Impact of a Nursing Triage Line on the Use of Emergency Department Services in a Military Hospital, The

Military Medicine, Dec 2003 by O'Rourke, Kathleen M, Roddy, Mary, King, Rosemary, Custer, Michael, Et al

Statistical Analysis

Data analysis was conducted using the SAS statistical program. Chi-square tests for homogeneity were obtained for categorical variables. Logistic regression models were constructed to identify the impact of TAN referral on the appropriateness of the ED visits. All variables with a p

Results

Table I presents demographic and other variables of study participants by type of referral: non-TAN vs. all TAN referrals. For this analysis, data are provided on three levels of the disposition: urgent, emergent, and nonurgent. In the crude analysis, there was a marginally significant difference in acuity level of ER usage between the non-TAN and TAN groups (p = 0.0702), The comparison group (6.9%) was more likely than TAN clients (2.8%) to be triaged as urgent. Moreover, self-referred ED patients (46.9%) were less likely than TAN referrals to be triaged as inappropriate ED users (50.3%). When TAN patients for whom there were no clinic appointments were excluded from the crude analysis, the percentage of TAN referrals Waged as nonurgent decreased to 44.2%, and the between-group difference became statistically insignificant (p = 0.22, data not shown).

In the crude analysis, there were significant differences between non-TAN and TAN referrals with regard to demographic variables. TAN referrals more likely than non-TAN-referred patients to be between the ages of 18 and 50 years, female, and a spouse of an active duty individual.

TAN patients (12.9%) were more likely to leave the ED without treatment than were their non-TAN counterparts (8.7%; p = 0.106). However, 10.9% of the entire study population left the ED without being seen. When analyzed by acuity level, 17 (5.9%) of high acuity patients left as compared with 44 (16.1%) of the lower acuity patients (p

As a result of demographic differences between non-TAN and TAN ED clients, a multivariate analysis was conducted to determine whether there was any difference in ED use between the two groups when controlling for these potential confounders (Tables II and III). For the purpose of the multivariate analysis, patients with acuity levels of 1 to 3 were combined and defined as appropriate referrals. These patients were compared with those with acuity levels 4 to 5, defined as inappropriate or nonurgent referrals. When the analysis included all TAN referrals to the ED, TAN clients were 13% less likely to receive a triage level of 1 through 3 and deemed appropriate users of ED resources than were non-TAN ED users. This finding was not statistically significant (p = 0.452; Table II). However, when the 98 TAN clients for whom no clinic appointment was available were excluded from the analysis, TAN clients were 19% more likely to be categorized as appropriate triage than self-referred patients (Table III). Again, this difference was not statistically significant (p = 0.401). The demographic variables, age, sex, and family code (relationship of patient to sponsor) no longer significantly impacted appropriateness of ER usage in the multivariate model.