This article was downloaded by: [University of Cambridge] On: 18 August 2015, At: 09:29 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London, SW1P 1WG
Journal of Health Communication: International Perspectives Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uhcm20
Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age ab
a
b
c
a
Christine M. Gunn , Marina Soley-Bori , Tracy A. Battaglia , Howard Cabral & Lewis Kazis a
Department of Health Policy and Management, Boston University School of Public Health, Boston, Massachusetts, USA b
Women's Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USA c
Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA Published online: 19 Jun 2015.
Click for updates To cite this article: Christine M. Gunn, Marina Soley-Bori, Tracy A. Battaglia, Howard Cabral & Lewis Kazis (2015) Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age, Journal of Health Communication: International Perspectives, 20:9, 1060-1066, DOI: 10.1080/10810730.2015.1018628 To link to this article: http://dx.doi.org/10.1080/10810730.2015.1018628
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Health Communication, 20:1060–1066, 2015 Copyright # Taylor & Francis Group, LLC ISSN: 1081-0730 print/1087-0415 online DOI: 10.1080/10810730.2015.1018628
Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age CHRISTINE M. GUNN1,2, MARINA SOLEY-BORI1, TRACY A. BATTAGLIA2, HOWARD CABRAL3, and LEWIS KAZIS1 1
Department of Health Policy and Management, Boston University School of Public Health, Boston, Massachusetts, USA Women’s Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USA 3 Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA
Downloaded by [University of Cambridge] at 09:29 18 August 2015
2
Current breast cancer screening guidelines promote the use of shared decision making for women younger than 50 years of age, yet their effect on mammography utilization is largely unknown. This study aimed to examine the effect of two elements of shared decision making on the use of mammogram screening: patient-perceived choice and patient–provider communication. Data were obtained from HINTS 4, a nationally representative survey of the U.S. population, administered from 2011 to 2013. Choice was measured with the question ‘‘Has a doctor ever told you that you could choose whether or not to have a mammogram?’’ Communication was measured using a 7-item scale (range: 7–28; higher scores denote better communication). Binary logistic regression models assessed the effect of patient choice and communication on ever having a mammogram using weighted sample data. The sample included 1,085 women younger than 50 years of age: 31% of women perceived having a choice to undergo mammography. The mean patient–provider communication score was 22.8. Those who thought they were given a choice regarding mammography were more likely to have a mammogram relative to those who did not think a choice was given by the provider. Patient–provider communication had no significant association with mammography utilization. Patient perceived choice, but not patient-provider communication, is positively associated with mammography utilization in women younger than 50 years of age.
Breast cancer is the most prevalent cancer among women, accounting for an estimated 226,870 new cancer cases and 39,510 deaths in 2012 (American Cancer Society, 2012). While advances in early detection have certainly contributed to declines in mortality (Autier, Hery, Haukka, Boniol, & Byrnes, 2009; Webb et al., 2013) there remains controversy over the balance of risks and benefits for screening mammography for women younger than 50 years of age. In 2009, the breast cancer screening guidelines were updated by the U.S. Preventative Services Task Force (2009). These recommendations include more explicit language endorsing a shared decision-making (SDM) approach between doctors and patients, especially for women younger than 50 years of age, for whom widespread population screening is no longer recommended. Breast cancer screening is an ideal venue for the use of SDM because of the uncertainty surrounding evidencebased recommendations, variability in patient risk factors,
Address correspondence to Christine M. Gunn, Women’s Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, 801 Massachusetts Avenue, First Floor, Boston, MA 02118, USA. E-mail: [email protected]
and range of personal beliefs about screening (Sheridan, Harris, & Woolf, 2004). Research demonstrates that women highly value cancer screenings, particularly if they perceive themselves to be at risk for developing cancer (Davidson, Liao, & Magee, 2011). When patient values are in conflict with expert recommendations, it is an opportune time for physicians to engage in discussions with patients to support more informed decisions about the short- and long-term benefits and risks of screening. Yet, the prevalence of SDM in clinical practice and its effect on screening mammography utilization is largely unexplored. One of the difficulties in implementing SDM in practice concerns defining what this approach is. Models of SDM often represent domains of information exchange, deliberation, values=preferences, understanding, and decisions (Charles, Gafni &Whelan, 1999; Makoul & Clayman, 2006). The relation between SDM and utilization of services suggests that participating in SDM is associated with fewer visits, hospitalizations, diagnostic tests and elective procedures (Bertakis & Azari, 2011; Stacey et al., 2011). Elwyn and colleagues (2012) developed a model capturing the process of SDM focusing on three main elements: choice talk, option talk, and decision talk. Through a process of deliberation, these domains transform patient’s initial preferences into informed preferences, resulting in a decision (Elwyn
1061
Downloaded by [University of Cambridge] at 09:29 18 August 2015
Shared Decision Making and Mammography et al., 2012). These choice and option talk domains are represented as the process by which SDM is accomplished in the clinical setting, namely through providing a choice and communication with patients. The process-focused nature of this model renders it useful for measuring specific elements of SDM that relate to a decision and was used as a basis for this analysis. This article assesses patient choice and patient–provider communication, two essential elements of SDM (Elwyn et al., 2012; Makoul & Clayman, 2006), with their influence on utilization of screening mammography among a nationally representative sample of women. More specifically, we examined the influence of communication and sociodemographics on patient reports of being provided a choice to undergo a mammogram. Ciampa and colleagues have found that, especially for patients with low numeracy, patient perceptions of good patient–provider communication encourage participation in colorectal cancer screenings (Ciampa et al., 2010). It was hypothesized that higher levels of communication would be associated with the patient being provided a choice to undergo screening. We then determined whether having a choice to undergo screening affected utilization of mammography. The authors postulated that perceived patient choice is associated with less utilization compared with those not given a choice. There is significant heterogeneity in uptake of screening following SDM (Edwards et al., 2013). Our hypothesis that choice was associated with less utilization is based on previous literature that suggests a decline in general health care utilization in a SDM context (Stacey, Bennett, & Berry, 2011). Results from this analysis can be used to inform clinical practice by identifying the specific elements of SDM that are associated with utilization of cancer screening tests.
Method Study Sample Data were obtained from the Health Information National Trends Survey (HINTS), an ongoing, cross-sectional survey of the U.S. civilian, noninstitutionalized adult population administered biennially by the National Cancer Institute. Full details of the survey have been previously published elsewhere (Finney Rutten et al., 2012; Nelson, Kreps, Hesse, Croyle, & Willis, 2004). The survey provides researchers with estimates of the prevalence of cancerrelevant knowledge, attitudes, and behaviors based on a national probability sample. Access to the survey instruments and data is available through the National Cancer Institute (n.d.). HINTS 4 (2012) was used for the analysis in this study. Two cycles of surveys were administered by mail between October 2011 and January 2013. The survey sampling method consisted of a two-stage design. A stratified sample of addresses was first selected from a file of all residential addresses. The second stage sampled all adults either within households (Cycle 1) or one selected adult from each sampled household (Cycle 2) (‘‘Health Information National
Trends Survey 4: Cycle 1 Methodology Report’’ [National Cancer Institute, 2012] and the ‘‘Health Information National Trends Survey 4: Cycle 2 Methodology Report’’ [National Cancer Institute, 2013]). The survey was administered in English or Spanish, with a nominal incentive provided for participation. The mailing protocol used the Dillman approach for total design methodology to increase response rates (Dillman, 2009). These two cycles were combined for a total sample size of 7,589 participants. After restricting the sample to women younger than 50 years of age and excluding cases with missing values for the variables of interest, there were 1,085 cases. Demographic characteristics were comparable between the two cycles. All subsequent analyses included a cycle variable as a covariate to account for differences attributable to data collection time point. For analysis the sample was weighted at the individual subject level and adjusted using a jackknife variance estimation technique to include replicate weights. Sampling weights consist of three components: (a) respondent base weight; (b) adjustment for nonresponse; and (c) calibration adjustment. A full description of the methodologies used in calculating sample weights is published by the National Cancer Institute (2012). This study was reviewed by the Boston University Institutional Review Board and met the criteria for exemption. Variable Specification The primary outcome of interest was reported utilization of mammograms. Responses were collapsed into two categories: never had a mammogram versus ever had a mammogram. This dichotomous outcome was used because given the younger age of the sample, many women may not have yet begun serial screening and may have only just undergone their first mammogram. The key independent variable of interest was whether patients reported having a choice to undergo screening tests (‘‘Has a doctor ever told you that you could choose whether or not to have a mammogram?’’). The authors hypothesized that perceived patient choice is associated with less utilization compared with those not given a choice (Bertakis & Azari, 2011; Stacey, Bennett, & Barry, 2011). A second aspect of SDM, patient–provider communication, was assessed using a composite measure of seven communication questions that elicited from patients how often providers: (a) gave ‘‘you the chance to ask all health-related questions you had’’; (b) gave ‘‘the attention you needed to your feelings and emotions’’; (c) involved ‘‘you in decisions about your health care as much as you wanted’’; (d) made ‘‘sure you understood the things you needed to do to take care of your health’’; (e) explained ‘‘things in a way you could understand’’; (f) spent ‘‘enough time with you’’; and (g) helped ‘‘you deal with feelings of uncertainty about your health or health care.’’ This set of questions used in the measure was combined as an equally weighted sum for all items after a Cronbach alpha demonstrated excellent internal consistency reliability (a ¼ .92). The final measure had a response range from 7 to 28 with
Downloaded by [University of Cambridge] at 09:29 18 August 2015
1062 a higher score denoting better communication between patient and provider. It was hypothesized that better communication would be associated with the patient being provided a choice to undergo screening. Variables were chosen for inclusion based on a SDM framework published by Elwyn and colleagues (2012). Demographic characteristics include gender, race=ethnicity, marital status, age, education, income, health status, and insurance. Prior work indicates that these variables have a significant effect on utilization of health services (Blendon, Aiken, Freeman, & Corey, 1989; Joung, Van Der Meer, & Mackenbach, 1995; Makuc, Breen, & Freid, 1999). Many of these characteristics also influence the likelihood that SDM occurs in the medical encounter (Levinson et al., 2008; Siminoff, Graham, & Gordon, 2006). Two proxy measures of cancer risk were also included: personal cancer history and family cancer history. Statistical Analysis Bivariate associations were evaluated between sociodemographic variables, utilization, choice, and communication. We computed chi-square tests for categorical variables and t tests for continuous variables stratified by two categories and analysis of variance routines for more than two categories. A multivariable logistic regression model with perceived choice as the dichotomized dependent variable was constructed to understand the relation between the two SDM variables. Communication was the key independent variable with sociodemographic variables included as predictors. To test the independent effects of choice and communication on the utilization of screening mammography, a multivariable logistic regression was employed with choice and communication as the two independent variables of interest. The multivariable model adjusted for sociodemographic and cancer history variables. The model goodness-of-fit was assessed using the likelihood ratio chi-square test, the likelihood ratio test, and the C-statistic. For all analyses, statistical significance was set at the .05 level. Analyses were conducted in SAS 9.3 using PROC SURVEYLOGISTIC incorporating sampling weights. Odds ratios, with the associated 95% confidence interval, and p values are reported based on these models.
Results Descriptive statistics are displayed in Table 1. Each variable is presented with the original and weighted sample size and percentages. The mean and standard deviations are reported for continuous variables. The sample was racially representative of the U.S. general population: 68% were non-Hispanic White, 11% Black, 15% Hispanic, and 7% other. Most women were married (53%), had health insurance (87%), and rated their health as good or better (89%). 31.3% (SD ¼ 2.6) of women younger than 50 years of age reported they perceived having a choice to undergo mammogram screening. Participants rated patient–provider
C. M. Gunn et al. Table 1. Characteristics of the sample Variable
n
Race=ethnicity Hispanic 180 Non-Hispanic White 649 Black or African American 177 Other 93 Marital status Married or living as married 619 Divorced, separated or widowed 195 Single 285 Highest level of education Less than high school 53 High school graduate 146 Some college 319 College graduate 581 Income $35,000 369 $35,000–50,000 144 $50,000–75,000 188 >$75,000 398 Health insurance Yes 971 Regular provider Yes 793 General health Good or better 974 Personal history of cancer Yes 62 Family history of cancer Yes 838 Ever had mammogram Yes 515 Choice for mammogram Yes 350 Age, M (SD) 1,099 Communication, M (SD) 1,085
Weighted n
% (SD)
11,402,041 14.5 (2.6) 53,334,702 67.6 (4.0) 8,438,679 10.7 (1.8) 5,747,642 7.3 (1.0) 41,938,801 53.1 (3.8) 6,642,934 8.4 (1.3) 30,341,330 38.4 (2.8) 6,714,994 8.5 (3.4) 11,564,359 14.7 (2.1) 30,420,492 38.5 (2.3) 30,223,220 38.3 (3.6) 28,585,548 9,910,205 13,586,864 26,840,449
36.2 12.6 17.2 34.0
(6.1) (1.9) (1.8) (4.6)
68,584,115 86.9 (1.7) 54,317,629 68.8 (5.6) 70,394,878 89.2 (2.6) 3,212,682
4.1 (0.7)
59,762,147 75.7 (2.4) 30,176,847 38.2 (3.1) 24,663,239 31.3 (2.6) 33.4 (0.4) 22.8 (0.3)
Note. N ¼ 1,085, weighted N ¼ 77,277,150.
communication relatively high with a mean score of 22.8 (SD ¼ 0.3). Approximately 38% (SD ¼ 3.1) had ever had a mammogram performed. Mammogram screening was positively associated with having a choice for mammography (p ¼ .008), being married (p < .0001), having higher income (p ¼ .007), better self-reported health status (p ¼ .008), having a personal cancer history (p ¼ .0005), and having a regular provider (p ¼ .025) (data not shown in table). Increasing age (OR ¼ 1.03; 95% CI: 1.00–1.07), poorer health status (2.51; 1.08–5.87) and having a regular provider (1.69; 1.16–2.47) were significant and increase the odds of a patient perceived choice to undergo mammogram screening (see Table 2). Patients who rated their health status as fair or worse were more than twice as likely to report having a choice than those with better health (2.51; 1.08–5.87). Having a regular provider also increased the odds of reporting a choice by 69% compared
1063
Shared Decision Making and Mammography Table 2. Logistic model predicting patient-reported choice to undergo mammogram screening (yes=no)
Downloaded by [University of Cambridge] at 09:29 18 August 2015
Variable Communication Race (Ref. ¼ non-Hispanic White) Hispanic Black or African American Other Age Marital status (Ref. ¼ single) Divorced, separated, or widowed Married or living as married Education (Ref. ¼ high school graduate) College graduate Less than high school Some college Household income (Ref. < $35,000) $35,000 to
Journal of Health Communication: International Perspectives Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uhcm20
Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age ab
a
b
c
a
Christine M. Gunn , Marina Soley-Bori , Tracy A. Battaglia , Howard Cabral & Lewis Kazis a
Department of Health Policy and Management, Boston University School of Public Health, Boston, Massachusetts, USA b
Women's Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USA c
Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA Published online: 19 Jun 2015.
Click for updates To cite this article: Christine M. Gunn, Marina Soley-Bori, Tracy A. Battaglia, Howard Cabral & Lewis Kazis (2015) Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age, Journal of Health Communication: International Perspectives, 20:9, 1060-1066, DOI: 10.1080/10810730.2015.1018628 To link to this article: http://dx.doi.org/10.1080/10810730.2015.1018628
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Health Communication, 20:1060–1066, 2015 Copyright # Taylor & Francis Group, LLC ISSN: 1081-0730 print/1087-0415 online DOI: 10.1080/10810730.2015.1018628
Shared Decision Making and the Use of Screening Mammography in Women Younger Than 50 Years of Age CHRISTINE M. GUNN1,2, MARINA SOLEY-BORI1, TRACY A. BATTAGLIA2, HOWARD CABRAL3, and LEWIS KAZIS1 1
Department of Health Policy and Management, Boston University School of Public Health, Boston, Massachusetts, USA Women’s Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USA 3 Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA
Downloaded by [University of Cambridge] at 09:29 18 August 2015
2
Current breast cancer screening guidelines promote the use of shared decision making for women younger than 50 years of age, yet their effect on mammography utilization is largely unknown. This study aimed to examine the effect of two elements of shared decision making on the use of mammogram screening: patient-perceived choice and patient–provider communication. Data were obtained from HINTS 4, a nationally representative survey of the U.S. population, administered from 2011 to 2013. Choice was measured with the question ‘‘Has a doctor ever told you that you could choose whether or not to have a mammogram?’’ Communication was measured using a 7-item scale (range: 7–28; higher scores denote better communication). Binary logistic regression models assessed the effect of patient choice and communication on ever having a mammogram using weighted sample data. The sample included 1,085 women younger than 50 years of age: 31% of women perceived having a choice to undergo mammography. The mean patient–provider communication score was 22.8. Those who thought they were given a choice regarding mammography were more likely to have a mammogram relative to those who did not think a choice was given by the provider. Patient–provider communication had no significant association with mammography utilization. Patient perceived choice, but not patient-provider communication, is positively associated with mammography utilization in women younger than 50 years of age.
Breast cancer is the most prevalent cancer among women, accounting for an estimated 226,870 new cancer cases and 39,510 deaths in 2012 (American Cancer Society, 2012). While advances in early detection have certainly contributed to declines in mortality (Autier, Hery, Haukka, Boniol, & Byrnes, 2009; Webb et al., 2013) there remains controversy over the balance of risks and benefits for screening mammography for women younger than 50 years of age. In 2009, the breast cancer screening guidelines were updated by the U.S. Preventative Services Task Force (2009). These recommendations include more explicit language endorsing a shared decision-making (SDM) approach between doctors and patients, especially for women younger than 50 years of age, for whom widespread population screening is no longer recommended. Breast cancer screening is an ideal venue for the use of SDM because of the uncertainty surrounding evidencebased recommendations, variability in patient risk factors,
Address correspondence to Christine M. Gunn, Women’s Health Unit, Section of General Internal Medicine, Evans Department of Medicine, Boston University School of Medicine, 801 Massachusetts Avenue, First Floor, Boston, MA 02118, USA. E-mail: [email protected]
and range of personal beliefs about screening (Sheridan, Harris, & Woolf, 2004). Research demonstrates that women highly value cancer screenings, particularly if they perceive themselves to be at risk for developing cancer (Davidson, Liao, & Magee, 2011). When patient values are in conflict with expert recommendations, it is an opportune time for physicians to engage in discussions with patients to support more informed decisions about the short- and long-term benefits and risks of screening. Yet, the prevalence of SDM in clinical practice and its effect on screening mammography utilization is largely unexplored. One of the difficulties in implementing SDM in practice concerns defining what this approach is. Models of SDM often represent domains of information exchange, deliberation, values=preferences, understanding, and decisions (Charles, Gafni &Whelan, 1999; Makoul & Clayman, 2006). The relation between SDM and utilization of services suggests that participating in SDM is associated with fewer visits, hospitalizations, diagnostic tests and elective procedures (Bertakis & Azari, 2011; Stacey et al., 2011). Elwyn and colleagues (2012) developed a model capturing the process of SDM focusing on three main elements: choice talk, option talk, and decision talk. Through a process of deliberation, these domains transform patient’s initial preferences into informed preferences, resulting in a decision (Elwyn
1061
Downloaded by [University of Cambridge] at 09:29 18 August 2015
Shared Decision Making and Mammography et al., 2012). These choice and option talk domains are represented as the process by which SDM is accomplished in the clinical setting, namely through providing a choice and communication with patients. The process-focused nature of this model renders it useful for measuring specific elements of SDM that relate to a decision and was used as a basis for this analysis. This article assesses patient choice and patient–provider communication, two essential elements of SDM (Elwyn et al., 2012; Makoul & Clayman, 2006), with their influence on utilization of screening mammography among a nationally representative sample of women. More specifically, we examined the influence of communication and sociodemographics on patient reports of being provided a choice to undergo a mammogram. Ciampa and colleagues have found that, especially for patients with low numeracy, patient perceptions of good patient–provider communication encourage participation in colorectal cancer screenings (Ciampa et al., 2010). It was hypothesized that higher levels of communication would be associated with the patient being provided a choice to undergo screening. We then determined whether having a choice to undergo screening affected utilization of mammography. The authors postulated that perceived patient choice is associated with less utilization compared with those not given a choice. There is significant heterogeneity in uptake of screening following SDM (Edwards et al., 2013). Our hypothesis that choice was associated with less utilization is based on previous literature that suggests a decline in general health care utilization in a SDM context (Stacey, Bennett, & Berry, 2011). Results from this analysis can be used to inform clinical practice by identifying the specific elements of SDM that are associated with utilization of cancer screening tests.
Method Study Sample Data were obtained from the Health Information National Trends Survey (HINTS), an ongoing, cross-sectional survey of the U.S. civilian, noninstitutionalized adult population administered biennially by the National Cancer Institute. Full details of the survey have been previously published elsewhere (Finney Rutten et al., 2012; Nelson, Kreps, Hesse, Croyle, & Willis, 2004). The survey provides researchers with estimates of the prevalence of cancerrelevant knowledge, attitudes, and behaviors based on a national probability sample. Access to the survey instruments and data is available through the National Cancer Institute (n.d.). HINTS 4 (2012) was used for the analysis in this study. Two cycles of surveys were administered by mail between October 2011 and January 2013. The survey sampling method consisted of a two-stage design. A stratified sample of addresses was first selected from a file of all residential addresses. The second stage sampled all adults either within households (Cycle 1) or one selected adult from each sampled household (Cycle 2) (‘‘Health Information National
Trends Survey 4: Cycle 1 Methodology Report’’ [National Cancer Institute, 2012] and the ‘‘Health Information National Trends Survey 4: Cycle 2 Methodology Report’’ [National Cancer Institute, 2013]). The survey was administered in English or Spanish, with a nominal incentive provided for participation. The mailing protocol used the Dillman approach for total design methodology to increase response rates (Dillman, 2009). These two cycles were combined for a total sample size of 7,589 participants. After restricting the sample to women younger than 50 years of age and excluding cases with missing values for the variables of interest, there were 1,085 cases. Demographic characteristics were comparable between the two cycles. All subsequent analyses included a cycle variable as a covariate to account for differences attributable to data collection time point. For analysis the sample was weighted at the individual subject level and adjusted using a jackknife variance estimation technique to include replicate weights. Sampling weights consist of three components: (a) respondent base weight; (b) adjustment for nonresponse; and (c) calibration adjustment. A full description of the methodologies used in calculating sample weights is published by the National Cancer Institute (2012). This study was reviewed by the Boston University Institutional Review Board and met the criteria for exemption. Variable Specification The primary outcome of interest was reported utilization of mammograms. Responses were collapsed into two categories: never had a mammogram versus ever had a mammogram. This dichotomous outcome was used because given the younger age of the sample, many women may not have yet begun serial screening and may have only just undergone their first mammogram. The key independent variable of interest was whether patients reported having a choice to undergo screening tests (‘‘Has a doctor ever told you that you could choose whether or not to have a mammogram?’’). The authors hypothesized that perceived patient choice is associated with less utilization compared with those not given a choice (Bertakis & Azari, 2011; Stacey, Bennett, & Barry, 2011). A second aspect of SDM, patient–provider communication, was assessed using a composite measure of seven communication questions that elicited from patients how often providers: (a) gave ‘‘you the chance to ask all health-related questions you had’’; (b) gave ‘‘the attention you needed to your feelings and emotions’’; (c) involved ‘‘you in decisions about your health care as much as you wanted’’; (d) made ‘‘sure you understood the things you needed to do to take care of your health’’; (e) explained ‘‘things in a way you could understand’’; (f) spent ‘‘enough time with you’’; and (g) helped ‘‘you deal with feelings of uncertainty about your health or health care.’’ This set of questions used in the measure was combined as an equally weighted sum for all items after a Cronbach alpha demonstrated excellent internal consistency reliability (a ¼ .92). The final measure had a response range from 7 to 28 with
Downloaded by [University of Cambridge] at 09:29 18 August 2015
1062 a higher score denoting better communication between patient and provider. It was hypothesized that better communication would be associated with the patient being provided a choice to undergo screening. Variables were chosen for inclusion based on a SDM framework published by Elwyn and colleagues (2012). Demographic characteristics include gender, race=ethnicity, marital status, age, education, income, health status, and insurance. Prior work indicates that these variables have a significant effect on utilization of health services (Blendon, Aiken, Freeman, & Corey, 1989; Joung, Van Der Meer, & Mackenbach, 1995; Makuc, Breen, & Freid, 1999). Many of these characteristics also influence the likelihood that SDM occurs in the medical encounter (Levinson et al., 2008; Siminoff, Graham, & Gordon, 2006). Two proxy measures of cancer risk were also included: personal cancer history and family cancer history. Statistical Analysis Bivariate associations were evaluated between sociodemographic variables, utilization, choice, and communication. We computed chi-square tests for categorical variables and t tests for continuous variables stratified by two categories and analysis of variance routines for more than two categories. A multivariable logistic regression model with perceived choice as the dichotomized dependent variable was constructed to understand the relation between the two SDM variables. Communication was the key independent variable with sociodemographic variables included as predictors. To test the independent effects of choice and communication on the utilization of screening mammography, a multivariable logistic regression was employed with choice and communication as the two independent variables of interest. The multivariable model adjusted for sociodemographic and cancer history variables. The model goodness-of-fit was assessed using the likelihood ratio chi-square test, the likelihood ratio test, and the C-statistic. For all analyses, statistical significance was set at the .05 level. Analyses were conducted in SAS 9.3 using PROC SURVEYLOGISTIC incorporating sampling weights. Odds ratios, with the associated 95% confidence interval, and p values are reported based on these models.
Results Descriptive statistics are displayed in Table 1. Each variable is presented with the original and weighted sample size and percentages. The mean and standard deviations are reported for continuous variables. The sample was racially representative of the U.S. general population: 68% were non-Hispanic White, 11% Black, 15% Hispanic, and 7% other. Most women were married (53%), had health insurance (87%), and rated their health as good or better (89%). 31.3% (SD ¼ 2.6) of women younger than 50 years of age reported they perceived having a choice to undergo mammogram screening. Participants rated patient–provider
C. M. Gunn et al. Table 1. Characteristics of the sample Variable
n
Race=ethnicity Hispanic 180 Non-Hispanic White 649 Black or African American 177 Other 93 Marital status Married or living as married 619 Divorced, separated or widowed 195 Single 285 Highest level of education Less than high school 53 High school graduate 146 Some college 319 College graduate 581 Income $35,000 369 $35,000–50,000 144 $50,000–75,000 188 >$75,000 398 Health insurance Yes 971 Regular provider Yes 793 General health Good or better 974 Personal history of cancer Yes 62 Family history of cancer Yes 838 Ever had mammogram Yes 515 Choice for mammogram Yes 350 Age, M (SD) 1,099 Communication, M (SD) 1,085
Weighted n
% (SD)
11,402,041 14.5 (2.6) 53,334,702 67.6 (4.0) 8,438,679 10.7 (1.8) 5,747,642 7.3 (1.0) 41,938,801 53.1 (3.8) 6,642,934 8.4 (1.3) 30,341,330 38.4 (2.8) 6,714,994 8.5 (3.4) 11,564,359 14.7 (2.1) 30,420,492 38.5 (2.3) 30,223,220 38.3 (3.6) 28,585,548 9,910,205 13,586,864 26,840,449
36.2 12.6 17.2 34.0
(6.1) (1.9) (1.8) (4.6)
68,584,115 86.9 (1.7) 54,317,629 68.8 (5.6) 70,394,878 89.2 (2.6) 3,212,682
4.1 (0.7)
59,762,147 75.7 (2.4) 30,176,847 38.2 (3.1) 24,663,239 31.3 (2.6) 33.4 (0.4) 22.8 (0.3)
Note. N ¼ 1,085, weighted N ¼ 77,277,150.
communication relatively high with a mean score of 22.8 (SD ¼ 0.3). Approximately 38% (SD ¼ 3.1) had ever had a mammogram performed. Mammogram screening was positively associated with having a choice for mammography (p ¼ .008), being married (p < .0001), having higher income (p ¼ .007), better self-reported health status (p ¼ .008), having a personal cancer history (p ¼ .0005), and having a regular provider (p ¼ .025) (data not shown in table). Increasing age (OR ¼ 1.03; 95% CI: 1.00–1.07), poorer health status (2.51; 1.08–5.87) and having a regular provider (1.69; 1.16–2.47) were significant and increase the odds of a patient perceived choice to undergo mammogram screening (see Table 2). Patients who rated their health status as fair or worse were more than twice as likely to report having a choice than those with better health (2.51; 1.08–5.87). Having a regular provider also increased the odds of reporting a choice by 69% compared
1063
Shared Decision Making and Mammography Table 2. Logistic model predicting patient-reported choice to undergo mammogram screening (yes=no)
Downloaded by [University of Cambridge] at 09:29 18 August 2015
Variable Communication Race (Ref. ¼ non-Hispanic White) Hispanic Black or African American Other Age Marital status (Ref. ¼ single) Divorced, separated, or widowed Married or living as married Education (Ref. ¼ high school graduate) College graduate Less than high school Some college Household income (Ref. < $35,000) $35,000 to
