Quantifying Exposure to Illicit Drugs & Psychosocial Stress in Real Time

点击翻译按钮获取中文摘要

Assessment of episodes of drug use and psychosocial stress is complicated by the fact that each is often transient and difficult to recall accurately. Assessment of their causal connections with one another, and of their genetic and environmental determinants, is complicated by the complexity of the causal connections and by the elusive nature of what constitutes the environment. We are continuing our work to study the factors that influence drug use through collection of subjective, psychological and physiological measures, drug use and family histories, and environmental variables, including data about our participants neighborhoods and workplaces. In this project, we are assessing drug use and psychosocial stress in near-real time through ecological momentary assessment (EMA), in which participants use handheld electronic diaries to record events as they occur and to report recent or ongoing events in response to randomly timed prompts throughout the day. We are also maintaining real-time records of where the reported events occur by collecting GPS data to track their whereabouts with a spatial resolution of several meters. We use these data collectively in a method we are calling geographical momentary assessment (GMA). Our goal with GMA has little to do with knowing the specific Baltimore locations where drug-related behaviors occur, and everything to do with gaining generalizable knowledge about how activity spaces (the spaces in which daily activities occur) are associated with such behaviors and their precipitants. We are currently analyzing GMA data from approximately 300 patients in opioid-agonist maintenance many of whom continue to use opioids and cocaine. One goal of our research is to develop methods to determine when to deliver therapeutic content in just-in-time adaptive interventions (JITAIs) when users need it without making them trigger it with effortful input. We trained a randomForest algorithm to predict heroin craving, cocaine craving, or stress (reported on a smartphone three times per day) 90 min into the future, using 16 weeks of field data from 189 outpatients being treated for opioid-use disorder. We used only one form of continuous input (along with person-level demographic data), collected passively: an indicator of environmental exposures along the past 5 h of movement, as assessed by GPS. Our models achieved excellent overall accuracy-as high as 0.93 by the end of 16 weeks of tailoring-but this was driven mostly by correct predictions of absence. Positive predictive value (PPV) usually peaked in the high 0.70s toward the end of the 16 weeks. When the prediction target was more rare, PPV was lower. Our findings complement those of other investigators who use machine learning with more broadly based "digital phenotyping" inputs to predict or detect mental and behavioral events. When target events are comparatively subtle, like stress or drug craving, accurate detection or prediction probably needs effortful input from users, not passive monitoring alone. We also found that high overall accuracy (including high specificity) can mask the abundance of false alarms that low PPV reveals. Among the measures collected in our project are sleep times and sleep quality. Sleep disturbance is common in patients with opioid use disorder (OUD) receiving medication for addiction treatment, and differences between the effects of two primary agonist medications-methadone and buprenorphine-are not well understood. In patients receiving either methadone or buprenorphine treatment for OUD, we examined sleep continuity and architecture using ambulatory monitoring to gather both an objective measure for up to 7 days and a subjective measure of sleep. Patients treated with buprenorphine versus methadone did not differ on any measure of sleep continuity or architecture. Women had longer EEG-derived total sleep time than men, along with lower %N2 and greater %N3. Self-reported sleep differed from EEG-derived estimates: wake after sleep onset was greater by EEG than by diary, and total sleep time and sleep efficiency were lower by EEG than by diary. Patients treated with buprenorphine or methadone did not substantively differ in ambulatory measures of sleep. However, there was a discrepancy between objective and subjective sleep measures. Further confirmatory evidence would inform the development of sleep-related recommendations for OUD patients undergoing agonist treatment. In addition to investigating possible effects of medications used in opioid agonist treatment (OAT) on sleep, we have investigated the effect of timing of medication visits. Illicit opioids can also disrupt sleep, but it is unclear how they affect sleep in outpatients receiving OAT. Therefore, we used electronic diary entries and actigraphy to measure sleep duration and timing in opioid-dependent participants treated with methadone or buprenorphine. For 16 weeks, participants were assigned to attend our clinic under different operating hours in a crossover design: Early hours (07:00-09:00) vs. Late hours (12:00-13:00) for 4 weeks each in randomized order, followed for all participants by our Standard clinic hours (07:00-11:30) for 8 weeks. Throughout, participants made daily electronic diary self-reports of their sleep upon waking; they also wore a wrist actigraph for 6 nights in each of the three clinic-hour conditions. Drug use was assessed by thrice-weekly urinalysis. In linear mixed models controlling for other sleep-relevant factors, sleep duration and timing differed by drug use and by clinic hours. Compared to when non-using, participants slept less, went to bed later, and woke later when using illicit opioids and/or both illicit opioids and cocaine. Participants slept less and woke earlier when assigned to the Early hours. These findings highlight the role OAT clinic schedules can play in structuring the sleep/wake cycles of OAT patients and clarify some of the circumstances under which OAT patients experience sleep disruption in daily life. Treatment with opioid agonists is effective for opioid use disorder, but early discontinuation of treatment is a major obstacle to success. We used our intensive longitudinal method of EMA to gain insight into the effects of stress, mood and craving on drug use while people are being treated and investigate the processes that lead people to drop out of treatment. EMA was conducted for up to 17 weeks by obtaining multiple electronic diary entries per day from 238 participants being treated with methadone or buprenorphine. Survival analysis was used to study two outcomes: dropping out of treatment and noncompliance with EMA self-report requirements. Self-reports of stress, craving, and mood were used as time-varying predictors. Demographic and psychosocial variables measured with the Addiction Severity Index at the start of treatment were used as time-invariant predictors. Dropping out of treatment was more likely in participants with more reported hassles (a measure of stress), higher levels of cocaine craving, lower levels of positive mood, a recent history of emotional abuse, a recent history of being bothered frequently by psychological problems, and with buprenorphine rather than methadone as their medication. In contrast, study noncompliance was not significantly associated with any of the variables analyzed. Our results suggest that assessment of stress, craving and mood during treatment might identify people who are at greater risk of dropping out, and therapeutic interventions targeting these processes might increase retention. Cocaine use in clinical treatment trials is often measured via self-report, which can be inaccurate, or urine drug screens, which can be intrusive and burdensome. Devices that can automatically detect cocaine use and can be worn conveniently in daily life may provide seve