Development and preliminarily validation of the Complementary Medicine Disclosure Index

Development and preliminarily validation of the Complementary Medicine Disclosure Index

Journal Pre-proof Development and preliminarily validation of the Complementary Medicine Disclosure Index Erica McIntyre, Hope Foley, Helene Diezel, J...

2MB Sizes 2 Downloads 29 Views

Journal Pre-proof Development and preliminarily validation of the Complementary Medicine Disclosure Index Erica McIntyre, Hope Foley, Helene Diezel, Joanna Harnett, Jon Adams, David Sibritt, Amie Steel

PII:

S0738-3991(20)30007-0

DOI:

https://doi.org/10.1016/j.pec.2020.01.008

Reference:

PEC 6496

To appear in:

Patient Education and Counseling

Received Date:

2 August 2019

Revised Date:

13 January 2020

Accepted Date:

16 January 2020

Please cite this article as: McIntyre E, Foley H, Diezel H, Harnett J, Adams J, Sibritt D, Steel A, Development and preliminarily validation of the Complementary Medicine Disclosure Index, Patient Education and Counseling (2020), doi: https://doi.org/10.1016/j.pec.2020.01.008

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. 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. © 2019 Published by Elsevier.

Development and preliminarily validation of the Complementary Medicine Disclosure Index Erica McIntyrea * [email protected] Hope Foleya,b [email protected] Helene Diezelb [email protected] Joanna Harnetta,c [email protected] Jon Adamsa [email protected]

ro of

David Sibritta [email protected] Amie Steela [email protected]

-p

Affiliations

Australian Research Centre in Complementary and Integrative Medicine, Faculty of Health, University of Technology Sydney, Ultimo, NSW

b

Endeavour College of Natural Health, Fortitude Valley, Brisbane, QLD

c

The University of Sydney, School of Pharmacy, Faculty of Medicine and Health, Sydney, Australia

lP

re

a

Highlights

ur

na

*Corresponding author: Faculty of Health, University of Technology Sydney Phone: +61 (02) 9514 XXX PO Box 123 Broadway NSW 2007 Australia

Developed a tool measuring reasons for complementary medicine disclosure behaviour



The Complementary Medicine Disclosure Index had acceptable construct validity



Index consists of two second-order measurement models: non-disclosure and disclosure



Concerns about the medical encounter the most important reason for non-disclosure

Jo



1

Abstract Objective. Non-disclosure of complementary medicine (CM) use to doctors is associated with health risks. No standardised, validated instrument exists to measure reasons for CM use disclosure behaviour to doctors in clinical research or practice. This study aims to develop and validate an index that measures the relative importance of reasons for CM disclosure and non-disclosure. Methods. Using data from the Complementary Alternative Medicine Use Health Literacy Disclosure Study (N=2019), we developed a CM Disclosure Index (CMDI) using a formative measurement approach. The adequacy of the measurement models was assessed by conducting variance-based structural equation modelling using partial least squares to analyse multicollinearity, significance and

ro of

relevance of the formative indicators to their relative primary constructs.

Results. The CMDI consists of two second-order measurement models, each with three sub-domains, and demonstrated acceptable construct validity indicating the index is a useful measure to identify

-p

the reasons for CM use disclosure behaviour.

Conclusion. The CMDI provides a preliminary tool to measure the relative importance of the reasons

re

for CM use disclosure and non-disclosure to doctors.

Practice Implications. Understanding patients’ reasons for disclosure and non-disclosure can assist in

lP

developing targeted interventions to both patients and practitioners to facilitate effective patientpractitioner communication and improve patient safety.

1. Introduction

na

Keywords: Complementary Medicine; Disclosure; Communication Barriers; Decision Making; Instrument Development

ur

The use of complementary medicine (CM) is prevalent throughout the world [1]. CM by definition (i.e. complementary) is generally not utilised within conventional medicine [2] and the use of these

Jo

treatments has the potential for direct and indirect risks [3]. For example, concomitant use of herbal medicines with pharmaceutical drugs carries the risk of herb-drug interactions if not appropriately managed [4]. The appropriate management of such risk requires identification of concomitant use and relies on communication between the patient and health care provider, particularly in primary care such as general practice, which often acts as a gatekeeper service to further health care access. Despite the importance of managing concomitant use for patient safety and optimal clinical outcomes, many patients neglect to disclose CM use to their primary care provider. The most recent international

2

review [5] identified substantial rates of CM non-disclosure across different health care settings with the rate of non-disclosure ranging between one in every four patients to four out of every five patients seeking medical care; non-disclosure rates across the 14 included studies was identified as around 70%. While the contemporary literature houses methodological and focus variation (including differences in surveyed populations, and heterogeneity of both the measures used and the definitions of CM) [5], the evidence to date does reveal non-disclosure of CM use as a significant challenge to providing safe, effective patient care. Meanwhile, other literature has begun to examine the reasons why patients disclose or do not disclose their CM use to medical providers. Much of this work , predominantly using qualitative methods,

ro of

suggests the reasons why patients fail to disclose CM use relates to: their expectations of the clinical encounter (e.g., fear a negative response from providers regarding CM use); perceptions of providers’ CM knowledge (e.g., belief that medical providers lack knowledge of CM), and experiences of patientprovider communication (e.g., the provider failed to enquire about CM use) [6]. Conversely, while reasons for disclosure have rarely been examined, when reported they are reflective of the factors

-p

associated with non-disclosure; for example, patient expectations regarding a supportive response from providers [7], practitioners’ ability to provide relevant knowledge and advice [8], or patient

re

experience of providers enquiring about CM use in the consultation [7, 8].

Unfortunately, all this literature (focusing on the factors that facilitate or discourage patient

lP

disclosure/non-disclosure behaviours) provides only preliminary exploration of these topics. More robust research is needed using standardised, validated instruments. Currently, there is there is no such instrument for measuring the reasons for patient CM disclosure behaviour (i.e., disclosure or

na

non-disclosure) to their doctor. A validated instrument would provide consistent identification of the relative importance of patient reasons for disclosure and non-disclosure. Employing such an

ur

instrument will improve our understanding of the reasons for these behaviours, and generate findings that can better inform and improve clinical communication around concomitant use of CM and conventional medicine that will assist to mitigate potential associated risks. In response to this

Jo

identified gap, this paper aims to develop and validate a parsimonious multidimensional index for measuring the reasons for patient CM disclosure behaviour via empirical fieldwork conducted as part of a national Australian study of complementary and conventional medicine use, health literacy and medicine disclosure.

3

2. Methods 2.1 Study design and participants An online cross-sectional survey was conducted with a sample of Australian adults (18 years and over) as part of the Complementary Alternative Medicine Use Health Literacy Disclosure (CAMUHLD) Study [9]. A total of 2,025 people completed the survey, who were a representative sample of the Australian general population with respect to age, gender and state/territory of residence in comparison to national statistics [10, 11]. This study reports on a sub-sample (n = 726) of this dataset who reported using CM and consulting with their doctor in the previous 12-months. The Human Research Ethics Committee at Endeavour College of Natural Health provided ethics approval (20170242). Reciprocal

ro of

approval was also granted by the Human Research Ethics Committees at Charles Sturt University (H17048), The University of Sydney (2017/140) and The University of Technology Sydney (ETH171564).

-p

2.2 Recruitment

Participants were recruited by a marketing research company (Qualtrics) using purposive convenience

re

sampling. Email invitations were sent to members of a database who had agreed to participate in research. The email contained a link to the questionnaire, which was hosted on the online Qualtrics platform. Informed consent was implied when respondents clicked on a button to continue the survey

lP

after reading the information page. A small financial reimbursement was received by participants to reflect the time taken to complete the survey (approximately 15 minutes).

na

2.3 Survey instrument

The survey included 50 items that measured five topic areas: demographics, health status, health

ur

services use, complementary medicine literacy, and medicine disclosure and communication. The measures relevant to this study are described in the following sections.

Jo

2.3.1 Demographics

Categorical variables were used to collect demographic information including gender, and age. 2.3.2 Doctor communication and CM disclosure and non-disclosure Participants were asked about communication with their family doctor in the previous 12 months, with four response options: 1) I told them about ALL complementary and alternative medicines I was using, 2) I only told them about SOME of my complementary and alternative medicine use, 3) I DID NOT tell them about my complementary and alternative medicine use, and 4) I did not visit this type 4

of health professional. Those who responded either 1 or 2 were presented with 16 items that enquired about their reasons for CM disclosure. If they responded 2 or 3 they were presented with 17 items that measured their reasons for non-disclosure of CM use to their doctor. A detailed description of these items follows. 2.4 Complementary Medicine Disclosure Index (CMDI): Theoretical approach The purpose of the CMDI is to identify the reasons for CM use disclosure behaviour to doctors, and reflects two distinct behaviours “CM disclosure”, and “CM non-disclosure”. The constructs and their related domains were defined following a systematic review of the literature [5] and discussion with content experts (i.e., academics and health practitioners). These definitions are supported by the

ro of

index construction and validation process that follows. The structure of the CMDI is conceptualised as two distinct formative measurement models, as the composite indicator variables (items) define (cause) the characteristics of two latent constructs: “reasons for CM disclosure” (Domain 1) and “reasons for CM non-disclosure” (Domain 2) [12-15]. Each of these latent constructs are

-p

conceptualised as hierarchical second-order measurement models, as they each have three antecedent theoretical domains [14]. Domain 1, reasons for CM disclosure, is formed by three sub-

re

domains: facilitators of CM disclosure, health needs, and expectations of the medical encounter. Meanwhile, Domain 2, reasons for CM non-disclosure, is formed by three sub-domains: barriers to disclosure, beliefs about the need to disclose, and concerns about the medical encounter. Table 1 and

2.5

lP

2 presents the operationalisation of the CMDI domains and their relevant items (see section 3.3). CMDI: Item development (indicator specification)

na

The index items were initially developed following an extensive literature review [5]. Content validity of the items was qualitatively assessed by the research team, who are content experts, with

ur

consideration of the theoretical, practical, and cultural context of the construct being measured. Items were revised until consensus was reached that the items accurately formed the construct. A convenience sample of seven Australian adults assessed the items for comprehension, complexity,

Jo

and face validity, and the items were further refined based on their qualitative feedback. The items were rated on a 5-point Likert scale from strongly agree (1) to strongly disagree (5). A 5-point response option has used as it has demonstrated good statistical reliability, provides a neutral option with equal psychological distance between the two response categories (strongly agree and strongly disagree), and is considered brief enough to answer quickly while allowing for an adequate number of options for respondents to express their level of agreement [16]. Domain 1, reasons for CM non-disclosure, was measured by 17 items, with 16 items measuring Domain 2, reasons for CM disclosure (see Table 1). 5

2.6 Data cleaning There were 2,025 survey respondents. Of these, six cases were removed from the dataset as they were unreliable (no variance in responses, or repeat patterns in the data). The final data set contained 2,019 valid cases. For the current sub-study, 726 cases were eligible for analysis (had used CM, consulted with a doctor in the previous 12 months, and completed either the non-disclosure or disclosure items). Two hundred and twenty five participants completed the items measuring reasons for CM non-disclosure; four of these were excluded from analysis as there was no variance in their responses to these items. There were 630 valid responses to the reasons for CM disclosure items. See Figure 1 for a flow diagram.

ro of

Insert Figure 1 2.7 Data analysis

IBM SPSS Statistics Premium Edition Version 24 was used to analyse descriptive statistics related to

-p

the sample characteristics and check relevant statistical assumptions.

The CMDI was conceptualised as having two distinct second-order constructs: 1) reasons for CM non-

re

disclosure (Domain 1), and 2) reasons for CM disclosure (Domain 2). These constructs were evaluated as two separate formative-formative measurement models as they meet the criteria outlined by Hair

lP

[14]. As the CMDI was conceptualised as a formative measure it is not appropriate to apply statistical procedures relevant to classical test theory, such as internal reliability and exploratory factor analysis [12, 13, 17]. The software SmartPLS was used to conduct variance-based structural equation modelling

na

(SEM) using partial least square (PLS) in order to analyse collinearity, relevance and statistical significance of the formative indicators of the measurement models for Domain 1 and Domain 2 [14,

ur

15].

Collinearity occurs when there is a high correlation between two indicators [14]. In PLS-SEM variance inflation factor (VIF) values for all indicator items are used to assess collinearity and construct validity

Jo

[14]. VIF values less than 3.3 indicate acceptable collinearity [14] and sufficient construct validity [14]. Formative indicators are considered to be relevant if they contribute to the construct in a meaningful way [14]. The relative importance of an indicator in forming the construct is determined by the value of its outer weight, and its outer loading indicates the absolute importance [14]. The two second-order latent variables were created using a repeated indicator approach (Mode B), as the items predicting the first-order constructs also perfectly predict their relative second-order constructs [15, 18]. First, the indicator items were used to create the measurement model and obtain

6

the second-order latent variable score. Second, a new model was created using the latent variable score as the indicator of the second-order construct. Bootstrap resampling using 5,000 samples was used to determine the contribution of each indicator (item) to the latent construct. A staged decision-making process was used as recommended by Hair et al. (2017). First, indicator outer weights were assessed for their relative contribution to the firstorder constructs, and retained in the measurement model if they were significant (p < 0.05), or if the outer loading of the indicator was ≥ 0.50 [14]. Second, each indicator that did not fit this criteria was evaluated for the significance of their outer loading, and removed from the measurement model if they did not demonstrate absolute importance to the construct (i.e., outer loading was < 0.50 and

ro of

non-significant) and there was theoretical justification to do so [14]. 3. Results 3.1 Participant characteristics

-p

The majority of the 726 participants who discussed CM use with their doctor were female (57.2%, n = 415); one participant identified as “other” gender. Most of these participants (n=630) disclosed either

re

some or all of their CM use, which compared to 221 who either did not disclose their CM use or only

3.2 Construct validity

lP

disclosed some use.

The VIF values for all items of the CMDI Non-disclosure domain ranged between 1.04 and 2.37, and

na

between 1.21 and 2.47 for the CMDI Disclosure domain. These results suggest that each item uniquely contributes to their relative primary latent construct [14, 15].

ur

3.3 Assessment of relevance of items to first-order constructs Three items (2, 11 and 16) were removed from the CMDI Non-disclosure model as they did not meet the statistical criteria described in section 2.7. Two items (1 and 5) had non-significant outer weights

Jo

(p > 0.05), and outer loadings < 0.50. As deleting formative indicators has a large consequence on the nature of the construct [14], the theoretical importance of these indicators was evaluated; consequently, items 1 and 5 were retained in the final CMDI Non-disclosure model as the research team agreed on their theoretical importance to the construct (see Table 1 for indicator weights, and significance values). For the CMDI Disclosure model all items met the statistical criteria and were retained (see Table 2 for summary statistics).

7

of

Table 1. Item and domain operationalisation, indicator weights, and significance values for the initial and final CMDI Non-disclosure measurement model

ro

(Domain 1). Reasons for CM non-disclosure (second-order construct)

behavioural

Patient’s perceived practical and behavioural barriers to disclosing CM use to doctors Beliefs about need to disclose

about

OW

p value (OW)

OL

OW

p value (OW)

0.142

0.042

0.85

0.152

0.040

0.84

0.998

1.013

0.00*

0.999

0.995

0.00*

0.274

-0.061

0.77

-

-

-

0.136

0.007

0.96

-

-

-

They did not need to know (6)

0.412

0.033

0.76

0.406

0.018

0.86

It is none of their business (12)

0.726

0.530

0.00*

0.702

0.491

0.00*

Complementary and alternative medicines are safe (5)

0.170

-0.044

0.65

0.181

0.82

I did not think they would know anything about complementary and alternative medicine (10) I do not use complementary and alternative medicine regularly enough (16) I did not think they would understand my choice (3)

0.858

0.706

0.00*

0.881

0.022 0.740

-0.028

-0.096

0.24

-

-

-

0.673

0.134

0.30

0.672

0.135

0.25

I was worried they would judge me (4)

0.600

-0.031

0.78

0.599

0.78

I felt uncomfortable discussing it with them (8)

0.784

0.399

0.00*

0.787

0.031 0.401

I was worried they wouldn’t support my treatment decisions (9)

0.716

0.068

0.63

0.724

0.086

0.53

I was worried they would discourage my use of complementary and alternative medicine (13) They do not approve of my use of complementary and alternative medicine (14) I was worried they would respond negatively (15)

0.723

0.124

0.40

0.715

0.105

0.42

0.588

0.045

0.68

0.586

0.043

0.67

0.701

0.076

0.54

0.701

0.077

0.51

They did not ask me about my complementary and alternative medicine use (1) There was not enough time in the consultation (7) I forgot to mention it (11)

I did not think it was important (2)

medical

Jo u

Concerns encounter

Beliefs reflecting patients concerns about the medical encounter with their doctor related to their disclosure of CM use. Includes perceptions about doctor’s beliefs that reflect the quality of therapeutic alliance.

Final model

OL

rn a

Patient’s beliefs reflecting the perceived need to disclose CM use to doctors.

Initial model

-p

and

Indicator (item number)

re

Practical barriers

and

lP

First-order construct conceptual definition

0.00*

0.00*

0.807

0.492

of

I previously had a negative experience when I disclosed using complementary and alternative medicine (17)

0.805

0.494

0.00*

Jo u

rn a

lP

re

-p

ro

Note. OL = outer loading, OW = outer weight, *p < .05.

0.00*

9

2).

ro

Reasons for CM disclosure (second-order construct)

of

Table 2. Item and domain operationalisation, indicator weights, and significance values for the initial and final CMDI Disclosure measurement model (Domain

Indicator (item number)

Facilitators

They asked me about my use of complementary and alternative medicine (21)

0.546

0.217

0.00*

I have a good relationship with them (22)

0.706

0.035

0.46

I felt comfortable discussing complementary and alternative medicine with them (23)

0.759

0.233

0.00*

They have a good attitude towards complementary and alternative medicine (27)

0.702

0.169

0.00*

They are open-minded (28)

0.781

0.194

0.00*

They support my use of complementary and alternative medicines (30)

0.657

0.086

0.06

They understand my treatment goals (31)

0.787

0.272

0.00*

They have my best interests at heart (32)

0.725

0.191

0.00*

I wanted them to fully understand my health status (18)

0.587

0.297

0.00*

I was concerned about drug interactions with the complementary and alternative medicine I was using (19) I wanted their approval of my complementary and alternative medicine use (25)

0.505

0.161

0.00*

0.788

0.424

0.00*

I wanted their advice about complementary and alternative medicines (33)

0.858

0.478

0.00*

I thought they might know something about complementary and alternative medicine (20) I knew they would be willing to discuss my alternative medicine use (24)

0.674

0.262

0.00*

0.812

0.380

0.00*

I knew they would understand about my complementary and alternative medicine use (26) I thought they could help with my treatment decisions (29)

0.818

0.301

0.00*

0.792

0.339

0.00*

re

lP

rn a

Aspects of the medical encounter that facilitate the disclosure of CM use to doctors. Includes perceptions about doctor’s beliefs and attitudes that reflect the quality of therapeutic alliance. Also includes beliefs about doctor’s attitudes towards CM and behaviours that facilitate disclosure. Does not include expectations of the medical encounter. Health needs

-p

First-order construct and conceptual definition

Jo u

Patients perceived health needs that facilitate disclosure of CM use to doctors.

Expectations of the medical encounter

Beliefs about what the patient will experience in the medical encounter with their doctor. Excluding beliefs about health needs.

Final model OL

OW

p value (OW)

Note. OL = outer loading, OW = outer weight, *p < .05.

10

3.4 Assessment of second-order constructs Practical and behavioural barriers, beliefs about need to disclose, and concerns about medical encounter each have significant (<.001) coefficients supporting the formative construct reasons for CM non-disclosure. Similarly, facilitators, health needs and expectations of the medical encounter had significant (<.001) coefficients, confirming the formative construct reasons for CM disclosure. Table 3 presents the path coefficients between the first and second-order constructs in each CMDI measurement model. Figure 2 and 3 show the final CMDI Non-disclosure and Disclosure measurement models respectively.

R2

Constructs

ro of

Table 3. Path coefficients between first and second-order constructs. Path

P value

coefficient

Reasons for CM non-disclosure

1.00

Beliefs about need to disclose

re

Concerns about medical encounter Reasons for CM disclosure

< .001

.31

< .001

.70

< .001

.37

< .001

.37

< .001

.35

< .001

1.00

lP

Facilitators Health needs

.13

-p

Practical and behavioural barriers

Expectations of the medical encounter

na

Note. R2 = 1 in formative second-order constructs as it is fully explained by its indicators. 4. Discussion and Conclusion

ur

4.1 Discussion

This study is the first to develop and evaluate measurement instruments for the reasons for patient

Jo

CM disclosure and non-disclosure to doctors. The final version of the CMDI demonstrated acceptable construct validity indicating the index is a useful measure to identify the reasons for CM use disclosure behaviour. Our study found that the three sub-domains (facilitators, health needs, and expectations of the medical encounter) had an equally important influence on the reasons for CM disclosure. Overall, the items forming the disclosure domain were statistically much stronger than those forming the non-disclosure domain. This may partially be explained by disclosure being an active behaviour that is easier to measure than non-disclosure, which is a passive behaviour [19]. However, as the volume of research informing the design of the instrument was far greater for the items in the non-

disclosure model compared to the disclosure model [5], the theoretical basis for the non-disclosure domain is strong. Similarly, the decision to retain two items despite their non-significant outer weights and loadings was theoretically based. The first of these items measured a belief about the safety of CM, which is one of the primary concerns underpinning the need for improved disclosure of CM use to physicians [3, 4]. The second item measured whether or not their doctor asked about their CM use, which has been found to be an important influencer of CM use disclosure [5, 20]. Our study results suggest that concerns about the medical encounter have the most important influence on the reasons for non-disclosure of CM use, followed by beliefs about the reason to disclose, and practical and behavioural barriers being the least important. This is consistent with the

ro of

most recent literature review that identified a fear of the health practitioner’s disapproval as the most common reason for patients not disclosing CM use [5]. This finding is concerning given that a strong therapeutic relationship is deemed essential for shared-decision making and achieving safe, effective patient care [21]. CM users have been found to place a high value on their relationship with their clinicians; which is likely to be reflected in our study’s population sample [22]. The reality of perceived

-p

physician disapproval has been questioned by observational research reporting negative or discouraging provider responses to CM disclosure amongst a minority of patients (representing less

re

than 20% of disclosers) [23-27], or were not reported at all [28]. In some studies, a substantial proportion of patients report positive (32-91%) or, less frequently, neutral responses (8-32%) from

lP

their physician with regards to disclosing CM use [24-30]. With this in mind, researchers should ensure that the data collected using our instrument is interpreted only as the perceptions and subjective

their patients.

na

experiences of CM users rather than objective measures of clinical interactions between doctors and

The results of our analysis also suggest that the practical and behavioural barriers domain of the

ur

reasons for CM non-disclosure construct may more accurately reflect extrinsic barriers outside the patients control. This is consistent with non-disclosure being a non-expressed, passive action that may not always be an active patient choice [31, 32], but rather an outcome of how the health practitioner

Jo

directs commutation during a consultation [7, 8]. This may partly explain why the item “I forgot to mention it” did not fit within this sub-domain, as it measures an intrinsic patient behaviour. In addition, the item “They did not ask me about my complementary and alternative medicine use” was not a strongly performing indicator of practical and behavioural barriers. Theoretical support for keeping this item in the CMDI is strong given that a lack of enquiry by a clinician has frequently been shown to influence CM disclosure [5, 20]. The construct practical and behavioural barriers may be too

12

ambiguous, which may also help explain why this sub-domain contributed the least amount of variance in non-disclosure behaviour. Future research should consider revaluating this construct. The primary strength of this study was the use of a thorough theoretical approach based on previous literature that directly informed CMDI development. Each domain within the final models uniquely contributed to their relative primary latent construct of CM disclosure and non-disclosure respectively, suggesting the CMDI is a comprehensive measure of each construct [14, 15]. The utility of the CMDI for prospective assessments of the reasons for CM disclosure and non-disclosure should be considered in future research. For example, it would be beneficial to determine the degree to which patient’s reasons for disclosure behaviour change over time in response to specific interventions; such

ro of

as patient education about the importance of disclosure, or practitioner education about the importance of establishing a strong therapeutic relationship and asking about CM use. In addition, future research could consider the development of alternative reflective measures to advance our understanding of CM disclosure behaviours in clinical practice.

-p

This study provides a robust analysis of theory-based causal models of reasons for CM disclosure and non-disclosure behaviour. However, there are limitations to our study that need consideration in

re

future research. First, all item responses were presented in the same direction, resulting in possible response bias; future research should consider presenting some items in a reverse direction to minimise this effect. Second, the reported results are only applicable to this sample population and

lP

the model needs to be tested in a range of patient populations and cultural contexts. The indexes were developed to practically predict each construct; consequently, the indicator weights forming each dimension will vary across sample populations [33]. Third, although the items were developed

na

through rigorous methods it is possible that they may not have adequately captured the complexity of the disclosure and non-disclosure constructs. Finally, we were unable to test the external validity

ur

of the index, which should be a priority for future research on this topic. 4.2 Conclusion

Jo

This study is the first to develop and test a theoretically based index that seeks to determine the relative importance of reasons for CM disclosure and non-disclosure to doctors. The two formative measurement models that comprise the CMDI are likely to provide reasonable representation of the reasons for which individuals do or do not disclose CM use in different contexts. However, the index needs to be validated in other sample populations. The CMDI provides a preliminary tool to consistently measure the relative importance of the reasons for CM use disclosure and non-disclosure to doctors. Further developing and adopting the CMDI in future work on this critical topic would help provide a consistency and rigour necessary to inform practice-related improvements. Understanding

13

patients’ reasons for disclosure and non-disclosure can assist to develop targeted interventions to both patients and practitioners in order to help facilitate effective patient-practitioner communication and improve patient safety. 4.3 Practice implications Open communication between patients and practitioners about CM is a critical part of advancing the appropriate and safe use of these treatments. The CMDI is an instrument that can be used across several contexts to assist in understanding the factors that facilitate and inhibit such communication in clinical practice. For example, the CMDI can be used to understand patients’ reasons for disclosure and non-disclosure of CM use within medical education clinical placement programs, practice-based

ro of

research projects that aim to improve gaps in patient practitioner relations, and the development of public health initiatives that target both patients and practitioners. The overall practice implication for use of the CMDI is to facilitate effective patient-practitioner communication and improve patient safety.

-p

Funding sources

This work was supported by funding from Endeavour College of Natural Health and the Australian

re

Research Centre in Complementary and Integrative Medicine (UTS:ARCCIM) at the Faculty of Health,

lP

University of Technology Sydney. Conflict of interest

ur

Author statement

na

None.

Erica McIntyre, Amie Steel and Joanna Harnett were responsible for conceptualizing the study,

Jo

design of methodology, investigation and project administration. Amie Steel and Erica McIntyre were responsible for funding acquisition. Erica McIntyre and Hope Foley were responsible for data curation. Erica McIntyre was responsible for formal analysis. Erica McIntyre, Hope Foley, Helene Diezel, Amie Steel and Joanna Harnett wrote the initial draft of the manuscript. Jon Adams and David Sibbritt critically reviewed and edited the manuscript in addition to providing mentorship to the core research team throughout the project.

14

Acknowledgements HF was supported by an Australian Government Research Training Program Scholarship while working on this manuscript. JA was supported by an Australian Research Council Professorial Future Fellowship while working on this manuscript (Grant FT140100195). AS’s academic position is supported by funding from the Blackmores Foundation, Bioceuticals and the Jacka Foundation of Natural Therapies. JH’s academic position was supported by a philanthropic donation from Blackmores Pty Ltd during the course of this study. The authors were solely responsible for conceptualising the study and the work

ro of

presented in this manuscript.

References

[1] P. Harris, K. Cooper, C. Relton, K. Thomas, Prevalence of complementary and alternative medicine (CAM) use by the general population: a systematic review and update, Int J Clin Pract 66 (2012) 924-

-p

939.

[2] World Health Organization, WHO Traditional Medicine: Definitions, WHO (2015).

re

[3] J.L. Wardle, J. Adams, Indirect and non-health risks associated with complementary and alternative

lP

medicine use: An integrative review, Eur J Integr Med 6(4) (2014) 409-422. [4] S. Mamindla, P. K.V.S.R.G, B. Koganti, Herb-drug interactions: An overview of mechanisms and

na

clinical aspects, Int J Pharm Sci Res 7 (2016) 3576-3586. [5] H. Foley, A. Steel, H. Cramer, J. Wardle, J. Adams, Disclosure of complementary medicine use to medical providers: a systematic review and meta-analysis, Sci Rep 9 (2019) 1573.

ur

[6] M.T. Chao, M.A. Handley, J. Quan, U. Sarkar, N. Ratanawongsa, D. Schillinger, Disclosure of

Jo

complementary health approaches among low income and racially diverse safety net patients with diabetes, Patient Educ Couns 98 (2015) 1360-1366. [7] B. Shelley, A. Sussman, R. Williams, A. Segal, B. Crabtree, ’They don’t ask me so I don’t tell them’: Patient-clinician communication about traditional, complementary, and alternative medicine, Ann Fam Med 7 (2009) 139-147.

15

[8] B. Shah, B. Lively, M. Holiday-Goodman, D. White, Reasons why herbal users do or do not tell their physicians about their use: A survey of adult ohio residents, J Pharm Tech 22 (2006) 148-154. [9] J. Harnett, E. McIntyre, A. Steel, H. Foley, J. Adams, D. Sibritt, Use of complementary medicine products: A nationally representative cross sectional survey of 2019 Australian adults, BMJ Open (2019) e024198. [10] Australian Bureau of Statistics, 2016 Census QuickStats, 2017. (Accessed 14 Dec 2017). [11] A. Steel, E. McIntyre, J. Harnett, H. Foley, J. Adams, D. Sibbritt, J. Wardle, J. Frawley,

ro of

Complementary medicine use in the Australian population: Results of a nationally-representative cross-sectional survey, Sci Rep 8 (2018).

[12] A. Diamantopoulos, H.M. Winklhofer, Index construction with formative indicators: An alternative

-p

to scale development, J Mark Res 38 (2001) 269-277.

[13] C.B. Jarvis, S.B. MacKenzie, P.M. Podsakoff, A critical review of construct indicators and

re

measurement model misspecification in marketing and consumer research, J Cons Res 30 (2003) 199218.

lP

[14] J.F. Hair, G.T.M. Hult, C.M. Ringle, M. Sarstedt, A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), 2nd ed., Sage, Los Angeles, USA, 2017.

na

[15] P.B. Lowry, J. Gaskin, Partial least squares (PLS) structural equation modeling (SEM) for building and testing behavioral causal theory: When to choose it and how to use it, IEEE Trans Prof Commun

ur

57 (2014) 123-146.

[16] C.C. Preston, A.M. Colman, Optimal number of response categories in rating scales: reliability,

Jo

validity, discriminating power, and respondent preferences, Acta Psychologica 104 (2000) 1-15. [17] S.B. MacKenzie, P.M. Podsakoff, N.P. Podsakoff, Construct measurement and validation procedures in MIS and behavioral research: Integrating new and existing techniques, MIS Quart 35 (2011) 293-334. [18] J.-M. Becker, K. Klein, M. Wetzels, Hierarchical latent variable models in PLS-SEM: Guidelines for using reflective-formative type models, Long Range Plann 45 (2012) 359-394.

16

[19] D.E. Montano, D. Kasprzyk, Theory of reasoned action, theory of planned behavior, and the integrated behavioral model, Health behavior: Theory, research and practice (2015) 95-124. [20] E. Ben-Arye, I. Halabi, S. Attias, L. Goldstein, E. Schiff, Asking patients the right questions about herbal and dietary supplements: cross cultural perspectives, Complement Ther Med 22 (2014) 304310. [21] M. Kunneman, V.M. Montori, A. Castaneda‐Guarderas, E.P. Hess, What is shared decision making?(and what it is not), Ac Emergency Med 23 (2016) 1320-1324.

ro of

[22] H. Foley, A. Steel, Patient perceptions of clinical care in complementary medicine: A systematic review of the consultation experience, Patient Educ Couns 100 (2016) 212-23.

[23] N.M. Alaaeddine, S.M. Adib, H.M. Alawieh, S.M. Adibilly, M.M. Khalil, S.E. Assaad, M.C. Khayat,

-p

Use of herbal medications and their perceived effects among adults in the Greater Beirut area, Le Journal Médical Libanais. The Lebanese Medical Journal 60 (2012) 45-50.

re

[24] A. Djuv, O.G. Nilsen, A. Steinsbekk, The co-use of conventional drugs and herbs among patients in Norwegian general practice: a cross-sectional study, BMC Complementary And Alternative Medicine

lP

13 (2013) 295.

[25] M.A.H. Levine, S. Xu, K. Gaebel, N. Brazier, M. Bédard, K. Brazil, L. Lohfeld, S.M. MacLeod, Self-

na

reported use of natural health products: a cross-sectional telephone survey in older Ontarians, The American Journal Of Geriatric Pharmacotherapy 7 (2009) 383-392.

ur

[26] F. Naja, M. Alameddine, L. Itani, H. Shoaib, D. Hariri, S. Talhouk, E. Tiralongo, The use of complementary and alternative medicine among Lebanese adults: results from a national survey,

Jo

Evidence-based complementary and alternative medicine : eCAM 2015 (2015) 682397. [27] W. Najm, S. Reinsch, F. Hoehler, J. Tobis, Use of complementary and alternative medicine among the ethnic elderly, Altern Ther Health Med 9(3) (2003) 50. [28] F.A. Stevenson, N. Britten, C.A. Barry, C.P. Bradley, N. Barber, Self-treatment and its discussion in medical consultations: how is medical pluralism managed in practice?, Social Science & Medicine 57 (2003) 513-527.

17

[29] B. Chin-Lee, W.J. Curry, J. Fetterman, M.A. Graybill, K. Karpa, Patient experience and use of probiotics in community-based health care settings, Patient Preference And Adherence 8 (2014) 15131520. [30] C. Ozturk, G. Karayagiz, Exploration of the use of complementary and alternative medicine among Turkish children, Journal of Clinical Nursing 17 (2008) 2558-2564. [31] J. Jou, P.J. Johnson, Nondisclosure of complementary and alternative medicine use to primary care physicians: findings from the 2012 National Health Interview Survey, JAMA Intern Med 176 (2016)

ro of

545-546. [32] T. Arcury, R. Bell, K. Altizer, J. Grzywacz, J. Sandberg, S. Quandt, Attitudes of older adults regarding disclosure of complementary therapy use to physicians, J Appl Gerontol 32 (2013) 627-645.

-p

[33] A.M. Hardin, J.C.-J. Chang, M.A. Fuller, G. Torkzadeh, Formative measurement and academic

re

research: In search of measurement theory, Educ Psychol Meas 71 (2010) 281-305.

Jo

ur

na

lP

Figure 1. Flow diagram of eligible cases included in analysis.

18

ro of -p re lP na ur Jo Figure 2. Final CMDI Non-disclosure measurement model.

19

ro of -p re lP

Jo

ur

na

Figure 3. Final CMDI Disclosure measurement model.

20

21

ro of

-p

re

lP

na

ur

Jo