Published ahead of print on May 8, 2003, doi:10.1164/rccm.200211-1343OC
© 2003 American Thoracic Society The Sarcoidosis Health QuestionnaireA New Measure of Health-related Quality of LifeDivision of Pulmonary and Critical Care Medicine, Department of Medicine, University of North Carolina at Chapel Hill, Chapel Hill; Division of Pulmonary and Critical Care Medicine, Department of Medicine, East Carolina University, Greenville, North Carolina; and Division of Pulmonary and Critical Care Medicine, Department of Medicine, Medical University of South Carolina, Charleston, South Carolina Correspondence and requests for reprints should be addressed to Christopher Cox, M.D., M.P.H., Duke University Medical Center, Division of Pulmonary and Critical Care Medicine, Box 3221, Durham, NC 27710. E-mail: christopher.cox{at}duke.edu
The Sarcoidosis Health Questionnaire (SHQ) is a health-related quality-of-life (HRQL) instrument we designed in a two-part study to allow patients to describe their satisfaction with life as it is affected by sarcoidosis. In the Development Study, we created the SHQ from a 151-item pool generated from interviews with 107 patients, a search of the relevant literature, and discussion with sarcoidosis experts. Using clinical impact methodology and the questionnaire responses of a separate group of 149 patients, we reduced the total number of items to 29. The final SHQ has three domains: Daily Functioning, Physical Functioning, and Emotional Functioning. In the Validation Study, performed with a different group of 111 patients, we found that the SHQ had good internal consistency reliability, as well as evidence of content, criterion, and construct validity based on its comparison with other measures of HRQL (Medical Outcomes Study 36-Item short form and the St. George's Respiratory Questionnaire) and of mental health (the Center for Epidemiologic Study—Depression Scale), as well as with clinical variables including spirometry. The SHQ alone was sensitive to differences in HRQL based on the number of involved organ systems. In conclusion, we have developed a self-completed HRQL questionnaire for U.S. patients with sarcoidosis.
Key Words: health status indicators • quality of life • questionnaires • sarcoidosis Sarcoidosis is a multisystemic disease of unknown etiology with no known cure, characterized by variable clinical manifestations and an unpredictable course (1). Appreciated primarily as a pulmonary condition, sarcoidosis may involve nearly any organ system and can affect persons of all races and ages (1). Although sarcoidosis has a low mortality rate, it can become a chronic condition associated with wide-ranging physical and mental disability (1–3). Because current therapies for sarcoidosis do not alter disease progression (4), the primary aim of treatment is the relief of symptoms. However, clinicians are challenged in their assessment of patients with sarcoidosis because there are no validated tools that allow either reliable measurement of disease activity or accurate prediction of disease course (1). Because most patients with sarcoidosis have pulmonary involvement, evidence of clinical improvement has been based traditionally on spirometry values and chest radiograph appearance. However, these measures fail to address other systemic manifestations of sarcoidosis and have been shown in other chronic pulmonary conditions to correlate weakly with measures that reflect issues of specific importance to patients, such as health-related quality of life (HRQL) (5). Therefore, assessing the HRQL of persons with sarcoidosis could complement existing physiologic measures by helping clinicians understand an individual's unique perceptions of how disease and treatment affect their satisfaction with life in diverse areas of functioning (6). Reports describing the HRQL of patients with sarcoidosis are limited to a few small cross-sectional studies from northern Europe (7–9) and the United States (10), in which a variety of generic and respiratory-specific questionnaires were used to measure HRQL. However, generic HRQL questionnaires are unlikely either to identify small but clinically significant differences in HRQL between patients with sarcoidosis or to be sensitive to subtle yet important changes over time in the HRQL of a patient with sarcoidosis (6, 11–14). On the other hand, it is doubtful that respiratory-specific instruments can reflect adequately the overall impact of this multisystemic disease on a person's satisfaction with life, especially if extrapulmonary manifestations are present. Developing a sarcoidosis-specific HRQL questionnaire could address these concerns with existing questionnaires while also providing a unique outcomes measure for use in future clinical trials. In this study, our primary aim was to create a short, self-completed measure of HRQL for persons with sarcoidosis of varying disease burdens and organ involvement, using accepted methodologic standards (15) and the guidelines of past research (11–14). We aimed to develop a questionnaire with both discriminatory and evaluative properties (6), although in this article we report only on testing of its discriminatory properties. Our study was performed in two parts with separate groups of patients (see Figure E1 in the online supplement): in Part 1, the Development Study, we developed and pretested the Sarcoidosis Health Questionnaire (SHQ). In Part 2, the Validation Study, we evaluated the reliability and validity of the SHQ.
Principles of Instrument Development We defined quality of life as an individual's overall satisfaction with life and functional status and health-related quality of life as the quality of life as it is affected directly and indirectly by their perceived health (16). We framed our research design using the theoretical model of Wilson and Cleary (17).
Development Study
Item Reduction.
Validation Study
Statistical Analysis Patients were excluded if they were less than 18 years of age, were organ transplant recipients, had active cancer, were not fluent in English, had impairments preventing the completion of study questionnaires, or did not have biopsy-proven sarcoidosis (1). The institutional review boards of the University of North Carolina at Chapel Hill (Chapel Hill, NC), the Medical University of South Carolina (Charleston, SC), and East Carolina University (Greenville, NC) approved our research protocol.
Patients The clinical characteristics of the two groups of study patients are shown in Table 1 . A total of 13 patients in the Development Study and 8 in the Validation Study did not complete all questionnaires fully and were excluded from further analyses. For the remaining 149 Development Study patients, the median age was 44 years; most were African-American (80%) and female (69%). The characteristics of the remaining 111 Validation Study patients were nearly identical to those of Development Study patients except for the percentage currently being treated for sarcoidosis (70 versus 52%, p = 0.002) and the percentage treated with oral corticosteroids (66 versus 50%, p = 0.03), both of which were greater among Development Study members. Validation Study participants had median percent predicted FEV1 and FVC values that were greater than 80%, higher than observed in past assessments of HRQL among patients with sarcoidosis (7, 10).
Development Study Frequencies, importance scores, and impact scores (frequency x importance score) for each of the SHQ's 29 items calculated from Development Study participant responses are shown in Table 2 . Problems commonly identified by patients included low energy (84%), worrying about their sarcoidosis worsening (83%), having a sensation of uncomfortable breathing during daily activities (81%), and feeling bodily pain (79%). Impact scores for the 29 items ranged from 2.41 to 3.74. We arranged the SHQ items in three domains guided by the concept of "clinical sensibility" (32) and our a priori predictions of areas of importance to patients: Daily Functioning (13 items), Physical Functioning (6 items), and Emotional Functioning (10 items). We also used exploratory principal components factor analysis to ensure that our interpretation of item relationships was reasonable and found that a one-factor solution existed (30) (see additional description in the online supplement). Final items in the unidimensional, multidomain questionnaire were written at a sixth grade reading level and were scored using a Likert Scale, with responses ranging from 1 ("all of the time") to 7 ("none of the time"). We used Cronbach's coefficient  , a measure of the proportion of variance in an observed score attributable to the "true score" being studied (and thus the proportion of variance that is reliable), to assess internal consistency reliability of the domains of the new questionnaire (30). Coefficient values may range from 0 (low reliability) to 1 (perfect reliability). The coefficient values of the Daily Functioning, Physical Functioning, and Emotional Functioning domains were 0.92, 0.84, and 0.87, respectively, indicating excellent internal consistency reliability.
Validation Study We tested the criterion, convergent, and divergent validity of the SHQ (6) in the Validation Study. The criterion validity of the SHQ was addressed by evaluating the discriminatory performance of the SHQ relative to legitimate "gold standards" of patient-centered and physiologic outcomes measures. Table 3 demonstrates differences in scores on the SHQ, the SF36, and the SGRQ by clinical characteristics including mental health and pulmonary function. The SHQ total score demonstrated statistically significant differences between all groups examined in these bivariate analyses except by the percent predicted FEV1. Only the SGRQ was able to show statistically significant differences in HRQL scores between the groups divided by FEV1 values.
We also tested the criterion validity of the SHQ by evaluating its ability to discriminate between patients on the basis of their burden of organ involvement. Table 4 demonstrates that the SHQ was the only HRQL questionnaire able to demonstrate (in all domains) statistically significant differences between persons with one, two, or three or more organ systems involved by sarcoidosis. Because of the small number of patients with involvement of certain organ systems, further testing of specific organ effect was not possible. SHQ scores were also significantly different (p < 0.001) for self-reported frequency (frequent, occasional, rarely) of symptoms perceived to be related to sarcoidosis.
The convergent validity of the SHQ as a discriminative assessment of HRQL was supported by Pearson correlations demonstrating strong, statistically significant associations (all p < 0.001) between the SHQ total score and other questionnaires and clinical variables (Tables 5 and 6) , including all SGRQ domain scores (all r > –0.65), the SF36-PCS (r = 0.75), and the SF36-MCS (r = 0.68). The SHQ Physical Function score demonstrated significant correlations (r > 0.75) with the SF36-PCS and all SGRQ domains, although it had a much weaker association with the SF36 mental composite score (r = 0.40, p < 0.001). It also had the highest correlation of any SHQ domain with spirometric values as well as stronger correlations (r = –0.31, p < 0.001) with the MRC Dyspnea Scale than was possessed by either the SGRQ (all r < 0.28 and p < 0.04) or the SF36 Physical Component score (r = 0.22, p = 0.019). The SHQ Emotional Functioning score was associated with both depression (r = 0.66, p < 0.0001) and the SF36 Mental Component score (r = 0.75, p < 0.0001), although it showed no statistically significant relationship with either dyspnea or spirometry. The Daily Functioning domain of the SHQ had stronger relationships with the SF36 Vitality, General Health, and Social Functioning domains (all r > 0.70, p < 0.001) than did the other SHQ domains. Finally, the SHQ total and domain scores all demonstrated high correlations (r > 0.54, p < 0.0001) with patients' self-rated quality of life.
We assessed the divergent validity of the SHQ by testing its relationship to variables that we expected would correlate only weakly with HRQL. As we expected, education, age, and smoking status demonstrated low correlations with SHQ domain and total scores as shown in Table 6. In addition, race, marital status, employment status, and possession of health insurance were not strongly associated with SHQ domain and total scores (all r < 0.1 and p > 0.05). Further evidence of the discriminant properties of the instrument was provided by the normal distribution of total SHQ scores (median, 4.0; interquartile range, 2.4–5.8). This distribution reflected the absence of either high ("ceiling effect") or low ("floor effect") score clustering (see Figures E2–E5 in the online supplement).
Using exploratory principal components factor analysis, we again found that the SHQ was unidimensional (one-factor solution) (see additional data in the online supplement). We also found that the good internal consistency reliability of the SHQ was preserved among Validation Study participants for Daily Functioning, Physical Functioning, and Emotional Functioning domains (Cronbach's
The varied manifestations of disease in patients with sarcoidosis may result in physical, emotional, and social problems that, in turn, can diminish their perceived HRQL (2). However, only a few published reports from Europe (2, 7–9) and the United States (3, 10) have described the health outcomes of these patients. To facilitate the measurement of health concerns that affect quality of life among patients with sarcoidosis, we have developed and performed early validity and reliability testing of a sarcoidosis-specific HRQL questionnaire, the Sarcoidosis Health Questionnaire (SHQ), by adhering to established methodologic guidelines (11) and by following a recognized framework of questionnaire design (11–14). The SHQ requires no investigator supervision, can be scored easily, requires only about 10 minutes to complete, and appears more sensitive than other HRQL questionnaires to the extent of sarcoidosis organ system involvement. Importantly, the SHQ includes only items that patients with sarcoidosis felt were important. We believe that use of the SHQ may improve communication between clinicians and patients by facilitating discussion (31) about issues that may not be commonly addressed during an encounter, such as depression, or easily measured, such as musculoskeletal pain and fatigue (2). This may be especially important because clinicians may not accurately recognize elements of illness that are important to patients (33) or appreciate the variability in different individuals' unique responses to particular symptoms. The SHQ could be used also to prioritize existing problems by their importance to patients as well as to assist in treatment decision-making by measuring the individual responses of patients to illness (34). It could also help clinicians focus more attention on a patient's individual perceptions of illness than clinical measures of their disease (34). As the first outcomes measure with evidence of validity designed specifically for use among persons with sarcoidosis, we believe that the SHQ could also be administered in clinical trials to assess the impact of particular interventions on patient satisfaction with life. Our results should be considered in the context and limitations of our research design. First, we developed the SHQ as a discriminative and evaluative questionnaire, but have not yet tested the evaluative properties of the SHQ by testing its responsiveness to change in HRQL over time. Related to this, the minimal clinically important difference in SHQ score has not been assessed through longitudinal testing, although HRQL instruments that use a seven-point scoring scale typically demonstrate clinically significant changes in HRQL at 0.5-unit intervals (35). Even though the convergent validity of the SHQ appears acceptable by virtue of its strong correlations with other complementary HRQL questionnaires and clinical variables, it is important to recognize that the SHQ includes some items with wording similar to those found in these same questionnaires, possibly inflating observed correlations. Because we examined relatively few variables with expected low correlations in testing the divergent validity of the SHQ, the true strength of its divergent validity should be evaluated further. Nevertheless, both the strength and direction of associations observed between the SHQ, other HRQL measures, and both related and unrelated clinical variables reflected our general expectations. The effect of pharmacologic treatment on our results is important to address. Oral corticosteroids are the primary therapy given to symptomatic patients with sarcoidosis, yet are associated with a wide array of potential adverse effects that could have been reflected in the items included in the SHQ. Even steroid-sparing agents such as methotrexate and hydroxychloroquine possess troublesome side effects that could affect the perceived HRQL of an individual. Had we excluded patients who were prescribed pharmacologic therapy, we could have narrowed greatly the clinical diversity of our population and therefore created a questionnaire that would not have been applicable to the broader population of patients with sarcoidosis. In this scenario, only items important to healthier patients who did not take sarcoidosis-specific medications would have been included. Nevertheless, the reader should be cautioned when interpreting questionnaire scores of patients who are prescribed pharmacologic therapies for sarcoidosis, because separating the "disease effect" from the "drug effect" in patients' HRQL may be difficult. Although patients who were prescribed medications for sarcoidosis had lower SHQ scores (indicating lower perceived HRQL) than those who took no sarcoidosis medications, we found that treated patients also had greater organ system burdens of disease, more severe dyspnea, lower FEV1 values, and more frequent symptoms (see Table E1 in the online supplement). This suggests that the SHQ likely samples more than just medication effect. Also, even though all the items included in the SHQ were important not only to the great majority of those both taking sarcoidosis medications and those who took no medications for sarcoidosis, it is possible that some items may be more responsive to the concerns of those receiving therapy, such as those addressing weight gain and medication dependence (see the online data supplement for further discussion). Because the decision to prescribe therapy is based on the individual's unique disease manifestation, patient communication of symptoms importance, and physician-specific decision-making regarding the potential usefulness of such therapy, there are multiple complex layers of potential confounding that may distort the scores of any HRQL questionnaire. Future prospective study will be helpful in understanding more clearly how specific medications may impact patient perceptions of HRQL. We observed that the SHQ had a relatively weak association with spirometry values, which, although similar to other findings with HRQL questionnaires (5, 7, 36), may surprise many clinicians. However, correlations between spirometry measures and the Short Form 36, the St. George's Respiratory Questionnaire, and the Medical Research Council Dyspnea Scale were low as well—lower, in fact, than those reported by others (5, 10). It is possible that the respiratory-specific discriminatory power of these HRQL questionnaires was limited by the low prevalence of airflow abnormalities and only moderate severity of dyspnea in our stable outpatient population. However, support for the overall discriminatory power of the SHQ was provided both by its separation of patient HRQL compromise on the basis of total organ system involvement and the normal distribution of SHQ scores, reflecting the absence of either ceiling or floor clustering. Future administration of the SHQ in populations with a greater variety of respiratory impairment as well as by comparison with complementary measures of pulmonary function (37) could help clarify its true sensitivity to differences in respiratory function. There are also relevant issues to address that relate to our study participants. Because of the university medical center population we sampled, it is possible that our results may reflect a tertiary care center bias and therefore may not be generalizable to patients monitored in community settings. However, our patient population is comparable in clinical characteristics to those enrolled in the largest prospective epidemiologic study of sarcoidosis studies yet undertaken, the National Heart, Lung, and Blood Institute-funded ACCESS study (38), and is therefore likely to reflect the diversity of U.S. patients with sarcoidosis at least reasonably well. Our study population could have clinical characteristics that may limit our questionnaire's generalizability to non-U.S. populations (2, 7–9). The majority of our patients with sarcoidosis were of African ancestry, a group that may experience a more severe disease course (1) and have less access to health care (39) than persons with European, primarily white, backgrounds. As a result, the items included in our HRQL questionnaire may differ from those included had the study been performed within another population. It is possible that differences in SHQ scores may exist among European or Asian patients with sarcoidosis simply because of unique cultural, social, and linguistic characteristics. However, it is probable that the SHQ may be useful within a variety of sarcoidosis populations and clinical settings because of the diversity of the nearly 400 patients involved in our study. We must caution others that the SHQ will require further validity testing among patients living outside the United States, although investigators may find that adding or deleting specific items from the SHQ, using established methodologic techniques, may improve its usefulness within these populations. Finally, although we have intended to assess HRQL with the SHQ, readers may wonder how the SHQ incorporates related concepts of quality of life, functional status, and health status. Quality of life is a person's overall satisfaction with issues of importance to them whereas functional status is the subjective ability to perform in physical, social, role, and mental activities of daily living (17, 40). Health status represents not only the effect of health and disease on functional capacity, but the ability to take pleasure in these activities of daily life (40). Although we have formally conceptualized HRQL as the specific effect of health and disease on a person's quality of life, we recognize that this theoretical construct itself exists among overlapping areas of quality of life, health status, and functional status. This definition of HRQL is an economic description of a construct with a tremendous diversity of meanings in the current literature. However, because there is no universally accepted definition of HRQL, this particular conceptual framework may not represent the construct completely as appreciated by all readers. In conclusion, we have developed for persons with sarcoidosis a new measure of HRQL, the Sarcoidosis Health Questionnaire, that has demonstrated early evidence of both reliability and validity. It also appears to be the first validated outcome measure specific to the study of sarcoidosis. Further research is required to evaluate the validity of its evaluative properties, as well as to assess its performance among non-U.S. populations of patients with sarcoidosis.
The authors thank Amelia Anderson, Julie Montenegro, Jeanie Mascarella, Robert DeVellis, Tim Carey, Sue Tolleson-Rinehart, the NRSA research group, Cathey Kaelin, and the patients who participated for their assistance with this study.
Supported by a Public Health Outcomes grant from the University of North Carolina at Chapel Hill School of Public Health (C.E.C.). This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: C.E.C. has no declared conflict of interest. J.F.D. has no declared conflict of interest. C.D.B. has no declared conflict of interest. Y.P.K. has no declared conflict of interest. M.A.J. is on a Medical Advisory Board for Centocor. Received in original form November 17, 2002; accepted in final form May 7, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
