PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Arch Gen Psychiatry. Author manuscript; available in PMC 2010 September 22.
Published in final edited form as:
PMCID: PMC2943840
NIHMSID: NIHMS233662

Real-time electronic-diary reports of cue exposure and mood in the hours before cocaine and heroin craving and use

Abstract

Context

In Ecological Momentary Assessment (EMA), participants electronically report their activities and moods in their daily environments in real time, enabling a truly prospective approach to the study of acute precipitants of behavioral events. EMA has greatly enhanced the study of tobacco addiction, but has rarely been attempted in individuals with cocaine or heroin addiction.

Objective

To prospectively monitor the acute daily-life precipitants of craving for, and use of, cocaine and heroin.

Design

Cohort study.

Participants

A volunteer sample of 114 cocaine- and heroin-abusing outpatients who were being treated with methadone provided EMA data on handheld computers for 14,918 person-days (mean 130.9 days per participant, range 6–189). Of those 114, a total of 102 (63 men, 39 women) provided acute pre-craving or pre-use data and were thus included in the present analyses.

Main outcome measures

Changes in reports of mood and exposure to 12 putative drug-use triggers at random intervals during the five hours preceding each self-reported episode of drug craving or use, analyzed via repeated-measures logistic regression (SAS GLIMMIX macro).

Results

During the five hours preceding cocaine use or heroin craving, most of the 12 putative triggers showed linear increases. Cocaine use was most robustly associated with increases in reports of “Saw Drug” (p<.0001), “Tempted to use out of the blue” (p<.0001), “Wanted to see what would happen if I used” (p<.0001), and “Good mood” (p<.0001). Heroin craving was most robustly associated with increases in reports of “Sad” (p=.0002) and “Angry” (p<.011). Cocaine craving and heroin use showed few reliable associations with any of the putative triggers assessed.

Conclusions

These findings confirm that polydrug-abusing individuals can provide behavioral data in their daily environments using handheld computers, and that those data can reveal orderly patterns, including prospectively detectable harbingers of craving and use, which may differ across drugs.

The study of craving and lapse during attempts to quit smoking has been quietly revolutionized by the use of handheld data-collection devices that enable study participants to report, in real time, their activities and moods in their daily environments1. This methodology is referred to as experience sampling2 or Ecological Momentary Assessment (EMA)3. A PDA (Personal Digital Assistant or PalmPilot) configured for use in EMA is referred to as an Electronic Diary (ED). EMA comprises two complementary types of data collection. The first type occurs through random prompts, enabling assessment of the base rates of exposure to putative craving/lapse triggers such as stressors and drug-associated cues4. The second type of data collection is participant-initiated; participants are typically instructed to initiate an EMA entry immediately after each episode of craving or use. These event-driven data provide near-real-time self-report of such episodes.

EMA data collected through random prompts enable a truly prospective approach to the question of what events preceded a specific episode of use. For example, Shiffman and Waters5 use such data to show that lapses to cigarette smoking were preceded by increases in negative affect over a time course of hours, but not days—in other words, over a time frame that could not have been captured if participants had merely made daily visits to a laboratory for assessment.

EMA studies of illicit-drug use are far more rare. For example, to determine whether stress predicts relapse to cocaine or heroin use, there is only a small body of non-EMA prospective studies whose time frames were not sensitive to episodes of acute stress. The general finding from such studies has been that prospective measures of stress can predict relapse only when aggregated over time frames of three to twelve months, not week by week68. The effects of more rapid increases in stress cannot be determined without EMA data. The near-absence of EMA studies presumably reflects reluctance to entrust handheld electronic devices to individuals from the relevant populations, along with concern that meaningful real-time reports cannot be collected during use of intoxicating drugs. However, some studies have shown that EMA is possible in users of illicit drugs, providing new insights into the natural history of Ecstasy use9 and demonstrating that cannabis smokers10 and homeless users of crack cocaine11 will use and return cellphones issued for EMA recording. Larger datasets would help to resolve longstanding theoretical questions about the precipitants and process of drug craving and use, and may also lead to better outcome measurement in clinical trials12 and even to new forms of tailored, on-the-spot treatment interventions (see Comment).

In this study, we used EMA to monitor craving for, and use of, cocaine and heroin in a large sample of methadone-maintained outpatients. (Methadone maintenance, though highly effective, does not entirely eliminate heroin use in all patients.) Here we present data on the time course of mood and cue exposure in the hours before self-reported episodes of craving and drug use.

Methods

Participants and Setting

Participants were methadone-maintained cocaine- and heroin-using outpatients at a treatment-research clinic in Baltimore, Maryland. Inclusion criteria were: (1) age between 18 and 65; (2) evidence of physical dependence on opioids (by self-report and physical examination); and (3) evidence of cocaine and opioid use (by self-report and urine screen). Exclusion criteria were: (1) schizophrenia or any other DSM-IV psychotic disorder; history of bipolar disorder; current Major Depressive Disorder; (2) current dependence on alcohol or any sedative-hypnotic (by DSM-IV criteria); (3) cognitive impairment severe enough to preclude informed consent or valid self-report; and (4) medical illness that would compromise participation in the study.

We aimed to enroll at least 100 participants for adequate power to examine differences among episodes of craving, episodes of use, and random prompts. A total of 130 participants enrolled, of whom 114 attended clinic long enough to be issued a PDA. Those 114 carried PDAs for 14,918 person-days (mean 130.9 days per participant, median 162.5, range 6–189). Of those 114, a total of 102 (63 men, 39 women) provided acute pre-craving or pre-use data and were thus included in the present analyses. Their mean age was 40.7 (SD 7.9, range 20–58), and they had completed a mean of 11.8 years of education (SD 1.5, range 7–15). Self-reported race/ethnicity was 61% white, 36% African-American, and 3% Hispanic. At treatment entry, 36% were unemployed, 33% were employed full-time, and most of the others were employed part-time. In the 30 days before treatment entry, they had used heroin on a mean of 29.2 days (SD 3.4, range 5–30; the participant reporting 5 days of use had transferred from a community methadone program) and cocaine on a mean of 19.6 days (SD 9.3, range 4–30). The main route of heroin use was reported as intravenous (62%) or intranasal (38%); the main route of cocaine use was reported as smoked (49%), intravenous (43%), or intranasal (7%).

Study Design

The study was designed to assess the natural history of craving and lapse against a background of methadone maintenance, weekly drug counseling, and abstinence reinforcement; all participants received the same treatment. Participants attended clinic 7 days a week for up to 20 weeks; methadone was administered daily (target dose 100 mg/day); urine drug screens were conducted three times per week. Abstinence reinforcement (vouchers given in exchange for urine specimens negative for cocaine, opiates, or both) was in place from weeks 7–18 (12 weeks total; up to $2310 in vouchers were available for participants continuously abstinent from cocaine and opiates); voucher procedures were similar to those used in our prior studies13. Participants were also assigned target quit dates for heroin (first day of week 5) and cocaine (end of week 6, just before vouchers became available). The use of quit dates was intended to make the study similar to prior EMA studies in cigarette smokers, in which EMA data were collected during ad lib smoking and during subsequent quit attempts; there were no consequences in our study for nonadherence to the quit dates.

A PalmPilot (PDA) was issued to each participant at the end of week 3. Our internally developed Transactional Electronic Diary (TED) software14, running on the PDAs, triggered 5 random prompts per day for 5 weeks, then 2 random prompts per day for 20 weeks. Random prompts were timed to occur only during each participant’s typical waking hours, which were programmed in for each day of the week before the PDA was issued. Participants were also instructed to initiate an event-contingent (EC) entry whenever they craved without using or used cocaine or heroin or both drugs. At each EC or randomly prompted (RP) entry, participants reported where they were, whom they were with, and what they were doing and answered questions about their mood.

Among the items assessed in each RP entry were putative triggers of craving and lapse, taken from Heather et al.’s15 adaptation of Marlatt & Gordon’s16 taxonomy of relapse triggers (derived from open-ended interviews and from checklists). Each item began “Within the past hour…” and was answered yes or no. The items were: “I felt bored”; “I felt angry or frustrated”; “I felt worried, anxious, or tense”; “I felt sad”; “I felt others were being critical of me”; “I was in a good mood and felt like celebrating”; “I felt ill, in pain, or uncomfortable”; “I saw heroin or cocaine”; “Someone offered me heroin or cocaine”; “I wanted to see what would happen if I used just a little heroin or cocaine”; “Out of the blue, I felt tempted to use heroin or cocaine.” Participants were asked to specify whether the drug in question was heroin, cocaine, or both. Based on clinical experience, we added the item “Within the past hour, I handled $10 or more in cash” (we chose $10 because our patients typically report their drug use in ten-dollar units).

The PDA models used were the original Palm Zire and its successor, the Palm Zire 21 (Palm Inc, Sunnyvale, California). Each participant was assigned two PDAs that were swapped weekly so that we could recharge the batteries after uploading the data. Our TED software locked all nonstudy functions of the PDA to reduce its street value. Each participant could lose or break one PDA and have it replaced; a second loss resulted in discharge from the study (this occurred with 15 participants). Participants who completed the study without having lost or broken any PDAs were allowed to keep one PDA (with its functions unlocked) or return it for $50. To enhance accuracy in reporting, participants were assured that the clinic staff would not see the content of their EMA entries; however, counselors could see a log of each participant’s number of missed RPs. Participants who missed more than 15% of RPs more than 2 weeks in a row were to be discharged from the study (no participants had to be discharged for this). There were no rules concerning the completeness of EC entries.

The study was approved by the IRB of the NIDA Intramural Research Program, and each participant gave written informed consent before being enrolled.

Statistical analysis

To evaluate the time course of exposure to putative relapse triggers prior to episodes of craving and drug use, we examined the RP entries in each of the five hours preceding each EC entry (with craving or use treated separately). Thus, all data shown here are from RP entries, but the selection of RP entries was determined by their proximity to a subsequent EC entry. There were fewer EC entries for heroin than for cocaine; to provide enough power for data analysis, entries involving only heroin were combined with those involving both heroin and cocaine. Justification for combining the data this way includes the observation that the daily and weekly temporal patterns of “heroin and cocaine” entries were more similar to those of “heroin only” entries than “cocaine only” entries (data not shown).

To determine whether the patterns in the figures represented statistically significant changes over time, we used repeated-measures logistic regressions (SAS GLIMMIX macro), with Hours Before Use (1, 2, 3, 4, or 5) as the within-subject independent variable. The pre-craving and pre-use datasets were analyzed in parallel. A separate GLIMMIX was performed for each putative trigger, and each GLIMMIX included a control term for the number of records that each participant contributed to the dataset. A first-order autoregressive error structure provided the best fit to the data. Contrast coefficients were used to test for a linear trend in Hours Before Use. The criterion for significance was p ≤ .05, with trends noted at p ≤ .10. Regardless of statistical significance, the effect size of the linear trend was also expressed as a correlation coefficient 17, 18.

An additional set of control analyses is described in the Results section.

Results

A total of 34,400 RPs were issued; of these, 26,969 were answered (mean 75% per participant, median 80%, range 22–98%), for a mean of 1.8 per day (median 1.9, range 0.02–4.1). The rate of PDA nonrecovery (breakage or loss) was one for every 226 person-days of use. 2,500 EC entries were initiated (mean 21.0 per participant, median 14.5, range 0–131), for a mean of 0.2 per day (median 0.1, range 0–2.2). Of the EC entries, 37% reflected episodes of drug use and 63% reflected episodes of drug craving.

Of the 102 participants included in the present analyses, 95 provided RP data during the five hours before episodes of craving, and 80 provided RP data during the five hours before episodes of use. The present report includes data from 1,555 RP entries (858 pre-craving only, 632 pre-use only, and 65 both). Of these, 850 involved cocaine alone (378 pre-craving only, 435 pre-use only, and 37 both), 227 involved heroin alone (170 pre-craving only, 47 pre-use only, and 10 both), and 478 involved both drugs (310 pre-craving only, 150 pre-use only, and 18 both). As mentioned above, “heroin alone” and “both drug” episodes were combined for analysis.

Overview of pre-use and pre-craving data for cocaine and heroin

The putative triggers we assessed showed significant linear increases before cocaine use (Fig 1), and heroin craving (Fig 4), but not before cocaine craving (Fig 2) or heroin use (Fig 3).

Figure 1
Real-time participant reports of putative triggers of cocaine use in the five hours of random-prompt entries preceding each episode of use. F and p values reflect linear trends in Hours Before Use from repeated-measures logistic regressions (SAS GLIMMIX ...
Figure 2
Real-time participant reports of putative triggers of cocaine craving in the five hours of random-prompt entries preceding each episode of craving. Details are the same as those for figure 1, except that df’s are (1,114) and the number of datapoints ...
Figure 3
Real-time participant reports of putative triggers of heroin use in the five hours of random-prompt entries preceding each episode of use. (Episodes of use of heroin and cocaine together are also included.) Details are the same as those for figure 1, ...
Figure 4
Real-time participant reports of putative triggers of heroin craving in the five hours of random-prompt entries preceding each episode of craving. (Episodes of use of heroin and cocaine together are also included.) Details are the same as those for figure ...

Pre-use data for cocaine

All 12 putative triggers showed significant linear increases in the hours before use (Fig 1). In terms of effect sizes, the increases were especially robust for “Saw drug,” “Tempted to use out of the blue,” and “Wanted to see what would happen if I used just a little.” Among the items assessing mood, “Good mood” showed the largest increase.

Pre-craving data for cocaine

Only three of the 12 putative triggers showed significant linear changes in the hours before craving (Fig 2). The effect sizes were relatively small, reflecting the fact that the changes were less orderly than in the pre-use data; for one of the three triggers (“Saw drug”), the change was actually a linear decrease.

Pre-use data for heroin

Again, only three of the 12 putative triggers showed significant linear changes (Fig 3), and again, the effect sizes were relatively small, the changes were not very orderly, and for one of the three triggers (“Was offered drug”), the change was actually a linear decrease.

Pre-craving data for heroin

Seven of the 12 putative triggers showed significant linear increases in the hours before use (Fig 4). In terms of effect sizes, the increases were most robust for items assessing mood, especially “Sad.”

Direct comparison of pre-use and pre-craving data one hour before EC entries

For heroin, “Saw Drug” was reported more often before use than before craving [F(1,10) = 9.84, p = .01]. (This analysis included 132 datapoints from 56 participants; degrees of freedom reflect the number of datapoints per participant.) For cocaine, the following were reported more often before use than before craving: “Good Mood” [F(1,13) = 65.32, p < .0001], “Sad” [F(1,13) = 21.44, p = .0005], “Saw Drug” [F(1,13) = 14.09, p = .002], “Tempted out of the Blue” [F(1,13) = 12.82, p = .003], “Was Offered Drug” [F(1,13) = 11.94, p = .004], and “Bored” [F(1,13) = 7.45, p = .017]. (These analyses included 118 datapoints from 55 participants.) Other comparisons were not statistically significant.

Control analyses from periods of abstinence

We examined the RP data for instances in which mood or cue exposure underwent increases similar to the ones seen here without being followed by reports of craving or use. To be confident that we were examining periods of abstinence, we restricted these control analyses to the participants with at least one week of urine-verified simultaneous abstinence from heroin and cocaine (n = 42, with 1–25 weeks of abstinence). Then, to ensure that this subsample was comparable to the total sample of 102, we reran the main “pre-use” analyses on the subsample; the pattern of results was similar to what we saw in the total sample (data not shown). We then created a dataset of RP entries from each participant’s first week of verified abstinence, excluding all RP entries that occurred up to five hours prior to an episode of craving. To have anchor points comparable to the EC-entry time points, we determined the three most common times of day for drug use among these 42 particpants: 2:00–3:00 PM, 5:00–6:00 PM, and 8:00–9:00 PM. Finally, we looked for linear trends in the RPs leading up to those time points (operationalized as 2:30, 5:30, and 8:30 PM); significant increases would be evidence for “false positive,” nonpredictive buildups in exposures to putative triggers.

Of the 36 associations thus tested (12 at each control timepoint), only three showed significant increases: “Worried” prior to 2:30 PM [F(1,98) = 7.09, p = .009, r = .26], “Sad” prior to 5:30 PM [F(1,103) = 6.33, p = .013, r = .24], and “Angry” prior to 8:30 PM [F(1,94) = 4.94, p = .029, r = .22]. In each instance, the percentage of “yes” responses appeared lower than what was seen before episodes of use (control percentages peaked at 5% for “Worried” before 2:30, 7% for “Sad” before 5:30, and 16% for “Angry” before 8:30; data not shown).

Comment

These findings offer an unprecedented glimpse into the buildup of mood changes and cue exposure in the hours before cocaine and heroin use or craving during outpatient treatment. The findings confirm that polydrug-dependent individuals can provide data about their daily activities in real time using handheld computers, and that those data can reveal orderly patterns, including prospectively detectable harbingers of craving and use.

We hasten to acknowledge that this first look at our data is limited in several important ways. For example, we have not yet examined the effects of participants’ physical locations, companions, or activities, nor the differential harbingers of craving and use episodes that participants specifically attributed to each of the putative triggers after the fact. We also cannot rule out the possibility of reactivity—for example, that RP reports of increased temptation to use drugs may have become self-fulfilling prophecies, causing rather than indexing the increased likelihood of subsequent use; alternatively, the use of the ED may have had therapeutic effects (many of our participants reported having found EMA helpful in their attempts to remain abstinent19, and none reported the opposite). However, reactivity to assessment has not been prominent in other EMA studies in which it was formally assessed20, 21. Another limitation is that we did not correct for multiple tests of significance; for our control analyses of periods of abstinence, this was actually the most stringent way to look for “false positive” increases in trigger exposure, and for our main analyses, we felt that it was acceptable because we emphasized the overall pattern of results (e.g. more associations for cocaine use than for cocaine craving) rather than the importance of any one assocation, and in fact we believe that the analyses with few significant findings (e.g. predictors of cocaine craving and heroin use) should be considered negative. Finally, we know that not every episode of drug use was reported, though the temporal resolution of urine drug testing makes the number of unreported uses difficult to determine. Direct comparison of EMA data and urine data will be addressed in future analyses. The occurrence of unreported episodes of drug use (or craving) does not change the conclusions to be drawn from the present analyses, except for the caveat that unreported episodes might have different triggers than reported ones. In future studies, we will be attempting to reinforce participants for accuracy in reporting of drug use.

Bearing these limitations in mind, we wish to highlight some interesting aspects of our findings.

For heroin, the putative triggers we assessed showed linear increases before craving, but not before use. Our speculative explanation for this is that our methadone-maintenance procedure, with its dose ceiling of 100 mg/day, was sufficient to dissociate such triggers from heroin use but was not always sufficient to suppress craving. Another possibility is that the associations will emerge as we examine our data more specifically in terms of the circumstances of use, as mentioned in the previous paragraph. Yet another possibility is that the increases in trigger exposure for heroin occurred over a longer time scale than the five hours we examined in the present analyses.

For cocaine, the putative triggers we assessed showed the opposite pattern: linear increases before use, but not before craving. This cannot be attributed to there being a paucity of episodes of cocaine craving without cocaine use; our dataset had similar numbers of each. One possibility is that cocaine craving is less discrete and more multifarious than heroin craving. Again, however, associations might emerge as we examine our data more specifically in terms of the circumstances of craving.

The statistically significant increase in “good mood” before episodes of cocaine use is, to our knowledge, the first prospective evidence for celebratory lapses, a phenomenon whose existence has been questioned22.

The base rate of endorsement of having handled at least $10 in cash in the past hour was unexpectedly high (an interesting finding in itself, perhaps reflecting our participants’ largely cash-based financial lives and their reliance on public transportation), but its association with cocaine use and heroin craving was also very clear, supporting the widespread clinical impression that cash is a trigger, and suggesting that handling cash should be included in formal taxonomies of relapse precipitants. Of course, some of the increase in handling cash before cocaine use probably reflects preparatory activities for purchasing cocaine—a marker rather than a trigger. To whatever extent the association may be causal, it does not militate against the use of cash payments for research participants, because there is strong evidence that cash payments need not precipitate drug use23. In future analyses, we will look more closely at circumstances under which cash is or is not a seeming trigger of craving or use.

Two of the putative triggers that increased most robustly before cocaine use were highly nonspecific in their content (“I wanted to see what would happen if I used just a little,” “Out of the blue, I felt tempted”). We had included these items because they were part of a published taxonomy of retrospective, attributive explanations for relapses15, 16, but we had hoped to show that most episodes of use are predictable in more specific terms. Such nonspecific precipitants may thwart classification of lapses into “stress-induced,” “cue-induced,” etc., though, again, a closer look at the circumstances of use and craving in our data may help determine whether such lapses truly occurred “out of the blue.” If further analyses confirm that episodes of craving and drug use can be divided in terms of precipitation by cue exposure or mood changes, this would greatly enhance assessment of the effectiveness of new relapse-prevention medications, some of which may be differentially effective for “cue-induced” vs. “stress-induced” relapse24.

Our control analysis of data from periods of urine-verified abstinence suggests that, for the most part, the putative triggers we assessed do not simply increase at particular times of day, producing false-positive predictions. Nonetheless, a few such “blips” did emerge; this finding is similar to a finding by Shiffman and Waters in cigarette abstainers5. To have clinically acceptable positive predictive values, the triggers we assessed will probably need to be combined with data on physical locations, companions, and activities. If further analyses along these lines show that ED data predict use and craving both sensitively and specifically, then real-time detection of such responses on their ascending limb could prompt on-the-spot PDA-based tailored feedback2528 that could potentially be developed as an effective therapeutic intervention to prevent episodes of drug use.

Acknowledgments

Funding/Support: This research was supported by the NIDA Intramural Research Program.

This research was supported the Intramural Research Program (IRP) of the National Institute on Drug Abuse (NIDA), National Institutes of Health. We wish to thank the NIDA IRP Archway Clinic staff for data collection.

Footnotes

None of the authors has any conflicts of interests to disclose.

References

1. Shiffman S, Paty J, Gnys M, Kassel J, Hickcox M. First lapses to smoking: within-subjects analysis of real-time reports. Journal of Consulting and Clinical Psychology. 1996 Apr;64(2):366–379. [PubMed]
2. Barrett L, Barrett D. An introduction to computerized experience-sampling. Social Science Computing Review. 2001;19:175–185.
3. Stone A, Shiffman S. Ecological momentary assessment (EMA) in behavioral medicine. Annals of Behavioral Medicine. 1994;16:199–202.
4. Shiffman S, Gwaltney CJ, Balabanis MH, et al. Immediate antecedents of cigarette smoking: an analysis from ecological momentary assessment. Journal of Abnormal Psychology. 2002 Nov;111(4):531–545. [PubMed]
5. Shiffman S, Waters AJ. Negative affect and smoking lapses: a prospective analysis. Journal of Consulting and Clinical Psychology. 2004 Apr;72(2):192–201. [PubMed]
6. Hall S, Havassy B, Wassermann D. Commitment to abstinence and acute stress in relapse to alcohol, opiates and nicotine. Journal of Consulting and Clinical Psychology 1990. 1990;58:175–181. [PubMed]
7. McMahon R. Personality, stress, and social support in cocaine relapse prediction. Journal of Substance Abuse Treatment. 2001 Sep;21(2):77–87. [PubMed]
8. Wasserman D, Weinstein M, Havassy B, Hall S. Factors associated with lapses to heroin use during methadone maintenance. Drug and Alcohol Dependence. 1998 Nov 1;52(3):183–192. [PubMed]
9. Hopper JW, Su Z, Looby AR, et al. Incidence and patterns of polydrug use and craving for ecstasy in regular ecstasy users: an ecological momentary assessment study. Drug and Alcohol Dependence. 2006 Dec 1;85(3):221–235. [PubMed]
10. Lukasiewicz M, Benyamina A, Brousse P, Falissard B, Reynaud M. An ecological momentary assessment of craving for cannabis and tobacco using SMS (Short Message Service): A pilot study and future development. ISBRA World Congress on Alcohol Research; Sydney, Australia. 2006.
11. Freedman MJ, Lester KM, McNamara C, Milby JB, Schumacher JE. Cell phones for ecological momentary assessment with cocaine-addicted homeless patients in treatment. J Subst Abuse Treat. 2006 Mar;30(2):105–111. [PubMed]
12. Epstein DH, Preston KL. The reinstatement model and relapse prevention: a clinical perspective. Psychopharmacology. 2003 Jul;168(1–2):31–41. [PMC free article] [PubMed]
13. Epstein DH, Hawkins W, Covi L, Preston KL. Cognitive-behavioral therapy plus contingency management for cocaine use: findings during treatment and across 12-month follow-up. Psychology of Addictive Behaviors. 2003;17(1):73–82. [PMC free article] [PubMed]
14. Vahabzadeh M, Epstein DH, Mezghanni M, Lin J-H, Preston KP. An Electronic Diary Software for Ecological Momentary Assessment (EMA) in Clinical Trials. Proceedings of the 17th IEEE Symposium on Computer-Based Medical Systems (CBMS); 2004. pp. 167–172.
15. Heather N, Stallard A, Tebbutt J. Importance of substance cues in relapse among heroin users: comparison of two methods of investigation. Addictive Behaviors. 1991;16:41–49. [PubMed]
16. Marlatt GA, Gordon JRE. Relapse Prevention: Maintenance Strategies in the Treatment of Addictive Behaviors. New York: Guilford; 1985.
17. Rosnow RL, Rosenthal R. Computing contrasts, effect sizes, and counternulls on other people’s published data: general procedures for research consumers. Psychological Methods. 1996;1(4):331–340.
18. Rosnow RL, Rosenthal R, Rubin DB. Contrasts and correlations in effect-size estimation. Psychological Science. 2000 Nov;11(6):446–453. [PubMed]
19. Epstein DH, Preston KL, Schmittner J. 114 cocaine and heroin abusers, 226 PalmPilots: initial experiences with Ecological Momentary Assessment at a methadone clinic. Paper presented at: 69th Annual Scientific Meeting of the College on Problems of Drug Dependence; Quebec City, Canada. 2007.
20. Hufford MR, Shields AL, Shiffman S, Paty J, Balabanis M. Reactivity to ecological momentary assessment: an example using undergraduate problem drinkers. Psychol Addict Behav. 2002 Sep;16(3):205–211. [PubMed]
21. Litt M, Cooney N, Morse P. Reactivity to alcohol-related stimuli in the laboratory and in the field: predictors of craving in treated alcoholics. Addiction. 2000 Jun;95(6):889–900. [PubMed]
22. Baker TB, Piper ME, McCarthy DE, Majeskie MR, Fiore MC. Addiction motivation reformulated: an affective processing model of negative reinforcement. Psychol Rev. 2004 Jan;111(1):33–51. [PubMed]
23. Festinger DS, Marlowe DB, Croft JR, et al. Do research payments precipitate drug use or coerce participation? Drug and Alcohol Dependence. 2005 Jun 1;78(3):275–281. [PubMed]
24. Shaham Y, Shalev U, Lu L, De Wit H, Stewart J. The reinstatement model of drug relapse: history, methodology and major findings. Psychopharmacology. 2003 Jul;168(1–2):3–20. [PubMed]
25. Gruber K, Moran P, Roth W, Taylor CB. Computer-assisted cognitive behavioral group therapy for social phobia. Behavior Therapy. 2001;32:155–165.
26. Newman MG, Kenardy J, Herman S, Taylor CB. Comparison of palmtop-computer-assisted brief cognitive-behavioral treatment to cognitive-behavioral treatment for panic disorder. Journal of Consulting and Clinical Psychology. 1997 Feb;65(1):178–183. [PubMed]
27. Przeworski A, Newman MG. Palmtop computer-assisted group therapy for social phobia. Journal of Clinical Psychology. 2004 Feb;60(2):179–188. [PubMed]
28. Spaniel F, Vohlidka P, Hrdlicka J, et al. ITAREPS: Information Technology Aided Relapse Prevention Programme in Schizophrenia. Schizophr Res. 2007 doi: 10.1016/j.schres.2007.09.005. [PubMed] [Cross Ref]