Computer-automated dementia screening using a touch-tone telephone.

BACKGROUND: This study investigated the sensitivity and specificity of a computer-automated telephone system to evaluate cognitive impairment in elderly callers to identify signs of early dementia. METHODS: The Clinical Dementia Rating Scale was used to assess 155 subjects aged 56 to 93 years (n = 74, 27, 42, and 12, with a Clinical Dementia Rating Scale score of 0, 0.5, 1, and 2, respectively). These subjects performed a battery of tests administered by an interactive voice response system using standard Touch-Tone telephones. Seventy-four collateral informants also completed an interactive voice response version of the Symptoms of Dementia Screener. RESULTS: Sixteen cognitively impaired subjects were unable to complete the telephone call. Performances on 6 of 8 tasks were significantly influenced by Clinical Dementia Rating Scale status. The mean (SD) call length was 12 minutes 27 seconds (2 minutes 32 seconds). A subsample (n = 116) was analyzed using machine-learning methods, producing a scoring algorithm that combined performances across 4 tasks. Results indicated a potential sensitivity of 82.0% and specificity of 85.5%. The scoring model generalized to a validation subsample (n = 39), producing 85.0% sensitivity and 78.9% specificity. The kappa agreement between predicted and actual group membership was 0.64 (P<.001). Of the 16 subjects unable to complete the call, 11 provided sufficient information to permit us to classify them as impaired. Standard scoring of the interactive voice response-administered Symptoms of Dementia Screener (completed by informants) produced a screening sensitivity of 63.5% and 100% specificity. A lower criterion found a 90.4% sensitivity, without lowering specificity. CONCLUSIONS: Computer-automated telephone screening for early dementia using either informant or direct assessment is feasible. Such systems could provide wide-scale, cost-effective screening, education, and referral services to patients and caregivers.

Lay person-based screening for early detection of Alzheimer's disease: development and validation of an instrument.

Symptoms of cognitive impairment reported to telephone interviewers by caregivers of 272 patients were analyzed with respect to research diagnoses of dementia. All patients received neuropsychological evaluation for establishing the research diagnoses. A data mining program that used machine learning algorithms produced an optimized binary decision tree for differentiating patient groups according to all available information. The results of this analysis were used to help four dementia experts create a dementia screening instrument amenable to application and scoring by nonclinical personnel. The validity of the resulting instrument was then evaluated in an independent sample of 103 patients administered neuropsychological testing within the previous 60 days. The psychometric properties of the empirically derived scale and its performance for discriminating control from probable or possible Alzheimer's patients indicate strong potential for use as a dementia screener for the general population.

A modeling analysis of young drivers' judgments of accident risk due to alcohol use and other driving conditions.

Male (60) and female (60) college students who were under the legal drinking age completed a personal driving experience questionnaire and judged the probability of an accident occurring in 100 driving scenarios, half involving an intoxicated male driver and half a sober male driver. The scenarios also varied in terms of the safety levels of five other driving factors. Mathematical models derived from the probability judgments permitted measurement of the influence of the various driving factors on perceived risk. The driving factors in order of their influence on the accident probability judgments were driver intoxication-nonintoxication, vehicle speed, road conditions, driver mood and driving experience, which did not significantly differ, and the driver's experience with alcohol use. Accident probability estimates of the alcohol scenarios increased as an additive function of the other driving factors, whereas increased risk of an accident in the no-alcohol scenarios was more dependent upon multiple unsafe factors. Female students judged the probability of an accident higher overall than did male students. Subject drinking classification (abstain-light, moderate or heavy) was not related to accident probability estimates.

Individual differences in alcohol responsivity: physiological, psychomotor and subjective response domains.

OBJECTIVE: Psychometrically stable measures of individual differences in response to consuming alcohol were developed as part of an ongoing project investigating "high-tolerant" drinking drivers. The derived measures were correlated with diagnostic criteria of alcohol abuse and dependence, as well as drinking frequency, heavy drinking and tolerance estimations. METHOD: Male subjects (N = 94), 21 to 59 years old, participated in each of two identical alcohol consumption sessions (mean dose = 0.78 g/kg). Multiple response measures obtained before and after drinking (mean BAC = 78 mg/dl, descending) were used to compute postconsumption regression residuals that were factor analyzed. RESULTS: Factors reflecting psychomotor, subjective and physiological responsivity to alcohol were extracted. Confirmatory factor analyses indicated consistent psychomotor and physiological factor structures across the two testing sessions, but a bifurcation of the subjective measures. Test-retest correlations of the responsivity residuals indicated medium to moderately large effect sizes (mean = .32, range = .15-.51). Residuals summated within each response domain indicate moderate internal consistency (mean = .56, range = .45-.72) and higher reliabilities than single measure indices (range = .53-.59). Summated residuals within the response domains suggest diminished responsivity is associated with more severe alcohol-use diagnoses and were differentially predictive of self-reported drinking practices. CONCLUSIONS: Alcohol responsivity is multidimensional. Individual differences across different domains can be measured reliably and are associated with different patterns of alcohol use and abuse. Greater attention should be given to the specific domain of alcohol responsivity measures obtained in past and future research that relates individual differences to alcohol-related problems.

Young drivers' evaluation of driving impairment due to alcohol.

Ninety-six college students, 18-20 years of age, were selected from three drinking categories (abstain-light, moderate, and heavy) based on their self-reported drinking habits. Subjects rated the relative importance of three driving components (attention, control/maneuvering, and emergency responses) for safe driving and the impairment each component and overall driving ability would sustain after driver drinking. Impairment ratings were made of the effects of drinking the number of drinks that would have resulted in the rater having blood alcohol concentrations (BACs) of .05%, .10%, and .15%. Subjects also estimated the number of drinks that they could consume in one hour and then safely drive after after different time periods, the time they should wait before driving after drinking various amounts of alcohol, and the number of beer, wine, wine cooler, and whiskey drinks that would cause them to reach a .10% BAC. While the driving components were rated to be of approximately equal importance for safe driving, the emergency responses component was judged to be more impaired following alcohol consumption than any of the other components, or overall driving ability. Heavy drinkers judged that there was significantly less driving impairment due to alcohol than did light or moderate drinkers in the case of overall driving ability and all of the driving components except emergency responses. Heavy drinkers also judged it safe to consume a number of drinks before driving that would result in higher mean BACs than those of light and moderate drinkers. Both drinking category and gender differences were found in the estimated BAC values that would have occurred at the time of driving after waiting periods judged by subjects to be sufficient for safe driving.

When alcohol and water don't mix: diving under the influence.

The impairing effects of alcohol on divers' abilities to perform shallow-water entry dives were investigated under controlled conditions. Each of 13 male recreational divers, 21 to 35 years old, performed three shallow-entry dives during each of seven diving sessions at various blood alcohol concentrations (BACs). The mean group BACs of Diving Sessions 1-7 were 0, 17, 40, 71, 100, 123 and 97 mg/dl, respectively. Diving performances were videotaped and rated on their relative risk of injury due to contact with the pool bottom if the depth had actually been 3.5 feet. The data indicate progressive and significant impairment of diving performance at BACs of 40 mg/dl (or 0.04 g/dl, or 0.04%) and higher. Corroborating evidence of alcohol impairment was obtained from administering Field Sobriety Tests following Sessions 1, 5, 6 and 7; almost all subjects failed the tests at the higher BAC sessions. An analysis of the self-ratings of diving performances indicated that the divers were not aware of either their degraded performance or the increased risk of injury resulting from diving after drinking.

Self-report of drinking using touch-tone telephone: extending the limits of reliable daily contact.

OBJECTIVE: Retrospective self-reports of alcohol consumption are ubiquitous in the alcohol research field. Time frames of these reports range from a week to a year or longer and are subject to several influences of bias that may have important clinical, epidemiological and methodological consequences. In order to specify drinking patterns more precisely, a study was conducted to monitor alcohol consumption on a daily basis. METHOD: Subjects (N = 51) responded for 112 days on an Interactive Voice Response system by entering their data daily using the touch-tone pad of their telephone. Each day, subjects answered 11 questions relating to drinking (including quantity) and to variables believed to affect consumption (e.g., stress level). RESULTS: The overall response rate was 93.0%. Subjects reported consuming at least one drink on 51.2% of all 5,151 reporting days (mean number of drinks reported = 4.6). Following completion of the study, subjects were also asked to recall consumption retrospectively using a standard quantity-frequency questionnaire. CONCLUSIONS: The present study demonstrated that: (1) data can be collected on a daily basis efficiently, and (2) traditional methods of data collection (e.g., quantity-frequency) result in a significant underreporting bias for heavier drinkers.

Validation of daily self-reported alcohol consumption using interactive voice response (IVR) technology.

OBJECTIVE: This study assesses the validity of daily self-reported drinking data obtained using an automated touch-tone interactive voice response (IVR) system. METHOD: Subjects (N = 30) reported alcohol consumption daily for 28 days using the IVR system. Concurrently, breath and saliva samples were obtained each night for objective determination of blood alcohol concentrations (BACs). Partners living with the subjects provided collateral reports daily. Retrospective drinking records were obtained from both partners at the outset of the study and from the target subjects at the end of the study, using timeline follow-back procedures referencing the target subjects' drinking over the previous 28-day period. RESULTS: Subjects reported drinking on 55.2% of the 840 possible subject days, and positive BAC readings were obtained on 25.9% of these days. The overall correlation between self-report and measured BAC was .72. Within-subject correlations between daily IVR reports and measured BACs ranged from -.07 to .92, with a mean of .57. The correlations between self-reported drinking and the collateral reports ranged from .18 to 1.0, with a mean of .89. Correlations between the daily self-reports and the timeline follow-back records obtained at the end of the study ranged from -.22 to .96, with a mean of .51. CONCLUSIONS: IVR technology provides an innovative, user-friendly methodology for obtaining valid measures of daily alcohol consumption. The validity of these measures may be differentially highest for frequent, heavy drinkers, a group for whom traditional assessment methods often produce the most biased underestimates.

Cycles of alcohol dependence: frequency-domain analyses of daily drinking logs for matched alcohol-dependent and nondependent subjects.

OBJECTIVE: This study was undertaken to evaluate possible differences in the patterns of daily alcohol use, in terms of periodicity or rhythmicity of consumption, between alcohol-dependent and nondependent drinkers otherwise matched on the basis of quantity and frequency of alcohol use. METHOD: Frequency-domain time-series analysis was applied to daily drinking logs of six alcohol-dependent and six nondependent male subjects matched for consumption quantity and frequency who had participated in a larger 112-day study of alcohol consumption patterns using an innovative touch-tone telephone method of daily self-report. Spectral analysis partitioned variance in the drinking logs among statistically independent frequency cycles of relatively heavier and lighter drinking. The percentage of variance accounted for by fast (< 4 days), medium (7 +/- 3 days) and slow (> 10 days) cycle periods was compared for diagnostic group differences. RESULTS: Effective matching precluded consumption quantity, frequency or variability differences between the groups. Spectral analysis of the time-series data revealed group differences in the rhythmicity, or cyclical characteristics, of the drinking logs. The drinking logs of alcohol-dependent subjects exhibited more pronounced weekly cycles, whereas drinking logs of nondependent subjects fluctuated over longer periods. CONCLUSIONS: Temporal patterns of regular alcohol use may be an important behavioral variable influencing the seriousness of consequences associated with drinking, which may partly elucidate the relatively weak relationship between dependence symptomatology and consumption quantity-frequency.

Comparison of interactive voice response and written self-administered patient surveys for clinical research.

This study compared the use of an interactive voice response (IVR) system with a written survey to collect data. Consecutive patients seen in an orthopedic clinic completed the Short Musculoskeletal Function Assessment (SMFA) by either IVR or written questionnaire and then were asked to complete the form again 3-7 days later using the opposite modality. Patient response rates were analyzed for differences between the self-administered IVR and written modes of administration. Three orthopedic clinics participated in the study. No significant differences between method of administration or individual differences in response per patient were found. Therefore, no significant differences in measurement are found when IVR is compared to the written SMFA.

Meeting the need for public education about dementia.

Research continues to advance the knowledge of pathophysiology and development of effective methods for treating patients with Alzheimer disease and other dementias. Dissemination of information is likely to be slowest among the general population, who may be the first to recognize dementia symptoms but may also be reticent to discuss concerns because of fear, embarrassment, and/or inadequate knowledge. The feasibility of providing public education and access to dementia resources was studied using a toll-free interactive voice response (IVR) telephone system. Public interest in this service and willingness to use this technology were evaluated in a 1-month study conducted in a predominantly rural upper Midwest county (population of 102,565). One hundred ninety-three calls were received during November 1999, with an average length of 9 minutes and 29 seconds. One in six calls lasted 15 minutes or longer. One third of the calls were received outside typical business hours (8:00 AM to 6:00 PM). Concern for a parent or grandparent was the most frequent reason (50.6%) given for the call. Self-concern was indicated by 24.7% of the callers. Callers provided positive feedback. Such IVR technology may provide a cost-effective bridge to the "digital divide" existing among elderly, lower socioeconomic status, and rural populations underrepresented as computer and Internet users.

Effectiveness of antidepressant pharmacotherapy: the impact of medication compliance and patient education.

This study was designed to investigate the impact of a time-phased patient education program (RHYTHMS) on medication compliance and treatment outcomes of primary care patients diagnosed with major depression and started on antidepressant pharmacotherapy. Two hundred forty-six depressed patients, diagnosed and treated at one of three outpatient clinics affiliated with the Kaiser-Permanente Northwest Region (KPNW) healthcare system, were randomly assigned to either receive or not receive (usual care) the educational materials by mail. Depression severity and functional impairment affecting patients' quality of life were assessed at baseline and 4, 12, and 30 weeks later. Self-reported impressions of improvement and patient satisfaction with treatment were also assessed at follow-up. Clinical assessment data were obtained using an interactive voice response (IVR) system. Study subjects were compensated $5, $10, $15, and $25 for completing each assessment (Baseline to Week 30, respectively). Upon study completion, prescription fill data of the subjects were extracted from the KPNW Pharmacy System for analysis of medication compliance. Most of the study subjects (63.5%) responded to the pharmacotherapy treatment by study end-point. Few statistically significant differences in either treatment outcomes or duration of medication compliance were found between the treatment groups, and significant differences found were of fairly small magnitude. Patients not receiving the educational materials initially exhibited a more positive response to treatment (Week 4), but this difference did not persist at later follow-ups and was associated with significantly higher relapse rates. A strong time-dose relationship was evident between the duration of the initial treatment episode and treatment outcomes at follow-up, but randomized treatment assignment did not influence the duration of initial medication compliance. Educational programs designed to encourage depressed patients to obtain adequate pharmacotherapy likely provide medical benefits. Such benefits appear to be relatively subtle and methodological differences between studies contribute to inconsistent conclusions concerning observed benefits. The intent of providing time-phased educational materials to patients is to maximize the relevance of such information by synchronizing it with typical recovery processes and issues. This study suggests that additional efforts at engaging patients earlier after the initiation of treatment might be of most benefit.

Computerized assessment of depression and anxiety over the telephone using interactive voice response.

We examined the reliability and validity of computer-administered versions of the Hamilton Depression (HAMD) and Hamilton Anxiety (HAMA) Rating Scales that were administered over the telephone using Interactive Voice Response (IVR). In two identical studies (HAMD: N = 113, HAMA: N = 74), both the IVR- and clinician-administered versions were administered in a counterbalanced order to a heterogeneous sample of subjects with psychiatric disorders and controls. Both the IVR HAMD and HAMA demonstrated adequate internal-consistency reliability (.90 and .93, respectively) and test-retest reliability (.74 and .97, respectively). The correlation between the IVR and clinician was high (HAMD = .96; HAMA = .65). The mean score difference between the IVR and clinician versions was less than one point for both the HAMD (.69 of a point) and HAMA (.60 of a point). It took subjects 12.23 minutes to complete the IVR HAMD, compared to 15.21 minutes for the clinician version; and 11.27 minutes for the IVR HAMA, compared to 15.33 minutes for the clinician (p < .001 for both comparisons). Subjects rated the clinician better in the areas of how much they liked being interviewed and how well they were able to describe their feelings. However, they were significantly more embarrassed with the clinician than with the IVR. Results support the psychometric properties of the IVR versions of the HAMD and HAMA scales. IVR technology presents new opportunities for expanding the utility of computerized clinical assessment.

Administration of the Hamilton Depression Rating Scale using interactive voice response technology.

We compared the reliability and equivalence of scores on the Hamilton rating scale for depression obtained by an interactive voice response (IVR) system and by human raters among 367 subjects 18 to 79 years old. Structured clinical interviews indicated that 93 of these subjects met the DSM-IV criteria for current major depression. We also examined short forms using the IVR data. The results obtained with the long (17-item) IVR interview correlated with those of the face-to-face interview, as did the results of the short (six- and eight-item) IVR interviews. In a random sample of 264 subjects, linear transformations of IVR results produced scores equivalent to clinical scores in the validation sample. We conclude that computerized assessment of the severity of depression by touch-tone telephone presents expanded opportunities for outcome research.