Development and first validation of the shared decision-making questionnaire (SDM-Q)

Abstract

Objective

Due to a lack of valid German instruments measuring shared decision-making (SDM), a theory-driven questionnaire (SDM-Q) to measure the process of SDM was developed and validated.

Methods

As a theoretical basis steps of the SDM process were defined in an expert panel. Item formulation was then conducted according to the Delphi method. For the first validation on a mixed sample Rasch analysis was used to eliminate items not fitting the construct thus receiving a unidimensional scale.

Results

After eliminating 4 items the remaining 11 form a unidimensional scale with an acceptable reliability for person measures (0.77) and very good reliability for item difficulties (0.95). Analysis of subgroups revealed a different use of items in different conditions. Furthermore the scale showed high ceiling effects.

Conclusion

A new theory-driven instrument to measure the process of SDM has been developed and validated by use of a rigorous method revealing first promising results. Yet the ceiling effects require the addition of more discriminating items, and the different use of items in different conditions demands an in depth analysis.

Practice implications

While the concept of SDM is constantly receiving more attention in medical practice, its valid and reliable measurement remains challenging.

Introduction

In recent years shared decision-making (SDM) has been advocated as an appropriate approach for patient physician communication and decision-making [1]. The first definition of this concept can be found in a report on making health care decisions by the President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research. Published in 1982 it describes SDM as a process which is based on mutual respect and partnership [2]. More attention to SDM in the literature was given in the late 1990s when several conceptualisations led to a stronger interest in decision-making between patient and physician. According to Charles et al. SDM implies that at least two individuals are involved in the process of making a treatment decision [3]. Both parties exchange key information about treatment options before they agree on a preferred choice. Another practical contribution of Towle and Godolphin [4] proposed competencies for physicians and patients to engage in SDM. In contrast Elwyn et al. [5] put more emphasis on the physicians’ side highlighting the need for eliciting and responding to patients’ understanding and their preferences. Further specifications of the decision-making process were developed by Giersdorf et al. [6]. In a recent systematic review on conceptual definitions of SDM Makoul and Clayman [7] stated that there is no common definition so far and proposed a model that consists of the essential practical elements of SDM integrating concepts from the existing literature. Since the various conceptual contributions in the 1990s a lot of research has been done on SDM and its influence on outcomes ranging from patient satisfaction, adherence, and decisional conflict to health status [7], [8].

The German Ministry of Health funded a research consortium on SDM from 2001 until 2005 with 10 projects researching the implementation of SDM in different scenarios such as breast cancer, depression or hypertension. The overall aim of this research consortium was the investigation of intervention effects resulting from specific training programs for physicians and the use of patient information materials and decision aids. Since these intervention strategies need adequate measures to evaluate their effectiveness, members of all 10 projects joined a comprehensive methods team on methodological issues in SDM to create a basis for common evaluation strategies. For the purpose of fully capturing process and effects of SDM both the views of physicians and patients need robust measures to enable comparisons and conclusions [9]. Valid instrument development should ideally include clear, theory-driven definition of the construct, item selection embedded in qualitative inquiry and rigorous development according to psychometric principles [9]. Yet the construct of SDM has been stated as very complex with some aspects being easier to measure than others which can render instrument development somewhat challenging [9].

With regard to available measures Elwyn et al. [9] conducted a systematic review on observational instruments and stated that none existed up to now to measure the extent of SDM. Consequently they developed the OPTION scale to assess the degree of SDM that is practiced by physicians in consultations [10], [11].

As for patient-report questionnaires the German Consortium's methods team generated a review following search strategies via “BIOSIS + Medline + PreMedline + LWW (Lippincott Williams and Wilkins) (1966–2002)” and “Psyndex plus Lit & AV (1977/2002)” [12]. The resulting instruments of this search measure different aspects of SDM such as patients’ preferences for information and participation [13], decisional conflict [14], doctor facilitation of participation and patients’ information seeking behavior [15] as well as risk communication and confidence in decision-making [16], and satisfaction with decision-making [17].

The German research consortium decided to address several relevant constructs within SDM: patients’ preferences for information and participation as well as process and outcome of medical decision-making. The autonomy preference index (API) [13], the control preference scale [18] the patients’ perceived involvement in care scale (PICS) [15] and the combined outcome measure for risk communication and treatment decision making (COMRADE) [16] were therefore chosen to address these.

Apart from the PICS for which the German version had already been validated [19], all instruments had to be translated into German and examined for validity taking differences in language use, cultures and health care systems into account. The instruments were used in comprehensive samples within the consortium ranging from N = 598 for the COMRADE to N = 1154 for the API, with conditions including hypertension, depression, multiple sclerosis and others. Given that one of the main variables studied in the research consortium was the extent of shared decision-making, the results from using PICS and COMRADE in the consortium were of particular interest. Preliminary data with PICS appeared to replicate the original factor structure in its German validation [19] whereas for COMRADE the two factors were highly correlated and item responses showed high ceiling effects [20]. This did not correspond with the original UK validation of COMRADE, probably due to translation, culture and health care system differences. Since the PICS had not been specifically developed to measure the process of SDM and the German version of the COMRADE did not show good validation data, consequently a new theory based instrument, the shared decision-making questionnaire (SDM-Q) was developed in the methods team. This paper aims to describe the development process, the first use of the SDM-Q as well as psychometric analyses for its first validation.