This article was downloaded by: [Saint Joseph's University]On: 26 August 2013, At: 17:11Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Women & HealthPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/wwah20

Development of a Shorter Version of theOsteoporosis Knowledge Assessment ToolIvana Tadic a , Dejan Stevanovic b , Ljiljana Tasic a & NadaVujasinovic Stupar ca Department of Social Pharmacy and Pharmaceutical Legislation,University of Belgrade Faculty of Pharmacy, Belgrade, Serbiab Department of Psychiatry, General Hospital Sombor, Sombor, Serbiac Institute of Rheumatology, University of Belgrade School ofMedicine, Belgrade, SerbiaAccepted author version posted online: 06 Dec 2011.Publishedonline: 10 Feb 2012.

To cite this article: Ivana Tadic , Dejan Stevanovic , Ljiljana Tasic & Nada Vujasinovic Stupar (2012)Development of a Shorter Version of the Osteoporosis Knowledge Assessment Tool, Women & Health,52:1, 18-31, DOI: 10.1080/03630242.2011.635246

To link to this article: http://dx.doi.org/10.1080/03630242.2011.635246

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 tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand 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 Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial 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

Women & Health, 52:18–31, 2012

Copyright © Taylor & Francis Group, LLCISSN: 0363-0242 print/1541-0331 online

DOI: 10.1080/03630242.2011.635246

Development of a Shorter Versionof the Osteoporosis Knowledge

Assessment Tool

IVANA TADIC, MPharmDepartment of Social Pharmacy and Pharmaceutical Legislation, University of Belgrade

Faculty of Pharmacy, Belgrade, Serbia

DEJAN STEVANOVIC, MDDepartment of Psychiatry, General Hospital Sombor, Sombor, Serbia

LJILJANA TASIC, PhDDepartment of Social Pharmacy and Pharmaceutical Legislation, University of Belgrade

Faculty of Pharmacy, Belgrade, Serbia

NADA VUJASINOVIC STUPAR, PhDInstitute of Rheumatology, University of Belgrade School of Medicine, Belgrade, Serbia

The aim of the authors of this study was to develop a short ver-

sion of the Osteoporosis Knowledge Assessment Tool to be used for

the target population of young adult Serbian females as an eas-

ily implemented add-on questionnaire. The 20-item Osteoporosis

Knowledge Assessment Tool was translated and culturally adapted

using the Principles of Good Practice for the Translation and Cul-

tural Adaptation Process for Patient-Reported Outcomes Measures.

The validation study was conducted on a sample of 250 female

students studying at the Faculty of Pharmacy at the University

of Belgrade, during a two-month period (November–December

2010). The difficulty index, item-total correlations, and internal

consistency were calculated first. Afterward, confirmatory factor

analysis was applied to test the structure of the Osteoporosis Knowl-edge Assessment Tool models and develop a short version. The

mean total Osteoporosis Knowledge Assessment Tool score was 8.31

(possible range 0–20). The confirmatory factor analysis fit indices

indicated poor fit of the data to the originally hypothesized struc-

Received July 16, 2011; revised October 16, 2011; accepted October 21, 2011.The authors declare that no conflicting financial interests exist.Address correspondence to Ivana Tadic, MPharm, Department of Social Pharmacy and

Pharmaceutical Legislation, University of Belgrade Faculty of Pharmacy, Vojvode Stepe 450,11000 Belgrade, Serbia. E-mail: ivana.tadic@pharmacy.bg.ac.rs

18

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 19

ture. The confirmatory factor analysis fit indices, difficulty indices,

and content validity allowed trimming of the original model and

development of a short version with nine items. The average chi-

square value for the Osteoporosis Knowledge Assessment Tool short

version was 31.79 (p D 0.240, SE D 0.176) with Bollen–Stine

bootstrap p D 0.249, Tucker–Lewis Index D 0.925, Comparative

Fit Index D 0.944 and Root Mean Square Error of Approxima-

tion D 0.027. The Osteoporosis Knowledge Assessment Tool thus

had acceptable characteristics and may be used for osteoporosis

knowledge assessment.

KEYWORDS knowledge, osteoporosis, attitudes

INTRODUCTION

During the past few decades, the prevalence of osteoporosis has increased.Therefore, health care professionals, as well as the general population, needto be more aware of this disease and its consequences (Pande et al., 2000).Besides the provision of adequate health care and accessibility to availabletreatments, education about osteoporosis is important not only for patientsbut also for healthy people (Cadarette et al., 2007). With increased knowl-edge of osteoporosis through appropriate education, preventive behaviorcan be implemented, and various risk factors for osteoporosis can be reduced(Pande et al., 2000; Werner, 2005). Education about osteoporosis and devel-opment of preventive behavior is especially important for women, becauseosteoporosis occurs more frequently among women than among men. Thus,it is essential that education and osteoporosis prevention begin as early asin adolescence.

Different questionnaires have been used to evaluate different aspectsof osteoporosis knowledge, including: prevention, risk factors, treatment,disease characteristics, and its consequences. Knowledge about osteoporosishas been evaluated in different populations: adolescents (Anderson, Chad, &Spink, 2005), young adults (Chan et al., 2007), women of different ages andeducation levels (Baheiraei et al., 2005; Drozdzowska, Pluskiewicz, & Skiba,2004; Kasper et al., 1994; Ailinger, Lasus, & Braun, 2003), men (Gaines &Marx, 2011), nurses (Ailinger & Emerson, 1998; Zhang & Chandran, 2011),and pharmacists (Lai et al., 2008). Because osteoporosis is most prevalentamong women, different tools have been developed and tested in womenof different age groups: from adolescence through middle age (OsteoporosisKnowledge Test—OKT (Chan et al., 2007; Gammage et al., 2009; Baheiraeiet al., 2005)), postmenopausal women (Osteoporosis Questionnaire—OPQ(Pande et al., 2000), Malaysian Osteoporosis Knowledge Toll—MOKT (Laiet al., 2008) and Osteoporosis and You (Cadarette et al., 2007)), and com-

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

20 I. Tadic et al.

bining groups older than 22 years of age (Facts about Osteoporosis Quiz—FOOQ (Ailinger, Lasus, & Braun, 2003; Ailinger, & Emerson, 1998)). TheOsteoporosis Knowledge Assessment Tool (OKAT) was developed for theAustralian female population ages 25–44 years, and covers all aspects ofosteoporosis knowledge mentioned above (Winzenberg et al., 2003).

Knowledge about osteoporosis has not been studied in Serbia. One ofthe main reasons for this is the lack of validated questionnaires. As a part of amulti-center project initiated in Serbia in 2010, concerning different aspectsof osteoporosis, the authors examined knowledge about and presence ofthe osteoporosis risk factors in young women. For this project an easilyimplemented, add-on questionnaire was needed for screening purposes.The OKAT was found to be suitable because it covered most aspects ofosteoporosis knowledge. Thus, the aim of the authors in this study was todevelop a valid but short version of the OKAT for young Serbian women.

METHODS

Participants

Twenty female students participated in the questionnaire translation and cul-tural adaptation process as volunteers. They were not part of the validationstudy group.

For the validation step, a convenience sample of female students study-ing at the Faculty of Pharmacy at the University of Belgrade was used.Using the criterion that at least 10 participants per item in a questionnaireis a sufficient number for factor analysis (Fayers & Machin, 2007), at least200 participants were needed, because the OKAT had 20 items. All of theparticipants had to meet the eligibility criteria (female; 18–25 years of age;ability to read and write Serbian). The principal author randomly selected300 female students using the undergraduate students’ registry. The studentswere selected through a randomized cluster sampling. Sixty students wereselected from each academic year. The students were contacted during reg-ular lecture attendance, and details of the study were explained to them. Ofall contacted students, 250 agreed to participate (83.33% response rate), andprovided signed informed consent. All of the participants were evaluated foreligibility by the principal author. The principal author checked the age andnationality of the students by reviewing the students’ identification cards.The ability to speak, understand, and read Serbian was evaluated throughinterview. All of the participants who agreed to participate met the eligibil-ity criteria. The participants anonymously completed the self-administeredquestionnaire under the authors’ supervision in prescheduled groups. Theparticipants were sitting apart from each other and were asked to answer allquestions being given a maximum of 20 minutes to do so. After completion,

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 21

the questionnaires were reviewed by the authors for completeness. All 250questionnaires were completely filled out (all questions were answered).Data collection took place during November and December of 2010.

The study was approved by the Ethics Committee of the Faculty ofPharmacy at the University of Belgrade.

OKAT Questionnaire

The OKAT covers knowledge related to preventive behavior, risk factors,treatment, and consequences of osteoporosis. It is a 20-item questionnaire,and every item has three possible answers: true, false, and do not know. Acorrect response was scored as 1 and incorrect (as well as do not know) as 0.The final score was calculated by summing the number of correct answers.The total possible range of scores was from 0 (the lowest level of knowledge)to 20 (the highest level of knowledge). The OKAT questionnaire has goodpsychometric characteristics, and all items are supported by one underlyingfactor-knowledge (Winzenberg et al., 2003).

Translation and Cultural Adaptation

of the OKAT Questionnaire

The translation and cultural adaptation of the OKAT were performed inseveral steps, according to the Principles of Good Practice Translation andAdaptation of Patient Reported Outcomes Measures developed by the Inter-national Society for Pharmacoeconomics and Outcomes Research (ISPOR)(Wild et al., 2005). Using these principles, the authors aimed to achieve asemantic, idiomatic, experiential, and conceptual equivalence between theoriginal and the Serbian version. The first step was preparation. Permissionto use the OKAT was obtained from its developer, T. M. Winzenberg. Thesecond step was forward translation. Two translators made independentforward translations, ‘‘T1’’ and ‘‘T2’’ (English–Serbian). Both translators (na-tive Serbian speakers) were familiar with the topic and research concept.The third step was reconciliation of these two translations in version ‘‘T12’’through discussion between the two translators and the project manager.During backward translation (the fourth step), two translators translated theSerbian version ‘‘T12’’ back into English. The fifth step included a backtranslation review to compare the conceptual equivalence between the orig-inal OKAT questionnaire and back translations. In the sixth step (cognitivedebriefing), the questionnaire was tested on 20 students in the presence oftwo researchers. During several semi-structured interviews, students wereasked to comment on the simplicity, clarity, and relevance of the questions.This testing was performed to check the interpretation, clarity, and under-standability of every item. All items in the questionnaire were felt to becomprehensive, precise, and relevant for osteoporosis assessment according

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

22 I. Tadic et al.

to the students’ opinions, so they were left unchanged, and no item wasadded, replaced, or omitted. The final Serbian version of the questionnaire(named the OKAT-S) was evaluated for technical errors.

After completion of the above steps, a panel meeting of all of the authorswas organized to verify the content of every item. Content validity wasperformed by reviewing whether the items corresponded to the conceptualdefinition of the questionnaire (Fayers & Machin, 2007). Additionally, thecontent of an item was considered inappropriate in the following instances:the wording was not precise, it was confusing or misleading, or it requiredfrom a participant to have a specific clinical experience with osteoporosis.If the content of an item appeared inappropriate (by consensus of all theauthors), the item was a candidate for exclusion during the development ofthe short version of the questionnaire.

After the translation process, a final testing was performed on the samplestudy group.

Statistical Analysis

Descriptive statistics included mean, standard deviation (SD), and difficultyindex for all items. The difficulty index was estimated as the number of cor-rect responses divided by the total number of responses. A question with anindex of difficulty of 0.75 or higher indicated that in 75% or more responsesthe question was answered correctly, which means that the question wasmost likely too easy (Winzenberg et al., 2003). Conversely, the difficultyindex of 0.10 or lower indicated that in at least 10% of responses the questionwas answered correctly, showing that the question was too hard.

Item-total correlations were explored using Spearman’s correlation coef-ficient between the items and the total score. Items should have substantiallycorrelated to the underlying measured concept (Stevanovic et al., 2009). Ifthe correlation coefficient was low (<0.2), the item was a candidate forexclusion during the development of the short version of the questionnaire.

Internal consistency of the questionnaire was assessed using Cronbach’scoefficient. Additionally, a procedure in which every item was omitted fromthe questionnaire was also used to test further the internal consistency ofthe scale.

Construct validity was assessed using factor analysis. Factor analysis teststhe patterns amongst the correlations (exploratory factor analysis—EFA) orhow well the data fit the hypothesized questionnaire structure (confirmatoryfactor analysis—CFA) (Fayers & Machin, 2007). CFA is preferred over EFAin validation studies when a prior theoretical model is present. Additionally,CFA allows model modifications as well as finding the most suitable modelwhile revising or shortening the questionnaire (Fayers & Machin, 2007).This study used CFA to evaluate the model with one underlying factorsuggested in the original study (Winzenberg et al., 2003), and to find a

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 23

shorter version of the questionnaire (OKAT-S-short). For a short version,the authors aimed to find a model that best fitted the data if a smallernumber of items were present. To compare model fits, likelihood tests wereperformed. As deviation of the data from a normal distribution could affectthe results of maximum likelihood tests and values of standard errors, theBollen–Stine bootstraps method and associated tests of overall model fit wereused to test the effects of non-normality in the underlying database (Ste-vanovic, 2009). When the model was established, goodness of fit betweenthe hypothesized models and the sample data were tested and confirmedby several parameters: Chi-square, Tucker–Lewis Index (TLI), ComparativeFit Index (CFI), and Root Mean Square Error of Approximation (RMSEA).As the Chi-square test depends on several factors (sample and model size,distribution of variables, and omitted variables), other indexes were alsocalculated (Stevanovic, 2009). The model was considered acceptable if theTLI and CFI were >0.9 and excellent if TLI and CFI were >0.95. RMSEA hadto be <0.08 for the model to be acceptable and <0.05 for the model to beexcellent (Stevanovic, 2009; Hu & Bentler, 1998). Factor loadings for itemsthat were lower than 0.2 could indicate inappropriate correlations betweenthe items and underlying factor ‘‘knowledge.’’ Therefore, this was a criterionfor item exclusion. All analyses were conducted using Analysis of MomentStructure Version 7 (AMOS 7).

RESULTS

The OKAT-S questionnaire was completed by 250 female students (meanage 21.29 years, SD D 1.83 years). The mean knowledge score was 8.31(SD D 2.85) out of a total possible score of 20. It took about three minutesfor the participants to complete the questionnaire.

From the content validation step, it was determined that the OKAT-Spossessed culturally appropriate items with sufficient context for measuringknowledge about osteoporosis. Only two items (numbers 11 and 20) pos-sessed insufficient content characteristics to assess knowledge. Item 11 (‘‘Itis easy to tell whether I am at risk of osteoporosis by my clinical risk factors’’)was found to be inappropriate, because it assumed that a person alreadyknew what the clinical osteoporosis risk factors were. Item 20 (‘‘There are

no effective treatments for osteoporosis available in Serbia’’) was deemedinappropriate, because it assumed that a person previously knew what aneffective treatment was and availability of such treatment in Serbia. Both ofthese items required specific clinical knowledge about osteoporosis.

The difficulty indices for items 1 and 7 were higher than 0.75 (0.97 and0.79, respectively), and for items 10 and 13 they were extremely low (0.09and 0.07, respectively) (Table 1). Twelve items had item/total correlationshigher than or equal to 0.20. Cronbach’s alpha for the OKAT-S questionnaire

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

24 I. Tadic et al.

TABLE 1 Basic Psychometric Characteristics of the OKAT-S

ItemIndex ofdifficulty

Item-totalcorrelation

Cronbach’salphaa

1. Osteoporosis leads to an increased riskof bone fractures

0.97 0.34 0.56

2. Osteoporosis usually causes symptoms(e.g., pain) before fractures occur

0.30 0.22 0.56

3. Having a higher peak bone mass at theend of childhood gives no protectionagainst the development ofosteoporosis in later life

0.28 0.16 0.57

4. Osteoporosis is more common in men 0.74 0.22 0.565. Cigarette smoking can contribute to

osteoporosis0.54 0.25 0.55

6. White women are at highest risk offracture as compared to other races

0.18 0.20 0.56

7. A fall is just as important as low bonestrength in causing fractures

0.79 0.20 0.56

8. By age 80, the majority of women haveosteoporosis

0.63 0.25 0.55

9. From age 50, most women can expectat least one fracture before they die

0.37 0.15 0.57

10. Any type of physical activity isbeneficial for osteoporosis

0.09 0.06 0.58

11. It is easy to tell whether I am at risk ofosteoporosis by my clinical risk factors

0.26 0.13 0.57

12. Family history of osteoporosis stronglypredisposes a person to osteoporosis

0.71 0.26 0.55

13. An adequate calcium intake can beachieved from two glasses of milk aday

0.07 0.09 0.57

14. Sardines and broccoli are good sourcesof calcium for people who cannot takedairy products

0.56 0.12 0.57

15. Calcium supplements alone canprevent bone loss

0.22 0.17 0.56

16. Alcohol in moderation has little effecton osteoporosis

0.38 0.20 0.56

17. A high salt intake is a risk factor forosteoporosis

0.20 0.01 0.59

18. There is a small amount of bone loss inthe 10 years following the onset ofmenopause

0.39 0.30 0.54

19. Hormone therapy prevents furtherbone loss at any age after menopause

0.38 0.23 0.55

20. There are no effective treatments forosteoporosis available in Serbia

0.25 0.27 0.55

Note. aCronbach’s alpha if item was deleted.

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 25

was 0.57. When any of the items were deleted, the alpha coefficient did notchange significantly (range 0.54–0.59), all of which indicated relatively poorinternal consistency.

The first tested model was the original hypothesized structure of 20items. The average chi-square value from the 2000 bootstrap samples was268.73 (p < 0.001, SE D 0.609) with Bollen–Stine bootstrap p D 0.002, TLI D

0.538, CFI D 0.587, and RMSEA D 0.048. Factor loadings ranged from �0.012to 0.421.

The second tested model in CFA included 14 items. Six items wereomitted from the first model: items 1 and 7 had a high proportion of correctresponses indicated on the difficulty indices, and items 10 and 13 had lowproportions of correct responses on the difficulty indices, while items 11 and20 were considered inappropriate (according to the translation step). CFAwas performed again on the second model that excluded these six items. Thesecond model showed better characteristics than the first one. The averagechi-square from the 2000 bootstrap samples was 115.216 (p D 0.003, SE D

0.341), with Bollen–Stine bootstrap p D 0.021, TLI D 0.637, CFI D 0.693, andRMSEA D 0.045. Factor loadings ranged from �0.049 to 0.442. The secondmodel showed that items 3, 9, 14, 15 and 17 had factor loadings lower than0.2 which met the criteria for item exclusion (Table 2).

CFA was performed after every sequentially excluded item. Eachsubsequent model showed better characteristics than the previous one(Table 3).

The last (seventh) CFA model had the best possible characteristics. Thismodel consisted of nine items: 2, 4, 5, 6, 8, 12, 16, 18, and 19. In this model,the average chi-square value from the 2000 bootstrap samples was 31.79 (p D

0.240, SE D 0.176), values of TLI and CFI were over 0.90, and RMSEA waslower than 0.05 which indicated acceptable data fit to the unidimensionalstructure (single-factor model) (Table 3). Factor loadings were higher than0.2, which is acceptable, and ranged from 0.27 to 0.42 (Figure 1). Thus,a short version of the OKAT in Serbian that was developed in this studyincluded nine items (items 2, 4, 5, 6, 8, 12, 16, 18, and 19), with Cronbach’salpha coefficient of 0.55. The mean knowledge score calculated from thenine-item version of questionnaire was 4.25 (SD D 1.95), which is 47.20% ofthe maximum possible score.

DISCUSSION

This is the first study in Serbia of which the authors are aware that evalu-ated a questionnaire about knowledge regarding osteoporosis in the youngadult female population. This was also the first known study that includedpharmacy students. The study population will become future primary healthcare professionals who will be in a position to educate patients about the

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

26 I. Tadic et al.

TABLE 2 Factor Loadings of Items by CFA Models

CFA Model (excluded item)

Item 1st

2nd(1, 7, 10,

11, 13, 20)3rd(17)

4th(14)

5th(3)

6th(9)

7th(15)

1. Osteoporosis leads to an increasedrisk of bone fractures

0.42

7. A fall is just as important as lowbone strength in causing fractures

0.24

10. Any type of physical activity isbeneficial for osteoporosis

0.07

11. It is easy to tell whether I am atrisk of osteoporosis by my clinical riskfactors

0.14

13. An adequate calcium intake canbe achieved from two glasses of milka day

0.10

20. There are no effective treatmentsfor osteoporosis available in Serbia

0.38

17. A high salt intake is a risk factorfor osteoporosis

�0.01 �0.05

14. Sardines and broccoli are goodsources of calcium for people whocannot take dairy products

0.09 0.05 0.06

3. Having a higher peak bone massat the end of childhood gives noprotection against the development ofosteoporosis in later life

0.19 0.14 0.14 0.14

9. From age 50, most women canexpect at least one fracture beforethey die

0.17 0.17 0.18 0.17 0.16

15. Calcium supplements alone canprevent bone loss

0.18 0.18 0.19 0.19 0.19 0.18

2. Osteoporosis usually causes symp-toms (e.g., pain) before fractures oc-cur

0.34 0.30 0.29 0.29 0.29 0.31 0.33

4. Osteoporosis is more common inmen

0.38 0.33 0.33 0.33 0.34 0.37 0.38

5. Cigarette smoking can contribute toosteoporosis

0.34 0.35 0.36 0.36 0.35 0.35 0.34

6. White women are at highest risk offracture as compared to other races

0.25 0.28 0.29 0.29 0.29 0.28 0.27

8. By age 80, the majority of womenhave osteoporosis

0.34 0.38 0.37 0.37 0.39 0.39 0.40

12. Family history of osteoporosisstrongly predisposes a person to os-teoporosis

0.35 0.32 0.33 0.33 0.32 0.30 0.27

16. Alcohol in moderation has littleeffect on osteoporosis

0.32 0.34 0.34 0.35 0.34 0.33 0.33

18. There is a small amount of boneloss in the 10 years following theonset of menopause

0.41 0.44 0.44 0.44 0.44 0.44 0.42

19. Hormone therapy prevents furtherbone loss at any age after menopause

0.33 0.35 0.34 0.34 0.35 0.36 0.36

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 27

TABLE 3 CFA Models’ Characteristics

CFAa

model(number

of items)

Excluded items

(from the 1st CFAmodelb and every

previous model) Chi-square

Bollen-

Stinebootstrap

p TLIc CFId RMSEAe

1st (20) 268.734(p < 0.001)

0.002 0.538 0.587 0.048

2nd (14) 1. Osteoporosis leads to anincreased risk of bone fractures

115.216(p D 0.003)

0.021 0.637 0.693 0.045

7. A fall is just as important as lowbone strength in causing fractures

10. Any type of physical activity isbeneficial for osteoporosis

11. It is easy to tell whether I am atrisk of osteoporosis by my clinical

risk factors13. An adequate calcium intake can

be achieved from two glasses ofmilk a day

20. There are no effective treatmentsfor osteoporosis available in Serbia

3rd (13) 17. A high salt intake is a risk factorfor osteoporosis

80.575(p D 0.092)

0.159 0.817 0.848 0.031

4th (12) 14. Sardines and broccoli are goodsources of calcium for people who

cannot take dairy products

69.283(p D 0.079)

0.132 0.818 0.851 0.034

5th (11) 3. Having a higher peak bone massat the end of childhood gives no

protection against the developmentof osteoporosis in later life

57.147(p D 0.088)

0.122 0.834 0.867 0.035

6th (10) 9. From age 50, most women canexpect at least one fracture before

they die

43.576(p D 0.152)

0.167 0.88 0.907 0.031

7th (9) 15. Calcium supplements alone can

prevent bone loss

31.794

(p D 0.24)

0.249 0.925 0.944 0.027

Note.aCFA D Confirmatory Factor Analysis.bThe first CFA model consists of following items: (1) Osteoporosis leads to an increased risk of bone

fractures; (2) Osteoporosis usually causes symptoms (e.g., pain) before fractures occur; (3) Havinga higher peak bone mass at the end of childhood gives no protection against the development of

osteoporosis in later life; (4) Osteoporosis is more common in men; (5) Cigarette smoking can contributeto osteoporosis; (6) White women are at highest risk of fracture as compared to other races; (7) A fall

is just as important as low bone strength in causing fractures; (8) By age 80, the majority of womenhave osteoporosis; (9) From age 50, most women can expect at least one fracture before they die;

(10) Any type of physical activity is beneficial for osteoporosis; (11) It is easy to tell whether I am atrisk of osteoporosis by my clinical risk factors; (12) Family history of osteoporosis strongly predisposes

a person to osteoporosis; (13) An adequate calcium intake can be achieved from two glasses of milk aday; (14) Sardines and broccoli are good sources of calcium for people who cannot take dairy products;

(15) Calcium supplements alone can prevent bone loss; (16) Alcohol in moderation has little effect onosteoporosis; (17) A high salt intake is a risk factor for osteoporosis; (18) There is a small amount of

bone loss in the 10 years following the onset of menopause; (19) Hormone therapy prevents further boneloss at any age after menopause; and (20) There are no effective treatments for osteoporosis available in

Serbia.cTLI D Tucker Lewis Index.dCFI D Comparative Fit Index.eRMSEA D Root Mean Square Error of Approximation.

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

28 I. Tadic et al.

FIGURE 1 Final CFA model for the OKAT-S-short.

importance of preventive behavior. Therefore, they need to have adequateknowledge about osteoporosis. However, the mean knowledge score in-dicated that the female students were not familiar with osteoporosis. The20-item OKAT-S mean total correct score was only 41.55% of the maximumscore. The finding were similar to a study that included women between25–44 years of age, with almost half of them having a university degree(45%), and whose level of knowledge was 44% of the maximum correct score(Winzenberg et al., 2003). To determine if the OKAT score was influencedby ‘‘difficulties’’ with the items, the difficulty index was calculated for eachitem. Items 1 and 7 were too easy to answer, in contrast to items 10 and13 that were too difficult. The original study indicated that the easy itemswere 1, 4, and 10 (Winzenberg et al., 2003). Items 1, 7, 10, and 13 couldfalsely increase/decrease the total score, so they were removed from thequestionnaire.

Results of content validity pointed out that items 11 and 20 did not relateto the content of the questionnaire. Both of these items required previousspecific clinical experience about osteoporosis, so they were excluded.

In the original study, an item-total correlation was considered to begood if it was higher than 0.20 (Winzenberg et al., 2003). In this study, eight

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 29

items (3, 9, 10, 11, 13, 14, 15, and 17) had item-total correlation coefficientsbelow 0.20, which might indicate that these items did not accurately reflectthe latent construct—osteoporosis knowledge. Furthermore, some of theseitems also showed inadequate characteristics in CFA analysis (factor loadingslower than 0.20). All of them were subsequently excluded. The value ofCronbach’s alpha (0.57) for the OKAT-S was poor and lower than the valueof the original questionnaire (0.69) (Winzenberg et al., 2003). When anyof the items were deleted, the alpha coefficient remained almost the same(range: 0.54–0.59), indicating relatively poor internal consistency.

The results of maximum likelihood tests were much better for the finalOKAT-S-short questionnaire compared with OKAT-S which was the basis forrejecting the original model (Floyd & Widaman, 1995). As chi-square dependson sample size, TLI, CFI, and RMSEA were calculated as parameters ofgoodness of fit statistics. TLI and CFI values also confirmed the need to rejectthe first model. However, the RMSEA value indicated that the model wasexcellent. All of the goodness of fit indices had certain disadvantages. A con-clusion should not be drawn based on the value of one single index. To avoidmisinterpretation, the authors followed the recommendations for the two-index presentation strategy and used the cut-off values for the recommendedfit indices (Hu & Bentler, 1998). Based on the results of statistical tests,content validity and values of factor loadings (CFA), the authros trimmedthe first model and checked whether more than one construct existed inthe questionnaire. Six additional models were tested until the final 9-itemversion of OKAT was arrived at. This one-factor model had satisfactoryvalues of all the examined fit indices and factor loadings values. When theconstruct of the final OKAT-S-short version was looked at, six items exploredthe recognition of risk factors, two items related to the consequences ofosteoporosis, and one to the osteoporosis treatment. Five items excludedfrom the original OKAT questionnaire were related to preventive behavior(items 3, 14, and 15 were excluded because of low factor loadings anditems 10 and 13 because of their low difficulty index). Other studies thathave examined knowledge of osteoporosis risk factors used questionnairesthat additionally examined beliefs and preventive behavior for osteoporosis(Anderson, Chad, & Spink, 2005; Chan et al., 2007; Sedlak, Doheny, & Jones,2000). Although the authors have shown that OKAT-S-short has both goodpsychometric characteristics and measures knowledge about osteoporosis,knowledge was related mainly to the risk factors. Future studies shouldinclude testing of preventive behavior to obtain a comprehensive pictureof osteoporosis knowledge.

Several limitations in this study were identified. First, participants wererecruited from only one study site, and only included students studyingpharmacy. Therefore, the results cannot be generalized to other populations.Second, ‘‘false’’ and ‘‘do not know’’ responses were scored in the same way(as 0) as they both reflected lack of knowledge. Therefore, the scale ranked

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

30 I. Tadic et al.

only two levels of knowledge, and the statistical analysis took into accountonly two options: correct and incorrect responses. Third, the Cronbach’salpha values of the first and the last (short) versions of the OKAT were lowerthan 0.7, which indicates relatively poor internal consistency. However, thefit indices of the nine-item model indicated that the fit of the data wasexcellent. The authors accepted the nine-item model as the best, although theCronbach’s alpha value remained similarly poor as in the first model. Fourth,no comparable knowledge assessment tool was used to evaluate convergentvalidity; test–retest was not tested. As a consequence, the temporal stabilityof the version was not reported, as well as its responsiveness. The resultscould only be compared with one study, which previously used the OKATquestionnaire (Winzenberg et al., 2003). This could be a limitation for cross-cultural comparisons.

In summary, for the young Serbian female population, a short versionof the OKAT-S showed better psychometric characteristics than the longerversion. The 9-item developed version (OKAT-S-short) has acceptable char-acteristics and it may be used for further osteoporosis knowledge assess-ments.

REFERENCES

Ailinger, R. L., and J. Emerson. 1998. Women’s knowledge of osteoporosis. Appl

Nurs Res 11:111–4.Ailinger, R. L., H. Lasus, and M. A. Braun. 2003. Revision of the facts on osteoporosis

quiz. Nurs Res 52:198–201.Anderson, K. D., K. E. Chad, and K. S. Spink. 2005. Osteoporosis knowledge, beliefs,

and practices among adolescent females. J Adolesc Health 36:305–12.Baheiraei, A., J. E. Ritchie, J. A. Eisman, and T. V. Nguyen. 2005. Psychometric

properties of the Persian version of the osteoporosis knowledge and healthbelief questionnaires. Maturitas 50:134–9.

Cadarette, S. M., M. A. Gignac, D. E. Beaton, S. B. Jaglal, and G. A. Hawker.2007. Psychometric properties of the ‘‘Osteoporosis and You’’ questionnaire:Osteoporosis knowledge deficits among older community-dwelling women.Osteoporos Int 18:981–9.

Chan, M. F., W. S. Kwong, Y. L. Zang, and P. Y. Wan. 2007. Evaluation of anosteoporosis prevention education programme for young adults. J Adv Nurs

57:270–85.Drozdzowska, B., W. Pluskiewicz, and M. Skiba. 2004. Knowledge about osteoporo-

sis in a cohort of Polish females: The influence of age, level of education andpersonal experiences. Osteoporos Int 15:645–8.

Fayers, P. M., and D. Machin. 2007. Quality of Life: Assessment, Analysis and In-

terpretation of Patient Reported Outcomes, 2nd ed. Chichester: John Wiley &Sons.

Floyd, J. F., and F. K. Widaman. 1995. Factor analysis in the development andrefinement of clinical assessment instruments. Psychol Assessment 7:286–99.

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3

Development of a Shorter Version of the OKAT 31

Gaines, J. M., and K. A. Marx. 2011. Older men’s knowledge about osteoporosis andeducational interventions to increase osteoporosis knowledge in older men: Asystematic review. Maturitas 68:5–12.

Gammage, K. L., C. Francoer, D. E. Mack, and P. Klentrou. 2009. Osteoporosishealth beliefs and knowledge in college students: The role of dietary restraint.Eat Behav 10:65–7.

Hu, L., and P. M. Bentler. 1998. Fit indices in covariance structure modeling: Sen-sitivity to underparameterized model misspecification. Psychological Methods

3:424–53.Kasper, M. J., M. G. Peterson, J. P. Allegrante, T. D. Galsworthy, and B. Gutin.

1994. Knowledge, beliefs and behaviors among college concerning preventionof osteoporosis. Arch Fam Med 3:696–702.

Lai, P. S., S. S. Chua, S. P. Chan, and W. Y. Low. 2008. The validity and reliabil-ity of the Malaysian osteoporosis knowledge tool in postmenopausal women.Maturitas 60:122–30.

Pande, K. C., D. Takats, J. A. Kanis, V. Edwards, P. Slade, and E. V. McCloskey.2000. Development of a questionnaire (OPQ) to assess a patient’s knowledgeabout osteoporosis. Maturitas 37:75–81.

Sedlak, C. A., M. O. Doheny, and S. L. Jones. 2000. Osteoporosis education programs:Changing knowledge and behaviors. Public Health Nurs 17:398–402.

Stevanovic, D. 2009. Serbian KINDL questionnaire for quality of life assessments inhealthy children and adolescents: Reproducibility and construct validity. Health

Qual Life Outcomes 7:79.Stevanovic, D., D. K. Tepavcevic, B. Jocic-Jakubi, M. Jovanovic, T. Pekmezovic,

A. Lakic, et al. 2009. Health-related quality of life measure for children withepilepsy (CHEQOL-25): Preliminary data for the Serbian version. Epilepsy Behav

16:599–602.Werner, P. 2005. Knowledge about osteoporosis: Assessment, correlates and out-

comes. Osteoporos Int 16:115–27.Wild, D., A. Grove, M. Martin, S. Eremenco, S. McElroy, A. Verjee–Lorenz, et al. 2005.

Principles of good practice for the translation and cultural adaptation processfor patient-reported outcomes (PRO) measures: Report of the ISPOR task forcefor translation and cultural adaptation. Value Health 2:94–104.

Winzenberg, T. M., B. Oldenburg, S. Frendin, and G. Jones. 2003. The design of avalid and reliable questionnaire to measure osteoporosis knowledge in women:The osteoporosis knowledge assessment tool (OKAT). BMC Musculoskelet Dis-

ord 4:17.Zhang, R. F., and M. Chandran. 2011. Knowledge of osteoporosis and its related risk

factors among nursing professionals. Singapore Med J 52:158–62.

Dow

nloa

ded

by [

Sain

t Jos

eph'

s U

nive

rsity

] at

17:

11 2

6 A

ugus

t 201

3