Patient-level factors are more salient than a legislation prohibiting minors in bars in predicting unintentional injury hospitalizations

This study was designed as a retrospective observational study involving patients aged 10 to 19 with and without positive blood alcohol concentration (cases and controls, respectively) over a 10-year period. Inclusion criteria were all of young patients aged 10 to 19 identified in the Illinois State Trauma Registry (ISTR) following presentation to level I and level II trauma centers. ISTR is a mandatory reporting database maintained by the Illinois Department of Public Health containing information about all traumas. This database is de-identified with respect to name and hospital, but includes patient demographic information, such as home address, gender, age, race, physiological data, mortality and discharge outcomes, and incident/scene address information. Cases were young patient encounters with a positive whole BAC test result at the time of admission, whereas controls were the young patients with a negative whole BAC test result at the time of admission. Only patients with complete records were used in the analysis. In addition, because we sought to determine whether a legislation that prohibit minors under 21 years old in establishments that serve alcohol is more salient than individual level factors in predicting traumatic unintentional injury hospitalization, we obtained city-level survey data of local liquor control authorities from the Illinois Liquor Control Commission (ILCC) conducted in 2013. Patient data (n = 6,139, patients with complete records) were then linked with the city-level alcohol-related legislation (n = 514 cities).

Illinois is no exception to the social burden of underage drinking consequences: alcohol related homicide, suicide, and aggravated assault represent the largest costs for the state, and when compared to other states in the US, “the harm from underage drinking averages $1,439 per youth” [27]. In 2012, young people ages 12 to 20 accounted for 9% of all alcohol-abuse treatment hospitalizations in the state. For this reason, Illinois has made prevention of underage drinking a top public health priority, and has implemented laws and penalties for driving under the influence (DUI), for underage drinking [28], and evidence-based strategies for reducing underage drinking. These include but not limited to the establishment of minimum drinking age (effective January 1, 1980), and the Illinois Zero Tolerance legislation that make it illegal for those under the age of 21 to drink and drive (Effective January 1, 1995). In addition, limiting the availability of alcohol (social, economic and physical) to young people through policies (e.g., a Class A misdemeanor to parents or guardians who knowingly permit their residence to be used by those under age 21) [28]. The adoption of a national minimum drinking age of 21 for “purchase or public possession” of alcohol [29], with strategies implemented via a comprehensive prevention approach. However, in Illinois, the detailed epidemiology of the relationship between the impacts of alcohol policies on unintentional injury has not been fully described.

We focused on unintentional injuries, defined as those not resulting from self-harm or suicide attempts as used in previous studies [30, 31]. In Illinois, each hospitalization at a trauma center is given an external cause of injury (E-Code) based on the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM), which is used to distinguish between intentional and unintentional injuries (Additional file 1).

Patient level covariates were selected based on previous literature and theory [15, 20]. Before analysis, the decision was made to classify race into white versus minorities (Chinese, Japanese, Hawaiian, Filipino, Korean, Asia Indian, Vietnamese, Samoan, Black or African American, American Indian, Alaska Native and Aleut). Patients who reported self-pay or government-subsidized programs, including Medicaid/Medicare were considered to have “no private insurance,” with those whose likely source of payment included any kind of private (e.g., HMO/PPO/POS) or commercial insurance reported as having “private insurance.” Patients whose method of payment was listed, as other, unknown, not billed for any reason, was empty or not applicable were reported as “other” insurance.

A city-level alcohol-related legislation used in the current study was obtained from the Illinois Liquor Control Commission (ILCC)‘s survey of local liquor ordinances conducted since August 22, 2013, with the data updated as of July 15, 2015. Among other questions, the survey asked city representatives to respond to a question that asked 1) “are minors under 21 years old allowed in bars and taverns”, with yes or no response options. The legislation was selected because it has been more widely adopted by 57.4% of cities that participated in the survey (n = 514 of 895 cities) compared to those often used in previous studies (e.g., “dry” towns, on-premises sale or consumption of liquor in bars, restaurants, and banquet halls) [8]. We then linked the legislation information to patient information by using “city name” as a unique identifier using Microsoft Access. For analysis, categorical variables were labeled as categorical for ease of interpretation in the mixed models. For example, variables city, county, zip code, BAC level, admission year and cause of injury was treated as nominal while variables under 21 allowed in bars or taverns, sex, insurance status and race were labeled as scale. Fig. 1 presents the data linkages and final study sample used in the current study of the patients aged 10 to 19, derived in stages.

Descriptive and demographic data are presented as frequency and percent whereas the χ² test was used to test for differences in patient level factors between young patients that tested positive for BAC and those that tested negative for BAC. Annual numbers, percentages and age-adjusted incidence rates of traumatic unintentional injuries (per 100,000 population) and 95% confidence intervals (CIs) were estimated (based on first encounter, excluding 0.38% repeat hospitalizations) and by age group, race and insurance status. Age-adjusted incidence rates were age-adjusted to the 2000 U.S. Standard Population (Census P25–1130) for all traumatic unintentional injuries; with crude rates for hospitalization calculated using the 2010 US Census (denominator population).

For the primary analyses, mixed-effects logistic regression models were estimated to evaluate the effect of an alcohol legislation on traumatic unintentional injury hospitalizations after adjusting for patient-level variables age, sex, insurance and race, and the interactions of legislation with age group. Interactions assessed whether the legislation may not be effective to a certain age group. The mixed-effects analysis were used to account for dependence resulting from patients being nested within cities. Predictors were added sequentially to the model, including BAC level, to evaluate the unique effect of BAC level after inclusion of the independent variables, in which the fixed effects can be interpreted as conditional on cities, with all random effects fixed at zero (e.g., unit-specific models). At the patient level, continuous predictors (age) was centered near their grand mean; binary predictors remained uncentered. Patient-level means for all predictors were included to represent contextual effects (e.g., the incremental effects of city characteristics after controlling for patient characteristics), all of which were retained regardless of statistical significance for proper interpretation of patient-level effects. Finally, the necessity for random slopes was evaluated separately for each patient-level predictor.