Preparation and Planning for Future Care in the Deep South: Adapting a Validated Tool for Cultural Sensitivity

Abstract

Background and Objectives

This study describes the adaptation and validation of Sörensen et al. (2017)’s preparation for future care (PFC) scale with diverse samples including rural dwelling African Americans and certified nursing assistants (CNAs), and subsequent psychometric development.

Research Design and Methods

Responses to the five-subscale PFC survey from 33 rural African American men across 12 months and cognitive interviews with a subset of 12 of these men are described. Psychometric refinement included descriptive qualitative analyses of consultations with experienced lay research advisors (N = 4 and N = 7) regarding potential changes to the PFC and a confirmatory factor analysis of the resultant scale (N = 138).

Results

Cognitive interviews with rural African American men revealed difficulty understanding Eurocentric questions. Emergent themes included emotional avoidance of planning, considerations of nursing homes and possible care providers, and coping strategies. In two consultation meetings, trained lay research advisors recommended language modifications to the original questions and response options. Factor analyzing the resultant scale revealed support for the original subscale constructs (acceptable fit: χ² = 205.03, df = 124, p < .001; root mean square error of approximation = .069 [.052–.085]; comparative fit index = .93; Tucker–Lewis index = .91).

Discussion and Implications

PFC and engagement in advance care planning is uncommon among African Americans, possibly due to distrust of and lack of cultural competency among health care professionals. The resulting tool and response options may be used as an interview guide/survey with African Americans to gain understanding about their preparation for future health care needs.

Health disparities increase morbidity and risk of mortality as well as reduce the ability to achieve optimal health outcomes among disadvantaged groups, including people of color, those with low education and income, and rural-dwelling individuals (Centers for Disease Control, 2013). Many of the health disparities in ethnic/racial groups in the United States can be traced to factors including historical distrust of the medical and research communities secondary to the Tuskegee Syphilis Study (Corbie-Smith, Thomas, & George, 2002; Freimuth et al., 2001) and lack of cultural competence of the health care workforce (Betancourt, Green, Carrillo, & Ananeh-Firempong, 2003). These and other factors motivate the low rates of African Americans (AAs) engaging in advance care planning and making preparations for future care, such as completing advance directives or discussing end-of-life treatment preferences with potential proxy decision makers (Crowther, Huang, & Allen, 2015; Huang et al., 2016; Melhado & Bushy, 2011). Positive outcomes are associated with engaging in these preventive health behaviors, however, including knowledge of and adherence to one’s treatment wishes (Houben, Spruit, Groenen, Wouters M., & Janssen, 2014), lower stress and anxiety for family and loved ones (Detering, Hancock, Reade, & Silvester, 2010), and lower health care costs during the last week of life (Zhang et al., 2009).

Prior research has identified a progression of four actions that underlie preparation for future care (PFC): developing an awareness of the need for future care, gathering information about potential needs and services/resources to meet those needs, making decisions about one’s preferences regarding sources and types of future care, and making concrete plans for service use (e.g., executing an advance directive) (Sörensen & Pinquart, 2001). Another dimension of PFC, avoidance of future care thoughts, interferes with this progression and, in a study of older primary care patients, has been linked to more depression symptoms 2 years later (Sӧrensen, Mak, Chapman, Duberstein, & Lyness, 2012). In contrast, more concrete planning is related to lower likelihood of being diagnosed with depression among primary care patients (Sörensen et al., 2012) and better subjective coping with health transitions and uncertain futures (Sörensen & Pinquart, 2000).

Although PFC has been studied cross culturally (Pinquart, Sӧrensen, & Davey, 2003), much of the prior research regarding future planning behaviors has included samples of predominantly older, non-Hispanic White, urban-dwelling women. The initial item generation for the original, 29-item PFC scale was based on qualitative interviews with older women in East Germany and Utah (Sörensen & Pinquart, 2000). This measure was validated by surveying rural and urban individuals aged 64–92 years in East Germany and in Utah and Georgia, United States, with only 9% of the sample identifying as AA (Sörensen & Pinquart, 2001). Notably, the survey approach used here may have missed racial/ethnic differences in language usage and interpretation of questions in the PFC. Subsequent studies using the PFC measure in the long form (Pinquart et al., 2003) did not probe the instrument’s content validity with AAs before applying it in survey designs. For the recently published short form (Sörensen, Chapman, Duberstein, Pinquart, & Lyness, 2017), an additional 150 urban AAs were surveyed. Although in a factor analytic validation the authors found a good fit for equal loadings and intercept models, the error variances varied between AA and Whites, suggesting that certain items may not have comparable response patterns across race. Additionally, primarily urban and few rural AAs were included in the short-form validation. Given that AAs’ sociocultural values regarding advance directives and future care planning often vary from the Eurocentric perspective (Huang et al., 2016; Melhado & Bushy, 2011), and the differences in factor structure noted by Sörensen et al. (2017), further methodological work is necessary for the successful use of the PFC measure in diverse populations.

The purpose of this multimethod study is to (a) describe changes in PFC scores over a 12-month period among 33 rural AA men who were participants in a research study focused on prostate cancer screening knowledge (Oliver et al., 2018), (b) present information from cognitive interviews with 12 of these men at baseline regarding their reactions to and understanding of the PFC survey questions affecting the validity of their responses, (c) summarize results of consultation with two trained lay research advisory groups about potential adaptations to the 15-item PFC survey (Sörensen et al., 2017) for better translation to disadvantaged populations, and (d) examine the factor structure of the resultant adapted scale within diverse samples and in comparison with the original constructs. Our goal in integrating the results of several studies with diverse methodologies is the creation of a survey or interview tool that may improve culturally competent use of the PFC and understanding of the processes involved in PFC among predominantly AA respondents.

Participants for All Studies

Substudy 1: Piloting the PFC Among Rural AA Men

A convenience sample was recruited with the assistance of local community leaders and through snowball sampling. As per the needs of the parent project regarding prostate cancer screening knowledge (Oliver et al., 2018), inclusion criteria were as follows: (a) identifying as a rural-dwelling AA man, (b) age 40–74 years, (c) no personal history of having a diagnosis of prostate cancer, and (d) English speaking.

Setting and Sample

Participants were recruited in three counties within the Black Belt region of Alabama, a rural area with a large AA population, so named because the area consists of the richest soil and a primarily agricultural economy. Greene, Hale, and Sumter Counties’ 2011 median household income was $24,226, $30,051, and $22,186, respectively, compared to $43,253 for Alabama (Alabama Department of Public Health, 2005). Greater than 80% of Greene County, 58% of Hale County, and over 73% of Sumter County residents are AA (US Census Bureau, 2015).

Participants were 33 AA rural-dwelling men with a mean age of 54.61 (standard deviation [SD] = 8.3; range 40–71). Of the sample, 15.2% had not completed high school, 42.4% had a high school education or equivalent, 30.3% had trade school or some college, and 12.1% had a college degree or higher. Fifty-one percent reported at least some difficulty in paying for basic needs. Thirty of the original 33 AA men at baseline completed the 12-month interviews (90.91% retention; Oliver et al., 2018); there were no differences between respondents completing the 12-month interviews and those who did not.

Substudy 2: Subsample for Cognitive Interviews

The purposive subsample included 12 of the original 33 AA men aged 40–65 years (M age = 53.50; SD = 8.64) from the three rural communities described previously, four men from each community. Roughly 42% (N = 5) of these men had a high school education, 41.7% (N = 3) had trade school or some college, and 16.6% (N = 2) completed college or had an advanced degree. There were no differences in men who completed cognitive interviews versus those who did not (N = 21) in age or income adequacy.

Substudy 3: Consultation With Lay Research Advisors

In order to gain specific recommendations to respond to the issues arising from these cognitive interviews, we contracted with two advisory groups to gain further insight into potential changes to PFC short-form items that might increase their clarity and cultural competency: the Wisconsin Network for Research Support Community Advisors on Research Design and Strategies (CARDS; Kaiser, Thomas, & Bowers, 2017) and a Patient-Centered Outcomes Research Institute (PCORI)-funded Project Advisory Council (PAC) in Sumter County (Sharing Opinions and Advice About Research in the Deep South [Project SOAR]). The primary scientific goal was to glean information from experienced community advisory groups/lay research advisors on wording of PFC items and response options.

Community Advisors on Research Design and Strategies

A one-and-a-half hour telephone consultation session was conducted with four members of the CARDS group to review the PFC items and response options (Kaiser et al., 2017). The CARDS consultation group consisted of two non-Hispanic White women who were research staff, an AA woman and an AA man who were community members recruited from senior and women’s groups, the food pantry, and parenting programs. All four group members were between the ages of 30 and 59. CARDS members are community-dwelling adults in Wisconsin who review materials developed for research, education and outreach those who have completed a training program on giving effective feedback. They meet with research clients to discuss proposed projects and review written materials and project plans.

Project SOAR PAC in Sumter County

Another one-and-a-half hour in-person consultation session was conducted with the Sumter County PAC (N = 7) including 2 AA men, 4 AA women, and 1 non-Hispanic White woman. These rural-dwelling individuals ranged in age from 33 to 77 (M = 50.83; SD = 15.65). Members consist of lay community stakeholders trained to work collaboratively with research partners to provide useful, directive, and appropriate feedback on various research topics and materials to be used within their communities. These materials may include recruitment flyers, survey development, and intervention design and implementation.

Substudy 4: Participants and Procedure for Factor Analysis

An independent participant sample of certified nursing assistants (CNAs) with significant professional exposure to death was used to increase the diversity of respondents as part of a dissertation (MKE). Exposure to death was thought to motivate these younger individuals to consider their own preparations for future care, thus increasing the relatability of the PFC scales for them. Exploration of the factor structure of the newly developed measure was conducted with this sample. Data were collected in person from CNAs (N = 138) who were employed in one of five local skilled nursing home facilities. Twenty-eight participants came from a pilot phase of data collection and 110 participants from the full study phase. The majority of the CNAs were women (N = 134; 97.8%) and self-identified as AA (N = 129; 93.5%) with a mean age of 38.26 (SD = 11.78).

Measures for All Substudies

Basic demographics including age and race/ethnicity were collected across samples.

PFC short-form used in the pilot, cognitive interview, and lay research advisor consultation

The process of preparation for future health care was measured with five subscales, each consisting of three items, measured with a 5-point Likert-type scale (1 = not at all true of me to 5 = completely true of me; (Sörensen et al., 2017). The subscales include Awareness (current sample Cronbach’s alpha = .80); Avoidance (current sample Cronbach’s alpha = .77); Gathering Information (current sample Cronbach’s alpha = .85); Decision Making (current sample Cronbach’s alpha = .61); and Concrete Planning (current sample Cronbach’s alpha = .67). PFC subscales have good 4-week test–retest reliability; predictive validity with regard to knowledge of services, sense of security about the future, and purchase of long-term care insurance; and discriminant validity with regard to decision styles and control beliefs (Sörensen et al., 2017). However, the validity of this scale for use among rural AA men is unknown.

My Aging Preparation Scale

My Aging Preparation Scale (MAPS; unpublished, 2013) (Table 1) was developed based on the original version of the short-form PFC (Sörensen et al., 2017) and the results of cognitive interviews and input from the two lay research advisory groups described earlier. The MAPS purports to measure the five latent constructs assessed in the original PFC (awareness, avoidance, gathering information, developing preferences, and concrete planning factors), while being more accessible to respondents.

Table 1.

Original 15-Item Preparation for Future Care (PFC) and Recommended Changes by Community Advisors on Research Design and Strategies (CARDS) and Project Sharing Opinions and Advice About Research in the Deep South

PFC	My Aging Preparation
AW1. I pay close attention to how my physical and mental capabilities are changing to assess whether I may soon need help or care	AW1.a. I pay attention to how my body is changing to decide if I may need help or care as I age AW1.b. I pay attention to how my mind is working to decide if I may need help or care as I age
AW2. I pay attention to information in the media on the risks of needing help or care in old age	AW2. I pay attention to information in the media about possible health problems as I age. (Prompt the participant to find out what the word “media” means to them)
AW3. Talking to other people has made me think about whether I might need help or care in the future	AW3. When I talk with other people about age-related health problems, I think about whether I may need help or care as I age
AV4. I try not to think about things like future loss of independence	AV4. I stay away from, or avoid, thinking about things like not being able to take care of myself as I age (e.g., not being able to drive, not being able to feed or clothe myself)
AV5. I don’t like to think about the risk of needing help or care in the future	AV5. I do not like to think about needing help or care as I age
AV6. I avoid negative topics like future dependence	AV6. I do not like to talk about the possibility of needing help or care as I age
GI7. I have compared different options for obtaining help or care in the future	GI7. I have compared different choices for help or care as I age
GI8. I have gathered information about options for care by talking to friends and/or relatives	GI8. I have gathered information about choices for care as I age by talking to friends and/or relatives
GI9. I have gathered information about options for care by talking to health care professionals (doctors, nurses, home health care agencies)	GI9. I have gathered information about choices for care as I age by talking with doctors, nurses, or other medical staff
	GInew. I understand the information I have gathered about choices for care as I age
DM10. I know what options for care I don’t want	DP10. I know what choices I do NOT want
DM11. I know my general preferences for care in the future even though I am not sure how I will get what I want	DP11. Even though I am not sure how I will get it, I know what I want for care as I age
DM12. If I ever need help or care, I can choose between several options that I have considered in some depth	DP12. I have thought carefully about my choices for care as I age
CP13. I will not consider certain types of care under any circumstance	CP13. As of today, I know I do not want certain types of care no matter what (e.g., using a feeding tube, going into a nursing home…)
CP14. I have explained to someone close to me what my care preferences are	CP14. I have told someone close to me about the care I want
	CPnew. I believe the person(s) I told understands what I want
CP15. I have written down my preferences for care	CP15. I have written down what I want for care

Note: AW = awareness; AV = avoidance; GI = gathering information; DM = decision making; CP = concrete planning.

Procedures for All Substudies

Substudy 1: Procedure for Piloting the PFC Among Rural AA Men

Data were collected between November 2011 and December 2013 with the approval from the University of Alabama Institutional Review Board (IRB). Written informed consent was obtained at baseline. Interviews were conducted in participants’ homes or another convenient location in the rural community (e.g., churches). Survey questions were read to the individuals and response cards were available to facilitate understanding. Six- and twelve-month interviews were conducted by telephone. Participants received $25 for completing baseline and 6- and 12-month follow-up assessments for a total of $75 across the course of the project.

Substudy 2: Cognitive Interview Procedures With Rural AA Men Regarding the PFC

Due to a desire to explore culturally determined responses and the challenges rural men had with the wording of some PFC items, standardized qualitative cognitive interviews were collected from a subsample of men (N = 12) from the larger study between August 2012 and February 2013 with the approval from the University of Alabama IRB. Cognitive interviews explore the decision- and response-making process for each survey item and grouping of items intended to measure a unique construct (e.g., awareness, avoidance, gathering information, decision making, and concrete planning). Beatty and Willis (2007) and Beatty (2003) recommended the most common procedure for cognitive interviewing that entails administering survey questions while collecting additional verbal information about the survey responses. This information is then used to evaluate the response to explore whether the question is generating valid information from the respondents. Analysis of cognitive interviews provided investigators with information regarding rural-dwelling AA men’s comprehension of items, the measure’s usefulness for assessing reported PFC activities, and the face validity of the short-form PFC survey in a unique sample.

The “think aloud” and intensive probing technique was used during cognitive interviews with participants (Beatty & Willis, 2007). This method was chosen because the goal was to both explore the comprehension of items on the PFC and attempt to better understand issues (thoughts and feelings) that arose as participants considered future health care needs. During this process, the participant was asked a question from the PFC subscale and then asked follow-up probes about their answer and their understanding of the content or meaning of the question. Scripted and spontaneous probes were used to obtain the greatest amount of relevant information (Willis, 2005). In general, men were asked (a) if the question was clear or if it contained terms that were difficult to understand, (b) what feelings/emotions arose in considering the question, and (c) what ideas or issues came to mind in considering the question (i.e., “think out loud as much as possible”). Men received $10 for participation in cognitive interviews.

NVIVO 10 software was used in qualitative coding of cognitive interview transcripts. Data from the digital recordings were transcribed verbatim by a professional transcriptionist. According to Sandelowski (2000), qualitative description is the method of choice when straight descriptions of phenomena are desired within a new and exploratory line of inquiry, such as cognitive interviewing, to better understand the validity of a survey in a unique population. The scientific goal was to stay close to the data and to the participants’ own words and reaction to survey items.

Best standards of qualitative methodology that support validity are rigor, trustworthiness, and an awareness of reflexivity, credibility, and believability (Russell & Gregory, 2003). In this study, we increased the trustworthiness of our findings by directly examining reflexivity, or what the coder brings to the coding of qualitative data, through the use of investigator triangulation. A two-member analysis team with experience and expertise in qualitative methodology (e.g., RSA and JSO) independently analyzed transcripts and developed themes. This investigator triangulation helped to keep investigators aware of potential biases and facilitated solid evidence for the interpretation of the data (Thorpe & Holt, 2007). Each transcript was read independently while initial themes and subcategories in the text were identified. Discrepancies were infrequent and were discussed until resolution was achieved. The analysis team kept detailed notes as part of an audit trail (Bradley, Curry, & Kelly, 2007).

Substudy 3: Procedure for Lay Consultation

During the digitally recorded telephone consultation session with Wisconsin CARDS, a research staff member read an item from the short-form PFC and then all members of the CARDS group discussed their understanding of and reaction to the item. Members of our research team were told to ask any specific questions of the CARDS group during the discussion. The session was interactive with frequent exchanges between our research staff and CARDS group members. The CARDS staff (Kaiser et al., 2017) recorded in copious notes everything that the CARDS group members said and our research team focused on listening.

Data from the digital recording were transcribed verbatim by a professional transcriptionist. The CARDS group used their notes from the consultation session to create and send a revised version of the PFC to our research team. Notably, because the recommendations of Project SOAR were provided in person, any recommended changes contrasting with the recommendations of the CARDS group were noted and recorded by our own research staff.

Substudy 3: Procedure for Factor Analysis of the Revised Measure

Prior to data collection, nursing home administrators from each of the facilities were contacted and asked to meet in order to obtain letters of support. During this meeting, the purpose and procedures of the study was explained and administrators had an opportunity to ask questions. After IRB approval, the study was advertised to potential participants via IRB-approved flyers in areas visible to staff members (e.g., staff break rooms, nurse’s station) and word of mouth. Some facilities made announcements to staff to inform them of the study; in these cases, research staff made an extra effort to be clear that the study was entirely voluntary and in no way required of them by their employer.

All interested parties received an informed consent document describing the procedures and minimal potential risks (e.g., possible discomfort over presented topics, fatigue). Participants who voluntarily agreed to be in the study completed the consent and accompanying packet of surveys. Participants were encouraged to enjoy snacks provided by the research staff during or after completion of the surveys. After completion of the packet, participants were given $10 in the pilot study, $5 in the main study, and thanked for their time and efforts. Research assistants asked participants to sign a receipt acknowledging receipt of the cash payment, and informed them their signed receipts were kept separate from their survey answers.

A confirmatory factor analysis of the revised measure created from consultation with lay research advisors in comparison with the factor structure of the original short-form PFC (Sörensen et al., 2017) was conducted using MPlus Software (Muthén & Muthén, 2010). Baseline model specifications included uncorrelated latent factors and residuals, and the variance for the first indicator for each of the five factors was set to 1. Given the impact of sample size on chi-square, three other fit statistics were preselected to determine model fit (Hooper, Coughlan, & Mullen, 2008): root mean square error of approximation (RMSEA; acceptable fit <.06); comparative fit index (CFI; acceptable fit >.90); and Tucker–Lewis index (TLI; acceptable fit >.90).

Results

Substudy 1 and 2: PFC Pilot Results and Discussion

Changes in rural-dwelling AA men’s preparations for future health care needs across 12 months are reported by subscale in Table 2. Repeated measures analysis of variance revealed that rural-dwelling AA men’s Awareness of future health care needs increased across time F(2, 31) = 4.70, p = .018. Similarly, their Avoidance increased across time, F(2, 31) = 3.59, p = .042, as did Gathering Information, F(2, 31) = 3.74, p = .037. In all cases, the greatest mean-level increase occurred at 6 months with decreases at the 12-month follow-up, but 12-month levels did not return to baseline levels. There were no significant differences across time in Decision Making or Concrete Planning. A purposive sample of 12 men who actively engaged in the discussion of preventive health behaviors and future health care needs were recruited to participate in cognitive interviews to further explore potential culture-specific responses to the PFC. “Active engagement” in interviews was defined as responding to interview questions in the prostate cancer screening knowledge study with elaboration of opinion across multiple sentences (all interviews conducted by JSO).

Table 2.

Repeated Measures Analysis of Variance for Subscales of the Preparation for Future Care Measure Administered to Rural-Dwelling African American Men (N = 33)

	Baseline	6 Months	12 Months
Subscales	M (SD)	M (SD)	M	F	p
Awareness	4.13 (0.74)	4.60 (0.47)	4.46 (0.56)	4.70	.018
Avoidance	3.00 (1.24)	3.62 (1.16)	3.31 (1.30)	3.59	.042
Gathering information	3.31 (1.04)	3.95 (0.95)	3.71 (1.07)	3.74	.037
Developing preference	3.58 (1.06)	4.06 (0.81)	4.02 (0.84)	1.95	.163
Concrete planning	3.05 (1.08)	3.24 (1.26)	3.17 (1.13)	0.23	.797

Three themes emerged from the cognitive interviews as follows: (a) issues with the wording and clarity of questions, (b) emotional reactions to the PFC questions, and (c) thoughts about PFC. Numbers after each quote indicate which man in the sample of cognitive interviewees made the statement. Five of the 12 men had issues with the wording or the clarity of the short-form PFC items. Notably, none of the men stated that they fully understood the wording of the items and all had at least one suggested change. Three men asked interviewers to “break down” early PFC items. One man asked directly what “future dependence” meant. Another man requested that the interviewer pause between or during questions to facilitate understanding. Six men found it difficult to respond to the question about preparation through knowing what care options they definitely would not want in the future. One stated, “I don’t know whether you can say you know about what you don’t want” (20061).

Second, respondents reported two emotional reactions to the PFC questions. Three men reported feeling good, stating that thinking about future care helped them feel knowledgeable and prepared. Eleven men mentioned avoidance, or a desire not to think about the need for future care because there is “power in thoughts and words.” This theme directly relates to one of the PFC subscales and has emerged in prior research about planning for future health care (Sörensen & Pinquart, 2000). One man stated simply, “if it ain’t bothering me, I don’t want to bother it … just worry about today, what’s happening in your life today” (30101). Another said, “I couldn’t answer that now because I might say I wouldn’t, but you just never know what your feeling will be down near the end when something like happens.” Participants mentioned not wanting to think of the need for future care because they have good insurance coverage or benefits, are in good health (e.g., “I don’t feel like, you know, it applies to me,” 10081), or try to maintain a positive attitude. One stated, “I’ve always been the type of person that looks after everybody else. I never thought about me getting in that, you know … that condition” (20091). Four men reported feeling sad during the cognitive interviews. Experiences of sadness were often associated with thoughts of aging and functional decline or with considering the situation of older individuals the men knew who needed and were receiving help or care. Other men described how their faith in God helped them maintain a sense of peace about future health care needs because they believed that God would be in control of their situation.

Finally, several subthemes emerged with regard to the third theme, thoughts regarding PFC. These subthemes included (a) thoughts of nursing homes, (b) who will care for me, and (c) coping strategies (e.g., acceptance: “It’s part of life”). Eight men described reactions to the possibility of future nursing home placement, with most expressing concern and a desire to avoid such a placement due to lack of quality care in that environment. However, other men acknowledged the possibility of future nursing home placement:

For example, if I get older or something, and if my wife passes or something, and my kids, you know … I get too old for them to take care of me, they’re gonna need to find somewhere, you know, different places to put people, like a nursing home and stuff like that. (10031)

A major concern was related to who would provide needed help and what help would be needed by the men in the future. Eleven men mentioned specific treatments such as medical checkups, ventilators/respirators, and getting home care. Another stated, “I wish that I could choose the one that I would like to be there with me, but you know, I know that’s gonna be hard to do” (30051). He also stated that he hoped, “that my kids can treat me the same way I’m treating my parents” (30051). He went on to express concern about home health care workers, “it goes through my mind about the type of care I don’t want: someone to come out that just got it for a job and not really have it in their heart, just their mind but not their heart.” Participants also mentioned suffering and observations of others they had known who received unwanted treatments. One stated, “I’m gone, don’t need to put me on no, you know, machines just to keep me pumped all up” (20091). Another expressed hesitancy due to planning for an unknowable future circumstance, “some people don’t want to be revived, but once you’re in it how do you know what you want before you get there?” (20121).

Seven men mentioned coping strategies. One man stated, “It seems like the more I know, the better I feel about things because I can handle it better” (10021). Seven men mentioned gathering information through the media, discussions with others, and doing their own research or paying attention to “advertising” about needs during old age. The veracity of available information was a concern. One man stated, “just gather the information and then kinda investigate it myself to make sure it’s reliable what they’re saying” (10021). Another said that considering the PFC questions, “lets you know that you need to be, you know, listening and gathering information as much as you can, so you’ll be educated and informed and you can make an intelligent decision” (20121). Ten men described a desire to make concrete plans through either having conversations with others or “writ[ing] it down.” Men most frequently mentioned having such conversations with their physician, but also described talking with their wife and children or other relatives and friends. One man stated, “I’m making plans for it now—discussing things with my kids and knowing what they think … I’m working towards it, so I kinda got a plan of what I want” (30061). Although men considered writing down their future wishes helpful, none of them had a written advance directive. One stated, “It probably should be in writing, I just haven’t done it yet” (20031). One man stated, “you don’t want to just put it all in the doctor’s hands or put it all in somebody else’s hands, because they might not have your interests at heart” (20121).

In summary, the 12 cognitive interviews with purposively sampled rural-dwelling AA men revealed problems with the clarity of PFC items, specific emotional reactions to the topic, and several cognitive reactions to PFC.

Substudy 3: Consultation With Lay Research Advisors Results and Discussion

The original short-form PFC and the revised interview tool resulting from the consultation sessions is provided in Table 1. Items in red indicate new or significantly revised items. Underlined text indicates terms within an item to be emphasized in interview delivery and bold text indicates significantly revised terms within an item.

The CARDS group recommended consistently using the term “as I age” across PFC items to increase clarity: “I would say ‘as I age.’ That’s a real sensitive topic … It’s very sensitive and with ageism and all that stuff, you know. I would say ‘as I age.’” The CARDS group also recommended changing the response options for every item in the PFC. Rather than the original “not at all true of me,” the CARDS group recommended “absolutely not true of me” to better differentiate this response from the option “not really true of me.” The Project SOAR group agreed with the changes proposed by the CARDS and recommended a three-response option consisting of “absolutely,” “maybe,” and “absolutely not.”

Overall, the CARDS group recommended simplifying the wording of PFC items (revised Flesh–Kincaid Reading Level = 3.6) and breaking multicomponent questions into component parts. For example, an original item in the Awareness subscale that asked about reactions to how “physical and mental capabilities are changing” was reworded in two separate items about “how my body is changing” and “how my mind is working” (e.g., awareness item one is now item 1a and 1b; see Table 1). The CARDS and Project SOAR groups reinforced changes such as providing examples of confusing terms (e.g., newspapers, television, and radio for “media”). The final revised scale was called “My Aging Preparations” or MAPS.

Substudy 4: Confirmatory Factor Analysis Results and Discussion

Notably, 93.5% of the CNA sample identified as AA. Cronbach’s alpha values for the MAPS in this sample ranged from questionable to good, with most subscales generating acceptable alpha values. The awareness subscale showed good reliability (α = .81) and the Gathering Information subscale also showed good reliability (α = .87). Avoidance and Concrete Planning showed acceptable reliability (α = .78 for both subscales). Developing Preferences was the subscale showing the weakest internal consistency, with questionable range alpha (α = .65). Overall, the entire MAPS showed good reliability with an alpha of .83.

Confirmatory factor analysis in the sample of CNAs (Table 3) revealed that the preexisting five-factor model including the latent constructs of Awareness, Avoidance, Gathering Information, Developing Preferences, and Concrete Planning had a close fit and was plausible. Results indicated acceptable fit: χ² = 205.03, df = 124, p < .001; RMSEA = .069 [90% CI: .052–.085]; CFI = .93; TLI = .91. Due to the impact of sample size and the number of indicators, RMSEA and CFI/TLI proved to be more reliable fit statistics.

Table 3.

Goodness of Fit Indices for My Aging Preparation Scale (MAPS) 5-Factor Model in Certified Nursing Assistant Sample

	χ2 (df)	Δ χ2 (df)	RMSEA (90% CI)	CFI	TLI
Model 1: 5-factor model	232.33 (125)	—	.079 (.063 to .095)	.91	.89
Model 2: 5-factor model, correlated residuals*	205.03 (124)	27.3 (1)	0.69 (.052 to .085)	.93	.91

Notes: CI = confidence interval; RMSEA = root mean square error of approximation; CFI = comparative fit index; TLI = Tucker–Lewis index.

*Residual variance between MAPS1 and MAPS2 were correlated.

General Discussion

These results extend prior research regarding the PFC scale by creating a revised, culturally sensitive tool that can be used with validity in diverse samples including AAs, rural-dwelling individuals, and individuals with significant exposure to death (e.g., CNAs) for the purpose of starting discussions about future care and assessing the extent of PFC needs. Additional research is needed to further explore the retest reliability and validity of the measure with diverse populations; however, the refinement with a Flesh–Kincaid Reading Level of 3.6 results in better item comprehension than for the original short-form PFC (Sörensen et al., 2017).

Preparing for healthy aging and future care follows an informed decision-making model (Briss et al., 2004), including patient and community interventions with a focus on improving knowledge, reducing decisional conflict or uncertainties, and increasing individual patient’s level of participation in decision making. Despite the benefits of planning in advance, most adults never complete an advance directive documenting their future treatment preferences (Huang et al., 2016; Melhado & Bushy, 2011). In general, many adults attempt to avoid “negative” thoughts, feelings, memories, physical sensations, and other internal experiences—even when doing so may create harm in the long-run (Hayes, Strosahl, & Wilson, 1999). This pattern of behavior has been termed “experiential avoidance” and qualitative studies including our own cognitive interviews reveal a tendency to avoid thinking of future dependency and need for care. It is possible that information selectivity and positivity bias among older adults with the goal of optimizing emotional experience and well-being (Blanchard-Fields, Jahnke, & Camp, 1995) also influence this avoidance, as is suggested by Mather and Carstensen (2005). Notably, the concerns and issues raised by the 12 rural AA men who participated in cognitive interviews are consistent with those expressed by older women in East Germany and Utah (Sörensen & Pinquart, 2000). Perhaps a complex blend of issues including experiential avoidance and positivity bias may promote avoidance.

Importantly, the existence of significant racial/ethnic differences in advance care planning (Crawley et al., 2000; Hopp & Duffy, 2000; Krakauer, Crenner, & Fox, 2002; Kwak & Haley, 2005) may also be explained by the history of discrimination against AAs in the health care system and differences in health literacy (Volandes et al., 2008) due to educational disparities. Future culturally sensitive interventions such as that conducted by Huang and colleagues (2016) should seek to enhance health literacy by providing information about risks, benefits, and alternatives to engagement in advance planning and PFC needs.

As in any research, this study has limitations. First, our samples varied by the primary research question addressed (e.g., whether PFC responses change over time, whether PFC question stems and responses are fully understood, how PFC question stems may be modified to increase understanding, how a revised scale compares to the original scale’s factor structure). This made it organizationally difficult to integrate the mixed methods used across our studies. Second, much of our work was cross-sectional and correlational; therefore, questions of causality and developmental change across more than 12 months are not addressed. However, this is one of few studies to focus on any longitudinal patterns in PFC and the only one to follow older AAs across time. Because qualitative cognitive interviews revealed less than full understanding of the question stems and response options, our longitudinal results should be interpreted with caution. Third, our focus has been the advance care planning and PFC behaviors of AA adults. Hence, by design our findings extend psychometric exploration of the PFC to this specific racial/ethnic group rather than a more diverse sample including Hispanics or Asian Americans. Moreover, great diversity exists in any racial/ethnic group and issues of intersectionality (Ghavami, Katsiaficas, & Rogers, 2016) are not considered herein.

In spite of these limitations, our findings have significant scientific and practice implications as we were able to gain insight into quantitative and qualitative reactions to the PFC needs in a specific demographic (AA adults) that has traditionally been difficult to recruit into health care research (Corbie-Smith et al., 2002). Moreover, prior qualitative PFC research has focused on women (Sörensen & Pinquart, 2000); hence, our cognitive interviews with rural dwelling AA men add significantly to our understanding of these constructs. Additionally, longitudinal findings that Awareness of future health care needs, Avoidance and Gathering Information increased across time are partly consistent with earlier work by Mak and Sörensen (2012), who found similar increases for all PFC subscales except Avoidance over a 7-year period. Larger time spans may be needed to detect significant change in Decision Making and Concrete Planning.

Advance care planning and preparing for future care can be considered preventive health behaviors that promote positive health outcomes (Allen, Phillips, Pekmezi, Crowther, & Prentice-Dunn, 2009; Payne, Prentice-Dunn, & Allen, 2010). Although past research suggests that the benefits of engaging in advance care planning and preparing for future care include knowledge of and adherence to treatment wishes (Houben et al., 2014), lower stress and anxiety for family and loved ones (Detering et al., 2010), and lower health care costs during the last week of life (Zhang et al., 2009), many of the older rural AA men in our study preferred not to discuss future care. This is consistent with findings in other samples, in which 18% actively endorsed avoidance of PFC (Pinquart & Sörensen, 2002). Furthermore, both in our interviews and other research, (Harris, Allen, Dunn, & Parmelee, 2013) spirituality plays a significant role in the process of PFC. Perhaps connecting spirituality with the future planning process, as well as continued exploration and use of culturally sensitive surveys and interview tools, will open possibilities for more effective planning and extension of its health benefits to diverse populations.

Funding

This work was supported by funding from the National Institute of Nursing Research (R21NR012250; Oliver and Allen, MPIs) and the Patient Centered Outcomes Research Institute (contract #1097; Allen, PI). No conflicts of interest are reported. The confirmatory factor analysis reported herein was conducted as part of Dr. Morgan K. Eichorst’s dissertation at The University of Alabama.