The present study aimed to elucidate organizational factors related to weapons confiscation in hospitals under conditions typical of hospitals in the United States. This study utilized a cross-sectional survey design. This study examines weapons confiscation and its associations with metal detector presence, and other factors including hospital settings (urban/rural settings, region where hospital is located), type of hospital, individual hospital departments where firearm incidents may have occurred (emergency department or other), and hospital policies concerning staff being armed.
This study utilized a validated anonymous on-line cross-sectional survey to collect data about security programs from security professionals who were members of the International Association of Healthcare Security and Safety (IAHSS; http://www.iahss.org/) and working in the United States. The typical member of this organization is a security professional, often the director of security, at a hospital or within a hospital system. Survey respondents were predominantly male, white, and English speaking at home. Of the 77 survey respondents, only three respondents were female, two were African American, and 1 was Hispanic. The survey participants represented security programs across 34 states, with the highest number of respondents coming from California (5); Illinois (6); North Carolina (5); Pennsylvania (8); Florida (4); and Massachusetts (4). Facilities described by survey participants were urban (49%), suburban (30%), or rural (21%), and most often described as offering general acute care hospital services (92%), trauma hospital (56%), psychiatric services (49%), and outpatient clinics (70%). Programs associated with hospitals typically were in the range of 200—400 beds (39%), less than 200 beds (23%) or greater than 400 beds (29%). As such, the survey participants represent a wide-range of facility locations and types, and, thus, this survey provides a broad picture of security programs and policies across the United States.
This survey was broad in scope and asked questions that covered a wide variety of issues related to firearms and other weapons encountered in a healthcare setting. The survey notice was sent out to over 2,200 IAHSS members on three occasions over a 6 month period of time with approximately 2 months between each contact. The survey notice was embedded in the IAHSS monthly electronic E-mailed newsletter. Details of survey design, validation, and sampling can be found in Blando et. al (2018). This analysis was confined only to those survey respondents who worked for healthcare facilities in the United States of America.
In this assessment, factors that may contribute to the frequency of weapons confiscation in a healthcare facility were assessed. The dependent variable was weapons confiscation. Weapons included any device where the purpose and intent is that it could be used to injure an individual. Examples of commonly encountered items include guns, knives, shanks, hammers, clubs, scissors, screw drivers, and razor blades.
An analysis of the self-reported frequency of weapon confiscation by the survey respondents was analyzed in connection with potential explanatory predictors using Chi-Square tests and logistic regression, both unadjusted and adjusted.
Predictors/covariates:
Predictor/covariate variables examined included the following: a) perception of their facility to be at high risk of violence (yes vs. no); b) whether the survey respondent reported their facility decision to arm guards was a response to violence in the community (yes vs. no); c) whether the state where they work has open-carry firearm laws (yes vs. no), d) whether the hospital where they work currently use metal detectors (yes vs. no); e) the type of facility, including whether their facilities were trauma hospitals (yes vs. no) or psychiatric hospitals (yes vs. no); f) the type of community, categorized as rural, suburban, or urban; g) the number of firearms by state registered with the Federal Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) (above national median value or below); h) the state violent crime rate (above national median value or below), i) whether the state where they work requires mandatory background checks to purchase a firearm (yes vs. no), j) the size of their facility based on the number of inpatient hospital beds, k) and the Giffords Law Center to prevent gun violence (http://lawcenter.giffords.org/scorecard/; accessed April 14, 2018) firearms law grade given to the state where the survey respondent works (grades A & B versus grades C—F).
Outcome:
The outcome measure was the response to the survey question about the frequency of security confiscating weapons with four multiple choice options (daily, weekly, monthly, or less than monthly). The confiscation frequency was dichotomized by collapsing responses for daily and weekly into the “frequent” category, and monthly and less than monthly into the “non-frequent” category.
Analysis
Continuous predictor variables were dichotomized by coding those above the median as “high” and those below as “low” based on the following median values: the median statewide violent crime rate was 377 incidents per 100,000 people, the median registered firearms per capita was 0.0035 per person, and the median number of inpatient hospital beds was 300. Two-by-two tables of the confiscation frequency category by each listed categorical predictor or covariate were examined, and each distribution of counts in the tables evaluated for statistical significance using a Chi-square or Fisher’s Exact test. A P value of ≤0.05 was considered statistically significant. A univariate logistic regression was also run to determine the univariate unadjusted odds ratio of the predictors with regard to the outcome. Potential explanatory variables that were statistically significant at p ≤0.05 where then included in a multivariate logistic regression model (Hosmer and Lemeshow, 1989). Odds ratios and maximum likelihood estimates from the logistic regression were estimated according to the logistic procedure run in SAS v9.3.
The adjusted multivariate logistic regression model included the five predictor/covariate variables that were statistically significant at p ≤0.05 in the univariate analysis (see Table 1). All five variables considered were assessed for multicollinearity by computing variance inflation factors (VIF) and assessed for interactions using SAS v9.3. The final model included these five predictive factors as they were either significantly associated with the outcome and/or potential confounding factors.
Table 1
Unadjusted Associations of Hospital, Policy, and Legal Factors with Frequent (daily or weekly) Weapons Confiscation in Hospitals
Potential Predictors of frequent weapons confiscation
|
Frequency
|
Unadjusted Odds Ratio
|
95% Confidence Interval
|
P-value
|
*a) Health care facility perceived to be at high risk of violence (n=76)
|
Yes = 62
No (ref)= 14
|
12.99
|
1.60 - 105.50
|
0.016
|
*b) Decision to arm guards was a response to violence in the community (n=66)
|
Yes = 23
No (ref)= 43
|
4.33
|
1.47 - 12.70
|
0.0077
|
c) State allows open-carry of firearms (n=76)
|
Yes = 37
No (ref) = 39
|
1.095
|
0.44 - 2.72
|
0.85
|
*d) Use of metal detectors (n=75)
|
Yes = 37
No = 38
|
2.89
|
1.11 - 7.49
|
0.029
|
e) Trauma hospital (n=76)
|
Yes = 43
No = 33
|
1.52
|
0.60 - 3.85
|
0.38
|
*e) Psychiatric hospital
(n=76)
|
Yes = 38
No = 38
|
4.94
|
1.83 - 13.31
|
0.0016
|
*f) The type of community (n=76)
|
Suburban & urban = 60
Rural (ref) = 16
Missing = 1
|
7.00
|
1.46 - 33.50
|
0.015
|
g) The number of firearms registered (n=76)
|
High = 30
Low (ref) = 46
|
0.87
|
0.34 - 2.21
|
0.76
|
h) Statewide violent crime rate (n=76)
|
High = 35
Low (ref)= 41
|
0.85
|
0.34 - 2.13
|
0.73
|
I) Statewide mandatory background checks for firearm purchases ( (n=73)
|
Yes = 69
No (ref) = 4
|
0.23
|
0.023 - 2.30
|
0.21
|
J) The size of the facility based on the number of inpatient hospital beds (n=76)
|
High = 31
Low = 45
|
0.99
|
0.39 - 2.50
|
0.98
|
k) firearms law grade given to the state where the survey respondent works (n=76)
|
A&B (higher grades) = 24
C, D, F (lower grades) (ref)= 52
|
0.53
|
0.095 - 3.00
|
0.48
|
* statistically significant