End-of-day reports of daily hassles and stress in men and women with opioid-use disorder: Relationship to momentary reports of opioid and cocaine use and stress

Abstract

Background and Aims:

Stress can be validly assessed “live” or by a summary evaluation of the very recent past. Using smartphone-based ecological momentary assessment (EMA) combined with end-of-day (EOD) entries, we assessed the association between daily hassles, stressful events and use of opioids and cocaine, in opioid- and cocaine-using men and women

Methods:

For up to 16 weeks, 161 outpatients in opioid-agonist treatment who reported cigarette smoking carried smartphones on which they reported stressful events (SEs) and drug use (DU) and completed an EOD questionnaire to report hassles encountered throughout the day, current perceived stress, cigarettes/day, and current mood. We compared EOD responses on days with and without SE and DU reports and on days when thrice-weekly urine drug screens indicated opioid or cocaine use or abstinence.

Results:

Participants (N=161) made 11,544 EOD entries; EMA SEs were reported on 861 (7.5%) days, and DUs on 1685 (14.6%) days. The most frequently reported hassles in EOD entries were “not enough money” (31.4% of daily reports) and maintaining abstinence (18.7%). Total EOD hassles showed small but statistically significant associations [odds ratios (95% CIs)] with EMA SEs [1.09 (1.06–1.13)], DUs [1.08 (1.06–1.10)], and urine-positive opioid [1.06 (1.04–1.09)] and cocaine [1.03 (1.00–1.06)] results. Men and women had similar rates (mean/day (SD)) of hassles: men 2.25 (3.55); women 2.55 (3.76) (F_1,159=0.53, p=0.47).

Conclusions:

Daily hassles, reported at the end of the day, are associated with both same-day stressful events and drug use. Monitoring hassles and devising specific coping strategies might be useful therapeutic targets.

1. Introduction

Research across a broad range of methodologies and types of participants has shown that stress is an important factor in drug use and relapse. Supporting evidence comes from laboratory-animal studies showing that stressors increase drug seeking (Bossert et al., 2013), human laboratory studies showing that greater susceptibility to an experimental stressor predicts shorter time to relapse after release from inpatient treatment (Sinha et al., 2006; Back et al., 2010), interview-based studies of reasons for relapse in patients across various types of addictions (Marlatt and Gordon, 1985), and epidemiological studies showing, for example, that economic losses in the 2008–2009 recession were associated with increases in alcohol consumption and alcohol-related problems (Mulia et al., 2014).

Our research clinic has been examining stress and addiction in a naturalistic way, using smartphone-based ecological momentary assessment (EMA) in patients being treated with opioid agonists for opioid-use disorder (Preston and Epstein, 2011; Preston et al., 2017). Because recovery from addiction is a day-to-day effort, the relevant stressors are not restricted to major events such as the loss of a job or the death of a loved one; day-to-day hassles have important effects of their own (DeLongis et al., 1988). These effects have been studied at the biological level. For example, self-reported daily hassles have been associated with higher inflammatory markers and coagulability (Jain et al., 2007). Using EMA, we have shown that stress ratings are significantly higher in the presence of drug-use triggers than in their absence, and that drug craving is strongly associated with the severity of both background stress (measured at randomly prompted times) and with the occurrence of discrete stressful events (Preston and Epstein, 2011; Preston et al., 2017). We have also shown a strong association between tobacco smoking and cocaine use and craving (Epstein et al., 2010).

In the same participants whose EMA data we have reported previously (Preston et al., 2017), we used smartphones to administer an end-of-day (EOD) questionnaire at each participant’s usual bedtime. EOD assessments can be more extensive than ambulatory EMA assessments because they do not interrupt ongoing activity. Conceptually, EOD assessment complements EMA by capturing the gestalt of a day’s experience: summaries of very recent events or feelings can differ from live reports of the same events or feelings, and assessment on both levels can offer insights relevant to behavioral health (Ariely and Carmon, 2000). Our EOD assessments used items adapted from the Hassles scale (DeLongis et al., 1988) and the Perceived Stress Scale (Ingram et al., 2016), standard measures that would not have been well suited to momentary administration several times per day.

In this paper, we combine our EOD data with some of our EMA data to examine relationships between EOD summary states (hassles, perceived stress, mood) and live reports of stressful events and opioid and cocaine use (referred to herein as drug use) covering the same day. Because we have found sex differences in the experience and effects of stress among people with substance-use disorders (Kennedy et al., 2013; Moran et al., 2018), we assessed sex differences in the present analyses. Our overall objectives were: (1) to characterize the hassles experienced by outpatient opioid and cocaine users and how these varied by sex, and (2) to assess how whole-day EOD reports of hassles, as well as EOD reports of current mood, cigarettes/day, and perceived stress, were associated with live reports of drug use and stress events from the same days, and how these associations differed by sex.

2. Methods

2.1. Participants

The participants in this study were enrolled in a large 46-week natural-history study of stress, geographical location, and drug use. During screening, participants completed the Addiction Severity Index (ASI; McLellan et al., 1985) and the Diagnostic Interview Schedule (DIS-IV; Robins et al., 1995) and were given physical examinations and psychological testing. The main inclusion criteria were: age 18 to 75 years, physical dependence on opioids, and (due to the behavioral-geography focus of the parent study) residence in Baltimore City or one of the surrounding counties. The main exclusion criteria were: history of any DSM-IV psychotic disorder, bipolar disorder, or current Major Depressive Disorder; current dependence on alcohol or sedative-hypnotics; cognitive impairment precluding informed consent or valid self-report; conditions that preclude urine collection; or medical illness or medications that would compromise research participation. Use of other drugs was not an exclusion, and most participants used multiple other drugs in addition to opioids.

This study was reviewed and approved by the ational Institutes of Health Addictions Institutional Review Board. Participants gave prior written informed consent and were paid for completing the research components of this study.

2.2. Procedure

Participants attended clinic five to seven days a week; methadone or buprenorphine was administered daily, and individual counseling was given weekly. Urine was collected under observation and screened for illicit drugs thrice weekly.

After two weeks of treatment, each participant was trained to use a smartphone as an electronic diary (ED) and then carried it for up to 16 weeks, during weeks 3–18 of the study. The EDs were programmed with our electronic-diary software (Vahabzadeh et al., 2004). Participants were asked to indicate their typical times of waking and sleeping for each day of the week. Participants used the EDs to make four types of entries: randomly prompted entries, self-initiated (event contingent) reports of drug use, self-initiated reports of stressful events, and end-of-day entries. In the present study, we report on the results of end-of-day entries on days with and without self-initiated reports of stressful events and drug use.

End-of-day entries were prompted by the ED entry each day approximately one hour before the participant’s self-reported bed/sleeping time. The questions (Table 1) included ratings of the participant’s current mood (using a single question with four mutually exclusive choices: angry/annoyed/afraid, happy, sad, or content), unreported drug use, cigarette consumption, a 5-question version of the Perceived Stress Scales (PSS), and a 32-item list of hassles. The PSS items were modified to reflect daily administration (Whitehead and Bergeman 2012). The five items included the four-item version of the Perceived Stress Scale (PSS; Cohen and Williamson, 1988). The fifth item “Today, how often did you find that you could not cope with all the things that you had to do?” was included because of our interest in coping as a drug-abstinence skill. The list of hassles was drawn from a list of 52 items developed by DeLongis and colleagues (1988) to investigate the relationship between stress and health. We shortened the list to make it more amenable to delivery via smartphone by eliminating or combining items that we deemed to be unlikely to be encountered in our drug-using population, such as investments, yardwork, car maintenance, conserving energy, clients, etc. We added items that seemed pertinent to our participants and research interests, such as neighborhood safety, trying to be abstinent, looking for employment, and things that happen at the research clinic itself.

Table 1.

Full text of the “End-of-Day” Questionnaire administered via smartphone

Question	Response Options
Which word(s) best describe how you are  youfeeling right now? (check one only)	Angry / Annoyed / Afraid; Happy / Excited;  Content / Relaxed; Sad / Depressed
Did you use any drugs at all today without  reporting it?	Yes, No; [If response was yes, the participant  was asked to report the type and amount of  drug used]
How many cigarettes did you smoke today?	0 cigarettes; 1–5 cigarettes; 6–10 cigarettes;  11–15 cigarettes; 16–20 cigarettes; more than  20 cigarettes (Note: 20 cigarettes are in a  pack)
Today, how often did you feel unable to  control important things in your life?*	0 Never; 1 almost never; 2 sometimes; 3  fairly often; 4 very often
Today, how often did you feel confident  about your ability to handle your personal  problems?*	0 Never; 1 almost never; 2 sometimes; 3  fairly often; 4 very often
Today, how often did you feel things were  going your way?*	0 Never; 1 almost never; 2 sometimes; 3  fairly often; 4 very often
Today, how often did you find that you could  not cope with all the things that you had to  do? *	0 Never; 1 almost never; 2 sometimes; 3  fairly often; 4 very often
Today, how often did you feel that difficulties  were piling up so high that you could not  overcome them?*	0 Never; 1 almost never; 2 sometimes; 3  fairly often; 4 very often
Which of the following items were a hassle  for you today?
Your children**	YES, NO, or NOT APPLICABLE
Your parents, parents-in-law, or other  relatives**	YES, NO, or NOT APPLICABLE
Your spouse or partner, if you have one**	YES, NO, or NOT APPLICABLE
Time spent with the family**	YES, NO, or NOT APPLICABLE
The health or well-being of a family  member**	YES, NO, or NOT APPLICABLE
Family-related obligations**	YES, NO, or NOT APPLICABLE
Your friends**	YES, NO, or NOT APPLICABLE
Having to do too many things without help***	YES, NO, or NOT APPLICABLE
Sex**	YES, NO, or NOT APPLICABLE
Things that happen at Archway***+	YES, NO, or NOT APPLICABLE
Looking for work***	YES, NO, or NOT APPLICABLE
If employed, your job**	YES, NO, or NOT APPLICABLE
Enough money**	YES, NO, or NOT APPLICABLE
Transportation***	YES, NO, or NOT APPLICABLE
Having a place to live***	YES, NO, or NOT APPLICABLE
Your neighborhood’s safety***	YES, NO, or NOT APPLICABLE
Your neighbors**	YES, NO, or NOT APPLICABLE
Crime***	YES, NO, or NOT APPLICABLE
Your smoking**	YES, NO, or NOT APPLICABLE
Your drinking**	YES, NO, or NOT APPLICABLE
Things related to your drug use***	YES, NO, or NOT APPLICABLE
Trying to be abstinent***	YES, NO, or NOT APPLICABLE
Your physical appearance**	YES, NO, or NOT APPLICABLE
Your medical care**	YES, NO, or NOT APPLICABLE
Your health or physical abilities**	YES, NO, or NOT APPLICABLE
The weather**	YES, NO, or NOT APPLICABLE
News events**	YES, NO, or NOT APPLICABLE
Cooking and housework**	YES, NO, or NOT APPLICABLE
Taking care of paperwork (e.g. paying the bills, filling out forms)**	YES, NO, or NOT APPLICABLE
Legal matters**	YES, NO, or NOT APPLICABLE
Being organized**	YES, NO, or NOT APPLICABLE
Not having enough to do***	YES, NO, or NOT APPLICABLE

^*Items from the Perceived Stress Scale (Cohen and Williamson, 1988)

^**Items from or modified from the Hassles and Uplifts Scale (DeLongis et al., 1988)

^***Items not from or modified from the Hassles and Uplifts Scale

⁺Archway refers to the treatment research clinic where the study was conducted.

For the stress-event entries, we asked participants to initiate an entry any time they felt “more stressed, overwhelmed, or anxious than usual” and to check whichever applied when they initiated an entry (multiple responses were possible). Participants rated the severity of each stress event and indicated the cause of the feeling by selecting from a list. Additional details about the stressful-event entries can be found in Preston et al. (2017).

For the drug-use entries, we asked participants to initiate an entry any time they used a drug for nonmedical purposes. Participants were asked to indicate yes or no for use of heroin, cocaine, other opiates (Percocet, oxycodone, etc.), marijuana, methamphetamines, benzodiazepines, “street” methadone or buprenorphine, alcohol, and other, followed by questions about the amount, reasons for use, time since use, ratings of craving, and context of use. Reports of using heroin and other opioids were combined as a measure of opioid use.

The participants were paid $10 each week for completing at least 82% of their random prompts and end-of-day prompts or were given a warning if they did not meet the criterion. If the participant did not meet the 82% completion criterion for two weeks after being warned, they were removed from the study and assisted with transfer into community-based addiction treatment

2.3. Data analysis

Urine was screened by immunoassay for benzoylecgonine (a metabolite of cocaine) and morphine (a metabolite of heroin and codeine), oxycodone, buprenorphine, and methadone. A specimen was considered opioid positive if any opioid was detected other than methadone or buprenorphine as prescribed for study participation.

Invalid values of PSS score, hassles, and mood were indicated as missing values in the data analyses. A PSS score was considered invalid if the participant answered all five items as zeroes; this occurred for 855 (7.4%) of records. A hassles score was considered invalid if the participant did not answer all questions; this occurred for 330 (2.9%) of records. All mood variables were considered missing if the participant endorsed multiple moods since they were intended to be mutually exclusive categories; this occurred for 363 (3.1%) of records.

Hypothesis tests were performed to assess sex differences on clinical and demographic characteristics (listed in Table 2). Two-sample t-tests were used for continuous variables, and Fisher’s exact tests were used for categorical variables.

Table 2.

Clinical and demographic characteristics

	Total Sample	Participants who smoked
		All	Men	Women
N	182 [135 (74.2%)  men]	161	117 (72.7%)	44 (27.3%)
Opioid agonist medication
Methadone	107 (58.8%)	96 (59.6%)	70 (59.8%)	26 (59.1%)
Buprenorphine/naloxone	75 (41.2%)	65 (40.4%)	47 (40.2%)	18 (40.9%)
Race n (%)
African American	118 (64.8%)	103  (64.0%)	73 (62.4%)	30 (68.2%)
White	60 (33.0%)	54 (33.5%)	40 (34.2%)	14 (31.8%)
Marital Status n (%)
Married	24 (13.2%)	21 (13.0%)	17 (14.5%)	4 (9.1%)
Never married	111 (61.0%)	100  (62.1%)	73 (62.4%)	27 (61.4%)
Separated/divorced/widowed	47 (25.8%)	40 (24.8%)	27 (23.1%)	15 (29.5%)
Employment Status n (%)
Full Time	84 (46.2%)	73 (45.3%)	58 (49.6%)	15(34.1%)
Part Time	42 (23.1%)	39 (24.2%)	29 (24.8%)	10 (22.7%)
Unemployed	46 (25.3%)	40 (24.8%)	24 (20.5%)	16 (36.4%)
Retired/disability/controlled	10 (5.5%)	9 (5.6%)	6 (5.1%)	3 (6.8%)
Age
Years mean (SD)	42.3 (9.5)	42.4 (9.8)	42.6 (10.0)	41.8 (9.2)
Education
Years mean (SD)	12.1 (1.5)	12.1(1.6)	12.0 (1.5)	12.3 (1.6)
Days used in last 30
Heroin mean (SD)	19.3 (11.9)	20.0 (11.5)	20.5 (11.1)	18.8 (12.5)
Other Opioid mean (SD)	8.1 (10.3)	7.5 (9.8)	7.1 (9.6)	8.4 (10.1)
Cocaine mean (SD)	4.5 (8.5)	4.9 (8.9)	4.9 (9.3)	4.9 (7.8)
Years Using
Heroin mean (SD)	14.5 (10.2)	14.5 (10.0)	16.0 (10.0)	10.7 (9.3)*
Other Opioid mean (SD)	1.5 (2.7)	1.4 (2.4)	1.1 (1.9)	2.1 (3.3)
Cocaine mean (SD)	6.0 (8.1)	6.2 (8.1)	6.4 (8.5)	5.7 (7.0)

^*Significantly different for men vs. women (two-sample t-test t=−3.13, df=82.8, p=0.0024); sex comparisons on other variables in this table did not reach statistical significance (p<0.05).

The 32 items from the hassles scale were reduced to eight items through a combination of mixed-regression modeling and cluster analysis, as described below.

In the first set of mixed-regression models, the independent variable was sex, and the dependent variable was either the total number of daily hassles or the proportion of hassles accounted for by each of the individual 32 hassles. In the second set of mixed-regression models, the independent variable was each of the 32 individual hassles (y/n for each day), and the dependent variables were whether a particular type of event-contingent (EC) entry had been made that day (stress entry or drug entry, in separate models). These analyses were mostly for description and data reduction.

The cluster analysis was performed on individual items of the hassles scale, to examine the similarity of items in terms of response frequency. For the cluster analysis, the proportion that each item was endorsed was calculated for each participant. So that the proportions were all based on the same set of participants and responses, the dataset was limited to records having no missing responses on any of the 32 items. A distance matrix for the item proportions was calculated using Euclidean distances, and hierarchical agglomerative clustering was performed using Ward’s method. The results of the cluster analysis were plotted as a dendrogram, and this plot was examined to determine the items that were similar in terms of response frequency.

For a third set of mixed-regression models, a subset of eight hassles items was selected from the 32 items that were considered most relevant to substance-use outcomes based on results of the cluster analysis and the first two sets of mixed-regression models. The selection of items was based on overall frequency of endorsement as well as strength of association with stress events and drug-use events. In these models, the independent variable was each of the eight individual hassles (y/n for each day), and the dependent variables were whether a particular type of EC entry had been made that day (stress entry or drug entry, in separate models) or whether the participant’s urine tested positive for opioids or cocaine (in separate models) in the time frame associated with that end-of-day entry.

For a fourth set of mixed-regression models, the same dependent variables were tested with different independent variables: total PSS score (each day), presence or absence of 4 mood-related responses (y/n for each day), and a high degree of smoking (≥10 CPD y/n for each day).

The mixed models were fit using SAS Proc Glimmix, SAS version 9.4. For all models, within- vs. between-person effects were decomposed according to the method of van de Pol and Wright (2009) in which outcome measures are centered on person-level means; the denominator degrees of freedom were set to reflect the number of participants rather than the number of measurement occasions, as recommended by Bolger and Laurenceau (2013). To examine possible gender differences, mixed regression model sets 3 and 4 included sex as a covariate; these models were also run separately for men and women and the sex-specific results are shown in the supplementary tables¹. Alpha was set at .05, two-tailed, for all analyses.

3. Results

3.1. Participants and EMA reporting and compliance

Data were collected between July 14, 2009 and June 4, 2015. Of 226 individuals who signed consent, 182 provided EMA data for at least two weeks, 12 provided EMA data for less than two weeks, and 32 left the study before providing any EMA data. Because we were interested in the relationship between smoking and stressors, we included only those participants who reported smoking tobacco in at least one EMA entry. Twenty-one participants never reported smoking in any of the EMA entries and were removed from the database, leaving 161 participants. Among the 161 participants, 117 (72.7%) were men, and 44 (27.3%) were women. The demographic and drug-use histories of all 182 participants who provided at least two weeks of data and the 161 participants whose data are reported by gender and combined are shown in Table 2. Statistically significant sex differences were found for years of heroin use: men had used heroin for 16.0 years on average compared to 10.7 mean years of use for women (t= −3.13, df=82.8, p=0.0024).

Participants provided EMA data for an average of 14.3 weeks; 40.4% completed the entire 16 weeks. The dataset includes 11,544 daily observations, 8597 (74.5%) made by male participants and 2947 (25.5%) made by female participants. Stress events were reported on 861 (7.5%) of daily observations. Male participants reported stress events on 8.13% (699/8597) and female participants reported stress events on 5.50% (162/2947) of daily observations. Drug use was reported on 1685 (14.6%) of daily observations. Male participants reported drug use on 14.2% (1222/8597) of observations, and female participants reported drug use on 15.7% (463/2947) of observations. Urine specimens were cocaine-positive on 33.3% (3640/10935) of daily observations and opioid-positive on 37.0% (4042/10935) of daily observations; urine drug results were missing for 609 out of 11,544 daily observations.

Participants reported smoking zero cigarettes on 569 (4.9%) of daily reports, 1–5 cigarettes on 2548 (22.1%) of daily reports, 6–10 cigarettes on 3691 (31.8%) of daily reports, 11–15 cigarettes on 2366 (20.5%) of daily reports, 16–20 cigarettes on 1881 (16.3%) of daily reports, and more than 20 cigarettes on 423 (3.7%) of daily reports; for the remaining 86 daily reports, smoking data were missing.

The most common mood adjective was Content, which was endorsed on 70.2% (7,854/11,181) of days. The frequencies of Happy (11.4% = 1,272/11,181) and Angry/Annoyed/Afraid (11.6% = 1,295/11,181) were similar; sad was the least frequently endorsed mood (6.4% = 711/11,181). Mood data was missing for 363 daily reports.

The perceived stress score ranged from 0 to 20, having a mean (SD) of 6.14 (3.64) and median of 6.00. PSS score was missing or invalid for 1003 daily reports.

3.1.1. Hassles in end-of-day entries.

The total number of daily hassles on any one day ranged from 0 to 25. The median number of hassles was 1; the mean (SD) was 2.33 (3.61); the 75^th percentile was 3 total hassles. The most frequent item endorsed was “not enough money,” which was endorsed on 31.4% of daily reports (Figure 1). The next most common items were maintaining abstinence (18.7%), your smoking (16.2%), transportation (15.7%), your drug use (13.4%), weather (11.2%), and spouse/partner (10.8%).

Figure 1.

Proportion of daily observations on which each item of the Hassles scale was endorsed, graphed as proportion of (A) total observations (n=11,544), (B) observations within each sex (men n=8597; women n=2947), (C) observations occurring on the same day as a real-time report of a Stress Event (stress events n=861 vs. no stress events n=10683), and (D) observations occurring on the same day as a real-time report of a Drug Use Event (drug use reported n=1685 vs. no drug use reported n=9859).

The proportion of daily hassles by item, stratified by sex, is shown in Figure 1. The number of total hassles was not significantly associated with sex (F_1,159=0.53, p=0.47). The median number of hassles per day was 1 for both men and women. For men, the mean (SD) was 2.25 (3.55), 75^th percentile was 3 total hassles (range 0–25); for women the mean (SD) was 2.55 (3.76), 75^th percentile was 4 total hassles (range 0–22). Sex differences in individual items were too small to be statistically significant after Bonferroni adjustment (p<0.0016).

In the analyses of the relationship between each of 32 daily hassles and same-day, real-time reporting of stressful events, every hassles item was endorsed more frequently on days when stress events had been reported (Figure 1). Stressor/hassle associations that achieved statistical significance (within-person effects, p<0.0016) were found for hassles involving: children (F1,147=12.1, p=0.0007), relatives (F_1,158=15.5, p=0.0001), family obligations (F_1,158=20.0, p<0.0001), having too much to do without help (F_1,159=13.7, p=0.0003), the research clinic (F_1,159=20.5, p<0.0001), not having enough money (F_1,159=26.5, p<0.0001), trying to be abstinent (F_1,159=11.6, p=0.0008), paperwork (F_1,159=15.1, p=0.0002), legal matters (F_1,159=15.9, p=0.0001), and being organized (F_1,159=14.0, p=0.0003).

In the analyses of each of the 32 daily hassles and same-day, real-time reporting of drug use (Figure 1), the associations that achieved statistical significance (within-person effects, p<0.0016) were for hassles involving: time with family (F_1,157=10.6, p=0.0014), family obligations (F_1,158=15.9, p=0.0001), friends (F_1,159=35.1, p<0.0001), having too much to do without help (F_1,159=23.6, p<0.0001), drug use (F_1,159=57.2, p<0.0001), trying to be abstinent (F_1,159=116.8, p<0.0001), and weather (F_1,159=11.9, p=0.0007).

For the cluster analysis of hassles, we limited the dataset to records with no missing responses on any of the 32 hassles items; this reduced the dataset to 4361 daily observations on 133 participants. The dendrogram resulting from the cluster analysis is shown in Figure 2. Three clusters are evident: the left-most cluster contains 6 items that reflect basic needs (“enough money,” “transportation,” “having a place to live”) or substance use (“your smoking,” “things related to your drug use,” “trying to be abstinent”). The middle cluster contains 6 items reflecting family relationships: “relatives,” “time with family,” “family health,” “family obligations,” “spouse,” and “too much to do without help.” The right-most cluster contains the remaining 20 items, a varied list that includes health concerns, work, legal matters, and personal safety.

Figure 2.

Results of hierarchical agglomerative cluster analysis that shows the similarity among items of the Hassles scale with respect to frequency of endorsement in the subset of participants whose Hassles data were complete (N=133).

Associations between the total number of hassles plus a subset of eight hassles with and stress and substance use are shown in Table 3. The total number of daily hassles had small but statistically significant within-person associations with same-day stress events and self-reported and urine-detected measures of cocaine and opioid use. The eight individual hassles differed with respect to their associations with stress and substance use. The hassles “family obligations”, “trying to be abstinent”, and “your drug use” all showed statistically significant positive relationships with same-day stress events as well as with self-reported and urine-detected opioid and cocaine use. The hassle “too much do without help” was significantly associated with all of these measures with the exception of cocaine use. The hassles “not enough money” and “having a place to live” were associated with an almost doubled likelihood of a stress event. The effect of sex was not statistically significant in any model, indicating that any sex differences in the effects of hassles on our stress and drug-use outcomes examined were too small for us to detect. Sex-specific results are shown in supplementary table S1². The hassle “family obligations” was associated with a more than tripled likelihood of opioid use for women but not men, and the hassle “too much to do without help” was associated with a more than doubled likelihood of opioid use for women, but not for men.

Table 3.

Total Hassles score, and eight individual Hassles items, in relation to same-day occurrence of substance use and stress events (n=161).

	Stressful Event (EMA)	Drug Use Event (EMA)	Opioid Use (urine screen)	Cocaine Use (urine screen)
Total Hassles	1.09 (1.06–1.13)	1.08 (1.06–1.10)	1.06 (1.04–1.09)	1.03 (1.00–1.06)
Female sex	0.86 (0.54–1.37)	1.48 (0.89–2.48)	2.08 (0.73–5.89)	1.02 (0.26–4.11)
Family  Obligations	2.00 (1.47–2.71)	1.72 (1.31–2.25)	1.58 (1.19–2.10)	1.54 (1.04–2.27)
Female Sex	0.82 (0.51–1.32)	1.59 (0.95–2.65)	2.22 (0.78–6.29)	1.04 (0.26–4.15)
Too Much w/o  Help	1.71 (1.29–2.28)	1.78 (1.41–2.24)	1.60 (1.24–2.06)	1.26 (0.90–1.77)
Female Sex	0.84 (0.52–1.35)	1.50 (0.90–2.52)	2.13 (0.75–6.07)	1.06 (0.26–4.25)
Trying  Abstinent	1.56 (1.21–2.02)	2.65 (2.22–3.17)	2.24 (1.84–2.72)	1.81 (1.43–2.30)
Female Sex	0.82 (0.50–1.33)	1.29 (0.79–2.10)	1.65 (0.59–4.59)	0.78 (0.20–3.02)
Your Drug Use	1.46 (1.11–1.91)	2.12 (1.74–2.57)	1.66 (1.33–2.06)	1.74 (1.34–2.26)
Female Sex	0.84 (0.52–1.34)	1.46 (0.87–2.43)	1.99 (0.72–5.54)	0.97 (0.24–3.81)
Enough Money	1.82 (1.45–2.29)	1.25 (1.06–1.49)	1.06 (0.89–1.27)	0.92 (0.74–1.15)
Female Sex	0.83 (0.52–1.32)	1.50 (0.89–2.52)	2.20 (0.77–6.32)	1.03 (0.26–4.13)
Transportation	1.16 (0.88–1.53)	1.13 (0.92–1.38)	1.23 (1.00–1.51)	0.91 (0.70–1.17)
Female Sex	0.90 (0.56–1.46)	1.56 (0.93–2.61)	2.36 (0.83–6.68)	1.10 (0.27–4.45)
Having Place to  Live	1.70 (1.22–2.37)	0.90 (0.69–1.18)	1.15 (0.86–1.53)	0.65 (0.46–0.91)
Female Sex	0.80 (0.49–1.31)	1.37 (0.82–2.30)	1.99 (0.73–5.45)	0.95 (0.24–3.74)
Your Smoking	1.38 (1.05–1.81)	1.11 (0.89–1.38)	1.00 (0.80–1.25)	0.80 (0.61–1.04)
Female Sex	0.84 (0.53–1.35)	1.55 (0.93–2.56)	2.15 (0.76–6.11)	1.10 (0.28–4.41)

All participants were smokers and provided at least two weeks of EMA data. Within-person effects are shown as odds ratios (95% CIs); bold type indicates p<0.05; each model adjusted for between-person effects. Odds ratios are interpreted as the change in risk of the outcome associated with a unit increase in the independent variable (i.e. an OR of 2 means a doubling of risk associated with a unit increase in the DV, an OR of 0.5 denotes a halving of risk associated with a unit increase in the DV).

3.2. Perceived stress, mood, and smoking in end-of-day entries

Associations between other end-of-day measures (PSS total score, mood ratings, and self-reported smoking) and stress and substance use are shown in Table 4. End-of-day PSS total score had small but statistically significant within-person positive associations with stressful events, drug-use events, and urine-screen measures of substance use. In general, end-of-day negative moods (angry/annoyed/afraid, sad) were associated with greater likelihoods of stress events and opioid use, while end-of-day positive moods (content, happy) were associated with lower likelihoods. End-of-day angry/annoyed/afraid mood was associated with higher likelihood of stressful events and with a greater likelihood of opioid use. End-of-day content mood was associated with a lower likelihood of a stress event and a lower likelihood of cocaine use. End-of-day happy mood was associated with decreased likelihood of opioid use. Smoking at least 10 cigarettes per day was associated with opioid use and with drug events (but not stress events). The effect of sex was not statistically significant in any model, indicating that any sex differences in these measures were too small for us to detect. Sex-specific results are shown in supplementary table S2³.

Table 4.

Perceived Stress Scale total scores (PSS Total) and mood, in relation to same-day occurrence of substance use and stress events (n=161).

	Stress Event	Drug Event	Opioid Use	Cocaine Use
	(EMA)	(EMA)	(Urine Screen)	(Urine Screen)
PSS Total	1.05 (1.01– 1.08)	1.05 (1.03– 1.08)	1.05 (1.02– 1.07)	1.10 (1.07–1.13)
Female Sex	0.84 (0.52– 1.36)	1.56 (0.96– 2.54)	2.03 (0.71– 5.77)	1.09 (0.28–4.25)
Angry…Afraid	1.42 (1.11– 1.82)	0.97 (0.79– 1.19)	1.23 (1.00– 1.51)	1.12 (0.88–1.44)
Female Sex	0.88 (0.55– 1.42)	1.46 (0.87– 2.43)	2.07 (0.73– 5.87)	1.01 (0.25–4.03)
Content Mood	0.67 (0.55– 0.81)	0.92 (0.79– 1.08)	0.87 (0.74– 1.01)	0.80 (0.66–0.97)
Female Sex	0.89 (0.55– 1.43)	1.47 (0.88– 2.45)	2.15 (0.76– 6.11)	1.06 (0.26–4.24)
Happy Mood	1.15 (0.87– 1.52)	0.85 (0.66– 1.09)	0.73 (0.58– 0.91)	0.89 (0.67–1.17)
Female Sex	0.87 (0.54– 1.41)	1.46 (0.88– 2.44)	2.11 (0.74– 5.98)	1.02 (0.25–4.06)
Sad Mood	1.37 (0.97– 1.93)	1.47 (1.14– 1.90)	1.91 (1.42– 2.57)	2.04 (1.43–2.90)
Female Sex	0.88 (0.54– 1.42)	1.50 (0.90– 2.50)	2.20 (0.77– 6.27)	1.09 (0.27–4.37)
Smoking ≥10 CPD	1.29 (0.98– 1.70)	1.68 (1.38– 2.05)	1.41 (1.15– 1.73)	0.99 (0.79–1.25)
Female Sex	0.88 (0.55– 1.43)	1.48 (0.89– 2.48)	2.15 (0.75– 6.16)	1.05 (0.26–4.25)

All participants were smokers and provided at least two weeks of EMA data. Within-person effects are shown as odds ratios (95% CIs); bold type indicates statistical significance (p<0.05); each model adjusted for between-person effects. Odds ratios are interpreted as the change in risk of the outcome associated with a unit increase in the independent variable (i.e. an OR of 2 means a doubling of risk associated with a unit increase in the DV, an OR of 0.5 denotes a halving of risk associated with a unit increase in the DV).

4. Discussion

One of the most striking findings from the hassles data was the prevalence of material needs. By far, the most common hassle was not having enough money; this was reported in over 30% of end-of-day (EOD) entries by both men and women. Other frequently reported hassles involved transportation problems and having no place to live. This result is consistent with the fact that fewer than 50% of these participants were employed full time. As serious as these problems are, hassles pertaining to material needs showed only modest associations with drug-use outcomes in our study. Not having enough money was associated with a statistically significant but small elevation in self-reported drug use overall; associations were stronger between material-needs hassles and stress events. Reporting a “not having enough money” hassle at the end of the day was associated with almost twice the likelihood of having reported a stressful event that day; reporting “not having a place to live” as a hassle at the end of the day was also positively associated a higher likelihood of having reported a stress event that day.

Although we saw only small relationships between financial distress and drug use at a daily level, other data suggest that the two problems can ultimately worsen one another. Siahpush and colleagues (2009) found that smokers with financial stress were more likely to want to quit smoking, yet less likely to succeed. Heavy drinking and smoking have been linked to increases in financial strain in older men and those with less education (Shaw et al., 2011). Severe economic loss (job or housing loss) has been shown to increase the risk of alcohol dependence and abuse (Mulia et al., 2014). Heavy smoking may in fact contribute to financial stress in that the money spent on cigarettes may make it difficult for low-income individuals to have money to pay for necessities (Widome et al., 2015).

Hassles pertaining to substance use were also prevalent: maintaining abstinence, smoking, and drug use were each cited in over 10% of EOD entries. These hassles showed consistent positive associations with drug use: on days when “trying to be abstinent” was reported as a hassle, there was approximately twice the likelihood of an EMA report of a drug-use event, or of an opioid-positive or cocaine-positive urine. Substance-related hassles showed smaller but statistically significant positive associations with stress events in the entire sample. These results highlight the stressful nature of becoming and remaining abstinent and the need for treatment focusing on refusal skills. The clustering of substance use hassles with basic needs hassles sheds light on the degree to which substance use has become a problem central to the lives of those engaged in treatment for opioid use.

Role-related hassles, such as family obligations and having too much to do without help, were less prevalent in this sample, being endorsed in less than 10% of daily reports, but were tightly coupled to drug use, particularly in women. This might reflect the taxing effects of “emotional labor” (labor requiring consistent, seemingly cheerful attention to others’ needs), which is shouldered disproportionately by women in both workplace and family spheres (Guy and Newman, 2004; Du et al., 2018).

Smoking was common in this study, and heavier smoking was associated with use of other drugs. The dataset for this analysis was limited to smokers because relatively few participants (21 out of 182 who provided at least two weeks of EMA data) reported no smoking during the study period. EOD reports of having smoked at least ten cigarettes that day were associated with increased likelihood of having used other drugs that day, particularly opioids. A recent report (Weinberger et al., 2017) showed that continued smoking among a nationally representative sample of individuals who met DSM-IV criteria for substance abuse or dependence was associated with greater odds of relapse; the authors advocated incorporating smoking cessation and prevention efforts into treatment for other drugs of abuse. Our findings support that proposal.

On a more general methodological level, our findings add to the evidence that EOD assessments can be useful in EMA studies, reducing participant burden with low recall bias (Schneider and Stone, 2016). Broderick and colleagues (2009) found that EOD ratings of pain showed good comparability to multiple randomly scheduled within-day EMA assessments and were better than a seven-day recall approach. The hassles scale used in this study showed good criterion validity: as might be expected, hassles were more frequently reported in EOD entries on days when stressful events had been reported by EMA. Every individual hassle on the 32-item EOD scale was endorsed more frequently on days that had EMA stress events compared to days without EMA stress events. This relationship was less apparent when we compared days with and without EMA drug-use reports (Figure 1); however, drug-use days were associated with EOD reports of hassles pertaining to drug use (“trying to be abstinent” or “your drug use”).

The main limitation of the study is its observational design. The need to reduce participant burden and the nature of the smartphone interface led us to shorten or modify the questionnaires from their original paper versions. However, the ambulatory nature of the assessments enabled us to detect relationships that have been elusive in controlled human -laboratory settings. A further limitation is that the sample was limited to smokers; thus, the results may not generalize to people with SUDs who are non-smokers. Additionally, the instrument used to capture stress, which consisted of the four -item plus one additional item, may not have been optimal for our research purposes. Although the four-item PSS had been previously tested and recommended by Cohen and Williamson (1988), a more recent study found that it did “not fit its proposed model” (Ingram et al., 2016), casting doubt on its validity. Another limitation is that we did not examine whether the association between daily hassles and drug use might be moderated or mediated by differences in stress reactivity or mood regulation.

Nevertheless, this study adds to the growing body of work that characterizes the stresses that patients face during addiction treatment, in order to inform behavioral therapy and lead to improved clinical outcomes. A unique strength of this research is its focus on daily hassles and minor stressors, which are recognized as important to overall health but have not been thoroughly studied in the context of addiction and relapse. Cumulative stressful life events and chronic stress are associated with decreased cardiac autonomic function in healthy adults (Lampert et al., 2016). Among patients with Crohn’s disease, daily hassles are associated with increased disease activity, reduced quality of life, and lower economic status (Sarid et al., 2018). Daily hassles have a bidirectional association with increased depressive symptoms among unaccompanied refugees (Keles et al., 2017). In older adults, intensity of hassles predicts mortality (Jeong et al., 2016). Daily hassles have been associated with cortisol levels in people with anxiety (Vasiliadis et al., 2013), and with higher allostatic-load components in people exposed to environmental and social stressors (Mair et al., 2011). Our study’s focus on individual as well as total hassles, as well as its use of EMA to capture how participants feel “in the moment” and minimize recall bias, further adds to the value of the current findings.

In conclusion, the end of day assessment of hassles was a useful addition to ecological momentary assessment. Hassles pertaining to not having enough money and substance use were prevalent and differentially associated with stressful events and drug use. Men and women reported similar rates of daily hassles. Monitoring hassles and devising specific coping strategies might be useful therapeutic targets

Highlights.

• Daily hassles were associated with stressful events and drug use.

• Hassles pertaining to not having enough money and substance use were prevalent.

• Men and women reported similar rates of daily hassles.

• End of day assessments can be useful in ecological momentary assessment studies.