effects of intimate partner violence versus poverty on …violence against women / october...

VIOLENCE AGAINST WOMEN / October 2001Sutherland et al. / VIOLENCE VERSUS POVERTY

Effects of Intimate Partner ViolenceVersus Poverty on Women’s Health

CHERYL A. SUTHERLANDEast Fife Women’s Aid

CRIS M. SULLIVANDEBORAH I. BYBEEMichigan State University

This article investigated whether women’s physical health symptoms were due to abuse,poverty, or both. A community sample of 397 women, about half of whom had beenassaulted by an intimate partner, was interviewed about their income, experience ofphysical abuse, and physical health. Hierarchical multiple regression revealed that bothincome and physical abuse contributed to women’s rates of physical health symptoms.Abuse contributed to the variance in physical health beyond that predicted by incomelevel alone. Findings suggest that abuse by an intimate partner or ex-partner negativelyaffects women’s health and is especially detrimental to the health of low-income women.

Intimate partner violence is a serious health concern for a significantnumber of women worldwide (Bachman & Saltzman, 1995; Coun-cil on Scientific Affairs, American Medical Association, 1992;Plichta, 1996). Cross-sectional and longitudinal investigations ofwomen from domestic violence shelter programs, emergencyrooms, and primary health clinic settings consistently demon-strate that intimate partner violence places women at risk forphysical health problems (Abbott, Johnson, Kozoil[ORKOZIOL?]-McLain, & Lowenstein, 1995; Alpert, 1995; Coker,Smith, Bethea, King, & McKeown, 2000; Eby, 1996; Sutherland,Bybee, & Sullivan, 1998; Tollestrup et al., 1999).

Compared to nonbattered women, survivors of physical abuseare more likely to suffer multiple physical health symptoms(Coker et al., 2000; Council on Scientific Affairs, 1992; Eby, 1996;

1122

AUTHORS’ NOTE: This research was supported by National Institute of MentalHealth Grant R01 MH44849, awarded to the second author.

VIOLENCE AGAINST WOMEN, Vol. 7 No. 10, October 2001 1122-1143© 2001 Sage Publications

Hamberger, 1994) and chronic health problems (Coker et al., 2000;Drossman, Talley, Leserman, Olden, & Barrio[OR BARREIRO?],1995; Talley, Fett, & Zinsmeister, 1995). Frequently reportedsymptoms include those associated with sleep problems such asfatigue, insomnia, and recurring nightmares (Eby, 1996; Suther-land et al., 1998); headaches; chest pain; back and limb problems;disturbing physical sensations (Coker et al., 2000; Drossman,1994; Drossman et al., 1995); stomach and gastrointestinal prob-lems (Coker et al., 2000; Drossman et al., 1995; Scarinci, McDon-ald-Haile, Bradley, & Richter, 1994; Talley, Fett, Zinsmeister, &Melton, 1994); respiratory problems such as choking sensations,hyperventilation, and asthma (Abbot et al., 1995); and gyneco-logical symptoms such as pelvic pain and menstrual problems(Coker et al., 2000; Eby, Campbell, Sullivan, & Davidson, 1995).

The extent to which women experience physical health symp-toms is directly linked to their experiences of violence. Womenwho experience more frequent and severe physical abuse aremore likely to report higher rates of health problems than arewomen who experience less violence (Campbell & Lewandowski,1997; Eby, 1996; Sutherland et al., 1998; Sutherland, Bybee, &Sullivan, in press). Research has demonstrated also that women’sphysical health symptoms gradually decrease over time as theirexperiences of abuse decline (Sutherland et al., 1998).

In many cases, the symptoms women experience appear unre-lated and have an unspecified origin. Health care providers oftencharacterize the complex array of symptoms as manifestations ofstress and have found them difficult to diagnose and treat. As aresult, some physicians have provided inadequate or harmfultreatment for women’s health symptoms without ever addressingthe underlying cause of the complaints. Improved identificationprocedures in emergency rooms and primary care facilities haverevealed that vague pain complaints when combined with otherphysical health symptoms may be strong indicators of intimatepartner violence (Attala, 1994; Drossman et al., 1995; Naumann,Langford, Torres, Campbell, & Glass, 1999; Saunders, Hamberger,& Hovey, 1993). When asked directly, women often attributephysical health problems to their partners’ physical and psycho-logical abuse (Abbott et al., 1995; Eby, 1996).

Despite the recent advances in knowledge about the relation-ship between intimate partner violence and women’s subsequent

Sutherland et al. / VIOLENCE VERSUS POVERTY 1123

physical health problems, it remains unclear whether health symp-toms that have been attributed to abuse are actually at least par-tially the result of women living in poverty. Prior research hasbeen based almost exclusively on the experiences of low-incomewomen, and in studies where a broader range of income levels hasbeen represented, income has not been examined as a factor deter-mining health outcomes (Campbell & Lewandowski, 1997; Camp-bell & Soeken, 1999; Coker et al., 2000; Tollestrup et al., 1999).

For decades, researchers have demonstrated a strong link be-tween poverty and poor health (for relevant literature reviews,see Feinstein, 1993; Kaplan & Keil, 1993). Living in poverty placeswomen at risk for numerous health problems, many of which par-allel those reported by survivors of intimate partner violence.Low-income women are more likely to suffer chronic health prob-lems (Kington & Smith, 1997; Luepker et al., 1993; Lynch, Kaplan, &Shema, 1997; Stronks, Van De Mheen, Van Den Bos, & Mackenbach,1997) and poorer general health status (Dunn & Hayes, 2000;Humphries & Van Doorslaer, 2000; Stronks, Van De Mheen, &Mackenbach, 1998) than are women with higher incomes. Theyendure unique stresses associated with living in poverty, such asincome instability, frequent moves, lack of transportation, andpoor housing conditions (Belle, 1990; Dunn & Hayes, 2000; Fuller,Edwards, Sermsri, & Vorakitphokatorn, 1993). They experience agreater number of financial stressors (e.g., job losses, unpaid bills,and inadequate housing) and have insufficient financial resourcesto address those and other undesirable events (e.g., frequent ill-ness and legal problems) than do middle- or high-income women(Dunn & Hayes, 2000; Mcleod & Kessler, 1990; Stronks et al., 1998).

In addition to the stresses associated with living in poverty, sev-eral other factors may predispose low-income women to poorhealth outcomes. Women with low incomes are more susceptibleto various illnesses and communicable diseases such as hepatitisand tuberculosis than are women with higher incomes because ofsubstandard housing and sanitation conditions (Collins et al.,1999). They are more likely to be exposed to toxic environmentalsubstances (Havenaar & Van Den Brink, 1997) and have limitedaccess to quality health care and preventive services (Alter, Naylor,Austin, & Tu, 1999; Bindman et al., 1995; Kington & Smith, 1997;Stronks et al., 1998; Weissman, Fielding, Stern, & Epstein, 1991).The combined influence of these factors places low-income women

1124 VIOLENCE AGAINST WOMEN / October 2001

at greater risk for chronic illness and other health problems. Onthe other hand, women with higher incomes have greater accessto resources, including quality health care and health mainte-nance programs (e.g., health club memberships). They have thefinancial resources to purchase nutritious foods and maintainhealthy eating habits and are more likely than women with lowincomes to participate in health promotion activities (Collins et al.,1999; Robinson, Caraher, & Lang, 2000).

Many survivors of intimate partner violence have reported simi-lar concerns, including unemployment, lack of transportation, sub-standard housing, and financial difficulties (Browne, Salomon, &Bassuk, 1999; Eby, 1996; Sullivan, Tan, Basta, Rumptz, & Davidson,1992). Because domestic violence investigators have generallyrelied on samples of low-income women, the extent to whichthese concerns are related to conditions of poverty versus inti-mate partner violence is unclear. Eby’s (1996) comparative studyof low-income abused and low-income nonabused women didindicate that low-income abused women reported higher stresslevels and poorer health outcomes than did the low-income non-abused women. However, without the inclusion of women from abroad range of income levels, the extent to which poverty contrib-utes to women’s stress and health has remained inconclusive. Thegoal of the present study was to determine whether intimate part-ner violence has a significant effect on women’s health beyondthat which can be explained by poverty. To that end, this studyinvolved interviewing women representing a broad range ofincomes, approximately half of whom had also experienced domes-tic violence.

METHOD

Participants were recruited into the study through advertise-ments in three newspapers, two of which were delivered free ofcharge to households in a medium-sized urban area in the upperMidwest. Four different advertisements were placed betweenAugust 1996 and June 1997.

To increase the likelihood of recruiting equivalent numbers ofwomen who had experienced intimate partner violence and womenwho had not experienced intimate partner violence and to maxi-mize the range of income levels in the sample, each of the four


advertisements was specifically designed to attract women whofit a particular profile. The first advertisement was designed torecruit middle-income women who had been physically harmedby an intimate partner, the second to recruit low-income womenwho had been physically harmed by an intimate partner, and thethird and fourth to recruit middle-income and low-income women,respectively, with no reference to experiences of physical abuse.The advertisements instructed interested women to contact theresearch office to get more information about the “Women’sHealth” project and to determine their eligibility for participation.

Women were considered eligible if they met the following crite-ria: (a) They were current residents of the same midsized Mid-western city, and (b) to reduce the confounding effect of age onreported health symptoms, they were between the ages of 18 and45. Each eligible woman was scheduled for a face-to-face inter-view at a time and location convenient for her. Interviews withtrained female interviewers took place in several different set-tings; most were scheduled at a local community organization,but some were conducted in women’s homes or at the local uni-versity. Of 439 women who scheduled interviews, 397 (90%) com-pleted them.

MEASURES

Data were collected through face-to-face interviews using astructured questionnaire. In addition to demographic informa-tion, the following constructs were assessed: physical abuse,income level, and physical health symptoms.

Physical abuse. Physical abuse was conceptualized as one ormore incidents in which the woman was physically harmed by anintimate partner/ex-partner or spouse/ex-spouse within the prior6 months. The relationship was considered intimate if the partici-pant indicated that the partner was or had been at least a boy-friend or girlfriend. If a woman had been casually dating and didnot consider any relationship intimate, she was not probed aboutthe occurrence of physical harm within the previous 6 months. Amodified 12-item version of the Conflict Tactics Scales (Straus,1979) was used to assess experiences of physical abuse during the


6 months prior to the interview. For each item about a specific typeof violent act (e.g., “kicked you”), responses ranged from 1 (never)to 6 (more than 4 times a week). Each participant’s scale score wascalculated as the mean frequency rate with which she experiencedthe different forms of physical abuse. The 12 items demonstratedhigh internal consistency (α = .93). Because this variable was posi-tively skewed (197 women received a score of 0 because they hadnot experienced physical abuse in the past 6 months), a log trans-formation was applied prior to analysis.

Income level. Income level was operationalized as the woman’shousehold income, adjusted for the number of adults and chil-dren supported by the income and expressed as a percentage ofthe U.S. Census Poverty Threshold Index (U.S. Department ofCommerce, Bureau of the Census, 1996). Each woman’s annualincome was divided by the appropriate poverty threshold (i.e.,income defined as the poverty threshold for the household con-figuration in 1996), then multiplied by 100 to yield a continuousvariable. One hundred percent poverty indicated the person’sannual income was at the poverty level, which for 1996 was equiv-alent to $8,163 for a single person or $10,815 for an adult and onechild.

Physical health symptomatology. Physical health symptom-atology was assessed through a modified version of the Cohen-Hoberman Inventory of Physical Symptoms (CHIPS) (Cohen &Hoberman, 1983). The original scale was modified to includephysical symptoms that are frequently reported by women withabusive partners, such as choking sensations, high blood pres-sure, and pelvic pain (Abbott et al., 1995; Campbell, 1989; Eby et al.,1995). Participants were asked to rate how often in the past 6 monthsthey had been bothered by each of 30 physical health symptoms.Their responses were rated on a 6-point scale ranging from 0(never) to 5 (more than 4 times per week). Previous research has indi-cated that this modified version of CHIPS is a valid and reliableindex of physical health symptomatology (Cohen & Hoberman,1983); it has been used successfully with samples of battered andnonbattered women (Eby, 1996; Sutherland et al., 1998). In the cur-rent sample, the internal consistency of the 30-item scale was .91.


RESULTS

RESEARCH PARTICIPANTS

Demographic characteristics of the study participants are pre-sented in Table 1. The participants were primarily White/Cauca-sian, in their 30s (M = 34, SD = 7.68), employed at least part time(66%), had some college experience (74%), and cared for two chil-dren (SD = 1.55). Most women (66%) said they were in an intimaterelationship at the time of the interview. Despite the medianhousehold monthly income of $1,600 (SD = $1,616), 129 women(33%) were surviving on monthly incomes below the povertythreshold for the number of people supported by that income.Nearly 20% of the women interviewed were uninsured for medi-cal care. As intended by the recruitment strategy, 52% of the par-ticipants reported that an intimate partner or ex-partner hadphysically harmed them within the 6 months prior to the interview.

The demographic characteristics of this sample were generallyrepresentative of the midsized Midwestern city population.According to 1990 census data (U.S. Department of Commerce,Bureau of the Census, 1990), the ethnic, educational, and employ-ment distributions of the current sample were comparable to thecity’s demographic characteristics. Women in this sample were aslikely to have private health insurance or at least Medicaid orMedicare as were people in the area in 1990 (Health Status Advi-sory Group Report, 1993). Despite the apparent similarities be-tween this sample and the city’s population, the sample medianannual income of $19,200 was substantially lower than the 1989estimated median household income for the city ($26,398) (U.S.Department of Commerce, Bureau of the Census, 1990).

PHYSICAL ABUSE

The 205 women who had been physically abused by an inti-mate partner in the previous 6 months reported between 3 and 8different types of physical abuse (M = 5.52, SD = 2.97). Eighteenwomen (9%) endorsed 1 type of physical abuse, and 3 women(1%) indicated their partners had used all 12 forms of violence. Onaverage, women’s partners or ex-partners assaulted them two or



TABLE 1Demographic Characteristics of the Sample (N = 397)

Characteristic N %

Race/ethnicityWhite/Caucasian 279 70Black/African American 66 17Multiethnic 25 6Hispanic/Latina 20 5Asian/South Pacific 4 1Native American 3 1

Age (mean = 34 years)17 to 29 years 121 3030 to 39 years 154 3940 to 45 years 122 31

Percentage poverty level≤ 125% (low income) 169 43126% to 200% 76 19201% to 600% (middle income) 146 37> 600% (high income) 6 2

Employment statusFull time 194 49Part time 69 17Temporary/seasonal 14 4Not employed 120 30

Relationship statusLiving together/married 153 38Not currently dating 120 30Girlfriend/boyfriend 108 28Dating, not girlfriend/boyfriend 16 4

Housing statusRenting apartment or house 172 43Own or buying a house 162 41Staying with friends/relatives 49 12Renting a room 6 2Homeless or other shelter 8 2

Education levelLess than high school 36 9High school graduate/GED 37 9Vocational/trade school 30 8Some college 191 48Bachelor’s degree 71 18Postbachelor’s degree 32 8

Medical insurancePrivate insurance 232 59Medicaid/Medicare 88 22None 77 19

Have children (mean = 2) 295 74

three times a month. The types of violence they used are presentedin Table 2. These findings coincide with reports from previous re-search on the number, severity, and frequency of physical assaultsby intimate partners (Eby, 1996; Sullivan & Bybee, 1999; Suther-land et al., 1998).

INCOME LEVEL

The sample’s household monthly incomes ranged from $0 tomore than $6,000. However, when accounting for family size,most women’s incomes fell into what are typically consideredlow- to middle-income ranges. Forty-three percent of the samplehad incomes at 125% of the poverty level or less; an additional19% were between 125% and 200% of poverty, and 37% werebetween 200% and 600% of the poverty level. The remaining 2%(6 women) had incomes that were higher than six times the pov-erty threshold. These six women, none of whom reported recentphysical abuse, were removed from further analysis.

PHYSICAL HEALTH SYMPTOMS

Overall, women endorsed an average of 16 different healthsymptoms (M = 16, SD = 6.35) at least once a month or less (M =1.27, SD = .78). One in four women said they were bothered by atleast 21 of the symptoms on the checklist. The percentages ofabused and nonabused women who reported each physical healthsymptom are presented in Table 3. The most common symptomsreported by both groups were indicative of pain and fatigue, suchas feeling low in energy, sleep problems, headaches, muscle ten-sion or soreness, back pain, and fatigue. Blurred vision and handstrembling were the symptoms cited least. Abused women weresignificantly more likely to report all but six of the symptoms (i.e.,sleep problems, headaches, muscle tension or soreness, back pain,and pelvic pain). Generally, the rate at which women experiencedeach symptom coincided with previous estimates from a commu-nity sample of low-income women (Eby, 1996) and a sample ofwomen who had recently left a battered women’s shelter (Suther-land et al., 1998).

Table 4 shows the women’s mean physical health symptomscores relative to their income levels and whether they had ex-


perienced intimate partner violence within the past 6 months. Ateach level of income, women who had been abused reported sig-nificantly higher levels of physical health symptoms.

CORRELATIONS BETWEEN INDEPENDENT VARIABLESAND WOMEN’S PHYSICAL HEALTH

Pearson correlation coefficients were derived to examine therelationship between the independent variables (income and phys-ical abuse) and women’s physical health symptoms. A low tomoderate correlation was found between income and women’sphysical health symptoms (r = –.27, p < .05); women with lowerincomes reported higher rates of physical health symptoms thandid women with higher income levels. A moderate correlationwas found between women’s experiences of physical abuse andphysical health symptoms (r = .43, p < .05); women who experi-enced more severe physical abuse reported higher levels of physi-cal health symptoms than did women who experienced lesssevere or no physical abuse. The correlation between income andwomen’s experiences of intimate partner violence was significantand moderate (r = –.31, p < .05); women with lower income levelsreported significantly more severe rates of physical abuse thandid women with higher income levels.


TABLE 2Types of Violence Reported by Women

Who Experienced Physical Abuse in Past 6 Months (N = 205)

Type of Abuse N %

In the last 6 months, how often did he/she:Push, shove, or grab you 198 97Throw something at you 127 62Slap you with an open hand 117 57Kick you or hit you with a fist 115 56Hit you or try to hit you with an object 114 56Choke you 111 54Beat you up 104 51Break your glasses or tear your clothing 86 42Tie you up or physically restrain you in some way 83 41Threaten you with a gun or knife 48 23Use a gun or knife against you 25 12Burn you 14 7

We also examined the possible influences of age and race differ-ences. Younger women were found to have lower income levels(r = .25, p < .05) and higher physical abuse scores (r = –.26, p < .05).White women (n = 279) had higher income levels (r = .19, p < .05)and lower physical abuse scores (r = –.19, p < .05) than non-Whitewomen (n = 118). However, neither age nor race (i.e., White vs.non-White) was significantly associated with women’s physicalhealth symptom scores. Controlling for age and race in separatepartial correlation analyses did not significantly affect the direc-tion or magnitude of relationships between physical health symp-toms and income or physical abuse.


TABLE 3Percentage of Nonabused and Abused Women

Who Reported Physical Health Symptoms (N = 397)

% Nonabused % AbusedPhysical Health Symptom (n = 192) (n = 205) ra

Feeling low in energy 91 95 .09*Sleep problems 85 90Headaches 85 83Muscle tension or soreness 83 82Back pain 78 83Constant fatigue 71 78 .09*Nightmares 63 71 .09*Acid stomach or indigestion 60 71 .11*Diarrhea 60 59Heart pounding or racing 52 70 .19*Muscle cramps 52 64 .12*Feeling weak all over 47 67 .20*Stomach pain 47 65 .18*Dizziness 47 61 .13*Severe aches and pains 42 62 .20*Migraine headaches 42 62 .20*Pelvic pain 42 46Nausea and/or vomiting 40 57 .17*Numbness or tingling in parts of body 40 58 .19*Hot or cold spells 39 54 .16*Ringing in your ears 35 45 .10*Poor appetite 34 66 .32*Shortness of breath 33 52 .20*Pains in your heart or chest 33 50 .17*Faintness 31 45 .15*Blurred vision 30 41 .12*Hands trembling 22 44 .23*

a. *Pearson correlation coefficients, one-tailed significance test, p < .05.

The concomitant correlation between income and physicalabuse confirmed that the relationship among income, physicalabuse, and physical health was complex. Further analyses wereconducted to assess the independent and interaction effects ofincome and physical abuse on women’s physical health.

EFFECT OF INTIMATE PARTNER VIOLENCE ON WOMEN’SPHYSICAL HEALTH

Hierarchical multiple regression was employed to determine ifknowledge of physical abuse improved prediction of women’srates of physical health symptoms after accounting for differencesin income levels. Rate of physical health symptoms was regressedonto two independent variables (income and physical abuse) andone product term (Income × Physical Abuse) in three sequentialsteps. The first block of the regression equation contained theincome variable, and the second block of the equation containedthe physical abuse variable. The product terms comprised thethird block of the equation. Prior to computing the product term,the scores of the main effect variables (income and physical abuse)were centered; the mean score of each measure was subtractedfrom individual scale scores. This step was taken to reduce multi-collinearity between the independent variables and their productterms and to improve the interpretability of the coefficients.


TABLE 4Physical Health Symptom Scores by Income Level and Physical Abuse (N = 397)

Income Level and Physical Health Physical HealthAbuse Category Symptom Score M Symptom Score SD Significance Test

Low income t(df = 167) = 4.34,(≤ 125% poverty) p < .001

Nonabused (n = 53) 1.09 .658Abused (n = 116) 1.65 .826

Low-middle income t(df = 74) = 2.31,(126% to 200% poverty) p < .02


Middle income t(df = 144) = 4.89,(201% to 600% poverty) p < .001


The results of the hierarchical multiple regression are dis-played in Table 5. The relevant statistical test is the significance ofthe change in the squared multiple R (R2 change), which indicatesthat the new variable adds significant predictive power to themodel beyond what is already explained by other variables alreadyin the equation. The combined influence of income, physicalabuse, and the interaction between income and abuse accountedfor 21% (R2 = .21) of the variance in women’s physical healthsymptomatology. Both independent variables accounted for a sig-nificant portion of the variance in physical health symptoms;income contributed an R2change = .07, p < .05 (F = 29.82, p < .001),and physical abuse yielded an R2 change = .20, p < .05 (F = 24.32,p < .001). In addition, the Income × Physical Abuse interactionterm significantly contributed to the variance in physical healthsymptoms, R2 change = .01, p < .05 (F = 16.87, p < .001).

The regression plot of the interaction effect is presented in Fig-ure 1. To plot the interaction effect, simple slopes were calculatedbased on recommendations by Aiken and West (1991). First, threelevels (high, medium, and low) of the centered income score wereproduced by setting the cutoff points as one standard deviationabove the mean (high; +133.41), the mean (medium; 0), and onestandard deviation below the mean (low; –133.41). Three separateequations were generated to represent the regression of the depend-ent variable (i.e., physical health symptoms) on physical abuse atthe three levels of income. Finally, the restructured regressionequations were used to compute the predicted dependent vari-able scores for each level of income. The three separate regressionlines were then generated and plotted. The pattern of regressionlines indicated that the relationship between abuse and healthsymptoms was stronger for women with lower income levels.However, it is important to note that the interaction effect wassmall (R2 change = .01, p < .05). This means that although womenwith lower incomes appeared at greater risk for physical healthproblems when assaulted by an intimate partner, survivors withhigher incomes were also highly vulnerable to health problems.

DISCUSSION

The current study generated four key findings. First, acrossmost of the physical health symptoms, rates of endorsement were


significantly higher among the abused women than nonabusedwomen. Second, as anticipated, at all levels of income, womenwho had experienced abuse in the past 6 months had significantlyhigher physical health symptom scores than did women who hadnot been abused. Third, women’s experiences of abuse signifi-cantly contributed to their physical health symptoms beyond thatwhich could be explained by their income levels. The results of thehierarchical multiple regression analysis indicated that both incomelevel and experiences of abuse accounted for variance in women’sphysical health symptoms, yet women’s experiences of physicalabuse explained a significant portion of variance in physicalhealth symptoms even after controlling for income level.

Finally, the results of the hierarchical multiple regression revealeda significant interaction between income and physical abuse.Although the pattern of regression lines indicated a similar trendin the relationship between physical abuse and physical healthsymptoms across income levels, there was a slight distinction


TABLE 5Hierarchical Multiple Regression of Poverty and

Physical Abuse on Rates of Physical Health Symptoms

Physical Health Symptoms

Variables Ba t Significance of t

Block 1: IncomeF = 29.82, p < .0001R2 = .07Income (centered) –.002 –5.46 < .05

Block 2: Physical abuseF = 49.60, p < .0001R2 = .20R2 change = .132, p < .05Physical abuse (centered) .789 8.03 < .05

Block 3: Physical Abuse × Income InteractionF = 34.68, p < .0001R2 = .21R2 change = .008, p < .05Income × Physical abuse .002 2.02 < .05

NOTE: Correlation estimates were performed to examine effects of ethnicity and age onphysical health symptoms. No significant differences were found between samples ofWhite and minority groups, and neither was there a significant relationship between ageand physical health symptoms.a. Unstandardized regression coefficients are reported for the step in which they were en-tered into the model.

among them. Abuse was significantly more strongly associatedwith health symptoms at lower levels of income.

This interaction pattern is open to a number of interpretations.It is certainly plausible that for women with low incomes, physi-cal abuse exacerbated the stress associated with living in povertyand increased the likelihood of poor physical health outcomes.The mediating effect of stress on the relationship between intimatepartner abuse and physical health has been examined elsewhere(see Eby, 1996; Sutherland et al., in press), but the cross-sectionalnature of the research design limits conclusive assumptions aboutthe causal direction of the relationship.

It is also possible that low-income women’s restricted access tohealth care limited their ability to seek adequate medical attentionfor abuse-related injuries. For example, if abuse-related injuriesrequired expensive medication or long-term rehabilitation, low-income women may not have been able to afford the high cost of


Figure 1 Plotted Regression Lines Illustrating the Interaction Between Income Leveland Abuse

prescription drugs or the reduced pay for lost time at work. If leftuntreated, those injuries may have led to chronic health problems.Each of these factors may have contributed to the interactioneffect between income level and physical abuse. It is important tonote that although women with lower income levels appearedmore vulnerable to health problems associated with intimate part-ner violence, physical abuse had a resounding effect on women’sphysical health regardless of their income level. Further researchusing a longitudinal design is needed to fully explain the complexprocess by which intimate partner violence jeopardizes women’shealth.

The results of the current study corroborate prior research find-ings. Several researchers have demonstrated that intimate partnerviolence places low-income women at risk for numerous physicalhealth problems (Campbell & Lewandowski, 1997; Campbell &Soeken, 1999; Coker et al., 2000; Eby, 1996; Sutherland et al., 1998;Tollestrup et al., 1999). Furthermore, they have shown that low-income women’s experiences of health problems are directlyrelated to the frequency and severity of physical abuse. Otherresearchers have established a similar link between poverty andphysical health problems (Dunn & Hayes, 2000; Humphries &Van Doorslaer, 2000; Stronks et al., 1998).

Although the study did improve on previous research, it hadseveral methodological limitations. First, the range of incomeswas somewhat restricted. The overrepresentation of low-incomewomen among those who had been abused limits the general-izability of our findings to women with higher incomes. Ourinability to recruit more abused women with higher incomes wasprobably due to the following two factors: the social stigmaattached to discussing domestic violence for women with highincomes and the recruitment strategy used. Although the news-paper advertisements were a successful tool for encouragingwomen to participate in the study, a preferable strategy wouldhave been random selection of study participants. Only thosewomen who saw the advertisements and were willing to discusstheir health issues and/or experiences of abuse participated. It istherefore difficult to ascertain the degree to which their experi-ences of abuse or health problems differ from women who did notrespond to the advertisements.


Similarly, the restricted range of incomes among minority womenlimited our ability to sufficiently examine the effects of intimatepartner violence within the context of race and ethnic differences.Replication of the study findings using a more ethnically diversesample with a broader range of income levels would helpstrengthen the generalizability of the results and clarify some ofthese issues.

It could also be viewed as somewhat arbitrary to use the past6 months as the cutoff point for determining whether womenshould be classified as being abused. It is certainly possible that anumber of women classified as nonabused had experienced inti-mate partner violence in the more distant past and that this expe-rience still affected their health symptoms.

Basing the assessment of women’s physical health on self-report measures of physical health symptoms presented anothermethodological limitation. Self-report measures of physical healthsymptoms are subjective and may not discriminate well betweenphysical and psychological health problems. In addition, althoughsymptoms are an important dimension of a woman’s physicalhealth, they are not a comprehensive indicator of women’s overallhealth status. Further research using multidimensional instru-ments to assess chronic health problems, physical functioning,and general health perceptions as well as symptoms is needed tocapture the full effect of violence on women’s health.

Finally, the cross-sectional nature of the research design restrictsour ability to assert causal direction among the relationships. Bothabuse and income accounted for variation in women’s reportedphysical health symptoms, yet the extent to which each factorindependently caused subsequent health problems remains unclear.Perhaps women’s income levels were influenced by the abusethey experienced and/or their health problems. Clarification ofthese issues would require a longitudinal design.

Despite these methodological limitations, this study expandsprevious research efforts in several ways. First, half the samplehad not experienced physical abuse within the 6 months prior tothe study, which allowed for a more accurate estimation of theeffects of abuse on women’s health. Second, women were recruitedfrom the community rather than within a shelter or clinical set-ting, which increased the likelihood of assessing women’s experi-ences of abuse as well as their health status when they were not in


a state of crisis. Third, by including women from a range ofincome levels, it was possible to examine the extent to which pov-erty contributed to health problems. The current study clearlydemonstrated that both income and abuse were significant pre-dictors of women’s physical health. More importantly, it improvedon previous efforts to differentiate the effects of abuse from thosedue to poverty.

Future efforts should bear in mind that although intimate vio-lence is harmful to women at varying income levels, it appears tobe especially detrimental to women with low incomes. Proactiveinitiatives from social and health care services that typicallyaddress the needs of low-income women could help alleviate thehealth consequences of abuse. Guidelines for good practice inidentifying and responding to the individual needs of domesticviolence survivors as well as meeting the training needs of serviceproviders are essential. Furthermore, policies and practices regard-ing welfare reform should consider the effect recent changes havehad on the lives of women trying to flee domestic violence, espe-cially with respect to benefit allowance, subsidized housing allo-cations, and work requirements (Browne et al., 1999).

The health care community is an important component of anycomprehensive community approach to ending domestic vio-lence. Battered women often seek medical attention for abuse-related injuries as well as health problems that appear unrelatedto any specific injury or predisposing health condition. Earlyidentification of abuse is essential to eliminating violence andsubsequent health problems from women’s lives. Routine screen-ing procedures may be the most effective way to identify batteredwomen. Several researchers and practitioners have outlined inter-ventions health care providers can implement to assist survivorsof intimate partner violence (Campbell & Lewandowski, 1997;Naumann et al., 1999). These interventions include providingemotional support and mental health counseling, safety plan-ning, patient education, legal advocacy, referral to communityservices, and consistent documentation of abuse history.

The current study underscores the need for more proactive ini-tiatives to end violence in women’s lives. Previous longitudinalresearch has demonstrated that women’s reports of physical healthsymptoms decrease as the violence in their lives decreases(Sutherland et al., 1998). Developing comprehensive community


response initiatives that effectively address not only women’sneed for health care and other community resources but alsowomen’s need for protection from an intimate partner’s violenceis an essential step toward reducing women’s risk for long-termhealth problems.

REFERENCES

Abbott, J., Johnson, R., Koziol[OR KOZOIL?]-McLain, J., & Lowenstein, S. R. (1995).Domestic violence against women: Incidence and prevalence in an emergency depart-ment population. Journal of the American Medical Association, 273, 1763-1767.

Aiken, L. S., & West, S. G. (1991). Interactions between continuous predictors in multipleregression. In [BOOK EDITORS?] Multiple regression: Testing and interpreting interac-tions (pp. 9-27). Newbury Park, CA: Sage.

Alpert, E. J. (1995). Violence in intimate relationships and the practicing internist: New“disease” or new agenda? Annals of Internal Medicine, 123, 774-781.

Alter, D. A., Naylor, C. D., Austin, P., & Tu, J. V. (1999). Effects of socioeconomic status onaccess to invasive cardiac procedures and on mortality after acute myocardial infarc-tion. New England Journal of Medicine, 341, 1359-1367.

Attala, J. M. (1994). Risk identification of abused women participating in a women, infants,and children program. Health Care for Women International, 15, 587-597.

Bachman, R., & Saltzman, L. E. (1995). Violence against women: Estimates from the redesignedsurvey (NCJ-154348). Washington, DC: Department of Justice, Bureau of Justice Statistics.

Belle, [INITIAL?] (1990). Poverty and women’s mental health. American Psychologist, 45,385-389.

Bindman, A. B., Grumbach, K., Osmond, D., Komaromy, M., Vranizan, K., Lurie, N., Bill-ings, J., & Stewart, A. (1995). Preventable hospitalizations and access to health care.Journal of the American Medical Association, 274, 305-311.

Browne, A., Salomon, A., & Bassuk, S. S. (1999). The impact of recent partner violence onpoor women’s capacity to maintain work. Violence Against Women, 5, 393-426.

Campbell, J. C. (1989). Atest of two explanatory models of women’s responses to battering.Nursing Research, 38, 18-24.

Campbell, J. C., & Lewandowski, L. A. (1997). Mental and physical health effects of inti-mate partner violence on women and children. Psychiatric Clinics of North America, 20,353-374.

Campbell, J. C., & Soeken, K. L. (1999). Women’s responses to battering: Atest of the model.Research in Nursing and Health, 22, 49-58.

Cohen, S., & Hoberman, H. M. (1983). Positive events and social supports as buffers of lifechange stress. Journal of Applied Social Psychology, 13, 99-125.

Coker, A. L., Smith, P. H., Bethea, L., King, M. R., & McKeown, R. E. (2000). Physical healthconsequences of physical and psychological intimate partner violence. Archives of Fam-ily Medicine, 9, 451-457.

Collins, K. S., Schoen, C., Joseph, S., Duchon, L., Simantov, E., & Yellowitz, M. (1999).Health concerns across a woman’s lifespan: The Commonwealth Fund 1998 survey ofwomen’s Health. The Commonwealth Fund: Programs [Online]. Available: http://www.cmwf.org/programs/women

Council on Scientific Affairs, American Medical Association. (1992). Violence againstwomen: Relevance for medical practitioners. Journal of the American Medical Association,263, 3184-3189.


Drossman, D. A. (1994). Physical and sexual abuse and gastrointestinal illness: What is thelink? American Journal of Medicine, 97, 105-107.

Drossman, D. A., Talley, N. J., Leserman, J., Olden, W., & Barreiro, [OR BARRIO?]M. A.(1995). Sexual and physical abuse and gastrointestinal illness: Review and recommen-dations. Annals of Internal Medicine, 123, 782-794.

Dunn, J. R., & Hayes, M. V. (2000). Social inequality, population health, and housing: Astudy of two Vancouver neighborhoods. Social Science and Medicine, 51, 563-587.

Eby, K. K. (1996). Experiences of abuse and stress: A path model of their joint effects on women’spsychological and physical health. Unpublished doctoral dissertation, Michigan StateUniversity.

Eby, K. K., Campbell, J. C., Sullivan, C. M., & Davidson, W. S., II. (1995). Health effects ofexperiences of sexual violence for women with abusive partners. Health Care for WomenInternational, 16, 563-567.

Feinstein, J. S. (1993). The relationship between socioeconomic status and health: A reviewof the literature. Milbank Quarterly, 71, 279-322.

Fuller, T. D., Edwards, J. N., Sermsri, S., & Vorakitphokatorn, S. (1993). Housing, stress, andphysical well-being: Evidence from Thailand. Social Science and Medicine, 36, 1417-1428.

Hamberger, L. K. (1994). The battered woman: Identification and intervention. The FemalePatient, 19, 29-33.

Havenaar, J. M., & Van Den Brink, W. (1997). Psychological factors affecting health aftertoxicological disasters. Clinical Psychology Review, 17, 359-374.

Health Status Advisory Group. (1993). An assessment of the health status of the capital commu-nity: A brief report. Lansing, MI: Author.

Humphries, K. H., & Van Doorslaer, E. (2000). Income-related inequality in Canada. SocialScience and Medicine, 50, 663-671.

Kaplan, G. A., & Keil, J. E. (1993). Socioeconomic factors and cardiovascular disease: Areview of the literature. Circulation, 88, 1971-1978.

Kington, R. S., & Smith, J. P. (1997). Socioeconomic status and racial and ethnic differencesin functional status associated with chronic diseases. American Journal of Public Health,87, 805-810.

Luepker, R. V., Rosamond, W. D., Murphy, R., Sprafka, J. M., Folsom, A. R., McGovern, P. G., &Blackburn, H. (1993). Socioeconomic status and coronary heart disease risk factortrends: The Minnesota Heart Survey. Circulation, 88, 2172-2179.

Lynch, J. W., Kaplan, G. A., & Shema, S. J. (1997). Cumulative impact of sustained economichardship on physical, cognitive, psychological, and social functioning. New EnglandJournal of Medicine, 337, 1889-1895.

McLeod, J. D., & Kessler, R. C. (1990). Socioeconomic status differences in vulnerability toundesirable life events. Journal of Health and Social Behavior, 31, 162-172.

Naumann, P., Langford, D., Torres, S., Campbell, J., & Glass, N. (1999). Women battering inprimary care practice. Family Practice, 16, 343-352.

Plichta, S. B. (1996). Violence and abuse: Implications for women’s health. In M. M. Falik &K. S. Collins (Eds.), Women’s health: The Commonwealth Fund Survey (pp. 237-272). Balti-more: Johns Hopkins University Press.

Robinson, N., Caraher, M., & Lang, T. (2000). Access to shops: The views of low-incomeshoppers. Health Education Journal, 59, 121-136.

Saunders, D. G., Hamberger, L. K., & Hovey, M. (1993). Indicators of woman abuse basedon a chart review at a family practice center. Archives of Family Medicine, 2, 537-543.

Scarinci, I. C., McDonald-Haile, J., Bradley, L. A., & Richter, J. E. (1994). Altered pain per-ception and psychosocial features among women with gastrointestinal disorders andhistory of abuse: A preliminary model. American Journal of Medicine, 97, 108-118.

Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT)Scales. Journal of Marriage and the Family, 38, 75-88.


Stronks, K., Van De Mheen, H. D., & Mackenbach, J. P. (1998). Ahigher prevalence of healthproblems in low-income groups; does it reflect relative deprivation? Journal of EpidemiolCommunity Health, 52, 548-557.

Stronks, K., Van De Mheen, H., Van Den Bos, J., & Mackenbach, J. P. (1997). The interrela-tionship between income, health and employment status. International Journal of Epide-miology, 26, 592-600.

Sullivan, C. M., & Bybee, D. (1999). Reducing violence using community based advocacyfor women with abusive partners. Journal of Consulting and Clinical Psychology, 67, 43-53.

Sullivan, C. M., Tan, C., Basta, J., Rumptz, M., & Davidson, W. S., II. (1992). An advocacyintervention program for women with abusive partners: Initial evaluation. AmericanJournal of Community Psychology, 20, 309-332.

Sutherland, C., Bybee, D., & Sullivan, C. (1998). The long-term effects of battering onwomen’s health. Women’s Health: Research on Gender, Behavior, and Policy, 4, 41-70.

Sutherland, C., Bybee, D., & Sullivan, C. (in press). Beyond bruises and broken bones: Thejoint effects of stress and injuries on battered women’s health. American Journal of Com-munity Psychology.

Talley, N. J., Fett, S. L., & Zinsmeister, A. R. (1995). Self-reported abuse and gastrointestinaldisease in outpatients: Association with irritable bowel-type symptoms. American Jour-nal of Gastroenterology, 90, 366-371.

Talley, N. J., Fett, S. L., Zinsmeister, A. R., & Melton, L. J., III. (1994). Gastrointestinal tractsymptoms and self-reported abuse: A population-based study. Gastroenterology, 107,1040-1049.

Tollestrup, K., Sklar, D., Frost, F. J., Olson, L., Weybright, J., Sandvig, J., & Larson, M. (1999).Health indicators and intimate partner violence among women who are members of amanaged care organization. Preventive Medicine, 29, 431-440.

U.S. Department of Commerce, Bureau of the Census. (1990). General population characteris-tics [Online]. Available: http://www.census.gov

U.S. Department of Commerce, Bureau of the Census. (1996). Preliminary estimate of povertythresholds for 1996 [Online]. Available: http://www.census.gov/hhes/poverty/threshld.html

Weissman, J. S., Fielding, S. L., Stern, R. S., & Epstein, A. M. (1991). Delayed access to healthcare: Risk factors, reasons, and consequences. Annals of Internal Medicine, 114, 325-331.

Cheryl Sutherland is an outreach development and training worker at East FireWomen’s Aid in Scotland and also is conducting research with the Greater Glas-gow Health Board. She received her Ph.D. in ecological/community psychology(1999) from Michigan State University. Her main research activities includeinvestigating the effect of domestic violence on women’s health and evaluatingstrategies to improve domestic violence survivors’ access to community services.Currently, she is assessing domestic violence survivors’ access to mental healthservices in Scotland.

Cris Sullivan is associate professor of ecological/community psychology at Michi-gan State University and director of evaluation for the Michigan CoalitionAgainst Domestic and Sexual Violence. Dr. Sullivan has been an advocate andresearcher in the movement to end violence against women since 1982. Herresearch has primarily involved examining the long-term effects of community-based interventions for battered women and their children and evaluating domes-tic violence and sexual assault programs.


Deborah Bybee is research associate professor of ecological/community psychologyat Michigan State University. She has a primary interest in quantitative methodsand how they can be used to understand complex, real-world phenomena, espe-cially those that change over time. Substantively, she has applied her methodologi-cal interest to a variety of areas, including advocacy for women with abusive part-ners, intervention with children who have witnessed domestic violence, housingassistance for individuals who are homeless and mentally ill, supported educationfor individuals with a mental illness, and mothering by women coping with a seri-ous mental illness.


effects of intimate partner violence versus poverty on …violence against women / october...

Documents