Effects of sedative-hypnotics on sleep quality among patients with insomnia: evidence from an observational, pre-post study in India

Background Insomnia continues to be neglected globally, despite its high prevalence. Guidelines by the health regulatory agencies call for studies to evaluate the effect of sedative-hypnotics on sleep quality. Methods We conducted a pre-post observational study to evaluate sleep quality among 186 inpatients receiving short-term oral sedative-hypnotic therapy in a tertiary care teaching hospital in Kozhikode (Kerala), India. Using Pittsburgh Sleep Quality Index_Past-Week (PSQI_PW) questionnaire, patients were interviewed upon hospital admission and at follow up after ≥1-week of sedative-hypnotic therapy. Additionally, we interviewed 36 physicians to understand the current clinical perception about sedative-hypnotics. Results Mean (SD) age of the study patients was 59 (7.5) years. Majority (63.4%) of the patients were men. Of the various primary diagnoses for hospitalization, cardiovascular disease was the most common (22.6%, n = 49). Sedative-hypnotic therapy improved the mean (SD) PSQI_PW overall score by 6.79 points (pre: 12.70 (3.5) vs. post: 5.91 (2.8); p < 0.0001). Statistically significant improvements in sleep duration, latency, efficacy, and day dysfunction were observed. Higher proportion of study patients were prescribed benzodiazepines (73.7%) compared to zolpidem (26.3%). Patients treated with zolpidem reported higher improvements in mean overall PSQI_PW scores compared to those treated with benzodiazepines, however these differences were not statistically significant upon adjusting for age, gender and primary diagnosis for hospitalization. Qualitative interviews indicate that that physicians consider zolpidem to be safer and more efficacious. Conclusions In our study, sedative-hypnotic therapy helped improve sleep quality among the hospitalized patients. More studies evaluating the comparative efficacy and safety of zolpidem vs. benzodiazepines – including among patient groups with varying demographic and clinical characteristics – are needed. India must develop evidence-based treatment guidelines to inform the clinical practice around the use of sedative-hypnotics.


Background
Insomnia continues to be overlooked despite its association with life threatening comorbidities like cardiovascular diseases and diabetes. Addressing the deleterious effects of insomnia poses a challenging task, especially in India, considering patients' attitude about the disease and the suboptimal healthcare system [1]. A recent report claimed that over 90% of the Indians fail to receive adequate sleep, with up to 58% believing that their work efficiency is affected by poor sleep quality and 11% falling asleep during work. However, only 2% of those with sleep-deprivation discuss their insomnia with a physician [2]. Mental disorders are a strong social stigma in India; insomnia often goes underreported like other mental disorders (such as anxiety and depression) because consulting a psychiatrist is often considered a taboo [3,4]. A global survey entitled "Better Sleep, Better Health: A global look at why we're still falling short on sleep"conducted across 13 countries including the USA and UKreported a very high prevalence of poor sleep in India [1]. In South India, the urban population seems to be more predisposed to difficulty in initiating and maintaining sleep than their rural counterparts [5]. Lack of awareness, coupled with inadequate health seeking behaviour, possibly underemphasizes the importance of healthy sleep [5]. The prevalence of insomnia is also high in North Indiawith over 75% patients reporting poor sleep quality upon hospital admission [6].
Sleep-wake cycle is greatly disrupted with ageing. Sleep apnoea and insomnia appear to be the major sleep problems among the elderly; approximately 50% of the elderly report difficulty initiating as well as maintaining sleep [7]. Alteration of sleep architecture includes longer peripheral stages of sleep (N1 and N2) and shorter stages of deep sleep (N3) or REM sleep. Insomnia in geriatric population is frequent, yet unresolved [8]. Although many diseases often contribute to insomnia, one cannot generalize that all comorbidities have a deleterious effect on sleep [9]. On the contrary, inadequate sleep is associated with higher risk as well as poorer prognosis of chronic diseases such as diabetes and cardiovascular diseases [10,11]. Moreover, poor subjective sleep quality and inadequate sleep duration adversely affect the overall health-related quality of life [12]. In particular, sleep quality and duration are often compromised among the hospitalized patients due to pain, acute illness, anxiety, medical intervention, environmental noise and lights [13,14].
Sedative-hypnotic drugs (primarily benzodiazepines (BZDs) and non-BZDs) are the mainstay for insomnia. To ensure the rational use of sedative-hypnotics, regulatory bodies (such as UK's National Institute for Health and Care Excellence) have recommended further studies to evaluate the impact of hypnotics on the improvement of sleep quality and daytime sedation. Pittsburgh Sleep Quality Index (PSQI), a standardized questionnaire by the University of Pittsburgh, is a reliable tool to evaluate sleep quality [15]. Employing this questionnaire, we evaluated the sleep quality of inpatients undergoing shortterm oral sedative-hypnotic therapy in a tertiary care hospital setting in South India.

Methods
This observational, pre-post study was designed to assess the sleep quality among in-patients undergoing oral sedative-hypnotic therapy. We obtained a modified oneweek version of Pittsburgh Sleep Quality Index Past Week (PSQI_PW) from the University of Pittsburgh. We conducted in-depth patient interviews on the day of hospital admission (pre-score) and on the day of discharge or 7 days after hospital admission (post-score), whichever happened earlier. We also report on the factors influencing sleep quality, the sedative-hypnotic prescribing pattern, and patient-reported adverse effects of sedativehypnotic drugs. We use the terms 'therapy' and 'treatment' interchangeably in this manuscript.

Study setting and patient sample
The study was conducted in an 800 bedded tertiary care teaching hospital in Kozhikode (Kerala state), India during November 2017 to May 2018. We enrolled all the inpatients (n = 186) who complained of difficulty initiating and/or maintaining sleep (insomnia) and were prescribed oral sedative-hypnotic on the day of hospital admission. These patients had varying primary diagnoses from various specialty departments. Patients with chronic insomnia (sleep trouble occurring at least 3 nights per week for at least 3 months) or those on longterm sedative-hypnotics therapy were excluded [16]. We also excluded all children, pregnant women, and the patients undergoing surgery.

Data collection and statistical analysis
We collected patient scoresbased on PSQI_PW instructionsto assess the four variables of sleep quality i.e. sleep duration, sleep efficiency, latency and day dysfunction. See Table 1. Individual scores of these variables were added up to calculate the total pre-and post-score (i.e. before and after the sedative-hypnotic therapy, respectively), ranging from 0 (better) to 21 (worse) for each patient. We also interviewed the patients (or caretaker for psychiatric patients) using PSQI_PW questionnaire. The patient case files were followed up each day to accurately record the occurrence of adverse events.
Following a week long oral sedative-hypnotic therapy, we re-interviewed the patients to calculate the postscores. We conducted follow up interviews via phone for the patients discharged within 7 days of hospital admission. Differences in mean sleep duration, mean sleep latency and mean sleep efficiency were compared. Due to high correlation between these continuous measures at two measurement times, a linear mixed model was used at alpha significance level of 0.05. An unstructured correlation structure was chosen because of limited measurement times (pre-post), and we used the Ken-Ward-Roger approximation method to compute denominator degrees of freedom [17]. Additionally, patient-reported adverse effects related to sedativehypnotic therapy were recorded. McNemar's statistical test was performed to compare the pre and post categorical values for day dysfunction. We employed SAS 9.4 for all the analyses.
We also conducted in-depth interviews with 36 physicians identified through convenience sampling, in order to understand the current clinical perception of sedative-hypnotic use.
On the day of hospital admission, 58% (n = 108) of the patients experienced difficulty initiating sleep (sleep   Table 3). The mean (SD) sleep efficiency score was 2.28 (0.56). See Table 3.
Prior to sedative-hypnotic treatment, about 80% (n = 149) reported abrupt awakening at night -mostly due to pain (14%) and dyspnoea (10.8%). About 10% patients experienced severe broken sleep without an identifiable cause and reported waking up multiple times at night. Majority of the sleep disturbances could be attributed to an underlying health condition i.e. pain among fracture patients, anxiety among cardiovascular disease patients, and dyspnea in patients with respiratory diseases. See Additional File 1: Figure S1.
In contrast, only 26.3% (n = 49) received non-BZD hypnotics i.e. zolpidem. Zolpidem was preferred for patients with respiratory diseases, probably due to risk of respiratory depression associated with BZDs. Other hypnotics i.e. Z-drugs (zaleplon, zopiclone) and melatonin receptor agonist (ramelteon) were not prescribed to any study patient. See Fig. 1.
Since six statistical tests were used in the pre and post treatment comparison (i.e. one each for sleep duration, latency, efficiency, day dysfunction, primary diagnosis and medication), we applied Bonferroni correction by multiplying p-value by 6.
Upon sedative-hypnotic therapy, the overall mean (SD) PSQI total score improved by 6.79 points (pre: 12.70 (3.5) vs. post: 5.91 (2.8); p < 0.0001) after 7 days. Comparing patient groups by primary diagnosis for hospitalization, the mean PSQI_PW score improvement was the highest among patients with fractures (pre: 15.71 vs. post: 5.79; p < 0.0001) followed by those with infections (pre: 13.28 vs. post: 4.96; p < 0.0001). Difference in the pre-post PSQI_PW scores was relatively lower but were statistically significant in patients with psychiatric disorders (pre: 13.75 vs. post: 7.25; p < 0.0001) and respiratory diseases (pre: 13.83 vs. post: 7.46; p < 0.0001). Patients with cardiovascular diseases showed least improvement in sleep quality (pre: 11.01 vs. post: 5.25; p < 0.0001). See Table 4(a). Table 4(b) reports mean pre-and post-treatment PSQI_PW scores among patients stratified by administered sedative-hypnotic drug. Among patients who received BZDs, the mean differences in PSQI_PW score ranged from − 6.17 (clonazepam) to − 6.59 (lorazepam). The pre-post differences (i.e. improvements) were statistically significant for all BZD patient sub-groups, except those who received nitrazepam. Patients who received zolpidem experienced statistically significant improvements in sleep quality; the pre-post mean difference in PSQI_PW score was − 7.91 (pre: 14.32 vs. post 6.41; p < 0.001).
Overall, the mean pre and post PSQI_PW scores were 12.7 and 5.9 (percent improvement: 53%), respectively.   For comparisons among patient sub-groups, the mean pre and post treatment PSQI_PW scores were compared using a linear mixed model, adjusting for age and gender. Table 5 compares the mean PSQI_PW improvements in patients administered BZDs versus zolpidem. In the unadjusted analysis, the percentage pre-post improvement among the study patientsoverall and in each primary diagnosis group except infectious diseaseswas relatively higher among those treated with zolpidem (Fig. 2). However, upon adjusting for age, gender and primary diagnosis for hospitalization, the difference in    mean PSQI_PW improvements among zolpidem and BZDs groups were not statistically significant. Overall, 10.2% (n = 19) patients reported adverse effects following initiation of sedative-hypnotic therapy, of which 74% were aged > = 60 years. Of the patientreported adverse effects, daytime drowsiness was the most common (n = 6) followed by early-morning awakening (n = 3), nightmares (n = 3) and dysgeusia i.e. altered taste (n = 2). None of the patients encountered falls or minor accidents. Patients treated with zolpidem did not report early morning awakening or daytime sedation (Additional File 1: Table S4).
For all the study patients, the short-term sedative-hypnotic therapy dosing was tapered and discontinued. While this would reduce any risk of rebound insomnia, 7% (n = 13) reported persistent insomnia following discontinuation.

Qualitative results
To understand the physicians' standpoint with respect to the current dilemma in choosing the right sedative-hypnotic, we conducted qualitative interviews with the physicians using a semi-structured questionnaire. Of the 36 physicians surveyed from various speciality departments, 72% (n = 26) preferred zolpidem for its higher (perceived) efficacy than BZDs in improving sleep quality. Furthermore, 53% (n = 19) physicians believed that zolpidem is also safer than BZDs for inpatients.
However, prescribers' opinion is inconsistent with the observed prescribing pattern of sedative-hypnotics in the hospital, with BZD being more frequently prescribed.
"BZDs often cause respiratory depression in susceptible individuals (patients with respiratory diseases). When such patients complain of insomnia, Z-drugs seem to be the best treatment option." -Pulmonologist A This opinion however varies from one speciality department to another. A psychiatrist seems more likely to prescribe BZDs, owing to its anxiolytic effects. Zolpidem, on the other hand -being devoid of anxiolytic effectsis less preferred in such circumstances.
"Most of the patients who report poor sleep also complain of anxiety at bedtime, especially prior to the day of a major diagnostic procedure. Tackling anxiety with a BZD appears to be more prudent than merely inducing sleep." -Cardiologist X Some physicians told that they also choose BZD because it is less expensive than zolpidem. Physicians, especially in the lower and middle income countries like India, often find the need to consider patients' ability to afford prescription medicines [18].

Discussion
We found that inpatients who complained of trouble initiating or maintaining sleep had poor or very poor sleep quality scores. Our findings are in accordance with previous reports of the untreated burden of insomnia among inpatients, especially the elderly [8,9,19]. A previous study in Brazil reported that abrupt awakenings among inpatients were primarily caused by comorbid conditions, uncomfortable bedding, nursing routine, fear, pain, breathlessness and room temperature. Patients also attributed insomnia to the comorbid condition such as fracture-induced pain, lung disease induced cough, or dyspnoea [19]. Poor sleep not only affects overall health, but also has negative impact on one's appearance, mood, motivation and concentration [1].
In our study, short-term oral sedative-hypnotic therapy improved sleep quality among majority of the inpatients who experienced insomnia. Comorbidityinduced insomnia benefited from sedative-hypnotic therapy; maximally in patients with hospitalizations due to fractures. The small proportion (7%) of the study population who reported persistent insomnia after sedative-hypnotic discontinuation also reported very poor prognoses; neither did their existing health condition or underlying diseases improve upon hospitalization, nor did they benefit much from sedative-hypnotic therapy.
Literature indicates that sleep quality is a proxy prognostic marker for comorbid conditions [20][21][22][23]. Therefore, timely diagnosis and management of insomniawith lifestyle/behavioural modifications and/or sedative-hypnotics therapy as appropriateis important. With respect to sedative-hypnotic therapy, the choice of drug, dose frequency and duration of therapy must be customized to each patient's characteristics and condition of severity [24]. However, in India, there is lack of clinical guidelines for prescribing of sedative-hypnotics among patients with varying diagnoses and circumstances. Our qualitative interviews indicate that utilization of short-term sedative-hypnotic therapy in patients with comorbidity-induced insomnia is largely driven by physician's judgement.
The treatment of insomnia remains somewhat unchanged for decades, ever since the introduction of BZDs in the 1960s and Z-drugs in the 1990s. From 2006 through 2011, zolpidem prescriptions underwent a steady rise in the US [25]. Europe has witnessed a decline in BZD prescriptions with a corresponding rise in Z-drug prescriptions. This trend continues in spite of contrary guidelines issued by the UK's National Institute for Health and Care Excellence (NICE), which favour the least expensive hypnotic i.e. BZDs [26][27][28]. The European Guidelines for Diagnosis and Treatment of Insomnia (2017) recommends BZD and Z-drugs as first line pharmacotherapy choice, but does not mention the differential impact of these two drugs on sleep quality [29].
Although zolpidem went on to become the most prescribed sedative-hypnotic in the US and UK over the last decade, little has this transition affected the practice in India [30,31]. Our findings are concordant with most of the recent drug utilization evaluation studies in India, in which BZDs continue to be preferred largely over Zdrugs [31,32]. Only 26% of the patients were prescribed zolpidem, the only Z-drug prescribed.
Although some literature hints the superiority of zolpidem over BZDs in terms of physician preference, the overall comparative evidence regarding sleep quality is inconclusive [33,34]. In our analyses comparing zolpidem and BZDs, while zolpidem led to greater improvement in overall sleep quality among the patient, the differences in improvement were not statistically significant upon adjusting for age, gender and primary diagnosis for hospitalization. Our qualitative interviews suggest that majority of the physicians support zolpidem; however, introducing a change in the existing practice is likely to be slow because junior doctors tend to follow their seniors, who have been habituated to prescribing BZDs for decades. The tendency to follow key opinion leaders is likely to be driven by the 'prestige-based hierarchy'prevailing in India [35]. Thus, individual clinical judgements of key opinion leaders appear to override clinical efficacy when the practice is primarily physician-dominated.
The inclination towards BZDs persists in India possibly due to the paucity of literature supporting the superiority of Z-drugs over BZDs. As mentioned earlier, India neither possesses nor follows standardized treatment guidelines that govern the use of sedatives. The poorly understood concept of safety and adverse effect profile of Z-drugs vs. BZDs presents a problem especially among geriatric patients and other high risk groups [36]. Nevertheless, with physicians in both the US and UK preferring zolpidem over the years, the prevailing popularity of BZDs in India is unanticipated, given the fact that Indian pharmaceutical regulations are preferentially biased towards 'imported' rules of practice.
On another note, poor marketing practices might have also contributed to zolpidem's low utilization in India. A controversial advertisement campaign by Abbott Pharmaceuticals in 2011aiming to promote its brand of zolpidem in Indiaevoked widespread criticism by doctors. While the pharma-giant probably overexploited the findings of a study that associated poor sleep with increased mortality due to cardiovascular disease, the doctors pointed out that the practice of 'pill popping' is more harmful [37].
Cost concerns do play a vital role in the choice of drugs, especially in the rural and low socioeconomic status areas in India. BZDsbeing relatively cheaperare thus preferred over zolpidem and other Z-drugs. Despite the fact that this pattern in India complies with the NICE recommendations of prescribing hypnotic of minimal cost, sleep quality is not a factor to be neglected. With India passing The Mental Healthcare Act of 2017, ailments mentioned in the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-V) become eligible for coverage under India's health insurance policies [38]. As our qualitative results suggest, the choice of the sedative-hypnotic must also take into account the patients' underlying diseases. This creates the compelling need for appropriate evidence-based treatment guidelines for sleep disorders among inpatients. We recommend more studies (such as prospective trials and/or real-world studies) evaluating the comparative efficacy and safety of zolpidem vs. benzodiazepines among patient groups with varying demographic and clinical characteristics, in order to inform treatment guidelines and clinical practice.
Our study, however, has few limitations. Our analyses did not adjust for some potential confounderssuch as patient's health status, severity of underlying condition and other comorbiditiesthat may influence the outcomes of sedative-hypnotic drugs in management of insomnia. Furthermore, our findings from a single tertiary healthcare centre in Kerala state cannot be extrapolated to other Indian states with different health indicators, disease burden, patient demographics and socio-economic status.

Conclusions
In conclusion, with sleep quality being an important prognostic marker during hospitalization, lack of standard guidelines for utilization as well as continuous monitoring of the efficacy of sedative-hypnotics needs to be addressed. Our study suggests the possible superiority of zolpidem over BZDs in improving sleep quality in hospitalized patients, however these results are not conclusive. We underline the need of more clinical studies comparing the impact of zolpidem and BZDs on sleep quality in order to inform clinical guidelines and promote evidence-based use of sedative-hypnotics.
Additional file 1. Supplementary information on patient demographics and sleep quality. support from SK. SK also facilitated data collection while interviewing psychiatric patients. AS, SRM and GS developed the analysis plan, and conducted the data analyses and statistical tests. GS and SP drafted the first draft of the manuscript with inputs from AS, SRM and MKU. GS, AS, SRM, SP and MKU conducted the literature review, interpreted the results, and revised and edited the subsequent versions of the manuscript to its final stages. All the authors approved the final manuscript. The views expressed in this manuscript are of the authors and not necessarily of their affiliation institutions.