Results

During the study period, 5,602 patients at the two centers were recorded as meeting the definition of MV related injury (CMC- 3,638, PCMH- 1,964) according to the screening log maintained by the Research assistants. Of these 4,257 were eligible for enrollment. Reasons for ineligibility were given as admitted to the hospital (574), deceased (23), found upon questioning not to have MV related injury (25), age < 18 years (504), and previous entry into this study (54), and other (165).

Of the 4,257 eligible for screening, consent to participate in the research could not be obtained in 1,470 (26%), leaving 2,844 consenting to the interview. Reasons for not getting consent were refusal by patient, family or attorney (360), severity of injury (322), concerns about confidentiality (197), treatment preventing access (130), language barrier (193), too intoxicated (32), and under arrest (13) or otherwise not available or not completed (166). The age group distribution of those screened is shown in Figure 1.

Figure 1
Age Group Distribution of Those Screened
Figure 1 - Age Group Distribution of Those Screened
d

Of the 2,787 screened with TWEAK, 388 (13.9%) were TWEAK positive, 269 out of 1,752 (15.4%) at CMC and 119 out of 1,035 (11.5%) at PCMH (Table 1). TWEAK positive patients were considered high risk for AA/AD and were eligible for randomization. No patients were eligible to be randomized based on breath alcohol criteria.

Table 1
TWEAK CMCPCMHTotal
Negative14839162399
Positive 
(Score of 2 or more)
269119388
Total175210352787

Table 2 shows the TWEAK scores by site.

Table 2
TWEAK ScoreCMCECUTotal
014028562258
1201131
217975254
338745
45059109
5271138
615621
710616
8437
9718

Table 3 shows the age, gender and race of those who were TWEAK positive. Patients who were older than 56 years were less likely to screen positive for AA/AD. Males were three times as likely to screen positive as females, and African Americans were less likely to screen positive relative to whites. Interracial effects of the interviewers and the patients on the likelihood of screening positive could not be determined due to sample size.

Table 3
AgeTWEAK + (%)
Score of 2 or more
TWEAK - (%)Odds Ratio
(95% CI)
18-2046 (14.8)265 (85.2)3.862 (2.001-7.455)
21-35225 (16.9)1105 (83.1)4.531 (2.497-8.221)
36-55105 (12.1)762 (87.9)3.066 (1.660-5.661)
56+12 (4.3)267 (95.7)-
Gender*    
Female114 (7.7)1374 (92.3)-
Male268 (21.5)977 (78.5)3.306 (2.616-4.178)
Race**    
White183 (17.6)858 (82.4)-
African American187 (12.0)1369 (88.0)0.640 (0.513-0.799)
Hispanic4 (8.3)44 (91.7)0.426 (0.151-1.201)
Other2 (11.8)15 (88.2)0.625 (0.142-2.757)
American Indian09 (100)NA
Asian014 (100)NA

* Gender not specified in 54 subjects
** Race not known or not specified in 102 subjects

As seen in Figure 2, the 388 patients in the TWEAK Positive group were randomly assigned to the intervention group (n = 130) or to the control group (n= 157). One hundred and one subjects were lost in follow-up efforts and are counted as missing data1.

Of the patients who screened as TWEAK positive, 97.0% agreed to be called for follow-up. Follow-up data were available from 265 patients at three months, and 243 at six months. Of those receiving the intervention, 25 out of 130 (19.2%) received a formal evaluation, which compares quite favorably to the 7 out of 157 (4.5%) in the control group [OR = 5.1, 95%, CI = 2.128 12.235).

Of the group that received the intervention and agreed to a formal evaluation, 21 out of 43 (48.8%) received the evaluation compared to 4 out of 87 (4.6%) who did not agree to a formal evaluation.

Figure 2 - Data Summary
d


Of those contacted at 3 months, 7 out of 265 reported being involved in a motor vehicle crash during that period (2.6%). We do not know whether they were a driver or a passenger in the subsequent crash. All 7 patients were in either the control group or the group that refused intervention. This is equivalent to an annualized crash rate of 106/1000, compared to the crash rate of 25/1000 nationally.10 No patient who received the intervention and agreed to an evaluation had a subsequent crash. No patient reached for follow up had a crash between the 3-month and 6-month follow up.

The number of patients who were randomized and accepted the intervention was insufficient to allow for a multivariate analysis to determine specific factors that predict which patients are more likely to comply with referral for evaluation.

We are planning a follow up study to compare subsequent DWI convictions in the groups receiving the intervention and those who did not. We will also attempt to look at the group that refused the intervention to determine whether the group had more previous DWI charges than those who accepted the intervention.

Continue to Discussion Return to Contents