The influence of co-morbidities on prescribing pharmacotherapy for insomnia: evidence from US national outpatient data 1995–2004

Summary

Objective: Patients with insomnia are likely to have other co-morbidities that could affect pharmacotherapeutic choices. This study examined the prevalence and impact of co-morbidities on the pharmacological treatment of insomnia.

Study design: A retrospective data analysis of the National Ambulatory Medical Care Survey from 1995 to 2004, comprising patients with a diagnosis of insomnia, was conducted. Multivariate logistic regression models were used to predict the impact of co-morbidities on pharmacotherapy for insomnia.

Results: A total of 5,487 unweighted patient visits with insomnia were identified, representing 161.4 million patients in the US. Approximately 38% of these patients had at least one co-morbid condition. Patients with mental co-morbidities, especially anxiety, had decreased likelihood of receiving pharmacotherapy for insomnia.

Conclusions: Mental co-morbidities such as episodic mood disorder, anxiety and depression are prevalent in patients with insomnia. However, many co-morbid patients do not receive pharmacological therapy specific for insomnia.

Keywords::

• insomnia
• co-morbidity
• pharmacotherapy
• benzodiazepine
• non-benzodiazepine

Introduction

Sleep difficulty is a commonly reported complaint in primary care practice in the US1. Among the sleep difficulties, insomnia is considered to be the most common disorder, affecting one-third of adult Americans1. Approximately 20% of patients are reported to have an occasional basis of insomnia, whilst approximately 10% of patients suffer from chronic insomnia1–3. Insomnia can have a severely negative impact on individuals’ lives and is associated with mood disturbances, difficulties with concentration, decreased cognitive functioning, increased absenteeism, impaired work performance, worse health status and poor quality of life3–6. Insomnia also poses a significant economic burden on the healthcare system, mainly because of increased use of medical services and increased healthcare utilisation in the US. The annual direct cost of insomnia was estimated to be $14 billion whilst the indirect cost was approximately $28 billion7.

Traditionally, insomnia is treated with behavioural therapy and/or pharmacotherapy8,9. Among the pharmacotherapeutic classes of medications, benzodiazepines and non-benzodiazepines are commonly used treatment modalities. These agents (except eszopiclone and zolpidem extended release) are indicated for the short-term management of insomnia10. However, long-term use of these medications has not been studied extensively in randomised clinical trials. Furthermore, these medications have the potential to cause adverse drug reactions (ADRs) such as residual daytime sedation, cognitive impairment, motor incoordination, dependence, addiction and drug abuse11–14. However, some studies have shown that ADRs are lower with non-benzodiazepine medications compared with older benzodiazepine medications15–19.

Insomnia, if not treated in a timely manner, can be associated with long-term complications such as depression and anxiety disorder20. Patients with insomnia commonly experience co-morbid mental disorders such as anxiety, mood disorders, dementia and substance abuse disorders18,20. Several studies have also documented that insomnia is considered to be an independent predictor of depression21–24. A study conducted by Gottleib et al reported that insufficient sleep (<8 hours/night) was associated with an increased risk of hypertension25. However, the impact of co-morbidities on the pharmacological treatment of insomnia has not been studied in detail as yet. Insomnia together with other co-morbid conditions can severely impair a patient's health status and quality of life. This study is important in terms of examining the influence of co-morbidities on prescribing pharmacotherapy for the management of insomnia. Hence, the objectives of this study were: (i) to determine the prevalence of co-morbidity in patients with insomnia treated in outpatient settings in the US using National Ambulatory Medical Care Survey (NAMCS) data; and (ii) to examine the impact of patient co-morbidity on the prescribing of pharmacological treatment for insomnia in outpatient physician settings in the US.

Methods

Study design

A cross-sectional study design was used to determine the prevalence and impact of patient co-morbidities on the treatment patterns for insomnia in US outpatient settings. NAMCS data from 1995 to 2004 was used for this study. The NAMCS is a national probability sample survey conducted by the Division of Health Care Statistics of the National Center for Health Statistics and the US Centers for Disease Control and Prevention. The NAMCS makes use of a multistage probability sampling design that is comprised of probability samples of primary sample units (PSUs), physician practices within PSUs and patient visits within practices. This survey contains data on individual outpatient office visits and is then weighted to reflect national estimates describing the use of ambulatory medical care services in the US. The survey is based on a sample of visits to non-federally employed office-based physicians (excluding those in the specialties of anaesthesiology, radiology and pathology) who are primarily engaged in direct patient care. The physician or a member of the physician's staff provides information about patient sociodemographics, physician specialty, reasons for the visit, source of payment for the visit, patient's complaints and symptoms, diagnoses made, second- or first-time visit, medication prescribed, and therapeutic and preventive services recommended for each visit. To obtain these national estimates, each individual record is assigned an inflation factor called the patient visit weight, which is used to obtain the total number of office visits made in the US26. All estimates from the NAMCS are related to the number of patient visits and subject to sampling variability^26.

Patient data

Patients aged ≥18 years with a physician visit and primary/secondary diagnosis of insomnia in US outpatient settings were included in this study. Clinical diagnoses of insomnia/insomnia of multiple causation were identified using the 9th Revision of the International Classification of Disease (ICD-9) codes (307.40, 307.41, 307.42, 307.47, 307.48, 780.5, 780.50, 780.52, 780.55, 780.56 and 780.59) reported in the patient visit record of the NAMCS dataset. Office visits were also selected as insomnia-related if insomnia was reported in the ‘reason of visit’ field or if any insomnia-specific medications were prescribed at the time of visit. The list of medications for the identification of insomnia-related visits was restricted to those that only had first-line indications for insomnia in order to avoid confounding by indication. This was done only for patients who were identified as having sleep difficulty through medication prescription.

Patient-specific information was extracted from the NAMCS data, including age, gender, race, residence in a metropolitan versus non-metropolitan area, insurance provider, length of visit, number of medications prescribed, established versus new patient, and geographic region. Physician practice characteristics were also obtained from the dataset, including whether they were the owner of a solo practice, medical specialty and geographic region.

Patient co-morbidities

Patient co-morbidities were also identified using ICD-9-CM codes. The patient co-morbidity was examined by creating indicator variables for each individual co-morbidity as well as creating an index of patient co-morbidity severity, the Charlson index27. This index is a simple, readily applicable method of estimating risk of death from co-morbid disease and can be used in longitudinal studies27. The Charlson co-morbidity severity index has since been validated for several other health outcome estimations besides death28–30. The co-morbidity index assigns weights for a number of major conditions (ranging from 1 for conditions such as myocardial infarction to 6 for metastatic solid tumour and AIDS). The index severity score for each patient is calculated by totalling the assigned weight for each of the co-morbidities that the patient suffers from30.

Prescription medications for insomnia

The literature has documented two major pharmacotherapeutic classes of medications used for the treatment of insomnia in the US; the benzodiazepines (e.g. flurazepam, quazepam, triazolam, estazolam and temazepam) and the non-benzodiazepines (e.g. zolpidem and zaleplon)31. For patients with insomnia identified using ICD-9 codes or insomnia as the ‘reason of visit’ who were also prescribed medication at the time of visit, medications were not restricted to the above list as there would be no confounding by indication.

Statistical analysis

The unit of analysis was an individual patient visit. Data were stratified according to patient characteristics, physician specialty, resulting diagnosis and medications prescribed. Several categorical variables were created for predictor variables extracted from the NAMCS dataset to control for potential confounding bias. The χ2 statistic was used to identify differences in categorical variables. Data were analysed taking into account the nature of complex survey design. Multivariate weighted logistic regression models were used to examine the potential predictors of pharmacological treatment for insomnia. Two separate regression models were developed in which one was used for identifying predictors of pharmacotherapy. The comparison group was patients with insomnia who did not receive any pharmacotherapy at the time of visit. Another logistic regression model was used to estimate predictors of non-benzodiazepine therapy in the subset of patients receiving pharmacotherapy, where the comparison group was patients who received benzodiazepines. Co-morbidity was included as the primary predictor variable in all analyses. First, the analysis was performed using the Charlson co-morbidity index as the indicator of co-morbidity, followed by subgroup analysis of patients with inclusion of an individual type of co-morbid condition. Two specifications were not combined in one model to avoid grouping effects. Univariate, bivariate and multivariate analyses of data were performed using STATA™ software version 9.0 (Stata Corp., College Station, TX).

Results

Patient characteristics

A total of 5,487 unweighted patient visits for insomnia of multiple causation were identified from 1995 to 2004. Although this appears to be a small sample, the NAMCS data do not allow us to capture the actual number of patients with insomnia; they only reflect the number of randomly selected patient visits with a diagnosis of insomnia. The data represent 161.4 million patient visits in the overall US population calculated using sampling weight or inflation factor provided in the NAMCS dataset to obtain estimates for the entire population. Table 1 gives the characteristics of the study population. Official visits for insomnia were more common in females (60.4%) and White patients (90.1%). There was an increasing prevalence of insomnia with age, and this was highest in adults aged ≥65 years (30.7%). Approximately 88% of office visits were made by established patients (i.e. patients were already seen by a physician before this visit) with a diagnosis of insomnia.

Table 1.  Descriptive statistics of the study population (unweighted population size = 5,487 cross-sectional patient visits from 1995 to 2004).

Variable	Measure	Proportion (%)
Patient demographics
Age groups (years)	18–34*^†	12.06
	35–49^†	29.29
	50–64^†	27.94
	≥65*^†	30.71
Gender*	Male	39.58
Race^†	Hispanic	7.13
	White	90.14
Established (vs. new) patients		88.34
Source of payment	Private insurance*^†	44.41
	Public insurance*^†	38.53
	Other	17.06
Co-morbidities	Presence of mental co-morbidity*^†	41.21
	Charlson co-morbidity index*	Mean = 0.22 (sd 0.68)
List of five most prevalent	Episodic mood disorders*^†	14.94
co-morbidities	Anxiety*	15.60
	Essential hypertension	10.15
	Depressive disorder*^†	7.02
	Diabetes	3.50
Physician/practice characteristics	Physician owner of solo practice (only 2001–2004)*	38.49
Region	Northeast	21.72
	Midwest*	20.41
	South	33.97
	West^†	23.89
Practice in metropolitan area	Yes*^†	83.96
Physician specialty	Family practice and internal medicine	32.73
	Psychiatry*^†	29.80
	Neurology*	9.02
	Other*^†	28.45
Pharmacotherapy	Prescribed medication therapy at visit	32.55
	Benzodiazepine prescription rate among those receiving insomnia medications	34.32

sd, standard deviation.

* Significantly correlated with medication therapy prescription at p ≤0.05 significance level using χ2 tests.

^† Significantly correlated with benzodiazepine medication prescription at p ≤0.05 significance level using χ2 tests.

Patient co-morbidities

Of patients who made office visits for insomnia 38% were reported to have at least one co-morbid condition. Approximately 41% of these patients (i.e. those reporting at least one co-morbid condition) had a concomitant diagnosis of a mental co-morbidity. The mean Charlson co-morbidity index in patients with insomnia was low (mean ± standard deviation 0.22 ± 0.68). Figure 1 illustrates the prevalence of the most common co-morbidities in patients with insomnia over the 10-year period. Among the specific co-morbidities examined, anxiety (15.6%) was the most common co-morbidity followed by episodic mood disorders (14.9%) (Table 1; Figure 1). Similarly, there was a stable prevalence of hypertension (10.1%), depression (7%) and diabetes (3.5%) in patients with insomnia over the 10-year period (Table 1; Figure 1).

Figure 1.  Prevalence of co-morbidities in patients with insomnia (1995–2004).

Physician characteristics

Approximately 38% of physicians were the owner of a solo practice from 2001 to 2004. When categorised according to the area of specialty, approximately 33% of patients visited family practice and internal medicine providers. However, approximately 30% of patients consulted psychiatrists and only 9% referred to neurologists. In addition, 32.6% of patients who made official visits for insomnia received some type of pharmacotherapy. Among patients who received some form of pharmacological treatment, 34.3% of patients received benzodiazepine prescription.

Predictors of pharmacotherapy

Table 2 details the predictors of pharmacotherapy for the treatment of insomnia for a weighted sample of 107.4 million patients. A multivariate logistic regression model indicated that patients aged 50–64 years had a 44% increased likelihood of receiving pharmacotherapy for insomnia compared with those aged 18–34 years (reference group) (odds ratio (OR) 1.44, 95% confidence interval (CI) 1.08–1.92). The odds of receiving pharmacotherapy for insomnia in established patients were 38% higher than in new patients (OR 1.38, 95% CI 1.02–1.88).

Table 2.  Predictors of type of therapy prescribed for sleep difficulty during outpatient visit (weighted sample size = 107.4 million) by logistic regression^‡.

Independent variable	Dependent variable
	Medication therapy prescribed (OR (95% CI))	Medication therapy prescribed accounting for type of co-morbidity (OR (95% CI))
Charlson co-morbidity index	1.02 (0.91–1.14)	–
Presence of mental co-morbidity	0.64 (0.50–0.83)*	–
Number of co-morbidities	1.09 (0.98–1.22)	–
Anxiety	–	0.55 (0.43–0.70)*
Episodic mood disorders	–	0.89 (0.68–1.16)
Essential hypertension	–	0.84 (0.64–1.09)
Depressive disorder	–	0.99 (0.70–1.39)
Diabetes	–	0.99 (0.67–1.47)
Age (years)^†
35–49	1.39 (1.04–1.85)*	1.40 (1.05–1.87)*
50–64	1.44 (1.08–1.92)*	1.49 (1.12–1.99)*
≥65	1.27 (0.93–1.74)	1.37 (1.01–1.87)*
Male gender	0.99 (0.84–1.17)	0.99 (0.83–1.17)
White race	1.15 (0.87–1.52)	1.13 (0.86–1.49)
Established patient	1.38 (1.02–1.88)*	1.42 (1.05–1.92)*
Insomnia as primary diagnosis	0.44 (0.30–0.63)*	0.40 (0.29–0.55)*
Metropolitan area	1.27 (1.02–1.58)*	1.26 (1.01–1.57)*
Region^†
Northeast	0.82 (0.64–1.05)	0.82 (0.64–1.05)
Midwest	0.63 (0.49–0.80)*	0.62 (0.48–0.79)*
South	0.84 (0.68–1.04)	0.82 (0.67–1.02)
Source of payment^†
Public insurance	1.16 (0.94–1.44)	1.18 (0.95–1.46)
Other	0.95 (0.75–1.20)	0.94 (0.74–1.19)
Specialty^†
Psychiatry	1.70 (1.32–2.20)*	1.40 (1.11–1.77)*
Neurology	0.73 (0.54–0.99)*	0.68 (0.51–0.93)*
Other	0.81 (0.66–1.00)*	0.82 (0.66–1.00)*

OR, odds ratio; CI, confidence interval; NAMCS, National Ambulatory Medical Care Survey.* Significant at p <0.05.^† For individual categorical independent variables with more than two categories, the dropped reference categories are age group 18–34 years, East region, private insurance, and family practice and internal medicine specialty.^‡ All analyses were weighted using the NAMCS sampling weights. Variables significant in univariate analyses are included in the multivariate models.

Patients with a primary diagnosis of insomnia had a 56% lower likelihood of receiving pharmacotherapy than those with a secondary diagnosis of insomnia (OR 0.44, 95% CI 0.30–0.63). Patients with mental co-morbidities had 36% decreased likelihood of receiving pharmacological treatment for insomnia than those without mental co-morbidities (OR 0.64, 95% CI 0.50–0.83). Subgroup analysis of a type of mental co-morbidity further revealed that patients with co-morbid anxiety had a 45% decreased likelihood of receiving pharmacological treatment for insomnia than those without anxiety (OR 0.55, 95% CI 0.43–0.70).

When the impact of physician specialty on the prescribing behaviour for pharmacotherapy of insomnia of multiple causation was studied, it was observed that patients visiting psychiatrists had two times higher odds of receiving pharmacotherapy for insomnia than those visiting family practice and internal medicine providers (OR 1.70, 95% CI 1.32–2.20). Patients visiting a neurologist had a 27% decreased likelihood of receiving pharmacotherapy for insomnia than those seeing family practice and internal medicine providers (OR 0.73, 95% CI 0.54–0.99).

Predictors of non-benzodiazepine prescription

Table 3 details patient and physician predictors of non-benzodiazepine prescription therapy among patients who received some type of pharmacotherapy for insomnia. Patients aged ≥65 years had a 57% lower likelihood of receiving a non-benzodiazepine prescription compared with patients aged 18–34 years (OR 0.43, 95% CI 0.23–0.79) among patients who were prescribed an insomnia-specific medication during the visit. Patient visits with public insurance (compared with private insurance) as a primary payer source were associated with decreased likelihood of non-benzodiazepine prescription (OR 0.58, 95% CI 0.39–0.86).

Table 3.  Physician and patient predictors of non-benzodiazepine prescription for sleep difficulty at outpatient visits (weighted sample size = 54.0 million) by logistic regression^‡.

Independent variable	Dependent variable
	Non-benzodiazepine prescription (OR (95% CI))	Non-benzodiazepine prescription accounting for type of co-morbidity (OR (95% CI))
Charlson co-morbidity index	1.06 (0.87–1.29)	–
Number of co-morbidities	1.14 (0.95–1.36)	–
Presence of mental co-morbidity	1.21 (0.75–1.95)	–
Anxiety	–	0.97 (0.61–1.54)
Episodic mood disorders	–	1.95 (1.27–3.01)*
Essential hypertension	–	1.56 (0.98–2.49)
Depressive disorder	–	1.73 (0.91–3.31)
Diabetes	–	1.01 (0.49–2.09)
Age (years)^†
35–49	0.55 (0.30–1.00)*	0.52 (0.28–0.94)*
50–64	0.61 (0.34–1.11)	0.58 (0.32–1.06)
≥65	0.43 (0.23–0.79)*	0.41 (0.22–0.75)*
Hispanic race	0.57 (0.33–1.00)*	0.57 (0.32–1.01)
Insomnia as primary diagnosis	2.73 (1.27–5.85)*	2.51 (1.18–5.32)*
Source of payment^†
Public insurance	0.58 (0.39–0.86)*	0.59 (0.40–0.88)*
Other	0.57 (0.38–0.86)*	0.58 (0.38–0.87)*
Practice in metropolitan area	1.50 (1.01–2.22)*	1.44 (0.97–2.12)
Region^†
Northeast	1.63 (1.05–2.53)*	1.60 (1.03–2.50)*
Midwest	1.15 (0.73–1.80)	1.16 (0.74–1.82)
South	1.14 (0.78–1.66)	1.14 (0.78–1.67)
Specialty^†
Psychiatry	0.81 (0.49–1.36)	0.68 (0.45–1.03)
Neurology	1.05 (0.59–1.87)	1.06 (0.59–1.90)
Other	0.72 (0.50–1.03)	0.75 (0.52–1.08)

OR, odds ratios; CI, confidence interval; NAMCS, National Ambulatory Medical Care Survey.

*Significant at p <0.05.

^†For individual categorical independent variables with more than two categories, the dropped reference categories are age group 18–34 years, private insurance, East region and internal medicine physician.

^‡All analyses were weighted using the NAMCS sampling weights.

Patients with a primary diagnosis of insomnia had three times greater likelihood of receiving a non-benzodiazepine prescription than those with a secondary diagnosis of insomnia among those who received some type of pharmacotherapy for the treatment of insomnia (OR 2.73, 95% CI 1.27–5.85). Among specific co-morbidities examined, patients with co-morbid episodic mood disorders were the only ones associated with a nearly two-fold increase in likelihood of receiving a non-benzodiazepine prescription (OR 1.95, 95% CI 1.27–3.01).

Discussion

Co-morbidities are widespread in patients with insomnia in the US. Patients with insomnia commonly suffer from mental co-morbidities such as anxiety, depression and episodic mood disorders in US outpatient settings. A stable prevalence of co-morbid conditions such as hypertension and diabetes was also found in this population. The findings of this study are consistent with the existing literature18–24.

Patients’ demographic, socioeconomic and clinical characteristics have a significant impact on the prescribing patterns of physicians for pharmacological treatment of insomnia. Analysis of the NAMCS data suggested that older age and being an established patient were associated with increased likelihood of receiving pharmacotherapy for insomnia in US outpatient visits. Patients with a primary diagnosis of insomnia were less likely to be treated with pharmacological treatment than those with a secondary diagnosis of insomnia. Many times sleep disturbances are caused by the presence of a medical or psychiatric disorder. Studies have shown that patients with medical conditions such as hypertension, arthritis, heartburn and depression are at a higher risk of developing insomnia/sleep difficulties32. It is possible that these patients may suffer from occasional or transient insomnia, hence physicians are likely to treat these patients with some type of pharmacotherapy for short-term management. However, patients with a primary diagnosis of insomnia may suffer from chronic or severe insomnia and physicians may be considering multiple therapeutic options for the treatment of these patients such as only behavioural therapy, pharmacotherapy or both. The variation in the prescribing pattern for insomnia could be due to several reasons. There are numerous patient and physician characteristics that affect the prescribing practices of physicians for insomnia in the US. Other than patient co-morbidities, time-constrained patient visits, physician's knowledge about sleep difficulties, limited training, and patient and physician belief that insomnia/sleep difficulties are not important or not curable can act as barriers for the pharmacological treatment of insomnia33,34.

The results also revealed that patients suffering from insomnia with mental co-morbidities were less likely to receive pharmacological agents specifically indicated for insomnia. Particularly when patients were analysed by type of mental co-morbidity, it was found that patients with co-morbid anxiety had a decreased likelihood of receiving pharmacotherapy for insomnia than those without anxiety. The authors could not identify the specific reasons for variations in the treatment patterns for patients with co-morbidities. However, there is the possibility that some of the patients may be receiving sedatives, anxiolytics or antidepressants such as trazodone, which is also used for the treatment of insomnia/sleep difficulties35. Although trazodone is not approved for the indication of insomnia/sleep difficulties, its off-label use is widespread in the US. There is also the possibility that these patients may be concomitantly using other antipsychotic medications that can have sedating effects or other side effects or drug interactions. Hence, physicians may be concerned about treating these patients with medications specifically indicated for insomnia. One cannot eliminate the possibility of preference of treating these patients with behavioural therapy. The study also highlights the fact that a lack of standardised guidelines for treating these patients with co-morbidities may make it difficult for clinicians to develop an adequate treatment plan. However, the authors could not conclude these points from the given analysis. Therefore, healthcare professionals should consider the impact of mental co-morbidities while treating patients with insomnia of multiple causation.

There are several physician characteristics that influence the prescribing of pharmacotherapy for insomnia32–34. Physician specialty has a significant impact on the prescription of pharmacotherapy. In this study population, the number of patient visits to a psychiatrist was considerably higher and, therefore, it is far more likely that a psychiatrist would code for the appropriate diagnoses since they are listed in DSM-IV compared with family practice or internal medicine providers. The authors found that patient visits to psychiatrists were much more likely to receive pharmacotherapy for insomnia compared with family practice and internal medicine providers. It is also possible that psychiatrists may be extensively using psychiatric medications including trazodone and antianxiety agents for treating co-morbidities in patients with insomnia. However, the study only examined the type of pharmacotherapy primarily indicated for the treatment of insomnia (hypnotics). Therefore, this study mainly analysed the hypnotic class of medications. It also reflects the expectation and the willingness of psychiatrists to treat behavioural and distressing sleep-related complaints. However, patient visits to neurologists were associated with a decreased likelihood of obtaining medications compared with family practice and internal medicine providers. This may be because psychiatrists are more informed about the severity or the impact of co-morbid conditions associated with insomnia. It is also likely that psychiatrists may be treating more complicated or severe cases insomnia where all other therapies (such as behavioural therapy) have already failed. However, this cannot be ascertained definitively from the dataset.

When the subgroup of patient visits resulting in pharmacotherapy prescription was analysed, visits by patients with co-morbid episodic mood disorders were two times more likely to result in a prescription of non-benzodiazepines. Among the patients who received some type of pharmacotherapy for insomnia, patients with a primary diagnosis of insomnia had a higher likelihood of receiving non-benzodiazepine prescriptions than those with a secondary diagnosis of insomnia. Decreased number and severity of side effects associated with the use of non-benzodiazepines compared with older benzodiazepines could be the reason for the increased use of these medications in this subgroup15–19.

The results also illustrated that patients aged ≥65 years and patients with public insurance were less likely to receive prescription of non-benzodiazepines. This study also confirms the finding that insomnia-related visits by patients aged ≥65 years were more likely to be associated with the prescription of benzodiazepines compared with visits by younger patients34. This may be due to a limited prescription drug coverage offered to Medicare beneficiaries34. The research has also documented that the type of insurance coverage has a significant impact on the prescription of pharmacotherapy36. Furthermore, some of the newer non-benzodiazepines are expensive compared with the generic version of older benzodiazepines. A recent study has shown that physician prescribing decisions are also influenced by the cost of therapy and out-of-pocket expenses incurred by patients37. Hence, there may be a decreasing trend of prescribing non-benzodiazepines to elderly patients and patients with public (Medicaid and Medicare) insurance. However, it will be interesting to examine the changes in prescribing patterns after the implementation of Medicare part D.

The literature has reported the impact of co-morbid conditions on the prescribing patterns and patient outcomes in different clinical condition such as diabetes, hypertension, anxiety and bipolar disorders. There are several studies that have examined the association between various factors such as patient's age, race, insurance coverage, out-of-pocket expenses and cost of therapy, and the physician prescribing behaviour for insomnia. However, there are sparse data available to examine the influence of patient co-morbidities in prescribing pharmacotherapy for the treatment of insomnia. This study not only accounted for different types of co-morbid conditions but also examined the severity of co-morbidities measured using the Charlson co-morbidity index. This study reflected treatment patterns for insomnia in the last 10-year time period. There is a significant variation in prescribing pharmacotherapy for insomnia in patients with different co-morbidities in the outpatient setting. Other than co-morbidities, patient age, insurance coverage and physician specialty also have a significant impact on prescribing pharmacotherapy for insomnia.

Healthcare professionals should take into consideration the impact of co-morbidities while treating patients with insomnia. The presence of co-morbid conditions such as anxiety, depression and mood disorder can cause long-term complications and significantly hamper an individual's work performance and quality of life38. Insomnia is challenging to treat for healthcare professionals owing to the lack of standardised guidelines addressing the needs of patients with co-morbid conditions. Hence, it is necessary to develop guidelines regarding the appropriate use of pharmacological treatment for insomnia in patients with other co-morbidities. It is also essential to develop and incorporate monitoring protocols as a part of a patient's treatment regimen. Taking into consideration the wide variation in the pharmacological treatment of insomnia, there is a need for a critical evaluation and management of emergent side effects and associated long-term complications of pharmacotherapy.

Limitations

There are several limitations of this study. Physician prescribing behaviour is a complex interaction of various patient-, physician-, treatment- and healthcare setting-related factors. However, there are several inherent limitations of the NAMCS database to evaluate prescribing behaviour. The NAMCS data do not provide sufficient information about several factors, including physician characteristics (such as their education, experience, age), patient–physician relationship (such as patient–physician interactions, therapy discussions) and therapy-related characteristics (such as safety, efficacy and side effects of therapy), which can influence prescribing patterns. However, this database provides a unique opportunity to study various physician-related characteristics, such as physician specialty, their Health Maintenance Organization status, ownership status and region of practice, as well as patient characteristic such as race and ethnicity, which can easily influence the decision-making process. Furthermore, the NAMCS database does not provide any information about the severity of sleep difficulty, so we could not distinguish between transient and chronic insomnia and could not examine its association with type of therapy prescribed. These data provide us with information about individual patients visit or patient–physician encounters and not actually about the individual patient per day. Therefore, no information about longitudinal treatment patterns or appropriateness of therapy could be obtained. The list of medications was also limited to only those with primary indications for insomnia in order to control the confounding effect of indications if the subject was identified with insomnia only through medication prescription. Furthermore, the NAMCS data provide information on no more than six medications received during each patient visit so information about the frequency of use of medications or past medication history could not be obtained. Physicians’ prescribing behaviour in a US outpatient setting was only looked at, hence no inferences regarding actual utilisation of these medications could be made.

This is the first study that has examined the impact of co-morbidities on physician prescribing behaviour for the treatment insomnia. Considerable variation was found in the prescribing patterns when controlled for several patient- and physician-related factors reported in the NAMCS database. However, future studies are required in this area evaluating the variation in behaviour of primary care physicians and psychiatrists separately. It will also be interesting to study how the prescribing behaviour changes over a period of time with the availability of newer pharmacotherapeutic agents for the treatment of insomnia.

Conclusion

Analysis of the NAMCS data from 1995 to 2004 indicated that co-morbidities such as episodic mood disorder, anxiety, hypertension, depression and diabetes are commonly prevalent in patients with insomnia. However, these patients with psychiatric co-morbidities are less likely to be treated with pharmacological agents specifically indicated for insomnia. Healthcare professionals should consider the impact of psychiatric co-morbidities while treating patients with insomnia. Several patient characteristics such as age and insurance coverage also influence the prescription of specific types of therapy for insomnia. The study found significant variations in the availability and access to pharmacotherapy for insomnia, especially for non-benzodiazepine class of medications, which need further exploration. The study warrants further research to explore the relationship between patient, physician and healthcare setting characteristics, and prescribing behaviour in patients with insomnia.

Acknowledgements

Declaration of interest:This study was funded by a research grant from Sanofi-Aventis. It does not specifically discuss any product of Sanofi-Aventis. R Balkrishnan is a paid consultant for Sanofi-Aventis. VN Joish is an employee of Sanofi-Aventis. MD Pawaskar, FT Camacho and RS Rasu have no conflicts of interests to declare.