If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
College of Pharmacy and Health Sciences, Purdue University, West Lafayette, Indiana, USAIndiana University Center for Aging Research, Regenstrief Institute, Indianapolis, Indiana, USACenter for Healthcare Innovation and Implementation Science, Indiana University, Indianapolis, Indiana, USA
Indiana University Center for Aging Research, Regenstrief Institute, Indianapolis, Indiana, USACenter for Healthcare Innovation and Implementation Science, Indiana University, Indianapolis, Indiana, USADepartment of Internal Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
Department of Internal Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USAClem McDonald Center for Biomedical Informatics, Regenstrief Institute, Indianapolis, Indiana, USA
Examine the association between race and time to pharmacologic treatment of insomnia in a large multi-institutional cohort.
Methods
Retrospective analysis of electronic medical records from a regional health information exchange. Eligible patients included adults with at least one healthcare visit per year from 2010 to 2019, a new insomnia diagnosis code during the study period, and no prior insomnia diagnosis codes or medications. A Cox frailty model was used to examine the association between race and time to an insomnia medication after diagnosis.
Results
In total, 9557 patients were analyzed, 7773 (81.3%) of whom where White, 1294 (13.5%) Black, 238 (2.5%) Other, and 252 (2.6%) unknown race. About 6.2% of Black and 8% of Other race patients received an order for a Food and Drug Administration-approved insomnia medication after diagnosis compared with 13.5% of White patients. Black patients were significantly less likely to have an order for a Food and Drug Administration-approved insomnia medication at all time points (adjusted hazard ratio [aHR] range: 0.37-0.73), and patients reporting Other race were less likely to have received an order at 2 (aHR 0.51, 95% confidence interval [CI] 0.28-0.94), 3 (aHR 0.33, 95% CI 0.13-0.79), and 4 years (aHR 0.21, 95% CI 0.06-0.71) of follow-up. Similar results were observed in a sensitivity analysis including off-label medications.
Conclusions
Patients belonging to racial minority groups are less likely to be prescribed an insomnia medication than White patients after accounting for sociodemographic and clinical factors. Further research is needed to determine the extent to which patient preferences and physician perceptions affect these prescribing patterns and investigate potential disparities in nonpharmacologic treatment.
Insomnia is a sleep disorder characterized by trouble falling asleep, staying asleep, early morning awakenings, and nonrestorative sleep, despite adequate opportunity.
Studies estimate that 10%-30% of adults in the United States suffer from the disorder, and women, older adults, those of lower socioeconomic status, and people with depression or anxiety are particularly vulnerable.
Prevalence and perceived health associated with insomnia based on DSM-IV-TR; international statistical classification of diseases and related health problems, tenth revision; and research diagnostic criteria/international classification of sleep disorders, criteria: results from the America insomnia survey.
Insomnia also disproportionately affects racial and ethnic minority groups, with Black and Hispanic people generally reporting a higher prevalence of the disorder.
Insomnia has been linked to significant morbidity, including impaired cognition, poor work performance and absenteeism, and increased risk of mood disorders.
There is also evidence that insomnia increases the risk of developing chronic illnesses such as type II diabetes, heart disease, and chronic kidney disease.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
Food and Drug Administration (FDA)-approved insomnia medications include benzodiazepines, sedative hypnotics, orexin antagonists, melatonin agonists, and barbiturates, although other medications are sometimes used off-label.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
Cognitive behavioral therapy for insomnia (CBT-I), an adaptation of traditional CBT-I developed to address sleep difficulties, is the primary nonpharmacologic treatment.
While clinical guidelines recommend CBT-I as the first-line option before trying pharmacotherapy, medication remains the most common form of treatment, potentially due to ease of access and use.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
Despite advances in treatment and the widespread availability of approved medications, untreated insomnia continues to incur considerable personal and societal cost.
Evidence suggests that racial disparities in treatment exist across many medical conditions, including depression, cardiovascular disease, and cancer, contributing to worse outcomes among minority patients.
These disparities are thought to be driven by various social and economic factors, including differential socioeconomic status, healthcare access, patient preference, and provider bias.
However, few studies have investigated potential disparities in the treatment of insomnia. To this end, our primary aim was to investigate the relationship between race and time to an order for an FDA-approved insomnia medication after insomnia diagnosis using retrospective electronic health record data from a multi-institutional health information exchange. Our secondary aims were to describe the prevalence and patterns of pharmacologic insomnia treatment in the study population.
Participants and methods
This retrospective cohort study was approved by the Institutional Review Board of Indiana University (#11732) and adheres to the reporting standards described in the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.
Study data were obtained from the Indiana Network for Patient Care (INPC), a regional health information exchange that receives information from seven health systems, 19 hospitals, laboratories, radiology centers, and primary care providers across the midwestern United States. The INPC network includes private, public, and safety net institutions, and its database contains demographics, diagnosis, laboratory, medication, medical notes, and healthcare utilization data.
Using the INPC, we identified adult patients (≥18 years) with a new insomnia diagnosis code (International Classification of Diseases, 9th or 10th clinical modification (ICD-9/10-CM)) between January 1, 2011 and December 31, 2019. Patients with an insomnia ICD-9/ICD-10-CM code or insomnia medication before 2011 were considered to have prevalent insomnia and excluded. Patients without any medication and/or diagnosis data during the study period were also excluded. Eligible patients were required to have at least one encounter in the INPC per year from 2010 to 2019 to ensure there was at least one year of data available before insomnia diagnosis. The sample size was determined by the number of patients who met the above criteria.
Data sources and variables
Age, sex, race, ethnicity, address, insurance, diagnosis, healthcare visit, and medication order data from 2010 to 2019 were extracted from the INPC. Age was defined at the time of insomnia diagnosis. Patient-reported race, the exposure of interest, was recorded as individual categories but grouped as “White,” “Black,” “Other,” and “unknown” for analysis purposes. Insurance status, defined as the insurance type reported at the healthcare visit that occurred closest to the date of insomnia diagnosis, was categorized as “government” (Medicare and/or Medicaid), “commercial,” and “other/unknown” and included as an individual-level social determinant of health. The national Area Deprivation Index (ADI) percentile for each patient was obtained using address data and the Neighborhood Atlas, which is available for free at University of Wisconsin School of Medicine and Public Health’s Neighborhood Atlas website (https://www.neighborhoodatlas.medicine.wisc.edu). The ADI is a validated composite measure of neighborhood disadvantage based on domains of income, education, employment, and housing quality at the census block group level, and was included as a measure of structural inequity.
ADI values range from 0 to 100 with higher values indicating more deprivation. We calculated Elixhauser mortality score for the year before insomnia diagnosis using ICD-9-CM/ICD-10-CM codes with Van Walravan weights.
The Elixhauser mortality score is a summary score derived from the Elixhauser Comorbidity Index, a popular method of categorizing comorbidities using ICD codes, that has been validated to predict mortality.
To adjust for the number of comorbidities, we derived a count variable that summed the number of unique ICD codes (at the three-digit level) a patient had in the year before diagnosis. Recent history of substance abuse was defined as having at least one ICD-9-CM/ICD-10-CM code for substance abuse in the medical record in the year before diagnosis. Type of visit associated with the initial insomnia diagnosis was determined from the encounter records and categorized as “emergency department or inpatient,” “outpatient,” or “other/unknown” for the analysis. Time to an insomnia medication, the outcome of interest, was defined as the number of days to an order for an FDA-approved insomnia medication after diagnosis, or days between diagnosis date and January 1, 2020 if no FDA-approved medication was ever ordered. FDA-approved insomnia medications included zolpidem, suvorexant, butabarbital, quazepam, estazolam, flurazepam, triazolam, tasimelteon, eszopiclone, temazepam, ramelteon, secobarbital, and zaleplon. For the purposes of this study, doxepin was not considered an FDA-approved insomnia medication as it is only insomnia-specific at doses of ≤6 mg
and dose information was not consistently available in the electronic health record data.
Statistical analysis
Descriptive statistics were expressed as means and standard deviations for normally distributed continuous variables, medians and interquartile ranges (IQR) for non-normally distributed continuous variables, and frequencies and percentages for categorical variables. Univariate comparisons were completed using Χ2 tests for categorical variables and Kruskal-Wallis tests for non-normal continuous variables. We used a conditional inference tree (CIT) to explore potential joint and nonlinear effects of variables. Potential nonlinearity identified by the CIT was subsequently validated using a multivariable fractional polynomial approach and likelihood ratio testing. Among the continuous variables (age, Elixhauser mortality score, national ADI percentile, and number of comorbidities), the best functional form for age and Elixhauser mortality score was quadratic, the best form for national ADI percentile was linear, and the best form for number of comorbidities was x−2.
A Cox frailty model was used to examine the relationship between race and time to an order for an FDA-approved insomnia medication after adjusting for sociodemographic and clinical factors. Clinic ID number was included as a random effect to account for clustering. The preliminary model included race, sex, ethnicity, age, national ADI percentile, insurance type, recent history of substance abuse, Elixhauser mortality score, type of encounter associated with the insomnia diagnosis, and number of comorbidities. Of note, the age and Elixhauser mortality score variables were centered to reduce the correlation between the linear and quadratic terms. We explored potential multiplicative interactions between the following pairs of variables: ADI and sex, ADI and race, ADI and age, sex and race, age and race, and age and ethnicity. After adding all interaction terms to the preliminary model, manual backward selection was performed by removing the least significant interaction term in an iterative fashion. After completing the selection procedure, no interaction terms were significant at α = 0.05. Thus, the final Cox model included the same variables as the preliminary model described above. After generating the final model, the proportional hazards assumption was examined for each variable using log-log plots.
We also performed two sensitivity analyses. In the first, we examined how race estimates changed when the definition of insomnia medication was expanded to include other medications commonly used to treat insomnia: doxepin, trazodone, amitriptyline, mirtazapine, melatonin, and chloral hydrate. Melatonin is an over-the-counter sleep aid, doxepin is used for insomnia at specific doses, chloral hydrate is a sedative, and trazodone, mirtazapine, and amitriptyline are sedating antidepressants sometimes used off-label for insomnia.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
In the second sensitivity analysis, we explored how estimates would change if the patients with missing ADI data (3.9%, n = 372) were assumed to have the highest possible value of ADI (100). We decided to use the highest ADI value in this sensitivity analysis because patients experiencing homelessness, who presumably have high deprivation, often have no address data and would therefore be missing ADI. Two-sided p values<.05 were considered statistically significant and all analyses were completed using SAS 9.4 (Cary, NC) except for the CIT, which was modeled using R version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Study sample
There were 250,159 patients with at least one healthcare encounter per year in the INPC during the study period, 12,336 of whom had a new insomnia ICD-9-CM/ICD-10-CM code between 2011 and 2019. In total, 2404 of the identified insomnia cases were ineligible due to having an order for an FDA-approved insomnia medication before the date of diagnosis. Patients without ADI and/or demographics data (n = 372) were excluded, resulting in a final analysis cohort of 9557 insomnia patients (Fig. 1). A comparison of baseline characteristics between those with and without ADI data is presented in Supplementary Table A.1 and the number of patients who received an insomnia diagnosis by year is presented in Supplementary Table A.2 (Appendix A). Only 1187 (12.4%) patients received an order for an FDA-approved insomnia medication after insomnia diagnosis. By race, 6.2% of Black patients, 13.5% of White patients, 8.0% of Other race patients, and 16.7% of unknown race patients received an order for FDA-approved pharmacotherapy after insomnia diagnosis (Table 1). The median time to medication after insomnia diagnosis was 961 days, or approximately 2.6 years, and the most frequently prescribed FDA-approved insomnia medications after diagnosis were zolpidem (63.4%), temazepam (14.0%), and eszopiclone (8.6%). Supplementary Table A.3 (Appendix A) presents a comparison of the first FDA-approved sleep medications prescribed by race.
Fig. 1Participant flow diagram of eligible and analytic populations. ICD, International Classification of Diseases; INPC, Indiana Network for Patient Care; ADI, area deprivation index
The study cohort was predominantly female (69%), White (75%), non-Hispanic (62%), and the median age at the time of insomnia diagnosis was 61 years (IQR 50-71). Most patients were insured through the government (Medicare or Medicaid, 76%) and the median national ADI percentile was 63 (IQR 46-81). Patients who received an FDA-approved insomnia medication were more likely to be younger, have a lower ADI percentile (indicating less deprivation), and have nongovernment insurance (Table 2). The most frequently documented Elixhauser comorbidities in the study cohort were hypertension (45%), diabetes (26%), depression (17%), and COPD (15%).
Table 2Baseline sociodemographic and clinical characteristics of insomnia cohort
Fig. 2 displays race-stratified Kaplan-Meier survival curves of time to the first order for an FDA-approved insomnia medication after diagnosis. For the multivariable Cox regression, the race and insurance variables were found to violate the proportional hazards assumption, so race*time and insurance*time interaction terms were added to the model. The model’s intraclass correlation coefficient was 0.17, indicating that 17% of the variance in outcome is explained by clinic and 83% of the variation is attributable to other factors. Table 3 reports the hazard ratios (HRs) for the prescription of an FDA-approved insomnia medication after insomnia diagnosis by race after accounting for socioeconomic, demographic, and clinical characteristics. Of note, race, age, number of comorbidities, insurance type, and encounter type were all statistically significant in the model. Compared with White patients, Black patients were less likely to have been prescribed an FDA-approved insomnia medication at all examined time points and the hazard decreased as the length of follow-up increased (Table 3). Although the HRs for receipt of an FDA medication were not different between White and Other race patients during the first year of follow-up, Other race patients had lower hazards of being prescribed an FDA-approved insomnia medication at 2, 3, and 4 years after insomnia diagnosis (Table 3).
Fig. 2Unadjusted Kaplan-Meier survival curves by race: receipt of an FDA-approved insomnia medication after diagnosis. FDA, Food and Drug Administration
Table 3Results from multivariable Cox frailty model: time to FDA-approved insomnia medication (n = 9557)
Variable
Unadjusted HR (95% CI)
Adjusted HR (95% CI)
Race, Black versus White
Time = 1 year
1.948 (1.506-2.519)
0.731 (0.564-0.947)
Time = 2 years
0.877 (0.688-1.119)
0.584 (0.457-0.748)
Time = 3 years
0.395 (0.284-0.549)
0.468 (0.344-0.636)
Time = 4 years
0.178 (0.112-0.283)
0.374 (0.248-0.563)
Race, Other versus White
Time = 1 year
2.262 (1.424-3.592)
0.811 (0.496-1.325)
Time = 2 years
0.722 (0.385-1.356)
0.514 (0.282-0.937)
Time = 3 years
0.231 (0.084-0.631)
0.326 (0.134-0.793)
Time = 4 years
0.074 (0.018-0.309)
0.207 (0.060-0.713)
Race, Unknown versus White
Time = 1 year
3.541 (2.510-4.995)
2.106 (1.713-2.589)
Time = 2 years
1.421 (0.912-2.213)
1.291 (1.053-1.583)
Time = 3 years
0.570 (0.287-1.131)
0.791 (0.614-1.019)
Time = 4 years
0.229 (0.087-0.603)
0.485 (0.348-0.676)
HR, hazard ratio; FDA, Food and Drug Administration; CI, confidence interval.
Multivariable race estimates were adjusted for age, sex, ethnicity, national area deprivation index percentile, insurance, number of comorbidities, Elixhauser mortality score, type of visit associated with initial insomnia diagnosis, and recent history of substance abuse. Bold values indicate statistical significance at α = 0.05.
After expanding the definition of insomnia medication to include non-FDA-approved drugs, 2324 of the original 9557 patients had an insomnia medication order before their insomnia diagnosis and were excluded. Of the remaining 7233 patients, 2561 (35.4%) had an order for an insomnia medication on or after diagnosis and median time to medication was 830 days (approximately 2.3 years). In this sensitivity analysis, the most frequently prescribed insomnia medications were trazodone (42.1%), zolpidem (21.5%), amitriptyline (10.4%), and mirtazapine (9.5%). Supplementary Table A.4 (Appendix A) presents a comparison of first sleep medications prescribed (both FDA and non-FDA) by race. Fig. 3 displays race-stratified Kaplan-Meier survival curves of time to the first order for any insomnia medication after diagnosis. In this sensitivity analysis, Black patients had a significantly lower risk of being prescribed an FDA-approved insomnia medication than White patients at 2, 3, and 4 years of follow-up. Patients reporting Other race were less likely than White patients to be prescribed an insomnia medication at 3 and 4 years of follow-up (Table 4).
Fig. 3Unadjusted Kaplan-Meier survival curves by race: receipt of any insomnia medication after diagnosis
Table 4Sensitivity analysis: race estimates, time to any insomnia medication (n = 7223)
Variable
Unadjusted HR (95% CI)
Adjusted HR (95% CI)
Race, Black versus White
Time = 1 year
0.899 (0.739-1.092)
0.920 (0.803-1.054)
Time = 2 years
0.848 (0.722-0.996)
0.681 (0.584-0.793)
Time = 3 years
0.800 (0.698-0.918)
0.504 (0.408-0.622)
Time = 4 years
0.755 (0.644-0.859)
0.373 (0.280-0.496)
Race, Other versus White
Time = 1 year
1.165 (0.815-1.665)
1.046 (0.826-1.325)
Time = 2 years
1.114 (0.835-1.485)
0.798 (0.604-1.055)
Time = 3 years
1.065 (0.837-1.356)
0.609 (0.412-0.900)
Time = 4 years
1.019 (0.807-1.285)
0.464 (0.273-0.789)
Race, Unknown versus White
Time = 1 year
1.080 (0.766-1.522)
1.040 (0.799-1.353)
Time = 2 years
1.015 (0.764-1.348)
0.751 (0.541-1.042)
Time = 3 years
0.955 (0.739-1.233)
0.542 (0.336-0.875)
Time = 4 years
0.898 (0.686-1.174)
0.392 (0.203-0.756)
CI, confidence interval; HR, hazard ratio. Multivariable race estimates adjusted for age, sex, ethnicity, national area deprivation index percentile, insurance, number of comorbidities, Elixhauser mortality score, type of visit associated with initial insomnia diagnosis, and recent history of substance abuse. Bold values indicate statistical significance at α = 0.05.
In the second sensitivity analysis, the 372 patients missing ADI data were assumed to have ADI = 100 (most deprivation) and added to the 9557 patients from the main analysis (total n = 9929). The race estimates from this sensitivity analysis were very similar to those obtained in the complete case analysis (Supplementary Appendix Table A.5).
Discussion
Our results indicate that patients belonging to racial minority groups were significantly less likely to be prescribed an insomnia medication after insomnia diagnosis than White patients, even when accounting for sociodemographic and clinical factors and healthcare utilization. Black patients were significantly less likely to have been prescribed an FDA-approved insomnia medication after insomnia diagnosis at all time points, and the probability monotonically decreased with longer duration of follow-up. While there was no significant difference for Other race patients during the first year of follow-up, they were significantly less likely to be prescribed an FDA-approved medication 2, 3, and 4 years after insomnia diagnosis. When expanding the outcome definition to include medications that are commonly used off-label to treat insomnia, there was no race difference in the first year after diagnosis. However, there was a significant difference at 2, 3, and 4 years for Black patients and at 3 and 4 years for Other race patients.
To our knowledge, this is the first study to investigate racial disparities in the pharmacologic treatment of insomnia. Nevertheless, previous research provides evidence of racial and ethnic disparities in prescribing pharmacotherapy for other medical conditions. For example, Massing et al. found that African-American patients with coronary heart disease had 40% lower odds of being prescribed lipid-lowering drugs than comparable White patients.
Studies have also found that patients belonging to racial minority groups are significantly less likely to be prescribed appropriate medications across a variety of psychiatric conditions, including depression, anxiety, and schizophrenia, even after adjusting for type of diagnosis and severity.
It is worth noting that most patients did not receive an insomnia medication after diagnosis (12.4% FDA, 35.4% FDA, or non-FDA), and the few who did often had to wait years. This suggests that providers may be undertreating insomnia as a whole and prefer giving off-label medications. It is unclear whether these findings reflect widespread use of CBT-I, the perception that insomnia is not a real problem, or physicians treating insomnia as a secondary condition. In line with previous literature,
our results showed that older patients and those with comorbidities are less likely to be prescribed insomnia medications. However, even when controlling for these (and other) factors, we found that White patients are overrepresented among those who received a medication order and have a shorter period between diagnosis and prescription.
In the sensitivity analysis, there was no difference in HRs by race during the first year, but Black race became significant in years 2, 3, and 4. Taken together with the main results that did show a difference for Black patients at all time points, this suggests that Black patients are more likely to be prescribed off-label medications, and when off-label medications are included, providers are prescribing insomnia medications approximately equally across race groups within the first year. However, given that most patients (77%) did not receive an insomnia medication until a year or more after their diagnosis, the lower HRs for Black and Other race patients at later time points remain relevant.
Because having at least one healthcare visit per year during the study period was an eligibility criterion of our study, it is unlikely that differential healthcare access drove the observed results. A more probable explanation is that a combination of both patient- and provider-level characteristics contributed to the differential prescribing patterns. In terms of patient characteristics, it is possible that there is a preference to avoid sedative-hypnotic medications among patients belonging to racial minority groups, although such preferences have not been well investigated for this type of medication. The lower likelihood of being prescribed an insomnia medication did persist in the sensitivity analysis that included, among others, sedating antidepressants frequently used off-label for insomnia; some studies suggest that African-American patients may have more reservations about using antidepressants than White patients and prefer counseling for depression over medication.
Of note, most patients did not receive an FDA-approved insomnia medication after diagnosis, and the median time to an order for an FDA-approved medication was approximately 2.6 years. This could reflect physicians following the published clinical guidelines for insomnia treatment, which recommends cognitive behavioral therapy (CBT-I) as the first-line option.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
However, research shows that Black patients are less likely to initiate or complete CBT for insomnia or other mental health disorders than White patients.
suggests that CBT-I for insomnia is equally effective across racial groups, so it is unlikely that minority patients responded better to CBT-I and did not require subsequent insomnia medications. It is also unlikely that minority patients experienced less severe insomnia and did not require medications as frequently; studies have found that Black individuals often have more severe and longer-lasting insomnia than their White counterparts.
could have felt that their concerns were ignored when they initially sought help for insomnia and were advised to try CBT-I or sleep hygiene rather than being prescribed medication. As a result, they may have decided against bringing up their insomnia in future visits. Previous literature shows that Black individuals are less likely to report sleep complaints than White individuals, despite experiencing worse sleep.
While patient characteristics may have played a role in the observed results, provider biases may have also contributed to discrepancies in prescribing patterns. There is evidence that providers’ assessments and clinical judgment can be heavily influenced by patient racial and ethnic traits; studies have shown that some healthcare providers implicitly associate non-White race with negative traits, including noncompliance and medication misuse, which affect both treatment decisions and outcomes.
Provider bias can also negatively affect patient-provider communication. For example, one study reported that providers were significantly less patient-centered, more verbally dominant, and had a less positive affect when communicating with African-American patients.
Given this evidence, it is possible that biases, stereotypes, and/or communication difficulties played a role in clinicians’ decisions to prescribe insomnia medications.
Strengths and limitations
This study has several strengths. First, our data were derived from a regional health information exchange containing real-world health data from multiple healthcare systems, including those with safety net hospitals, resulting in a large and diverse study cohort. We also included the ADI as an explanatory variable in the model to identify the relationship between race and insomnia treatment when adult socioeconomic status is fixed; ADI captures multiple aspects of socioeconomic status at the census block group level and may contribute meaningful information about structural inequality not captured by individual insurance status. Because we limited the sample to patients with at least one healthcare visit per year during the study period, we were able to investigate whether pharmacologic treatment disparities exist when all patients have consistent access to healthcare.
This study also has important limitations. We were unable to determine which patients tried CBT, which is recommended as the first-line insomnia treatment. While there is evidence of racial disparities in the initiation of CBT for insomnia and other psychiatric conditions, there is no specific investigation of referral patterns.
Given that clinicians are less likely to prescribe controlled substances to Black individuals in mental health settings, it is possible that there is also a preference to refer Black patients to behavioral interventions for insomnia. In the future, we plan to use natural language processing on medical note text to investigate CBT-I referral and initiation in this cohort. While the INPC receives information from most of the major healthcare systems in Indiana, encounters that occurred at non-INPC facilities may have been missed. Additionally, patients were required to have at least one healthcare encounter per year, so the results may not generalize to patients who use healthcare less frequently. Because we used real-world EMR data not collected for research purposes, there were missing data and it is possible that missingness correlates with specific physicians and/or institutions. We defined insomnia using ICD-9/-10 diagnosis codes, so cases would have been missed if they were not documented accordingly in the EMR.
Conclusions
The results of this study suggest that there may be racial disparities in the pharmacologic treatment of insomnia, even when key sociodemographic factors are controlled and healthcare access is guaranteed. Given relative inaccessibility of CBT-I and the fact that medication remains the most common form of insomnia treatment, these findings raise questions about the adequacy of insomnia management, particularly among minority patients. Additional work is needed to determine the extent that patient preferences and physician perceptions affect prescribing patterns, investigate potential racial differences in physician referrals for CBT-I, and formulate strategies to remove barriers to insomnia treatment for minority patients.
Author contributions
Concept and design: Holler, Boustani, Campbell, Dexter, Ben Miled, and Owora, Statistical analysis: Holler and Owora, Drafting the paper: Holler, Critical revision of the paper for important intellectual content: Holler, Boustani, Campbell, Dexter, Ben Miled, and Owora.
Declaration of conflict of interest
Drs. Ben Miled and Boustani have a financial interest in DigiCare Realized and could benefit from the results of this research. Dr. Boustani serves as a Chief Scientific Officer and co-Founder of BlueAgilis; and the Chief Health Officer of DigiCare Realized, Inc. He has equity interest in Blue Agilis, Inc; DigiCare Realized, Inc; Preferred Population Health Management LLC; and MyShift, Inc (previously known as RestUp, LLC). He serves as an advisory board member for Acadia Pharmaceuticals; Eisai, Inc; Biogen; and Genentech. These conflicts have been reviewed by Indiana University and have been appropriately managed to maintain objectivity. The remaining authors declare no competing interests.
Role of the funding source
Merck Sharp & Dohme Corp. was not involved in the design and conduct of the study, the analysis and interpretation of the data, preparation, review, or approval of the paper, nor were they involved in the decision to submit the paper for publication.
Funding
This work was supported in part by Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Rahway, NJ, USA.
Prevalence and perceived health associated with insomnia based on DSM-IV-TR; international statistical classification of diseases and related health problems, tenth revision; and research diagnostic criteria/international classification of sleep disorders, criteria: results from the America insomnia survey.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
Clinical practice guideline for the pharmacologic treatment of chronic insomnia in adults: an American Academy of Sleep Medicine Clinical Practice Guideline.
00032-3&id=gr1.jpg){kind=link}
00032-3&id=gr2.jpg){kind=link}
00032-3&id=gr3.jpg){kind=link}