Lifetime history of insomnia and hypersomnia symptoms as correlates of alcohol, cocaine, and heroin use and relapse among adults seeking substance use treatment in the United States from 1991 to 1994

Abstract

Aims

To examine the association between a lifetime history of insomnia and hypersomnia compared with no sleep disturbance and substance use patterns and amounts before and after a substance use treatment episode.

Design

Secondary analysis of data from the Drug Abuse Treatment Outcome Studies conducted from 1991 to 1994.

Setting

Data were collected at 96 substance use treatment programs in 11 United States cities including short-term in-patient, long-term residential, methadone maintenance, and outpatient drug-free treatment modalities.

Participants

Study samples included 7,168 adults at treatment entry and 2,965 at 12 months post-treatment entry whose primary substance use at entry was alcohol (14.7%), cocaine (62.7%), or heroin (22.6%).

Measurements

Lifetime history of insomnia and hypersomnia was assessed via self-report. Type and frequency of substance use were assessed at treatment entry. Substance use was also assessed 12 months following treatment completion. Associations were examined using linear and logistic regression with age, sex, race, education level, depression history, treatment modality, and in-treatment substance use as covariates.

Findings

Lifetime history of insomnia, hypersomnia, both or neither was reported by 26.3%, 9.5%, 28.0% and 36.2% of participants, respectively. Compared with no sleep disturbance, lifetime insomnia and hypersomnia were associated at treatment entry with unique substance use patterns and a higher frequency of any substance use (p < .001). All types of sleep disturbance were associated with higher rates of cocaine use at 12-month post-entry (ORs: 1.30–1.57).

Conclusions

There is evidence of an adverse association between substance use and sleep disturbance including higher frequency of all substance use before substance abuse treatment and higher rates of cocaine use after a treatment episode.

Introduction

Substance use is a major public health concern that interferes with mental and physical health, as well as overall functioning. Substantial evidence indicates that substance use and sleep disturbance are related, and that this relationship is likely bidirectional (1–8). Although many substances are associated with initial stimulatory or soporific effects (9), chronic substance use severely impairs sleep duration and quality, especially during periods of withdrawal. Additionally, despite the availability of effective treatments for sleep disturbance, there is evidence that illicit substances may be used in an attempt to alleviate sleep problems (10–12). Accordingly, sleep disturbance in the context of substance use is an important research focus, with potential implications for understanding and treating substance use and preventing relapse.

The relationship between sleep disturbance and substance use is complex in that substance use has been associated with both insomnia and hypersomnia. Insomnia is a widespread problem that affects one-third of the general population (13,14). Insomnia includes problems initiating and maintaining sleep, and specifically includes impairment related to difficulty falling asleep, staying asleep, or waking earlier than intended (15). Among individuals with insomnia, 30.0% also meet diagnostic criteria for comorbid alcohol use disorder and 14.4% experience comorbid substance use disorder (16). Insomnia is also associated with higher odds of meeting criteria for alcohol use disorder (OR=2.0, 95% CI [1.3–3.0]) and substance use disorder (OR=2.1, 95% CI [1.2–3.5]; (16). These results are consistent with, albeit higher than, comorbidity estimates from the NIMH Epidemiological Catchment Area program. Ford and Kamerow (17) reported that among individuals with insomnia, 7.0% also experience comorbid alcohol use disorder and 4.2% experience comorbid substance use disorder.

Hypersomnia involves daytime sleepiness or fatigue, daytime naps or sleep periods, protracted sleep periods (9 or more hours) that feel unrefreshing, and sleep inertia (15). Hypersomnia is associated with higher odds of alcohol use disorder (OR=2.9, 95% CI [1.7–4.8]) and substance use disorder (OR=3.6, 95% CI [2.0–6.4]) (16). Additionally, among individuals with hypersomnia, 36.1% meet criteria for alcohol use disorder and 22.9% meet criteria for substance use disorder (16). However, comorbidity estimates from the NIMH Epidemiological Catchment Area program are comparatively lower with 6.2% of individuals having both hypersomnia and alcohol use disorder and 4.7% of individuals having both hypersomnia and substance use disorder (17).

The relationship between sleep disturbance and substance use also appears to depend on the type of substance used (6–8), most notably between substances such as alcohol, cocaine, and heroin. There is strong evidence that sleep disturbance and alcohol are related. In particular, insomnia is present during various phases of alcohol use (10,18–20), and alcohol is reportedly used by 45% of patients with alcohol dependence as self-medication for sleep problems (10). Additionally, cocaine is associated with longer sleep onset latency and decreased sleep efficiency compared to age-matched controls as measured by polysomnography (21). The first four days of cocaine withdrawal are initially associated with insomnia symptoms, but then followed by increased desire for sleep and hypersomnia (22). Notably, 80% of individuals with increased desire for sleep during early-phase cocaine withdrawal self-medicated with alcohol and opiates (22). Although there is less data on sleep disturbance and heroin, both objective and subjective sleep disturbance, including insomnia symptoms, are related to heroin use (3,23–27).

Sleep disturbance may also be a risk factor for substance use relapse (28). Insomnia symptoms, and long sleep onset latency in particular, have been consistently linked to alcohol use relapse (10,29,30). There is also a growing literature on the relationship between sleep and cocaine relapse. A recent report found evidence that improvements in slow-wave sleep mediated the relationship between modafinil treatment and a higher rate of cocaine abstinence (31). Further, sleep disturbance has been linked to cocaine cravings after long-term withdrawal in rats, which could increase the likelihood of relapse (32). Less research has examined if hypersomnia or insomnia are prospectively related to heroin relapse (see similar conclusion in Brower & Perron, 2010). However, there is evidence that improvements in sleep problems may predict abstinence in opioid-dependent patients (33). Additionally, an experience sampling study also provided evidence that poorer sleep quality is related to more drug cravings among patients in residence for substance treatment (34). These findings highlight the importance of sleep disturbance in the context of substance use relapse. To the best of our knowledge, studies have not yet sought to systematically examine the role of insomnia or hypersomnia symptoms on substance use relapse in the context of a large, community-based substance use treatment study.

The aim of the present study was to test the hypothesis that adults seeking substance use treatment in the United States from 1991 to 1994 that report a lifetime history of insomnia symptoms, hypersomnia symptoms, or both insomnia and hypersomnia symptoms will have differing patterns of alcohol, cocaine, and heroin use compared to no sleep disturbance. This study will also test the hypothesis that individuals that reported a lifetime history of insomnia symptoms, hypersomnia symptoms, or both insomnia and hypersomnia symptoms at pre-treatment will use substances more frequently than individuals that reported no sleep disturbance symptoms. Last, this study will test the hypothesis that individuals that reported a lifetime history of insomnia symptoms, hypersomnia symptoms, or both insomnia and hypersomnia symptoms will be more likely to use alcohol, cocaine, and heroin 12-months post-treatment.

Methods

Design

Data come from the Drug Abuse Treatment Outcome Study (DATOS), which was a multisite prospective study on community-based substance use treatment from 96 programs in 11 United States cities from 1991 to 1994. The methods employed by DATOS are described elsewhere (35). Participants were included in this sample if sleep data was available and the primary substance use problem was alcohol (n=1,057), cocaine (n=4,494), or heroin (n=1,617). This resulted in 7,168 adults (2,451 female, mean age=32.8) with data at treatment entry. A follow-up sample was randomly selected from participants who completed the pre-treatment intake interview (35,36). Participants were interviewed 12-months after treatment termination, or 24-months after treatment termination for long-term methadone treatment. Of those targeted for the 12-month follow-up, 4,229 were eligible (88.4% of those targeted), 3,147 were located (74.4% of those eligible), and 2,966 were successfully interviewed (70.1% of those eligible) (35,36). One participant was not included in the present 12-month follow-up sample due to missing pre-treatment sleep data. A comparison of participants who completed only the pre-treatment intake interview and those that also completed 12-month follow-up interviews indicated that these samples have similar characteristics and are generally comparable (36). Participation in the DATOS study was voluntary, and individuals provided verbal and written informed consent (36).

Measures

Substance use (outcomes)

Primary substance use problem was determined from the pre-treatment substance use assessment, and was recoded by DATOS data managers. Primary substance use problem included alcohol (n=1,057), cocaine (n=4,494), or heroin (n=1,617) use. Substance use frequency was determined from the pre-treatment substance use assessment, and was recoded by DATOS data managers. Responses included “no use,” “less than once/week,” “1–3 times/month,” “1–2 times/week,” “3–4 times/week,” “5–6 times/week,” “daily or almost every day,” “2–3 times/day,” and “4+ times/day.” Post-treatment substance use was determined from a positive response to the question, “During the past 12 months, have you used (substance)?” assessed at the 12-month post-treatment follow-up. Post-treatment substance use was restricted to alcohol, cocaine, and heroin use.

Sleep disturbance (predictors)

A lifetime history of sleep disturbance symptoms was assessed at pre-treatment. Sleep disturbance symptoms were not assessed 12-months post-treatment. A lifetime history of insomnia symptoms was determined by a positive response to any of the following questions, “Have you ever had 2 weeks or more when nearly every night you had trouble falling asleep?”, “Have you ever had 2 weeks or more when nearly every night you had trouble staying asleep?”, or “Have you ever had 2 weeks or more when nearly every night you had trouble waking up too early?”. These questions reflect difficulties with sleep onset latency (SOL), waking after sleep onset (WASO), and early morning awaking (EMA), which are core components of the diagnostic criteria for insomnia (15). A lifetime history of hypersomnia symptoms was determined by a positive response to “Have you ever had 2 weeks or longer when nearly every day you were sleeping too much?”. A negative response to any of these insomnia or hypersomnia questions was categorized as no lifetime history of sleep disturbance symptoms. Lifetime history of sleep disturbance was coded as only insomnia symptoms (n=1,885), only hypersomnia symptoms (n=680), both hypersomnia and insomnia symptoms (n=2,008), or no sleep disturbance symptoms (n=2,595). No sleep disturbance symptoms was the reference category for all analyses.

Covariates

A lifetime history of depression symptoms was determined by participant responses to, “In your lifetime, have you had at least 2 weeks during which you felt very sad, blue, depressed, or you lost interest and pleasure in things you usually cared about or enjoyed?” Previous studies indicate that depression is associated with substance use (37–40) and sleep disturbance (41–43). Substance use during treatment was assessed at the 12-month post-treatment interview with the following question, “While you were in treatment, (on the average) how often did you use (substance)?”. Participants received treatment in four modalities: short-term inpatient treatment (n=2,425), long term residential treatment (n=2,123), outpatient drug-free treatment (n=1,426), or methadone maintenance treatment (n=1,194). Age, sex, race, age, education level, and depression history were measured at the pre-treatment assessment. Demographic characteristics by sleep disturbance type are displayed in Table 1.

Table 1.

Descriptive characteristics of sleep, substance use, and demographic variables by sleep disturbance type assessed at pre-treatment and substance use at 12-months post-treatment

	Insomnia symptoms		Hypersomnia symptoms		Both insomnia and hypersomnia symptoms		No sleep disturbance
Characteristic	Mean or n	SD or %	Mean or n	SD or %	Mean or n	SD or %	Mean or n	SD or %
Pre-treatment sleep disturbance	1885	26.3%	680	9.5%	2008	28.0%	2595	36.2%
12-month sleep disturbance	784	26.4%	289	9.7%	784	26.4%	1108	37.4%
Age	34.0	7.3	30.7	6.8	32.1	7.0	32.9	7.2
Gender (n=female)	623	33.1%	248	36.5%	825	41.1%	755	29.1%
Race
African-American	944	50.1%	331	48.7%	890	44.3%	1,485	57.2%
Caucasian	682	36.2%	255	37.5%	831	41.4%	744	28.7%
Hispanic	213	11.3%	79	11.6%	239	11.9%	300	11.6%
Other	46	2.4%	15	2.2%	48	2.4%	66	2.5%
Education Level
Grade school	92	4.9%	21	3.1%	105	5.2%	95	3.7%
High school	583	30.9%	206	30.3%	661	32.9%	824	31.8%
High school degree	733	38.9%	270	39.7%	750	37.4%	994	38.3%
Some college	317	16.8%	132	19.4%	329	16.4%	464	17.9%
Associate’s degree	81	4.3%	21	3.1%	79	3.9%	109	4.2%
Bachelor’s degree	65	3.5%	22	3.2%	77	3.8%	95	3.7%
Advanced degree	13	0.7%	8	1.2%	7	0.3%	14	0.5%
Depression history	1,057	56.3%	273	40.3%	1,437	72.0%	565	21.8%
Substance use type
Alcohol	310	16.4%	102	15.0%	313	15.6%	332	12.8%
Cocaine	1,069	56.7%	475	69.9%	1,271	63.3%	1,679	64.7%
Heroin	506	26.8%	103	15.1%	424	21.1%	584	22.5%
Substance use frequencya	5.3	2.3	4.9	2.2	5.3	2.2	4.8	2.3
Substance use during treatment assessed at 12-month follow-upb
Alcohol (use/no use)	160/595	21.2%	46/234	16.4%	149/606	19.7%	206/849	19.5%
Cocaine (use/no use)	139/614	18.5%	44/234	15.8%	134/617	17.8%	182/869	17.3%
Heroin (use/no use)	120/633	17.3%	29/251	10.4%	98/654	13.0%	143/910	13.6%
Substance use in past 12-months assessed at 12-month follow-upb
Alcohol (use/no use)	438/346	55.9%	160/129	55.4%	442/342	56.4%	595/512	53.7%
Cocaine (use/no use)	301/479	38.6%	118/170	41.0%	331/453	42.2%	362/744	32.7%
Heroin (use/no use)	157/623	20.1%	41/247	14.2%	122/661	15.6%	195/911	17.6%

^a4 = 3–4 times/week, 5 = 5–6 times/week, 6 = daily or almost every day;

^bPercentages correspond to number of cases reporting use divided by total number of cases by sleep disturbance type.

Data analysis

The aims of the study were tested with logistic and linear regression. The first aim was tested with binomial logistic regression with the logit command in STATA 14.1 (44). The independent variable (IV) was pre-treatment sleep disturbance group (0=no sleep disturbance, 1=insomnia symptoms, 2= hypersomnia symptoms, 3=both insomnia and hypersomnia symptoms) and the dependent variable (DV) was primary substance use problem at pre-treatment. Given that the primary substance use problem variable does not have a natural reference group for comparisons (e.g., “no substance use”), the variable was coded as follows: 0=cocaine, 1=alcohol; 0=cocaine, 1=heroin; or 0=heroin, 1=alcohol. The second aim was tested with linear regression with the regress command from STATA 14.1 (44). The IV was pre-treatment sleep disturbance group and the DV was pre-treatment substance use frequency. The third aim was tested with a binomial logistic regression with the logit command from STATA 14.1 (44). The IV was pre-treatment sleep disturbance group and the DV was substance use 12-months post-treatment (0=absence, 1=presence) for alcohol, cocaine, and heroin. All models also included age, sex, race, age, education level, and depression history assessed at pre-treatment as covariates. Analyses for the third aim also included treatment modality and in-treatment substance use as covariates.

Results

Lifetime history of sleep disturbance and primary substance use problem at pre-treatment

Descriptive statistics for pre-treatment sleep disturbance and substance use variables are presented in Table 1. A lifetime history of insomnia symptoms compared to no sleep disturbance symptoms was significantly related to a higher likelihood of reporting alcohol use compared to cocaine use and a higher likelihood of reporting heroin use compared to cocaine use (Table 2). A lifetime history of hypersomnia symptoms compared to no sleep disturbance symptoms was significantly related to a higher likelihood of reporting cocaine use and alcohol use compared to heroin use (Table 2). A lifetime history of reporting both insomnia and hypersomnia symptoms compared to no sleep disturbance symptoms was not significantly related to a higher likelihood of reporting a particular substance (Table 2).

Table 2.

Comparison of substance use and substance use frequency for participants with a lifetime history of insomnia, hypersomnia symptoms, or both insomnia and hypersomnia symptoms compared to no sleep disturbance symptoms assessed at pre-treatment.

	β	SE	z/t	p	OR	95% C.I.
Insomnia symptoms
Alcohol (reference: cocaine)	0.315	0.096	3.270	.001	1.37	[1.13, 1.66]
Heroin (reference: cocaine)	0.282	0.082	3.430	.001	1.33	[1.13, 1.56]
Alcohol (reference: heroin)	0.059	0.107	0.560	.579	1.06	[0.86, 1.31]
Substance use frequency	0.424	0.071	5.940	<.001	-	[0.28, 0.56]
Hypersomnia symptoms
Alcohol (reference: cocaine)	0.068	0.130	0.530	.599	1.07	[0.83, 1.38]
Cocaine (Reference: heroin)	0.370	0.128	2.900	.004	1.45	[1.13, 1.86]
Alcohol (reference: heroin)	0.369	0.162	2.270	.023	1.45	[1.05, 1.99]
Substance use frequency	0.050	0.098	0.510	.609	-	[−0.14, 0.24]
Insomnia and hypersomnia symptoms
Alcohol (reference: cocaine)	0.117	0.099	1.180	.238	1.12	[0.93, 1.37]
Heroin (reference: cocaine)	−0.007	0.088	−0.080	.934	0.99	[0.84, 1.18]
Alcohol (reference: heroin)	0.096	0.113	0.850	.398	1.10	[0.88, 1.37]
Substance use frequency	0.508	0.074	6.900	<.001	-	[0.36, 0.65]

Dependent variables: alcohol use, cocaine use, heroin use, or substance use frequency assessed at pre-treatment. Independent variables: a lifetime history of insomnia, hypersomnia symptoms, or both insomnia and hypersomnia symptoms compared to no sleep disturbance symptoms assessed at pre-treatment. Covariates: age, sex, race, education level, and depression history assessed at pre-treatment.

Lifetime history of sleep disturbance and substance use frequency at pre-treatment

A lifetime history of insomnia symptoms as well as reporting a lifetime history of both insomnia and hypersomnia compared to no sleep disturbance symptoms were significantly related to greater substance use frequency (Table 2). A lifetime history of hypersomnia symptoms was not significantly related substance use frequency (Table 2).

Pre-treatment sleep disturbance and substance use 12-months post-treatment

Alcohol, cocaine, and heroin use 12-months post-treatment for each sleep disturbance group are reported in Table 1. Compared to no sleep disturbance symptoms, participants that reported a lifetime history of insomnia symptoms at pre-treatment had a 30% increase in the odds of reporting cocaine use 12-months post-treatment (Table 3). In addition, individuals that reported a lifetime history of hypersomnia symptoms at pre-treatment compared to no sleep disturbance symptoms had a 52% increase in the odds of reporting cocaine use 12-months post-treatment (Table 3). Participants that reported a lifetime history of both insomnia and hypersomnia symptoms at pre-treatment also had a 57% increase in the odds of reporting cocaine use 12-months post-treatment (Table 3). In contrast, all three types of sleep disturbance assessed at pre-treatment (insomnia, hypersomnia, or both insomnia and hypersomnia) were not significantly related to alcohol or heroin use 12-months post-treatment (Table 3).

Table 3.

Alcohol, cocaine, and heroin use at 12-months post-treatment predicted by lifetime history of insomnia or hypersomnia symptoms compared to no sleep disturbance reported at pre-treatment.

	β	SE	z	p	OR	95% C.I.
Insomnia symptoms
Alcohol	0.104	0.107	0.970	.330	1.11	[0.90, 1.37]
Cocaine	0.260	0.112	2.330	.020	1.30	[1.04, 1.62]
Heroin	0.024	0.167	0.140	.886	1.02	[0.74, 1.42]
Hypersomnia symptoms
Alcohol	0.099	0.145	0.680	.495	1.10	[0.83, 1.47]
Cocaine	0.421	0.150	2.810	.005	1.52	[1.14, 2.04]
Heroin	−0.083	0.243	−0.340	.732	0.92	[0.57, 1.48]
Insomnia and hypersomnia symptoms
Alcohol	0.134	0.112	1.200	.231	1.14	[0.92, 1.42]
Cocaine	0.450	0.116	3.890	<.001	1.57	[1.25, 1.97]
Heroin	−0.214	0.182	−1.170	.241	0.81	[0.57, 1.15]

Dependent variables: Alcohol, cocaine, and heroin use (presence or absence) at 12-months post-treatment. Independent variables: a lifetime history of insomnia, hypersomnia symptoms, or both insomnia and hypersomnia symptoms compared to no sleep disturbance symptoms assessed at pre-treatment. Covariates: age, sex, race, education level, depression history, treatment modality, and in-treatment substance use.

Discussion

The present study tested hypotheses about the effect of a lifetime history of insomnia symptoms, hypersomnia symptoms, or both insomnia and hypersomnia symptoms compared to no sleep disturbance on alcohol, cocaine, and heroin use patterns, substance use frequency, and substance use 12-months post-treatment in a large-scale sample of adults seeking substance use treatment in the United States. A lifetime history of insomnia symptoms was associated with a greater likelihood of reporting alcohol and heroin use compared to cocaine use. These findings are consistent with other research that has shown that insomnia is related to alcohol and heroin use (3,10,18,19,24). This is also one of the first studies to indicate that insomnia symptoms may be more strongly related to alcohol and heroin use than cocaine use for adults seeking community-based substance use treatment in the United States. A lifetime history of hypersomnia symptoms was related to a greater likelihood of reporting cocaine and alcohol than heroin use. One possibility is that cocaine may be used to self-medicate hypersomnia symptoms; however, it may also be the case that cocaine withdrawal results in hypersomnia symptoms, which are then self-medicated with alcohol (22). This study adds to a growing literature on hypersomnia and substance use that has been previously limited by low rates of substance use (45). The current study did not find evidence that reporting both a lifetime history of insomnia and hypersomnia symptoms was related to a particular substance use pattern. Differentiating types of sleep disturbance such as insomnia and hypersomnia may be an important consideration in determining an individual’s primary substance use problem.

Compared to no sleep disturbance symptoms, individuals that reported a lifetime history of insomnia symptoms as well as both a lifetime history of insomnia and hypersomnia used substances more frequently. One explanation for these findings is that individuals who report insomnia symptoms may use substances to self-medicate the daytime impairment associated with insomnia (46,47). Alternatively, substance use and withdrawal are associated with insomnia symptoms (15), and it may be that the present findings reflect an increase in insomnia symptoms in combination with increased substance use. In contrast, a lifetime history of hypersomnia symptoms compared to no sleep disturbance symptoms was not related to substance use frequency. One possibility is that insomnia symptoms are a more salient sleep disturbance symptom given that the acute stages of substance use and withdrawal are connected with insomnia (19,21,22,48). Additionally, reporting both a lifetime history of insomnia and hypersomnia was related to higher substance use frequency, and may suggest that more sleep disturbance is associated with more frequent substance use.

A lifetime history of insomnia symptoms and hypersomnia symptoms may also be risk factors for cocaine relapse. Insomnia symptoms, hypersomnia symptoms, and both insomnia and hypersomnia symptoms were related to a higher likelihood of using cocaine 12-months post-treatment. These results add to a growing literature linking sleep and cocaine abstinence (31,49). The present study builds on these prior findings by providing evidence that a lifetime history of insomnia and hypersomnia symptoms are prospectively associated with post-treatment cocaine use in the context of a large, multisite community-based substance treatment study. One explanation for these findings is that substance use after treatment may be an attempt at alleviating withdrawal symptoms such as insomnia and hypersomnia (15,48,50). Indeed, there is evidence that sleep does not return to normal after substance use abstinence (10,48,51,52). Future research will be necessary to confirm the potential importance of a lifetime history of insomnia and hypersomnia symptoms on cocaine relapse, as well as identify the mechanisms involved in this relationship. Neither a lifetime history of insomnia nor hypersomnia symptoms reported at pre-treatment significantly predicted heroin or alcohol use 12-months post-treatment. Regarding heroin, the lack of an effect may be due to the relatively lower rate of post-treatment heroin use (17.4%). The absence of an effect for post-treatment alcohol use is surprising given prior evidence for this relationship (10), and further highlights the need for prospective studies with validated measures of sleep disturbance.

Although the findings from the present study provide further evidence for the relationship between sleep disturbance and substance use and relapse, there are limitations to this study. First, this study was cross-sectional and utilized retrospective self-reported sleep disturbance and substance use. Hence, it is not possible to determine if sleep disturbance causes substance use, if substance use causes sleep disturbance, or if other variables may account for these relationships. Experimental and prospective studies are needed to investigate sleep disturbance symptoms and substance use in order to determine the directionality of these results. Objective measures of sleep disturbance and substance use should also be a priority in future research. Second, the present study utilized data collected between 1991 and 1994, which may limit the generalizability of these effects to the current era. However, it is notable that evidence for a linkage between sleep disturbance and substance use has continued to emerge over time (53,54). Third, a substantial number of pre-treatment participants were not included in the 12-month post-treatment assessment. Although missing data is common in longitudinal substance use treatment studies (55), findings from the 12-month follow-up sample should be interpreted with caution as they may not be representative of the pre-treatment sample. Fourth, this study did not evaluate insomnia or hypersomnia diagnosis, but rather used questions that target a lifetime history of insomnia and hypersomnia symptoms. Hence, it is not possible to determine the temporal relationship between sleep disturbance and substance use. Further, the endorsement of sleeping too much may also reflect extended time in bed rather than hypersomnia (56,57). While these questions do reflect the core diagnostic criteria for insomnia as well as one core symptom of hypersomnia (15), future studies will benefit from implementing a prospective design with validated measures of insomnia and hypersomnia symptoms or a diagnostic interview that evaluates insomnia and hypersomnia diagnosis. Finally, the DATOS study included 96 treatment sites. However, treatment site identifiers were not included in the DATOS dataset due to confidentiality risks (58). As such, the findings from this study should be interpreted with caution as p-values and confidence intervals may be inaccurately estimated without accounting for site-level clustering.

In conclusion, these findings provide additional support for an association between substance use and sleep disturbance. Specifically, the study showed that lifetime history of insomnia and hypersomnia was associated with higher frequency of all substance use at treatment entry and higher rates of cocaine use at the 12-month post-treatment assessment among adults entering substance use treatment in the United States. Given the potential importance of this association, future research should use prospective designs to determine if screening for sleep disturbance can inform substance use treatment and relapse prevention efforts. Additionally, research may also benefit from examining treatments that target sleep disturbance in the context of substance use, particularly given evidence that sleep medications can improve sleep disturbance associated with marijuana withdrawal (59).