Risk Associated with Bee Venom Therapy: A Systematic Review and Meta-Analysis

Objective The safety of bee venom as a therapeutic compound has been extensively studied, resulting in the identification of potential adverse events, which range from trivial skin reactions that usually resolve over several days to life-threating severe immunological responses such as anaphylaxis. In this systematic review, we provide a summary of the types and prevalence of adverse events associated with bee venom therapy. Methods We searched the literature using 12 databases from their inception to June 2014, without language restrictions. We included all types of clinical studies in which bee venom was used as a key intervention and adverse events that may have been causally related to bee venom therapy were reported. Results A total of 145 studies, including 20 randomized controlled trials, 79 audits and cohort studies, 33 single-case studies, and 13 case series, were evaluated in this review. The median frequency of patients who experienced adverse events related to venom immunotherapy was 28.87% (interquartile range, 14.57–39.74) in the audit studies. Compared with normal saline injection, bee venom acupuncture showed a 261% increased relative risk for the occurrence of adverse events (relative risk, 3.61; 95% confidence interval, 2.10 to 6.20) in the randomized controlled trials, which might be overestimated or underestimated owing to the poor reporting quality of the included studies. Conclusions Adverse events related to bee venom therapy are frequent; therefore, practitioners of bee venom therapy should be cautious when applying it in daily clinical practice, and the practitioner’s education and qualifications regarding the use of bee venom therapy should be ensured.


Introduction
Bee venom is one of the most commonly encountered animal venoms and consists of various chemical agents that induce allergic reactions in the human body [1]. Bee venom therapy (BVT), in which bee venom is used for medicinal purposes, is available worldwide, but is primarily utilized in Asia, Eastern Europe, and South America [2]. The diverse therapeutic applications of BVT include various musculoskeletal conditions, such as arthritis and rheumatism, chronic recalcitrant neuralgia, arthralgia, and immune-related diseases. BVT is also used to desensitize patients to bee stings and thus inhibit allergic reactions [3] [4] [5].
Although the therapeutic utility of bee venom has been demonstrated, its safety profile is an important limiting consideration, because immune responses to BVT can range from trivial skin reactions that resolve over several days to life-threatening responses such as anaphylaxis [6] [7]. In a recent survey, the incidence of systematic reactions (SRs) in patients who received venom and inhaled-allergen subcutaneous immunotherapy was 13.60%, whereas the prevalence of SRs in patients that received bee venom immunotherapy (VIT) was 28.72% [8]. In another survey, 12.13% patients who received VIT experienced SRs (an average of 1.91 SR events per subject), suggesting that serious adverse events (SAEs) due to BVT are quite common [9].
The most significant issue related to the AEs of BVT is that the occurrence of SAEs is unpredictable. It is therefore necessary to determine the prevalence and nature of AEs related to various types of BVT, so that bee venom can be used safely in clinical practice. The aim of this systematic review was to provide summary information regarding the types of AEs related to BVT and their prevalence in treated patients.

Study selection
Types of studies. All types of clinical studies, including randomized controlled trials (RCTs) and randomized crossover trials, as well as observational studies, including cohort studies, case-control studies, case series, and case studies, were included in this study.
Types of participants. The subjects of the studies evaluated in this review included adults and pediatric patients, and the selection was not limited to studies of patients with specific diseases. Regardless of a patient's condition and disease status, studies were included if bee venom was used as a key intervention, and AEs that may have been causally related to BVT were reported.
Types of interventions. In this review, we included studies of bee sting acupuncture (BSA), a subcutaneous or intramuscular injection of bee venom for the purpose of acupoint stimulation (bee venom acupuncture [BVA], sweet bee venom [SBV]), and dried honeybee venom (apitoxin injections), as well as subcutaneous VIT for desensitization of venom immune reactions. BSA, BVA, SBV, and apitoxin injections usually involve the use of venom derived from bees (family Apidae), whereas VIT generally involves the use of venom from bees (family Apidae) and wasps (family Vespidae) concomitantly. Therefore, we included all types of venom therapy including both bee and wasp venoms. We also reviewed studies where bee venom was used alone or in combination with other treatments. However, studies describing bee stings resulting from random encounters (e.g., during resting or by attack), sting challenge tests, sublingual VIT, and irrelevant venom types were excluded from this study. We included RCTs comparing BVT with no treatment, normal saline injections, and conventional medications for relative risk assessment. Trials in which different types of BVT were compared with each other were excluded.
Types of outcome measures. The major aim of this review was to identify the frequency and types of AEs related to BVT. In case studies and case review series, the type of AE was classified into 1 of 3 categories: SR, skin problem (SP), and other (nonspecific reaction, symptom, or sign that was not an SR or SP). If an SR occurred as an AE, it was classified into 1 of 5 categories based on the Mueller classification (grade I, grade II, grade III, and grade IV) [10]. The causal relationship between BVT and AEs was also assessed in each study according to the WHO-UMC causality scale [11]. AEs were scored as certain when they clearly occurred after BVT, disappeared after withdrawal, and could not be explained by other diseases or treatments. AEs were scored as probable when the timing of the AEs and BVT indicated that they were most likely related, they disappeared as a probable result of the discontinuation of BVT, and the events were not induced by other diseases or treatment. AEs were scored as possible when they occurred after BVT treatment but no information was available on the relationship between their disappearance and the withdrawal of BVT and when they could potentially be explained by other diseases or treatments. In addition, AEs were scored as unlikely when the event and the BVT had an improbable causal relationship. AEs were scored as conditional/unclassified when the event occurred but more data were necessary for a conclusion to be reached. Finally, AEs were scored as unassessable/unclassifiable when they could not be evaluated properly owing to insufficient and/or contradictory information [12].
In audits and cohort studies, AE types were divided into SR, large local reaction (LLR), local reaction (LR), and other (nonspecific reaction, symptom, or sign that was not an SR, LLR, or LR). An LLR was defined as swelling exceeding 10 cm in diameter and lasting longer than 24 h, and an LR was defined as local pruritus, edema, or erythema [13]. Finally, the prevalence of AEs related to BVT was assessed through observational studies, including audits and cohort studies.

Data sources
The following 12 databases were searched: PubMed, EMBASE, the Cochrane Library, CINAHL, China National Knowledge Infrastructure (CNKI), Wanfang (China), Weipu (China), KoreaMED, the Korean Medical Database (KMBASE), the Korean Studies Information Service System (KISS), National Discovery for Science Leaders (NDSL) (Korea), and the Oriental Medicine Advanced Searching Integrated System (OASIS) (Korea). Bibliographic references in relevant publications (Journal of Pharmacopuncture) were manually searched to avoid missing eligible articles. The References sections of reviews on AEs of BVT were searched manually, and articles published through June 2014 were included. The search terms consisted of two parts: "BVT" (e.g., bee sting, apitoxin, or venom immunotherapy) and "adverse events" (e.g., adverse reaction, side effects, risk, or safe

Data collection and analysis
Study selection. Two independent reviewers (JHP and BKY) screened the articles for inclusion by title and abstract. If disagreements regarding the selection of a study could not be resolved through discussion, the final decision was made by the arbiter (THK).
Data extraction and management. One reviewer (JHP) read the full text of the articles selected for review and extracted the data using a standard data extraction form. Another reviewer (BKY) rechecked the data to ensure that it had been extracted appropriately. Any disagreement among the reviewers was resolved by discussion or by the arbiter (THK).

Quality assessment of AEs in RCTs.
To evaluate the quality of the detection and reporting of the AEs in the included RCTs, 7 items were assessed according to the CONSORT recommendations for harm data: (1) mention of AEs in the title or abstract, (2) mention of BVTrelated AEs in the introduction, (3) predefined definition of AEs related to BVT, (4) collection or monitoring method for AEs, (5) mention of the method for analyzing and presenting AEs, (6) mention of any patients who dropped out of the study owing to AEs, and (7) mention of the specific denominator for the analysis of AEs [14] [15]. The quality of each item was judged as good, moderate, bad, or not reported [12]. The quality of a study was scored as good if each item was reported clearly in the manuscript or in the registered protocol. If each item was reported, but not in detail, the methodological quality was scored as moderate. The quality of a study was scored as bad when any of the items were not appropriately reported. If an item was not described at all, it was recorded as not reported.

Statistical analysis
A meta-analysis of the RCTs was conducted if the incidence of AEs was clearly reported and the relative risk of AEs could be assessed because of similar study designs and intervention methods, including BVT types and control interventions, with minimal clinical heterogeneity. The relative risk of BVT and control interventions was assessed, and effects were calculated using Revman 5.2 software (http://ims.cochrane.org/revman).

Results
Through electronic and manual searching, 8,108 potentially relevant articles were identified, including 5,504 records from PubMed, EMBASE, the Cochrane Library, and CINAHL; 468 records from the Chinese databases; and 2,136 records from the Korean databases, from which 2,118 duplicate records were removed. Through a screening process involving the use of the titles and abstracts of identified records, we excluded 5,699 records that did not meet the inclusion criteria. The remaining 291 articles were reviewed for eligibility, and 146 articles were excluded, including experimental studies (32), reviews (57), surveys (3), studies without description of the assessment of AEs (43), and studies without relevant intervention or comparison groups (11). Finally, 145 studies, including 20 RCTs, 79 audits and cohort studies, 33 single-case studies, and 13 case series, were included in the review (Fig 1).

Case studies and case series
Thirty-three single-case studies and 13 case series were identified as described in Table 1 [2,7,. A total of 69 individual isolated cases were reported in 46 papers. Incidents were reported in 11 countries: Korea (37 cases), China (10 cases), the United States (7 cases), France (6 cases), Germany (2 cases), Turkey (2 cases), Canada (1 case), Italy (1 case), Russia (1 case), Saudi Arabia (1 case), and the Slovak Republic (1 case). The reported BVT methods included BSA (29), BVA (21), and VIT (19). Among the 69 AE cases, 58 cases were related to BVT, 6 cases were related to wasp venom treatment, and 5 cases were related to treatment with a mixture of bee venom and wasp venom. Among the 58 AE cases related to treatment with bee venom only, 30 SRs, 23 SPs, and 5 other cases, including cough; headache; uremia; anorexia; discoloration of the sclera; jaundice; painful cyclic uterine contractions; severe pain affecting the left shoulder, chest wall, and left arm; and muscular weakness in the left arm and hand, were reported. The 30 SRs related to treatment with bee venom only were classified as grade I (5 cases), grade II (10 cases), grade III (14 cases), and grade IV (1 case). The severity of the AEs related to BVT only were reported as moderate (34 cases) or severe (24 cases), and the causality was deemed to be probable for 49 cases and possible for 9 cases. Most practitioners were qualified practitioners (30 cases), and 4 patients were treated by unqualified personal with no medical training or licensure regarding BSA. One patient died after treatment by an unqualified BSA practitioner. In 23 cases, there was no description of the practitioner. A pre-treatment skin test for venom allergies was reported in only 10 cases, and it was almost always performed

Audits and cohort studies
AEs were also reported in 79 mainly retrospective audit studies that aimed to assess the safety of BVT (Table 2) . These studies were chiefly observational and included case-controlled and cohort studies. VIT (63 studies) was the most commonly used BVT method, followed by BSA (9 studies) and BVA (7 studies). The treatment protocol for VITs included conventional VIT, cluster VIT, rush VIT, ultra-rush VIT, specific immunotherapy, and rushspecific immunotherapy. Eleven studies were conducted in Spain, 10

RCTs and randomized crossover trials
Eighteen RCTs and 2 randomized crossover trials were included in this review (Table 3) [139][140][141][142][143][144][145][146][147][148][149][150][151][152][153][154][155][156][157][158]. One-hundred and forty-eight AEs related to BVT were reported in 397 participants. Seventeen patients ended their study participation owing to BVT-related AEs. For the BSA and BVA studies, all of the participants who were negative for skin allergy tests were included in the studies. With regard to the quality of the reporting of AEs, more than half of the items in the CONSORT AE reporting guidelines were not reported (52.14%). Most RCTs did not report the AEs in the title, abstract, or introduction, or report definitions of AEs and mention the methods for analyzing and presenting AEs. In 9 studies, the collecting and monitoring method for AEs involved retrospectively checking with the physician and/or participant, and the monitoring methods of 7 studies were not reported appropriately. Most studies reported the number of patients who stopped participating, as well as the specific denominator for the analysis of BVT-related AEs.
The meta-analysis of AE occurrence in the 4 RCTs assessing patients experiencing AEs showed that BVA increased the risk of AEs by 261% compared to the risk associated with normal saline control treatment (relative risk, 3.61; 95% CI [2.10, 6.20] , Fig 2).

Discussion
The aim of our systematic review was to summarize the evidence pertaining to BVT-related AEs by analyzing AE types and their prevalence in patients. We reviewed 145 studies, including 20 RCTs and randomized crossover studies, 79 audits and cohort studies, 33 single-case studies, and 13 case series. According to our findings, BVT can lead to AEs such as SRs, LLRs, LRs, SPs, and nonspecific reactions, some of which are serious.
In case studies and case series, we found that SRs comprised 51.72% of the AEs produced by bee venom. Moreover, the identified severe AEs included 14 cases of grade III SR and 1 case of grade IV SR (50.00% of the total SRs). We also found that there have been SAEs associated with BVT that urgently required subcutaneous adrenaline or steroid and oxygen therapy, with death occurring in 1 case [22] [51] [58]. Aside from SRs, AEs associated with BSA and BVA   mainly include SPs such as granulomas and plaques, which may be attributable to persistent local inflammation caused by venomic components or from the remaining stinger at the site of cutaneous injection [29]. In contrast, SRs resulting from BVT are mainly derived from anaphylaxis, hypersensitivity, and late-onset reactions [3] [58]. In 46 audits and cohort studies of VIT, the median incidence of AEs was 28.87%, and SRs occurred in 681/4844 (14.06%) participants. These results suggest a more frequent AE incidence in comparison with that in previous systematic reviews of VIT, which reported SR incidences of 11.5 to 11.8% [159] [160]. Interestingly, some studies found a complete lack of AEs related to BVT and a corresponding lack of SRs, and some studies have shown minor AEs, but no serious SRs [60] [117,118] [134].
Skin tests allow practitioners to distinguish whether BVT is an appropriate intervention for particular patients. In most RCTs and randomized crossover trials with BSA and BVA, participants were included if they showed negative responses in skin tests, whereas participants were included in VIT case studies and case series if they showed positive responses in skin tests. This difference in the participants included in each type of study does not seem to be directly related to the AEs associated with BVT; negative venom skin test results are not always a guarantee of VIT safety [94]. However, serious AEs can occur as a result of BSA and BVA in patients with positive skins tests. There is a report of a young, healthy adult who was sensitized to bee venom through BSA, and who was later stung by a bee and developed severe, life-threatening anaphylaxis [161].
Venom concentration and the frequency of venom administration can influence the severity and rate of incidence of AEs resulting from BSA and BVA. Unfortunately, we could not analyze the effect of venom concentration and administration frequency on the severity and rate of incidence of AEs because only limited numbers of RCTs were included in this review.
With regard to the quality of reporting of AEs in RCTs, CONSORT items were generally not reported properly. Future RCTs with BVT must adopt the CONSORT AE reporting guidelines to ensure transparency and accuracy. When designing protocols, methods of AE assessment based on the CONSORT AE reporting guidelines should be suggested in detail.
AEs related to BVA or VIT have been reported in various studies, including surveys [8] [162] [163] and reviews [159] [160] [164] [165]. However, in this paper, we extensively reviewed all types of BVT (BSA, BVA, SBV, apitoxin injection, and VIT). We focused on the incidence of AEs in audit and cohort studies related to BVT, and sought to provide an overview of the many types of AEs that were reported in case studies and case series. We performed this investigation through a comprehensive search of the literature.
This review has some limitations. First, the heterogeneity of intervention in the reviewed articles was high; thus, the exact AE incidence and risk associated with the treatment methods could not be calculated. Second, although different venoms were used in different therapies (bee [family Apidae] venom was mainly used in BSA and BVA, whereas venom of both bees [family Apidae] and wasps [family Vespidae] was used in VIT), AEs from VIT were not classified in terms of the type of venom, treatment protocol (conventional VIT, cluster VIT, rush VIT, ultra-rush VIT, etc.), or phase (induction and maintenance).
While it is evident that BVA clearly increases the risk of AEs in comparison with normal saline, our review revealed that BSA and BVA are often implemented without a skin test, and also showed that patients have experienced SAEs that can be fatal after receiving BSA from unqualified personnel. Therefore, in order to enhance the safety of BVT, a skin test should be conducted before BVT is administered, and the venom should be administered only by qualified individuals [166].
Based on the results of this review, several suggestions can be made to support effective clinical practice and future clinical trials with BVT. In order to support responsible use of BVT, educational materials on the safety and efficacy of BVT should be made available for patients. Moreover, practitioners should be aware of the various AEs associated with BVT, establish clinical guidelines to minimize the development of AEs, and develop and implement strict criteria for monitoring AEs once they occur.

Conclusion
AEs related to BVT are not uncommon. Therefore, BVT practitioners should pay careful attention to the incidence of AEs and patterns of AE occurrence in their patients. Additionally, education and qualification of BVT practitioners should be ensured based on appropriate training programs and clinical guidelines for monitoring of AEs related to BVA and BSA. Furthermore, when reporting AEs in RCTs evaluating BVT, researchers should describe AEs in detail according to the CONSORT recommendation for harm data to ensure transparency and accuracy.