Development and Validation of the Insulin Treatment Satisfaction Questionnaire

Transcription

1 CLINICAL THERAPEUTICS /VOL. 26, NO. 4, 2004 Development and Validation of the Insulin Treatment Satisfaction Questionnaire Roger T. Anderson, PhD, 1 Soren E. Skovlund, MSc, BSc, 2 David Marrero, PhD, 3 Douglas W. Levine, PhD, 4 Keith Meadows, DrPhil, 5 Meryl Brod, PhD, 6 and Rajesh Balkrishnan, PhD 7 1 Wake Forest University School of Medicine, Winston-Salem, North Carolina, 2 Novo Nordisk, Bagsvaerd, Denmark, 3 Indiana University School of Medicine, Indianapolis, Indiana, 4 University of South Carolina, Columbus, South Carolina, 5 City University, London, United Kingdom, 6 The Brod Group, Mill Valley, California, and 7 University of Texas Health Sciences Center School of Public Health, Houston, Texas ABSTRACT Background: Treatment of diabetes mellitus (DM) is complex, requiring multifaceted lifestyle change or regulation and, for many, self-regulation of insulin levels in the blood. Historically, daily insulin treatment has been viewed as burdensome to patients, prompting newer formulations and improved delivery methods. Objective: This multicenter, clinical study was designed to develop a conceptually sound, clinically meaningful, and psychometrically valid measure of insulin treatment satisfaction, applicable to a wide range of insulin therapies. Methods: A 3-phase iterative process was employed to develop and validate the Insulin Treatment Satisfaction Questionnaire (ITSQ): (1) conceptual development of items, (2) preliminary validation among patients with DM, and (3) confirmatory validation among patients with DM. Results: The ITSQ was validated with 170 patients in phase 2 and 402 patients in phase 3. Confirmatory factor analysis produced a 5-factor, 22-item instrument assessing regimen inconvenience, lifestyle flexibility, glycemic control, hypoglycemic control, and satisfaction with the insulin delivery device. Results for reliability and construct validity of the final version were consistent in both samples of patients treated with insulin, with different data collection methods. Internal consistency (using Cronbach α coefficient) of the subscales ranged from 0.79 to Test retest reliability (using Spearman rank correlation coefficients) ranged from 0.63 to ITSQ scores showed moderate to high correlation with related measures of treatment burden. The ITSQ differentiated among insulin delivery methods, glycosylated hemoglobin values, the number of times the patient required assistance administering insulin, and insulin adherence. Conclusion: In our study samples, the ITSQ appeared to be conceptually and psychometrically sound and applicable to a wide range of insulin therapies. (Clin Ther. 2004;26: ) Copyright 2004 Excerpta Medica, Inc. Key words: treatment satisfaction, insulin, diabetes mellitus. Accepted for publication February 5, Printed in the USA. Reproduction in whole or part is not permitted /04/$19.00 Copyright 2004 Excerpta Medica, Inc. 565

2 CLINICAL THERAPEUTICS INTRODUCTION The incidence of diabetes mellitus (DM), a leading cause of disability and health care costs in most Western countries, is increasing at epidemic proportions. 1,2 Among the various pharmacologic treatments, insulin is among the most widely used, encompassing all cases of type 1 DM and between 20% and 30% of adult cases of type 2 DM. 3 Insulin treatment of DM is complex, involving the self-regulation of insulin levels in the blood. Successful glycemic control is the primary clinical outcome. Most insulin regimens require that the patient self-administer doses on a daily basis according to a prescribed treatment plan. This need for self-care, coupled with the lifestyle restrictions imposed by insulin regimens, may play an important role in the patient s success in following treatment recommendations and achieving good glycemic control. Patients who are satisfied with their treatment in terms of low burden and easy integration with daily activities will likely manage their home treatment better and consequently could be expected to maintain positive physical and psychological health, whereas poor self-management of DM can lead to serious, disabling complications. With the emergence of new short-, medium-, and long-acting insulin analogues delivered by both pen and pump devices, 4 there is a complementary need to quantify the impact of these treatment options in terms of individual patients treatment burden. Currently, few well-tested surveys are available for describing and comparing patient experiences across a wide range of insulin treatment regimens. Measures are needed that have sufficient breadth of scope to characterize key patient experiences and features of insulin therapies that substantially affect daily life with DM. This includes identifying specific problem areas requiring attention from patient DM counselors and educators, to maximize the fit between insulin therapy and patient preference and lifestyle, and ultimately to enhance glycemic control through optimal patient adherence. The latter objectives bring treatment satisfaction to the fore. Conceptually, treatment satisfaction is a recipient s rating of the salient aspects of the process and results of the treatment. 5 To formulate an assessment, each individual weighs the relevant aspects of treatment and determines his or her overall degree of satisfaction through a cognitive process that reflects personal preferences for treatment, experiences during treatment, and the importance of each outcome relative to desired goals. In the context of DM, it is possible for a patient to be dissatisfied despite the clinical effectiveness of a therapy. A treatment plan of decreasing glycosylated hemoglobin (HbA 1c ) to stringent parameters (eg, <6%) with an intensive insulin treatment therapy may have intended negative consequences on a patient s daily life or well-being (eg, hypoglycemia) that may be perceived by the patient as outweighing the goal of optimal control and long-term health protection. Tools to measure patient satisfaction with insulin regimens are available in the literature. 6,7 However, most of these instruments have been developed for a specific delivery method (eg, pen systems) or formulation, potentially limiting their usefulness across various insulin treatments. Instruments that assess generic characteristics of DM treatments may be inconclusive or insensitive to unique differences in burden concerning insulin therapies. The purpose of this research program was to develop a conceptually sound, clinically meaningful, and psychometrically valid treatment satisfaction measure applicable for modern insulin therapies. METHODS Development of the Insulin Treatment Satisfaction Questionnaire (ITSQ) was conducted in 3 distinct phases: (1) conceptual development of the items, (2) preliminary psychometric validation, and (3) confirmatory psychometric validation. Phase I: Conceptual Development The conceptual basis for the ITSQ and its items were developed through a literature review, consultation with clinicians, and interviews with patients with DM. Focus groups were conducted to generate and revise an initial item pool. Four groups were formed of patients with either type 1 or type 2 DM from lowincome community health clinics and from a DM care clinic in an academic medical center. Englishlanguage MEDLINE listings (last 10 years) were searched using the following key words: diabetes care, insulin, burden, and satisfaction. Published reports and relevant questionnaires in the literature were examined to yield lists of general content clusters. The latter lists were given to DM health care providers for review to add or further specify the content. An initial 566

3 R.T. Anderson et al. patient focus group was held (comprising patients using syringe, pen, and pump methods of insulin delivery) to discuss general issues of treatment burden and to refine the list. A list of draft items was produced (based on this feedback) and discussed, and 3 iterative patient focus groups were held to develop the items in terms of concept clarity, conciseness, and relevance. All focus groups were audiotaped, transcribed, and content coded by trained experts. Phase II: Preliminary Validation Following item development, a prototype ITSQ instrument was tested in a sample of patients treated with insulin. A validation battery comprised questions on patients demographics, treatment history, and the following related instruments: the 12-Item Short-Form (SF-12, version 2) Health Survey, 8 which assesses general health and well-being; and the Insulin Management Self-Efficacy Scale, 9 which measures patients confidence in their ability to care for themselves with insulin. Other items assessed history of symptoms known to occur with insulin treatment and glycemic control adapted from Whitty el al. 10 The Problem Areas in Diabetes (PAID) Scale 11 measured the extent of problems perceived by the patient regarding different aspects of having DM. The World Health Organization Diabetes Treatment Satisfaction Questionnaire (WHO-DTSQ) 6 measured overall DM treatment satisfaction. * Items assessing global satisfaction with insulin developed for this study and subjected to cognitive testing during patient interviews for clarity, conciseness of meaning, and ease of response. The items assessed patients experiences during the past 4 weeks: In general, how satisfied are you with your insulin regimen? ; In general, how satisfied have you been with the type of insulin you use? ; In general, how satisfied have you been with the device or method you use to give yourself insulin? ; and In general, how satisfied have you been with your overall diabetes treatment (think about all aspects of your treatment including diet, exercise, tablets, insulin, etc)? The response scale was ordinal, ranging from 1 (extremely satisfied) to 6 (extremely dissatisfied). * Permission to use the WHO-DTSQ was granted by the World Health Organization regional office for Europe. Construct validity of the ITSQ total score, the primary score of this tool, would be affirmed if the score demonstrated moderate correlations (eg, r > 0.30) with measures of related concepts such as general DM treatment satisfaction (eg, WHO- DTSQ), problems with DM management (eg, PAID), and somatic symptoms relevant to DM. Thus, we hypothesized that respondents who reported regimen problems or symptoms or who give low ratings for general aspects of their insulin treatment would also report lower satisfaction on the ITSQ. Furthermore, we expected that global measures of health status (eg, SF-12) would demonstrate consistency between poorer health and lower satisfaction, reflecting inadequate DM control. As a measure of treatment burden, the ITSQ was expected to show relatively low correlations with physical well-being, a distinctly different concept than treatment satisfaction. Construct validity of the ITSQ subscale domains, representing core content areas of the survey, would be affirmed if each ITSQ subscale correlated more highly with pertinent measures than with the other less relevant subscales. Thus, scales assessing inconvenience and flexibility would correlate highest with measures of regimen problems (eg, PAID); a subscale assessing satisfaction with glycemic control would correlate highest with global ratings of satisfaction with insulin and treatment; and the subscale assessing satisfaction with devices would correlate highest with a global item on satisfaction with method of insulin delivery. Finally, we expected that because treatment satisfaction is intended as a core indicator of treatment quality, it would correspond to clinical outcomes of HbA 1c value and medication adherence. Participants were recruited from 3 medical centers in the United States by using patient lists of DM care centers at each location. The settings were primary care practices that offer comprehensive DM care to both publicly and privately insured patients. Eligible patients were aged 18 years, had no apparent cognitive impairment, had a diagnosis of DM, were treated with insulin therapy, and were able to read English. To ensure sample heterogeneity for patients treatment exposures and experiences, we drew a convenience sample of patients, overselecting type 1 DM and alternative insulin delivery methods such as pump and pen (versus syringe). From this group, 567

4 CLINICAL THERAPEUTICS participants were randomly selected for a substudy to complete the questionnaire again 1 week later in the physician s office to assess test retest reliability. Clinical information recorded from the patients charts included type of DM, HbA 1c value, whether the patient was prescribed insulin or insulin/oral hypoglycemic agents, and the number of bolus and basal injections per day. We also assessed self-reported adherence with insulin doses, operationalized as selfreported frequency (often to not at all) in the last 4 weeks of skipping an insulin dose because you forgot to take it. Exploratory factor analysis with orthogonal transformation was used to identify the ITSQ item structure for subsequent testing. Construct validity of the initial ITSQ factors was checked from correlations with related measures in the battery. An intermediate version of the ITSQ was produced for refinement. Phase III: Confirmatory Validation In the final step, the item structure of the ITSQ was retested in an independent sample with confirmatory factor analysis and refined to maximize model fit to produce a final ITSQ version. The confirmatory validation battery consisted of the revised ITSQ, questions on patients demographics and insulin treatment, and self-reported clinical information including most recent HbA 1c value, duration and type of DM, insulin product and regimen, and time taking insulin. The battery was administered via the Internet to patients treated for DM in a national Internet consumer panel. Patients with DM were contacted by , and only those matching the inclusion criteria were asked to participate. To ensure that the survey respondents had the required characteristics, we verified the original panel registration data with personal data on the questionnaire. Confirmatory factor analysis was performed with structural equation modeling to evaluate the model developed in the preliminary validation phase. Model fit was assessed using the chi-square test. Because it is well known that the chi-square test of significance is sensitive to sample size, 12 model fit was also evaluated in light of 3 other common fit indices: the normed chi-square (χ 2 /df ), the comparative fit index (CFI), 13 and the root-mean-square error of approximation (RMSEA). 14 Values of the normed chi-square close to 2 represent good fit. 12,15,16 Internal consistency reliability of the ITSQ total score and subscales was assessed using the Cronbach α coefficient. 17 Test retest reliability was examined using Spearman rank correlation coefficients. Known-groups validity was examined by comparing adjusted means of ITSQ factors for insulin delivery assistance (none vs some help), HbA 1c value ( 7.5% vs >7.5%), and insulin delivery method (syringe vs pen). The clinical meaningfulness of the ITSQ was assessed during the confirmatory study by correlating ITSQ scores with patients blood glucose levels and by assessing the ITSQ to discriminate patients by their HbA 1c values. Spearman rank correlation coefficients were used to measure the association of ITSQ total scores and subscale scores with related measures (PAID, WHO-DTSQ, Insulin Management of Diabetes, Self-Efficacy Scale, and DM symptoms) and with global ratings of satisfaction. Known-groups validity was assessed by tests of least square means (adjusted for age and sex) for ITSQ subscale and total scores between HbA 1c values ( 8% vs >8%), method of insulin delivery (syringe and vial vs pen, pump), and self-reported frequency of need for assistance taking insulin (never vs any need). ITSQ subscale scores were calculated by imputing the missing values based on the mean of the nonmissing items. All subscales were transformed to a scale of 0 to 100, with higher scores indicating better treatment satisfaction. RESULTS Phase I: Conceptual Development A total of twenty-five patients with DM participated in 4 focus groups. Participants were diverse in terms of age, sex, ethnicity, length of time and method of taking insulin, and type of DM. Findings from patient interviews, clinical expert review, and the literature were synthesized to generate a preliminary set of 32 items covering 6 general aspects of insulin treatment (ease of regimen, glycemic control, treatment burden or distress, interference in meal planning, lifestyle, and symptoms) and method of insulin delivery. Patients response options were based on an ordered scale. For example, for the question, How convenient is it for you to take all your daily insulin doses as prescribed? patients indicated their responses on a 7-point scale from extremely conve- 568

5 R.T. Anderson et al. nient (score of 1) to not at all convenient (score of 7). We did not test importance or preference weighting in the development and scoring of the ITSQ instrument because recent research suggests that assigning preference weights does not improve sensitivity to change. 18,19 Phase II: Preliminary Validation The preliminary validation sample included 170 patients (Table I). Respondents had a mean (SD) age of 48.7 (13.2) years, and 107 (62.9%) were women. Many had completed high school (83 [48.8%]); 44 (25.9%) had completed college and 36 (21.2%) had not completed high school. Of these 170 patients, 63 (37.1%) had type 1 DM, and their primary method of taking insulin was syringe and vial (120 [70.6%]), followed by pen injection (22 [12.9%]), pump (22 [12.9%]), and inhaler 4 (2.4%). In 55 respondents (32.4%), HbA 1c values were categorized as low (<7.5%), 36 (21.2%) were moderate (7.5% 8.5%), and 55 (32.4%) were high (>8.5%); 17 respondents (10.0%) took insulin QD, 73 (42.9%) BID, 32 (18.8%) TID, 29 (17.1%) QID, and 15 (8.8%) >4 times daily. Most respondents (95 [55.9%]) had had DM for >10 years, 31 (18.2%) for 6 to 10 years, and 19 (11.2%) for 3 to 5 years. Few respondents (24 [14.1%]) had had DM for 2 years. Likewise, the largest group of respondents (69 [40.6%]) had been taking insulin for >10 years, followed by 16.0% for 6 to 10 years, 31 (18.2%) for 3 to 5 years, and 41 (24.1%) for 2 years. Exploratory factor analysis of questionnaire responses in this sample identified a parsimonious 5-factor structure including ease and convenience of regimen, burden of regimen, lifestyle flexibility, glycemic control, and satisfaction with the insulin delivery device, with a total of 16 items. This structure was highly consistent with content areas identified in the focus groups. Internal consistency (Cronbach α coefficient) of the subscales ranged from 0.78 (lifestyle) to 0.92 (glycemic control), and for all ITSQ questions, it was Test retest reliability (Spearman rank correlation coefficient) in a randomly selected sample was moderate to high for all subscales, ranging from 0.63 to 0.94 (Table II). Based on the psychometric findings, 5 items were added to create a separate domain of hypoglycemia as distinct from overall glycemic control. An additional item was added to form an inconvenience subscale, collapsed from the burden and ease subscales. These procedures resulted in a 22-item ITSQ, which was submitted for confirmatory validation. Phase III: Confirmatory Validation Approximately 60% of patients contacted by agreed to participate in the study. The confirmatory validation sample included 402 respondents (Table I). They had a mean (SD) age of 55.6 (9.4) years, and 174 (43.3%) were women. Nearly equal proportions of the patients had completed high school (193 [48.0%]) and had completed college (199 [49.5%]). All had type 2 DM, and 328 (81.6%) used insulin delivered by syringe and vial, followed by pen injection (55 [13.7%]), pump (5 [1.2%]), and inhaler (1 [0.2%]). Among the respondents, 127 (31.6%) had HbA 1c characterized as low (<7.5%), 125 (31.1%) had moderate values (7.5% 8.5%), and 72 (17.9%) had high values (>8.5%). Insulin administration QD was most common (207 [51.5%]), followed by BID (45 [11.2%]) and 3 times per day (150 [37.3%]). Most patients (233 [48.0%]) had had DM for >10 years, 114 (38.4%) for 6 to 10 years, 34 (8.5%) for 3 to 5 years, and 21 (5.2%) for 2 years. A total of 87 patients (21.6%) had been taking insulin for >10 years, 129 (32.1%) for 6 to 10 years, 54 (13.4%) for 3 to 5 years, and 132 (32.9%) for 2 years. Confirmatory factor analysis of the 22-item ITSQ reproduced the original item structure and the proposed factor of hypoglycemia items (Appendix). Three of the 4 fit indices indicated that the model fit the data 20,21 : normed χ 2 = 2.35, CFI = 0.96, and RMSEA = The χ was (P < 0.001), reflecting the sensitivity of the latter test to sample size. Thus, the final ITSQ contained 22 items among 5 content clusters, forming a total ITSQ score transformed as 0 to 100 where 100 indicates complete satisfaction with insulin treatment. Table II shows the ITSQ subscale and total score means, percentage at ceiling (most satisfied), and internal consistency reliability (Cronbach α coefficient) of the final item set. Consistent with the results for the exploratory sample, the confirmatory sample showed high reliability, ranging from 0.79 (lifestyle flexibility) to 0.91 (glycemic control), and 0.93 for all ITSQ questions. Ceiling effects were relatively low, ranging from 16.2% (inconvenience of regimen) to 569

6 CLINICAL THERAPEUTICS Table I. Characteristics of preliminary and confirmatory validation samples (n = 170 and n = 402, respectively). * Characteristic Preliminary Sample Confirmatory Sample Age, mean (SD), y 48.7 (13.2) 55.6 (9.4) Sex, no. (%) Women 107 (62.9) 174 (43.3) Men 61 (35.9) 228 (56.7) Highest level of education, no. (%) Did not complete high school 37 (21.2) 10 (2.5) Completed high school or equivalent 83 (48.8) 193 (48.0) Completed college 44 (25.9) 199 (49.5) DM type, no. (%) Type 1 63 (37.1) 0 (0.0) Type 2 92 (54.1) 402 (100.0) HbA 1c value, no. (%) Low (<7.5%) 55 (32.4) 127 (31.6) Moderate (7.5% 8.5%) 36 (21.2) 125 (31.1) High (>8.5%) 55 (32.4) 72 (17.9) No. of times insulin taken per day, no. (%) 1 17 (10.0) 207 (51.5) 2 73 (42.9) 45 (11.2) 3 32 (18.8) 141 (35.1) 4 29 (17.1) 7 (1.7) >4 15 (8.8) 2 (0.5) Time with DM, no. (%) <6 mo 7 (4.1) 2 (0.5) 6 mo 1 y 2 (1.2) 2 (0.5) 1 2 y 15 (8.8) 17 (4.2) 3 5 y 19 (11.2) 34 (8.5) 6 10 y 31 (18.2) 114 (28.4) >10 y 95 (55.9) 233 (58.0) Time taking insulin, no. (%) <6 mo 10 (5.9) 34 (8.5) 6 mo 1 y 13 (7.6) 25 (6.2) 1 2 y 18 (10.6) 73 (18.2) 3 5 y 32 (18.8) 54 (13.4) 6 10 y 27 (15.9) 129 (32.1) >10 y 69 (40.6) 87 (21.6) Primary method of taking insulin, no. (%) Syringe and vial 120 (70.6) 328 (81.6) Pen injection 22 (12.9) 55 (13.7) Pump 22 (12.9) 5 (1.2) Inhaler 4 (2.4) 1 (0.2) Other 0 (0.0) 13 (3.2) DM = diabetes mellitus; HbA 1c = glycosylated hemoglobin. * Percentages based on n = 170 and n = 402. Number of respondents with missing values not shown. 570

7 R.T. Anderson et al. Table II. Comparison of results from the preliminary and confirmatory validation samples (n = 170 and n = 402, respectively). Preliminary Sample Confirmatory Sample No. of Ceiling, Cronbach α Test Retest No. of Ceiling, Cronbach α ITSQ Subscale Items Mean * (SD) % Reliability Reliability Items Mean * (SD) % Reliability Inconvenience of regimen (22.87) (20.33) Lifestyle flexibility (25.30) (22.54) Glycemic control (28.54) (24.34) Hypoglycemic control (20.06) Insulin delivery device satisfaction (22.39) (20.17) ITSQ total (20.85) (16.53) ITSQ = Insulin Treatment Satisfaction Questionnaire. * Scores range from 0 to 100 (lower score indicates less treatment satisfaction). Test retest period was ~2 weeks (n = 27). 7.5% (hypoglycemic control), and 1.2% for the total score. Construct validity of the ITSQ subscales and total score, summarized in Table III, showed moderate correlation with the PAID (r = 0.43 to 0.69), WHO-DTSQ (r = 0.42 to 0.74), DM symptoms (r = 0.28 to 0.54), and Insulin Self-Efficacy Scale (r = 0.24 to 0.67). Relatively low correlations were found between ITSQ subscales and SF-12 Health Survey physical component score as the ITSQ is not a measure of physical well-being; however, the moderate correlation with mental component score is evidence that the ITSQ overlaps with emotional distress. Table IV presents the known-groups validity of the ITSQ subscales and total score. Adjusted mean ITSQ total scores differentiated among insulin delivery groups: insulin delivery method (70.4 vs 76.7, P < 0.01), HbA 1c values (75.5 vs 69.6, P = 0.01), and less burden in terms of needing help with insulin regimen (73.1 vs 64.9, P < 0.01). The ITSQ subscales displayed targeted sensitivity to their intended concepts, with ITSQ lifestyle and flexibility discerning insulin delivery method (P = 0.01), ITSQ glycemic control discerning HbA 1c value (P = 0.01), and ITSQ hypoglycemic control and inconvenience discerning the need for help taking insulin (P = 0.03). Finally, as shown in Table V, patients reporting high compliance (ie, no skipped doses of insulin in the last 4 weeks) had lower HbA 1c values than those with partial compliance (ie, some skipped doses), as well as fewer symptoms and higher ITSQ satisfaction scores (P = 0.05, P = 0.001, and P = 0.003, respectively). Predictors of insulin adherence (data not shown) were number of prescribed doses and older age, consistent with the literature on medication adherence in general. 22 DISCUSSION Insulin treatment of DM is a complex regimen that requires self-regulating insulin levels to match physiologic needs. Although successful glycemic control is the primary outcome, adoption and maintenance of the requisite behaviors depend on the patient s expectations, self-efficacy, and day-to-day experiences with treatment. 23 Health care providers recognize that psychosocial issues play an important role in how effectively patients manage their DM. Insulin regimens that best fit patients expectations and lifestyle demands may promote adherence and effective HbA 1c control. Comparison of insulin therapies with different insulin profiles, different formulations, or different modes of delivery requires a comprehensive instrument that is clinically meaningful and psychometrically sound. The ITSQ proved to be conceptually sound and clinically relevant to important goals of glycemic control and adherence. In addition, its practicality was evident from the low incidence (<5%) of missing values in both preliminary and confirmatory samples. From the qualitative work based on patient interviews through the results of the quantitative analysis 571

8 CLINICAL THERAPEUTICS Table III. Construct validity of Insulin Treatment Satisfaction Questionnaire (ITSQ) subscales in testing sample (n = 170). * Global Satisfaction Insulin SF-12 Health Satisfied Satisfied Satisfied Satisfied ITSQ WHO- DM Self-Efficacy Survey, 8 with DM with Insulin with Type with Device/ Subscale PAID 11 DTSQ 6 Symptoms Scale 9 MCS/PCS Treatment Regimen of Insulin Method Inconvenience of regimen /0.03 (NS) Lifestyle flexibility /0.12 (NS) Glycemic control /0.20 (NS) Insulin delivery device satisfaction /0.16 (NS) ITSQ total /0.06 (NS) PAID = Problem Areas in Diabetes; WHO-DTSQ = World Health Organization Diabetes Treatment Satisfaction Questionnaire; DM = diabetes mellitus; SF-12 = 12-Item Short-Form Health Survey (version 2); MCS = mental component score domain; PCS = physical component score domain. * Spearman rank correlation coefficients are shown. All coefficients in table have statistical significance of P < 0.01 unless noted as NS (not significant; P > 0.05). Scores range from 0 to 100 (lower score indicates less treatment satisfaction). Higher score indicates higher burden. Higher score indicates less satisfaction. Higher score indicates higher symptom burden. Higher score indicates better health status. Satisfaction during the past 4 weeks (higher score indicates more satisfied). and final item reduction, we found 5 core dimensions of patients satisfaction with insulin treatment: inconvenience, lifestyle interference, glycemic control, hypoglycemia, and insulin delivery device. The treatment inconvenience factor is a composite of treatment ease, burden, and quality, graded along a continuum. We followed a rigorous process for the development of the ITSQ, involving content review by experts in the United States, Europe, and Japan, multiple focusgroup discussions, and use of 2 samples to test and confirm the results. The iterative approach of preliminary validation and confirmatory validation allowed improvement of the ITSQ and yielded a final 22-item version with a reliable internal structure and good construct validity of its subscales and total score. In addition, the ITSQ Table IV. Known-groups validity of the Insulin Treatment Satisfaction Questionnaire (ITSQ) subscales, grouped by clinical status according to least squares means. Need for Help with Insulin Delivery Method * HbA * 1c Taking Insulin Syringe and Vial Pen/Pump Low High No Yes ITSQ Subscale (n = 328) (n = 74) P (n = 127) (n = 197) P (n = 358) (n = 44) P Inconvenience of regimen Lifestyle flexibility Hypoglycemic control Glycemic control Insulin delivery device satisfaction ITSQ total HbA 1c = glycosylated hemoglobin. * Analysis of variance adjusted for age and sex. Analysis of variance adjusted for age, sex, and duration taking insulin. 572

9 R.T. Anderson et al. Table V. Mean patient outcomes (n = 130) by insulin treatment compliance group. Outcome High Compliance * Partial Compliance P HbA 1c value Symptom score ITSQ total score HbA 1c = glycosylated hemoglobin; ITSQ = Insulin Treatment Satisfaction Questionnaire. * No skipped doses. Some skipped doses. Student t test of group means. was reproducible (test retest) and had convergent validity with instruments that overlap the content areas assessed. Based on our investigation, the ITSQ appears to be clinically meaningful because it is able to discriminate between methods of taking insulin, difficulty with the regimen, recent HbA 1c values, and patients adherence to insulin therapy. Notably, patients with poor adherence to insulin had significantly lower treatment satisfaction and higher HbA 1c values, demonstrating that difficult or demanding treatment regimens may appreciably lower treatment effectiveness, which may be possible to achieve with less burdensome treatment. This study documented the validity and reliability of the ITSQ; it did not investigate relative performance of the ITSQ compared with other candidate measures in discriminating among various levels of insulin treatment satisfaction or other clinical outcomes. Future studies are needed to contrast and compare the benefits of treatment-specific versus generic instruments. Since the completion of validity testing, the ITSQ has been translated successfully into 8 other languages (Danish, Dutch, Finnish, French, German, Norwegian, Polish, and Swedish). Assessment of the burden of blood sugar monitoring and weight gain with therapy were outside the scope of this development, but may need to be assessed separately, depending on study focus. The ITSQ was developed specifically for use with insulin treatment and does not address satisfaction with noninsulin (tablet) medications, for which other instruments are needed. Another limitation of the ITSQ is that although its development included considerable input from researchers in Europe, Japan, and the United States, the primary validation involved samples obtained in the United States. Therefore, the validity of the ITSQ in international settings is unknown and must be tested. Although the ITSQ validation results were obtained from samples diverse in terms of DM type, age, and race/ ethnicity, the extent to which the ITSQ subscale scores perform equivalently among subgroups of patients is not certain. We did not find statistically significant interactions with patient characteristics; however, additional research is needed to address the question of structural invariance by group. Our study was cross sectional and was not capable of examining casual influences of low treatment satisfaction on clinically relevant outcomes. For example, patients taking more complicated regimens may have experienced more difficulty controlling their DM and therefore may have been less satisfied with their regimen. In the absence of a randomized controlled trial to test this, the data presented in this analysis could be limited by confounding. The use of the Internet to find patients could also have contributed to selection bias because Internet users may be better educated, wealthier, and potentially have better glycemic control. The global measures of satisfaction with diabetes treatment, insulin device, and adherence used in this study to validate the ITSQ produced results that are highly consistent with their underlying concepts. However, these measures were developed and pretested for this study and have not been formally validated. Therefore, the measurement error of the latter items is uncertain. A final limitation is that despite the strong construct validity and sensitivity to clinical indices, responsiveness to change (eg, changes in HbA 1c values) of the ITSQ was not examined in this cross-sectional study. Prospective studies are planned to address this issue. 573

10 CLINICAL THERAPEUTICS CONCLUSION This study has described the development and validation of a 22-item instrument designed to assess treatment satisfaction with a wide range of insulin therapies. Results from our 2 samples indicate the ITSQ is a conceptually and psychometrically sound measure of treatment satisfaction for conventional and modern insulin regimens. ACKNOWLEDGMENTS This study was supported by Novo Nordisk. The authors acknowledge the valuable contributions of Dr. Annabel Bowden, Quintiles, for assistance with manuscript preparation; Dr. Hitoshi Ishii, Tenri Hospital, for advice, adaptation, and testing of the ITSQ in Japanese; Dr. Sherwyn L. Schwartz for helping with data collection; and Kathleen Dziak, BA, Fabian Camacho, MA, and Vanessa Duren-Winfield at Wake Forest University School of Medicine for help with data collection. REFERENCES 1. Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic. Nature. 2001; 414: King H, Aubert RE, Herman WH. Global burden of diabetes, : Prevalence, numerical estimates, and projections. Diabetes Care. 1998;21: Beckles GL, Engelgau MM, Narayan KM, Herman WH, et al. Population-based assessment of the level of care among adults with diabetes in the U.S. Diabetes Care. 1998;21: Owens DR. New horizons alternative routes for insulin therapy. Nat Rev Drug Discov. 2002;1: Weaver M, Patrick DL, Markson LE, et al. Issues in the measurement of satisfaction with treatment. Am J Manag Care. 1997;3: Bech P, Moses R, Gomis R. The effect of prandial glucose regulation with repaglinide on treatment satisfaction, wellbeing and health status in patients with pharmacotherapy naive type 2 diabetes: A placebocontrolled, multicentre study. Qual Life Res. 2003;12: Cappelleri JC, Gerber RA, Kourides IA, Gelfand RA. Development and factor analysis of a questionnaire to measure patient satisfaction with injected and inhaled insulin for type I diabetes. Diabetes Care. 2000;23: Ware J Jr, Kosinski M, Keller SD. A 12-item Short- Form Health Survey: Construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34: Hurley A. Measuring self-care ability in patients with diabetes. The insulin management diabetes self-efficacy scale. In: Waltz C, ed. Measurement of Nursing Outcomes: Measuring Client Self-Care and Coping Skills. 4. New York: Springer; 1990: Whitty P, Steen N, Eccles M, et al. A new self-completion outcome measure for diabetes: Is it responsive to change? Qual Life Res. 1977;6: Welch GW, Jacobson AM, Polonsky WH. The Problem Areas in Diabetes Scale. An evaluation of its clinical utility. Diabetes Care. 1997;20: Bollen KA. Structural Equations with Latent Variables. New York: Wiley; Bentler PM. Comparative fix indexes in structural models. Psychol Bull. 1990;107: Browne MW, Cudeck R. Alternate ways of assessing model fit. In: Bollen KA, Long JS, eds. Testing Structural Equation Models. Newbury Park, Calif: Sage; 1993: Byrne BM. A Primer of LISREL: Basic Applications and Programming for Confirmatory Factor Analytic Models. New York: Springer-Verlag; Marsh HW, Hocevar D. Application of confirmatory factor analysis to the study of self-concept: First- and higher-order factor models and their invariance across groups. Psychol Bull. 1985;97: Cronbach L. Coefficient alpha and the internal structure of tests. Psychometrika. 1951;16: Jansen SJ, Stiggelbout AM, Nooij MA, Kievit J. The effect of individually assessed preference weights on the relationship between holistic utilities and nonpreference-based assessment. Qual Life Res. 2000; 9: Trauer T, Mackinnon A. Why are we weighting? The role of importance ratings in quality of life measurement. Qual Life Res. 2001;10: Hu L-T, Bentler PM. Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychol Methods. 1998;3: Hu L-T, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999;6:

11 R.T. Anderson et al. 22. Anderson RT, Ory M, Cohen S, McBride JS. Issues of aging and adherence to health interventions. Control Clin Trials. 2000;21(Suppl):171S 183S. 23. Casparie AF, van der Waal MA. Differences in preferences between diabetic patients and diabetologists regarding quality of care: A matter of continuity and efficiency of care? Diabet Med. 1992;12: Appendix. The 22-item Insulin Treatment Satisfaction Questionnaire (ITSQ). People who take insulin can have many different experiences with their treatment. Some people who take insulin may find it difficult and burdensome, whereas others feel that it is not much of a bother at all. The following questions are about your perceptions of your current insulin treatment and how it affects you in your daily life. When you think of your insulin treatment, please keep in mind the type of insulin you take, the dose or amount of insulin, your schedule for taking insulin, and the device or method you use to give yourself insulin. Please think about your experiences during the past 4 weeks when you answer the questions. Please answer each question by circling the number 1 that best represents your answer. If you are unsure about how to answer a question, please give the best answer you can. 1. How much of a bother is it for you to take all your daily insulin doses as prescribed? No bother at all A tremendous bother 2. How much does your current insulin treatment interfere with your ability to enjoy social or leisure activities? Does not interfere at all Interferes tremendously 3. How much does your current insulin treatment interfere with your work or school activities? (If you do not work or attend school, think about your regular daily activities.) Does not interfere at all Interferes tremendously 4. How much do you have to plan the timing of your meals or snacks around the insulin you currently use? No planning at all A tremendous amount of planning (continued) 575

12 CLINICAL THERAPEUTICS Appendix. (Continued) 5. How much do you have to plan what you eat with your current insulin treatment? No planning at all A tremendous amount of planning 6. How much do you have to plan your physical activities (such as exercise or strenuous household chores) around your current insulin treatment? No planning at all A tremendous amount of planning 7. How confident are you that you can avoid symptoms of low blood sugar (such as sweating, trembling, dizziness, blurred vision) with your current insulin treatment? confident confident 8. How confident are you that you can avoid severe episodes of low blood sugar that result in loss of consciousness (fainting or passing out) with the insulin you currently use? confident confident 9. In general, how bothered are you by symptoms of low blood sugar (such as sweating, trembling, dizziness, blurred vision) with the insulin you currently use? bothered bothered 10. How much do you feel that the insulin you are currently using increases the chances that you will experience low blood sugar? 11. How worried are you about experiencing low blood sugar during the night with the insulin you currently use? worried worried (continued) 576

13 R.T. Anderson et al. Appendix. (Continued) 12. How confident are you that you can avoid symptoms of high blood sugar (such as dry mouth, thirst, frequent urination, fatigue, increased appetite) with your current insulin treatment? confident confident 13. How satisfied are you with the stability of your blood sugar levels with your current insulin treatment? satisfied satisfied 14. Overall, how pleased are you with the blood sugar control you achieve with your current insulin treatment? pleased pleased 15. In general, how stressful is it for you to manage your current insulin treatment? stressful stressful 16. How burdensome is it for you to manage your current insulin treatment? burdensome burdensome The following questions are about your perceptions of your current method of taking insulin and how it affects your daily life. For these questions, you should only think about the device or method you use to give yourself insulin. 17. How easy is it for you to take the correct amount of insulin each time with your current method of taking insulin? easy easy 18. How convenient is your current method of taking insulin when you are away from home? convenient convenient (continued) 577

14 CLINICAL THERAPEUTICS Appendix. (Continued) 19. How much pain or other physical discomfort do you experience with your current method of taking insulin? No pain or discomfort A tremendous amount of pain or discomfort 20. How comfortable are you taking insulin in a public place (where people might see you) with your current method of taking insulin? comfortable comfortable 21. How much emotional distress or anxiety do you experience from your method of taking insulin? No distress or anxiety A tremendous amount of distress or anxiety 22. Overall, how satisfied are you with your current method of taking insulin? satisfied satisfied Address correspondence to: Roger T. Anderson, PhD, Department of Public Health Sciences, Wake Forest University School of Medicine, Medical Center Boulevard, Winston-Salem, NC rtanders@wfubmc.edu 578