Quantifying Exposure to Illicit Drugs & Psychosocial Stress in Real Time

量化非法药物的暴露程度

• 批准号: 10699649
• 负责人: David Epstein
• 金额: $ 87.57万
• 依托单位: NATIONAL INSTITUTE ON DRUG ABUSE
• 依托单位国家: 美国
• 资助国家: 美国
• 起止时间: 至
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10699649
• 关键词: Alkaloids; Anhedonia; Arousal; Behavior; Behavioral; COVID-19 pandemic; Categories; Clinic; Cocaine; Cocaine use disorder; Code; Complement; Data; Dropout; Drops; Drug usage; Drug user; Ecological momentary assessment; Enrollment; Environment; Environmental Exposure; Event; Exposure to; Factor Analysis; Future; Genetic; Goals; Heroin; Hour; Illicit Drugs; Impairment; Joints; Laboratories; Lead; Life; Literature; Methods; Mitragyna; Modeling; Moods; National Institute of Drug Abuse; Nature; Outpatients; Paper; Participant; Patients; Persons; Pharmaceutical Preparations; Physiological; Plants; Predictive Value; Prevalence; Prevention; Psychosocial Stress; Public Health; Publishing; Questionnaires; Reporting; Research; Rewards; Risk; Running; Sampling; Series; Ships; Site; Specificity; Stress; Structure; Surveys; Time; Work; addiction; base; craving; data reduction; digital medicine; drug craving; drug reward; experience; hedonic; heroin use; mental state; negative mood; non-drug; opioid use disorder; pandemic disease; pleasure; positive mood; response; retention rate; sensor; smartphone Application; trustworthiness

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 have continued to analyze and publish data we collected (before the Covid-19 pandemic) from outpatients at what had been our on-site clinic. Participants provided data in near-real time through ecological momentary assessment (EMA), in which a smartphone app prompted them to record events as they occurred and to report recent or ongoing events in response to randomly timed prompts throughout the day. Studies have included the following: --In work published in Nature Digital Medicine, we used 16 weeks of our already collected field data from 189 outpatients being treated for opioid use disorder (OUD), simulating live predictions of heroin craving, cocaine craving, or stress (reported via smartphone app 3x/day) 90 minutes into the future. We used only one form of continuous input (along with person-level demographic data), collected passively: an indicator of environmental exposures in the past 5 hours, via GPS. Our randomForest models achieved excellent overall accuracyas high as 0.93 by the end of 16 weeks of tailoringbut, as we clearly displayed, accuracy was driven mostly by correct predictions of absence (Figure 1). For predictions of presence, positive predictive value (PPV) usually peaked in the high 0.70s toward the end of the 16 weeks. We emphasized PPV because it reflects the trustworthiness of a given "alert." This is the crux of how a JITAI will be experienced by users in real time: not sensitivity ("what percentage of my cravings will be detected?") or specificity ("of the noncraving moments that constitute the bulk of my time, what percentage will be undisturbed by false alarms?"), but PPV ("is this craving alert to be believed?") and NPV ("does the app's silence right now mean I'm not at risk of craving?"). We found that PPV remained low for participants who rarely craved (figure 8 in the published paper, which is included in the package). I offered two conclusions: (1) With GPS tracks, we predicted moments of drug craving approximately as well as anyone had predicted future mental states with any amount of other passive-sensor (or EMA) data. (2) Nonetheless, those predictions probably included too many false alarms to sustain the live use of the app for lapse prevention. --To make better use of the mood-adjective data we collect via EMA, we have clarified how specific moods should be categorized. Like many EMA studies, ours have assessed mood via participant ratings of long lists of adjectives intended to assess a four-quadrant "mood circumplex." The circumplex reflects two continuums: low-energy to high-energy (arousal), and unpleasant to pleasant (valence). Data reduction is complicated by longstanding debates concerning, among other issues, whether ambivalent states are possible and whether moods form similar factor structures at two levels of analysis: within each person, and within moments for each person. We conducted a multilevel factor analysis of EMA mood ratings from 306 participants (at random moments thrice daily for up to 8 weeks, for a total of 39,321 person moments). We found that the best-fitting solution, within and between people, consisted of three factors: positive mood, negative mood, and low-arousal states. High-arousal states did not constitute a separate factor. To our knowledge, this is the first analysis of mood states to result in three categories rather than two (valence only) or four (valence versus arousal). This more "carved at the joints" coding of mood states may lead to more accurate prediction. --We also examined EMA-based predictors of early dropout from treatment. We found that patients who drop out initiate more stress-event entries than other patients, suggesting that they are not simply disengaged from treatment (or research): many are adherent but overwhelmed. Retention strategies can take that finding into account. --We used four weeks of EMA data from 47 healthy controls and 54 outpatients in treatment with medication for opioid use disorder (25 still using heroin, 29 largely abstinent) to examine assumptions about anhedonia, in two ways. First, we wanted to extend findings that people scoring high on anhedonia scales were shown, by EMA, to be "anything but flat and blunted," such that pleasant events in their daily lives induced "mood brightening" exceeding that of controls. Second, we wanted to extend findings that people with addictions did not report drug-specific anhedonia, in which drug reward is the only reward that is fully enjoyed. We assessed patients in their first four months of MOUD. It was not a foregone conclusion that treatment would render them hedonically intact: scientific literature and public-health messages suggest that anhedonia in OUD does not readily reverse. Indeed, via standard questionnaires, our OUD participants had a 41% prevalence of anhedonia at baseline. But EMA data showed that, in daily life, they encountered or undertook nondrug pleasures just as often as our healthy controls, and when they did so, their pleasure ratings were, at worst, only slightly lower than those of our healthy controls. Our findings challenged the view that addiction typically or irreversibly impairs responses to nondrug rewards, and challenged definitions of anhedonia as an obliteration of the capacity for pleasure. --Now, in a newly launched continuation of this project, we are using EMA to study a nationwide sample of people who use the drug kratom. (To complement the EMA data, we are running a substudy in which some participants take their kratom under our observation for pre-post behavioral and physiological assessment.) The nationwide EMA study has enrolled participants at a pace that is unprecedented for us, and we anticipate that the behavioral data will be exceedingly informative, especially because we are having participants ship samples of their purchased kratom project for laboratory analysis so we can examine relationships between specific alkaloid content and effects reported in real time.