Benzodiazepine prescription and length of hospital stay at a Japanese university hospital

Benzodiazepines are among the most commonly used classes of medication because of their multiple therapeutic actions as anxiolytics, sedative hypnotics, anticonvulsants, and muscle relaxants [1, 2]. However, it has been reported that benzodiazepines are often prescribed in hospitals and for surgical patients in the absence of appropriate documentation of the indications for their use [3–7]. Benzodiazepine use during hospital stays is an important factor in the excessive long-term use of these drugs and can be harmful especially to the elderly, who have increased risks of falls and fractures [8, 9] and cognitive and memory changes [10, 11] that could dramatically decrease their quality of life. Thus it is important to prescribe benzodiazepines for inpatients appropriately, considering the balance of benefit and risk.

Although the relationship between bed days and benzodiazepine prescription has been studied in several Western countries, the number of subjects was relatively small (<1,000) [12–16], the analysis was limited to the elderly [12, 13, 17] or to those at limited departments [12, 14, 15], or the diagnostic information was not controlled for in the analyses [4, 18]. The findings are inconsistent among such Western studies, and no empirical study has been conducted on these issues in Japanese hospitals.

Thus we quantitatively clarified the relationship between benzodiazepine prescription and length of hospital stay at our Japanese university hospital, taking into consideration the effects of important clinical factors; patient age, sex, and diagnosis; in this study.

The number of inpatients prescribed benzodiazepine was 4,282 (19.9%). The percentage of patients prescribed benzodiazepine was significantly higher for women than for men, and the percentage increased with age (Table 1). When divided into 10-year age categories, the percentage of patients prescribed benzodiazepine tended to increase with age, both for men and women (Figure 1). The percentage of patients prescribed benzodiazepine by ICD-10 diagnosis is shown in Figure 2. When the specific three-digit codes of ICD-10 were analyzed (n > 100), benzodiazepine prescription was most common for patients diagnosed with chronic renal failure (N18, 31.3%), followed by heart failure (I50, 29.3%), acute myocardial infarction (I21, 28.9%), malignant neoplasm of the liver and intrahepatic bile ducts (C22, 28.0%), post-procedural disorders of the nervous system (G97, 28.0%), malignant neoplasm of the bronchus and lung (C34, 27.7%), non-insulin-dependent diabetes mellitus (E11, 25.0%), and malignant neoplasm of the bladder (C67, 22.2%). The percentage of patients prescribed benzodiazepine by department is shown in Figure 3. It was highest at the department of neurology (35.8%), followed by cardiac surgery (35.8%), orthopaedics (26.3%), general surgery (24.1%), internal medicine (23.7%), neurosurgery (20.3%), and dermatology (20.2%).

		N	%BDZ	Odds ratio	(95%C.I.)a
Age, years
	18 to 44	5,023	14.0	---
	45 to 64	6,028	19.8	1.51	(1.37, 1.68)
	65 to 74	5,355	22.7	1.80	(1.63, 2.00)
	75 or older	5,083	23.0	1.83	(1.65, 2.03)
Sex
	Male	11,844	18.6	---
	Female	9,645	21.6	1.21	(1.13, 1.29)
Number of bed days
	4 or shorter	5,938	9.1	---
	5 to 13	5,258	16.5	1.98	(1.76, 2.22)
	14 to 49	4,917	22.8	2.96	(2.65, 3.31)
	50 or longer	5,376	32.7	4.86	(4.37, 5.40)

^aThe variables age and sex from logistic regression analysis were used to estimate a crude odds ratio for benzodiazepine prescription.

For the number of bed days, multiple logistic regression analysis was used to estimate the odds ratio for benzodiazepine use, adjusted by age, sex, and ICD-10 diagnosis.

The arithmetic mean (standard deviation) for bed days was 29.6 (35.7) days, the geometric mean 13.9 (4.0) days, and the median value (range) was 13 (1-254) days for the total sample. The percentage of patients prescribed benzodiazepine tended to increase with the number of bed days. According to the results of multiple logistic regression analysis, benzodiazepine prescription was significantly associated with bed days, even after adjusting for the effects of the three independent variables; age, gender, and ICD-10 diagnosis (Table 1).

One-fifth of the inpatients had been prescribed benzodiazepines during their hospital stay. The elderly (≥ 65 years) were more likely to have a benzodiazepine prescribed compared to the non-elderly, as were women compared to men. Although the percentage of benzodiazepine prescribed varied by department and ICD-10 diagnosis, it was significantly associated with a higher number of bed days, even after adjusting for the effects of age, gender, and ICD-10 diagnosis.

The percentage for benzodiazepine prescription was higher than was found for outpatients in our previous study (12%) [19, 20], but was rather low when compared with Western studies of inpatients, which have reported prevalences of 33-45% [3, 15, 18]. A possible explanation for this difference is the relatively short length hospitalization in our study. When the analysis was limited to those with ≥ 50 bed days, the percentage given benzodiazepine prescriptions was equivalent to that in a report from France (33%) [15]. However, this fact does not necessarily weaken the generalizability of our results to Japanese inpatients, because the arithmetic mean of bed days is reported to be 20 days for Japanese general hospitals and 36 days for all Japanese hospitals according to a national survey by the Japanese Ministry of Health, Labor, and Welfare [25]. The average number of bed days in our hospital (= 30 days) was typical of Japanese hospitals.

Another possible explanation is the Japanese unwillingness to admit to psychiatric symptoms. The Japanese people have high thresholds for not visiting psychiatric clinics [26] and tend to complain of somatic symptoms, even if they suffer from mental illnesses such as depression [27, 28]. There may have been some patients in our study for whom psychological complaints were masked and who were followed up without anti-psychotropic regimens. It is reasonable that specific diseases such as musculoskeletal and nervous system diseases and the corresponding departments of neurology and orthopedics had high prevalences of benzodiazepine prescription because increased muscle tension is often accompanied by psychological distress, and benzodiazepines have a therapeutic effect as a muscle relaxant [1, 2, 29]. It is also reasonable that patients hospitalized in the department of cardiac surgery had a higher prevalence of benzodiazepine prescription because of cardiovascular symptoms due to their disease and the extremely anxious situation before heart surgery [30]. Although these implications are inconclusive, extensive intervention would be needed for inpatients with such disease conditions and further study needs to be done to assess their psychological distress and the degree of patient satisfaction. Through such assessment, important factors may be found for how to shorten the number of bed days for such patients.

Because an increased number of bed days has been reported to be closely related to the anxiety of inpatients with a wide rage of general medical disorders [10], it is possible that those staying at a hospital for a long time are likely to be anxious about their disease condition and that such worries may lead the patient seeking to use benzodiazepines. This was a cross-sectional study, and it is also possible that the side effects of benzodiazepines may contribute to extended hospital stays [31]. There may be room for reducing medical costs by targeting patients who have been hospitalized for a long time [32, 33]. Although the data are preliminary and not listed in the table, the percentage prescribed benzodiazepine was 10.2% for patients with hospital expenses of less than 2,000 US$ (n = 5,562), whereas it was 32.0% for those with expenses of 8,500 US$ or more (n = 5,169). For example, it may be good to routinely screen the mental condition of inpatients using a simple questionnaire before prescribing benzodiazepines and to use a consultation-liaison service with psychiatric specialists for the treatment of patients prescribed benzodiazepines [34]. Particularly, the elderly should be targeted because of the high prevalence of benzodiazepine prescription to those aged = 65 years. Possible side effects of benzodiazepines, such as memory loss and increased risks for falls and fractures, are also issues of the elderly [8, 9, 11, 13, 35].

Before making conclusions, we must mention several study limitations. First, this was a database study, and the specific reasons for prescribing benzodiazepine were often unclear. For example, it is possible that some inpatients in the department of surgery may have been prescribed benzodiazepine as a pre-medication. However, a close relationship between benzodiazepine prescription and bed days was observed in non-surgical departments as well. Future tasks would be to gather information on the use of benzodiazepines from both inpatients and the physician-in-charge. Second, a cross-sectional study design prevented us from coming to conclusions regarding causal relationships between benzodiazepine prescription and the number of bed days. The use of benzodiazepine might be a simple marker for, rather than a cause of, prolonged hospitalization. That is, the chance of benzodiazepine prescription would be higher for patients with longer hospital stays because the percentage of benzodiazepine use increased with increased bed days. However, we can at least state that the monitoring of benzodiazepine prescription is important in hospitals, whether or not benzodiazepine prescriptions are a cause or a consequence. For those who have already been in the hospital for long time, the continuing appropriateness of benzodiazepine prescription should be checked. Finally, our findings cannot be compared with those at other medical institutions in Japan because, to our knowledge, this is the first study of its type in Japan. The hospital of the study is a university teaching hospital that has hundreds of physicians who work in all general departments that offer outpatient clinics; thus, any tendency in prescribing practices might reflect the effects of Japanese educational conferences and the advice of supervisors, rather than individual prescribing preferences [36]. To generalize our conclusions requires further studies at other Japanese institutions, as well as hospitals in other countries.

Despite these limitations, our findings have valuable implications. This is the first study to report data on inpatients that are relevant to the discussion of the use of benzodiazepines in Japanese hospitals. We found that the percentage of benzodiazepine prescriptions increased with the number of bed days and that patients prescribed benzodiazepines generally stayed for a longer period than those who were not prescribed benzodiazepines. Based on the finding of a high percentage of benzodiazepine prescription to elderly inpatients, pharmacological treatment should be considered carefully using evidence-based clinical guidelines; alternatively, physicians might refer patients to psychiatrists or specialists in mental health. Further studies are needed to determine the precise cause(s) of the high level of benzodiazepine prescription in Japanese hospitals.