| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH |
From the Center for Injury Research and Control, University of Pittsburgh, Pittsburgh, Pennsylvania; and Pacific Institute for Research and Evaluation, Calverton, Maryland.
Address reprint requests to: Harold B. Weiss, PhD, MPH, Center for Injury Research and Control, Presbyterian University Hospital, Suite B-400, University of Pittsburgh, 200 Lothrop St., Pittsburgh, PA 15213; E-mail: hw{at}injurycontrol.com.
 |
ABSTRACT |
|---|
TOP
 ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS: An International Classification of Diseases, 9th Revision (ICD-9-CM) pregnancy and injury criterion was applied to 1997 statewide hospital discharge data from 19 states (52% of the female population age 15–49 years). Cases were classified as assault-related with and without pregnancy-associated diagnoses. Pregnancy-associated and all women were compared using demographic specific rate ratios and severity stratification.
RESULTS: There were 7402 assault-related injuries among all women ages 15–49; 745 (10%) were pregnancy-associated. The incidence of assaults involving hospitalizations was 65 per 100,000 person-years for pregnant women versus 21 per 100,000 for all women (rate ratio 3.14; 95% confidence interval [CI] 2.04, 3.39). Pregnant women were younger (mean age 24.2 versus 30.8 years), their average length of stay was shorter (2.6 versus 4.0 days), their mean injury severity score was less (2.5 versus 4.9), and the median charge per stay was lower ($3351 versus $6775). Unadjusted age-specific rate ratios (pregnant versus all women) for assaults were significantly higher for younger women 15–19 (rate ratio 7.22; 95% CI 4.81, 8.38), but when restricted to cases with injury severity scores of at least 4, most differences in rate ratios disappeared, except in the youngest women.
CONCLUSION: While pregnant women were more likely to be hospitalized for assaults, most of the increase is associated with their lower hospital admission threshold and increased rates of both pregnancy and assaults among young and nonwhite women. Practitioners should perhaps think of pregnant women more as a "sensitive" rather than a "high-risk" group.
The National Institute of Justice has estimated that 79,000 women a year, mostly young, are hospitalized because of physical assault, usually perpetrated by male acquaintances.1 Many of these women are pregnant at the time of the assault, adding a host of concerns for the woman, her fetus, and clinicians. Violence against women during pregnancy is an issue that stirs broad interest and concern. It is disturbing to even imagine that violence can intrude upon this poignant period in a woman’s life, yet intimate partner violence, if it exists in a relationship prior to pregnancy, does not always stop because a woman becomes pregnant. Whether it is more likely to increase or decrease during pregnancy is not known.
The issue of violence against pregnant women has received attention in the literature through three perspectives. The most common addresses issues related to fetal outcome. These studies have examined the impact of violence on fetal outcomes such as low birth weight, prematurity, or mortality.2–10 Complementing these studies are those that explore the impact of violence on maternal health, looking at physical, reproductive, and psychological parameters of health and disease.11–15
Both fetal and maternal perspectives benefit from a third focus that measures the prevalence of violence against pregnant or recently pregnant women to answer the question of whether pregnancy changes the risk or nature of violence. Older literature often reported higher rates of violence toward pregnant women.16–18 However, these findings began to be questioned when it was pointed out that both violence and pregnancy rates are higher in younger women.19 Design weaknesses (non-population-based, small shelter, or clinic-based populations, lacking representativeness), differences in definitions of violence (physical, sexual, threats, psychological), different periods of coverage (violence around the time of pregnancy versus violence during pregnancy), and a lack of severity adjustment and comparison populations, have left the question of pregnancy assault risk unanswered.3,12,15,20,21
While most subjects of violence against women are not hospitalized, concentrating on hospitalized cases has several advantages. First, it focuses on serious injury. These are important because of the severity of the injury to the individual, the increased risk to the fetus, and the cost to society. Second, the existence of large population-based hospital discharge data systems makes it possible to examine prevalence and construct comparisons. Third, hospital data are comparable across states, making large-scale aggregation feasible. Fourth, discharge data contain charge information that can be used for modeling cost estimates. Finally, unlike clinic and emergency department settings where the encounter is brief, inpatients have more time to confide in and relate the abusive nature of the injuries to health care personnel.
The first population-based study of hospitalized maternal injury was conducted by Greenblatt et al.22 They looked at Maryland hospital discharge data for the period 1979–1990. Among 80,311 injured women aged 15–45 years, 2.7% were reported to be pregnant. By examining assault E-codes (E970-E969) in computerized discharge data (E-codes are part of the International Classification of Diseases-Clinical Modification, 9th Revision, or ICD-9-CM, dealing with injury mechanisms and intent) they reported that 10% of the hospitalized injuries involving pregnant women were assault-related and that the rate ratio (comparing pregnant patients to all women aged 15–45) for assault-related hospitalization was 1.14 (not statistically significant). While this study brought creative methodological approaches, it contained drawbacks including incomplete E-coding, use of diagnosis codes for screening that were not as refined or as expansive as desired, and it was done before accreditation mandates for hospital identification of victims of abuse were common. Recognizing these issues, the authors recommended that their analyses of pregnancy-associated injury hospitalizations be repeated.
This recommendation was taken up by a study by Weiss, which borrowed from the Greenblatt methods, applying them to Pennsylvania’s 1995 hospital discharge data.23 This study, with more diagnosis fields to search and an improved search algorithm, found 761 (4.6%) of injury discharges to women of reproductive age were associated with pregnancy. Rate ratios were significantly higher for assaults (rate ratio 3.04; 95% confidence interval [CI] 2.45, 3.78), with the increased risk concentrated in young women. This pilot study recognized the challenge of differentiating how much of the observed increases were due to increased injury rates versus increased hospitalization rates because of evidence that pregnant injured women are more likely than nonpregnant women to be hospitalized for less serious conditions.22,24 However, the small numbers of pregnancy-associated assaults in that study (n = 89) limited the utility of adjusting for this concern. In addition, there were no perpetrator codes in 1995 with which one might distinguish intimate partner violence from other forms of violence. The current study fills those gaps by focusing on assault-related hospitalizations from a large, population-based, multistate hospital discharge database.
|
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Data were solicited from states that mandated E-coding for 2 years or more or exhibited an E-code completeness rate of 90% or better and at least five diagnosis fields to search for pregnancy-associated codes. Three states with large populations and fairly good completeness (> 60%) but not mandated E-coding were also included. 1997 data was chosen as it was the first year that perpetrator-specific codes were used and it was followed by 2 years national regulations regarding hospital screening for domestic violence.
States were contacted and arrangements made to receive nonconfidential versions of statewide discharge data. Data were received from 19 states (Arizona, California, Florida, Maine, Maryland, Massachusetts, Michigan, Nebraska, New Hamphire, New Jersey, New York, Pennsylvania, Rhode Island, South Carolina, Utah, Vermont, Virginia, Washington, and Wisconsin) whose population made up 51.9% of US women aged 15–49. The 19 states represented the hospitalization experience of 36 million women who were residents of those states and 1.9 million resident births.25 The combined dataset covered complete counts from about 2000 hospitals reporting 176,267 injuries to women aged 15–49.
The data underwent extensive editing, filtering, groupings, and development of derived variables to enhance compatibility, coding validity, and usability. Detailed algorithms were applied to identify injuries and to exclude cases of noninjury (eg, surgical and medical care, place of injury codes, adverse drug effects, and late effects).
Costs, estimated in 1996 dollars, were input for each record using a model derived from charges listed in the discharge record and diagnosis codes. Inputs into the cost model included data from the National Medical Expenditure Survey, the National Health Interview Survey, the Civilian Health and Medical Program of the Uniformed Services, and from national and state hospital discharge systems. Monetary measures in this study included total hospital charges, lifetime medical costs, lifetime productivity loss, and lifetime monetized quality-adjusted life years. Cost methods are detailed elsewhere.26,27
Injury severity was calculated using ICDMAP-90 (Tri-Analytics Inc., Bel Air, MD), a computerized injury coder that assigns injury severity scores based on ICD-9-CM injury diagnoses. Injurity severity score is a widely used score based on an anatomically based threat-to-life scale that ranges from 1 (minor) to 75 (unsurvivable).28
Pregnancy association was defined by examining diagnosis fields for ICD-9-CM diagnostic codes including complications of pregnancy and childbirth, certain conditions originating in perinatal period, and "V" codes for normal and high-risk pregnancy, postpartum care after delivery, outcome of delivery, and antenatal screening.
The above steps were applied to all age and gender injury discharges from the 19 states (n = 1,220,506) progressively limiting the data to females 15–49 years (n = 176,267), with acute care visits (n = 156,713), with valid injury E-code assignments (n = 144,260), who were residents of the state (n = 137,887), with assault-related hospitalizations (n = 7402).
Incidence rates were calculated per 100,000 person-years. For the pregnant population, denominators were derived from state-specific birth data and adjusted downward to account for the 9-month period of gestation and the assumption that the pregnancies would not be detectable in the hospital discharge data during the first 2 months of pregnancy.
Rate ratios were constructed between pregnant and all women for different comparison groups. This comparison, rather than a pregnant versus "nonpregnant" group contrast, was done for several reasons. After subtracting pregnancy-associated cases, the referent group still contains some pregnant women in the first 2 months of their pregnancy and other pregnant women not detected by the diagnosis algorithm, thus, it would be a misnomer to label it a nonpregnant group. Secondly, because the desire was to compare pregnant women with nonpregnant women, the comparison takes into account the 5-month period of every pregnancy year in which pregnant women are not detectably pregnant (ie, pregnant women contribute person-years to both groups because they are not pregnant over an entire year).
Rate ratios were calculated by dividing the group-specific (age, race, mechanism, intent, etc) rate for pregnancy-associated injury discharges by the group-specific injury rate. Point and 95% CI estimates of the rate ratio, comparing the pregnant and all injured women aged 15–49, were computed per standard methods.29
Assaults to women of reproductive age were analyzed to present prevalence rates and rate ratios for specific subgroups. Next, to adjust for the increased propensity of pregnant women to be hospitalized because they are pregnant, assaults were reanalyzed only for cases with an injury severity score of four or greater.
|
RESULTS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Among injured females 15–49 years with a pregnancy-associated diagnosis, 14% (745/5498) were assault-related (65 per 100,000 person-years), for all injured women it was 5% (7402/137,887, 21 per 100,000 person-years). The rate ratio was 3.14 (95% CI 2.04, 3.39).
Among nonwhite injured females with a pregnancy-associated diagnosis, 21% (427/2082) were assault-related, while for whites it was 9% (235/2635). The rate of pregnancy-associated assaults was almost seven times higher in nonwhites (178 per 100,000 person-years) than whites (26 per 100,000 person-years). However, the rate ratio was elevated similarly among both whites (2.65; 95% CI 1.41, 3.03) and nonwhites (3.34; 95% CI 2.55, 3.69). Among nonwhites aged 15–19, the rate of pregnancy-associated assaults per 100,000 person-years was 341 (rate ratio 5.54; 95% CI 4.32, 6.73).
Pregnancy-associated assault victims were younger compared to all women 15–49 years (mean age 24.2 versus 30.8 years). The proportion of pregnancy-associated assaults within each age group climbed sharply after age 16, peaked at age 19, and declined slowly thereafter (Figure 1
). The pregnancy-associated rates and rate ratios were highest in the youngest age group, declining with age (Figure 2).
|
|
Pregnancy-associated assault cases were more likely to be paid for by Medicaid (rate ratio 4.49; 95% CI 4.06, 4.98). The median charge per visit was $3351 for pregnancy-associated assaults versus $6775 for all assaulted women. Respective total costs for lifetime medical loss sum and lifetime monetized quality-adjusted life years (rounded) were $4,926,000, $6,296,162, and $71,620,000 for pregnancy-associated assault cases and $89,245,000, $111,545,000, and $1,689,194,000 for all assaults.
Among the top three body part groupings, pregnancy-associated assault rates per 100,000 person-years and rate ratios were: trunk 14.7 (rate ratio 19.6; 95% CI 16.2, 23.7), face 8.9 (rate ratio 2.0; 95% CI 1.6, 2.4), and abdomen and pelvic organs 8.2 (rate ratio 3.8; 95% CI 3.1, 4.7).
Perpetrator coding was incomplete for both pregnancy-associated and all assaults. Among pregnancy-associated assaults, 22.6% were accompanied by a perpetrator-related E-code; for all assaulted women the figure was 8.8%. Among the cases that were perpetrator coded, 88.0% and 83.7% were spouse or partner related among pregnancy-associated and all assaults, respectively.
The mean injury severity score among the pregnancy-associated assaulted women was 2.5, while the mean injury severity score among all women was 4.9. Figure 3 shows the rate ratio of assault-related hospital discharges by severity group. There was a significantly increased rate ratio for minor injuries (injury severity score less than 4) but not for the moderate, serious, and severe injuries. This finding was used as the basis for the severity adjustment, which eliminated assault-related cases with minor injuries from rate comparisons.
|
details the frequency, rates, and rate ratios of selected characteristics for hospitalized assaults in the subgroup of seriously injured cases. Proportionally eliminating the less severe pregnancy-associated cases reduced most of the rate ratios to values not significantly different from one. The overall rate ratio fell to a non-significant 1.07 (95% CI 0.57, 1.28). However, rate ratios were significantly greater than one for a few subgroups, including the youngest age group (rate ratio 2.49; 95% CI 1.31, 3.63) and for firearm-related assaults (rate ratio 1.55; 95% CI 1.07, 2.23).
|
|
DISCUSSION |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
While this study described a significant increase in the nonage and nonrace stratified rate ratio for pregnancy-associated assaults, it showed the importance of examining the issue of pregnancy risk by: Age, because assaults against women occur most often in the young and young women are most likely to be pregnant; race, because rates of assaults are much higher in nonwhites and pregnancy rates are higher among nonwhites; and severity, because pregnancy-associated cases are much more likely to be admitted for less serious injuries.
This study demonstrated that both age and race-specific rate ratios were markedly reduced once adjusted for injury severity. For both whites and nonwhites overall, after severity adjustment, there was no significantly elevated rate ratio. Moderate increases in the rate ratio remained among the 15–19 years age group (significant in nonwhites, not significant for whites) and for firearm-related assaults.
While hospital discharge data have significant advantages, they also have disadvantages. Waller and colleagues recently described these as they relate to violence against women.30 They include concerns about quality and completeness of intent and perpetrator coding, difficulty detecting conditions that are not injury-related (stress, depression, and other diseases), and possible duplicate counts.
Like all secondary data analyses, hospital discharge data are affected by the quality of the data.31–33 For intentional injuries, methods for screening and documentation are not always specified or consistently applied. As long as these vagaries are consistently applied within and among hospitals, the results contrasting pregnant women may be more valid from a comparative standpoint, but less so from a vantage seeking accurate rates. Miscoding and undercounting will occur, but it is difficult to conjecture how systematic inclusions of pregnancy-associated codes among nonpregnant assaulted women, the type of error that could most affect the results, would happen. While the data suggested that most hospitalized assaults were spouse or partner related, the low percentage of perpetrator coded cases dictates caution with this interpretation. Regarding duplicate counts, individuals would have needed multiple admissions with both a pregnancy and an assault code, rendering multiple admissions in our study population less likely.
Other limitations stem from the etiologic nature of the study design. Individual women were not followed up, thus, the study did not elucidate violence patterns before, during, or after pregnancy. The study could not describe the relationship of violence to pregnancy intent, sexual assault, gestational age, previous births, parity, prenatal care, pregnancy outcome, marital status, amount of time spent with an intimate partner (exposure), or relationship of the fetus to the assailant. Understanding the roles these factors play remains for future longitudinal research to characterize. Finally, women in early pregnancy are not likely, or at best, much less likely, than women in later pregnancy, to have the pregnancy identified and coded during a hospital stay and would be misclassified into the nonpregnant group. Therefore, a diagnosis-based pregnancy definition, such as that used in the current study, does not measure early pregnancy-associated assault risks very well, whether it ends in spontaneous or induced abortion or progresses to birth.
It may be helpful for practitioners to think of pregnant women as a "sensitive" rather than a "high-risk" population. They are sensitive to being hospitalized because the hospitalization threshold appears to be considerably lower for pregnant women with trauma than their non-pregnant counterparts. The perceived increase in risk is mostly fostered by a combination of this lower threshold and higher rates of both pregnancy and assaults among young and nonwhite women. As a sensitive population, pregnant women may be a special group worth addressing for preventive efforts. However, such attention should probably take place in conjunction with broader efforts aimed at reducing the differential of the rate of assault by age, socioeconomic status, and race, regardless of current pregnancy status.
|
Footnotes |
|---|
This work was based in part on methods pioneered by Jesse Greenblatt, Andrew Dannenberg, and Christopher Johnson. We gratefully acknowledge the assistance of Karen Zuri in coordinating the state data requests.
Received February 11, 2002. Received in revised form March 28, 2002. Accepted May 2, 2002.
|
REFERENCES |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. Bullock L, McFarlane J. The birth-weight/battering connection. Am J Nurs 1989;89:1153–5.[Medline]
3. Gazmararian J, Lazorick S, Spitz A, Ballard T, Saltzman L, Marks J. Prevalence of violence against pregnant women. JAMA 1996;275:1915–20.[Abstract]
4. Parker B, McFarlane J, Soeken K. Abuse during pregnancy: Effects on maternal complications and birth weight in adult and teenage women. Obstet Gynecol 1994;84: 323–8.
5. Cokkinides VE, Coker AL, Sanderson M, Addy C, Bethea L. Physical violence during pregnancy: Maternal complications and birth outcomes. Obstet Gynecol 1999;93: 661–6.
6. Berenson A, Wiemann C, Wilkinson G, Jones W, Anderson G. Perinatal morbidity associated with violence experienced by pregnant women. Am J Obstet Gynecol 1994; 170:1760–9.[Medline]
7. Dye T, Tolivert N, Lee R, Kenney C. Violence, pregnancy and birth outcome in Appalachia. Paediatr Perinat Epidemiol 1995;9:35–47.[Medline]
8. Campbell J, Torres S, Ryan J, King C, Campbell DW, Stallings RY, et al. Physical and nonphysical partner abuse and other risk factors for low birth weight among full term and preterm babies: A multiethnic case-control study. Am J Epidemiol 1999;150:714–26.
9. McFarlane J, Parker B, Soeken K. Abuse during pregnancy: Associations with maternal health and infant birth weight. Nurs Res 1996;45:37–42.[Medline]
10. Murphy CC, Schei B, Myhr TL, Du Mont J. Abuse: A risk factor for low birth weight? A systematic review and meta-analysis. CMAJ 2001;164:1567–72.
11. Gazmararian J, Adams M, Saltzman L, Johnson C, Bruce F, Marks J, et al. The relationship between pregnancy intendedness and physical violence in mothers of newborns. Obstet Gynecol 1995;85:1031–8.[Abstract]
12. Campbell JC, Moracco KE, Saltzman LE. Future directions for violence against women and reproductive health: Science, prevention, and action. Matern Child Health J 2000;4:149–54.[Medline]
13. Campbell J. Abuse during pregnancy: Progress, policy, and potential. Am J Public Health 1998;88:185–7.
14. Campbell J, Pugh L, Campbell M, Visscher M. The influence of abuse on pregnancy intention. Womens Health Issues 1995;5:214–23.[Medline]
15. Gazmararian JA, Petersen R, Spitz AM, Goodwin MM, Saltzman LE, Marks JS. Violence and reproductive health: Current knowledge and future research directions. Matern Child Health J 2000;4:79–84.[Medline]
16. Gelles R. The violent home: A study of physical aggression between husbands and wives. Beverly Hills, California: Sage Publications, 1974.
17. Eisenberg S, Micklow P. The assaulted wife: ‘Catch 22 revisited.’ Women’s Rights Law Report 1997;3:138–61.
18. Berrios DC, Grady D. Domestic violence. Risk factors and outcomes. West J Med 1991;155:133–5.[Medline]
19. Gelles R. Violence and pregnancy: Are pregnant women at greater risk of abuse? J Marriage Fam 1988;50:841–7.
20. Campbell JC. Abuse during pregnancy: A quintessential threat to maternal and child health–so when do we start to act? CMAJ 2001;164:1578–9.
21. Ballard T, Saltzman LE, Gazmararian J, Spitz AM, Lazorick S, Marks JS. Violence during pregnancy: Measurement issues. Am J Public Health 1998;88:274–6.
22. Greenblatt JF, Dannenberg AL, Johnson CJ. Incidence of hospitalized injuries among pregnant women in Maryland, 1979–1990. Am J Prev Med 1997;13:374–9.[Medline]
23. Weiss H. Pregnancy-associated injury hospitalizations in Pennsylvania, 1995. Ann Emerg Med 1999;34:626–36.[Medline]
24. Poole G, Martin J, Perry K, Griswold J, Lambert C, Rhodes R. Trauma in pregnancy: The role of interpersonal violence. Am J Obstet Gynecol 1996;174:1873–8.[Medline]
25. National Center for Health Statistics, Centers for Disease Control and Prevention, US Department of Health and Human Services. 1997 Natality data set. 1999.
26. Gold M, Siegel J. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996.
27. Miller T, Lawrence B, Jensen A, Waehrer G, Spicer R, Lestina D, et al. Estimating the cost to society of consumer product injuries: The revised injury cost model. Bethesda, Maryland: U.S. Consumer Product Safety Commission, 1998.
28. MacKenzie E. Injury severity scales: Overview and directions for future research. Am J Emerg Med 1984;2: 537–49.[Medline]
29. Rosner B. Fundamentals of biostatistics. 4th ed. Belmont, California: Duxbury Press, 1994.
30. Waller AE, Martin SL, Ornstein ML. Health related surveillance data on violence against women. Violence Against Women 2000;6:868–903.
31. Smith G, Langlois J, Buechner J. Methodological issues in using hospital discharge data to determine the incidence of hospitalized injuries. Am J Epidemiol 1991;134:1146–58.
32. Sniezek J, Finklea J, Graitcer PL. Injury coding and hospital discharge data. JAMA 1989;262:2270–2.[Abstract]
33. Marganitt B, MacKenzie E, Smith G, Damiano AM. Coding external causes of injury (E-codes) in Maryland hospital discharges 1979–88: A statewide study to explore the uncoded population. Am J Public Health 1990;80:1463–6.
This article has been cited by other articles:
|
|
 |
|
  D. El Kady, W. M. Gilbert, G. Xing, and L. H. Smith Maternal and Neonatal Outcomes of Assaults During Pregnancy Obstet. Gynecol., February 1, 2005; 105(2): 357 - 363. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Splinters & Fragments Inj. Prev., March 1, 2003; 9(1): 96 - 96. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
4 by Selected Characteristics, 19 States, 1997