Only two studies have compared mefloquine with doxycycline. A trial in Thailand [10] involved little exposure to malaria and no placebo control, and a recent trial in Africa [11] showed a relatively low efficacy for each drug. Few blinded or placebo-controlled trials have assessed the tolerability of these drugs. Because mefloquine and doxycycline are the agents currently recommended for prophylaxis of malaria, we believed that a well-designed and carefully conducted direct comparison of the efficacy and tolerability of these two compounds was necessary. Distinguishing features of our trial include 1) intense exposure to malaria; 2) nonimmune study participants; 3) a double-blind, placebo-controlled design to permit unbiased assessment of tolerability; 4) use of a loading dose of mefloquine; and 5) careful monitoring of compliance.

Methods

Top Methods Results Discussion Author & Article Info References

Study Site and Participants

This trial was conducted from May to July 1994 in northeastern Irian Jaya, Indonesia, a lowland, partially cleared forest area 50 km south and east of the provincial capital, Jayapura. In this area, P. falciparum is resistant to sulfadoxine-pyrimethamine and both P. falciparum and P. vivax are resistant to chloroquine [12-15]. Approximately 330 soldiers were deployed to 18 military posts; a typical post consisted of 15 soldiers and was located at a road check point or in a local community.

Two hundred four soldiers who met all entrance criteria and agreed to participate were randomly assigned to one of three groups. The 140 soldiers from Ambon, Indonesia, had recently arrived (before study enrollment) and had received one dose of sulfadoxine-pyrimethamine; the 64 soldiers from Sumatra, Indonesia, had received weekly doses of sulfadoxine-pyrimethamine in Irian Jaya for 3 months. Before study prophylaxis began, all soldiers were given daily prophylaxis with doxycycline tablets for 4 to 6 weeks to allow clearance of sulfadoxine and pyrimethamine from their blood.

Our study was conducted in accordance with U.S. Army, U.S. Navy, and Republic of Indonesia regulations governing the protection of human participants. All volunteers gave written, informed consent.

Prestudy Evaluation and Radical Curative Treatment

All soldiers from military posts that were considered to have high malaria attack rates and that were within the study area were asked to participate. Before enrollment, volunteers had a medical history taken and underwent physical examination and laboratory testing. Initial laboratory testing consisted of a complete blood count (QBC CBC, Becton Dickinson, Mountainview, California) and assessment of glucose-6-phosphate dehydrogenase (G-6-PD) activity (G-6-PD spot test, Sigma, St. Louis, Missouri). Participants who had a history of frequent travel, chronic illness, allergy to one of the study drugs, or G-6-PD deficiency were excluded from participation. Before the study began, all participants received radical curative treatment to clear preexisting malaria parasites from the blood and liver. Treatment consisted of quinine sulfate (9 to 11 mg/kg of body weight three times a day for 4 days; Danbury Pharmaceuticals), doxycycline (100 mg twice a day for 10 days; Pfizer Indonesia, Jakarta, Indonesia), and primaquine (30-mg base/d for 14 days; Sanofi Winthrop, New York, New York).

Drug Supply, Randomization, and Packaging

Mefloquine hydrochloride in 250-mg tablets that contained a 228-mg base was provided by Hoffman-La Roche in Nutley, New Jersey. Placebo tablets that were identical in appearance to the mefloquine tablets were provided by Hoffman-La Roche in Basel, Switzerland. Doxycycline in 100-mg capsules and placebo capsules that were identical in appearance but not in content to the doxycycline capsules were provided in weekly blister packs by Pfizer Indonesia. Two investigators randomly assigned the participants to study groups and packaged the drugs; the randomization code was stored in individual envelopes in a locked box at the study site. All investigators and study personnel did not have access to or know the randomization code throughout the study. Block randomization was used (block size, 15); participants were allocated in equal numbers to the mefloquine, doxycycline, and placebo groups. Drugs were packaged into weekly zipper-lock plastic bags; each bag contained a mefloquine or mefloquine placebo tablet and a blister pack of seven doxycycline or doxycycline placebo capsules (double-dummy technique). During the radical cure, individual soldiers were assigned consecutive study drug numbers by military post and drug packages were labeled with each soldier's name, age, rank, post, and signature to ensure correct identification of the participant.

Study Initiation, Drug Dosage, and Documentation of Compliance

The radical curative treatment and start of study therapy were staggered according to military post over a 4-week period. Therapy with the study drugs was initiated after the last dose of primaquine in the radical curative treatment regimen was received. Participants received 100 mg of doxycycline per day and mefloquine placebo; 250 mg of mefloquine per week (after a loading dose of 250 mg per day for 3 days) and doxycycline placebo; or placebos for both drugs. A study employee visited each post every morning. This employee identified each soldier by name and signature, opened his packet of the study drug, and watched him swallow the pill or pills and drink water. Both the participant and the employee signed the daily record form. Each day, this form was returned to a supervisor at the main office for confirmation of the soldier's and worker's signatures. All participants were encouraged to eat at the time of drug administration.

The study employee searched for any soldier who was not present at the time of the employee's visit. If the participant was not found, the drug packet and a special form were left with the soldier's supervisor, who would later witness the drug administration. The same special form was used for soldiers who had planned to travel during the study period. For unanticipated travel, soldiers were given a standby packet of the study drug and a signature form, were told to store these items in their travel pack, and were instructed to provide double-signature documentation. Approximately every week, a supervisor and the investigators made unscheduled visits to the posts to identify problems. For any participant who developed malaria, records of previous drug administration were reviewed; these participants were carefully questioned about missed drug doses. In situations where a doxycycline or doxycycline placebo capsule was missed, two capsules were taken on the following day. If a mefloquine or mefloquine placebo tablet was missed, the tablet was taken the following day.

End Points

The primary end point for efficacy was the first occurrence of malaria, as documented by a positive malaria smear. Giemsa-stained thick and thin malaria smears were obtained weekly and when the participants had any of the following symptoms of malaria: headache, fever, chills, nausea, or vomiting. Smears were examined by oil immersion microscopy (magnification, x1000) and were considered negative if no asexual parasites were found in 200 ocular fields of the thick film. If the initial smear was positive, the participant was promptly visited by a physician; a second malaria smear was obtained; and blood was drawn for malaria culture, complete blood count, and measurement of drug levels. Participants with P. falciparum malaria were treated with quinine (9 to 11 mg/kg) three times per day for 4 days and with doxycycline (100 mg) twice per day for 10 days; participants with P. vivax malaria received chloroquine (600-mg base followed by 300-mg base 6 hours later and at 24 and 48 hours). After treatment for malaria, participants were offered prophylaxis with doxycycline (in tablet form rather than capsules).

Later, all positive smears and a sample of negative smears were read by a senior technician who was blinded to the initial smear result. Differences in interpretation of the smears were resolved by a third blinded reading. The randomization code was broken after all slide diagnoses had been confirmed and all data had been entered and checked.

Assessment of Study Drug Tolerability

Tolerability of the study drugs was assessed in five ways. In the first assessment, we directly compared the proportion of participants in the doxycycline and mefloquine groups who ever reported a given symptom at any time during the study (participants were asked daily if they had had any symptoms). Data obtained on the first day of the study (before the dose of the study drug) and during the last week of the study (to eliminate symptoms of malaria) were excluded from the daily reporting of symptoms in all three groups. In the second assessment, we compared the frequency of symptoms between the two drug groups and between each drug group and the placebo group (background rate) by using incidence density ratios or relative risk. This approach corrected for unequal person-days of observation among the groups.

In the third assessment, we compared the proportions of participants who had a symptom during the first week of the study, when nearly equal numbers of participants were present in all three groups and when the risk for intolerability of the study drugs was believed to be highest (during and after the administration of the mefloquine loading dose). The fourth assessment, an exit questionnaire, was performed by one investigator during the last month of the study to specifically ask about known or suspected intolerability to the study drugs. Only data from persons who were still receiving the study drug at the time of questioning were included. Participants were asked whether they felt healthier after participating in the study; which symptoms they thought were related to the study drug; whether they had dizziness, nausea, diarrhea, difficulty sleeping, and dreams; and whether they had regularly taken the drug with water and food. In the fifth assessment, participants who required discontinuation of study therapy or dose adjustment because of possible side effects were described in detail.

Statistical Analysis

Incidence rates (incidence density) for all cases of malaria and for cases of malaria caused by either P. falciparum or P. vivax were calculated for each group as the number of cases of malaria divided by total person-years of follow-up. Each soldier's person-time was included until the soldier left the study. Cumulative risk was calculated as the number of cases divided by the number of participants beginning study prophylaxis. Participants who did not complete the radical cure were excluded from the analysis. In the efficacy analyses, a mixed infection was considered to have been caused by P. falciparum. Protective efficacy was defined as the percentage reduction in the occurrence of malaria and was calculated as (1 – the relative risk) x 100, where the relative risk was defined as the ratio (drug: placebo) of either incidence rates or cumulative risks. Koopman method was used to obtain CIs for the ratio of two proportions [16] to calculate CIs for protective efficacy based on relative risk. Exact CIs for incidence density rates (cases/person-time) were based on the Poisson distribution. The exact conditional distribution (binomial distribution) was used to calculate CIs for protective efficacy on the basis of the incidence density ratio [17]. For the placebo group, the Kaplan-Meier product-limit method was used to estimate the cumulative probability of remaining free of malaria during follow-up for all cases of malaria combined and for cases of malaria caused by either P. falciparum or P. vivax (at the time when one type of malaria occurred, the other type was considered censored) [18]. The Fisher exact test was used to compare proportions of spontaneously reported symptoms. Unpaired t-tests were used to compare mean values of continuous outcome variables. Unless otherwise noted, all reported CIs are 95% CIs and all reported P values are two-sided. Epi Info 6.02 (Centers for Disease Control and Prevention, Atlanta, Georgia) and SPS 6.1 for Windows (SPS Inc., Chicago, Illinois) were used for both data management and standard statistical analyses.

Role of Funding Sources

Co-investigators from the Tropical Medicine Unit of F. Hoffmann-La Roche Ltd. in Basel, Switzerland, assisted with protocol design, regulatory issues, and quality assurance. Neither of the pharmaceutical companies that provided support played any role in gathering, analyzing, or interpreting the data.

Results

Top Methods Results Discussion Author & Article Info References

Study Participants, Compliance Documentation, and Follow-up

Three hundred thirty-one soldiers were deployed to the study area. Of these, 53 (16%) were not asked to participate because their military post was in an area of low transmission of malaria organisms, 34 (10.3%) refused to participate or to have blood drawn, and 40 (12.1%) were excluded from participation for the following reasons: Eleven were G-6-PD deficient, 6 frequently traveled, 5 had an underlying illness, 1 was allergic to the study drug, 11 were intolerant to quinine during radical curative treatment, and 6 left the area. Of the 331 soldiers, 204 (61%) met all entrance criteria and were randomly assigned to one of the three study groups (67 to the doxycycline group, 68 to the mefloquine group, and 69 to the placebo group). Sixteen of the 204 participants (7.8%) did not complete the study. Five doxycycline recipients (7.4%) withdrew from the study (3 because of travel from the area, 1 because of noncompliance with the study protocol, and 1 because of a false-positive malaria smear with concurrent illness); 7 mefloquine recipients (10.3%) withdrew (4 because of travel from the area, 2 because of noncompliance with study protocol, and 1 because of concurrent illness); and 4 recipients of both placebos (5.8%) withdrew (3 because of travel from the area and 1 because of a false-positive malaria smear).

The participants' baseline characteristics and history of malaria are summarized in Table 1. The groups were well balanced with regard to baseline variables and possible risk factors. Troops from Sumatra had baseline malaria smear evaluations while receiving prophylaxis with sulfadoxine-pyrimethamine: Five percent of these soldiers (3 of 64) were positive for P. falciparum and 30% (19 of 64) were positive for P. vivax; most were asymptomatic. The 140 men in the battalion from Ambon had been receiving prophylaxis with doxycycline for 10 days before we were given permission to collect baseline smears; none of these participants was positive for either organism.

Signature documentation of the 15 221 doses of the study drug was as follows: member of research team plus participant, 95.5% (14 542 doses); participant plus other witness (usually the post supervisor), 2.7% (416 doses); participant only, 1.3% (203 doses); no dose taken, 0.3% (48 doses); and no documentation, 0.1% (12 doses). No significant differences were seen in numbers of witnessed or missed doses between the study groups (data not shown).

Initially enrolled participants who did not develop malaria or who dropped out of the study were followed for a mean of 13.5 weeks (range, 12.7 to 14.6 weeks). An additional seven participants who were stationed at a post that enrolled late and who did not develop malaria were followed for a mean of 9.4 weeks.

Protective Efficacy

Table 2 shows the attack rate (per person-year) and protective efficacies of mefloquine and doxycycline relative to placebo. In the placebo group, 53 (77%) of the participants developed malaria (9.1 person-years), resulting in an attack rate of 5.8 cases per person-year (95% CI, 4.3 to 7.7 cases per person-year). Of these cases, 57% were caused by P. falciparum and 44% were caused by P. vivax. One mixed infection occurred; this infection was considered to have been caused by P. falciparum. In the placebo group, the first cases of P. falciparum and P. vivax malaria occurred 13 and 11 days, respectively, after the last dose of primaquine (Figure 1).

View larger version (15K): [in this window] [in a new window]   Figure 1. Kaplan-Meier plot (survival curve) of estimated cumulative probability of remaining free of malaria in 69 placebo recipients. Plots are for all cases of malaria (53 cases), cases of malaria caused by Plasmodium falciparum (30 cases), and cases of malaria caused by P. vivax (23 cases). The species-specific curves consider the other species as censored (removed) at the time of occurrence of malaria. The mefloquine and doxycycline groups were excluded from this plot for clarity. Short vertical bars represent the four participants who dropped out of the study; numbers in parentheses are the numbers of participants at risk.

No cases of malaria developed in the mefloquine group (16.9 person-years). On the basis of the reduction in incidence density, this yields protective efficacies of 100% (CI, 96% to 100%) for all cases of malaria, 100% (CI, 93% to 100%) for cases caused by P. falciparum, and 100% (CI, 91% to 100%) for cases caused by P. vivax. One case of malaria occurred in the doxycycline group (16.0 person-years), yielding protective efficacies of 99% (CI, 94% to 100%) for all cases of malaria, 98% (CI, 88% to 100%) for cases caused by P. falciparum, and 100% (CI, 90% to 100%) for cases caused by P. vivax.

Study Drug Tolerability and Laboratory Data

In the first assessment of tolerability, we directly compared the proportion of participants in the doxycycline and mefloquine groups who had reported a symptom at any time during the study (Table 3, columns 1 and 2). Unlike the placebo group, the doxycycline and mefloquine groups had a similar number of participants throughout the surveillance period; this minimized the bias introduced by differences in observation time. We found that significantly fewer soldiers in the doxycycline group reported any gastrointestinal symptoms, anorexia, any neurologic symptoms, dizziness, headache, and subjective fever than did soldiers in the mefloquine group. Cough was reported by more participants in the doxycycline group. Table 3 lists 20 symptom categories: Fewer doxycycline recipients than mefloquine recipients had symptoms in 16 categories; the same number of participants in both groups had symptoms in 3 categories; and more doxycycline recipients than mefloquine recipients had symptoms in 1 category.

In the second assessment, we compared the frequency of symptoms between the two drug groups and between each drug group and the placebo group (background rate) by using incidence density ratios or relative risk (Table 3, last 3 columns). By using this method to determine the total number of reported symptoms, we noted that both doxycycline and mefloquine were significantly better tolerated than placebo (P < 0.001 and P = 0.005, respectively) and that doxycycline was better tolerated than mefloquine (P = 0.006). No symptom occurred significantly more frequently in either the doxycycline or mefloquine group than in the placebo group.

During the first week of the study, a symptom was spontaneously reported by 69 (34%) participants: 30% of those receiving doxycycline (20 of 66; P > 0.2 [relative to placebo]), 37% of those receiving mefloquine (25 of 68; P > 0.2), and 35% of those receiving placebo (24 of 68). Gastrointestinal symptoms were reported by 22 (11%) participants during the first week of the study: 9.1% (P > 0.2 [relative to placebo]) of the doxycycline group, 16.2% (P = 0.09) of the mefloquine group, and 7.4% of the placebo group. No statistically significant differences were seen in the first week for any symptom, including neurologic symptom.

An exit questionnaire was conducted during the last month of the study. The only statistically significant finding was that more doxycycline recipients noted nausea if they did not take the drug with food: Twenty-two (35%) doxycycline recipients compared with 9 (14%) mefloquine recipients (P = 0.003) and 2 (11%) placebo recipients (P = 0.03) noted this symptoms.

Two soldiers were removed from the study because of illness: one (a doxycycline recipient) because of a serious neurologic event (acute hysteria) and a false-positive malaria smear and the other (a mefloquine recipient) because of a febrile illness of uncertain origin. An additional mefloquine recipient who had a history of head trauma and strabismus refused to take the drug for several days because of headaches and an increase in visual difficulty. At the time of evaluation, his symptoms were considered to be unrelated to the study drug. When his symptoms resolved, therapy with the study drug was resumed, and he successfully completed the study. No participant was withdrawn from the study because of intolerance to the study drugs or adverse effects that seemed to be related to the study drugs.

Complete blood counts obtained before and after the study revealed a significant increase in hematocrit in the doxycycline group (increase, 0.0147) when compared with changes in hematocrit in the mefloquine group (decrease, 0.0050; P = 0.005) and the placebo group (decrease, 0.0055; P = 0.002). The total leukocyte count, the differential, and the platelet count did not change significantly in any study group.

Discussion

Top Methods Results Discussion Author & Article Info References

This randomized, double-blind, placebo-controlled comparison of mefloquine and doxycycline showed that both agents were highly efficacious as prophylaxis for chloroquine-resistant P. falciparum and P. vivax malaria in a largely nonimmune population. Only one case of malaria occurred in the doxycycline group, and no cases occurred in the mefloquine group. However, in the placebo group, malaria developed in more than 50% of the participants during the first 6 weeks of the study and had developed in 77% by the end of the study. These results yield high protective efficacies with narrow 95% CIs. Our study also confirms that both drugs are very well tolerated.

Efficacy in this trial was as good as or better than efficacy in previous trials that evaluated mefloquine or doxycycline. Initial field trials of prophylaxis with mefloquine in semi-immune Thai villagers in 1977 showed that weekly (180 or 360 mg) and every-other-week (360 mg) regimens had high efficacy [19]. Since 1977, six studies have been performed near the border areas of Thailand; these studies used weekly mefloquine or mefloquine-sulfadoxine-pyrimethamine [20-25]. All but one of these studies showed 70% to 90% protective efficacy or effectiveness (efficacy refers to monitored compliance; effectiveness refers to unmonitored compliance) against P. falciparum malaria. Lobel and colleagues [26] found that weekly mefloquine had an effectiveness of 94% (CI, 86% to 97%; calculated as the percentage reduction in incidence density) compared with chloroquine in Peace Corps volunteers working in sub-Saharan Africa. In a large cohort study of travelers returning from East Africa, Steffen and colleagues [27] found that the prophylactic effectiveness of mefloquine was 91% (CI, 85% to 94%; calculated as the percentage reduction in incidence density).

Four controlled field trials evaluated the prophylactic efficacy or effectiveness of doxycycline in Thailand from 1985 to 1988. In 1985-1986, Pang and coworkers [28] showed that investigator-witnessed daily administration of doxycycline (at a dose equivalent to the 100-mg adult dose) compared with chloroquine yielded a protective efficacy of 87% (CI, 66% to 96%; calculated as the percentage reduction in incidence density) for P. falciparum malaria in semi-immune adolescent children. In a follow-up placebo-controlled trial in the same study sample [29], doxycycline (100-mg adult equivalent dose) provided protective efficacies of 89% (CI, 51% to 96%) for P. falciparum malaria and 96% (CI, 85% to 98%; calculated as the percentage reduction in incidence density) for P. vivax malaria. In 1986-1987, a double-blind, placebo-controlled cohort study of Thai soldiers [30] found that 100 mg of doxycycline per day provided protective efficacies of 98% (CI, 92% to 99%) for P. falciparum malaria and 81% (CI, –12% to 98%; calculated as the percentage reduction in incidence density) for P. vivax malaria. In 1987-1988, a follow-up randomized trial of effectiveness (73% compliance with doxycycline) in Thai soldiers [31] showed a protective effectiveness of 64% (CI, 36% to 79%) for P. falciparum malaria and 64% (CI, 44% to 76%; calculated as the percentage reduction in cumulative risk) for P. vivax malaria, relative to pyrimethamine-dapsone.

In addition, the efficacy of doxycycline has been assessed in two studies done in East Africa. A randomized, placebo-controlled clinical trial performed in semi-immune adolescent Kenyan children [11] found protective efficacies of 77% (CI, 55% to 88%) with mefloquine and 84% (CI, 66% to 92%; calculated as the percentage reduction in incidence density) with doxycycline. In a recent double-blind, placebo-controlled trial comparing azithromycin with doxycycline in semi-immune Kenyan adults, the protective efficacy of doxycycline was 93% (CI, 80% to 97%; calculated as the percentage reduction in cumulative risk) (Shanks GD. Personal communication).

Our report is a new contribution to the literature on the efficacy of doxycycline and mefloquine because our trial directly compared these drugs and showed that high efficacy is achievable. The high efficacy that we observed probably reflects that our participants were motivated to take prophylaxis, that compliance was closely monitored, that a loading dose of mefloquine was used, and that parasites were sensitive to the study drugs. In addition, if a doxycycline dose was missed, two doses were taken the following day.

The generalizability of findings from prophylaxis trials for malaria prevention is limited by three factors: parasite resistance patterns, immunity of study participants, and problems with compliance. In Irian Jaya, resistance to antimalarial drugs seems to be as great as or greater than the resistance seen in most malarious areas in the world. In this area, P. falciparum is resistant to sulfadoxine-pyrimethamine and both P. falciparum and P. vivax are resistant to chloroquine [12-15]. The high efficacy of mefloquine seen in our study, however, is not generalizable to areas where P. falciparum is highly resistant to mefloquine (for example, the Thai-Burmese and Thai-Cambodian border areas) [32-34].

Results of prophylaxis studies in semi-immune persons may not be generalizable to nonimmune populations. We believe that our sample is best classified as nonimmune to malaria. Immunity to malaria gradually develops after several exposures, is incomplete, and is lost after exposure stops [35]. A semi-immune state is reflected by decreased frequency and levels of parasitemia, as well as by parasitemia in the absence of clinical symptoms. The baseline characteristics of our participants (Table 1) suggest little recent exposure to malaria, as shown by medical histories, infrequency of splenomegaly, and normal hematocrits. Transmission of malaria organisms is very low in the areas from which the troops in our study originated (Baird K. Personal communication; Ambon, Indonesian Ministry of Health. Personal communication). As shown by their initial malaria smears, troops from Sumatra (64 of 204 participants) had been exposed to malaria for 3 months before the study while receiving sulfadoxine-pyrimethamine. We believe that any low-level clinical immunity present in this subset as a result of brief exposure to malaria did not influence outcome; thus, the results of our trial should be generalizable to travelers who have never had malaria. The fact that 49 of the 53 placebo recipients who developed malaria had two or more symptoms typical of malaria at the time of a positive smear during active case detection supports the classification of these recipients as nonimmune.

Doxycycline, with its daily dosing requirement, may be relatively less effective than mefloquine because of problems with compliance. To address this question, we followed an additional group of soldiers for 13.2 person-years (n = 83: 36 persons who did not participate in our study and 47 study participants in whom malaria had been treated). Our study team diagnosed all cases of malaria and provided all treatment but did not enforce compliance. Eight cases of malaria occurred (four caused by P. falciparum and four caused by P. vivax); this yielded a protective effectiveness of 89% (CI, 78% to 96%) for all cases of malaria and a protective effectiveness of 89% (CI, 73% to 98%) for cases caused by either P. falciparum or P. vivax. Overall, participants in this group were more likely to be exposed to malaria (many had previously developed malaria while in the placebo group of our study) and were less likely to be compliant (several did not wish to participate in the study) than the average participant. Effectiveness decreased in the group, but by less than we had anticipated.

Concern exists about the time required to achieve adequate mefloquine plasma concentrations after initiation of weekly prophylaxis (7 weeks) [36, 37]. Boudreau and colleagues [36] found that a 3-day mefloquine loading dose was well tolerated in U.S. soldiers and that this dose rapidly produced steady-state plasma mefloquine concentrations. Our trial and a large cohort study of Dutch soldiers deployed to Cambodia [32, 38] have confirmed that this regimen is very well tolerated.

None of the studies described above has identified significant problems with the tolerability of doxycycline or mefloquine. Gastrointestinal intolerance (especially when doxycycline is taken on an empty stomach), sensitivity to the sun, and vaginitis are the adverse effects most commonly associated with doxycycline [7]. Esophageal erosions may be an underappreciated adverse effect of doxycycline [39, 40]. Gastrointestinal intolerance, dizziness, and sleep disturbance have been most commonly reported with mefloquine [7, 36]. Cases of serious neuropsychiatric disorders (seizures, acute psychoses, and neurosis) have been reported in persons receiving mefloquine [27, 41-43]. A randomized, double-blind field trial directly compared the tolerability of mefloquine and doxycycline in U.S. soldiers deployed to Thailand [10, 44]. No differences in tolerability or gastrointestinal pathogens were found. Large cohort studies and a recent double-blind comparison have largely failed to demonstrate a difference in frequency of adverse effects with mefloquine and chloroquine [26, 27, 36].

Our trial is the first double-blind, placebo-controlled assessment of mefloquine and doxycycline tolerability. The five methods of assessment consistently showed that both drugs were very well tolerated. We witnessed one serious neuropsychiatric event in a doxycycline recipient but doubt that this event was associated with doxycycline. No participant was removed from the study because of a possible drug-related side effect. We found the relative risk for neurologic symptoms with mefloquine to be similar to or less than the risk seen with placebo in all categories except dizziness (relative risk, 1.42; P > 0.2). Fewer neurologic, gastrointestinal, and overall symptoms were reported in the doxycycline group than in the mefloquine and placebo groups. A possible explanation for this finding is that doxycycline prevented the occurrence of other infectious diseases. This explanation is further supported by the significantly increased hematocrit in the doxycycline group. In addition, tolerability was probably improved because we encouraged all participants to take the study drug with food.

The generalizability of our data on tolerability is limited by three factors: the study sample (young, healthy Asian men), the sample size (uncommon events were not detectable), and the geographic location [an area where infectious diseases are prevalent]. It is possible but unlikely that bias was introduced 1) when soldiers were exposed to quinine and doxycycline before the study or 2) because of the nonidentical contents of the doxycycline study capsules. In addition, incidence density data are misleading when the frequency of side effects is not constant with time and the study groups have different durations of follow-up. In our study, the total number of symptoms plotted by study week was fairly constant (data not shown). The best way to draw conclusions from data on tolerability is to seek consistent findings across methods of assessment.

We included a placebo group to determine the actual malaria attack rate. The use of the placebo group was deemed ethical because soldiers deployed to Irian Jaya usually receive only sulfadoxine-pyrimethamine or chloroquine prophylaxis for 3 months and then receive no prophylaxis. We provided 24-hour radio support and standby quinine that could be used to begin treating malaria if symptoms occurred and a microscopic diagnosis was not immediately available.

Our data suggest that both mefloquine and doxycycline, when taken as directed, are highly effective in preventing malaria in visitors traveling to areas where parasites are sensitive to these drugs. We suggest the mefloquine loading-dose regimen for persons who will enter a malarious area sooner than 1 week after initiation of prophylaxis and who do not have a contraindication to mefloquine. We believe that doxycycline should be recommended for persons who are likely to be compliant, who do not have a contraindication to the drug, who cannot tolerate mefloquine, or who are traveling in an area of mefloquine resistance. Doxycycline must be consumed with food to prevent gastrointestinal intolerance. Esophageal ulcers may be prevented if water and food are consumed after intake of the drug and if persons do not lie supine. Both doxycycline and mefloquine should be prescribed only for persons traveling to an area where malaria occurs, and both must be taken for 4 to 6 weeks after departure from the malarious area. Finally, physicians can reassure patients that intolerance to both drugs is uncommon.

Appendix: Research Working Group

U.S. Army Medical Component, Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand: Colin Ohrt, MD, George Watt, MD, DTM&H.

U.S. Naval Medical Research Unit 2, Jayapura and Jakarta, Indonesia: Thomas L. Richie, MD, PhD; Hendra Widjaja, MD; David J. Fryauff, ScD; Hasan Basri, MD; J. Kevin Baird, PhD; Purnomo, MSc; M. Awalludin Sutamihardja; Suradi; F. Stephen Wignall, MD.

Naval Medical Research Institute, Bethesda, Maryland: Stephen L. Hoffman, MD.

District Military Health Services, Jayapura, Irian Jaya, Indonesia: Januar Fitriadi, MD; Budi Santoso, MD; Lukas Hadiarso, MD.

Army Malaria Research Unit, Ingleburn, Australia: G. Dennis Shanks, MD.

Center for Health Research and Development, National Institutes of Health, Jakarta, Indonesia: Emiliana Tjitra, MD, MSc; Suriadi Gunawan, MD, DPH; Harijani Marwoto, MSc; Sri Oemijati, MD, DR; and P.R. Arbani, MD, MPH.

Battalions 143 and 731, Indonesian Army, Irian Jaya, Indonesia: Wendy Budiawan, MD; Bambang Sukanto, MD.

F. Hoffmann-La Roche Ltd., Basel, Switzerland: Jurg Handschin, PhD; Dieter Sturchler, MD.

Provincial Health Service, Jayapura, Indonesia: Bernardus Sandjaja, MD; Slamet Harjosuwarno, MD, MPH; Budi Subianto, MD.

Walter Reed Army Institute of Research, Washington, DC: Douglas Tang, PhD.

Drs. Richie, Widjaja, and Fryauff: U.S. Embassy, Attn: U.S. NAMRU-2/(PERSON'S NAME), Box 3 Unit 8132, APO AP 96520-8132, United States.

Dr. Fitriadi: Office of Military Command VIII/Trikora, (KESDAM VIII Trikora), Jalur Gura Besi, Jayapura 99111, Irian Jaya, Indonesia.

Dr. Hadiarso: Sub-Directorate of Preventive Medicine, The Indonesian Army Directorate of Health, Jalur Abdul Rahman Saleh No. 20, Jakarta Pusat, Indonesia.

Dr. Tjitra: Infectious Diseases Research Center, National Institute of Health Research Division, Ministry of Health, Jalur Percetakan Negara no. 29, Jakarta 10560, Indonesia.

Dr. Shanks: U.S. Army Medical Research Unit, Unit 64109 Nirobi, APO AE 09831-5000.

Dr. Handschin: Pharma Business Development and Strategic Marketing, F. Hoffmann-La Roche Ltd, Pharmacueticals Division, CH-4070 Basel, Switzerland.

Dr. Sandjaja: Health Training Center (Balai Pelatihan Kesehata), Jalur Kesehatan No. 1, Abepura, Jayapura 99351, Irian Jaya, Indonesia.

Dr. Tang: Division of Biometrics, Department of Biostatistics, Walter Reed Army Institute of Research, Walter Reed Army Medical Center, Washington, DC 20307-5100.

Dr. Watt: U.S. Army Medical Component, Armed Forces Research Institute of Medical Science, APO AP 96546, United States.

Dr. Wignall: Freeport Malaria Control, PO Box 61982, New Orleans, LA 70161-1982.