|
Article Excerpt
MANY CLINICAL AND EPIDEMIOLOGICAL studies require repeated contact with subjects, either during the initial intervention or at follow-up. Selective attrition of the study population during follow-up may threaten the validity of such studies (van Amelsvoort et al., 2004). There are contradictory findings concerning the association of loss to follow-up and age. In some previous research, a higher proportion of loss to follow-up in older study participants was revealed (Thomas et al., 2002); other studies found a lower proportion of loss to follow-up in older participants (Jooste et al., 1990; Pelisse and Barasso, 2003). According to at least one study, male participants were more often lost in follow-up (Claus et al., 2002); however, another study revealed no association between gender and loss to follow-up (de Graaf et al., 2000).
Some investigators have found a higher proportion of loss to follow-up in study participants with "lower educational qualifications" (Claus et al., 2002; de Graaf et al., 2000; Jooste et al., 1990); in another study, on high-risk injection drug users, a higher proportion of loss to follow-up in participants with a high school diploma occurred (Messiah et al., 2003). Alcohol misuse predicted a higher proportion of loss to follow-up (de Graaf et al., 2000; Hansen et al., 1985; Morrison et al., 1997). Among 16,915 adolescent orthodontic patients, Morrison et al. (1997) found that smokers had a 1.92-fold (95% confidence interval [CI]: 1.56-2.38) increased risk of loss to follow-up within 2 years, despite an elaborate program to minimize attrition. Psaty et al. (1994) found that, among white Americans, "current smoking" was a significant predictor of loss to follow-up. In a longitudinal study of 50- to 75-year-old Dutch men (Blanker et al., 2005), smoking and drinking were predictors of loss to follow-up in univariate analysis. Similar findings on smoking and loss to follow-up were reported in other settings, including high school or junior high school populations (Josephson and Rosen, 1978), a community coronary risk factor follow-up study (Jooste et al., 1990), and a cardiovascular cohort study (Thomas et al., 2002).
To the best of our knowledge, little is known about predictors of loss to follow-up in young patients during a screening and intervention study in an emergency department (ED). We do know that the efficacy of screening and brief intervention (SBI) programs on alcohol can be enhanced when an initial intervention is followed by booster sessions (Longabaugh et al., 2001). In smokers, meta-analysis data showed that tobacco-cessation interventions in hospitalized patients were effective only when an initial intervention was followed by booster interventions for at least 1 month (Rigotti et al., 2003). Loss to follow-up (or attrition), therefore, is not only an aspect of study validity but affects the therapeutic regimen in SBI strategies. The aim of this investigation was to evaluate risk factors for loss to follow-up in young patients with minor trauma in an ED setting who participated in a computerized SBI program on alcohol.
Method
This study was approved by the ethics committee of the Charite University Hospital, and written informed consent was obtained from all participants. The investigation was designed as a randomized controlled intervention study. All consecutive patients, 18 years or older, admitted between December 2001 and February 2003 to the ED of the Charite University Hospital of Berlin with acute trauma, were asked to participate. The study is described in detail elsewhere (Neumann et al., 2006). The intervention tested in this study was a computerized tailored brief advice (CTA) that focused on the patient's alcohol consumption. The written alcohol advice was delivered along with brief information on other modifiable health risks (e.g., smoking and illicit drug use). Patients were randomized to either the intervention group or the control group and were followed up at 3, 6, 9, and 12 months. Of 3,026 consecutive patients, 54 refused follow-up at baseline, and 126 were included during the initiation phase according to the protocol (Neumann et al., 2006). Of the remaining 2,846 participants, 284 datasets were incomplete, mainly due to missing data related to income. Therefore, analysis was based on 2,562 complete datasets.
To evaluate alcohol drinking behavior, patients were asked to complete the Alcohol Use Disorders Identification Test (AUDIT; Babor et al., 2001; Saunders et al., 1993), in which a score of 0-7 = "no alcohol problems," 8-15 = "moderate level of alcohol problems," and 16-40 = "high level of alcohol problems" (Miller et al., 1992). The best cutoff for alcohol intervention in this specific trauma population remained unclear at first; therefore, a cutoff of five AUDIT points for delivery of the written alcohol advice was chosen a priori in both genders. In this study, all patients who were currently smoking were classified as smokers. The participants' motivation to change their behavior was measured using the Readiness To Change (RTC) questionnaire for participants with moderate or high levels of alcohol problems (Rollnick et al., 1992) or smoking (Hannover et al., 2003)....
|
