Our study included more participants than a previous meta-analysis [19] (3,766 vs. 1,082), and strengthened the evidence that prophylactic antibiotic use is effective for the prevention of Lyme disease following a tick bite. Furthermore, our subgroup analysis revealed that patients who received a single dose (200 mg) course were shown to be less likely to develop Lyme disease than those given placebo (RR, 0.29 [95%CI: 0.14–0.60]), but there is no evidence of the effectiveness of a 10-day course and topical antibiotics course (RR, 0.28 [95%CI: 0.05–1.67] and 0.73 [95%CI: 0.25–2.08]), respectively. Our results support the strategy of a single-dose oral doxycycline therapy for prevention of Lyme disease.
As early as 2001, Nadelman et al. assessed the effect of doxycycline in the prevention of Lyme disease. However, the effectiveness estimated in the RCTs showed a wide confidence interval (RR = 0.13 [0.02–1.04]) [23]. Until recently, an RCT [9] with a relatively large sample size (n = 1,089) provided stronger evidence that a single dose of doxycycline can prevent the development of Lyme disease, following a bite from Ixodes ricinus (RR = 0.33 [0.15–0.70]). Our meta-analysis combined two RCTs and showed a more accurate CI (RR = 0.29 (0.14–0.60]). Additionally, two observational studies reported the results of doxycycline in the prevention of Lyme disease. Korenberg et al. [24] reported that none of the patients in the experimental group (n = 261) developed erythema migrans after receiving doxycycline (100 mg twice daily) for 3–5 days after the tick bite, whereas 5/97 patients developed erythema migrans in the control group which did not receive any antibiotics. Jackson et al. [25] reported the clinical application of doxycycline for Lyme disease prophylaxis, and the results indicated a high level of satisfaction with the pharmacy services provided, with no reports of subsequent development of Lyme disease symptoms or other side effects. However, the sample size of this study was small (n = 8).
Although our results support the use of antibiotics for the prevention of Lyme disease and the advantages of a single dose of doxycycline, routine use of antibiotic prophylaxis is not recommended after a recognized tick bite [17]. In our meta-analysis, we estimated that 50 patients (95%CI: 25–100) would need to be treated (NNT) with single-dose doxycycline to prevent one case of Lyme disease. Therefore, it is essential to determine who is at high risk of infection and who is worthy of treatment. For instance, animal studies have shown an exponential increase in the risk of B. burgdorferi infection after 48–72 h of deer tick attachment [26, 27]. Consequently, guidelines state that a tick bite is considered to be high-risk only if it was attached for more than 36 h [16]. Falco et al. reported that 52.5% of all tick bites had been attached for < 36 h [28], so the recommendation represents that nearly half of patients avoid receive antibiotics treatment. Additionally, Nadelman et al. [23] found that ticks which were partially engorged with blood (with incidence rate of 9.9%), rather than unfed ticks (incidence rate of 0%), were associated with the development of erythema migrans. Nadelman et al. [23] found that erythema migrans developed more frequently after bites from nymphal ticks than after bites from adult ticks, with an incidence rate of 5.6% and 0%, respectively. Harms et al. [9] revealed that 11.1% untreated patients with a B. burgdorferi-positive tick bite developed Lyme disease, and the NNT in this subgroup was only 10. These findings might provide valuable information for clinicians, but need further confirmation.
Antibiotic use has some side effects [13]. The major side effects of oral doxycycline include enterocolitis, anaphylaxis (including angioedema), Stevens–Johnson syndrome, severe urticarial reactions, and a lupus-like syndrome. Minor reactions of intravenous ceftriaxone include gastrointestinal symptoms of abdominal pain, nausea, vomiting, and diarrhea, and hypersensitivity reactions such as rash, pruritus, fever and chills, candidiasis, and local reactions at the injection site [14]. Although none of included studies reported serious side effects, there were still many minor side effects reported such as rash or nausea. The two studies revealed that incidence of mild side effects after using single-dose doxycycline was 5.9–30.1% [9, 23], which suggests that up to a third of patients are likely to suffer mild side effects. Furthermore, Nadelman et al. found that 18.2% of patients were recognized additional tick bites after enrollment, but during the 6-week study period, the participants needed repeated antibiotic prophylaxis, which would strongly increased the risk of side effects [23]. Availability of a universally acceptable and effective prophylactic agent with minimal side effects would be the ideal. Previous studies found that topical azithromycin was highly effective when applied topically at the sites of tick bites in mice [29, 30]. Although no effective evidences were found in human trials [3], a topical pharmacological prophylactic strategy is still worth exploring [31], given that minor side effects such as localized itching, redness, and dryness were reported only in 1.6% patients [3].
Our meta-analysis has some limitations. First, although we screened more than 4000 related articles, only six studies were eligible for final analysis. Second, we included 4 studies from USA and 2 studies from Europe, the difference of Ixodes species and B. burgdorferi subspecies between the USA and Europe may bring heterogeneity. Third, erythema migrans was considered as the main end point of all included studies evaluating antibiotic prophylaxis, since it is the most common clinical manifestation and only reliable marker of infection caused by B. burgdorferi infection. However, this end point was limited and could have resulted in underestimation of the actual incidence of B. burgdorferi infection. Fourth, Harm et al. study [9], which included 1689 participants, contributed 54.5% of the weight to the pooled results, but this study was the only one assessed as low quality (Jadad score 3). Therefore, our evidence is limited and further confirmation is needed. Last, we did not analyze the seroconversion results, because only few patients showed seroconversion even in the control group.