Subjects, study design, and setting
This cross-sectional study consists of three parts.
Firstly, a survey was carried out mapping the distribution of typical dentistry decisions at the Department of Prosthetic Dentistry, University Medical Center Hamburg-Eppendorf (UKE) in Hamburg, Germany over a two-week period.
Secondly, the study surveyed patients’ autonomy preferences (PAP) in relation to respective attitudes hold by dentists regarding dental treatment decisions in the Department of Prosthetic Dentistry at the UKE. A consecutive sample of 100 patients and all dentists of the clinic (n=16) were recruited. Both samples are not related to each other since for ensuring maximum anonymity of participating patients names of the corresponding dentists were not recorded.
Third, a cross sectional survey of PAP in dentistry, general practice (GP) and a multiple sclerosis (MS) outpatient unit was assessed and compared by consecutively recruiting 100 patients in the waiting areas in each setting (N=300 in total). The dental patients and the patients with multiple sclerosis were contacted at the UKE. The patients in the primary care setting were recruited in three different GP offices, which were members of a primary medical care research network.
The study protocol (PV3452) was reviewed and approved by the Institutional Review Board of the Medical Association in Hamburg, Germany. Written informed consent was obtained from all participants prior to their enrollment. Participants and researchers received no monetary compensation.
Measures
Medical decisions in dentistry
Decisions were registered by the dentists immediately after each appointment using a documentation sheet. This sheet was developed using a list of medical decisions, generated by a panel of two experienced dentists and an expert of methods, and underwent several evaluations to assure that the decision-categories were both exhaustive and disjunctive. Complete registration of decisions in the unit over a two weeks period was supposed to obtain an estimate of each decision’s frequency of occurrence.
Autonomy preferences
PAP was in this study assessed using two different methods, focusing on medical decision making in general and regarding specific medical problems. This proceeding provided occasion to carve out the extent of comparability of the two approaches. Focusing specific decisions in addition, also allowed for intra-individual comparisons of PAP.
The Dental Decision Questionnaire
Based on the dental decisions identified in the first part of the study, a questionnaire with 15 items was developed to assess PAP. This item set comprised decisions referring to a wide range of dental diagnostics and treatments and covered to the greatest extent the entire spectrum of dental care except orthodontics and oral surgery (Table 1).
The answering categories of the Control Preference Scale (CPS) [21] were used as the response scale for the items of this questionnaire. In the current study, the so called “pick-one” approach was applied providing an ordinal five–point response scale in a single item test evaluating a specific decision. The five steps range from “I want to make the decision alone” (0) to “I want the doctor to make the decision for me” (4).
The resulting Dental Decision Questionnaire (DDQ) was applied in the second part of our study to assess PAP. In addition, all participating dentists were asked to rate each decision on the list with regard to the decision’s sensitivity for patients’ preferences. For this assessment, decisions were provided as hypothetically.
The Medical Decision Questionnaire
Based on several medical problems, ten were selected and corresponding items with response scales according to the CPS format were created. This selection covered three medical domains (4 items for GP, 3 items for dentistry, 3 items for MS) and provided variation between diagnostic vs. treatment, and serious vs. trifling decisions. Again, the willingness to involve themselves into making the medical decision was hypothetically.
This Medical Decision Questionnaire (MDQ) was applied in the third part of the study. With one exception, all patient groups (dental, general medicine, multiple sclerosis) were provided with the same set of items. Items specifically focusing decisions in the field of multiple sclerosis were not given to dental or GP patients.
The Autonomy Preference Inventory
The Autonomy Preference Inventory (API) [22] originally consists of six items prompting the patients to indicate their AP referring to medical decisions in general. The API presents statements indicating more or less autonomous attitudes exemplified by standard situations and provides for each statement a 5-point Likert scale ranging from “I completely agree” (0) to “I totally disagree” (4). However, since polarization of items is varying, items assessed either preference for low or high autonomy, scores of items assessing preference for low autonomy were reversed to be in accordance with DDQ and MDQ response scale. Therefore, higher API scores represent lower patient autonomy. The API also provides a set of items assessing information needs. However, in our study 6 AP items were applied. Regarding Cronbach’s alpha, the API scale turned out to show a much higher internal consistency when item 4 and 6 were excluded (6-item scale: alpha = .59; 4-item scale: alpha = .81). This finding is in line with previously published data of the API [22]. All analyses were, therefore, based on the four-item API. The API was applied only in the third part of the study.
Statistical Analyses
For analysis of PAP as assessed with API, DDQ, and MDQ, measures for central tendency (means) and variability (standard deviation; SD) were calculated for the entire population and subgroups considering the scores as quasi continuously scaled. Our statistical approach involved several steps corresponding to the three parts of the study.
Firstly, distribution of decisions in the studied department was analyzed descriptively by presenting frequencies and percentages.
Secondly, doctors’ and patients’ attitudes on whether and how much decisions should be shared were compared using unpaired t-tests for each of the 15 dental decisions in the DDQ.
Third, variation of PAP within an individual and with respect to setting and type of medical decision was investigated using the scores of API and MDQ. Intra-individual variation of PAP was approached by calculating Pearson correlations between API mean scores and each of the single MDQ items. Moreover, for each patient range between lowest and highest PAP as indicated for seven (dental and GP patients) or ten (MS patients) MDQ items, respectively, was calculated. In addition, intra-individual standard deviations of MDQ items within each patient were calculated as a measure of to which extent specific PAP varied within individual patients. The intraindividual variability in answering the 7 decision item, fitting to all participants was also calculated as a Intraclass correlation coefficient based on an unifactorial ANOVA. This was followed by comparing PAP between patients of three different settings using analysis of variance (ANOVA). For this analysis, means of API and MDQ were used. Finally, the impact of the specific type or character of a decision was studied using a repeated measurement design and the variants of decisions as within group factors in an ANOVA.
Alpha correction for multiple testing was performed using Bonferroni correction. To compensate for violation of sphericity Greenhouse and Geisser corrections were used.
Missing values were estimated using the corresponding individual mean value, if at least 3 of the 7 common items (for API 1 of 4) had been completed.