Psychometric properties of belief measures about osteoporosis and its control

Psychometric properties of belief measures about osteoporosis and its control

Accepted Manuscript Psychometric properties of belief measures about osteoporosis and its control Lacey Alexander, Piyaorn Wajanatinapart, Diane Lauv...

709KB Sizes 0 Downloads 55 Views

Accepted Manuscript Psychometric properties of belief measures about osteoporosis and its control

Lacey Alexander, Piyaorn Wajanatinapart, Diane Lauver PII: DOI: Reference:

S0897-1897(16)30183-5 doi: 10.1016/j.apnr.2017.08.006 YAPNR 50946

To appear in:

Applied Nursing Research

Received date: Revised date: Accepted date:

13 February 2017 4 August 2017 31 August 2017

Please cite this article as: Lacey Alexander, Piyaorn Wajanatinapart, Diane Lauver , Psychometric properties of belief measures about osteoporosis and its control, Applied Nursing Research (2017), doi: 10.1016/j.apnr.2017.08.006

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Psychometric Properties of Belief Measures about Osteoporosis and its Control Lacey Alexander, MS, RNa,c,d, PhD Candidate Piyaorn Wajanatinapart, PhD, RNb, Nursing Instructor Diane Lauver, PhD, RN, FAANa, Professor

c

Faculty of Nursing, Suan Dusit University in Bangkok, Thailand.

Corresponding author, [email protected]

d

AN

US

Jonas Nurse Leader Scholar 2016-2018

Lacey Alexander

M

4120 Signe Skott Cooper Hall

ED

701 Highland Avenue

AC

CE

PT

Madison, WI 53705

IP

b

T

UW-Madison School of Nursing, Cooper Hall, 701 Highland Ave, Madison, WI 53705.

CR

a

ACCEPTED MANUSCRIPT 2 Introduction The risk of osteoporosis (OP) increases with age, with the rate of OP diagnoses peaking at about 85 years of age. At this age, 1 in 3 women and 1 in 10 men have an OP diagnosis (Looker, Borrud, Dawson-Hughs, & Shepherd, 2012). OP is associated with serious consequences, such as fractures,

T

limited daily function, reduced socialization, and death (David et al., 2010).

IP

Preventive behaviors for OP exist, yet people do not often engage in them. Primary preventive

CR

behaviors are those that prevent illness from occurring (Cohen, Chavez, & Chehimi, 2010), such as getting adequate dietary calcium and vitamin D for OP (Gahche et al., 2011; Johnson, Lacher, Pfeiffer,

US

Schleidcher, & Sempos, 2011). Secondary prevention behaviors are those that control illnesses through

AN

early detection and diagnosis, such as bone mineral density (BMD) screening (Cohen et al., 2010). Less than one-third of people who could benefit from BMD screening for OP have had such screening (Lim,

M

Hoeksema, & Sherin, 2009; Curtis et al., 2008). For conciseness, we will use the term “control” to refer

ED

to both primary and secondary prevention behaviors for OP in this paper. Nurse leaders recommended that studying and improving health behaviors be priorities for

PT

nursing science (Henly et al., 2015). Although knowledge about a disease may be necessary to guide

CE

preventive behavior regarding that disease, it is not sufficient. However, beliefs regarding disease and corresponding preventive behaviors can explain or predict preventive behavior. Beliefs can support or

AC

interfere with people’s adoption of preventive behaviors. For example, if people believe that OP is likely and important, and if they believe that preventive measures are useful, then they are more likely to engage in the corresponding preventive behaviors for OP than to avoid them (Lauver, 1992; Skinner, Tiro, & Champion, 2015). Health beliefs are amenable to change (Johnson, 1999; Skinner et al., 2015) and nurses have the essential skills to deliver psycho-educational interventions to address knowledge and beliefs regarding preventive behaviors.

ACCEPTED MANUSCRIPT 3 We had proposed a health promotion study in which we planned to address older adults’ beliefs about OP and its control in order to improve behaviors to control OP. To evaluate predicted changes in beliefs about OP, we sought corresponding measures of beliefs that were based on theories of health behaviors (Johnson, 1999; Ryan & Deci, 2008; Skinner et al., 2015) and had good psychometric properties. We sought measures that reflected a breadth of content and would likely be sensitive to

IP

T

change in our future intervention (Wajanatinapart, Alexander & Lauver, 2014). However, we had difficulty finding such belief measures. The purpose of the current study was to assess content validity

CR

and internal consistency reliability of belief measures regarding behaviors to control OP, based on

US

relevant behavioral theories. Background

AN

From the health belief model (Skinner et al., 2015), we proposed that perceived susceptibility

M

and perceived severity are relevant to health beliefs about OP. Perceived severity refers to one’s beliefs about the severity of a health problem and its possible consequences. Perceived susceptibility refers to

ED

one’s beliefs about the susceptibility of developing a health problem (Skinner et al., 2015). In critical

PT

reviews, researchers have concluded that perceived susceptibility and severity of disease had been correlated positively with corresponding behavior (Skinner et al., 2015); these data offer support for the

CE

construct validity of these concepts.

Scholars have documented that perceived self-efficacy from social cognitive theory (Kelder,

AC

Hoelscher, & Perry, 2015), the health belief model (Skinner et al., 2015), and perceived competence from self-determination theory (SDT; Barsevick & Lauver, 1991; Williams et al., 2006) consistently correlate positively with engaging with a corresponding behavior. On close reading about these concepts (Skinner et al., 2015; Williams et al., 2006) we maintain that perceived self-efficacy and perceived competence reflect the same concept. They both refer to the belief that one can execute certain behaviors to reach an aim, such as eating food that is rich in vitamin D to reduce risks of OP (Kelder et al., 2015; Skinner et al., 2015; Williams et al., 2006). Researchers have correlated positively perceived competence with corresponding behaviors (Teixeira et al., 2012).

ACCEPTED MANUSCRIPT 4 We sought existing, theory-based, belief measures that were relevant to OP and its control but did not find adequate measures for our use. Kim and colleagues (1991) had developed and revised (Gendler et al., 2015) an Osteoporosis Health Belief Scale (OHBS) that measured beliefs such as susceptibility and competence. This scale has had adequate to good internal consistency reliability and construct validity. However, some of these questions reflected dated content and the scale was

IP

T

relatively long at 42 items. Given that nurse researchers likely would want to measure variables in

CR

addition to beliefs, such as knowledge or barriers, a shorter beliefs scale would be desirable for use. We also located the Risk Behavior Diagnosis (RBD) scale, developed by Witte and colleagues

US

(1996), because Dassow and Nayak had used it in their studies of screening, including for OP (Dassow, 2005; Nayak, Roberts, Chang, & Greenspan, 2010). Witte’s scale included perceived susceptibility and

AN

perceived severity of disease and risk behaviors, and self-efficacy about getting appropriate care. On

M

closer examination, Witte and colleagues (1996) had designed the RBD to study beliefs about condom use to prevent human papillomavirus (HPV). They found the RBD scale to have content, construct, and

ED

predictive validity in the context of HPV screening.

PT

When Dassow (2005) and Nayak’s team (2010) had used the RBD scale, they did not report reliability or content validity of their measures in the context of OP. Dassow had revised the RBD scale to

CE

assess women’s beliefs about OP and cancer screening with the assumption that his revised scale would be reliable because the original RBD scale was reliable (Witte et al., 1996). Later, Nayak and colleagues

AC

(2010) used Dassow’s version of the RBD scale to assess beliefs about OP screening. This team justified using Dassow’s version of the RBD scale by citing Witte and team’s (1996) psychometric evaluations and Dassow’s (2005) study.

ACCEPTED MANUSCRIPT 5 We maintained that Dassow’s and Nayak’s team’s assumptions about the applicability of the RBD to OP prevention were problematic. In terms of validity across studies, the bio-psycho-social characteristics of preventive behaviors could differ significantly. he type of belief items relevant to HPV prevention would not necessarily be the same as those for OP control. Preventing HPV involves safer sex behaviors and interpersonal dynamics whereas preventing OP involves getting adequate nutrition (e.g.,

IP

T

Ca and D) which can be done relatively independently of a significant other. In terms of reliability,

CR

Dassow’s and Nayak’s team’s claims are false because reliability is not a constant characteristic of a measure, but varies with the context and sample with which it is used (Devon et al., 2007; Polit & Beck,

US

2006). Researchers cannot generalize about the reliability from one measure across significantly differing situations or contexts.

AN

Thus, we planned a study with one purpose to: (1) assess content validity of perceived

M

susceptibility and perceived severity of OP from an expert sample. In addition, we had purposes to: (2) assess clarity of our questions and (3) describe internal consistency reliability of perceived susceptibility,

ED

perceived severity, and perceived competence about OP control with a sample of older adults. In this

PT

paper, we also discuss issues involved with the adoption of measures from prior research to new

Method

CE

situations without careful consideration about how the contexts could influence the measures.

AC

Design. We used a cross-sectional, descriptive design. We sought one sample, a panel of nurse experts, to assess the content validity of our belief measures about OP and its control. The panel was comprised of five clinicians. Three participants were experienced nurses with graduate degrees. Two of these three were faculty members in a School of Nursing with expertise in medical-surgical nursing and gerontology. One of these worked with older adults in the hospital. A fourth nurse was baccalaureateprepared nurse who worked in the community with older adults. A fifth respondent was a physician in geriatrics. In addition, we recruited a sample of community-dwelling adults to evaluate the clarity of the scales and to respond to scales for our assessment of internal consistency reliability. Settings. For aim 1, regarding content validity, nurse researchers coordinated data collection from experts who worked in a Midwestern university, an associated hospital, and a community

ACCEPTED MANUSCRIPT 6 agency. For aims 2 and 3, regarding clarity and internal consistency reliability, the research team (i.e., faculty, students, and a professor emeritus) invited community-dwelling adults aged 60 and older in a Midwestern state to participate. By design, the research team sought participants from communities that differed in population density (urban, suburban, and rural) and location (at least five different counties) to minimize potential bias that could occur by recruiting participants from only one type of

T

location.

IP

Measures. We had two versions of our questionnaires; one version was for experts and the other

CR

was for older adults. For aim 1, we asked experts to rate measures for content validity. More specifically, we asked them to rate clarity and relevance from 1 ‘not at all’ to 4 ‘extremely.’ If experts rated a set of

US

items less than 3, then we asked experts to provide rationale.

AN

For aim 2, we asked participants in the older adult sample to identify text that was not clear. At the bottom of each page of questions, we asked, “Were there any words or phrases that were not clear for

M

you?” If so, we asked participants to circle the words/phrases that were unclear. If items were unclear or

unreliable (Polit & Beck, 2006).

ED

vague, then the validity of the responses could be threatened and the responses could be inconsistent and

PT

For aim 3, we asked older adults to respond to belief items so we could assess internal consistency

CE

reliability of scales. We describe the response options below about each belief measure. Ideally, we would have assessed content validity first and then incorporated experts’ feedback

AC

in revisions prior to assessing reliability. However, this was not feasible for the following reasons. We had just been funded to implement a funded, psychoeducational program to control OP. Within a few weeks of receiving funding for the program, our physician colleagues suggested that we incorporate belief measures regarding OP control in our program evaluation (Dassow, 2005; Nayak et al., 2010;). We desired to maintain good collaborations with physician partners, yet we did not think these belief measures had reliability and validity for reasons described earlier. We also had to implement the program promptly to meet the expectations of funders. Given this real-life situation, we decided to assess the content validity and internal consistency simultaneously. By doing so, we could have some evaluation of these measures and implement our program shortly thereafter.

ACCEPTED MANUSCRIPT 7 Belief measures. Perceived susceptibility scale. We selected all the items that reflected perceived susceptibility from the 12-item RBD scale used by Dassow (2005) and Nayak and colleagues (2010). Among these three items, one item was, “I am at risk for getting osteoporosis.” By retaining all the relevant items, our findings could be reviewed with prior findings about this scale. We created five new items to broaden

T

the type of content in the scale and be relevant to our planned program. Three new items included the

IP

words “other people my age.” We thought that this phrase could increase consistency with which

CR

respondents would interpret the item as opposed to an item without this phrase. For the new, eight-

US

item scale, respondents used a 4-point scale from ‘strongly disagree’ to ‘strongly agree’. Perceived severity scale. We selected all the items that reflected perceived severity from the

AN

RBD scale. Among these three items, one item was, “I believe osteoporosis is a serious disease.” By retaining all of the relevant items, our findings could be reviewed with prior findings about this scale.

M

We created two new items to broaden the type of content (e.g., beliefs about drug companies), for five

ED

items total. Respondents used a 4-point scale from ‘strongly disagree’ to ‘strongly agree’. Individually and together, the research team reviewed all scales carefully for health literacy and made revisions

PT

accordingly.

CE

Perceived competence scale. We based our measure of perceived competence regarding OP control on a 4-item scale of perceived competence regarding preventive behaviors that individuals

AC

largely can control, such as eating and taking medications (SDT, 2015; Williams et al., 2006). We chose not use items that Dassow (2005) or Nayak and colleagues (2010) identified as perceived self-efficacy because the focus of their items was different from ours. Their items focused on people’s abilities to obtain screening tests, necessitating use of health services, and did not include primary prevention strategies such as calcium or vitamin D intake. We retained wording from previous studies (Self-Determination Theory, 2015) about perceived competence, such as, “I feel confident in my ability to.” We substituted the phrases “to get enough calcium/vitamin D” or “to take in the calcium/vitamin D I need” for terms such as “to maintain a healthy diet” (SDT, 2015). With four items for each of two categories (i.e., calcium and vitamin D), we had eight items. One perceived competence question was, “I feel confident in my ability to get

ACCEPTED MANUSCRIPT 8 enough calcium (from food, drink, or other sources) now and in the future.” Respondents used a 7-point, Likert-type, response rating scale ‘strongly disagree’ to ‘strongly agree.’ By using these responses, we could compare our findings to other studies using the SDT measure of perceived competence. Scholars have demonstrated that this measure has had predictive validity in longitudinal, experimental studies; it has explained preventive health behaviors (Fortier et al., 2012;

T

Williams et al., 2004; Williams et al., 2006). We obtained total scores for this and other belief scales by

IP

averaging results from participants.

CR

Procedure

Our university institutional review board approved this study. For aim one, about content

US

validity, we invited the expert panel to participate by mail. Completion of the questionnaires signaled

AN

their consent. We did not assess the content validity of a proposed, perceived competence scale. One, this measure was based on one of the two theories (SDT) that guided our study. Two, we reasoned that the

M

context in which the prior scale had been evaluated was more similar to, than different from, controlling

ED

OP such as eating or taking medications/supplements. To clarify, Williams and colleagues (2004; 2006) had evaluated this scale in preventive contexts to control hyperlipidemia and diabetes which involved

PT

individuals’ activities such as eating and taking medications. These researchers documented the scale had

2006).

CE

content and predictive validity, as well as high internal consistency (.90; Williams and colleagues (2004;

AC

For our aims about clarity of questions and internal consistency reliability, our research team recruited older adults from community settings that served older adults; 35 (66%) were from centers and 18 (34%) were from personal contact. We excluded people living in long-term care facilities because they would have fewer choices over their preventive behaviors for OP. We invited participants to complete the scales confidentially and stated that we would assume that their completion implied consent. In our consent process, we told volunteers that we sought their input on our questions so we could improve them for later use in research. At the beginning of the set of questionnaires, and at the bottom of each page, we asked older adult participants to circle phrases that were not clear to them. Participants either completed the questionnaires immediately and returned them in person, or, took

ACCEPTED MANUSCRIPT 9 them home to complete and return in postage-paid envelopes. Data analysis We used SPSS for our analyses. We examined the data for missing values and patterns of missing data (Fox-Wasylsyshyn & El-Masri; 2005). Aim 1 for content validity. Polit and Beck (2006) recommend computing content validity indices

T

(CVIs) for clarity and relevance. We computed these by item (I-CVI) and by scale (S-CVI). We combined

IP

expert ratings of ‘2’ or ‘1’ into a category to reflect disagreement and ratings of ‘3’ or ‘4’ into a category

CR

to reflect agreement. Then we summed experts’ scores and divided the sum by the total of experts

US

(N=5). We used > .80 to determine acceptability of I-CVI and S-CVIs; the maximum score was 1.00. If an ICVI or S-CVI were less than .80, then we either revised or omitted the items, while striving to balance

AN

breadth and reliability of the scales.

Aim 2 for clarity of items. We reviewed questionnaires returned by older adult participants for

M

the number of circles that they made. We counted the number of circles by item.

ED

Aim 3 for internal consistency reliability. We assessed the initial reliability of the beliefs scales with Cronbach’s alphas. We used >.80 as an ideal criterion (DeVon et al., 2007; TAN, 2009), while

PT

recognizing that Cronbach alphas of > .70 and < .80 are viewed as acceptable for new measures (Polit

CE

and Beck, 2006). If the Cronbach’s alpha for a scale was < .80, then we considered revising items reliability (e.g., when participants indicated items were not clear). Or, we decided to omit items that

AC

contributed to low reliabilities. If we omitted items, then we re-examined the reliabilities of the revised, shorter scales. We also examined whether we could improve initial Cronbach alphas for the scales by considering experts’ comments on validity. Results Sample description. Among 56 questionnaires administered to older adults, 53 participants returned questionnaires anonymously. Our sample of community-dwelling adults was mostly White (96%) and female (80%) with a mean age of 77 (sd = 7.3) and a range of 28 years; 63% were married and 63% of the sample had an education less than a bachelor’s degree. Ninety-four percent said they had enough resources to meet daily needs. All participants had insurance; 86% (n=44) had Medicare

ACCEPTED MANUSCRIPT 10 while 72.5% had private insurance. Eighty-two percent of the sample denied feeling anxious or depressed. The most frequent health problems they reported were pain or discomfort (65%, n=33), high blood pressure (51%), musculoskeletal issues (47%), and eye conditions (31%). Most reported they had no problems with walking (60%), usual activities (78%), or self-care (92%). Missing data. We had no missing data from the expert sample. Among 53 questionnaires

T

returned by the second sample of older adults, we excluded two questionnaires. This was because less

IP

than half of the items had been answered on these questionnaires. To address our aims, we evaluated

CR

51 questionnaires.

US

We report the missing data by scale here. For the perceived susceptibility scale, we had 8 responses missing among 408 possible responses (8 items x 51 participants), 1.96%. For perceived

AN

severity items, we had 1 response missing among 255 possible responses (5 items x 51 participants), 0.39%. For perceived competence about calcium items, we had 3 responses missing among 204

M

possible responses (4 x 51), 1.47%. For the perceived competence about vitamin D items, we had 7

ED

responses missing among 204 possible responses, 3.43%. The missing data were from participants recruited at four different sites. We did not observe a pattern of missingness in our raw data. Using

PT

Little’s MCAR test, we found that missing data met the assumption of MCAR (Little, 1988). We

CE

substituted missing data with imputations because the amount of missing data was small and we wanted to maximize cases available for analyses (Azur, Stuart, Frangakis & Leaf, 2011; Royston, 2004).

AC

Content validity. For the perceived susceptibility scale, the refined item and scale CVIs from the content experts were summarized. The CVIs for clarity and relevance were high, that is, 1.00 and .96 respectively. The refined item and scale CVIs for the perceived severity measure were high, .93; and acceptable for new measures, .73 respectively. Clarity. We summarized the number of words or phrases that participants circled to indicate the terms could be clearer. We found 7 circled phrases for perceived susceptibility, 3 circled phrases for perceived severity (e.g., “significant” disease) and 16 circled phrases for perceived competence (e.g., “now” and “maintaining”). Also, two people circled a term in the directions for perceived competence questions (e.g., “were”). Given that we had 21 items and 51 respondents, we had 1071 potential

ACCEPTED MANUSCRIPT 11 responses. Considering 26 total words or phrases questioned for clarity among 1071 potential responses, we had a 2.4% rate of words or phrases questioned for clarity. Internal consistency reliability. The initial Cronbach alphas for these measures varied from a low of .40 for perceived severity, to a high of .96, for perceived competence. See Table 2 for the initial and revised reliability data on belief scales. The initial reliability was >.90 for perceived competence so we

T

did not change the number of items. After revisions, the subsequent alphas for perceived severity were

CR

Discussion

IP

acceptable, .79, and high, .88, for perceived susceptibility.

Although OP is common (David et al., 2010), many people do not engage in preventive behaviors

US

to control OP (Curtis et al., 2008). Health beliefs about diseases and corresponding preventive behaviors

AN

are common influences on health behaviors. Importantly, beliefs can be altered by nursing interventions (Johnson, 1999; Skinner et al., 2015).

M

By examining the content validity by experts, clarity by participants, and internal consistency

ED

reliability of belief measures about OP control, we have aligned our research with current priorities for nursing science about prevention (Henly et al., 2015). We also addressed a gap in prior research

PT

because we had been unable to locate psychometrically sound and evidence-based measures of beliefs

CE

regarding OP control. We have extended interdisciplinary research on beliefs about screening and beliefs about OP (Dassow, 2005; Gendler et al., 2015; Kim et al., 1991; Nayak et al., 2010).

AC

By evaluating the content validity and internal consistency reliability for perceived susceptibility and severity of OP we have advanced research by Dassow (2005) and Nayak et al. (2010), in particular. Although these researchers had revised existing scales for new contexts that were different from the context for which original items had been designed, they did not assess the content validity or the reliability for their revised scales. In this paper, we have provided a critique of the ways in which prior researchers have applied former measures to new contexts without careful consideration and logical reasoning for doing so. We note that reliability and validity are not constant characteristics of measures but rather, are dependent upon the context in which measures are used (Devon et al., 2007; Polit & Beck, 2006).

ACCEPTED MANUSCRIPT 12 Regarding aim 1, content validity by experts, we documented high CVIs on clarity and relevance for the revised perceived susceptibility scale. We documented a high CVI on clarity for our revised perceived severity scale but a lower than desired .73 for relevance. In a review of the spontaneous comments made by experts, two experts questioned the relevance of the item about OP being a “significant” disease because this item seemed repetitive of an earlier item about OP being a “serious”

T

disease. These comments indicate that reviewers considered the “set” of items in their response rather

IP

than just a particular item.

CR

Regarding aim 2, clarity by participants, we found limited indications by participants to suggest terms were unclear. This may be explained by the research team striving to have directions and items

US

written at an 8th grade reading level. Or, this rate may be low because participants may have been

AN

focused more on responding to items and less on marking terms that were not clear. Regarding aim 3, we observed high internal consistency reliabilities for perceived susceptibility

M

and perceived competence. Our finding about reliability for perceived competence regarding OP

ED

extends research on this scale, based on SDT (Williams et al., 2006). We also observed good internal consistency reliability for perceived severity. However, our internal consistency reliabilities can be

AC

CE

PT

interpreted cautiously given our modest sample size.

ACCEPTED MANUSCRIPT 13 A strength of our sample is that we did not recruit from a clinic population, which would have inherent biases such as having interest in, or access to, health care. We also acknowledge limitations of our sample. One, it was predominately White, female older adults living in Midwestern communities. Nevertheless, this sample reflects the population from which it was drawn. Two, our sample size was modest for describing internal consistency reliabilities; our reliability estimates may not be stable.

T

We acknowledge a limitation in our design. Optimally, our team would have assessed content

IP

validity and internal consistency reliability in a sequential process. However, we faced competing

CR

priorities to evaluate measures and to implement a newly funded program. We reasoned that evaluating content validity and internal consistency reliability of our measures prior to our planned intervention

US

was preferable to using measures that had not been evaluated for such psychometric properties. Also,

AN

we did not evaluate other psychometric properties such as construct validity or test-retest reliability. A limitation of our measures is that our items did not address physical activity, although activity

M

is related to OP. We did not include items about activity because our intervention was planned with

ED

local community center staff. They did not want our intervention for OP to focus on activity because they had other programs that addressed activity. This is an example of compromising among multiple

PT

partners in community-based research.

CE

Our revised measures are relevant to nursing scholars. Nurses can assess and address adults’ perceptions regarding OP and its control. If clients’ beliefs are inaccurate, then clinicians can provide

AC

clarifying, evidence-based information. Beyond individual clarification, clinicians are in ideal positions to deliver psychoeducational interventions to modify beliefs regarding OP and its control. Our measures are representative of content that nurses would address in a community-based, psychoeducation program about OP and its control. Future researchers could replicate our assessments with a larger and diverse sample. Having a much larger sample size would enabled researchers to conduct additional analyses (e.g. factor analyses) among the sets of items. Researchers can extend our work and assess the stability, such as test-retest reliability, of our refined beliefs measures regarding OP and its control. Researchers could also consider adding psychometrically sound measures about physical activity if their intervention focused on activity

ACCEPTED MANUSCRIPT 14 to control OP. Also, researchers can examine the construct validity of these scales in future studies to evaluate whether they are correlated with behavior and clinical outcomes as expected. Conclusion With feedback from experts and reports from lay older adults, we refined measures of beliefs regarding OP and its control for subsequent consideration for use by nurses and clinical researchers.

T

With data from experts, we documented excellent content validity for perceived susceptibility but less

IP

than desirable content validity for perceived severity. Using data from community-dwelling older

CR

adults, we documented that our belief measures were overall clear and had good to excellent internal

US

consistency reliabilities. Clinicians could consider using our measures if their practice population is similar to ours or collaborating with researchers to replicate this study. In describing our search for

AN

psychometrically sound belief measures regarding OP and its control, we have discussion concerns

AC

CE

PT

ED

M

about using former measures in new contexts without careful considerations and evaluation.

ACCEPTED MANUSCRIPT 15 References Azur, M. J., Stuart, E. A., Frangakis, C., and Leaf. P. J. (2011). Multiple imputation by chained equations: What is it and how does it work? International Journal of Methods in Psychiatric Research. 20(1), 40-49.

T

Babatunde, O. T., Himburg, S. P., Newman, F. L., Campa, A., & Dixon, Z. (2011). Theory-driven

IP

intervention improves calcium intake, osteoporosis knowledge, and self-efficacy in community-

CR

dwelling older Black adults. Journal of Nutrition Education and Behavior, 43(6), 434-440. http://dx.doi.org/10.1016/j.jneb.2010.07.004

US

Barsevick, A., & Lauver, D. (1991). An assessment of the concept of expectation: Its usefulness in nursing

AN

practice. In P. Chinn, (Ed.), An anthology of caring, (pp. 211-223). New York: National League for Nursing.

M

Cohen, L., Chavez, V., & Chehimi, S. (2010). Prevention is primary: Strategies for community well-being (2nd

ED

ed.). San Francisco, CA: Jossey-Bass.Curry, C. L., Hogstel, O. M., Davis, C. G., & Frable, J. P. (2002). Population-based osteoporosis education for older women. Public Health Nursing, 19(6), 460-

PT

469. http://dx.doi.org/10.1046/j.1525- 1446.2002.19607.x

CE

Curtis, J. R., Carbone, L., Cheng, H., Hayes, B., Laster, A., Matthews, R., ... & Delzell, E. (2008). Longitudinal trends in use of bone mass measurement among older Americans, 1999–

AC

2005. Journal of Bone and Mineral Research, 23(7), 1061-1067. http://dx.doi.org/10.1359/jbmr.080232 Dassow, P. (2005). Setting educational priorities for women's preventive health: Measuring beliefs about screening across disease states. Journal of Women's Health, 14(4), 324-330. http://dx.doi.org/10.1089/jwh.2005.14.324 David, C., Confavreux, C. B., Mehsen, N., Paccou, J., Leboime, A., & Legrand, E. (2010). Severity of osteoporosis: What is the impact of co-morbidities?. Joint Bone Spine, 77, S103-S106. http://dx.doi.org/10.1016/s1297-319x(10)70003-8

ACCEPTED MANUSCRIPT 16 Deci, E. L., & Ryan, R. M. (2008). Self-determination theory: A macrotheory of human motivation, development, and health. Canadian Psychology, 49(3), 182-185. http://dx.doi.org/10.1037/a0012801 DeVon, H. A., Block, M. E., Moyle‐Wright, P., Ernst, D. M., Hayden, S. J., Lazzara, D. J., … & Kostas‐

T

Polston, E. (2007). A psychometric toolbox for testing validity and reliability. Journal of Nursing

CR

IP

Scholarship, 39(2), 155-164. http://dx.doi.org/10.1111/j.1547-5069.2007.00161.x Fortier, M. S., Duda, J. L., Guerin, E., & Teixeira, P. J. (2012). Promoting physical activity: Development

US

and testing of self-determination theory-based interventions. International Journal of Behavioral

AN

Nutrition and Physical Activity, 9(1), 20. http://dx.doi.org/10.1186/1479-5868-9-20 Francis, K. L., Matthews, B. L., Van Mechelen, W., Bennell, K. L., & Osborne, R. H. (2009). Effectiveness of

M

a community-based osteoporosis education and self-management course: A wait list controlled

ED

trial. Osteoporosis International, 20(9), 1563-1570. http://dx.doi.org/10.1007/s00198-009-08340

PT

Fox‐Wasylyshyn, S. M., & El‐Masri, M. M. (2005). Handling missing data in self‐report

CE

measures. Research in Nursing & Health, 28(6), 488-495. http://dx.doi.org/10.1002/nur.20100

AC

Gahche, J., Bailey, R., Burt, V., Hughes, J., Yetley, E., Dwyer, J., ... & Sempos, C. (2011). Dietary supplement use among US adults has increased since NHANES III (1988-1994). NCHS Data Brief, (61), 1. Gendler, P. E., Coviak, C. P., Martin, J. T., Kim, K. K., Dankers, J. K., Barclay, J. M., & Sanchez, T. A. (2015). Revision of the osteoporosis knowledge test: Reliability and validity. Western Journal of Nursing Research, 37, 1623-1643.

ACCEPTED MANUSCRIPT 17 http://dx.doi.org/10.1177/0193945914537565 Kelder, S. H., Hoelscher, D., & Perry, C. L. (2015). How individuals, environments and health behaviors interact: Social cognitive theory. In K. Glaz, B, K. Rimer, & K. Viswanath (Eds). Health behavior: Theory, research, and practice. John Wiley & Sons, San Francisco, CA, 75-94.

IP

T

Henly, S. J., McCarthy, D. O., Wyman, J. F., Heitkemper, M. M., Redeker, N. S., Titler, M. G., ... & Conley,

CR

Y. P. (2015). Emerging areas of science: Recommendations for nursing science education from the council for the advancement of nursing science idea festival. Nursing Outlook, 63, 398-407.

US

DOI: http://dx.doi.org/10.1016/j.outlook.2015.04.007

AN

Johnson, C. L., Lacher, D. A., Pfeiffer, C. M., Schleicher, R. L., & Sempos, C. T. (2011). Vitamin D status: United States, 2001-2006. US Department of Health and Human Services, Centers for Disease

M

Control and Prevention, National Center for Health Statistics.

PT

Health, 22(6), 435-448.

ED

Johnson, J. E. (1999). Self‐regulation theory and coping with physical illness. Research in Nursing &

http://dx.doi.org/10.1002/(sici)1098-240x(199912)22:6%3C435::aid-nur2%3E3.0.co;2-q

CE

Kim, K. K., Horan, M. L., & Gendler, P. E. (1991). Osteoporosis knowledge tests, osteoporosis health

AC

belief scale, and osteoporosis self-efficacy scale. Research in Nursing & Health, 14(2), 155-163. http://dx.doi.org/10.1002/nur.4770140210 Lauver, D. (1992). A theory of care-seeking behavior. Journal of Nursing Scholarship, 24(4), 265-271. http://dx.doi.org/10.1111/j.1547-5069.1992.tb00734.x Lauver, D. R., Settersten, L., Kane, J. H., & Henriques, J. B. (2003). Tailored messages, external barriers, and women’s utilization of professional breast cancer screening over time. Cancer, 97(11), 2724-35. http://dx.doi.org/10.1002/cncr.11397 Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association. 83(404),1198-1202.

ACCEPTED MANUSCRIPT 18 Lim, L. S., Hoeksema, L. J., & Sherin, K. (2009). Screening for osteoporosis in the adult US population: ACPM position statement on preventive practice. American Journal of Preventive Medicine, 36(4), 366-375. http://dx.doi.org/10.1016/j.amepre.2009.01.013 Looker, A. C., Borrud, L. G., Dawson-Hughes, B., Shepherd, J. A., & Wright, N. C. (2012).Osteoporosis or low bone mass at the femur neck or lumbar spine in older adults, United States, 2005-2008. US

IP

T

Department of Health and Human Services, Centers for Disease Control and Prevention,

CR

National Center for Health Statistics.

Nayak, S., Roberts, M. S., Chang, C. H., & Greenspan, S. L. (2010). Health beliefs about osteoporosis and

AN

http://dx.doi.org/10.1177/0017896910364570

US

osteoporosis screening in older women and men. Health Education Journal, 69(3), 267-276.

Polit, D. F., & Beck, C. T. (2006). The content validity index: Are you sure you know what’s being

M

reported? Critique and recommendations. Research in Nursing & Health, 29(5), 489-497.

ED

http://dx.doi.org/10.1002/nur.20147

PT

Royston, P. (2004). Multiple imputation of missing values. The Stata Journal. 4(3), 227-241. Self-Determination Theory. (2015, September 20). Retrieved from

CE

http://www.selfdeterminationtheory.org/

AC

TAN, S. (2009). Misuses of KR-20 and Cronbach’s alpha reliability coefficients. Education and Science, 34(152) 101-112.

ACCEPTED MANUSCRIPT 19 Teixeira, P. J., Carraça, E. V., Markland, D., Silva, M. N., Ryan, R. M. (2012). Exercise, physical activity, and self-determination theory: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 9(1), 1-30. http://dx.doi.org/10.1186/1479-5868-9-78 Skinner, C. S., Tiro, J., & Champion, V. L. (2015). The health belief model. In K. Glaz, B, K. Rimer, & K.

T

Viswanath (Eds). Health behavior: Theory, research, and practice. John Wiley & Sons, San

CR

IP

Francisco, CA, 75-94.

Wajanatinapart, P, Alexander, L, & Lauver, D. (2014). Developing measures of perceptions about

US

osteoporosis and its preventive behaviors among older adults. Presented at the Midwest Nursing

AN

Research Society annual conference. St. Louis, MO, 4/28/14.

Williams, G. C., McGregor, H. A., Sharp, D., Kouides, R. W., Levesque, C. S., Ryan, R. M., & Deci, E. L.

M

(2006). A self-determination multiple risk intervention trial to improve smokers’ health. Journal

ED

of General Internal Medicine, 21(12), 1288-1294. http://dx.doi.org/10.1111/j.15251497.2006.00621.x

PT

Williams, G.C., McGregor, H., Zeldman, A., Freedman, Z., & Deci, E. (2004). Testing a self-determination

CE

theory process model for promoting glycemic control through diabetes management. Health

AC

Psychology, 23, 58-66. http://dx.doi.org/10.1037/0278-6133.23.1.58

ACCEPTED MANUSCRIPT 20 Table 1 Content validity indices (CVI) for items by measure by scale for belief measures + ___________________________________________________________________________________________________________________________________________________________________________________

Initial number

Initial item – CVI#

of items

Initial scale – CVIs items

Subsequent number of

Subsequent item – CVI

Subsequent scale - CVIs

T P

___________________________________________________________________________________________________________________________________________________________________________________

I R

Belief Scales

___________________________________________________________________________________________________________________________________________________________________________________

Clarity Perceived

8

0.60-1.00

Relevance 0.40-1.00

Clarity Relevance 0.93

0.85

Clarity 5

5

0.60-1.00

0.40-1.00

0.68

0.68

D E

Severity

0.80-1.00

1.00

0.96

0.80-1.00 0.60-1.00

0.93

0.73

U N

A

M 3

Clarity Relevance

C S

1.00-1.00

Susceptibility

Perceived

Relevance

___________________________________________________________________________________________________________________________________________________________________________________ Footnotes: +N = 5 experts rated items for content validity

#

T P

E C

Reflects the minimum to maximum CVI by item within measures.

C A

ACCEPTED MANUSCRIPT 21 Table 2 Internal consistency reliabilities of belief scales, initially and after revisions (N = 51) ____________________________________________________________________________________________________________________________

Initial number of items Per scale

Initial Cronbach alpha

After revisions number of items per scale

After revisions Cronbach alpha+

T P

____________________________________________________________________________________________________________________________

I R

Belief Scales ____________________________________________________________________________________________________________________________

Perceived Susceptibility

8

.76

6

.88

Perceived Severity

5

.40

3

.79

Perceived Competence~

8

.96

~

~

C S

A

___________________________________________________________________________________________________________________________

Note: +We examined whether we could improve Cronbach alphas by deleting selected items. ~ Authors did not revise this scale.

D E

T P

C A

E C

U N

M

ACCEPTED MANUSCRIPT 22 Table 3 Belief items for perceived susceptibility, severity, and competence scales ______________________________________________________________________________________________________________________________________________________________________

Perceived susceptibility items*

Source of item

Retained

Omitted

______________________________________________________________________________________________________________________________________________________________________

I am too young to have osteoporosis.

New item

X

My chances of falling are high when compared to other people my age.

New item

My chances of breaking a bone are high, compared to other people my age.

New item

X

My chances of getting osteoporosis are high, compared to other people my age.

New item

X

I am at risk of getting osteoporosis. People in the news claim the risks of getting osteoporosis are higher than they are.

(Nayak, et al., 2010)

It is possible that I will get osteoporosis.

(Nayak, et al., 2010)

D E

M

(Nayak, et al., 2010)

X

I R

C S

U N

A

It is likely that I will get osteoporosis.

T P

X

X

X X

______________________________________________________________________________________________________________________________________________________________________

Perceived severity items*

T P

Source of item

Retained

Omitted

______________________________________________________________________________________________________________________________________________________________________

If I were to have osteoporosis then I would be stooped over.

E C

I believe osteoporosis is a serious disease. I believe osteoporosis is a severe disease. I believe osteoporosis is a significant disease.

C A

Drug companies have focused on osteoporosis mostly to sell their products.

New item

X

(Nayak, et al., 2010)

X

(Nayak, et al., 2010)

X

(Nayak, et al., 2010)

X

New item

X

______________________________________________________________________________________________________________________________________________________________________

Perceived competence** Two sets of four items; one about calcium, one about vitamin D

Source of item

Retained

Omitted

______________________________________________________________________________________________________________________________________________________________________

I feel confident in my ability to get enough calcium/vitamin D (from food, drink, or other sources) now and in the future.

(Self-Determination Theory, 2015)

X

ACCEPTED MANUSCRIPT 23

I now feel capable to take in the calcium/vitamin D I need (from food, drink, or other sources) over time.

(Self-Determination Theory, 2015)

X

I am able to keep on getting calcium/vitamin D in my diet over the long term.

(Self-Determination Theory, 2015)

X

I am able to meet the challenge of taking in the calcium/vitamin D I need from food, drink, or other sources.

(Self-Determination Theory, 2015)

X

T P

I R

______________________________________________________________________________________________________________________________________________________________________

Note:

*Scale: Strongly disagree (1) to strongly agree (4) **Analogous to self-efficacy; Scale: Not all true (1) to very true (7)

C S

A

U N

D E

T P

C A

E C

M