The current study employed a Delphi procedure to define the content of a minimal dataset for outcome measurement in the field of ABI. The expert panel identified 12 domains of importance and reached consensus on how to measure 11 out of the 12 selected domains.
No consensus was reached on how to settle the overlap between the measurement instruments that were selected for the domains ‘participation’ and ‘quality of life’. Respondents showed a clear pattern of reasoning for their decisions: experts either wanted to limit the length of the MDS-ABI and thus chose the USER-P, or considered it important to include a validated measure of quality of life, leading them to choose the USER-P in combination with the 12-Item Short Form Health Survey (SF-12) [
43]. Since this dichotomy reflects a matter of personal preference, we believe a supplementary Delphi round would have yielded similar results. Initially, we adopted the inclusion of both the USER-P and the SF-12 in the MDS-ABI, considering the fact that this option received the highest preference scores. However, after the first draft version of the MDS-ABI was established, it appeared that the SF-12 - although freely available - cannot be scored without the use of purchased software. Consequently, the answer with the second-highest preference scores, i.e. applying the USER-P for measuring both ‘participation’ and ‘quality of life’, was put into practice in constructing the concept MDS-ABI. As a result, the MDS-ABI does not strictly cover a measure of quality of life.
Features of the draft MDS-ABI
Administration of the clinician-rated elements of the draft MDS-ABI will take approximately 25–40 min, based on the added average administration duration of all the measurement instruments. Completion time for the patient-reported questionnaires and screening questions is approximately 30 min. Because of its wide array of outcome domains and relatively limited administration length, the MDS-ABI could be used for screening purposes in health care to assist anamnesis, diagnosis and determination of appropriate treatment and to serve as an outcome measure in research contexts. The draft MDS-ABI contains existing measures that are available in Dutch as well as other languages; most of these measures are frequently used in international settings. All measurement instruments are of sufficient methodological quality and are accessible without restrictions. The measurement instruments and screening questions are not specific to conditions within the spectrum of ABI. Therefore, the MDS-ABI can facilitate quick, standardised data collection on all ABI-patients.
Since the current MDS-ABI consists of several patient-reported outcome measures, it is intended for use with adult patients who are capable of self-reporting on suitable domains. For ABI patients who are unable to complete these questionnaires, a so-called ‘proxy-module’ for all possible domains will be developed which can be administered by a professional or informal caregiver to retrieve essential information on injury outcomes. Likewise, since the MDS-ABI contains questionnaires that are irrelevant for patients in the acute phase of their injury, the authors recommend administering the HADS, FSS, and USER-P only in later stages of ABI.
Given that the current study was aimed at identifying important outcome domains for patients with ABI, no measurement instrument was included with regard to caregiver burden. Although ABI can induce a great and lasting burden on caregivers [
44], identifying a suitable measure for the caregivers of persons with ABI was beyond the scope of this research. Future steps in the development of the MDS-ABI need to focus on incorporating caregiver experiences as well, especially in the proxy-module.
Strengths and limitations
The use of a Delphi procedure is a particular strength of the current study. As a Delphi procedure usually is undertaken through multiple rounds of surveys, it allows respondents to adjust their opinion based on other participants’ knowledge, thereby transforming several personal opinions into a broad consensus [
45]. Responses are anonymous; therefore, the Delphi procedure incorporates the advantages of group-based decision making without the disadvantages of face-to-face meetings such as practical implications and dominant characters [
46].
The Delphi panel was composed of 45 healthcare professionals employed in a variety of sectors and disciplines of ABI health care. Accordingly, the results are considered to reflect the general opinion of experts across the field. In consulting experts rather than launching an established MDS-ABI, we anticipated stimulating bottom-up evolution of the dataset to improve support for the use of the MDS-ABI.
Given the fact that experts could access every proposed measurement instrument using survey hyperlinks, we estimate that respondents based their opinions on the quality and appropriateness of measurement instruments rather than familiarity alone.
The current study has several limitations. First, despite of the diversity of the healthcare sectors and disciplines represented by the Delphi panel, the sample was drawn from the professional network of the research team and consisted solely of Dutch participants; hence, it may not represent the opinion of all experts in the national and international field. Nonetheless, all selected measurement instruments are available in English and therefore, the MDS-ABI is suitable for international use. Nonwithstanding, we acknowledge that preferences for specific instruments may differ among countries. For instance, to date, the USER-P has mostly been used in Dutch studies.
Second, persons with ABI were not part of the actual Delphi procedure, although they defined crucial outcome domains during a group meeting preceding round one. ABI patients were not enrolled in the surveys because they lack expertise on measurement instruments and their psychometric properties. However, the experiences of persons with ABI will be evaluated in the consecutive feasibility study, in which the relevance of the selected domains and measures will be evaluated.
Moreover, some remarks should be made on the selected measurement instruments in the MDS-ABI. As the aim of a minimal dataset is to serve as a screening instrument of confined length, the selection of measurement instruments was limited to instruments with a short administration duration. A comprehensive assessment of certain domains arguably offers a better understanding of complaints than could be obtained by using shorter screening instruments. Similarly, there are widely used instruments of high quality that have been developed for patients with a specific diagnosis. Although the inclusion of such measures would not be feasible for a minimal dataset aimed at the heterogeneous group of people with ABI, comprehensive measures such as personality questionnaires, and measures designed for specific diagnoses such as scales assessing arm functioning after stroke, can be added to the MDS-ABI to preference.
Furthermore, only those instruments that were validated and translated into Dutch at the time of the first round of the Delphi procedure complied with the requirements and were proposed to the participants. As a result, recently developed measures were not considered. As outcome measurement after ABI is a developing field, the MDS-ABI will need to be updated regularly in the future.
Last, no suitable instrument was identified for measuring the domains ‘communication’ and ‘social support’. Since both domains were marked as essential for the MDS-ABI, respondents opted to use screening questions. As no validated screening questions for these domains exist, these items were composed by the Delphi panel. Therefore, their sensitivity and feasibility need to be evaluated in future research.
Future research
As a next step, we aim to evaluate the feasibility, usefulness and relevance of the MDS-ABI with healthcare professionals and with ABI patients in several sectors of Dutch health care. On the long term, the MDS-ABI needs to be updated according to future developments in the field of minimal data collection, such as the development and implementation of comprehensive item banks. Such item bank systems, for example PROMIS [
11], can be administered more efficiently using Computerized Adaptive Testing (CAT) [
47]. However, this application is still under construction for most domains and therefore currently unavailable for use as part of the MDS-ABI. Once fully operative, CAT-administered item banks of the self-reported domains could be integrated into the MDS-ABI. Nonetheless, we value the additional clinician-rated parts of the MDS-ABI to obtain a complete image of the patient’s status.