This study examining the preliminary psychometric properties of the SCI-R in adults with T2DM in the UK demonstrated evidence supporting its structure, reliability, divergent validity and known groups validity. Although a uni-dimensional scale could not be confirmed using CFA, exploratory analyses supported a 13-item uni-dimensional scale (with satisfactory reliability), consistent with the findings of the US validation . The internal consistency of the 13-item unidimensional scale was satisfactory, and also consistent with the US validation.
Despite identifying a general factor from which a total score can be computed, the lack of convergent validity for the majority of items indicates that they are relatively disparate, confirming previous findings that different aspects of self-care do not correlate highly [26, 27], and reflecting the multidimensional nature of diabetes self-care .
Indeed, a range of independent behaviours are required for optimal self-management and individuals may choose to undertake certain self-care activities without necessarily taking on others. For example, an individual may record blood glucose results diligently but not think it an important part of his/her self-management to read food labels. This may be due to some aspects of self-care being more/less consistent with others, the value/emphasis placed on each activity by healthcare providers or reflect the variable ease/difficulty of incorporating various self-care behaviours into one’s routine on a regular basis. As has been found with knowledge , scores for individual aspects of self-care activities may be more predictive of various outcomes (e.g. HbA1c) than the total score. In light of these findings (and mixed support for a uni-dimensional scale), we recommend that SCI-R items are scored individually as well as summed to form a total score for some purposes.
Despite our findings that uni-dimensionality may not be necessary (or expected) when assessing self-care behaviours in T2DM, the responsiveness (or sensitivity to change) of individual items may be an issue. Like other measures of self-care behaviour , many items were prone to ceiling effects, which may be a factor of some aspects of self-care being easier or considered more important than others to undertake consistently. This was not reported in the US validation despite similar total scores but the SCI-R was found to be responsive following a psychological or cholesterol-intensive intervention . One of the challenges of using self-report measures of self-care is that they are likely to be prone to social desirability bias (i.e. the individual’s natural tendency to respond to items in a way that he/she believes others would value). Despite being cost-effective and practical, self-report of self-care behaviours is considered by many to be problematic for this very reason. Tools are available to assess the individual’s tendency to respond in socially desirable ways [29, 30] but we were unable to use those in the context of our study and they may not be practical in most research or clinical practice scenarios.
Another challenge facing investigators is determining the significance of any observed change in scores, as statistical significance can often be achieved with large sample sizes. Furthermore, when working clinically, it is not possible to ascertain the statistical significance of a difference between scores at two consecutive consultations for a single patient. In response to this challenge, the minimal clinically important difference (MCID) is a statistical technique that can be defined as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate, in the absence of troublesome side effects and excessive cost, a change in the patient’s management” . The FDA’s draft guidance on the use of patient-reported outcome measures in support of labeling claims  encouraged developers and researchers to identify a MID or MCID as a benchmark for interpreting the clinical importance or relevance of study results to patients – though the more recent definitive guidance has omitted this recommendation . In this UK study, the MCID for the SCI-R total scale was established for the first time, indicating that a minimum change of four points would be required for the change to be considered clinically meaningful, though a more conservative MID suggested that a change of >6.5-7 points would be needed.
In support of its divergent (or discriminant) validity, and as expected, we were able to demonstrate that responses to the SCI-R were largely unrelated to measures of treatment satisfaction (DTSQ) and psychological well-being (W-BQ28). Known-groups validity was partially supported. As expected, we found that those with a lower HbA1c (≤7.5% (≤58 mmol/mol)) reported greater engagement in self-care behaviours. We also found that those with a longer duration of diabetes (>16 years) reported greater engagement in self-care behaviours overall. This finding may be due to the fact that those individuals have had more time to adapt positively to living with diabetes. An alternative explanation may be that increased self-care in those with a longer duration is confounded with the development of complications, increasing the individual’s perceived severity of and susceptibility to negative outcomes and, thus, increasing their engagement with self-care activities. However, we found no differences in total self-care between those with and without complications or between the different insulin treatment algorithms.
Strengths and limitations
The current study offered the opportunity to establish the psychometric properties of the SCI-R in a large cross-sectional study of adults with type 2 diabetes to the UK. While the analyses reported here offer preliminary evidence of validity and reliability, full psychometric validation also involves assessment of test-retest reliability, predictive validity and responsiveness, which requires longitudinal data. This study is also limited by the lack of opportunity for assessing the convergent validity of the measures, i.e. the SCI-R was the only measure of diabetes self-care included in the study. It would have been ideal to assess convergent validity by correlating SCI-R scores with scores on other instruments measuring aspects of diabetes self-care, such as the Summary of Diabetes Self-Care Activities (SDSCA) , a measure of the frequency of performing diabetes self-care tasks. Finally, this study was conducted using data from the AT.LANTUS Follow-on study, which followed up on approximately two-thirds of those in the original AT.LANTUS trial. While the demographics and clinical characteristics of those in our sample were similar to those at the end of the original trial, those who agreed to participate in the AT.LANTUS Follow-on study may well have been more likely to follow their recommended treatments, hence contributing to the ceiling effects observed here.