Physical violence against women remains a major public health challenge and human rights issue Devries et al., 2013. Globally, more than 35% of women experienced either intimate partner violence or non-partner physical violence García-Moreno et al., 2013. Physical violence against women is an important risk factor for women’s poor health resulting in a wide range of short and long-term health consequences, including the incidence of unintended pregnancies Jeyaseelan et al., 2007; Pallitto et al., 2013, increased risk of reproductive and sexually transmitted infections García-Moreno et al., 2013; Durevall and Lindskog, 2015 and mental health issues for the victim Skogstad et al., 2014; WHO, 2014. Physical violence also has a significant impact on the family of abused women Widom et al., 2014 and considerable negative externalities such as psychological stress even for those witnessing the violence Geffner, 2014; Schiff et al., 2014.
Although physical violence against women manifests as a worldwide public health concern, up to 70% of women in low- and middle-income countries WHO, 2005 and more than 45% in Africa compared to 32.7% in high-income countries WHO, 2013 experience physical violence. In the Gambia, more than 40% of women aged 15–49 years experienced physical violence in the past 12 months in 2013, of which 24% sustained injuries GBoS/ICF, 2014. Previous studies have indicated that injuries from physical violence against women are perpetrated by known persons including spouses and friends in the home, during the night Hofner et al., 2009; Tingne et al., 2014, injured by fist punching, leg kicking or struck by an object Wong et al., 2014, injury to the head, neck, face and the upper limbs Brennan et al., 2006.
Despite several African countries have documented violence against women Andersson et al., 2007; Abramsky et al., 2011, they seldom use injured victims from physical violence to investigate risk factors for physical violence in women. Thus, identifying risk factors associated with injury from physical violence in women is essential to developing interventions aimed at preventing violence. Commonly identified risk factors for physical violence against women include younger age, low education Trinh et al., 2016; Ahmadi et al., 2017, unemployment status Jeyaseelan et al., 2007, financial dependence Fageeh, 2014, low economic status Doku and Asante, 2015, alcohol consumption Devries et al., 2014, being married Kouyoumdjian et al., 2013 divorced or separated Mohsen et al., 2017, in a polygamous marriage Ali et al., 2014, being previously victimized McCoy et al., 2013; Sapkota et al., 2016 and been abused during childhood or brought up by a single parent Chan, 2014.
While the above-mentioned studies investigated injury patterns and risk factors for physical violence against women; it may be difficult to infer those results into the Gambian context due to differences in socioeconomic, environmental and cultural factors. Accordingly, we conducted a case-control study to determine injury patterns and identify risk factors associated with physical violence among women in the Gambia.
2.1. Study settings and participants
From October 2016 through May 2017, we conducted a case-control study with study participants recruited from emergency rooms (ERs) and outpatient departments (OPDs) of government-managed healthcare facilities located in six districts. These districts are located within the two local urbanized administrative regions (West Coast Region and Kanifing Municipality), which accounts for 60% of the country’s population GBoS, 2013. A simple random sample of eight health facilities were selected to represent the different tiers of the healthcare system in The Gambia, which included one tertiary health facility (Serrekunda General Hospital), one district hospital (Brikama District Hospital), one major health centre (Faji Kunda), and five minor health centers (Gunjur, Bakau, Banjul’nding and Serrekunda). These health facilities treat a broad range of conditions including patients with all injury types. Private healthcare facilities were excluded from the study because they do not offer 24-h ER/OPD services to all patients.
Cases were female patients aged ?15 years who sought medical treatment for injuries from physical violence during the study period. An injury from physical violence was defined as any injury or physical pain that had been intentionally caused by another person Hirschinger et al., 2003. Controls comprised of female patients aged ?15 years who sought treatment for injuries from traffic crashes, falls, sports, and other non-violence causes and were matched to each case by health facility, date of injury from physical violence and age and. Patients were excluded from the study if they were unable to verbally communicate with data collectors, unable recall details of the violent incident due to injury, could not provide a written consent or were minors. In total, 194 case-control pairs met the inclusion criteria and were included in the analysis.
All participants provided written informed consent before participating in the study. The study protocol was reviewed and approved by the University of The Gambia Research and Publication Committee and The Gambia Government/Medical Research Council Joint Ethics Committee on human subjects’ research. The Ministry of Health Social Welfare also granted approval to conduct the study at each participating health facility.
ERs/OPDs staff trained on the administration of the questionnaire, collected information on sociodemographics (e.g. age, height, weight, ethnicity, marital status, educational level, employment status, household income level and childhood upbringing), injury characteristics (e.g. date and time of injury, place of injury, mechanism of injury, nature of the injury, body part injured, severity of the injury and physical violence perpetrator), lifestyle behaviors in the past week (i.e. cigarette smoking, alcohol consumption, and illicit drug use), experience of verbal abuse, physical threats or physical abuse in the past 12 months, social supports, and risk-taking behaviors.
To ensure data quality, twice weekly visits by the researchers (PB and ES) were made to study sites to collect completed questionnaires, check for accuracy, and to ensure adherence to the study protocol. Questionnaires were doubled checked, double entered and cleaned in Microsoft Access.
Injury severity was assessed using the Kampala Trauma Score II (KTSII) which was developed in 1996 by the Injury Control Centre-Uganda Owor and Kobusingye, 2001. The KTSII scores five parameters during the patient’s assessment: age (in years), respiratory rate, systolic blood pressure, neurologic status and score for serious injuries on admission. The scores are further categorized into three levels: mild (9~10), moderate (7~8) and severe (?6) injuries. The KTSII has been validated and found to be a good measure of injury severity in most sub-Saharan African countries Weeks et al., 2014; Haac et al., 2015; Seid et al., 2015.
Social support was assessed using the 12-item Multidimensional Scale of Perceived Social Support (MSPSS) which measures the level of support that an individual perceives in three domains (family, friends, and significant others) Zimet et al., 1988. The MSPSS has been used in the USA and Africa populations and reported to have high reliability (alpha coefficients of 0.91~0.94) Canty-Mitchell and Zimet, 2000; Stewart et al., 2014.
The revised Domain-Specific Risk-Taking Scale (DOSPERT) was used to assess risk-taking behaviors which evaluate the likelihoods that respondents might engage in behaviors from six risk domains (i.e., Ethical, Gambling, Investing, Health/Safety, Recreational, and Social) Blais and Weber, 2006. The DOSPERT has been validated and used in a wide range of settings, populations, and cultures, including South Africa Szrek et al., 2012 which has similar demographic characteristics to The Gambia. Three domains of Health/Safety, Recreational, and Social were used in this study. A high score indicates greater risk-taking level for each of the three domains.
2.4. Statistical analysis
Injury patterns of case patients were presented as numbers with percentages. Distribution of sociodemographics, lifestyle behaviors in the past week, the experience of verbal abuse, physical threats and physical abuse in the past 12 months, social supports, and risk-taking behaviors were compared between cases and controls using Pearson’s Chi-squared test for categorical variables and Student’s t-test for the continuous variables. To avoid large type II errors in variable selection and biased inferences, variables with a p-value of ?0.25 in the bivariate logistic analysis were included in the multivariable analysis Maldonado and Greenland, 1993. A forward stepwise conditional logistic regression was used to identify independent relationships of potential risk factors for injuries from physical violence in which adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) were computed. Matching variables of health facility, date of physical violence and age were forced into the multivariable model, and variables with p-values of <0.05 were considered statistically significant. The Hosmer-Lemeshow goodness of fit and likelihood ratio tests were used to evaluate the appropriateness of the model Hosmer and Lemesbow, 1980. All analyses were performed using the Statistical Analysis Software (SAS) package (vers. 9.4 for Windows; SAS Institute, Inc., Cary, NC, USA). 3. Results Out of a total of 217 female patients diagnosed with injuries from physical violence at eight ER/OPD units, 194 patients yielding 89.4% response rate were matched to 194 control patients on health facility, date of injury from physical violence and age. Distributions of injury patterns of 194 violently injured women are shown in Table 1. Injuries from physical violence occurred most frequently in the daytime (66.4%) and home (61.3%). The most common mechanism of injuries from physical violence was being struck by an object (33.2%), fist punching/leg kicking (28.0%). Injuries to the head/neck/face (44.5%) and upper extremity (22.0%) were the most common body parts. More than half (55.2%) of case patients had mild injuries (KTSII score of 9~10). The most common perpetrators were friends (55.8%) and strangers (34.4%). Comparisons of sociodemographics between the 194 cases and 194 controls are shown in Table 2. Compared to control patients, case patients were significantly more likely to have been a Fula, Christians, had lived in a compound with ?2 households, had more than five female siblings and had been brought up by a grandparent during childhood. No significant differences in age, marital status, age at first marriage, duration of the first marriage, educational level, employment status, household income and the number of male siblings between the case and control patients were detected. Comparisons of behavioral and social characteristics between cases and controls are shown in Table 3. Compared to control patients, case patients were significantly more likely to have had the experience of being physically threatened, verbally or physically abused in the past 12 months. As for lifestyle behaviors, case patients were significantly more likely than the control patients to have consumed alcohol in the past week. Cases were similar to controls in terms of cigarette smoking, illicit drug use in the past week and social supports or risk-taking behaviors. Results of conditional logistic regression analyses of risk factors for injuries from physical violence among Gambian women are shown in Table 4. After adjusting for confounders, compared to Mandinka women group, Fula women (OR, 2.45; 95% CI, 1.06~5.66) were more likely to have an injury from physical violence. Participants who had lived in ?2 households (OR, 3.07; 95% CI, 1.22~7.72) and had more female siblings (OR, 3.10; 95% CI, 1.38~6.97) were more likely to have an injury from physical violence than their counterparts. Compared to the participants who had been brought up by both parents, those who had been brought up by grandparents were (OR, 3.34; 95% IC, 1.06~10.51) fold more likely to have an injury from physical violence. Participants who had been verbally (OR, 3.04; 95% CI, 1.56~5.96) or physically abused (OR, 3.36; 95% CI, 1.34~8.39) were more likely to have an injury from physical violence than their counterparts. 4. Discussion In this case-control study, we described injury patterns and identified risk factors for injuries from physical violence among women in the Gambia who sought care at urban ERs/OPDs. Our results revealed that being a Fula woman, living in a compound with multiple households, having more than five female siblings, being brought up in childhood by grandparents and being verbally or physically abused in the past 12 months were independently associated with injury from physical violence among women in the Gambia. As with previous studies in the US Ranney et al., 2009; Jacovides et al., 2013, a large proportion of injuries from physical violence were assaulted by friends, in the daytime, in the home and were caused by hitting with objects or punched with fist / kicked with legs. The head, neck, and face were the most common injured body parts. These findings highlight the extent to which injury from physical violence may be associated with frequent social interaction with friends, colleagues and family members during the daytime. Our finding of the Fula ethnic group having an increased risk for injury from physical violence could be associated with early marriage. In the Gambia, Fula women marry at an early age GBoS/UNICEF, 2010 which has been found to be associated with violence victimization Hong Le et al., 2014; Rahman et al., 2014; Peterman et al., 2015. More than 60% of the Fula women in the Gambia are married before age 18 years GBoS/UNICEF, 2010 which is more than 50% as found in our study. The association between living in a compound with multiple households and injuries from physical may reflect the structural and sociocultural characteristics of the Gambian society where families live in communes, and often times consisting of two or more generations GBoS/ICF, 2014. This situation could be challenging, particularly where women from different backgrounds are married to male siblings and live together in one extended family compound. Consequently, misunderstandings and rivalry due to cooking turns, shared facilities, gossiping, teasing, as well as other issues relating to quarrels between children may have resulted in physical confrontation among women causing injury. In addition, physical violence against women may also be precipitated by male partners through jealousy regarding infidelity since it has been reported that physical violence against women is associated with a recent breakup in relationships Hirschinger et al., 2003. As for the association between having more than five female siblings and the risk of injury from physical violence; possible reasons for this is that having numerous children is synonymous with polygamous marriages Ali et al., 2014 where contentious rivalry among co-wives may be extended to their children. This pattern of violence could also be related to outgrown younger siblings trying to physically challenge older siblings. It is also possible that cultural norms for older sibling physically assaulting younger siblings as a means for protection or social control Tucker et al., 2010 may have resulted in physical assault causing injury. Additionally, spending time with a violent-prone female sibling has been found to increase the risk of later victimization Taylor et al., 2015. Contrary to previous findings Hayslip Jr and Kaminski, 2005; Sandberg, 2016, we found women who had been brought up by grandparents to have had an increased risk for injuries from physical violence. Prior studies have indicated that children who are cared for by grandparents are deprived of parental love and can often exhibit violent behavior Hameed-ur-Rehman and Sadruddin, 2012; Laeheem, 2013. One possible explanation of our finding may be related to old age and chronic health problems associated with grandparents Keles et al., 2007 which may make them less likely to have full supervisory control over the grandchildren. As a result, children may be engaged in peer-motivated delinquent behaviors exposing them to the risk of victimization outside of the home Jackson et al., 2013. Another possible explanation may be due to the pampering of grandchildren by their grandparents which might have influenced their behaviors negatively. As found in Thai study children pampered by parents were more likely than their counterparts to be violent Laeheem, 2013. Consistent with studies in the US and Sweden, Kaufman et al., 2016; Pratt-Eriksson et al., 2016, the previous victimization in the past 12 months was a risk factor of injury from physical violence among women in the Gambian. This finding is in line with the violence cycle theory Walker, 2009, which suggests that women tolerate violence perpetrated against them and would not seek help even after repeated episodes of violent events. Several other reasons, including prejudice against female victims, fear of embarrassment and retaliation, family privacy, concern for children and cultural norms Gracia, 2004; Peerzada and De Sousa, 2016; Sapkota et al., 2016 might have also contributed to the recurrence of injury from physical violence in this group of women. Therefore, support from family, community and institutional levels, as well as knowledge of when and where to seek care or support if they had suffered from violence, may be important factors to consider when designing violence interventions for women. This study has several limitations. First, the small sample size may have limited our ability to observe the effects of other important variables in female victimization such as marital status, educational level, employment status, alcohol consumption and illicit drug use. The lack of information on lifestyle and behavioral factors on the perpetrator for adjustment purposes is also a possible limitation. Second, the results of this study might not be generalizable to all women who sought treatment at the ER/OPD due to injuries from physical violence. The study focused only on patients from government-managed health facilities in urban areas and those treated in private health facilities or live in rural areas might have different characteristics from our study population. Third, misclassifying the cause of injury for control patients as non-violent could not be completely eliminated, which might have underestimated the effects of risk factors on injuries from physical violence. Finally, we cannot exclude the possibility of recall and social desirability bias from both the participants and interviewers. Therefore, findings of this analysis should be interpreted with caution. 5. Conclusions Factors such as being a Fula woman, having more than five female siblings, living in a compound with multiple households, being brought up by grandparents, being verbally or physically abused in the past 12 months were independently associated with injuries from physical violence among women in the Gambia. These risk factors could serve as useful targets in physical violence preventive interventions among women in the Gambia, as well as to identify women at risk of revictimization, particularly in the home environment.