We have described the development of an outcome measure for quantifying patients' experiences of treatment for MS relapses, to be used as the primary outcome measure in a randomised controlled trial of service delivery . Importantly, the MSRMS satisfies current health service aspirations towards a patient led service . It is a patient reported measure, which enables relapse management to be evaluated from their perspective. In addition, patients themselves determined the domains of the MSRMS, therefore allowing them to define how services are evaluated.
The MSRMS satisfied most published criteria for robust measurement, with its psychometric performance being reasonably constant across two independent samples. However, scores in the postal survey sample were better distributed than in the RCT sample. There could be a number of reasons for this, including differences in the characteristics of the sample. The RCT sample included more people who received steroids at home, who may have reported better experiences of relapse management , due to a number of reasons, for example, the RCT sample may have known that outcomes were being assessed, whereas the postal survey sample received standard NHS care, which may have been more variable.
All four MRMS subscales had notable floor effects in the RCT sample indicating that the minimum possible score was commonly seen. From a clinical perspective, this finding might be taken to imply that many people considered that their relapse management could not have been better. This could, of course, be due to a number of reasons, for example, patients may not have wished to be critical about a service on which they rely, or they may have a positive rapport with the care team that make them forgive any deficiencies. Also from a conceptual perspective, it seems unlikely that care for these people could not have improved, even though floor effects and skewed positive responses are common in surveys that elicit patients' views of their health care . Alternatively the floor effects could represent limitations in the measurement range covered by our scale. A qualitative evaluation of people at the floor end of the scale should help to determine whether their experiences could be improved, and will provide an evidence base for future development of the MSRMS.
The fact that we removed almost 70% of items from the preliminary scale may suggest that important information regarding patients' views on relapse management may have been discarded. However, the aim of the item generation stage was to be as inclusive of a wide range of patients' views as possible (inpatient, outpatient and home). Our over-inclusiveness at the item generation stage led to the inclusion of items that were not applicable across the different modes of treatment. Consequently, these items went on to fail the psychometric criteria, and hence were removed.
Results from this study improve our understanding of patients' experiences of relapse management. First, the process of scale development has defined four key domains for intervention and measurement (access to care, interpersonal care, information, coordination of care). This provides a clinical framework for evidence-based patient-focused relapse management. Second, the scale gives clinicians a robust mechanism for measuring the outcomes of relapse management. Of note, 75% of patients endorsed "excellent" on the single item question of quality of care, supporting the concern that simple patient satisfaction scores and global indicators of quality, may present limited and optimistic pictures , unlikely to be sufficient to provide the evidence base to evaluate and improve complex service delivery .
This study has some limitations. First, the sample sizes were relatively small, compared to other scale development studies. Second, although the interview sample included patients from two hospital sites, the postal survey and the trial sample were recruited from one hospital setting. Further psychometric evaluation of the scale is needed in a larger diverse sample of patients across a number of hospital sites. Cognitive interview techniques  could also be used to identify whether any other issues of importance regarding MS relapse management could be added. It is also important to evaluate the measure using other psychometric paradigms, for example Rasch analysis or item response theory [28, 29]. Additional items could also be generated that could discriminate amongst those who rate the care as excellent. Rasch analysis in particular can aid the selection of these additional items. Rasch analysis can also further assess the unidimensionality of the scales. Finally, although patients' perceptions of care delivery are crucial, it is important that correlations with an objective measure of the quality of medical care are established before any public policy decisions can be made.
In conclusion, we have developed a measure of patients' experiences of treatment for MS relapses. It has robust psychometric properties, and could be extended to compare patients' experiences of relapse management across other treatments and locations (e.g. GP surgeries).