Domestic-Violence Hospital Emergency Department Visits

face of a prisoner

The U.S. National Electronic Injury Surveillance System All-Injury Program (NEISS AIP) makes readily available nationally representative data on injury-related visits to hospital emergency departments, including injury-related visits caused by violence. The National Crime Victimization Survey (NCVS) also includes such data, as does the National Violence Against Women Survey (NVAWS). With careful estimation, NEISS provides provides better-quality data on injury-related visits from domestic violence than does NCVS or NVAWS. Estimates based on NEISS indicate that, in 2001, women and men made respectively 360,000 and 240,000 visits to hospital emergency departments for injuries from domestic violence.

To facilitate estimation and comparison, NEISS, NCVS, and NVAWS perpetrator categories are condensed into five perpetrator groups. The underlying NEISS report includes 10 perpetrator categories, while the NCSV and NVAWS include 20 and 55 perpetrator categories, respectively. The available categories in all three surveys are sufficient to form reasonable and roughly comparable perpetrator groups for intimate partner violence and domestic violence. The statistics estimated here concern men and women, defined as persons ages 18 and older. The surveys include age variables. Hence statistics for other age/sex groups could be similarly estimated.

NEISS has a large share of unspecified-perpetrator incidents. In 2001, “violence – unspecified perpetrator” amounted to 84% and 55% of violent injuries recorded in the NEISS sample for men and women, respectively. Unspecified perpetrators of violence against men and against women account for only 3% and 2% of injury-related visits to hospital emergency departments in the National Crime Victimization Survey (NCVS), 1995-2004. NVAWS, for incidents recalled across childhood and adulthood, reports less than 1% unspecified perpetrator injury-related visits to hospital emergency departments. NVAWS cannot credibly identify perpetrators better than NEISS and NCVS across a much large time span. That’s merely one of many serious weaknesses of NVAWS.

Different categorizations of multiple-perpetrator incidents contribute to relatively large NEISS unspecified-perpetrator shares. In the 2001 NEISS, multiple perpetrators caused 6.4% and 3.7% of violent injuries to men and women respectively. NEISS does not further describe multiple perpetrators. Multiple-perpetrator incidents in NEISS are swept into the unspecified-perpetrator incident group. NCVS, in contrast, allows multiple perpetrator categories to be reported for an incident involving multiple perpetrators. Hence NCVS allows a single incident to be allocated (fractionally) to multiple perpetrators. For 1995-2004, NCVS indicated that multiple perpetrators caused for men and women, respectively, 38% and 16% of violence-related injuries that prompted a hospital emergency department visit, and 8% and 4% of all victimizations. The NVAWS included one multiple perpetrator category, “male & female stranger.” Less than 0.1% of victimizations reported in detail in the NVAWS were for this perpetrator category. Multiple perpetrator incidents seem to have been described in the NVAWS as separate incidents, reported in detail subject to a biasing perpetrator-category constraint.

NEISS incident location helps to categorize incidents with multiple and unspecified perpetrators. For adjusted estimates, unspecified-perpetrator incidents that occurred within the home are divided by sex and allocated to domestic violence by the sex-specific share of domestic violence in perpetrator-specified incidents within the home. Unspecified-perpetrators incidents that occurred outside the home are similarly allocated by sex to domestic violence. The validity of NEISS statistics on perpetrator-subcategories of violence depend on the extent to which NEISS specified perpetrators are representative of unspecified perpetrators, controlling for sex and location.

NCVS estimates here do not separately account for series incidents. In NCVS, a series incident encompasses a six or more similar incidents in a six-month period. To be treated as a series incident, the respondent must not be able to “recall enough details of each incident to distinguish them from each other.”^ Only the most recent incident in a series incident is reported in detail in the NCVS. From 1995 to 2004 among persons 18 years old and older, series incidents accounted for 3% of incidents leading to hospital emergency department visits. Domestic violence often produces repeated incidents. Estimates of intimate partner violence against females showed rates 25% to 49% lower than weighting series incidents for 6 to 10 times, where 10 is the cap on the frequency reported.^ Those estimates include all forms of violence, including verbal threats that are classified in the NCVS as simple assaults. Series incidents that include hospitalization for each incident in the series cannot be determined. The frequency of such (annual) series incidents is likely to be much lower than the frequency of series incidents in general.

For analyzing serious injuries from violence, NEISS provides much better quality data than NVAWS. In addition to its implausible identification share of perpetrators across a lifetime recall period and its problematic treatment of multiple-perpetrator incidents, NVAWS has a confusing question frame and unclear incident bounding. NVAWS had high sample attrition and no sample weighting. NVAWS integrated relatively narrow policy concerns closely with the collection of the data. NVAWS treated men and women separately. Because NEISS is clearly superior to NVAWS, NVAWS will not be considered further here.

With reasonable adjustment for unspecified perpetrators, NEISS provides better estimates of serious injury from domestic violence than does the NCVS. Similarly to the National Hospital Ambulatory Care Survey (NHAMCS), NEISS shows about five times as many injury-related hospital emergency-department visits from violence as does the NCVS.^ NEISS collects data on violent injuries from hospitals emergency departments. NCVS collects such data from survey subjects’ recall. NEISS data collection is thus more directly related to violent incidents in time and circumstances. Moreover, the NEISS sample of violent-injury-related visits to hospital emergency departments in 2001 is about 300 times as large as the corresponding NCVS sample of incidents from 1995-2004. The NHAMCS credibly identifies about 2.5 times as many emergency department visits than does the NCVS even after a wide range of adjustments.^ NEISS figures are close to those of the NHAMCS. NEISS provides much more credible statistics for emergency departments visits than does the NCVS. NEISS is weaker than NCVS in identifying perpetrators. Reasonable adjustments can address that NEISS weakness. Overall, the better NEISS coverage of injury-related hospital emergency department visits from violence makes NEISS also better for estimating such visits from domestic violence.

The adjusted NEISS estimates are reasonable relative to the 1994 Survey of Injured Victims of Violence (SIVV). The SIVV violence total is roughly comparable with that from the 2001 and 2008 NEISS. The SIVV, with its focus on violent injuries, has a lower share of unspecified perpetrators than the NEISS: 35% and 18% for men and women, respectively, compared to 84% and 55% in the 2001 NEISS. After adjusting the NEISS data to allocate unspecified perpetrators, the 2001 NEISS has domestic violence shares in total violence of 29% and 67% for men and women, respectively. That’s reasonably comparable with the SIVV figures 21% and 60%. The NCVS, in contrast shows a much lower domestic violence share in total violence, particularly for men. That suggests that NCVS interviewees are reluctant to identify domestic violence.

For detailed data and comparison of NEISS, SIVV, and NCVS statistics, see dataset on serious injuries from domestic violence, intimate-partner violence and other violence by victim sex.

Leave a comment (will be included in public domain license)

Your email address will not be published. Required fields are marked *