Minimally Important Differences of the UCLA Scleroderma Clinical Trial Consortium Gastrointestinal Tract Instrument

MATERIALS AND METHODS

Patient characteristics and study methods have been published^5,8. In brief, patients with SSc and GIT involvement were invited to participate at the following 3 scleroderma centers in the United States: UCLA, Los Angeles, CA; University of Michigan, Ann Arbor, MI; and University of Texas at Houston, Houston, TX. The protocol was approved by the institutional review board at each institution (UCLA approval no. 07-08-004-04), and each subject signed a consent form prior to completing the questionnaires. In addition to completing the paper-and-pencil UCLA SCTC 2.0 questionnaire⁹, patients reported their age, sex, race/ethnicity, and level of education. Each physician did a physical examination to determine the type of SSc (limited or diffuse cutaneous) and provide their GIT diagnoses.

Patients were re-administered the UCLA SCTC GIT 2.0 during their second clinic visit. We used 3 different patient-reported anchors. Patients rated their overall and upper and lower GIT involvement: (1) “Compared to your LAST VISIT, how would you rate your overall gastrointestinal symptoms?”; (2) “Compared to your LAST VISIT, how would you rate your upper gastrointestinal symptoms (such as heartburn, nausea, vomiting, bloating or gas or air in the stomach)?”; and (3) “Compared to your LAST VISIT, how would you rate your lower gastrointestinal symptoms (such as diarrhea or constipation)?”. Responses were provided using a categorical response scale: “Much better; somewhat better; almost the same; somewhat worse; or much worse.”

For Reflux and Distension/Bloating scales, we used the overall and upper GI items as anchors to estimate the MID. For Diarrhea, Constipation, and Fecal Soilage scales, the overall and lower GI scales were used as anchors. For the Emotional Well-being and Social Functioning scales and the total GIT score, we used all 3 rating items as anchors.

The MID was estimated by examining change in scores of different GI scales (Time 2 – Time 1) in subjects who reported they were somewhat better or somewhat worse. MID have been found to range between effect sizes (ES) of 0.20 and 0.50¹⁰. ES is the ratio of observed change to a measure of variance¹¹ and was defined as (mean score for individual scale at Time 2 – mean score for individual scale at Time 1)/SD baseline. Because we had multiple anchors for each scale, we present individual MID estimates and as an average across different anchors.

To assess the usefulness of an anchor, previous research has recommended reporting the correlation between the anchor and the change score; for example, a correlation of zero will make the anchor useless and a correlation of at least 0.30–0.35 has been suggested^10,12. We assessed the association between the anchors and the change scores for scales using the Spearman rank-order correlations to account for the ordinal level of measurement of the anchors.

All analyses were performed using Stata software version 10.2 (Stata Corp., College Station, TX, USA).

RESULTS

DISCUSSION

MID estimates provide a benchmark for interpretation of results by helping researchers and clinicians understand whether HRQOL score differences between 2 treatment groups are meaningful, or if changes within one group over time are meaningful⁷. For example, an average change of 0.15 points (on a 0–3 scale) for a patient-reported measure may be statistically significant in a clinical trial, but may not be perceived as beneficial by the subjects. It is also important to note that MID estimates are applicable at the group level and not at the individual level. Other statistical tests have been recommended to assess statistical significance at an individual level^13,14.

The UCLA SCTC GIT 2.0 was developed to assess severity of SSc-related GIT symptoms and effects of GIT symptoms on emotional well-being and social functioning^5,8. In this study, we present MID estimates for improvement and worsening that are applicable for interpretation of scores in clinical trials and observational studies. Our study is in alignment with other studies that have shown that an effect size of 0.20 to 0.50 corresponds to the MID for patient-reported outcome measures^14,15,16. Also, the MID estimates were larger than those for the “no change” group, providing confidence in our estimates.

Previous reports have shown that the MID estimates may differ for worsening compared to improvement groups^14,17,18. Therefore, we decided to present the MID estimates separately for improvement and worsening groups. Our results are in agreement with the published data. On average, our MID estimates for the improved group were larger than those for the worsened group. The only exception was the Social Functioning scale, where MID estimates were 0.07 for the improved group and 0.21 for the worsened group.

Although we show that an improvement of 0.26 points (on a 0–3 scale) in the Reflux scale is the MID estimate, it should not be interpreted that a change of less than 0.26 points is not clinically important, as there is an inherent uncertainty around MID estimates. Previous studies have reported this uncertainty around the MID estimates^19,20, hence experts recommend using several anchors. In addition, they suggest gathering data from both observational and clinical trials to support confidence in MID estimates¹⁰, as it is unlikely that a single MID estimate is applicable to all patient populations. Despite this uncertainty, these data can be used to interpret clinical trial data and observational studies.

Our study has several strengths. Our MID estimates are based on a large sample size in patients at 3 US scleroderma centers. Second, we prospectively incorporated anchors in order to estimate the MID.

Our study also has limitations. As reported⁵, we used only patient-reported anchors to estimate MID. We did not include radiological test measures such as the gastroesophageal endoscopy and breath test in this study. Future studies should corroborate our estimates using these tests and different anchors. Second, our study population generally had mild to moderate GIT disease (self-rated), with only 9% of patients stating severe to very severe GIT disease. The estimates may differ by severity of illness¹⁰. In addition, as previously seen^14,21, the majority of patients in our study considered themselves about the same between the 2 timepoints. Therefore, these data should be considered preliminary and should be confirmed with larger cohorts and/or clinical trials.

We provide MID estimates for the UCLA SCTC GIT 2.0 scales. This information can aid in interpreting scale scores in future randomized controlled trials and observational studies.