Attachment and Health Care Relationships in Low-Income Women with Trauma Histories: A Qualitative Study

Abstract

Few studies have examined the relationship between low-income, traumatized women and their health care providers. In this study we interviewed 23 women from primary care and social service settings for the underserved about trauma, attachment, psychiatric symptoms, and reports of their interactions with primary care providers. Nearly all reported trauma exposure, and 17% had current posttraumatic stress disorder. About half were categorized as Unresolved with regard to attachment state of mind. Analyses of a health experiences interview showed that women with Unresolved attachment reported significantly more negative interactions with providers. Attachment may play a role in the relationship between trauma and health care interactions with providers, indicating the need for further study of this relationship and suggesting intervention strategies to help both parties contribute to a more collaborative process.

Women experience substantial exposure to traumatic experiences, including physical and sexual violence (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). These interpersonal trauma experiences are linked with emotional problems, depression and anxiety disorders (including posttraumatic stress disorder [PTSD]), and interpersonal and relationship problems (e.g., Green, 1994; Roth, Newman, Pelcovitz, Van der Kolk, & Mandel, 1997). Trauma and its associated outcomes may be particularly relevant to the health care system because rates of interpersonal trauma are high among primary care patients (McCauley et al., 1997; Walker et al., 1999), with rates of current PTSD in primary care samples ranging up to 23% (Liebschutz et al., 2007; Stein, McQuaid, Pedrelli, Lenox, & McCahill, 2000). Trauma exposure has been linked to decreased routine or preventive health care (e.g., Rheingold, Acierno, & Resnick, 2004), and trauma and PTSD predict negative physical health outcomes and increased health care costs (Walker et al., 1999). These problems and outcomes seem likely to affect not only the health of traumatized individuals but their relationships with their health care providers. Because of the link between trauma, psychiatric disorders, and relationship problems, attachment states of mind may provide a lens through which to understand relationships between primary care providers (PCPs) and their patients.

Provider–patient relationships, although not intimate in and of themselves, are very important to patients and require collaboration in order for medical treatments to be effective. It is likely that the PCP–patient relationship reflects similar dynamics to those that patients display in their relationships more generally (Alexander, 1992; Dozier, Stovall, & Albus, 1999). Furthermore, patients often want or need to talk with their physicians about their intimate relationships (e.g., sexual problems, domestic violence), in part because patients with mental health concerns prefer to be treated in primary care (Katon, Unützer, Wells, & Jones, 2010), and among low-income uninsured populations in particular, specialty mental health care is rarely available. Although a PCP may not need to know a particular patient’s attachment classification, having a better understanding of how attachment may influence doctor–patient relationships could help provide PCPs with generalized information that they may find useful in dealing with patients who may be puzzling or difficult or those who display contradictory behaviors. If PCPs feel that patients are not “doing their part,” the PCPs may become dismissive and, at the extreme, provide lower quality care. If patients feel that providers are dismissive, rude, or uninterested, this may affect their attendance at follow-up visits or adherence to suggested treatments. Patients who feel uncomfortable with their providers may be less likely to provide important information or call when they have a problem with a medication, and they may be more likely to wait until a crisis has developed to seek care. It therefore seems useful that the PCP have a general understanding of what may be driving some patient behaviors.

State of mind regarding attachment refers to the quality of the adult’s discourse when discussing attachment issues. Assessments of attachment state of mind usually use the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1996). The three traditional states of mind assessed in the AAI are Autonomous/Secure, Dismissing, and Preoccupied. Briefly, adult attachment states of mind are categorized as Autonomous/Secure when they present a coherent, consistent picture of relationships and relationship influences; as Dismissing when they idealize one or more attachment figures and/or show lack of memory for attachment experiences; and Preoccupied when discourse is characterized by angry involvement or is rambling. A fourth attachment state of mind, explicitly linked to trauma exposure, is used if the interview shows signs of unresolved experiences of trauma. It is called Unresolved and is superimposed on the three main classifications. It represents a breakdown of strategy and is characterized by lapses in reasoning or discourse when discussing abuse or loss (Hesse, 1999). Examples of such lapses would include things like falling silent in the middle of discussing loss or trauma, then shifting abruptly to another topic, or discussing people who have died in the present tense or as if they are simultaneously both dead and alive (Lyons-Ruth & Jacobvitz, 1999).

Multiple studies have shown a relationship between exposure to violence and trauma and less secure or Unresolved attachment states of mind (Aspelmeier, Elliott, & Smith, 2007; Bakermans-Kranenburg & van IJzendoorn, 2009; Roche, Runtz, & Hunter, 1999; van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999). By trauma we (and other authors) refer to the types of experiences that are associated with the development of PTSD, such as sexual assault and abuse, physical assault and abuse, and violent loss. Psychopathology in general, and anxiety disorders specifically, are associated with insecure and Unresolved attachment (Bakermans-Kranenburg & van IJzendoorn, 2009; Dozier et al., 1999; Fonagy et al., 1996). van IJzendoorn and Bakermans-Kranenburg (1996) showed that 40% of the clinical samples in their meta-analysis were classified as Unresolved and only 8% as Secure. An at-risk sample of adolescent mothers showed an association between childhood abuse and Unresolved attachment on the AAI, which in turn was associated with more dissociation, self-identity confusion, and relationship problems (Bailey, Moran, & Pederson, 2007). Unresolved attachment is associated with both borderline personality disorder and dissociative disorders (Slade, 1999). Unresolved and Dismissing classifications are also overrepresented among those with backgrounds of low socioeconomic status (van IJzendoorn & Bakermans-Kranenburg, 1996).

Dozier and Bates (2004) noted that treatment relationships often function as attachment relationships. Bassuk, Dawson, Perloff, and Weinreb (2001) found that poor women with interview-diagnosed PTSD, compared to those without PTSD, were significantly more concerned that they did not get good medical care, trusted their doctors less, found medical staff to be more rude, felt that staff did not understand their problems, and felt they had to wait too long for appointments. These findings are consistent with studies that indicate that women who have experienced interpersonal violence may perceive their providers as less understanding and less respectful and report being more dissatisfied with their care (e.g., McNutt, van Ryn, Clark, & Fraiser, 2000; Yam, 2000). These studies suggest the possibility that trauma exposure and PTSD may affect relationships with providers in negative ways.

Our aim in this exploratory study was to examine the reported health care experiences of low-income women with trauma and the impact of attachment state of mind on the relationships that women described with their health care providers. Unresolved attachment has been shown to have the strongest and most consistent link with trauma, the focus of our research program, so this specific state of mind was the most critical to highlight. We hypothesized that women with Unresolved attachment would report more negative experiences with their providers than would other women.

METHOD

Participants

Twenty-three women were recruited in the Washington, DC, metropolitan area from three sites: a community primary care practice that serves as a training site for family medicine residents, a private nonprofit community organization that provides individuals with comprehensive social services, and a transitional housing program for homeless low-income families. The Georgetown University Institutional Review Board approved the study, and all of the women provided informed consent.

Most of the women (78%) were African American, 13% were White, and 9% were of other ethnicities. About half of the women (56%) were divorced or separated, 35% had never been married, and 9% were married or cohabiting. Most had a high school education (74%). The mean age ofthe sample was 42.4 (range = 22–58 years, SD = 8.87). Moreover, 43% of the participants were working; 39% were unemployed; and 17% were home-makers, students, or retirees. Regarding insurance, 22% were uninsured and 78% had Medicaid or a city-funded health insurance program; 83% received government assistance (e.g., food stamps, Supplemental Security Income).

Measures

Stressful Life Events Screening Questionnaire (SLESQ; Corcoran, Green, Goodman, & Krinsley, 2000; Goodman, Corcoran, Turner, Yuan, & Green, 1998)

The SLESQ was used to assess history of exposure to 13 traumatic events covered in the Diagnostic and Statistical Manual of Mental Disorders and described in behavioral language. It has good test–retest reliability for number of events reported and good correspondence between the questionnaire and interview versions.

PTSD module from the Structured Clinical Interview for DSM IV Axis I Disorders–Non-Patient Edition (SCID; First, Spitzer, Gibbon, & Williams, 1996)

The SCID assesses Axis I diagnoses and is based on the Diagnostic and Statistical Manual of Mental Disorders (4th ed.). Following a short discussion to choose the most stressful event from the SLESQ, the SCID PTSD module for past and current diagnosis was administered, referring to a particular event. In a prior study of college women, we found a kappa of 1.00 for the interrater reliability of lifetime PTSD using the SCID (Green et al., 2005).

Patient Health Questionnaire (PHQ; Spitzer, Kroenke, & Williams, 1999)

Current depression was assessed via self-report using the PHQ. Questions take the form of “Over the last 2 weeks, how often have you been bothered by any of the following problems?” and include symptoms such as “little interest or pleasure in doing things” and “feeling down, depressed, or hopeless.” There is good agreement between PHQ diagnoses and those of independent mental health professionals (for the diagnosis of any one or more PHQ disorder, kappa = 0.65; overall accuracy = 85%; sensitivity = 75%; specificity (George = 90%).

AAI et al., 1996)

The AAI assesses states of mind with regard to attachment, which refers to the conscious and unconscious rules individuals have developed for organizing attachment-related experiences, feelings, and thoughts. Interviewees are asked to describe relationships with parents during childhood, recall specific memories, and conceptualize relationship influences. The reliability of the AAI categories, described earlier, is high over time (78% were categorized in the same main category after 2 months) and across interviewers (no effects), although the Unresolved category is somewhat less stable (61% were categorized in the same main category after 2 months; Bakermans-Kranenburg & van IJzendoorn, 1993). Distributions of categories are relatively independent of language, country of origin, culture, and socioeconomic status (Bakermans-Kranenburg & van IJzendoorn, 2009; van IJzendoorn & Sagi, 2008). Because of the focus on trauma in this study, we emphasize the comparison of the women who were categorized as Unresolved compared to the rest (Not Unresolved).

Healthcare Experiences Interview (HEI)

This semistructured, qualitative interview was designed for the study. It began as follows: “Now I would like to learn more about your experiences in the health care system, with your doctors, nurse practitioners, nurses, and other staff people, and what it has been like seeking and getting help for your physical problems.” It covered positive and negative interactions with health care providers, preferences for providers, discussions about trauma with providers, perceived links between trauma and health, and suggested improvements to the system. The goal was to encourage women to talk freely about their health care experiences, so examples were sought, and women were encouraged to tell the interviewers anything not covered that might be relevant. For example, the question about positive experiences was as follows:

I would like for you to think about some of your experiences with doctors, nurse practitioners, nurses, or other health providers in the past year or so. Can you tell me about a good discussion or experience you have had in the past year or so, one that you felt positive about? (Describe) What made it feel/seem like a positive experience for you?

Procedure

Signs were posted in the above settings asking English-speaking women to volunteer for individual research interviews. The first and second authors (Bonnie L. Green and Stacey I. Kaltman) conducted 3-hr interviews with women in their respective health care or service setting after author Mary Dozier, an expert in the AAI who was trained and certified by the developers, trained both to conduct the AAI. The AAI and the HEI were audio-recorded and fully transcribed by a transcription service. The AAI transcripts were sent to Dr. Dozier for coding.

Data Analysis

HEI coding

The HEI was approximately 90 min in duration. We conducted a qualitative analysis of the transcripts using a grounded theory approach and multiple iterations of data review and coding (Glaser & Strauss, 1967). A qualitative software program, ATLAS.ti, was used to sort and organize narrative statements and later to facilitate the coding of the data. Members of the research team served as initial coders and started by reading through several transcripts looking for general themes. The team then discussed and developed tentative codes based on their discussion of the themes that emerged across the interviews. They then looked at whetherthe tentative codes provided adequate coverage by reviewing the content of additional transcripts. Additional review and discussion by team members led to further refinement of codes. Once the team formed a consensus on the final coding, two members of the team each analyzed all HEI transcripts using these codes. The codes assigned by the two raters were compared, and a third member of the team resolved any coding differences.

Our intent was to examine the negative statements made by the participants about their health care providers as a proxy for difficult relationships because we did not have recordings of interactions between the patients and providers and so ultimately could not know what actually took place. However, because the participants were asked about both negative and positive health care experiences, we examined both. Although the negative and positive statements about providers occurred most often during the portion of the interview when we specifically asked for these experiences, they sometimes arose in other parts of the interview; we coded them regardless of when they appeared.

The coding of the open-ended questions resulted in codes that described positive or negative statements about relationships with providers or the health care system. These codes were subsequently combined into larger categories that were used for final analyses. The negative statements that fell into the areas of incompetence and active rudeness of the provider were combined to compose a Negative–Active category. Passively being ignored or overlooked and criticisms of the system (e.g., long waits, insurance problems) were combined into a Negative–Passive category. The positive statements in the areas of communication, expressions of concern, and relationship with provider were combined into a Positive–Relationship category, whereas competence of provider and informativeness of provider were combined into a Positive–Technical category.

The numbers of statements made by each woman for each of the four summary categories and for the total number of codes were compared between women who received attachment classifications of Unresolved versus Not Unresolved using chi-square analysis and independent samples t tests. All statistical analyses were carried out using SPSS 12.0 for Windows XP.

RESULTS

All but one woman reported having experienced a traumatic event on the SLESQ. On average the women endorsed 4.74 (SD = 2.42) types of trauma exposures. Exposures included child abuse/molestation (44%), rape/attempted rape (44%), domestic violence (39%), robbery (26%), traumatic loss (35%), life-threatening illness (30%), and other (27%). Lifetime PTSD was identified in 30.4% of the women, whereas current PTSD waspresent in four women (17%). Current depression by self-report was present in 22% of the women.

Attachment

The classification of Unresolved attachment described more than half of the sample (52%; n = 12). In the three-way breakdown, 30% (n = 7) of the women were categorized as Autonomous, 44% (n = 10) were Preoccupied, and 22% (n = 5) were Dismissing. One subject could not be classified. She was included in the Not Unresolved group for analyses because she did not meet criteria for the Unresolved classification. Neither PTSD nor depression was associated with Unresolved attachment.

Perceptions of Health Care

According to responses given in the HEI, fewer than half of the women (39%) felt that their provider was meeting their needs. Only about one quarter of the women (26%) had discussed their trauma history with a provider. Of those who had not had such a discussion, only one quarter to one third (5 women; 28%) wished to do so. In response to the open-ended questions about relationships with health care providers, most of the women (74%) made both positive and negative statements. All but three women (87%) made negative statements, and all but three (87%) made positive statements (different women). Examples of the different types of statements can be found below.

Negative Passive Statements

More than 60% of the sample (61%) made negative passive statements about being passively ignored or overlooked. In general these statements referred to experiences in which the patients felt that their concerns were not taken seriously, there was a lack of responsiveness or follow-up on the part of providers, or the providers were behaving in a manner that lacked empathy or sensitivity. Some examples include the following:

They didn’t take care of that. And because I was also trying to get assistance, public assistance, paperwork that needed to be filled out by the doctor, they didn’t fill that out. I just became ignored.

… here I am, waiting for results. You don’t communicate. You don’t call, and breast cancer runs on my mother’s side of the family and you are not telling me anything.

It was kind of a nonchalant attitude and my feelings were hurt and I was angry.

A similar proportion of participants (57%) made negative passive statements that referred to criticism of the health care system. These statements referred to frustrations within the clinic setting, such as long wait times, limited time spent with the PCP, and a lack of continuity in providers between visits. Participants also described concerns regarding the larger health care system, including difficulties accessing or maintaining health care coverage and the bureaucracy involved in getting adequate care from various providers and organizations who have limited communication between them.

That’s the most difficult part for me. You wait and wait and wait. I think that there is somewhere in the book that you just supposed to wait… . Not just saying everybody has to be quick …, but certain things should not take as long… . And time is precious to some people, you know, they don’t have time to just wait. Not saying that I’m above anything … that’s just the way it is.

It seems like sometimes when you are at the clinic they look you over real quick … and say, do you have enough of your medicine. And they would write you another prescription real quick so you could go. And it don’t seem like … you get as close to the doctor as you would in a regular situation, if you had insurance.

Negative Active Statements

More than half of participants (52%) made negative active statements related to incompetence. These statements described the patients’ perceptions that they received unnecessary procedures, were prescribed the wrong medications, and were misdiagnosed. They also described providers who made medical errors and used inadequate infection control precautions.

Out at the C Hospital, I went for a mammogram and they kept talking about “this breast this,” and I tell them I’m having pains in this [other] breast; and they wanted to do a biopsy on this one, and I said no, you need to do it on this [other] one. So I said to hell with it. So I didn’t go back.

… I ended up having a partial hysterectomy, which I know today did not have to take place, but it was because of the type of health care that I was getting.

I was diagnosed with something that I don’t have. Come on now. That was very irresponsible.

More than one third of participants (39%) made negative active statements about experiences in which they were treated rudely or poorly. These experiences involved clinic staff at all levels, including receptionists, nurses, PCPs, and specialty providers.

… the people they send you to talk to are supposed to be nurses. They are supposed to help you find a primary care. They talk to you like you are stupid. Like because you are on this [Medicaid] you are some sort of low life, … and I don’t like that. Because no matter what I have been through, I ain’t no dummy… . And you are supposed to help me.

If I had a question, they treated me like I was an idiot for asking.

There are some women that feel like anything that happened to you of the sexual nature you deserved it … She was supposed to be a social worker, and you would think she would know better but her response to me was that sometimes … these things happen to them, more so than if they were really paying attention to what was going on.

Positive Relationship Statements

More than one third of participants (39%) made positive relationship statements regarding their relationship with the provider. The participants spoke explicitly about their relationship with their PCPs, suggesting that there was trust and warmth that had, in some cases, taken time to build. They mentioned providers who asked them about their lives and some who even shared about their own lives as a way to connect with them.

… I have been coming here now for like, about two years. And I feel that I have established a rapport with Dr. X, and that she knows what’s going on with me, and I can talk to her about different things.

Dr. X is very positive. She talks to you. She wants to know about your personal life and what’s going on with you.

A similar proportion of participants (38%) made positive relationship statements about communication. The participants most frequently described encounters in which they felt comfortable talking with their PCPs and when they felt that their PCP really listened.

One of my primary care doctors I was pretty open with about my background, that I didn’t have to be open with her about… . That wasn’t her field, but she was a good listener. I will put it that way… . It was just good to talk to someone.

She’s a good doctor because she takes the time to listen to you. So it’s like her medical side and then she also has this warm, friendly, down-to-earth type of personality. I pretty much feel comfortable talking to her …

More than one third of participants (35%) made positive relationship statements related to expressions of concern. The participants described encounters in which providers and other clinic staff showed care and concern for them through their words and behaviors.

After she took my vitals she did stuff, how are you, she let me know she was more concerned doing my physical… . But when you go a step further and say “how are you feeling, how are you doing, have you been depressed lately,” that says to me that they care about your overall well-being.

He was really nice. He would always make jokes. He was funny. And he seemed like he really cared. He was really sensitive.

Positive Technical Statements

Almost half of the sample (48%) made positive technical statements regarding the competence of the provider. Participants described many qualities that their PCPs displayed as evidence of their competence, including being very observant, thoroughness, professionalism, technical skill, and using innovative treatments.

I have had older doctors and to me they were the best doctors. People who like to look at you when you walk through the door and they can begin to tell you what your problem is… . I will never forget this lady; she was fantastic.

They examined me so well. They take their time and examine you real well.

More than 20% (22%) of participants made positive technical statements about the informativeness of the provider. The participants described encounters in which their providers provided them with information in a way that they could understand it, which often led to increased feelings of comfort.

After I was admitted, they all came to my room. And that’s when they explained to me like, maybe the next day, that what was going on was, after they took an [electrocardiogram] and let me see the results of it, then they brought another machine that actually took a picture of my heart. And then that’s when they explained it.

… he brought out the medical books and showed me pictures of how it can look, what it can progress to. He was very informative and I felt really good after leaving there, knowing that somebody opened up and explained to me what I’m going through … And I felt really good getting that information. Not knowing was … scary.

Attachment and Health Care Perceptions

Unresolved attachment on the AAI (compared to ratings of Not Unresolved) was significantly related to total number of negative statements about health care, t(21) = 2.15, p < .05 (Not Unresolved, M = 1.63, SD = 1.12; Unresolved, M = 2.58, SD = 1.00). Positive statements showed a trend only when we compared the two groups, t(21) = 2.04, p = .054 (Not Unresolved, M = 1.36, SD = 0.92; Unresolved, M = 2.25, SD = 1.14). Because both positive (trend only) and negative statements were related to attachment state of mind, the total number of positive plus negative statements was also related, t(21) = 2.56, p < .02 (Not Unresolved, M = 3.00, SD = 1.67; Unresolved, M = 4.83, SD = 1.75).

DISCUSSION

This study was designed to explore the reported health care experiences of low-income women with trauma histories, their attachment states of mind, and the relationship between the two. We focused on Unresolved attachment because of its link with trauma, and indeed more than half of the sample was categorized as Unresolved. We hypothesized that women with Unresolved attachment would report more negative experiences with their providers than those who were Not Unresolved. As hypothesized, these women reported more negative encounters than women with Not Unresolved states of mind. Although the overall relationship between attachment and positive statements did not reach significance, there was a trend, that may have been significant in a larger sample, for Unresolved women to make more positive statements as well.

We found very high rates of interpersonal violence in the sample. This is not surprising given that we were recruiting in settings that served the uninsured, where trauma rates are high, and given the association between trauma exposure and economic status (e.g., Breslau, Davis, & Andreski, 1995; Kessler et al., 1999). Furthermore, we advertised the interview as pertaining to very stressful experiences, so individuals with such experiences may have been more likely to volunteer. Rates of the psychiatric disorders we assessed were also much higher than in the general population (e.g., Kessler et al., 1994, 1995) but consistent with estimates in primary care (Liebschutz et al., 2007; Stein et al., 2000), where most of the women were recruited.

Attachment states of mind in the study population were distributed differently than those reported in a normative sample. In their meta-analysis with the AAI, van IJzendoorn and Bakermans-Kranenburg (1996) reported that 55% of normal mothers across studies were classified as Autonomous in four-way classifications, 16% as Dismissing, 9% as Preoccupied, and 19% as Unresolved. In their overview paper of the first 10,000 AAI interviews, Bakermans-Kranenburg and van IJzendoorn (2009) reported that across all of the clinical and at risk samples, the rate of Unresolved attachment was 38% in the four-way distribution. In our sample 52% of the women were classified as having Unresolved attachment. The higher rates in our sample likely related to high rates of violence exposure, lower social class, relatively high rates of lifetime psychiatric disorder, and recruitment in primary care, all factors that predict Unresolved attachment.

Whereas Stovall-McClough and Cloitre (2006) found that those with a diagnosis of PTSD had Unresolved classifications more frequently than those without PTSD, and several studies of PTSD and self-reported attachment found that insecure attachment was associated with higher levels of PTSD symptoms (Bakermans-Kranenburg & van IJzendoorn, 2009), we were not able to link PTSD or depression to Unresolved attachment, possibly because of our small sample.

The fact that Unresolved women also reported more positive statements was unanticipated but fits with a style that is characterized by cognitive and affective disorientation, confusion, and dissociation (Hesse, 1999). Indeed the key defining characterization of this classification is that there is no consistent strategy, so changes of strategy would be the norm, unlike in individuals with consistent but insecure attachment. For example, Dozier (1990) suggested that patients classified as Dismissing typically appear to care providers as if they are invulnerable, whereas those with Preoccupied attachment may impress physicians as needy and dependent. At present there is little research or clinical description that helps us to understand how women who are classified in an interview as Unresolved (based on their discussions of loss and trauma) actually behave on a day-to-day basis with providers and others, so we are left to hypothesize about this and add our findings to this literature. Slade (1999) described interactions with Unresolved patients in a mental health encounter and noted the difficulty of working with these patients because of their dissociated and distorted affect, resulting in a slow and painstaking recreation of what may have happened in their earlier lives. Authors using somewhat different classification systems based on self-reported attachment have described patient behavior in individuals with “fearful” or “dismissing” classifications, both of which areroughly comparable with the Unresolved classification used in our study. Thompson and Ciechanowski (2003) described “fearful” patients as being mistrustful of themselves and others when attempting to cope with distress, and therefore they present a double message of help seeking and help rejecting when they are threatened with illness. Hunter and Maunder (2001), describing “disorganized” attachment, discussed the likelihood of incoherence in the patient, with histories difficult to obtain and hard to sort out. Patients may want staff to respond to them but not have faith that they can do so. These authors warned that such patients’ attitudes can generalize and that staff may therefore end up blaming each other or becoming very discouraged. Staff may be pulled between wanting to rescue the patients and wishing they would leave.

Although we were not able to establish that the women in this study who endorsed both positive and negative statements were reporting about the same provider, that seems a strong possibility given the lack of consistent strategy that is associated with the classification and its strong association with borderline disorders. A style that shifts from positive to negative and back again is likely to be, at the least, confusing and discouraging to a PCP (Thompson & Ciechanowski, 2003). In our study of PCPs and their encounters with trauma patients (Green et al., 2011), the most often mentioned “difficult” patient was the help-seeking but noncompliant/help-rejecting patient, one who, for example, “calls almost every day and leaves a message on my voicemail for this or that or the other, but doesn’t show up for the regular appointment” (Green et al., 2011, p. 39). Even if PCPs understand that there is a problem they may feel unprepared to respond appropriately (Green et al., 2011). PCPs may become frustrated about engaging the patient and may respond by behaving in a less empathic way. Alternatively, patients may just perceive them that way.

We do not argue here that providers need to know the attachment style of patients in order to best interact with them. We have attempted to provide some information about how patients with disturbed attachment might perceive their providers so that we could understand how attachment might manifest in these relationships. Our findings provide information relevant to developing educational programs for PCPs to give them a better perspective on how trauma patients may be challenged in their capacity to interact in a clear and straightforward way as a full partner in the treatment relationship. Such knowledge may provide a broader perspective for the PCP that helps him or her be more patient and flexible with these individuals.

These issues could be approached in several ways. One is to train PCPs about trauma and its effects on mental health; the range of ways in which individuals with trauma backgrounds, disturbed attachment, and mental disorders may present and perceive PCPs; along with practical suggestions for how better to support and engage these individuals in collaboration. Based in part on the present study and the one mentioned above with PCPs, our research group recently adapted a training for mental health professionals working with trauma patients to be appropriate for PCPs, who see patients for a much shorter time (Green et al., 2010). This training engages the PCP in taking the perspective of the patient, appreciating the myriad effects of having a serious trauma history, and thinking of new approaches to difficult patients based on this knowledge, with specific examples and suggestions. To date this approach appears promising based on ratings of trained PCPs’ interactions with standardized patients and on feedback from their actual patients (Green et al., 2010). Screening for trauma exposure and symptoms of mental disorders may give PCPs a “heads up” to potential concerns and mental health symptoms. Patient education about trauma and mental health may also help patients to appreciate how they may have been affected by earlier experiences. And finally, although we are not aware of any such interventions that focus on trauma, patient empowerment interventions that help coach patients about how to approach their doctors and how to focus on the issues that are most important to them for a particular visit would likely help patients who may typically be unfocused or confused to make better use of their time with the provider (Roter, Stashefsky-Margalit, & Rudd, 2001).

This study has a number of limitations, so the findings should be viewed with caution. Our sample does not represent a typical primary care population, even one in a low-income setting. Thus, rates of psychiatric disorders and the attachment category distribution may not be representative. The small sample size was associated with reduced power to find significant differences. The sample also had very high rates of violence, and lack of variability on exposure to violence may have attenuated individual differences. The system-related problems that the women described indeed may be real system-level problems that attenuate individual differences. Likewise, rude or other negative behaviors by PCPs, if they did actually occur, may or may not be attributable to patient behaviors. We make no assumptions about whether the encounters described by the women actually happened or that they happened as reported. Our goal was to understand women’s perceptions of relationships with their health care providers, because we assume that these perceptions drive behavior and emotional reactions to the providers and the settings.

In conclusion, attachment state of mind appears to be related to perceptions of health care experiences, suggesting that attachment may be a useful construct to help frame some thinking about intervening with trauma patients. Studies going forward could explore this potential link in more depth. Adding recordings and scoring of actual dialogue between doctors and patients, along with the assessment of attachment states of mind, would help to clarify whether these negative/mixed perceptions are grounded in actual encounters. In either case, provider and patient education and training may help improve the quality of care for trauma patients. To the extent that women feel mistreated in their medical care, regardless of the reality, there is room to intervene. Patients with Unresolved attachment may be experienced as difficult or puzzling, and providers may need relevant information about trauma, attachment, and mental health problems, as well as specific practical suggestions for how to interact with these patients, to minimize relationship struggles and negative interactions and maximize the extent to which patients participate positively and actively in their own health care.