We identified 240 women and 1380 men OEF/OIF Veterans who met the eligibility criteria for this study. The mean age of the study population was 33 years (Table ), with approximately 60% each being single, without any outside health insurance, and self identifying as being of white race. Over 60% had served in the military for less than 3 years, with the majority (70%) having served in the Army. During the observation period, 54% (874/1620) of the study population did not have any visits to the local VAMC for outpatient health care. Bivariate analyses showed that users and non-users of medical services did not differ significantly in age (p = 0.30), or length (p = 0.84) or branch (p = 0.23) of service. We observed a significant difference in utilization by gender, in that women were significantly more likely than men to have utilized care [p = .01].
Sample Characteristics by Health Care Utilization Status
Women non users compared to men non users (Table ) were significantly younger(mean age 29 vs. 33 years, p < 0.001) and were more likely to be from the army service branch (75% vs. 68%, p = 0.04). There was no significant difference by race (p = 0.7), marital status(p = 0.13) having private health insurance (p = 0.6), a service-connected disability (p = 0.7) or length of service (p = 0.3).
Sample Characteristics of Veterans Not Using Care by Gender
Among the individuals who utilized VA healthcare (746/1620), more women utilized care than men (53% vs. 45%, p < 0.01). Women health care users were more likely than men health care users to be younger (Table ) (mean age 31 vs. 34 years, p < 0.001), single (74% vs. 62 %, p < 0.001) and non-white (24% vs. 12%, p < 0.001). Men and women did not differ in having private health insurance (p = 0.7), a service-connected disability (p = 0.8) or length of service (p = 0.2). Women and men users showed no differences in the mean ± standard deviation (SD) (9.7 ± 10.6 vs. 9.47 ± 9.79, p = 0.8) and median (6.5 versus 6.0, p = 0.6) number of outpatient visits (Table ). The range of study time was between 0.5 years and 2.3 years for the visits. The average observation time for women and men was similar (0.71 ± 0.61 vs. 0.79 ± 0.63 years, p = 0.22). The average number of outpatient visits per person per year, for women and men was 16.3 and 15.4 visits per year, respectively (p = 0.7).
Sample Characteristics of Veterans Utilizing Care by Gender
Results of initiation of care analysis
Multivariable logistic regression analysis showed that, after adjusting for age, race, marital status, private health insurance, service-connected disability, length of service and type of service branch, women were more likely to initiate care than men (OR = 1.47; 95% CI: 1.09, 1.98) (Table ). Other characteristics associated with initiation of care included having a ≥30% service connected disability (OR = 2.73; 95% CI: 1.93, 3.85) and having private health insurance (OR = 1.44; 95% CI: 1.15, 1.86). For the logistic regression interaction model, the following interaction terms were included: sex*race, sex*marital status, sex*covered by health insurance, sex* service connected disability. No interaction terms were significant so they were omitted from the final model.
Adjusted Logistic and Negative Binomial Regression Models for Outpatient Care Use
Results of rate of care analysis
In the adjusted negative binomial regression model, the intensity of utilization among health services users was similar for women and men (incident rate ratio (IRR) = 1.07; 95% CI: 0.90, 1.27) (Table ). The rate did not differ by race (IRR = 0.90; 95% CI: 0.75, 1.09) or by percentage of service-connected disability (IRR = 1.03; 95% CI: 0.88, 1.21). Veterans with longer length of service(≥ 3 years) had significantly higher intensity of use (IRR = 1.14; 95% CI: 1.00-1.30). Veterans with health insurance were less likely than those without insurance to continue utilizing outpatient care at the VA (IRR = 0.77; 95% CI: 0.67, 0.88). The interaction model investigated the interactions between sex and age, marital status, race, and service branch, but none of these interaction terms was significant and they were omitted.