Rural Location and Relative Location: Adding Community Context to the Study of Sexual Assault Survivor Time Until Presentation for Medical Care

Abstract

Despite a strong empirical base linking community context and proximity to resources to individual health care access, studies examining predictors of sexual assault survivor time until presentation for medical care have not yet examined these relationships. This study addresses this gap. The data included retrospective records on a sample of 1,630 female survivors who reported their sexual assault to law enforcement and were subsequently seen by a sexual assault nurse examiner (SANE) in one of eight Alaskan communities between the years 1996 and 2006. Logistic regression models were used to determine whether delays in presentation (presentation 12 hr or more after assault) differed for women presenting in unique communities (rural location), and between those whose assault and exam occurred in different communities versus occurring in the same community (relative location). Although rural location did not seem to have a unique impact on time until presentation, differing locations (i.e., relative location) of assaults and exams increased the likelihood of delays in presentation. Non-American Indian/Alaska Native race/ethnicity and knowing one’s assailant(s) also increased the likelihood of delays. These results indicate that in addition to a need for further research, there is a need for more appropriate and reliable sexual assault medical services across communities, and that survivors assaulted by known assailants should be targeted in efforts to reduce time until presentation.

Keywords

sexual assault time until presentation community context rurality

Introduction

Approximately 1.1 out of every 1,000 persons age 12 and older in the United States were raped or sexually assaulted in 2013 (Truman & Langton, 2014). Despite the importance of prompt medical care for survivors who want it, there is significant variability in how long it takes for them to seek and receive that care (Adefolalu, 2014; Dunlap, Brazeau, Stermac, & Addison, 2004; McCall-Hosenfeld, Freund, & Liebschutz, 2009; Millar, Stermac, & Addison, 2002; Nesvold, Friis, & Ormstad, 2008; Resnick et al., 2000). Consequently, having a greater understanding of which factors might affect sexual assault survivor time until presentation can inform efforts to improve the distribution of justice for these survivors.

Although several studies have made important contributions to a shared understanding of the variation in time until presentation among sexual assault survivors (Adefolalu, 2014; McCall-Hosenfeld et al., 2009; Millar et al., 2002; Nesvold et al., 2008), none have specifically examined the role of community context and relative location to services. This oversight is a serious short-coming considering the influence of community rurality on perceived barriers to help-seeking among rape survivors (Logan, Evans, Stevenson, & Jordan, 2005) as well as on differences in help-seeking from law enforcement between rural and urban sexual assault survivors (Rennison, Dragiewicz, & DeKeseredy, 2013; Ruback & Ménard, 2001). The current study therefore seeks to add to the empirical literature on predictors of sexual assault survivor time until presentation for medical care, with a particular focus on the role of the location of the exam and the relative locations of the assaults and exams.

Time Until Presentation and Justice for Survivors

Delayed presentation for medical care among sexual assault survivors is linked to a decrease in police involvement (Dunlap et al., 2004; Nesvold et al., 2008); a decreased likelihood of receiving a variety of medical services (Millar et al., 2002); a decrease in the number and types of services received (e.g., physical exam, tests for STIs; Dunlap et al., 2004); less effective services such as emergency contraception (Piaggio, von Hertzen, Grimes, & Van Look, 1999), HIV post-exposure prophylaxis (New York State Department of Health AIDS Institute, 2013), and tests for the presence of date-rape drugs (LeBeau & Montgomery, 2010); and difficulties in documenting physical evidence (e.g., injuries may heal as time passes and hair and semen left by the assailant may be lost; World Health Organization, 2003; Young, Bracken, Goddard, Matheson, & the New Hampshire Sexual Assault Medical Examination Protocol Project Committee, 1992). Not only do these outcomes have direct consequences for survivors, they could also have implications for help received by the survivor in the future; that is, it is possible that if a sexual assault survivor’s experiences with formal services are not helpful or her needs are not met, she may decide not to seek help in the future because of that experience (Kennedy et al., 2012).

Known Correlates of Time Until Presentation

Because prompt medical care is important for the health of and justice for sexual assault survivors, researchers have sought to understand differences in time until presentation by examining characteristics of the assault event and the survivors in the United States (McCall-Hosenfeld et al., 2009), Canada (Millar et al., 2002), Norway (Nesvold et al., 2008), and South Africa (Adefolalu, 2014).

Characteristics of assault events

The assailant’s use of verbal threats (as opposed to none) is significantly associated with earlier presentation (McCall-Hosenfeld et al., 2009; Millar et al., 2002). The use of severe violence during the assault (e.g., the use of physical coercion, the use of weapons, assaults resulting in injury) is also associated with earlier presentation (McCall-Hosenfeld et al., 2009; Millar et al., 2002). More penetrative assaults and assaults that occurred in a victim’s area (e.g., home, job site, hotel, friend’s home) have been found to be associated with delays in presentation (Nesvold et al., 2008). Assaults involving anal penetration and those in which the survivor was held captive have also been found to be associated with delays in presentation (Adefolalu, 2014). A known assailant(s) was associated with delays in presentation in all the studies reviewed (Adefolalu, 2014; McCall-Hosenfeld et al., 2009; Millar et al., 2002; Nesvold et al., 2008), except one (Logan, Cole, & Capillo, 2007).

Characteristics of survivors

With respect to race/ethnicity, survivors in Canada who identified as First Nations/Aboriginal, Hispanic, or South Asian presented earlier than did White, Black, or Asian women (Millar et al., 2002). A survivor’s fear of the perpetrator and fear that family members would not believe the assault had occurred were also associated with delayed presentation (Adefolalu, 2014). Survivors with a clear memory of the assault presented later than those who had only a suspicion that an assault had occurred or who had only a vague description of the assault (Nesvold et al., 2008). In some studies, being younger was associated with delays (Adefolalu, 2014; Nesvold et al., 2008), but in others was not (McCall-Hosenfeld et al., 2009; Millar et al., 2002).

Community Size and Location as Potential Correlates of Time Until Presentation

Although the characteristics of sexual assault events and survivors are important for understanding delays in time until presentation, it is also important to remember that sexual assaults and the events that unfold thereafter occur within a community context. Communities in which individuals are situated have characteristics, such as their geography or their social structure, which may differentially affect time until presentation. Although both of the aforementioned characteristics are often described using the same concept of rurality (Weisheit, Wells, & Falcone, 1995), the mechanisms through which each would affect time until presentation are distinct.

In the case of rurality’s effect on individual access to health care, the geographical accessibility of resources is often a key focus: rural–urban differences in using resources may be viewed as a product of rural residents having fewer resources within easy travel distance than urban residents (Borders & Booth, 2007; Brems, Johnson, Warner, & Roberts, 2006). Rural rape survivors indicated that a lack of resources and transportation options were barriers to accessing health and mental health services in focus group discussions, whereas urban rape survivors did not (Logan et al., 2005). More recently, the United States Government Accountability Office published a report indicating that the number of sexual assault forensic examiners in six states did not meet the need for exams, particularly in rural areas (Government Accountability Office, 2016). Rural survivors also identified long wait times for police due to large and isolated geographic areas as being a barrier to criminal justice system services, a barrier also not identified by the urban survivors (Logan et al., 2005). These findings, therefore, indicate that the geography of a sexual assault survivor’s context may affect time until presentation for medical care through the mechanisms of differing travel distance to appropriate services and/or, relatedly, few transportation options to cover that distance.

The mechanisms through which a community’s social structure (i.e., its social life and social order) may affect time until presentation are quite different from those of its geography. Rural areas are those that are typically characterized by social connections that are more complete (based on a knowledge of others’ personal backgrounds rather than their formal role positions) than those in urban areas (Weisheit et al., 1995). Concerns about privacy, confidentiality, and/or stigma have been identified as barriers to accessing a variety of health care services in rural areas by a variety of demographic populations (Leston, Jessen, & Simons, 2012; Logan, Stevenson, Evans, & Leukefeld, 2004; Sexton, Carlson, Leukefeld, & Booth, 2008). More importantly, these concerns have also been identified as barriers to health, mental health, and criminal justice service use by rape survivors (Logan et al., 2005).

Not only are concerns about privacy, confidentiality, and/or stigma identified as barriers to service use in rural areas, but they are also identified as having differential effects based on rurality versus urbanicity. For example, a survey of physical and behavioral health care providers in rural and urban settings in Alaska and New Mexico found that confidentiality limitations were rated more strongly as a barrier to health care delivery in rural areas than in urban areas, with providers in the smallest rural areas giving these issues the strongest rating of being problematic (Brems et al., 2006). Further, privacy and perceived stigma have been identified as reasons behind differences in mental health care access between rural and urban residents (Hoyt, Conger, Valde, & Weihs, 1997; Rost, Fortney, Fischer, & Smith, 2002).

Lastly, rural areas may also be culturally different from urban areas, perhaps in part based on the demographics of the communities (e.g., Alaska Native or American Indian communities), which could influence norms around seeking help or talking about sexual assault. To refer back to Millar et al.’s (2002) finding that survivors who identified as First Nations/Aboriginal, Hispanic, or South Asian presented earlier than did White, Black, or Asian survivors, there is some indication that individuals of differing race/ethnicities are differentially affected when seeking help. Although many other factors could be affecting these differences, it is possible that cultural norms could play a role. Given that some rural communities are predominantly American Indian/Alaska Native, there could be an additive effect of rurality and culture in these communities that affects delays in presentation.

Whether the effect of context is a result of geographic rurality, social structural rurality, cultural rurality, or entirely different community characteristics, it has been found that the importance of assault event and survivor characteristics in predicting whether women report violence committed against them varies across geographic contexts (Rennison et al., 2013). Further, rape survivors divided into rural and urban focus groups identified differing barriers to health, mental health, and criminal justice service use (Logan et al., 2005). In conjunction with the finding that rural counties have lower rates of survivors reporting sexual victimizations to police than do urban counties (Ruback & Ménard, 2001), the above body of literature indicates that rural and urban community contexts could differentially influence a sexual assault survivors’ time to presentation for help to both law enforcement and medical care professionals.

Goal of Study

The primary goal of this study was to test how community context affects sexual assault survivor time until presentation for medical care. It was hypothesized that survivors presenting in rural communities would be significantly more likely to present later than survivors presenting in the most urban community. It was also hypothesized that survivors presenting in communities different from those of the assault would be significantly more likely to present later than survivors presenting in the same community as the assault. A secondary goal of this study was to re-examine commonly used predictors of time until presentation to test the consistency of findings across various samples and methodologies. It was hypothesized that known assailants and survivor intoxication at the time of assault would be associated with delayed presentation, and that survivor race/ethnicity would also be related (although the direction of the relationship was not specified due to mixed prior findings).

Methods

Setting

The data from this study come from rural and urban communities across the state of Alaska. Sexual assault is a prevailing concern in this state, as 37% of adult Alaskan women have reported that they have experienced sexual violence at least once in their lifetime (UAA Justice Center, 2010). Alaska has for many years had the highest average forcible rape rate (per the FBI’s Uniform Crime Report) in the United States, as evidenced in 2012 with a rate of 79.7 per 100,000, followed by South Dakota with a rate of 70.2 and Michigan with a rate of 46.4 (United States Department of Justice, 2014).

Sexual violence rates are not Alaska’s only exceptional characteristic. Alaska is also unique because of the extreme differences in types of rural and urban communities across the state. Alaska is larger than Texas, California, and Nevada (the next three largest states) combined (Indian Law and Order Commission, 2013), but in contrast to its vast physical space, it has one of the smallest populations in the United States with just over 735,000 residents (United States Census Bureau, 2014). Although many of these residents live in the state’s largest cities such as Anchorage, Juneau, and Fairbanks, many residents also live in small and/or remote communities (Indian Law and Order Commission, 2013).

Sampling

To better address the immediate needs of sexual assault survivors, Alaska has adopted the use of Sexual Assault Response Teams (SARTs) in several communities across the state. These teams include law enforcement, victim advocates, and Sexual Assault Nurse Examiners (SANEs; Rosay & Henry, 2008), and have been used since the 1970s in hundreds of communities across the United States (Greeson & Campbell, 2012). Through the SART protocol, sexual assaults reported to law enforcement are referred to a SANE if there is need for medical attention, if there is a likelihood of collecting forensic evidence, and/or if the minimum legal requirements of proof have been met. The SANEs are specifically trained to interact with survivors of sexual assault, and the exams follow a standard protocol that includes the documentation and collection of forensic evidence (Rosay & Henry, 2008). As it relates to survivors receiving medical care, it is important to note that SARTs typically operate under a “victim-centered” philosophy, meaning that they prioritize ensuring survivors’ choices regarding which services and systems the survivor becomes involved with (Greeson & Campbell, 2012).

The data for this study came from the Alaska SANE data set, which was compiled by Rosay and Henry (2008) from archived SANE evaluations conducted in Anchorage from 1996 to 2004; in Bethel and Fairbanks in 2005 and 2006; and in Homer, Kodiak, Kotzebue, Nome, and Soldotna in 2005. The original study population included 1,699 cases. Male survivors and statutory assault survivors were excluded from the sample for the present study to avoid confounding conceptual differences between male and female survivors (Bullock & Beckson, 2011; Davies, 2002; Du Mont, Macdonald, White, & Turner, 2013) and between assault and statutory assault (Glosser, Gardiner, & Fishman, 2004). The final sample, therefore, consisted of 1,630 females of all ages who contacted law enforcement and were subsequently examined by a SANE for a non-statutory sexual assault in one of eight Alaskan communities between 1996 and 2006. It should be emphasized that these survivors first chose to report their assaults to law enforcement and were then triaged to the SANE, and no survivors were required to report the assault to law enforcement through the SANE/SART process.

Dependent Variable: Time Until Presentation

One dichotomous variable captured whether survivors presented within 12 hr of their assaults. To reiterate, all of these women first reported to a law enforcement officer and were then triaged to the SANE, and thus their time until presentation includes the amount of time for the survivors to report to law enforcement and for law enforcement to connect them with medical care.

Predictors of Time Until Presentation

Location

Operationalizing rurality is often complicated by the “multidimensional” nature of rurality: Areas can be classified as rural based on their demographic/geographic, economic, social structural, and/or cultural compositions (Weisheit et al., 1995). All four of these dimensions are important for understanding rural–urban differences, but the demographic dimension, where rural areas are those that are sparsely populated and geographically isolated (Weisheit et al., 1995), is the most commonly used. Although common, this method of measurement has its own complications for statistical modeling due to the inherently small numbers in rural areas. For reasons that will be further explained in the Analytic Plan, this study used seven dummy variables (each with Anchorage as its reference category) to account for the eight individual communities (cities or municipalities per the United States Census) in which the exam occurred as the measure of community context. Descriptively, these communities can be classified as urban or rural using four ordinal categories established by Brems et al. (2006): “small rural” (pop. less than 3,500), “rural” (pop. 3,500-14,999), “small urban” (pop. 15,000-34,999), and “urban” (pop. 35,000 or more) per the 2010 United States Census. These classifications and other Census data related to each community are listed in Table 1.

Table 1.

Community Characteristics.

Location of SANE	Data Collection Years	Study n	Population 2010	Rurality Classification	Percent American Indian/Alaska Native^a	% of Survivors Presenting in Same Community in which They Were Assaulted^b
Anchorage	1996-2004	1,328	291,826	Urban	7.9	94.8
Fairbanks	2005-2006	139	31,535	Small Urban	10.0	74.2
Bethel	2005-2006	100	6,080	Rural	65.0	40.0
Homer	2005	9	5,003	Rural	4.1	66.7
Kodiak	2005	3	6,130	Rural	9.9	66.7
Kotzebue	2005	21	3,201	Small Rural	80.8	66.7
Nome	2005	16	3,598	Rural	65.3	56.3
Soldotna	2005	14	4,163	Rural	8.7	27.9

The eight communities (cities or municipalities per the United States Census) in this study are classified as urban or rural using four ordinal categories established by Brems, Johnson, Warner, and Roberts (2006): “small rural” (pop. less than 3,500), “rural” (pop. 3,500-14,999), “small urban” (pop. 15,000-34,999), and “urban” (pop. 35,000 or more) per the 2010 United States Census.

Relative location. Percents are for n (1630) after imputing values for missing data using Multiple Imputation (MI).

Relative location

Whether the assault took place in same community as the SANE (yes or no) was included because having to travel to another community for the exam could lengthen the delay between assault and presenting to the nurse examiner. It was hypothesized that survivors whose exams were in communities different from their assaults would present later than those whose assaults and exams were in the same communities.

Other predictors

Rural areas can be difficult to analyze quantitatively because they are by definition characterized by small numbers. To maintain statistical power when working with small numbers, a parsimonious model with as few variables as possible is desired. Therefore, for this study, variables were included only if they were conceptually expected and/or empirically established to influence time until presentation and vary by community context. Because of strong collinearity between the year in which survivors presented and the communities to which they presented, the year was not included as a temporal control.

The race/ethnicity of the survivor was included because prior studies have linked race/ethnicity to variations in how female survivors mentally process rape (Bletzer & Koss, 2006) and to variations in delays between sexual assaults and seeking medical care (Millar et al., 2002). Further, race/ethnicity was expected to vary by context, given the strikingly different variations in the percent of each community that was American Indian and Alaska Native (AIAN) per the 2010 Census (See Table 1). For the current study, the race/ethnicity of the survivor was measured using one dichotomous variable: whether the survivor identified as AIAN or not.¹ No specific hypotheses were formulated as to how race/ethnicity would affect time until presentation in this study.

The survivor’s relationship with the assailant(s) was included because it has been found to be significantly associated with delays in presentation in most analyses (Adefolalu, 2014; McCall-Hosenfeld et al., 2009; Millar et al., 2002; Nesvold et al., 2008), with only one study failing to find an association (Logan et al., 2007). Not only is the survivor–assailant relationship one of the most consistent predictors of delays in presentation, it was also expected that survivor–assailant relationships would vary by community context. The likelihood of knowing the assailant may be expected to be greater in small communities than in large communities because of the denser nature of rural social connections (Annan, 2006, 2011; Ruback & Ménard, 2001). This study used three categories to capture the relationship between the survivor and the assailant: (a) stranger, (b) known non-partner (including acquaintances, friends, former spouses or partners, relatives, authority figures), and (c) partner (including spouses). In the case of multiple assailants, the most intimate relationship was used.² Stranger was used as the reference category, and it was predicted that survivors assaulted by partners would be significantly more likely to delay presentation than those assaulted by strangers, and that those assaulted by known non-partners would be more likely to delay presentation than those assaulted by strangers.

Whether or not the survivor was alcohol intoxicated at the time of the assault was included in the analyses, despite somewhat varied empirical support. For example, although the survivor’s intoxicant exposure at the time of the assault was associated with later presentation in McCall-Hosenfeld et al.’s (2009) bivariate comparisons, it was not significant in their multivariate analyses nor was it significant in either Adefolalu (2014) or Millar et al.’s (2002) studies. There is, however, fair reason to believe that alcohol intoxication could impair judgment and/or influence self-blame on the survivor’s behalf, and thus delay reporting and ultimately time until presentation. Blame could be externally placed on the survivor by the responding officers, which could lead them to hesitate to classify the incident as a sexual assault and thus delay their triaging the survivor to medical care. Further, although empirical measures of Alaska’s rural–urban differences in alcohol use prevalence are difficult to locate, it is known that rates of alcohol-induced deaths are higher in Alaska’s rural areas than its urban areas (Hull-Jilly & Casto, 2013), and is also largely understood by criminal justice professionals in the state that rates of alcohol consumption are more problematic in rural areas (Alaska State Troopers, 2011), potentially indicating a larger problem with alcohol use in rural communities.³ It was therefore hypothesized that survivors who were intoxicated at the time of their assault would present later than those who were not.

Missing Data

Following the rule that more than 5% missing data points in a small-to-moderately sized data set qualifies as a high amount (Tabachnik & Fidell, 2013), four variables (time until presentation, relationship with the assailant, relative location, and intoxication) were identified as potentially problematic. There were significant differences between those cases with data for these four variables and those with missing data. These significant differences indicated that the missing data points for each variable were not missing completely at random (MCAR). List-wise deletion of cases with missing data would, therefore, introduce bias into the analyses and decrease their statistical power (Meyers, Gamst, & Guarino, 2013). Missing data for each of these four variables were therefore estimated using the Multiple Imputation (MI) algorithm in IBM’s Statistical Package for the Social Sciences (SPSS) version 22 with the Missing Values add-on module (IBM, 2011).⁴

Survivor and Assault Characteristics

Following the imputation, 58.3% of the sample presented within 12 hr. Most survivors were American Indian/Alaska Native (56.2%) and roughly two thirds were alcohol intoxicated at the time of their assault (66.9%). Very few of the survivors were assaulted by their partners (5.5%), especially when compared with those assaulted by known non-partners (77.6%). Most assaults occurred in the same community as the SANE (88.2%).

Analytic Plan

As indicated earlier in the Methods section, the eight communities in which survivors presented were used to measure community context instead of the demographic categories of rurality used by Brems et al. (2006). This method was used because there were too few communities with a sufficient number of survivors per community to conduct a multi-level model (i.e., mixed effects), but including a group-level variable (i.e., rurality) in a fixed effects model (e.g., logistic regression) can force the model to assume “ . . . that the regression coefficients apply equally to all contexts” (Luke, 2004, p. 7). Looking at each community separately in a fixed effects model, however, does not force the model to make such assumptions and would allow for looking at differences between communities. Binary logistic regression analyses were therefore conducted in IBM’s Statistical Package for the Social Sciences (SPSS) version 22 with the Missing Values add-on module (IBM, 2011) using the pooled parameter estimates for the missing values on time until presentation, relationship with the assailant(s), relative location, and survivor intoxication.⁵

Two models were run. In Model 1, only the seven community dummy variables were included (with Anchorage as the reference community). In Model 2, the community variables along with the other predictors (relative location, race/ethnicity, survivor intoxication, relationship with the assailant(s)) were included.

Results

Results are displayed in Table 2. Both models had non-significant chi-square values for the Hosmer and Lemeshow test, indicating they were a good fit for the data. In Model 1, survivors presenting in Bethel (a rural community) were significantly different from those presenting in Anchorage (the most urban community). Specifically, survivors presenting in Bethel were less than half as likely as those presenting in Anchorage to present within 12 hr. In Model 2, with the individual-level factors included, there were no unique community differences. Survivors presenting in the same community in which they were assaulted, however, were 3.1 times more likely than those presenting in a different community to present within 12 hr. Non-AIAN survivors were 0.60 times as likely as AIAN survivors to present within 12 hr. Finally, survivors assaulted by a known non-partner were 0.33 times as likely as those assaulted by strangers to present within 12 hr, and survivors assaulted by a partner were 0.40 times as likely to do so. Neither model explained a large amount of variance, with a Cox and Snell R² of .01 for the first and .09 for the second model. Comparison of model fit showed a statistically significant reduction in the −2 log likelihood ratios from Model 1 to Model 2, indicating the second model provided the best fit of the two models.

Table 2.

The Odds of Presentation Within 12 Hr.

	Model 1 (n = 1,630)					Model 2 (n = 1,617)
	B	SE	Exp(B)	95% CI for Exp(B)		B	SE	Exp(B)	95% CI for Exp(B)
	B	SE	Exp(B)	Lower	Upper	B	SE	Exp(B)	Lower	Upper
Community^a
Bethel	−0.72	0.22	0.49**	0.32	0.74	−0.24	0.25	0.79	0.48	1.30
Fairbanks	−0.28	0.19	0.76	0.52	1.10	−0.01	0.20	0.99	0.66	1.47
Homer	0.57	0.79	1.77	0.38	8.27	1.95	1.09	7.05	0.82	60.44
Kodiak	−1.10	1.23	0.33	0.03	3.75	−0.89	1.27	0.41	0.03	4.93
Kotzebue	0.52	0.51	1.69	0.61	4.64	0.70	0.55	2.02	0.69	5.90
Nome	−0.28	0.53	0.76	0.27	2.14	0.05	0.56	1.05	0.35	3.16
Soldotna	0.09	1.07	1.09	0.12	10.13	1.30	1.21	3.65	0.29	45.88
Relative location^b						1.14	0.20	3.13***	2.12	4.64
Race/ethnicity^c						−0.51	0.15	0.60**	0.44	0.81
Alcohol intoxication^d						0.28	0.15	1.32	0.97	1.80
Relationship^e
Known non-partner						−1.11	0.20	0.33***	0.22	0.49
Partner						−0.92	0.30	0.40**	0.22	0.71
−2 log likelihood	(avg.) 2,190.13					(avg.) 2,040.01
Cox and Snell R²	(avg.) .01					(avg.) .09
Nagelkerke R²	(avg.) .02					(avg.) .12
Hosmer and Lemeshow	(avg.) χ² = 0.00, df (1), p = 1.00					(avg.) χ² = 9.26, df (7), p = .27

Note. CI = confidence interval. These numbers are pooled estimates because Multiple Imputation (MI) was used to replace missing values for several variables. See discussion on Missing Data in the Methods section. Missing data for race/ethnicity were not imputed because this variable was missing too few cases to be problematic (less than 1%). The number of cases for race/ethnicity and therefore Model 2 was 1,617.

Ref = Anchorage.

Ref = Different location.

Ref = American Indian/Alaska Native.

Ref = Sober.

Ref = Stranger.

p < .05. **p < .01. ***p < .001.

Discussion

Rural Location

Survivors presenting in Bethel were more likely to delay presentation than those presenting in Anchorage, although this difference disappeared when controlling for the other variables in Model 2. It could very well be that differences in relative location accounted for the significant difference between Bethel and Anchorage in Model 1. Referring back to Table 1, only 40% of those presenting in Bethel were assaulted in the same community as their exam, as opposed to 94.8% in Anchorage. The question remains, however, why the other communities with similarly low rates of assault and presentation in the same community were not significantly different from Anchorage in Model 1. It is possible that the numbers in these other rural communities were too low to provide enough statistical power to detect a difference, and differences are not necessarily expected between Fairbanks (also an urban community) and Anchorage. It is also possible that using the SANE location as the measure of community context was too rough a proxy to capture the effects of rurality on time until presentation.

Many of the assumptions about why rurality would affect time until presentation (presented in the Introduction) involve the assault occurring in a rural location (and thus forcing the survivor to overcome travel and resource barriers) and the survivor being from a rural community (and thus forcing the survivor to overcome anonymity and confidentiality barriers). The measure of where the medical exam occurred obviously does not actually measure those two factors, and is therefore a very rough (albeit the best available in this particular study because data on the other two are not available) proxy for the mechanisms initially proposed. Because of this significant limitation, more research is needed in this area, using a measure of rurality that captures the survivor’s and the assault’s community context, as well as variables that try to assess the mechanisms of access, anonymity, and confidentiality concerns.

In addition to a more precise measure of rurality, larger numbers are needed. The total number of communities was relatively small (because they were necessarily limited to where SANE had been implemented), as well as was the number of survivors presenting in rural communities. These small numbers could have resulted in an ability to detect an effect where one might actually exist. A larger sample that includes more rural areas could overcome this limitation. With its large sample size and geo-located data of crime victims (including sexual assault survivors), to really assess the impact of these contextual variables, it would be desirable for the National Crime Victimization Survey (NCVS) be modified to include time until presentation variables.

Relative Location

Survivors whose assault was in the same community as their exam were more likely than those presenting in different communities to present earlier, as hypothesized. This finding is not surprising—if the assault and exam were in different locations, the time until presentation would incorporate the travel time between them and would therefore be longer.

What remains unknown is why those survivors who presented in different communities than their assault did so. Was it because there were services in the communities they were assaulted in but they were not comfortable using them? Or were there no services available? Both are plausible but untested explanations for the current findings. It is also unknown exactly why presenting in different communities than the assault resulted in delays, which has distinct policy and practice implications. For example, are the delays a result of having to drive 5 hr to access medical services, a result of not being able to drive and thus waiting for a plane to transport them to services, or some other reason? Measures that could capture these various differences are beyond the scope of the SANE examinations and therefore require data collection through different means.

Despite the above unknowns, the relative location finding does suggest that if we wish for sexual assault survivors to receive services as soon as possible, appropriate and reliable resources are needed in as many communities as possible to perform sexual assault examinations. Although it is unknown as to whether services were available in the communities these survivors were assaulted in but they chose not to use them, it is arguable that if the services in the location of their assault had been appropriate and reliable, they would choose to use them. Appropriate and reliable resources would therefore be services that are not only known, but also resources that can ensure confidentiality and effectively combat stigma around sexual assault victimization.

Race/Ethnicity

American Indian/Alaska Natives were more likely to present earlier rather than later than were non-AIANs. This finding is somewhat consistent with Millar et al.’s (2002) finding that survivors who identified as First Nations/Aboriginal, Hispanic, or South Asian presented earlier than did White, Black, or Asian women (Millar et al., 2002). Although they offered no explanation for this finding, Millar et al. suggested that further explorations need to be made into understanding the role of societal attitudes regarding culpability for the attack, and into personal and cultural belief systems around blame and societal standards. It is also worth noting that not only were Millar et al.’s racial categories somewhat different, their measure of time was an ordinal measure with four categories. That both studies found race/ethnicity to be significant using varying measures indicates that the race/ethnicity of the survivor deserves further study. It is also important to explore this predictor further because Nesvold et al. (2008) and McCall-Hosenfeld et al. (2009) both included some measure of race/ethnicity and neither study found this variable to be significant. Until the reason for these differences can be teased out (is it the measurement of race/ethnicity; the measurement of time; something different?), it is difficult to extrapolate too much on the findings around race/ethnicity from this study.

Relationships with Assailants

Consistent with the prior literature (Adefolalu, 2014; McCall-Hosenfeld et al., 2009; Millar et al., 2002; Nesvold et al., 2008) and with the hypothesis regarding this variable, the survivors’ relationships with their assailants were significantly related to time until presentation, with survivors who were more intimately related to their assailants being more likely to present later than those who were less related. Survivors with more well-known assailants are likely to struggle more to define their situation as an assault (Kahn, Jackson, Kully, Badger, & Halvorsen, 2003; Koss, 1985; Littleton, Breitkopf, & Berenson, 2008; Orchowski, Untied, & Gidycz, 2013), which could therefore lead to delays in help-seeking (Ullman, 2010). Alternatively, survivors who know their assailants could be more hesitant to report the assault out of feelings of not wanting to get the assailant in trouble or out of fear of reprisal from the assailant or other mutual acquaintances. To echo McCall-Hosenfeld et al.’s (2009) suggestion, “Preventative efforts must include public health campaigns to better inform survivors that assault by a known assailant is a crime, and early access to medical care may mitigate adverse outcomes” (p. 595).

Delays on behalf of the survivor, however, are only half of the problem when considering that these survivors were triaged to medical care by law enforcement officers (LEOs). It is possible that LEOs who respond to sexual assaults could be more hesitant to classify more intimate attacks as assaults due to misconceptions about these crimes, which could lead to delays in presentation. Although the SARTs are designed specifically to respond adequately to sexual assault, Greeson and Campbell (2012) found that many organizations involved in SARTs, particularly police departments, have limited resources to devote to trainings specific to sexual assault. It is therefore suggested here that LEOs be better trained to understand that all sexual assault is a crime (including intimate sexual assault). Greeson and Campbell suggest that SARTs could apply for mini-grants to bring trainings or presentations on specialized topics to their officers.

The solid relationship between known assailants and delays in presentation, along with the above suggested remedies, are particularly salient considering the finding from the Bureau of Justice Statistics analysis of NCVS data that 57% of sexual assault survivors knew their assailants (Catalano, Smith, Snyder, & Rand, 2009).

Limitations

As discussed in the Methods section, the time until presentation variable includes the time it took each survivor to report her assault to law enforcement as well as the amount of time it took law enforcement to triage her to the SANE. Within this data set, there is no way to untangle these two periods of time (assault to law enforcement and law enforcement to SANE) which involve conceptually different antecedents and implications. Without these distinctions, it is difficult to interpret time until presentation differences.

Despite this limitation, the findings of this article still have research and policy value. This is the first article to look at community context as a predictor of time until presentation, and therefore, creates a basis for studying the nuances of differences between communities. The conceptual argument has been made for why differences might exist, and further studies can use this argument to justify their analyses, and can use the methods of this work as a starting point for making improvements. Policy-wise, these results provide insight into several variables that should perhaps receive different approaches when considering how to reduce time until presentation.

Another limitation to recognize is that the process through which survivors were selected for this study is not independent of the outcome of interest. That is to say, time until presentation can only exist if the survivor contacted law enforcement about her sexual assault and presented for medical care. It is therefore possible that the impact of the independent variables on time until presentation may be confounded with those variables’ impact on the survivor’s likelihood of presenting. Jones, Alexander, Wynn, Rossman, and Dunnuck (2009) found women not reporting assault to police were more likely to be employed, have recently used drugs or alcohol, a known assailant, longer time intervals before seeking medical care, a previous relationship with the assailant, a greater desire to not see their assailant jailed, and a greater fear that police would be insensitive or blame them for the assault. This selection problem is especially relevant to the question of rural and urban differences, as it could take longer for survivors to come into contact with law enforcement in rural areas than in urban, thus decreasing the likelihood of collecting forensic evidence, and thus decreasing the likelihood that they would present to the SANE. Unfortunately, because these data were collected from SANE records, it is not possible to learn about survivors who reported their sexual assault to law enforcement but were not triaged to a SANE and to therefore model these selection dynamics. The results of this study therefore can only be applied to survivors who reported to law enforcement and presented for medical care.

Conclusion

This study has established that survivors who present in communities different from those in which the assault occurred have a greater likelihood of presenting later, pointing to a need for appropriate and reliable resources for sexual assault survivors in as many communities as possible. Further, it has added to mixed findings around the effects of race/ethnicity on time until presentation—indicating the need for further conceptualizing and study to make better policy recommendations. Lastly, it has re-affirmed the importance of known assailant(s) in affecting time until presentation, which has significant policy and practice implications that can be acted on with confidence. More broadly, these results provide insight into problems surrounding sexual assault in the state with the highest rates of sexual violence in the country, and this study has laid the foundation for studying community context as a potential correlate of time until presentation. It could be argued that Alaska’s exceptionalism and the unique characteristics and experiences of the individuals examined in this study may limit the generalizability of this study’s findings, but as Rosay and Henry (2008) argued, “ . . . there is no indication that results from Alaska are somewhat less generalizable than results from any other single state” (p. 32). Even if these results are not perceived as generalizable and therefore do little to inform a larger body of literature, they can provide insight into problems surrounding sexual assault in the state with the highest rates of sexual violence in the country, a contribution that is doubtlessly important.

Footnotes

Acknowledgements

The authors would like to thank Dr. André Rosay for access to the Alaska SANE Data Set, and for his input on the development of this article. The authors would also like to thank the anonymous faculty reviewers in the Criminal Justice Department at Temple University for their suggestions and support.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Ingrid Diane Johnson is a PhD student in the Criminal Justice Department at Temple University. Her research focuses on the help-seeking processes of survivors of sexual assault and intimate partner violence, and how formal and informal networks can improve those processes. She currently works as a research assistant for the Juvenile Justice Translational Research on Interventions for Adolescents in the Legal System (JJ-TRIALS) project.

Matthew L. Hiller is an associate professor in the Department of Criminal Justice at Temple University. Before moving to Temple in 2004, he worked as an associate research scientist at the Institute of Behavioral Research at Texas Christian University (1997-2000) and as an assistant research professor in the Center for Drug Abuse Research at the University of Kentucky (2000-2004). His research focuses on substance abuse treatment in the criminal justice system. Areas of expertise within this include response to treatment, treatment retention and process, mental illness and co-occurring disorders, treatment motivation, coercion and treatment, and research on implementation science and the adoption of evidence-based practices by criminal justice programs. His skill set includes program evaluation, research design, quantitative methods, meta-analysis, and survey research. He has more than 60 articles and book chapters and has served as principal investigator or co-investigator on numerous research grants funded through the National Institute of Drug Abuse, Center for Substance Abuse Treatment, Bureau of Justice Assistance, and the National Institute of Justice. He currently is finishing research projects with the Criminal Justice Collaborating Council in Waukesha, Wisconsin, where he has been conducting on-going process and outcome evaluations of a DUI court for multiple repeat offenders and a drug court for heroin-dependent probationers.

References

Adefolalu

A. O.

(2014). Fear of the perpetrator: A major reason why sexual assault victims delayed presenting at hospital. Tropical Medicine & International Health, 19, 342-347.

Alaska State Troopers. (2011). 2011 annual drug report. Retrieved from http://dps.alaska.gov/AST/ABI/docs/SDEUreports/2011%20Annual%20Drug%20Report.pdf

Annan

S. L.

(2006). Sexual violence in rural areas: A review of the literature. Family & Community Health, 29, 164-168.

Annan

S. L.

(2011). “It’s not just a job. This is where we live. This is our backyard”: The experiences of expert legal and advocate providers with sexually assaulted women in rural areas. Journal of the American Psychiatric Nurses Association, 17, 139-147.

Bletzer

K. V.

Koss

M. P.

(2006). After-rape among three populations in the Southwest: A time of mourning, a time of recovery. Violence Against Women, 12, 5-29.

Borders

T. F.

Booth

B. M.

(2007). Research on rural residence and access to drug abuse services: Where are we and where do we go? Journal of Rural Health, 23, 79-83.

Brems

Johnson

M. E.

Warner

T. D.

Roberts

L. W.

(2006). Barriers to healthcare as reported by rural and urban interprofessional providers. Journal of Interprofessional Care, 20, 105-118.

Bullock

Beckson

(2011). Male victims of sexual assault: Phenomenology, psychology, physiology. Journal of the American Academy of Psychiatry and the Law, 39, 197-205.

Catalano

Smith

Snyder

Rand

(2009). Female victims of violence. Retrieved from http://www.bjs.gov/content/pub/pdf/fvv.pdf

10.

Davies

(2002). Male sexual assault victims: A selective review of the literature and implications for support services. Aggression and Violent Behavior, 7, 203-214.

11.

Du Mont

Macdonald

White

Turner

(2013). Male victims of adult sexual assault: A descriptive study of survivors’ use of sexual assault treatment services. Journal of Interpersonal Violence, 28, 2676-2694.

12.

Dunlap

Brazeau

Stermac

Addison

(2004). Acute forensic medical procedures used following a sexual assault among treatment-seeking women. Women & Health, 40, 53-65.

13.

Glosser

Gardiner

Fishman

(2004). Statutory rape: A guide to state laws and reporting requirements. Retrieved from https://aspe.hhs.gov/report/statutory-rape-guide-state-laws-and-reporting-requirements-summary-current-state-laws

14.

Government Accountability Office. (2016). Sexual assault: Information on training, funding, and the availability of forensic examiners. Retrieved from http://www.gao.gov/assets/680/675879.pdf

15.

Greeson

M. R.

Campbell

(2012). Sexual Assault Response Teams (SARTs): An empirical review of their effectiveness and challenges to successful implementation. Trauma, Violence, & Abuse, 14, 83-95.

16.

Hoyt

D. R.

Conger

R. D.

Valde

J. G.

Weihs

(1997). Psychological distress and help seeking in rural America. American Journal of Community Psychology, 25, 449-470.

17.

Hull-Jilly

Casto

L. D.

(2013). State epidemiologic profile on substance use, abuse and dependency. Retrieved from http://dhss.alaska.gov/dbh/Documents/Prevention/EPI2013.pdf

18.

IBM. (2011). IBM SPSS missing values 20. Retrieved from http://www.csun.edu/sites/default/files/missing-values20-64bit.pdf

19.

Indian Law and Order Commission. (2013). Reforming justice for Alaska Natives: The time is now. Retrieved from http://www.aisc.ucla.edu/iloc/report/files/Chapter_2_Alaska.pdf

20.

Jones

J. S.

Alexander

Wynn

B. N.

Rossman

Dunnuck

(2009). Why women don’t report sexual assault to police: The influence of psychosocial variables and traumatic injury. Journal of Emergency Medicine, 36, 417-424.

21.

Kahn

A. S.

Jackson

Kully

Badger

Halvorsen

(2003). Calling it rape: Differences in experiences of women who do or do not label their sexual assault as rape. Psychology of Women Quarterly, 27, 233-242.

22.

Kennedy

A. C.

Adams

Bybee

Campbell

Kubiak

S. P.

Sullivan

(2012). A model of sexually and physically victimized women’s process of attaining effective formal help over time: The role of social location, context, and intervention. American Journal of Community Psychology, 50, 217-228.

23.

Koss

M. P.

(1985). The hidden rape victim: Personality, attitudinal and situational characteristics. Psychology of Women Quarterly, 9, 192-212.

24.

LeBeau

M. A.

Montgomery

M. A.

(2010). Challenges of drug-facilitated sexual assault. Forensic Science Review, 22, 1-6.

25.

Leston

J. D.

Jessen

C. M.

Simons

B. C.

(2012). Alaska Native and rural youth views of sexual health: A focus group project on sexually transmitted diseases, HIV/AIDS, and unplanned pregnancy. American Indian and Alaska Native Mental Health Research: The Journal of the National Center, 19, 1-14.

26.

Littleton

Breitkopf

C. R.

Berenson

(2008). Beyond the campus: Unacknowledged rape among low-income women. Violence Against Women, 14, 269-286.

27.

Logan

T. K.

Cole

Capillo

(2007). Differential characteristics of intimate partner, acquaintance, and stranger rape survivors examined by a Sexual Assault Nurse Examiner (SANE). Journal of Interpersonal Violence, 22, 1066-1076.

28.

Logan

T. K.

Evans

Stevenson

Jordan

C. E.

(2005). Barriers to services for rural and urban survivors of rape. Journal of Interpersonal Violence, 20, 591-616.

29.

Logan

T. K.

Stevenson

Evans

Leukefeld

(2004). Rural and urban women’s perceptions of barriers to health, mental health, and criminal justice services: Implications for victim services. Violence and Victims, 19, 37-62.

30.

Luke

D. A.

(2004). Multilevel modeling. Thousand Oaks, CA: Sage.

31.

McCall-Hosenfeld

J. S.

Freund

K. M.

Liebschutz

J. M.

(2009). Factors associated with sexual assault and time to presentation. Preventive Medicine, 48, 593-595.

32.

Meyers

L. S.

Gamst

Guarino

A. J.

(2013). Applied multivariate research: Design and interpretation (2nd ed.). Thousand Oaks, CA: Sage.

33.

Millar

Stermac

Addison

(2002). Immediate and delayed treatment seeking among adult sexual assault victims. Women & Health, 35, 53-64.

34.

Nesvold

Friis

Ormstad

(2008). Sexual assault centers: Attendance rates, and differences between early and late presenting cases. Acta Obstetricia et Gynecologica Scandinavica, 87, 707-715.

35.

New York State Department of Health AIDS Institute. (2013). HIV prophylaxis for victims of sexual assault. Retrieved from http://www.hivguidelines.org/clinical-guidelines/post-exposure-prophylaxis/hiv-prophylaxis-for-victims-of-sexual-assault/

36.

Orchowski

L. M.

Untied

A. S.

Gidycz

C. A.

(2013). Factors associated with college women’s labeling of sexual victimization. Violence and Victims, 28, 940-958.

37.

Piaggio

von Hertzen

Grimes

D. A.

Van Look

P. F. A.

(1999). Timing of emergency contraception with levonorgestrel or the Yuzpe regimen. The Lancet, 353, 721.

38.

Rennison

C. M.

Dragiewicz

DeKeseredy

(2013). Context matters: Violence against women and reporting to police in rural, suburban, and urban areas. American Journal of Criminal Justice, 38, 141-159.

39.

Resnick

H. S.

Holmes

M. M.

Kilpatrick

D. G.

Clum

Acierno

Best

C. L.

Saunders

B. E.

(2000). Predictors of post-rape medical care in a national sample of women. American Journal of Preventative Medicine, 19, 214-219.

40.

Rosay

A. B.

Henry

(2008). Final report: Alaska sexual assault nurse examiner study. Retrieved from https://www.ncjrs.gov/pdffiles1/nij/grants/224520.pdf

41.

Rost

Fortney

Fischer

Smith

(2002). Use, quality, and outcomes of care for mental health: The rural perspective. Medical Care Research and Review, 59, 231-265.

42.

Ruback

R. B.

Ménard

K. S.

(2001). Rural-urban differences in sexual victimization and reporting: Analyses using UCR and crisis center data. Criminal Justice and Behavior, 28, 131-155.

43.

Sexton

R. L.

Carlson

R. G.

Leukefeld

C. G.

Booth

B. M.

(2008). Barriers to formal drug abuse treatment in the rural South: A preliminary ethnographic assessment. Journal of Psychoactive Drugs, 40, 121-129.

44.

Tabachnik

B. G.

Fidell

L. S.

(2013). Using multivariate statistics (6th ed.). Boston: Pearson Education.

45.

Truman

J. L.

Langton

(2014). Criminal victimization, 2013. Retrieved from http://www.bjs.gov/index.cfm?ty=pbdetail&;iid=5111

46.

UAA Justice Center. (2010). Intimate partner violence and sexual violence in the State of Alaska: Key results from the 2010 Alaska Victimization Survey. Retrieved from http://justice.uaa.alaska.edu/research/2010/1004.avs_2010/1004.07a.statewide_summary.pdf

47.

Ullman

S. E.

(2010). Talking about sexual assault: Society’s response to survivors. Washington, DC: American Psychological Association.

48.

United States Census Bureau. (2014). State and county quickfacts: Alaska. Retrieved from http://quickfacts.census.gov/qfd/states/02000.html

49.

United States Department of Justice. (2014). Crime in the United States, 2012. Retrieved from http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2012/crime-in-the-u.s.-2012/violent-crime/violent-crime

50.

Weisheit

R. A.

Wells

L. E.

Falcone

D. N.

(1995). Crime and policing in rural and small-town America: An overview of the issues. Retrieved from https://www.ncjrs.gov/pdffiles1/Digitization/154354NCJRS.pdf

51.

World Health Organization. (2003). Guidelines for medico-legal care for victims of sexual violence. Retrieved from http://apps.who.int/iris/bitstream/10665/42788/1/924154628X.pdf

52.

Young

W. W.

Bracken

A. C.

Goddard

M. A.

Matheson

, & the New Hampshire Sexual Assault Medical Examination Protocol Project Committee. (1992). Sexual assault: Review of a national model protocol for forensic and medical evaluation. Obstetrics & Gynecology, 80, 878-883.