Nội dung

Health and Quality of Life Outcomes BioMed Central Open Access Research Beyond satisfaction: Using the Dynamics of Care assessment to better understand patients' experiences in care Bruce Rapkin1, Elisa Weiss*1, Rosy Chhabra3, Laura Ryniker1, Shilpa Patel1, Jason Carness1, Roberto Adsuar1, Wendy Kahalas2, Carol DeLaMarter2, Ira Feldman2, Judy DeLorenzo2 and Ellen Tanner2 Address: 1Department of Psychiatry and Behavioral Sciences, Memorial Sloan-Kettering Cancer Center, New York, USA, 2AIDS Institute, New York State Department of Health, Albany, USA and 3Department of Pediatrics, Albert Einstein College of Medicine, Bronx, USA Email: Bruce Rapkin - rapkinb@mskcc.org; Elisa Weiss* - weisse1@mskcc.org; Rosy Chhabra - rchhabra@aecom.yu.edu; Laura Ryniker - RynikerL@mskcc.org; Shilpa Patel - patels12@mskcc.org; Jason Carness - carnessj@mskcc.org; Roberto Adsuar - adsuarr@mskcc.org; Wendy Kahalas - wxk07@health.state.ny.us; Carol DeLeMarter - cmd13@health.state.ny.us; Ira Feldman - isf01@health.state.ny.us; Judy DeLorenzo - jpd07@health.state.ny.us; Ellen Tanner - etv02@health.state.ny.us * Corresponding author Published: 10 March 2008 Health and Quality of Life Outcomes 2008, 6:20 doi:10.1186/1477-7525-6-20 Received: 12 July 2007 Accepted: 10 March 2008 This article is available from: http://www.hqlo.com/content/6/1/20 © 2008 Rapkin et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Abstract Background: Patient perceptions of and satisfaction with care have become important indicators of the quality of services and the relationship of services to treatment outcomes. However, assessment of these indicators continues to be plagued by measurement problems, particularly the lack of variance in satisfaction data. In this article, we present a new approach to better capture patient perceptions of experiences in care, the Dynamics of Care (DoC) assessment. It is an in-depth approach to defining and assessing patients' perspectives at different junctures in care, including their decisions about whether and where to seek care, the barriers encountered, and the treatments and services received. Methods: The purpose of this article is to describe, validate, and discuss the benefits and limitations of the DoC, which was administered as part of a longitudinal study to evaluate the New York State HIV Special Needs Plan (SNP), a Medicaid managed care model for people living with HIV/AIDS. Data are from 426 study respondents across two time points. Results: The results demonstrate the validity and value of the DoC. Help seeking decisions and satisfaction with care appear to be situation-specific, rather than person-specific. However, barriers to care appear to be more cross-situational for respondents, and may be associated with clients' living situations or care arrangements. Inventories in this assessment that were designed to identify potential deterrents to help seeking and difficulties encountered in care demonstrated clear principal component structures, and helped to explain satisfaction with care. The problem resolution index was found to be independent from satisfaction with care and the data were more normally distributed. DoC data were also associated with subsequent utilization and change in quality of life. Conclusion: The DoC was designed to be a flexible, integrated measure to determine individuals' salient service needs, help seeking and experiences in care. One of the many strengths of the assessment is its focus on specific problems in context, thus providing a more sensitive and informative way to understand processes in care from the patient's perspective. This approach can be used to direct new programs and resources to the patients and situations that require them. Page 1 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 Background There is mounting evidence that variations in perceived quality of health care among people with HIV/AIDS affect patient behavior, especially adherence to medication regimens and other physician recommendations, as well as health outcomes [1-9]. As a result, client satisfaction has become the most important direct feedback to providers on the quality of services and the relationship of services to treatment outcomes for many health care organizations [2]. Cherin et al. (2001) note that "patient satisfaction with services, as a cornerstone of quality care, has emerged as an important focus of managed care organizations (MCOs) over the last decade. Member dissatisfaction with services has been demonstrated to have a direct impact on plan disenrollment" (p. 105). Funding agencies have also placed an emphasis on clients' perceptions of care. The Health Resources and Services Administration (HRSA), in providing guidance to its Ryan White CARE Act grantees, has recommended that service planners and providers pay close attention to client satisfaction as they develop and implement their service delivery systems [10]. The emphasis in the private and public sectors on perceived quality of care has led health and managed care industry research organizations (e.g., Group Health Association of America, RAND Corporation, The Measurement Group) to invest considerable time and expense in developing patient care satisfaction instruments tailored to managed care organizations. These instruments tend to define satisfaction in this context as a patient's perception of the quality of physicians, access to services, communication with providers and administrative staff, and of the success of their treatment [3]. Despite widespread consensus that perceived quality of care, and patient satisfaction in particular, is important to measure, and despite much investment in instrument development, the assessment of patient satisfaction has been and continues to be plagued by critical measurement issues [11]. There is substantial debate about whether patient satisfaction can be measured reproducibly and meaningfully [12-16]. Ware's work (1978) highlights the bias in patient satisfaction questionnaires due to acquiescent response set (ARS), a tendency to agree with statements of opinion regardless of content. More than twenty years after Ware's seminal article discussed this problem, researchers are still encountering negatively skewed distributions in satisfaction measures [3,15]. Moreover, Ware (1978), in his review, found that biases were greatest for groups reporting lower educational attainment or less income, which makes this problem of paramount concern in the study of quality of care for many segments of the HIV/AIDS-affected population. The literature also suggests that when we look at satisfaction with managed care http://www.hqlo.com/content/6/1/20 programs among patients with HIV/AIDS, men consistently rate such programs higher than women [3], and that attitudes about what aspects of care delivery are most important vary by gender and race/ethnicity [4]. Several studies suggest that health status per se, rather than degree of improvement in health status due to medical care, also influences satisfaction ratings; there is some evidence that healthier patients tend to report greater satisfaction with health care [11,14,17]. Statistical adjustments can be made to deal with sociodemographic or health status confounds that may occur when trying to predict patient behaviors such as adherence. However, if the purpose is quality improvement, adjustment hides what may be important problems and hinders the development of innovative solutions for different patient groups [14]. In HIV care, studies suggest that it may be particularly important to understand how different patient groups experience care and what people with different backgrounds value, so that interventions and changes in care delivery practices can be appropriately tailored [2,4,7,12,18]. Given the pitfalls of satisfaction measurement and its debatable value for quality improvement, there has been a call for new approaches to the assessment of patient perceptions of care. A number of researchers have emphasized the need for increased attention to narrative, specifically patients' detailed accounts of what went well and what did not, as well as reports about what did or did not happen when patients received care, what they experienced in light of what they value, and whether needs have been met [4,5,8,14,19]. Cleary and Edgman-Levitan (1997) note that this kind of information tends to better reflect the quality of care and is also more "interpretable and actionable for quality improvement purposes" (p. 1608). This is particularly the case when looking at the quality of care for patients with complex needs and many contacts with providers, since these patients are likely to perceive different aspects of care differently (e.g., a patient with HIV may be satisfied with help with drug use but not with the support provided for medication adherence). Cleary and Edgman-Levitan (1997) have shown that people with complicated conditions do not describe their experiences in terms of single visits with one provider, but in terms of episodes of care, and these authors call for developing ways of collecting quality information for entire episodes of care (p.1611). These concerns led us to develop a novel approach to the assessment of health service use and quality, which we refer to as the "Dynamics of Care" assessment. The purpose of this article is to describe, validate, and discuss the benefits and limitations of the Dynamics of Care assessment, which was developed for a longitudinal evaluation Page 2 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 study called "Choices in Care." The study was designed to evaluate New York State's Special Needs Plan – a comprehensive care model created for people with HIV and their children that was developed as an alternative to both Medicaid fee-for-service (FFS) and to Medicaid managed care [20]. The larger Choices in Care evaluation examines access to care and quality of care as well as patient reported outcomes such as treatment adherence, risk behaviors, and quality of life. In this paper, we focused solely on the validation of the Dynamics of Care assessment. Methods Sample Individuals were eligible to participate in the Choices in Care study if they enrolled in one of the Special Needs Plans (SNPs) in the prior three months, or if they received care through the traditional Medicaid FFS system. Enrollment in a SNP is voluntary, and the majority of people with Medicaid in New York City who are HIV+ are still receiving care through the FFS model. Thus, we oversampled SNPs enrollees to facilitate comparison with FFS. New SNP enrollees and FFS recipients were eligible for the study if they were 18 years of age or older, spoke either English or Spanish, resided in New York City, had the cognitive capacity to complete the informed consent and a 1 hour interview, and were not incarcerated at the time of the recruitment call. The Special Needs Plan model began in 2003. To date, the three active plans have an enrollment of over 2000 individuals. Study recruitment began in September, 2003, following approval from the Institutional Review Boards at Memorial Sloan-Kettering Cancer Center and the New York State AIDS Institute; recruitment concluded in January, 2007. We recruited new SNP enrollees within 45–90 days of their enrollment into one of the three plans, in order to get baseline information about their health status upon entering the plan, before they had an opportunity to make use of SNP services. New SNP Enrollees were identified from a roster of names provided monthly by the New York State Department of Health (NYSDOH) AIDS Institute. It is important to note that upon enrolling in a SNP, individuals were informed that they would be contacted for plan evaluation purposes. While all individuals had the opportunity to take part in the study, only a percentage were reached by phone and agreed to participate during the eligible timeframe (up to 90 days post enrollment). The comparison group of fee-for-service recipients was recruited concurrently as a convenience sample; they self-selected for the study by responding to flyers posted at Designated AIDS Centers, at community based organizations, and Adult Day Health Care programs. The majority of these recruitment sites provide care to HIV+ individuals under the fee-for-service model and http://www.hqlo.com/content/6/1/20 to people enrolled in one of the Special Needs Plans. For privacy reasons, it was not feasible to obtain a list of all HIV+ fee-for-service recipients from which to recruit. The purpose of the Choices in Care evaluation is to examine access to care, perceived quality of life, member satisfaction and patient reported outcomes among HIV+ adult medicaid recipients. Since the study's inception, 306 SNP enrollees and 322 FFS recipients have been recruited, for a total of 628 subjects. The study refusal rate among known eligible SNP enrollees was 12%. The majority of analyses reported here are based on data from 426 respondents who have completed the first and second (of five) study interviews. Procedures Study participants were interviewed five times, at three month intervals, in order to compare patient needs, access to care, quality of care, and patient outcomes across the SNP and FFS samples, and within the SNP sample across the three plans. The present validation study draws on data from the baseline and three-month interviews. The informed consent process occurred at the time of the baseline interview. Each interview took 30–45 minutes to complete and was conducted either in person at the study offices in New York City or on the phone, whichever was more desirable for the participant. Interviews were conducted in English or Spanish, depending on the participant's preference. In addition to transportation reimbursement, participants received $25 for the first interview, $30 for the second, $35 for the third, $40 for the fourth and $45 for completing the fifth. Measures Information about personal characteristics, including selfidentified gender, age, race/ethnicity, whether the respondent's primary language is English or not, and education level (high school graduate or not), was obtained in the baseline interview. The Choices in Care study was designed to examine impact of care on patient reported outcomes. Thus, outcome measures were collected at baseline and again at six months, including health status, treatment adherence, risk behaviors, and quality of life. The main Dynamics of Care assessment occurred at the intervening three month interview, with brief follow-up questions occurring during the six month interview, including questions about the current status and resolution of events in care identified at the three month time point (see discussion of measure below). The Dynamics of Care assessment The Dynamics of Care assessment was designed to capture patients' experiences of care in detail, for the purpose of evaluation and quality improvement. It is an in-depth approach to defining and assessing patients' perspectives Page 3 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 at different junctures during an episode of care, including their decision whether and where to seek care, the barriers they encountered, and the treatments and services they received. The Dynamics of Care is an interviewer administered assessment that begins by asking respondents whether they experienced problems in the last three months in nine specific areas ("trigger events") that the Special Needs Plan was designed to address. These areas include: 1) Adherence to Medical Instructions; 2) Medical Problems; 3) Specialty & Inpatient Hospital Care; 4) Preventive Health Care & Screening; 5) Sexual Risk Behavior; 6) Family Planning; 7) Psychological Symptoms; 8) Substance Use; and 9) Life Circumstances & Demands. This first part of the measure, the Events in Care Screening Questionnaire (ECSQ) contains 54 yes/no items and serves to identify the areas of recent concern to each patient [21]. Respondents are also asked to rate satisfaction with the services that are available to them in each of these nine areas on a scale from 0 (Completely Dissatisfied) to 10 (Completely Satisfied), regardless of whether they identify recent events in care in a given area. Satisfaction ratings are used to identify areas for further discussion and probing. The main part of the Dynamics of Care interview explores respondent experiences, with respect to three selected events in care. The interview is structured with skip patterns to probe relevant questions depending upon how far individuals have progressed in regard to seeking and obtaining care. For each of the three trigger events, respondents are asked to provide a brief narrative description of the concern. They then delineate whether or not they sought care and what factors influenced that decision, including provider recommendations. Among those who were or are actively seeking care, we ask about any barriers and delays that they have encountered. For those who have already started receiving services it is determined which providers they have seen; respondents are then asked about experiences in care including the outcome of referrals, coordination of care, and communication with providers. Respondents are also asked to rate satisfaction with help received from the main provider involved with their care pertaining to each event. In addition, overall experience of problem resolution is explored including whether or not they are seeking or receiving care, or trying to handle a problem on their own. If a participant mentions problems in more than three of the nine areas, three events are selected for further probing based on the area of greatest concern to the patient, the area that the patient rated highest in terms of satisfaction with care available, and the area the patient rated lowest in terms of satisfaction with care available. If more than one area receives an equally high or an equally low satisfaction rating, interviewers were instructed to select area(s) that http://www.hqlo.com/content/6/1/20 tended to be identified as concerns less frequently, in our initial piloting of the ECSQ measure (i.e., concerns related to sexual risk, family planning, and substance use). This procedure effectively oversampled infrequent concerns, which made it possible to conduct analyses related to less frequent areas of concern. In sum, the Dynamics of Care assessment guides respondents to identify key events related to their care, and to describe what has transpired regarding that event up to that point. By tracing specific episodes in this way, it is possible to understand discrete circumstances and cause-effect associations based upon each individual's immediate experiences. In essence, this approach trades the breadth and lack of specificity inherent in global satisfaction ratings for a high degree of precision in the assessment of narrowly framed but highly salient experiences. As we shall detail in description of analyses below, the Dynamics of Care measure may be examined with respect to specific areas of care or summarized across areas to create a variety of different summary variables. Health care • Number of health care providers and composition of health care team: Respondents were asked to identify all of the health care providers that were part of their current health care team or whom they had seen in the six months preceding the baseline interview for care. • Plan enrollment status: Respondents were classified as a member of one of the NYSDOH Special Needs Plans, or as a Medicaid fee-for-service (FFS) recipient. Health history and health status • Years since HIV diagnosis: Respondents were asked the month and year when they first tested positive for HIV. • CD4 Count: Measured as less than or equal to 200, or greater than 200, based on respondent report. • Undetectable viral load: Determined by a response of "undetectable" to the question "Do you know or remember the results of your last viral load count?" • HIV-related diagnoses: ⴰ Total number is based on respondents' indication of whether they had ever been diagnosed with any of 11 different opportunistic infections, disorders and malignancies (e.g., pneumonia PCP, Kaposi sarcoma,). ⴰ Number of conditions for which the respondent was not receiving treatment was determined by asking respondents to indicate whether the condition had been completely treated, whether they were being treated at that time, or were never treated. Page 4 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 • Number of symptoms: Respondents were asked if they had any of 11 different psychological and physical symptoms in the past four weeks. Examples of symptoms include trouble with thinking, concentrating, or memory; fever, chills, sweats. • Physical and psychological quality of life composites from the MOS-SF36: This widely used 36-item measure of health related quality of life and functional status yields physical and psychological quality of life composites, each scored from 0–100 with higher scores indicating greater wellbeing [22]. Risk history and current risk behaviors The respondents were asked their history and current activity (within the last 3 months) in a variety of risk behaviors including: • History of and current hard drug (heroin, crack, other illicit drug, and injected drugs) use • Sexual history- having a sexual partner who is an injection drug user; having a sexual partner who is HIV seronegative; whether or not the respondent was ever a sex worker; and unprotected sex behavior • Criminal justice involvement • Tobacco use • Excessive alcohol use Personal characteristics Variables include: self-identified gender; age; race/ethnicity; whether the respondent's primary language is English or not; education level (high school graduate or not); number of sources of income (0–9); monthly gross income from all sources; housing stability (whether or not the respondent lives in a place of his/her own); whether the respondent self-identified as gay or bisexual; whether the respondent had a spouse or partner; whether the respondent was living with a partner; number of other adults and minor children in the household; and whether the respondent was living with another seropositive person. Analysis Analyses were based on data from 426 respondents who completed the first two study interviews. All data were analyzed using SPSS V12.0 for Windows. In the results section that follows, we first present the characteristics of the sample, including demographic information, health status, health care utilization, and risk behavior. We next describe our identification and sampling of Events in Care for further assessment in the Dynamics of Care. Finally, http://www.hqlo.com/content/6/1/20 we present a series of analyses of the psychometric and response properties of each section in the Dynamics of Care, including stage of help-seeking, barriers and delays encountered while seeking help, and perceived quality of interaction with main providers. For each section of the measure, we give the item response rates and intraclass correlations across respondents' specific areas of care. We also provide data on inter-item reliability and the principal component structure of summary scores. Lastly, we determine whether Dynamics of Care measures were related to demographic and health-related measures through a series of multiple regression analyses. Results Sample characteristics Detailed demographic characteristics are outlined in Table 1. With respect to respondents' involvement with the health care system, 33% of respondents were enrolled in one of the Special Needs Plans; 77% received care through traditional Fee-for-Service. Nearly all had a primary medical care provider. Most people in the sample reported having at least one case manager (82%) and at least one medical specialist (76%) within the six months preceding our baseline interview. Forty-one percent indicated that they have seen a dentist in the past six months; 66% have seen a therapist, psychologist, or psychiatrist; 24% had seen a social worker or other patient supporter, such as a patient advocate; and 7% had a physical therapist, occupational therapist, or rehabilitation therapist on their health care team. Thirty-three percent of respondents had seen a dietitian, nutritionist, or exercise specialist in the past six months. With respect to risk behaviors, 75% of the sample has a history of crack, heroin, or other injection drug use; 23% had used one or more of these drugs in the past three months. Fifty-nine percent had sex with someone who injects drugs, and 18% have traded sex for money, drugs, or other needs. Nineteen percent of respondents reported having unprotected sex in the past three months, and 7% had unprotected sex with someone who is seronegative. Twenty-one percent had smoked one or more packs of cigarettes a day in the past three months, and 10% reported that they had too much alcohol in the past three months. Events in Care The first part of the Dynamics of Care assessment entails identifying and prioritizing events of interest to the respondent. As noted above, we developed the ECSQ, a face-valid measure including 54 items that probe needs in nine areas of concern. In each area, we asked whether people living with HIV/AIDS (PWHAs), had specific problems or concerns in the past three months, whether they needed additional information, and whether their providers had raised concerns. Detailed results of the validation Page 5 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 http://www.hqlo.com/content/6/1/20 Table 1: Sample Characteristics Total Sample N = 426 Gender Female Average age Race/Ethnicity Black Hispanic Primary language Spanish Education level High school diploma or GED Bachelor's or higher degree Sexual orientation Gay or Bisexual Have spouse or partner Living situation Live with spouse or partner Live with adult other than spouse or partner Live with minor children Live with an individual who is seropositive Unstable housing ( Does not have a place of their own) Reported working Involved in the criminal justice system (at some time) Average number of years diagnosed CD4 Count greater than 200 Undetectable viral load Average number of opportunistic infections Had 2 or 3 conditions Had 4 or more conditions Reported never receiving treatment for one or more conditions Average number of physical and psychological symptoms Average score on the MOS physical composite scale Average score on the MOS psychological composite scale 49.5% 46 years 61% 39% 18% 64% 8.4% 26% 26% 21% 35% 19% 16% 10% 14% 48% 9.7 years 83% 46% 2.3 38% 24% 27% 4.27 63.7 (out of 100) 64.3 (out of 100) of the ECSQ are presented in a separate paper [21]. Responses to the Events in Care measure are detailed in Table 2. For each area, we indicate the proportion of PWHAs that endorsed each item as well as the proportion that endorsed any item in that category. The most prevalent areas of concern included Medical Problems (66%), Specialty and Inpatient Hospital Care (56%), Psychological Symptoms (54%), and Preventive Health Care & Screening (50%). Endorsement frequencies varied substantially across individual items. The most common specific concerns included a Psychological Symptom, "bothered by sadness or depression" (48%) and a Medical Problem Symptom, "becoming more tired and having less energy" (47%). Across all nine areas, 50% of PWHAs indicated that they needed additional information in at least one area and 48% stated that a provider had raised some concern regarding their care. On average, PWHAs endorsed items in 3.7 areas of concern (sd = 2.3), or a total of 1568 events in care. Selection of concerns for further assessment Due to practical limitations of time, we decided to ask respondents the Dynamics of Care questions in only three of their identified areas of concern. As mentioned above, in order to select a representative sample of clients' concerns for further discussion in the Dynamics of Care assessment, we first asked respondents to rate their satisfaction with services available to them in each area, in order to select areas in which they were most and least satisfied. Then, we asked them to indicate which area represented their "biggest concern or need." These ratings allowed us to follow-up on clients' most salient concerns, as well as additional areas of concern that varied in terms of level of satisfaction with services. When several different areas received a respondent's highest or lowest rating, interviewers were instructed to select one of the areas that were mentioned less frequently in our pilot data, specifically, Family Planning, Sexual Risk Behavior or Substance Use. In other words, we designed the process to over-sample areas that tended to be mentioned less frequently. Table 3 summarizes the results of this process. Life Circumstances & Demands (20%), Medical Problems (21%), and Psychological Symptoms (16%) were identified most often as respondents' biggest areas of concern, in contrast with Substance Use (5%), Sexual Risk Behavior (4%), and Family Planning (3%) (see Table 3, Column 1). Areas identified as being of greatest concern were probed further in the Dynamics of Care assessment. Among those people who had indicated concerns in a given area, areas most likely to be given the highest satisfaction ratings included Adherence to Medical Instructions (52%), Sexual Risk Behavior (51%), Preventive Health Care & Screening (48%) and Specialty and Inpatient Hospital Care (46%) (see Table 3, Column 2). Respondents were least likely to select the highest satisfaction rating for Services related to Substance Use (31%) and to Life Circumstances & Demands (32%). Based on our priorities for sampling, selection of events in care for subsequent probing ranged from 100% for Family Planning to 17% for Medical Problems (see Table 3, Column 3). Areas most likely to receive clients' lowest satisfaction ratings included Preventive Health Care and Screening (34%), Specialty and Inpatient Hospital Care (32%), and Psychological Symptoms (30%). Areas least likely to receive lowest ratings included Family Planning (11%) and Substance Use (15%) (see Table 3, Column 4). Since low satisfaction ratings were less prevalent than high, a higher proportion of these events were probed, ranging Page 6 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 http://www.hqlo.com/content/6/1/20 Table 2: Rates of trigger concerns over the past three months, by area and item (N = 428) Adherence to Medical Instructions 43% Difficulty taking medications, like forgetting, being late, prescriptions Difficulty following medical advice Any side effects that interfered with activities or made you feel worse Questions about treatment or lab tests Any provider expressed concern about adherence or make treatment changes 21% 8% 29% 14% 16% Medical Problems and Symptoms 66% Any new symptoms or health problems Pain or symptom gets noticeably worse More tired or have less energy Any medical problem interfered with activities or ability to care for yourself Any questions about health symptoms Seek treatment at an emergency room Provider expressed concerns about your symptoms, pain or energy level 28% 27% 47% 23% 29% 20% 21% Specialty & Inpatient Hospital Care 56% Required a medical specialist Any dental problems Any problems with vision Any elective surgery or procedure In hospital or institution (medical) Provider suggested a medical specialist 23% 23% 29% 8% 9% 18% General Wellness and Prevention 50% Want check-up or screening for blood pressure, diabetes, cancer, etc. Change habits: diet, smoking, exercise Have any questions about staying healthy to discuss with provider Provider said you were overdue for any procedure, screening or other tests 23% 38% 29% 11% Substance Use 11% Increased use of alcohol, marijuana, heroin, cocaine or drugs Your concerns about alcohol, drugs Provider concerned re: alcohol, drugs 7% 9% 8% Sexual Health and Risk Behavior 31% Find better ways to prevent sexual transmission of HIV or other infections Difficulty practicing safer sex Difficulty talking about HIV Want to discuss sexual risk and protection with provider Provider expressed concern about sexual risk or not protecting yourself 27% 6% 4% 8% 9% Family Planning and Birth Control 14% Want better ways to prevent pregnancy Made changes to birth control choices Considering changes to birth control Thinking about having a baby Want to discuss birth control or family planning with provider Provider had concerns, changes to birth control or family planning decisions 7% 2% 2% 9% 4% 3% Psychiatric Symptoms 54% Page 7 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 http://www.hqlo.com/content/6/1/20 Table 2: Rates of trigger concerns over the past three months, by area and item (N = 428) (Continued) Bothered by sadness or depression Felt panicky, anxious or out of control Confusion or difficulty concentrating interfered with activities or self-care Thoughts of harming yourself or others In hospital or institution (psychiatric) Questions about coping with feelings or changes in thinking or memory Provider suggested psychological help 48% 32% 29% 6% 2% 22% 20% Dealing with Life Circumstances 41% Medicaid coverage, loss of income, benefits, housing, or legal problems Needs of others (like children, family, partner) interfere with self-care Increased responsibilities at home Caring for someone sick or disabled Questions about how to handle problems with money, housing, family Provider discussed problems related to living situation and care giving 22% 4% 16% 12% 11% 11% from 50% of medical problems to 100% of family planning concerns (See Table 3, Column 5). Overall, the strategy of selecting respondents' greatest concern, considering other areas according to level of satisfaction, and over-sampling less common concerns yielded good representation across the nine areas under consideration. We were able to probe 969 events in care, which represents 62% out of a total of 1568 events indicated by the ECSQ. The distribution of these events ranged from 4% in the area of Substance Use to 16% in the area of Specialty and Inpatient Hospital Care (see Table 3). Overview of analysis of the Dynamics of Care We conducted a series of analyses on the Dynamics of Care assessment to better understand the psychometric Table 3: Rates of Selection of Areas to Probe in Dynamics of Care Interviews Adherence to Medical Instructions Medical Problems Specialty & Inpatient Hospital Care Preventive Health Care & Screening Substance Use Sexual Risk Behavior Family Planning Psychological Symptoms Life Circumstances & Demands 1 2 3 4 5 6 Chosen as Most Important Chosen as Most Satisfied % Chosen as 2nd Area to Probe Chosen as Least Satisfied % Chosen as 3rd Area to Probe % of All Events Chosen 10% 52% 63% 29% 83% 15% 21% 13% 41% 46% 17% 48% 27% 32% 50% 72% 15% 16% 7% 48% 30% 34% 58% 10% 5% 4% 3% 16% 31% 51% 43% 40% 87% 84% 100% 26% 15% 22% 11% 30% 86% 79% 100% 56% 4% 10% 5% 13% 20% 32% 30% 25% 66% 13% Explanation of Columns 1. Proportion of time each area was identified as the most important concern to discuss. 2. Proportion of time area was identified as one of the areas with most satisfactory care. 3. Proportion of time area was chosen when it was eligible for the second slot. 4. Proportion of time area was identified as one of the areas with least satisfactory care. 5. Proportion of time area was chosen when it was eligible for the third slot. 6. Proportion of time area was chosen for discussion in any slot out of all selected areas. Note: Up to three areas selected for discussion included PWHAs' biggest need or concern and areas of greatest and least satisfaction with available services. Since levels of satisfaction could be tied, interviewers were instructed to break ties by selecting areas that were mentioned less frequently in pilot work (family planning, sex risk and substance use. The goal was to balance the number of times we probed each area. As Column 6 demonstrates, we were successful in over-sampling these areas. Page 8 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 and response properties of each section. The Dynamics of Care assessment was designed to be used in several different ways. The focus on specific events in care makes it possible to examine similar issues across different clients and settings (for example, comparing all episodes in the area of adherence). It is also possible to combine like items across areas of assessment, to derive idiographic (that is, individually-defined) measures of processes in care that are comparable across individuals, despite differences in events in care. For example, it is reasonable to examine individuals' average satisfaction across providers. Similarly, we can indicate whether clients ever encountered a particular barrier in seeking care, irrespective of the specific events in care under consideration. For the present paper, we have focused analysis on summary ratings derived from the dynamics measurements. This provided an opportunity to consider how baseline individual differences in health care, health status, and personal characteristics related to experiences in care. In order to present analyses in a cogent fashion, we will focus on each section of the interview in sequence: Stage of Help-Seeking, Barriers and Delays Encountered while Seeking Help, Perceived Quality of Interaction with Main Providers and Problem Resolution. This is necessary because sample size changes from section to section (e.g., we only discuss barriers encountered while help-seeking among those respondents who actually sought help). This sequential approach also conveys how the assessment can highlight and focus on different junctures in care. For each section of the measure, we present information on item response rates, as well as item homogeneity (intraclass correlations) across respondents' sampled areas of care. We also provide information on data reduction procedures, including inter-item reliability (Cronbach's alpha coefficient) and principal component structure of various summary scores. These results are presented in the narrative or in separate tables, below. Finally, we describe results of a series of multiple regression analyses to determine whether Dynamics of Care measures were related to demographic and health-related measures in a coherent fashion. In order to filter the large number of predictor measures and focus in on the strongest and most reliable associations with each dynamics variable within the different sections, we conducted separate regression analyses to identify the strongest predictors from each set of measures, and then considered only those predictors in a final, overall analysis. Results of all regression analyses are presented in similar format, summarized in Tables 4679. We will refer back to these tables throughout the presentation of results. We also examined zero-order correlations to identify possible instances of multicollinearity, and will comment on those findings in the description of each analysis. http://www.hqlo.com/content/6/1/20 Stage of help-seeking and factors that influenced helpseeking decisions The first section of the Dynamics of Care interview assesses whether and when individuals started to seek help for each sampled event in care. To be selected, these events had to have been a concern at some point during the three months prior to the interview; however, the onset of these concerns could have occurred earlier. Thus, of the 969 events sampled for probing, 16% had emerged as problems in the last month and an additional 28% occurred in the past three months. Of the remaining events, 13% emerged within the past six months, 14% within the past year, and 29% had been problems for a year or more. Respondents had already started to receive help for 52% of the sampled events, were actively seeking help for 16%, and were considering seeking help for 21%. For the remaining 11% of events, respondents said that they would not consider seeking professional help. Across 387 individuals (of a possible 428) who identified at least one event in care, 80% were already receiving help for at least one of their sampled concerns, 34% were seeking help for at least one concern, 41% were contemplating help seeking in some area, and 23% mentioned at least one area in which they would not consider seeking professional help. Among 262 individuals for whom three events were sampled, intraclass correlations (ICC) in help seeking status indicators were relatively low across areas of care, ranging from .02 to .09. (ICC can be interpreted as a correlation between the measure and some indicator of class membership – in this case that events were all reported by the same individual). The timing of the onset of individuals' different concerns also had a moderate ICC = .23. These results suggest that the interview succeeded in eliciting discussion of individual problems and concerns that were distinct from one another. In order to study differences in help seeking, we created composite scores reflecting the proportion of events for which an individual was receiving, seeking, contemplating or avoiding help. Correlates of these composites are presented in Table 4. Help seeking status was strongly associated with differences in areas of concern. People dealing with life circumstances were less likely to be receiving assistance but more likely to be seeking assistance. Alternatively, people concerned about specialty care were more likely to be seeking assistance, and, as would be expected, were less likely to say that they would not consider professional help. People with substance use and wellness concerns were more likely to say that they were considering initiation of care but had not yet done so. Help seeking was also associated with presence of providers. As expected, receipt of help was generally associated with provider recognition of problems and presence of different types of providers. However, people seeking help Page 9 of 20 (page number not for citation purposes) Health and Quality of Life Outcomes 2008, 6:20 http://www.hqlo.com/content/6/1/20 Table 4: Regression Analyses – Stage of Help-Seeking % Already Receiving Help Dealing with Life Circumstances (0.21) 0.13 p < .001 Dental Care (0.16) Recent Unprotected Sex (-0.14) Number of Case Managers (-0.08) Psychiatric or Mental Health Care (0.14) Problems Recognized by a Provider (0.10) Any Employment (0.10) Criminal Justice Involvement (0.13) % Seeking Help Dealing with Life Circumstances (0.18) Specialty and Inpatient Care (0.15) 0.12 p < .001 Dental Care (-0.14) Physical QOL (-0.10) Spanish Speaking (-0.09) Number of Case Managers (0.12) Recent Unprotected Sex (0.15) Any Employment (-0.11) Lives with Other Adults (0.13) % Considering Help Substance Use (0.15) 0.07 p < .001 Problems Recognized by a Provider (-0.08) General Wellness and Prevention (0.11) CD4 Count Greater than 200 (0.09) Alcohol Use (0.08) % Would Not Consider Help Specialty and Inpatient Care (0.13) Spanish Speaking (0.11) Age (-0.11) Lives with Another HIV+ Person (0.09) 0.09 p < .001 Number of Specialists (-0.12) Dental Care (-0.11) Number of Case Managers (-0.08) Special Needs Plan Enrollee (-0.10) tended to report greater involvement with case managers, especially compared to people who were already connected to services. Help seeking was also associated with individual health status. People who reported poorer physical quality of life at baseline were more likely to be seeking care, while those with fewer symptoms were more likely to say that they would not consider professional help. Alternatively, recent sexual risk behavior and alcohol use were associated with seeking and contemplating care, respectively. Help seeking was also related to indicators of socioeconomic status, with employed individuals more engaged in care, in contrast with those who were recently involved in the criminal justice system or who had not completed high school. Ethnic differences also emerged. Spanish-speaking Hispanics were more likely to have been receiving help, but respondents who identified as Hispanic reported being less likely to not consider seeking help. Active help seeking was also associated with living with other adults and particularly other HIV+ individuals. Younger respondents were also more likely to be considering help. In order to better understand help seeking, the Dynamics of Care assessment includes a series of nine items that ask Number of Symptoms (-0.09) Latino Identity (-0.16) High School Graduate (-0.12) Criminal Justice Involvement (0.11) about beliefs and preferences that might deter help seeking. These items and their orthogonal varimax principal component structure are presented in Table 5. This analysis was conducted on events (not individuals), so that it would be possible to directly compare factor scores generated for different areas of concern. This four component solution accounted for 68% of the total variance with clear simple structure and item communalities ranging from .55 to .89. Intraclass correlations for these components were computed by comparing the component scores for each of an individual's three sampled areas of concern. The magnitude of ICCs varied by factor: For example, component 1, the belief that nothing would help and one had to learn to live with the problem, and component 3, the desire to get things back on track independently and to need to obtain more information, yielded high correlations of .40 and .45, respectively. This indicates that these two dimensions tended to be similar across all of an individual's areas of concern. Alternatively, component 4, the belief that a problem was not important enough to seek help, had an ICC of .16, indicating that this component was much more problem-specific. Component 2, feeling embarrassed, uncomfortable and judged about seeking care, had an intermediate ICC of .24. Page 10 of 20 (page number not for citation purposes)