1. Intimate Partner Violence, Health Care Utilization and Insurance Status: Results from a Large Population-Based survey
Kenneth J. Steinman, PhD, MPH
Amy E. Bonomi, PhD, MPH
Academy on Violence & Abuse
April 25, 2009
Supported by a grant from the Ohio Family Health Survey

2. Background
IPV (intimate partner violence) is associated with HCU (health care utilization)
(Coker et al., 2004; Helweg-Larsen, 2003; Rivara et al, 2007)
Few population-based studies
Little consideration of insurance status

3. Goals Assess IPV-HCU association in a population-based sample
Does association vary by insurance status?

4. Ohio Family Health Survey
September 2008-January 2009
n=50,944 (23,083 women <65)
Computer-Assisted Telephone Interviews
Random Digit Dialing
Over-sampling of ethnic minorities, select counties
Representative of Ohio adults & households
Response rate: 47%

5. Dependent Variables
During the past 12 months, how many times were you a patient in…
an urgent care center?
a hospital emergency room?
a hospital overnight?

6. IPV measure
During the past 12 months, how many times, if any, has anyone hit, slapped, pushed, kicked or physically hurt you?
Think about the time of the most recent incident involving a person or persons who hit, slapped, pushed, kicked or physically hurt you. What was that person’s relationship to you? (open-ended)

7. intimate partner violence
classifying a case of intimate partner violence
How many times…?
Not a case
0 times
1+ times
What was that person’s relationship to you?
01 Stranger
02 Coworker
03 Professional caretaker
10 S/he is my Child
11 S/he is my Stepchild
12 Another family member
13 Acquaintance/friend
97 OTHER
98 DK
99 REFUSED
04 Male/Female first date
05 Someone you were dating
06 Former boyfriend/girlfriend
07 Current boyfriend/girlfriend or fiancé
08 Spouse or live-in partner
09 Former spouse or live-in partner
intimate partner violence
Other violence

8. Other Variables Age, ethnicity, region Socioeconomic status
Income, education, home ownership
Insurance status
Uninsured; Medicaid; Employer-based; Other

9. Data Analysis Bivariate association Generalized Linear Model
Covariates with IPV
IPV with dependent variables
Generalized Linear Model
Poisson distribution, log link
Controlling for age, ethnicity, SES
Weighted data; complex survey design

10. IPV prevalence n Weighted % 95%CI Total 351 1.9 [1.6-2.2] Age 18-24 75
5.3
[ ]
25-34 93
2.1
[ ]
35-44 85
1.8
[ ]
45-54 72
1.1
[ ]
55-64 26
0.4
[ ]
Race/Ethnicity
white
259
1.7
[ ]
black 70
3.3
[ ]
Hispanic 22
2.7
[ ]
Asian
--*

11. IPV prevalence (2) n Weighted % 95%CI
Income as % of federal poverty level
<100%
145
4.3
[ ] %
133
2.1
[ ]
>300% 73
0.8
[ ]
Home ownership
owns home
173
1.3
[ ]
does not own home
176
3.5
[ ]

12. IPV prevalence (3) n Weighted % 95%CI Education < High school 44
3.8
[ ]
High school graduate
259
1.3
[ ]
College Graduate 48
0.2
[ ]

13. IPV prevalence (4) n Weighted % 95%CI Insurance type Medicaid 127 5.2
[ ]
employer-based
105
0.7
[ ]
uninsured 83
4.5
[ ]
Other a 36
1.8
[ ]
a Other insurance types include: Medicare only, directly purchased, unknown

14. Association of IPV with Health Care Utilization:
Prevalence Rate Ratio (unadjusted)
95% CI
Prevalence Rate Ratio (adjusted)a
Urgent care
2.1
[ ]
1.6
[ ]
Emergency room
2.2
[ ]
1.5
[ ]
Hospital Admission
1.7
[ ]
1.3
[ ]
a Adjusted for age, ethnicity, income, education, home ownership, insurance status

15. Association of IPV with Hospital Admission: Variation by Insurance Status
Uninsured
Medicare
Employer-Based
Other
(n=2,924)
(n=2,998)
(n=13,763)
(n=2,395)
PRR
95% CI
Urgent care
2.3
[ ]
1.4
[ ]
1.3
[ ]
1.9
[ ]
Emergency room
1.7
[ ]
[ ]
[ ]
[ ]
Hospital admission
1.2
[ ]
1.1
[ ]
[ ]
3.0
[ ]
PRR=Prevalence Rate Ratio (adjusted for age, ethnicity, income, education, home ownership)

16. Discussion IPV – HCU association holds in general population
IPV prevalence varies by insurance status
IPV – HCU association may vary by insurance status
Smaller effects for among women with employer-based insurance (EBI)?

17. Implications Urgent care may be important source of care
EBI samples may underestimate IPV-HCU association
Include insurance as a component of SES
Useful alternative measurement approach

18. Thank you