Article
Improving Sexual Health Communication Between Older Women and Their Providers: How the Integrative Model of Behavioral Prediction Can Help
Research on Aging 2014, Vol. 36(4) 450-466 ª The Author(s) 2013 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0164027513500055 roa.sagepub.com
Anne K. Hughes1, Ola S. Rostant2, and Paul G. Curran3
Abstract Talking about sexual health can be a challenge for some older women. This project was initiated to identify key factors that improve communication between aging women and their primary care providers. A sample of women (aged 60þ) completed an online survey regarding their intent to communicate with a provider about sexual health. Using the integrative model of behavioral prediction as a guide, the survey instrument captured data on attitudes, perceived norms, self-efficacy, and intent to communicate with a provider about sexual health. Data were analyzed using structural equation modeling. Self-efficacy and perceived norms were the most important factors predicting intent to communicate for this sample of women. Intent did not vary with race, but mean scores of the predictors of intent varied for African 1
School of Social Work, Michigan State University, East Lansing, MI, USA Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA 3 Psychology Department, Kenyon College, Gambier, OH, USA 2
Corresponding Author: Anne K. Hughes, School of Social Work, Michigan State University, 655 Auditorium Rd., Baker Hall, East Lansing, MI 48824, USA. Email: [email protected]
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
451
American and White women. Results can guide practice and intervention with ethnically diverse older women who may be struggling to communicate about their sexual health concerns. Keywords medical decision making, older adults, structural equation modeling
Introduction The World Health Organization considers sexual health to be an important component of quality of life and overall well-being (World Health Organization, 2006), but it is often overlooked as a concern for older women (Gott, 2005). Aging women experience high rates of some sexual problems (Laumann, Das, & Waite, 2008) and these problems can serve as important indicators of other serious health concerns, such as cardiovascular problems and depression (Basson & Schultz, 2007; Nusbaum, Lenahan, & Sadovsky, 2005). However, barriers to communication with a health care provider can influence whether or not older women get help for sexual health concerns (Lindau, Leitsch, Lundberg, & Jerome, 2006). Communication could be enhanced if the factors that play a role in whether or not older women will communicate with their health providers about sexual health were better understood. The integrative model of behavioral prediction (IMBP; Fishbein, 2000) proposes a model that identifies the factors that predict engagement in a behavior, such as communication. This study aims to increase understanding of how important the main variables of the IMBP are to an older woman’s intent to communicate with a health provider about sexual health.
Literature Review Sexual Health and Aging Sexual health is associated positively with life satisfaction (Woloski-Wruble, Oliel, Leefsma, & Hochner-Celnikier, 2010), greater purpose in life (Prairie, Scheier, Matthews, Chang, & Hess, 2011), and well-being (Avis et al., 2005; Davison, Bell, LaChina, Holden, & Davis, 2009; Laumann, Paik, & Rosen, 1999). In a global study of older men and women, subjective sexual well-being was found to be positively correlated with overall happiness (Laumann et al., 2005). Older women identify sexual health as a lifelong need (Woloski-Wruble et al., 2010), and continue to engage in sexual
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
452
Research on Aging 36(4)
activity well into older age (Jacoby & Revere, 2005; Trompeter, Bettencourt, & Barrett-Connor, 2012). However, aging is also associated with an increased incidence of problems in sexual functioning and associated relationship strain (Hartmann, Philippsohn, Heiser, & Ruffer-Hesse, 2004; Howard, O’Neill, & Travers, 2006; Ling, Wong, & Ho, 2008; Luftey, Link, Rosen, Wiegel, & McKinlay, 2009). With the incidence of some sexual problems as high as 49% among women aged 40–80 (Moreira et al., 2005) inclusion of sexual health in assessments or clinical encounters would provide an opportunity for providers to address this concern (Woloski-Wruble et al., 2010).
Communication About Sexual Health Because many sexual health concerns (e.g., anorgasmia, low desire, and some sexually transmitted diseases) do not present with physical signs, communication is especially relevant to their identification, treatment, and management (Sobecki, Curlin, Rasinski, & Lindau, 2012). Unfortunately, providers have struggled with communication about sexual health, especially across gender, age, and cultural differences (Burd, Nevadunsky, & Bachmann, 2005; Gott, Galena, Hinchliff, & Elford, 2004; Gott & Hinchliff, 2003). Factors that contribute to communication difficulties are provider discomfort, negative attitudes about sexuality and aging, and patient discomfort (Gott, 2005). Older female patients may be reluctant to talk because of embarrassment or shame or belief in the social stereotype that older women should not be concerned with sexuality beyond reproduction (Gott & Hinchliff, 2003). Younger age of the provider, perceived lack of interest on the provider’s part, and preference that the provider initiate discussions have also been found to impact sexual health communication for older women (Nusbaum, Singh, & Pyles, 2004).
The IMBP The IMBP proposes that to understand why someone engages in a particular behavior, we must understand that person’s intent to engage in the behavior. Intent is predicted by three primary variables: (1) attitudes toward the behavior, (2) perceived norms regarding the behavior, and (3) self-efficacy to perform the behavior, as well as skills and environmental constraints (Fishbein, 2000; Fishbein & Yzer, 2003). Intent is enhanced if one has a positive attitude toward the behavior, perceives that important people in their life will support the behavior, and has positive self-efficacy to perform the behavior (Fishbein, 2000; Fishbein & Yzer, 2003). Attitudes come from one’s feelings toward the behavior and are based on a person’s beliefs about the positive and negative
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
453
consequences of engaging in the behavior. Sexual health communication may be associated for some older women with negative outcomes, especially if they have had a negative experience with it in the past. Perceived norms are described as having two components: injunctive norms (e.g., the perception that other people that you are close to would support you adopting the behavior) and descriptive norms (e.g., whether or not those people engage in the behavior themselves). Older women’s intent can be impacted if their friends and family are generally supportive of discussing sexual health with providers and if those friends and family have spoken to a provider about their own sexual health. Self-efficacy refers to one’s belief that she could perform the behavior even under difficult circumstances (Fishbein, 2000; Fishbein & Yzer, 2003). In the area of sexual health communication, an example would be if a woman felt she was able to talk to her physician even if the sexual health topic was very difficult for her, say discussing a sexually transmitted disease or decreased sexual function. It is important to understand to what degree behavioral intent is under the control of attitudes, perceived norms, or self-efficacy (Fishbein & Yzer, 2003). The IMBP has been used in several studies to understand health behavior, such as predicting participation in cancer prevention activities (Smith-McLallen & Fishbein, 2009), HIV prevention (Rhodes, Stein, Fishbein, Goldstein, & Rotheram-Borus, 2007), and use of decision aids in primary care (Frosch, Legare, & Mangione, 2008). These research studies found that the influential factors in changing health behavior varied with the particular behavior under study as well as the sample population. Fishbein (2000) discusses the primacy of understanding behavior, so that we can implement interventions that most effectively target behavior change. A previous qualitative study (Hughes, 2011) found that for a convenience sample of 27 women in the Midwest (mean age ¼ 60.96 [SD ¼ 7.95]; 48% African American), perceived norms was the least important IMBP construct when these women considered whether or not to communicate about sexual health with a health care provider. Attitudes and self-efficacy were found to be more salient to this group of women, all of whom had intent to communicate with a provider about their sexual health. The current study was developed to examine whether the trends indicated in the previous qualitative work would be maintained in a larger, more representative sample of older U.S. women. This study was guided by two research questions: (1) How do the IMBP variables—attitudes, perceived norms, and self-efficacy—relate to intention to communicate about sexual health for older U.S. women? and (2) What is the relative strength of these three variables as predictors of an older woman’s intent to communicate with a health care provider about sexual health?
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
454
Research on Aging 36(4)
Method Sample The sampling frame consisted of members of SurveyMonkey Audience, a diverse group of people who are reflective of the U.S. population that uses the Internet (K. Campbell, SurveyMonkey Audience account representative, personal communication, September 29, 2011). Members of SurveyMonkey Audience participate in the completion of surveys on a voluntary basis and are compensated for their participation with points toward merchandise or donations to a charity of their choice. The principal investigator worked with a representative from SurveyMonkey to determine the sampling frame, which included women, aged 60 and above, one third of whom were non-White. The survey was made available online to this stratified sample. Participants had to be members of SurveyMonkey Audience to receive an invitation to participate in the online survey, and they chose whether or not to complete it.
Procedures The research protocol was approved by institutional review board of the university. During the fall of 2011, members of the sample were sent an e-mail notification from SurveyMonkey informing them of the study and a link to the online questionnaire, which was available for 1 week. The questionnaire included a cover letter that described the study as well as informed consent. No incentive was given to participants. Analysis was conducted using Mplus version 6.1 (Muthen & Muthen, 1998–2011).
Measures The questionnaire included demographic items, such as year of birth, race and ethnicity, level of education, income, and marital status. Items measuring the constructs of intent, attitudes, perceived norms, and self-efficacy were created by the principal investigator, and were guided by prior research on development of items to measure IMBP constructs (Busse, Fishbein, Bleakley, & Hennessy, 2010; Francis et al., 2004; Frosch, Legare, Fishbein, & Elwyn, 2009). The questionnaire was piloted with 10 women aged 60 and over from the community. The women provided feedback on the length of time it took to complete, their comprehension of the items, and what they thought the purpose of the study was. Changes based on their feedback were integrated into the final online questionnaire. Table 1 provides details on the
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
455
Table 1. IMBP Questionnaire Items. IMBP construct and response choice options Attitudes (5 items) –Semantic differential response choices
Perceived norms (5 items) –Semantic differential response choices (first item) –5 point Likert-type scale, strongly disagree to strongly agree (second to fifth items)
Self-efficacy (4 items) –5 point Likert-type scale, not difficult at all to very difficult (first item), not confident to very confident (second item), strongly disagree to strongly agree (third and fourth items)
Intent (3 items) –7 point Likert-type scale, strongly disagree to strongly agree
Item Talking to my primary care provider about my sexual health is: –likely/not likely to bring results –harmful/helpfula –will answer my questions/will not answer my questions –is pleasant for me/is unpleasant for me –not the right thing to do/the right thing to doa Most people who are important to me think that I should/I should not talk about my sexual health with my primary care providera I feel under social pressure to talk about my sexual health with my primary care provider People who are important to me want me to talk about my sexual health with my primary care provider Most people who are important to me talk about their sexual health with their primary care providers My primary care provider is comfortable talking with me about my sexual health How difficult is it for you to talk about your sexual health with your primary care provider? How confident are you that you could talk to your primary care provider about your sexual health?a Whether or not I talk about my sexual health with my primary care provider is under my controla My primary care provider’s comfort level impacts my ability to talk about sexual health at my visits I expect to talk about my sexual health with my primary care provider at my next visit I want to talk about my sexual health with my primary care provider at my next visit I intend to talk about my sexual health with my primary care provider at my next visit
Note. IMBP ¼ integrative model of behavioral prediction. a Indicates reverse scoring of the item in analysis.
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
456
Research on Aging 36(4)
individual items that were utilized to measure each of the IMBP constructs. Items were reverse scored when necessary and summed, with higher numbers corresponding to more of the construct (e.g., more positive attitudes, higher self-efficacy). Cronbach’s a coefficients were calculated for the IMBP variables in the model, with the following results: intent (.954), attitudes (.837), perceived norms (.625), and self-efficacy (.689).
Statistical Analysis Demographic characteristics of the sample were summarized using descriptive statistics. Confirmatory factor analysis and structural equation modeling were used to test the IMBP that examined the antecedents of intent to communicate about sexual health with a health care provider. A measurement model was tested for each of the four constructs (1) attitudes, (2) perceived norms, (3) self-efficacy, and (4) intent. Measurement models were evaluated and respecified before examination of the structural model (Bentler, 1989). To assess model fit multiple global fit indices were used, namely, the comparative fit index (CFI; Bentler, 1990), the Tucker–Lewis index (TLI; Tucker & Lewis, 1973), and the root mean square error of approximation (RMSEA; Brown & Cudeck, 1993). For the CFI and TLI indices values at or above .90 indicate adequate fit, for the RMSEA a value less than .08 is considered to indicate good fit (Bentler, 1995). Lagrange multiplier indices were used for post hoc model testing in order to identify theoretically relevant paths and enhance overall model fit (Bentler, 1995). We analyzed our models using full information maximum likelihood estimation that takes missingness into account (Enders & Bandalos, 2001; Muthen & Muthen, 1998–2011).
Results The invitation to participate in this study was sent to a total of 7,225 members of the SurveyMonkey Audience. Nine hundred and ninety-six respondents completed the survey for a response rate of 13.79%. Race of respondents included the following: 4.0% Native American, 2.7% Asian, 18.2% African American, 71% White, and 0.8% Native Hawaiian or Other Pacific Islander. Women of Hispanic ethnicity were 2.2% of the respondents. Because of the relatively large numbers of these races, women who were African American or White were included in the analysis for a sample size of 853 participants. Table 2 provides the demographic details of the sample.
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
457
Table 2. Demographic Characteristics of the Sample (N ¼ 853). Variable
N
Age Race White African American Education Some high school High school graduate/equivalency Some college Associate’s degree Bachelor’s degree Master’s degree Professional degree Doctorate degree Missing Total household income Less than US$19,999 US$20,000–US$39,999 US$40,000–US$59,999 US$60,000–US$79,999 US$80,000–US$99,999 US$100,000 or more Missing Marital/relationship status Married Widowed Divorced Separated Never married Living with a partner Missing
%
679 174
79.6 20.4
7 67 195 71 219 219 30 33 12
.8 8.0 23.2 8.4 26.0 26.0 3.6 3.9
67 174 185 159 97 134 37
8.2 21.3 22.7 19.5 11.9 16.4
405 109 191 21 77 30 20
48.6 13.1 22.9 2.5 9.2 3.6
M
SD
66.0
5.07
The first measurement model specified fit the data poorly (CFI ¼ .850, TLI ¼ .819, RMSEA ¼ .104). Evaluation of Lagrange multiplier indices identified five error covariances: two within the attitudes subscale, two within the perceived norms subscale, and one between Item 1 on the perceived norms subscale and Item 4 on the attitudes subscale. The error covariance within subscales likely reflects some shared measurement error. Shared error may also reflect the IMBP proposition that there are common background factors affecting behavior through their influence on the proximal
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
458
Research on Aging 36(4)
e
Attit 1
e
Attit 2
e e
Attit 3
e
Attit 5
e
PN 1
Attit 4
e
PN 2
e
PN 3
e
PN 4
e
PN 5
e
SE 1
e
SE 2
e
SE 3
.69 .70 .66 .67 .70 .74 .05 .40
Attitudes –.057
Perceived Norms
.127*
.97
.53 .52 .72 .91 .29
Intent
.89 .93
Intent 1
e
Intent 2
e
Intent 3
e
.430***
Self Efficacy χ2 (108) 553.13, p
Improving Sexual Health Communication Between Older Women and Their Providers: How the Integrative Model of Behavioral Prediction Can Help
Research on Aging 2014, Vol. 36(4) 450-466 ª The Author(s) 2013 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0164027513500055 roa.sagepub.com
Anne K. Hughes1, Ola S. Rostant2, and Paul G. Curran3
Abstract Talking about sexual health can be a challenge for some older women. This project was initiated to identify key factors that improve communication between aging women and their primary care providers. A sample of women (aged 60þ) completed an online survey regarding their intent to communicate with a provider about sexual health. Using the integrative model of behavioral prediction as a guide, the survey instrument captured data on attitudes, perceived norms, self-efficacy, and intent to communicate with a provider about sexual health. Data were analyzed using structural equation modeling. Self-efficacy and perceived norms were the most important factors predicting intent to communicate for this sample of women. Intent did not vary with race, but mean scores of the predictors of intent varied for African 1
School of Social Work, Michigan State University, East Lansing, MI, USA Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA 3 Psychology Department, Kenyon College, Gambier, OH, USA 2
Corresponding Author: Anne K. Hughes, School of Social Work, Michigan State University, 655 Auditorium Rd., Baker Hall, East Lansing, MI 48824, USA. Email: [email protected]
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
451
American and White women. Results can guide practice and intervention with ethnically diverse older women who may be struggling to communicate about their sexual health concerns. Keywords medical decision making, older adults, structural equation modeling
Introduction The World Health Organization considers sexual health to be an important component of quality of life and overall well-being (World Health Organization, 2006), but it is often overlooked as a concern for older women (Gott, 2005). Aging women experience high rates of some sexual problems (Laumann, Das, & Waite, 2008) and these problems can serve as important indicators of other serious health concerns, such as cardiovascular problems and depression (Basson & Schultz, 2007; Nusbaum, Lenahan, & Sadovsky, 2005). However, barriers to communication with a health care provider can influence whether or not older women get help for sexual health concerns (Lindau, Leitsch, Lundberg, & Jerome, 2006). Communication could be enhanced if the factors that play a role in whether or not older women will communicate with their health providers about sexual health were better understood. The integrative model of behavioral prediction (IMBP; Fishbein, 2000) proposes a model that identifies the factors that predict engagement in a behavior, such as communication. This study aims to increase understanding of how important the main variables of the IMBP are to an older woman’s intent to communicate with a health provider about sexual health.
Literature Review Sexual Health and Aging Sexual health is associated positively with life satisfaction (Woloski-Wruble, Oliel, Leefsma, & Hochner-Celnikier, 2010), greater purpose in life (Prairie, Scheier, Matthews, Chang, & Hess, 2011), and well-being (Avis et al., 2005; Davison, Bell, LaChina, Holden, & Davis, 2009; Laumann, Paik, & Rosen, 1999). In a global study of older men and women, subjective sexual well-being was found to be positively correlated with overall happiness (Laumann et al., 2005). Older women identify sexual health as a lifelong need (Woloski-Wruble et al., 2010), and continue to engage in sexual
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
452
Research on Aging 36(4)
activity well into older age (Jacoby & Revere, 2005; Trompeter, Bettencourt, & Barrett-Connor, 2012). However, aging is also associated with an increased incidence of problems in sexual functioning and associated relationship strain (Hartmann, Philippsohn, Heiser, & Ruffer-Hesse, 2004; Howard, O’Neill, & Travers, 2006; Ling, Wong, & Ho, 2008; Luftey, Link, Rosen, Wiegel, & McKinlay, 2009). With the incidence of some sexual problems as high as 49% among women aged 40–80 (Moreira et al., 2005) inclusion of sexual health in assessments or clinical encounters would provide an opportunity for providers to address this concern (Woloski-Wruble et al., 2010).
Communication About Sexual Health Because many sexual health concerns (e.g., anorgasmia, low desire, and some sexually transmitted diseases) do not present with physical signs, communication is especially relevant to their identification, treatment, and management (Sobecki, Curlin, Rasinski, & Lindau, 2012). Unfortunately, providers have struggled with communication about sexual health, especially across gender, age, and cultural differences (Burd, Nevadunsky, & Bachmann, 2005; Gott, Galena, Hinchliff, & Elford, 2004; Gott & Hinchliff, 2003). Factors that contribute to communication difficulties are provider discomfort, negative attitudes about sexuality and aging, and patient discomfort (Gott, 2005). Older female patients may be reluctant to talk because of embarrassment or shame or belief in the social stereotype that older women should not be concerned with sexuality beyond reproduction (Gott & Hinchliff, 2003). Younger age of the provider, perceived lack of interest on the provider’s part, and preference that the provider initiate discussions have also been found to impact sexual health communication for older women (Nusbaum, Singh, & Pyles, 2004).
The IMBP The IMBP proposes that to understand why someone engages in a particular behavior, we must understand that person’s intent to engage in the behavior. Intent is predicted by three primary variables: (1) attitudes toward the behavior, (2) perceived norms regarding the behavior, and (3) self-efficacy to perform the behavior, as well as skills and environmental constraints (Fishbein, 2000; Fishbein & Yzer, 2003). Intent is enhanced if one has a positive attitude toward the behavior, perceives that important people in their life will support the behavior, and has positive self-efficacy to perform the behavior (Fishbein, 2000; Fishbein & Yzer, 2003). Attitudes come from one’s feelings toward the behavior and are based on a person’s beliefs about the positive and negative
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
453
consequences of engaging in the behavior. Sexual health communication may be associated for some older women with negative outcomes, especially if they have had a negative experience with it in the past. Perceived norms are described as having two components: injunctive norms (e.g., the perception that other people that you are close to would support you adopting the behavior) and descriptive norms (e.g., whether or not those people engage in the behavior themselves). Older women’s intent can be impacted if their friends and family are generally supportive of discussing sexual health with providers and if those friends and family have spoken to a provider about their own sexual health. Self-efficacy refers to one’s belief that she could perform the behavior even under difficult circumstances (Fishbein, 2000; Fishbein & Yzer, 2003). In the area of sexual health communication, an example would be if a woman felt she was able to talk to her physician even if the sexual health topic was very difficult for her, say discussing a sexually transmitted disease or decreased sexual function. It is important to understand to what degree behavioral intent is under the control of attitudes, perceived norms, or self-efficacy (Fishbein & Yzer, 2003). The IMBP has been used in several studies to understand health behavior, such as predicting participation in cancer prevention activities (Smith-McLallen & Fishbein, 2009), HIV prevention (Rhodes, Stein, Fishbein, Goldstein, & Rotheram-Borus, 2007), and use of decision aids in primary care (Frosch, Legare, & Mangione, 2008). These research studies found that the influential factors in changing health behavior varied with the particular behavior under study as well as the sample population. Fishbein (2000) discusses the primacy of understanding behavior, so that we can implement interventions that most effectively target behavior change. A previous qualitative study (Hughes, 2011) found that for a convenience sample of 27 women in the Midwest (mean age ¼ 60.96 [SD ¼ 7.95]; 48% African American), perceived norms was the least important IMBP construct when these women considered whether or not to communicate about sexual health with a health care provider. Attitudes and self-efficacy were found to be more salient to this group of women, all of whom had intent to communicate with a provider about their sexual health. The current study was developed to examine whether the trends indicated in the previous qualitative work would be maintained in a larger, more representative sample of older U.S. women. This study was guided by two research questions: (1) How do the IMBP variables—attitudes, perceived norms, and self-efficacy—relate to intention to communicate about sexual health for older U.S. women? and (2) What is the relative strength of these three variables as predictors of an older woman’s intent to communicate with a health care provider about sexual health?
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
454
Research on Aging 36(4)
Method Sample The sampling frame consisted of members of SurveyMonkey Audience, a diverse group of people who are reflective of the U.S. population that uses the Internet (K. Campbell, SurveyMonkey Audience account representative, personal communication, September 29, 2011). Members of SurveyMonkey Audience participate in the completion of surveys on a voluntary basis and are compensated for their participation with points toward merchandise or donations to a charity of their choice. The principal investigator worked with a representative from SurveyMonkey to determine the sampling frame, which included women, aged 60 and above, one third of whom were non-White. The survey was made available online to this stratified sample. Participants had to be members of SurveyMonkey Audience to receive an invitation to participate in the online survey, and they chose whether or not to complete it.
Procedures The research protocol was approved by institutional review board of the university. During the fall of 2011, members of the sample were sent an e-mail notification from SurveyMonkey informing them of the study and a link to the online questionnaire, which was available for 1 week. The questionnaire included a cover letter that described the study as well as informed consent. No incentive was given to participants. Analysis was conducted using Mplus version 6.1 (Muthen & Muthen, 1998–2011).
Measures The questionnaire included demographic items, such as year of birth, race and ethnicity, level of education, income, and marital status. Items measuring the constructs of intent, attitudes, perceived norms, and self-efficacy were created by the principal investigator, and were guided by prior research on development of items to measure IMBP constructs (Busse, Fishbein, Bleakley, & Hennessy, 2010; Francis et al., 2004; Frosch, Legare, Fishbein, & Elwyn, 2009). The questionnaire was piloted with 10 women aged 60 and over from the community. The women provided feedback on the length of time it took to complete, their comprehension of the items, and what they thought the purpose of the study was. Changes based on their feedback were integrated into the final online questionnaire. Table 1 provides details on the
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
455
Table 1. IMBP Questionnaire Items. IMBP construct and response choice options Attitudes (5 items) –Semantic differential response choices
Perceived norms (5 items) –Semantic differential response choices (first item) –5 point Likert-type scale, strongly disagree to strongly agree (second to fifth items)
Self-efficacy (4 items) –5 point Likert-type scale, not difficult at all to very difficult (first item), not confident to very confident (second item), strongly disagree to strongly agree (third and fourth items)
Intent (3 items) –7 point Likert-type scale, strongly disagree to strongly agree
Item Talking to my primary care provider about my sexual health is: –likely/not likely to bring results –harmful/helpfula –will answer my questions/will not answer my questions –is pleasant for me/is unpleasant for me –not the right thing to do/the right thing to doa Most people who are important to me think that I should/I should not talk about my sexual health with my primary care providera I feel under social pressure to talk about my sexual health with my primary care provider People who are important to me want me to talk about my sexual health with my primary care provider Most people who are important to me talk about their sexual health with their primary care providers My primary care provider is comfortable talking with me about my sexual health How difficult is it for you to talk about your sexual health with your primary care provider? How confident are you that you could talk to your primary care provider about your sexual health?a Whether or not I talk about my sexual health with my primary care provider is under my controla My primary care provider’s comfort level impacts my ability to talk about sexual health at my visits I expect to talk about my sexual health with my primary care provider at my next visit I want to talk about my sexual health with my primary care provider at my next visit I intend to talk about my sexual health with my primary care provider at my next visit
Note. IMBP ¼ integrative model of behavioral prediction. a Indicates reverse scoring of the item in analysis.
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
456
Research on Aging 36(4)
individual items that were utilized to measure each of the IMBP constructs. Items were reverse scored when necessary and summed, with higher numbers corresponding to more of the construct (e.g., more positive attitudes, higher self-efficacy). Cronbach’s a coefficients were calculated for the IMBP variables in the model, with the following results: intent (.954), attitudes (.837), perceived norms (.625), and self-efficacy (.689).
Statistical Analysis Demographic characteristics of the sample were summarized using descriptive statistics. Confirmatory factor analysis and structural equation modeling were used to test the IMBP that examined the antecedents of intent to communicate about sexual health with a health care provider. A measurement model was tested for each of the four constructs (1) attitudes, (2) perceived norms, (3) self-efficacy, and (4) intent. Measurement models were evaluated and respecified before examination of the structural model (Bentler, 1989). To assess model fit multiple global fit indices were used, namely, the comparative fit index (CFI; Bentler, 1990), the Tucker–Lewis index (TLI; Tucker & Lewis, 1973), and the root mean square error of approximation (RMSEA; Brown & Cudeck, 1993). For the CFI and TLI indices values at or above .90 indicate adequate fit, for the RMSEA a value less than .08 is considered to indicate good fit (Bentler, 1995). Lagrange multiplier indices were used for post hoc model testing in order to identify theoretically relevant paths and enhance overall model fit (Bentler, 1995). We analyzed our models using full information maximum likelihood estimation that takes missingness into account (Enders & Bandalos, 2001; Muthen & Muthen, 1998–2011).
Results The invitation to participate in this study was sent to a total of 7,225 members of the SurveyMonkey Audience. Nine hundred and ninety-six respondents completed the survey for a response rate of 13.79%. Race of respondents included the following: 4.0% Native American, 2.7% Asian, 18.2% African American, 71% White, and 0.8% Native Hawaiian or Other Pacific Islander. Women of Hispanic ethnicity were 2.2% of the respondents. Because of the relatively large numbers of these races, women who were African American or White were included in the analysis for a sample size of 853 participants. Table 2 provides the demographic details of the sample.
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
Hughes et al.
457
Table 2. Demographic Characteristics of the Sample (N ¼ 853). Variable
N
Age Race White African American Education Some high school High school graduate/equivalency Some college Associate’s degree Bachelor’s degree Master’s degree Professional degree Doctorate degree Missing Total household income Less than US$19,999 US$20,000–US$39,999 US$40,000–US$59,999 US$60,000–US$79,999 US$80,000–US$99,999 US$100,000 or more Missing Marital/relationship status Married Widowed Divorced Separated Never married Living with a partner Missing
%
679 174
79.6 20.4
7 67 195 71 219 219 30 33 12
.8 8.0 23.2 8.4 26.0 26.0 3.6 3.9
67 174 185 159 97 134 37
8.2 21.3 22.7 19.5 11.9 16.4
405 109 191 21 77 30 20
48.6 13.1 22.9 2.5 9.2 3.6
M
SD
66.0
5.07
The first measurement model specified fit the data poorly (CFI ¼ .850, TLI ¼ .819, RMSEA ¼ .104). Evaluation of Lagrange multiplier indices identified five error covariances: two within the attitudes subscale, two within the perceived norms subscale, and one between Item 1 on the perceived norms subscale and Item 4 on the attitudes subscale. The error covariance within subscales likely reflects some shared measurement error. Shared error may also reflect the IMBP proposition that there are common background factors affecting behavior through their influence on the proximal
Downloaded from roa.sagepub.com at Southern Illinois University Carbondale on June 3, 2015
458
Research on Aging 36(4)
e
Attit 1
e
Attit 2
e e
Attit 3
e
Attit 5
e
PN 1
Attit 4
e
PN 2
e
PN 3
e
PN 4
e
PN 5
e
SE 1
e
SE 2
e
SE 3
.69 .70 .66 .67 .70 .74 .05 .40
Attitudes –.057
Perceived Norms
.127*
.97
.53 .52 .72 .91 .29
Intent
.89 .93
Intent 1
e
Intent 2
e
Intent 3
e
.430***
Self Efficacy χ2 (108) 553.13, p
