/46, RR 0.57, 95% CI 0.25 to 1.32; 93 participants, Analysis 6.3). Treatment failure was reported by Phimda 2007: 1/30 participants in the azithromycin group and 0/27 in the doxycycline group, but the difference was not statistically significant (57 participants, Analysis 6.4).
|Figure 4. Forest plot of comparison: 6 Azithromycin vs doxycycline, outcome: 6.1 Febrile after 48 hours.|
One trial compared telithromycin with doxycycline (Kim 2007). The length of fever between the two treatment groups was similar 20.5 (standard deviation 12.9) hours compared to 22.6 (standard deviation 21.4) hours, respectively, and the difference was not statistically significant (92 participants, Analysis 7.1). Relapse was reported only in the doxycycline group (1/45) over a two-week period of follow-up (92 participants, Analysis 7.3). Adverse events were reported in the telithromycin group (7/47) and in the doxycycline group (11/45, RR 0.61; 95% CI 0.26 to 1.43; Analysis 7.4). Gastrointestinal side effects were reported in the telithromycin group (3/47) and in the doxycycline group (6/45, RR 0.48, 95% CI 0.13 to 1.80; Analysis 7.5). No treatment failure was reported.
No trials reported deaths or serious complications.
There are several trials of scrub typhus treatment with antibiotics. The diagnostic criteria for scrub typhus in Sheehy 1973 study was based on mostly clinical features, so it is not clear how many participants actually had scrub typhus. Two trials (Brown 1978 and Song 1995) had more strictly laboratory diagnostic criteria, although the Weil-Felix test with Proteus OX-K is relatively insensitive. In Watt 2000, a serological dipstick test was used to screen mild scrub typhus for enrolment. The researchers then conducted a separate test to confirm the diagnosis. This led to some participants being excluded from the study after randomization. The other three trials (Kim 2004; Kim 2007; Phimda 2007) were based on WHO criteria to screen mild scrub typhus for enrolment. In Kim 2004 and Kim 2007, among participants who completed the treatment, not all of them were confirmed scrub typhus, however, the intention to treat analysis was performed based on participants who complete the treatment.
The uncertainty around diagnosis causes problems in clinical decision making, as well as in evaluating antibiotics. It appears in these studies participants with other infectious diseases were excluded, increasing the certainty of the diagnosis, but this remains a problem until better tests are more widely available.
The concealment of allocation was poorly reported, and only one study included in this review (Phimda 2007) reported adequate allocation concealment. Studies were small. Three recent studies used intention-to-treat analysis (Kim 2004; Kim 2007; Phimda 2007).
Overall, broad spectrum antibiotics are known to be effective in this condition, but trials are currently insufficient to determine whether and under what conditions one is more effective than another.
Implications for practice
The antibiotics tested appear to cure the condition, and there seem to be little to choose between the broad spectrum antibiotics tested, but trials are small.
Rifampicin seem to be more effective than doxycycline in areas where scrub typhus appears to respond poorly to conventional antibiotics (tetracycline and chloramphenical), and where doxycycline-resistance strain is suspected.
Clinicians should monitor the progress of patients in the light of reports of drug resistance.
Implications for research
Further research is required to evaluate antibiotics for scrub typhus. Trials would be more easily interpreted if reliable diagnostic tests were available. Such research could examine whether a single dose of doxycycline is as effective as a three to five day course of treatment.
Regimens for severe disease need to be evaluated, for example, comparing intravenous chloramphenicol with intravenous tetracycline.
Studies are also needed to evaluate alternative antibiotics, particularly in areas where scrub typhus appears to respond poorly to conventional antibiotics.
We would like to thank Professor Paul Garner for his great contribution to the previous version of this review.
We wish to thank Prof Virat Sirisanthana, Dr George Watt, and Dr George Wyatt for their advice and comments.
This document is an output from a project funded by DFID for the benefit of developing countries. The views expressed are not necessarily those of DFID.The authors take sole responsibility for the data presented and the views expressed.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Last assessed as up-to-date: 25 May 2010.
Protocol first published: Issue 2, 2000
Review first published: Issue 2, 2000
Contributions of authors
Ratana Panpanich: involved with the design and writing of the protocol, duplication of screening of titles and abstracts, inclusion/exclusion of full text papers, quality assessment of included studies, data extraction and analysis, initial draft of the review.
Qin Liu: co-ordinated the review update, involved with the duplication of screening of titles and abstracts, inclusion/exclusion of full text papers, quality assessment of included studies, data extraction, updated the data analysis and draft of the review.
Declarations of interest
We certify that we have no affiliations with or involvement in any organization or entity with a direct financial interest in the subject matter of the review (eg, employment, consultancy, stock ownership, honoraria, and expert testimony).
Sources of support
- Faculty of Medicine, Chiang Mai University, Thailand.
- Liverpool School of Tropical Medicine, UK.
- Department for International Development, UK.
- European Union Directorate General XII, Belgium.
REVIEW HISTORY (started 4 March 2002)
4 March 2002: Updated review received by editorial base. This includes a new trial (Watt 2000), and responses to comments from Assistant Editor and statistician: (1) slight change to the objective; adverse outcomes changed from "Number and seriousness of side effects" to "Number of adverse events"; Relative Risk used for binary outcomes (previously Peto odds ratio)
June 2010: Updated review received by editorial base. Primary outcomes amended so no longer include "death" as primary outcome. Added new trials.
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study