Introduction

Approximately one in three adults aged 65 years or over falls at least once a year [1]. With the rapidly aging society, falls among the elderly have become a critical health issue [2]. Among older adults, falling is a leading cause of unintentional injuries and accounts for two thirds of the deaths resulting from unintentional injuries. It is estimated that approximately 10–15% of people who fall suffer serious consequences, such as fractures [3]. Falls in the elderly patients could cause fractures of the hip, wrist, pelvis, proximal humerus, ankle, and elbow [4].

Many studies have identified risk factors of falling. One of the causes is pharmacotherapy, which can increase the risk of injurious falls and fracture via sedation, orthostatic reactions, balance disorder, muscle weakness, cognitive impairment, and/or osteoporosis [5]. In addition, drug clearance decreases as people become older due to the aging process, which leads to prolonged adverse effects [6]. The evidence of the relation between anticholinergic drugs (e.g., tricyclic antidepressants, proton pump inhibitors, and antiparkinsonian drugs) and the risk of injurious falls and fracture has been established by meta-analysis [7,8,9]. As far as we know, however, there is no systematic literature review or meta-analysis regarding the risk of injurious falls or fracture in relation to antihistamine use.

H1-antihistamines are drugs that block the H1-histamine receptors in the body [10]. They have been commonly used to treat histamine-mediated allergic reactions such as allergic rhinitis, atopic dermatitis, and urticaria since the 1940s [11, 12]. Generally, H1-antihistamines are categorized into two classes: first-generation antihistamines (FGAHs) and second-generation antihistamines (SGAHs). FGAHs have been widely used for decades, but they are likely to have adverse effects because they readily cross the blood–brain barrier (owing to high lipophilicity) and cross-react with other receptors [13]. As an alternative, SGAHs were developed in an attempt to reduce anticholinergic side effects associated with FGAHs.

According to the Screening Tool of Older Persons’ Prescriptions (STOPP) and Beers criteria, the two most frequently used instruments for assessing the suitability of prescription drug use in the elderly population, FGAHs are considered potentially inappropriate due to anticholinergic side effects [14, 15]. Nonetheless, the prevalence of FGAH prescriptions has remained high in clinical settings [16, 17]. Therefore, given the potential for adverse effects of FGAHs on the elderly population, a systematic review is needed for the synthesis of real-world evidence regarding the increased risk of injurious falls or fracture. The objective of this study was to investigate the association between FGAHs and injurious falls or fracture on the basis of published studies.

Materials and methods

The search strategy and data source

For this systematic review and meta-analysis, we followed the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines and the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [18, 19]. We developed a protocol in advance to specify the objective, outcome, eligibility criteria, search strategy, methods for study selection, data extraction, and data synthesis for this meta-analysis. Before conducting the literature search, we defined a structured research question following the Population, Intervention, Comparison, Outcome (PICO) format: “In elderly patients (Population), does taking first-generation antihistamines (Intervention), compared to not taking them (Comparison), increase the risk of falls or fracture (Outcome)?”

We searched two core databases (MEDLINE and EMBASE) and five Korean core databases (KoreaMed, KMbase, KISS, NDSL, and KiSTi) published by November 14, 2016 (the date of the last search) to identify studies on the association between FGAH use and the risk of injurious falls or fracture in the elderly population. The following search terms were used: “aged” and “elderly,” for population; “antihistamine*,” “histamine H1 antagonists,” “histamine H1 blockers,” “brompheniramine,” “carbinoxamine,” “chlorpheniramine,” “clemastine,” “cyproheptadine,” “dexbrompheniramine,” “dexchlorpheniramine,” “dimenhydrinate,” “diphenhydramine,” “doxylamine,” “hydroxyzine,” “meclizine,” “promethazine,” and “triprolidine” for intervention; “fall*,” “fracture*,” “accidental falls,” and “fractures, bone” for outcome. Search terms belonging to each group (population, intervention, and outcome) were co6mbined using “OR,” while population, intervention, and outcome were combined via “AND.” The FGAH active ingredients listed above were derived from 2012 Beers criteria.

Study selection

Two reviewers (H.C. and J.M.) independently screened the titles and abstracts for the eligibility criteria. Subsequently, the full text of the studies that potentially met the inclusion criteria was evaluated to reach a decision on final inclusion. Any disagreements were resolved by reaching a consensus through mutual discussion or consulting a third reviewer (H.-Y.K.).

Studies were included if they met the following eligibility criteria: (1) they presented original data from observational studies (e.g., a cohort study, case-control study, or case-crossover study); (2) the population of interest was the elderly population aged 65 years or older; (3) the outcome of interest was clearly defined as injurious falls or fracture; (4) the treatment of interest was antihistamine medication; (5) they reported the odds ratio (OR), the relative risk (RR), or the hazard ratio (HR) of injurious falls or fracture associated with the use of FGAHs, and the corresponding 95% confidence interval (CI). We also included studies that did not clearly provide information on the generation of antihistamines. Studies were excluded if both reviewers agreed that they did not meet the eligibility criteria. Nonetheless, studies were excluded if they were not specific to antihistamine use (e.g., anticholinergic drugs), even though they presented original data. Review articles, editorials, case series, conference abstracts, and letters to the editor were not included. Duplicate studies and studies for which we were not able to obtain full-text articles were excluded.

Data extraction and quality assessment

Two independent reviewers (H.C. and J.M.) extracted data from the selected articles on the following items: the name of the first author, the year of publication, the country where the study was conducted, study design, study population and baseline characteristics, exposure, the type of outcome, risk estimates (OR, RR, or HR and its 95% CI), and adjusted covariates.

The risk of bias in each observational study was evaluated using the Risk of Bias Assessment Tool for Nonrandomized Studies (RoBANS) version 2.0, which showed moderate reliability, promising feasibility, and validity [20]. RoBANS version 2.0 consists of the following six domains: selection of the participants, confounding variables, measurement of exposure, blinding of outcome assessments, incomplete outcome data, and selective outcome reporting. Two reviewers (H.C. and J.M.) independently categorized the quality of evidence into four levels (very low, low, moderate, or high) in accordance with the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) method [21]. Any controversy was resolved by discussion.

The main analysis and subgroup analysis

We investigated the association between the use of FGAHs and the risk of injurious falls or fracture by means of adjusted risk estimates, if available. Due to a lack of studies, we considered “injurious falls or fracture” one combined outcome to conduct the meta-analysis. To understand the underlying clinical and methodological heterogeneity across the studies, we performed subgroup analysis by type of study design, adjustment for any confounder, and generation of antihistamines. We performed the test for subgroup differences available in the Review Manager 5.3 software to determine whether the results for subgroups were significantly different.

Statistical analysis

A generic inverse-variance method was employed to pool the ORs of injurious falls or fracture associated with the use of FGAHs and the corresponding 95% CIs [22]. The random-effects model was chosen instead of the fixed-effect model, considering the high likelihood of between-study variance and the likelihood of producing more conservative estimates.

Heterogeneity was assessed statistically by means of Cochran’s Q statistic and Higgins I2 statistic [23, 24]. Statistical significance was set to the 10% level owing to low power of the test. The magnitude of heterogeneity was interpreted as not important (0–30%), moderate (30–50%), substantial (50–70%), or considerable (70–100%).

All analyses were conducted in the Review Manager 5.3 software (London, UK) from the Cochrane Collaboration. Data with a two-sided p value of less than 0.05 were considered significant.

Results

Search results

Figure 1 shows the strategy used to identify the relevant studies for the meta-analysis. After the search strategy was applied, 473 potentially relevant articles were identified in our initial literature search. This number was reduced to 411 articles after excluding duplication. After screening of the titles and abstracts, 371 studies were excluded, mainly because they were not relevant to our study objective. Three studies published before 1995 were excluded because we could not obtain full-text articles. After full-text assessment, five studies satisfied the eligibility criteria [25,26,27,28,29]. No additional studies were identified in our manual search of references in published articles.

Fig. 1
figure 1

The flowchart for identification of relevant studies

Full size image

Study characteristics

Table 1 summarizes the main characteristics of the five studies that were included in the analysis. They were published between 2011 and 2016. Among them, three were case-control studies [25,26,27], one was a cohort study [29], and one article was a case-crossover study [28]. The data sources of three studies were medical records [26,27,28], a patient-safety reporting system with computerized medical records [25], and national veterans affairs administrative data [29]. Geographically, four studies were from Asia [25,26,27,28], and one from the USA [29]. Three studies defined their outcome of interest as falls [25, 27, 28], one study as hip fracture [26], and one study as falls or fracture [29]. In three studies, results were adjusted for potential confounders [26, 27, 29]. As to the generation of antihistamines, two studies included FGAHs [26, 29], and three studies did not clearly define the generation of antihistamines [25, 27, 28]. As for the eligible studies that did not provide information on the generation of antihistamines, we tried to contact the authors for further information. Lee (2011) responded that all generations of antihistamines were used in the study [27]. Although we could not acquire relevant information from the other authors, we did not exclude these studies for the following reason: they deal with medication that causes sedation and drowsiness, implying high probability of FGAH [25, 27, 28]. The risk of bias analysis revealed little concern, aside from two studies where the key covariates were not adjusted for confounding variables [25, 28]. We conducted subgroup analysis because of this concern to make an unbiased interpretation. The quality of evidence was moderate according to the GRADE method.

Table 1 Characteristics of five studies included in the meta-analysis
Full size table

The main analysis and subgroup analysis

Five studies were included in the analysis, and the results showed increased risk of injurious falls or fracture in patients exposed to antihistamines without heterogeneity among the studies (OR 2.22, 95% CI 1.70–2.90, heterogeneity: p = 0.45, I2 = 0%). Although five studies were finally selected, we included only two studies in the main analysis that clearly focused on FGAHs [26, 29]. Figure 2 shows the pooled estimate of the two included studies assessing the risk of injurious falls or fracture relevant to the use of FGAHs. This use considerably correlated with the risk of injurious falls or fracture without heterogeneity (OR 2.03, 95% CI 1.49–2.76, heterogeneity: p = 0.41, I2 = 0%).

Fig. 2
figure 2

Antihistamine use and the risk of injurious falls or fracture according to the random-effects model. CI confidence interval, IV inverse variance, SE standard error.

Full size image

On the other hand, the results on all generations of antihistamines or without a clear mention of the generation of antihistamines in terms of injurious falls or fracture were also significant without heterogeneity (OR 2.89, 95% CI 1.71–4.89, heterogeneity: p = 0.42, I2 = 0%). Although no statistical heterogeneity was detected, we conducted subgroup analysis according to the type of study design, adjustment for any confounder, and the generation of antihistamines to assess potential clinical and methodological heterogeneity (Table 2). Overall, an increased risk of injurious falls or fracture associated with the use of antihistamines was observed in all subgroups. There was a tendency among case-control studies to be associated with a higher OR (OR 3.79, 95% CI 1.80–7.96, heterogeneity: p = 0.41, I2 = 0%) in comparison with the case-crossover study (OR 2.20, 95% CI 1.04–4.64, heterogeneity: not applicable). Two studies on antihistamines not clearly mentioning the generation and not adjusted for a potential confounder seemed to show lower risk estimates of injurious falls or fracture (OR 2.49, 95% CI 1.39–4.44, heterogeneity: p = 0.61, I2 = 0%) than did a study on all-generation antihistamines that was adjusted for a potential confounder (OR 5.80, 95% CI 1.67–20.18, heterogeneity: not applicable). Nevertheless, the test for subgroup differences did not yield statistically significant results.

Table 2 Subgroup analysis for the use of antihistamines and the risk of injurious falls or fracture
Full size table

Discussion and conclusions

We conducted a meta-analysis to examine the association between FGAH use and the risk of injurious falls or fracture on the basis of all eligible studies. In the main analysis, we obtained clear evidence that the relative risk of injurious falls or fracture is considerably increased among FGAH users (OR 2.03, 95% CI 1.49–2.76, heterogeneity: p = 0.41, I2 = 0%). Our results support other meta-analyses (regarding anticholinergic drugs), which have shown an increased risk of injurious falls and fracture associated with anticholinergic medication such as tricyclic antidepressants, proton pump inhibitors, and antiparkinsonian drugs [7,8,9].

On the other hand, there were two other studies that did not clearly indicate the generation of antihistamines, and one study that included all generations of antihistamines. The results of these studies showed higher OR without statistical heterogeneity (OR 2.89, 95% CI 1.71–4.89, heterogeneity: p = 0.42, I2 = 0%) compared to the main analysis. It is notable that the study population was hospitalized patients in these articles [25, 27, 28], whereas that of the main analysis was mostly outpatients. In addition, the outcome of these studies was injurious falls, whereas that of the main analysis included fracture that resulted from injurious falls [4]. The difference in severity in terms of study population and outcome measurement may explain why the main analysis showed lower OR than that of the other articles.

Four studies were conducted in Asian countries, and each study presented increased risk of injurious falls or fracture among antihistamine users. According to Dhanwal et al., who analyzed geographical variation of hip fractures, it is estimated that a half of hip fractures will occur in Asia in 2050 [30]. Appropriate interventions that are intended to reduce potentially inappropriate antihistamine prescribing may help to lower the incidence rate of hip fracture. It is notable that three of the included articles deal with falls of hospitalized patients in a hospital. Careful monitoring is needed when elderly patients are prescribed antihistamines because falls among elderly patients are becoming more prevalent in hospitals [31].

Our study has several potential limitations. First, only a few studies were included in the main analysis owing to limited information about the generation of antihistamines in the other included studies. Because SGAHs are known to have smaller anticholinergic side effects, thus possibly resulting in biased estimates, we excluded studies including all generations of antihistamines or not clearly mentioning the generation of antihistamines from the main analysis. Therefore, to establish a more robust relation between FGAHs and injurious falls or fracture, more empirical research is required. Second, we combined two different conditions—injurious falls and fracture—for the same outcome variable because there were relatively few eligible studies. As a result, clinical heterogeneity may be present. Nevertheless, these outcomes are not completely different, given that among older people, fracture is frequently caused by falls [4]. Especially, hip fracture is mostly a consequence of falls [32, 33]. Third, the study population varied among the studies in terms of gender distribution. Lee et al. (2016) mostly included female elderly patients, whereas the study by Alvarez et al. (2015) predominantly included elderly male patients [26, 29]. Fourth, it was not confirmed whether appropriate statistical methods (conditional logistic regression) were used in two matched case-control studies [26, 27]. Fifth, publication bias was not assessed by means of a funnel plot. Because of the relatively small number of studies, publication bias tests may be inappropriate [34]. Lastly, we compared FGAH users with non-FGAH users in the present study because there was no study comparing FGAH users versus SGAH users. Further real-world empirical research is needed to determine the impact of FGAHs and to confirm the safety of SGAHs. Despite these limitations, to the best of our knowledge, this study is the first systematic review and meta-analysis to explore the correlation between antihistamine use and the risk of injurious falls or fracture on the basis of real-world evidence.

In conclusion, our findings revealed that FGAH use is considerably associated with an increased risk of injurious falls or fracture as compared to nonuse of FGAH. Clinicians need to exercise caution when prescribing FGAHs to elderly patients. We expect that our findings will provide evidence supporting Beers and STOPP criteria that discourage prescription of FGAHs to elderly patients.