AIBD indicates autoimmune bullous disease.
Sebaratnam DF, Hanna AM, Chee S, Frew JW, Venugopal SS, Daniel BS, Martin LK, Rhodes LM, Tan JCK, Wang CQ, Welsh B, Nijsten T, Murrell DF. Development of a Quality-of-Life Instrument for Autoimmune Bullous DiseaseThe Autoimmune Bullous Disease Quality of Life Questionnaire. JAMA Dermatol. 2013;149(10):1186-1191. doi:10.1001/jamadermatol.2013.4972
Quality-of-life (QOL) evaluation is an increasingly important outcome measure in dermatology, with disease-specific QOL instruments being the most sensitive to changes in disease status.
To develop a QOL instrument specific to autoimmune bullous disease (AIBD).
A comprehensive item generation process was used to build a 45-item pilot Autoimmune Bullous Disease Quality of Life (ABQOL) questionnaire, distributed to 70 patients with AIBD. Experts in bullous disease refined the pilot ABQOL before factor analysis was performed to yield the final ABQOL questionnaire of 17 questions. We evaluated validity and reliability across a range of indices.
Australian dermatology outpatient clinics and private dermatology practices.
Patients and Exposure
Patients with a histological diagnosis of AIBD.
Main Outcomes and Measures
The development of an AIBD-specific QOL instrument.
Face and content validity were established through the comprehensive patient interview process and expert review. In terms of convergent validity, the ABQOL was found to have a moderate correlation with scores on the Dermatology Life Quality Index (R = 0.63) and the General Health subscale of the 36-Item Short Form Health Survey (R = 0.69; P = .009) and low correlation with the Pemphigus Disease Area Index (R = 0.42) and Autoimmune Bullous Disease Skin Disorder Intensity Score (R = 0.48). In terms of discriminant validity, the ABQOL was found to be more sensitive than the Dermatology Life Quality Index (P = .02). The ABQOL was also found to be a reliable instrument evaluated by internal consistency (Cronbach α coefficient, 0.84) and test-retest reliability (mean percentage variation, 0.92).
Conclusions and Relevance
The ABQOL has been shown to be a valid and reliable instrument that may serve as an end point in clinical trials. Future work should include incorporating patient weighting on questions to further increase content validity and translation of the measure to other languages.
Clinical Trial Registration
anzctr.org.au Identifier: ACTRN12612000750886
Autoimmune bullous disease (AIBD) is characterized by pathogenic autoantibodies directed against structural components of the skin, manifesting clinically as bullae and erosions. With the advent of corticosteroid therapy, these dermatoses are rarely life threatening; however, they have the capacity to severely impinge on a patient’s quality of life (QOL) because the burden comes from living with a chronic condition rather than dying of it. Physical symptoms such as bullae formation, pain, itch, and associated functional limitations can severely affect the patient’s QOL. The disfiguring nature of these symptoms and signs can have a significant burden on social function, independent of clinical severity. In addition, auxiliary considerations, such as financial implications and the adverse effects of treatments, all exert a detrimental effect on QOL.
For many dermatological conditions, a measure of QOL provides a suitable means to monitor disease activity and to evaluate the effectiveness of care. Generic QOL measures have been used by the broader medical community because of their breadth and applicability across different conditions and interventions; however, adequate assessment of QOL necessitates instruments that are capable of capturing any changes, however small, affecting the patient.1 Although generic measures have been tested and validated repeatedly for their psychometric properties, skin-specific QOL instruments have the advantage of being more sensitive and responsive. They generally explore a given domain in greater depth than the corresponding domain of a generic measure, enabling a heightened responsiveness to any changes that may occur.2 Previous studies evaluating QOL in AIBD have used generic QOL instruments such as the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36),3- 6 dermatology-specific instruments such as the Dermatology Life Quality Index (DLQI),7,8 and psychometric tools3,5,6,8 and qualitative evaluation.9 All studies have reported that patients with AIBD have a significant decrease in QOL compared with the greater population. These results have been obtained through the use of generic and dermatology-specific QOL instruments; however, no disease-specific QOL measure has yet been developed to assess QOL in AIBD.10,11
The literature indicates that disease-specific instruments are the tools most sensitive to changes in disease status. The advantage of disease-specific instruments is that questions are perceived to be of direct relevance to patients and to have high face validity, with the specificity of the questions maximizing the ability to detect any changes in disease process.12 The aim of our study was to develop a QOL instrument specific to patients with AIBD.
This study was granted ethical approval by the South Eastern Sydney and Illawarra Area Health Service Human Research Ethics Committee in 2006. Methods similar to those followed in the creation and validation of a QOL instrument for epidermolysis bullosa were used for the development of this QOL instrument for AIBD.13
Patients with a histological diagnosis of an AIBD, including pemphigus, bullous pemphigoid (BP), mucous membrane pemphigoid (MMP), epidermolysis bullosa acquisita (EBA), or linear IgA bullous dermatosis, were eligible for inclusion in the study. Patients were recruited for the item generation process from outpatient and specialized bullous clinics with letters of introduction and telephone calls to obtain consent for participation in the study. Twenty-six patients consented to involvement in the item generation process encompassing patients with pemphigus vulgaris (n = 12), BP (n = 10), pemphigus foliaceus (n = 1), MMP (n = 1), linear IgA bullous dermatosis (n = 1), and EBA (n = 1). Open, nonstructured interviews were conducted with patients and any family or friends present to foster an understanding of the ways in which bullous disease affects QOL. Initially we used open questions, allowing patients to express their views freely, but as the interview progressed, we used closed questions to refine our understanding of the issues, with reflective listening used throughout. Detailed documentation was completed throughout the interview process with direct quotations included to capture the voice of the patients. Nineteen patients were interviewed in person and 7 were interviewed by telephone, with interviews ranging from 10 to 60 minutes (mean duration, 30 minutes). A focus group was organized because, as suggested in the literature, focus groups are known to generate items and yield additional information.14 The 26 patients with bullous disease were invited by telephone or e-mail to participate in the focus group, but the response rate was low and the focus group ultimately consisted of 3 patients and 2 spouses. The session lasted longer than 2 hours, during which information gathered from the individual interviews was reinforced and discussed further. This process of item generation was continued until a saturation point was reached, with no new items being generated. All the patient comments were grouped into categories to form 88 items (Figure).
These 88 items were then combined or deleted to yield the 45-item pilot Autoimmune Bullous Disease Quality of Life questionnaire (ABQOL), which captured the issues established through the item generation process. Similar to other QOL instruments, the pilot ABQOL was self-administered, with close-ended questions and a range of possible responses on a Likert scale assigned numerical values (0-3) according to the degree of effect on QOL. The layout of the questionnaire was consistent, with responses spaced evenly, questions numbered to avoid omission of any items, and items measuring similar variables mixed throughout the questionnaire so that, in the event of patient fatigue, even partially completed surveys had the potential to offer information about a variety of QOL variables.
Letters of introduction were posted to dermatologists registered with the Australasian College of Dermatologists requesting that they invite their patients with AIBD to participate in the study. An additional 16 patients were recruited in this manner, with 28 more patients recruited through ongoing referrals to the bullous clinics of one of us (D.F.M.). In total, 70 patients completed the pilot ABQOL.
Six bullous experts ranked each question in order of importance, and this rank was then correlated with patient sensitivity scores. Insensitive items were defined as those in which more than 50% of patients responded with an extreme value (ie, a response of never or always), which is an accepted threshold in the literature.15 Omitting these items eliminates floor or ceiling effects that reduce the responsiveness of an instrument. To maximize the sensitivity of the instrument in capturing changes in disease status, questions pertaining to treatment and treatment-related adverse effects were eliminated. To refine the questionnaire further, we conducted a principal axis factor analysis on the remaining items. All statistical analyses were conducted using commercially available software (SPSS; SPSS Inc). To assess the dimensionality of the retained items, we performed an exploratory principal components analysis followed by Oblimin rotation.16 Significance was defined as a loading of more than 0.4. Item complexity occurred if an item loaded less than 0.4 or loaded more than 0.4 on more than 1 factor. If an item was complex, it was not discarded but assigned to the factor on which it loaded most heavily if the difference in loading was at least 0.1. Based on these criteria, a final ABQOL questionnaire was produced, consisting of 17 items.
Face and content validity were established through the comprehensive item generation process and review of the questionnaire by the series of bullous experts. Convergent validity evaluates the degree to which scores on one instrument converge with others purported to measure similar outcomes. This convergence was determined by correlating ABQOL scores with other measures of disease burden collected prospectively from our patients attending specialized bullous clinics. We hypothesized that the ABQOL would demonstrate a moderate correlation with the SF-36 and DLQI (Spearman R > 0.64) while also having unique attributes distinguishing it as a superior measurement tool in AIBD (Spearman R < 0.9). The ABQOL scores were also correlated with validated objective measures of disease activity, the Pemphigus Disease Area Index (PDAI)17 and the Autoimmune Bullous Skin Disorder Intensity Score (ABSIS).18 Moderate correlation was also expected with the PDAI and ABSIS scores, although to a lesser degree than that of the DLQI or SF-36, because disease activity is an entity distinct to QOL, and the 2 variables are known to not necessarily correlated.
Discriminant validity evaluates the degree to which one instrument identifies changes in a particular outcome compared with other similar instruments. Discriminant validity was assessed by comparing the proportions of insensitive items in the ABQOL and DLQI using the Fisher exact test.
Internal consistency is a measure of reliability and refers to the degree to which items in a given instrument correlate with one another. Internal consistency was measured using the Cronbach α coefficient, which also tested for construct validity. Construct validity was additionally assessed through the factor analysis process.
A subset of 20 patients was readministered the final questionnaire within 3 days after completing the pilot questionnaire. The results of both questionnaires were compared, and the percentage variation in response was calculated to measure the reliability of questionnaire scores on repeated administration.
Of the 70 patients included in the analysis, the sex distribution was approximately even, with 36 male and 34 female participants (Table 1). Patient age ranged from 23 to 95 years, with a mean age of 60.1 years. Most of the patients had pemphigus vulgaris (n = 28), followed by BP (n = 25), pemphigus foliaceus (n = 8), MMP (n = 4), EBA (n = 3), and linear IgA bullous dermatosis (n = 2). The mean (SD) ABQOL score for patients with pemphigus vulgaris was 11.5 (5.5); for BP, 8.4 (5.5); and for the remaining patients, 11.9 (8.9).
The Kaiser-Meyer-Olkin measure of sampling adequacy (0.675) and the Bartlett test of sphericity (P < .001) suggested that factor analysis of the data was appropriate.16 Although the Kaiser criterion (factors with an eigenvalue >1) and the Catell scree plot16 suggested retaining 5 factors, we selected the first 3 (psychosocial, symptom, and mucosal) based on comprehensibility (ie, psychometrically meaningful factors with at least 2 items). The 3 retained factors represented 52.7% of the cumulative variance of the 17 items. Nine items loaded on the psychosocial subscale and 5 questions loaded on the symptom subscale, including those concerning function (questions 4 and 5), suggesting that the impaired functioning depended on the extent of reported symptoms (questions 1, 2, and 3). Three questions loaded on the mucosal subscale (Table 2). Of the 17 questions, 4 questions demonstrated complexity. Items 11, 12, and 15 (anxiety, family/friends, and social life) loaded on the psychosocial and symptom subscales, suggesting that these items assessed 2 factors. These items were retained and assigned to the psychosocial subscale because they loaded considerably more toward the symptom factor. Item 4 (clothing changes) also was assigned to the symptom subscale because it more accurately represented that subscale (symptom factor loading, 0.64 compared with psychosocial factor loading, 0.56). Item 9 (embarrassment) loaded significantly on the symptom subscale (0.69) and just below the 0.4 cutoff on the psychosocial subscale (0.37). In assessing the face validity of this individual item, embarrassment was seen to be more relevant to the psychosocial rather than the symptom aspects of AIBD. Hence, despite the psychosocial loading being less than 0.4, this item was considered complex and assigned to the psychosocial subscale.
Face and content validity of the questionnaire were assessed through the comprehensive item generation process. In terms of convergent validity, the ABQOL showed poor convergence with the overall SF-36 questionnaire (R = 0.51) and function-specific subscales (R < 0.52) but significant convergence with the general health subscale (R = 0.69, P = .009). Moderate convergence was observed with the DLQI (R = 0.64) as expected. With regard to discriminative validity, the ABQOL was also found to have 7 of 17 insensitive items. This finding was significantly less than the DLQI, which had 9 of 10 insensitive items (Fisher exact test, P = .02) in the 51 patients for whom scores were available, but significantly more than the SF-36, which was found to have 17 of 36 insensitive items (Fisher exact test, P = .01) in the 27 patients for whom scores were available.
In terms of the measures of disease activity, data were available for 34 patients encompassing 210 clinical encounters. The correlation between the ABQOL and disease activity measures was low for the PDAI (R = 0.42) and the ABSIS (R = 0.48). Proportions of insensitive items could not be calculated for the ABSIS and PDAI owing to the largely dichotomous nature of the individual questions in these measurement tools (Table 3).
The construct validity and internal consistency were acceptable, with a Cronbach α of 0.84. The test-retest reliability of the questionnaire was acceptable, with a mean test-retest value of 0.92.
To our knowledge, the ABQOL represents the first disease-specific QOL instrument for use in AIBD, and our findings indicate that the instrument is a valid and reliable patient-based measure. The statistical analyses we performed indicate that the ABQOL is relatively free from error and will yield consistent results for a patient under similar conditions. No items were considered difficult, the response distribution was optimal for most questions, and the instrument was found to have high validity and reliability across a range of indexes. The ABQOL was significantly superior to the DLQI in terms of sensitivity, reflecting the precision of disease-specific QOL instruments. Extant QOL instruments that have been developed for the general dermatology population do not necessarily review the QOL domains relevant to patients with a given dermatosis.19 For instance, of the 120 patients interviewed for the development of the DLQI, only 1 patient had a blistering disease2; accordingly, such instruments may not tap into the QOL dimensions relevant in blistering conditions.
The ABQOL was developed with input from patients with all subtypes of AIBD, and the final questionnaire was honed on the basis of results obtained from a similarly representative cohort. Patients with pemphigus and BP were overrepresented in our study, but this overrepresentation reflects the higher prevalence of these dermatoses within the population with AIBD. Those diseases with more mucosal involvement, such as EBA and MMP, gave composite, equivalent scores to the patients with pemphigus vulgaris in the moderate range, and those for patients with BP were slightly lower. This finding fits with our impression of the relative severities of these disorders. The disease in most of the patients in our study was in remission with immunosuppressive therapy, which parallels the state of disease most patients will have during routine clinical follow-up. This population also accounts for some of the discrepancy observed between scores and other measures of disease burden. As expected, only a moderate correlation was observed between the ABQOL and DLQI because, although both instruments evaluate QOL, we believe that the ABQOL taps into the QOL domains specifically affecting patients with AIBD, which was reflected by the differences in scores. The correlation seen between the ABQOL and the SF-36 subscales was lower than expected, with the physical and functional subscales of the SF-36 generally having higher correlations than the psychosocial subscales (P < .001). Patient feedback suggested that these correlations may result from the general nature of questions in the SF-36 not capturing the psychosocial burden of AIBD and the less-than-optimal level of patient response due to high levels of patient fatigue on completing the validation questionnaires.
The intermittent, situational exacerbations of the emotional and psychological burden of dermatological disease are not effectively captured in the wording of the SF-36 items, hence reducing their content and face validity in quantifying psychosocial QOL effects. From a content and face validity aspect, the ABQOL is much better placed (as a disease-specific tool) to achieve accurate QOL quantification with regard to the psychosocial burden. This finding is supported by the fact that the ABQOL maintained a moderate correlation with the DLQI, a dermatology-specific tool that refers specifically to the impact of the disease.
The correlation among the ABQOL, PDAI, and ABSIS was lower than expected but plausible given that the insult to QOL experienced by patients is often independent of the visible disease burden. Even patients with relatively low disease activity experienced a detriment to their QOL, which highlights the need for an AIBD-specific QOL instrument because it may identify changes in QOL independent of changes in disease severity. The literature suggests that patients and dermatologists often have different perceptions of a patient’s QOL and that this discrepancy is associated with lower satisfaction with medical care.20 Results gleaned from the ABQOL can be used to evaluate QOL objectively and to analyze which aspects of disease most affect a patient (the symptoms themselves, social burden, etc), which should facilitate the provision of personalized care and increase patient satisfaction. In our analysis, the questions that consistently yielded the highest answers were question 4 (clothing changes; mean, 1.46), question 2 (itching; mean, 1.32), and question 9 (embarrassment; mean, 1.13), demonstrating the multiple QOL domains affected by AIBD.
The ABQOL can be used to quantify the effect of a patient’s AIBD on their QOL and capture changes in disease status, which may not be appreciable during routine clinical review. We hope that the ABQOL may be used to map the trajectory of their disease and to evaluate the effect of therapeutic interventions. The ABQOL also represents a promising patient-based measure that can be used as an end point in clinical trials. Given the variety of outcome measures used in studies of AIBD,21 the need for a uniform outcome in this setting is particularly compelling. Further studies in cohorts with more active disease or incident cases are needed because the ABQOL has been validated in patients with predominantly quiescent disease, and the responsiveness of the instrument needs to be established. Because relatively few of the patients with AIBD interviewed for this study had MMP, linear IgA bullous dermatosis, or EBA, once the questionnaire has been validated in foreign languages and cultures, we will be able to perform larger collaborative studies interviewing these rarer cohorts of patients to compare their QOL with that of patients with pemphigus and pemphigoid. The items related to adverse effects of treatments that were eliminated from the pilot ABQOL to construct the final questionnaire will be studied as a separate QOL issue related to therapy. Future work also includes revising the instrument by incorporating patient weighting on questions to further increase content validity and translation of the measure to other languages.
The ABQOL is the first QOL measure specific to AIBD and has been shown to be a valid and reliable tool. The specificity of the tool enables it to overcome the disadvantages of more generic QOL instruments and represents a promising patient-based measure to quantify disease burden, monitor disease activity, and map QOL dimensions as interventional targets.
Accepted for Publication: April 4, 2013.
Corresponding Author: Dédée F. Murrell, MA, BMBCh, MD, FACD, Department of Dermatology, St George Hospital, University of New South Wales, Gray Street, Kogarah, Sydney, New South Wales, Australia 2217 (firstname.lastname@example.org).
Published Online: August 7, 2013. doi:10.1001/jamadermatol.2013.4972.
Author Contributions: Drs Sebaratnam, Hanna, Chee, and Murrell had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sebaratnam, Chee, Martin, Murrell.
Acquisition of data: Sebaratnam, Hanna, Chee, Venugopal, Daniel, Rhodes, Tan, Wang, Welsh, Murrell.
Analysis and interpretation of data: Sebaratnam, Hanna, Frew, Venugopal, Wang, Nijsten, Murrell.
Drafting of the manuscript: Sebaratnam, Hanna, Daniel, Tan, Murrell.
Critical revision of the manuscript for important intellectual content: Sebaratnam, Hanna, Chee, Frew, Venugopal, Martin, Rhodes, Wang, Welsh, Nijsten, Murrell.
Statistical analysis: Sebaratnam, Hanna, Frew, Venugopal, Wang, Nijsten, Murrell.
Administrative, technical, and material support: Sebaratnam, Chee, Venugopal, Rhodes, Tan, Welsh, Murrell.
Study supervision: Venugopal, Daniel, Murrell.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported in part by grants from Dr Murrell, the Australasian Blistering Diseases Foundation, and the University of New South Wales Independent Learning Program to work with Dr Murrell (Drs Sebaratnam, Chee, Hanna, and Frew and Messrs Tan and Wang). Dr Sebaratnam received a travel grant from the European Society for Dermatological Research.
Previous Presentations: Preliminary data from this study were presented at the 41st Annual Meeting of the Medical Dermatology Society; March 4, 2010; Miami, Florida; and at the 41st Annual Scientific Meeting of the European Society for Dermatological Research; September 8, 2011; Barcelona, Spain.
Additional Contributions: Christopher McCormack, MDDS, FACD, assisted with patient recruitment. We thank the patients with AIBD who kindly volunteered their thoughts and time to assist with this project.