Tuebingen Cushing's Disease QOL Inventory: Part I
Tuebingen Cushing's Disease QOL Inventory: Part I
Inclusion criteria were age above 18 years, an unequivocally established diagnosis of pituitary-dependent CD proved by preoperative endocrinological tests and/or histopathological examination and immunostaining of tumour specimen.
Sixty-three CD patients were eligible for study participation. Thirteen (20·6%) patients were recruited after their admission for operative treatment at the neurosurgical department of the University Hospital in Tuebingen between January 2009 and October 2010. Fifteen newly diagnosed and not yet treated patients (23·8%) with pituitary-dependent CD were recruited with the help of the British Pituitary Foundation (http://www.pituitary.org.uk). The remaining 35 patients (55·6%) had been initially treated between December 2001 and December 2008. These 35 patients were invited for a follow-up visit at our department. The mean time since operative treatment in this group was 2·6 ± 2·0 years (range: 3 months to 9 years). During their visit, they were asked to imagine themselves in their individual pretreatment situation and retrospectively evaluate HRQoL. Additionally, all 35 patients were endocrinologically investigated. Relapse of CD was considered confirmed if serum cortisol levels were not suppressible by a low-dose (2 mg) dexamethasone suppression test in the presence of elevated urinary free cortisol. We found a relapse of their previous disease in ten of these 35 patients (28·6%). In the group that was investigated before operation, 7·7% presented with luteinising hormone/follicle stimulating hormone luteinising hormone/follicle stimulating hormone (LH/FSH) deficiency, 7·7% with growth hormone (GH) deficiency and 15·4% of the patients presented with a panhypopituitarism. In the group that was investigated retrospectively, we found LH/FSH deficiency in 16·1%, GH deficiency in 9·7% and Panhypopituitarism in 3·2%. Persistent hypocortisolism after transsphenoidal tumour removal was encountered in 20·0%.
The 63 patients were divided into six age groups that were similar in size (11 patients aged between 21 and 30 years, 10 between 31 and 40 years, 15 between 41 and 50 years, 11 between 51 and 60 years, 11 between 61 and 70 years and five older than 70 years).
Exclusion criteria comprised major comorbidities known to affect QoL, but not related to CD (e.g. chronic obstructive pulmonary disease, chronic renal failure, liver abnormalities, neurological disorders). Patients' ethnicity, sex, age, level of education, and preoperative serum cortisol and ACTH levels were assessed (Table 1).
To obtain normative data from a healthy population, a large sample of healthy controls was recruited via a German online internet version of the questionnaire (http://www.surveys-engine.com). They participated voluntarily, came from Germany and did not receive any payment. The questionnaire was linked to our department homepage. Authors' affiliations and an introductory explanation about the aim and purpose of the survey were given, and volunteers were invited to fill out the Tuebingen CD-25 and the WHOQoL-BREF. For the healthy subjects, the introductory sentence 'Because of my Cushing's disease…' of the Tuebingen CD-25 was changed into the introductory sentence 'In my daily life…'. The items and the wording of the questions were not changed to allow for a comparison between patients and healthy controls.
Information about age, gender and educational background was assessed online. A total of 1784 healthy controls (1210 women, 574 men) with a mean age of 28·9 ± 12·5 years (range 15–86 years) and a mean educational level of 14·1 ± 2·4 years (range 7–23 years) participated in the study. Volunteers younger than 15 years were excluded from the study. The healthy controls were assigned to seven age groups (i.e. 15–20, 21–30, 31–40, 41–50, 51–60, 61–70 and 70+ years) to obtain age-specific normative data. The majority (90·0%) of the 1210 female subjects respectively of the male subjects (82·4%) were younger than 40 years; however, sample sizes for older persons were also sufficient.
The research protocol was approved by the Ethics Committee of the University of Tuebingen (Tuebingen, Germany) and was performed under the tenets of the Declaration of Helsinki and its subsequent amendments. Written informed consent was given by all patients prior to participating in the study.
On the basis of the Food and Drug Administration (FDA) recommendations for the construction of patient-reported outcome measures (PRO) (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM193282.pdf), the concept and endpoint of the Tuebingen CD-25 are presented in Fig. 1. While the primary endpoint is the successful treatment of the disease, as assessed by the endocrinological investigation, the secondary endpoint is the regression of symptoms and signs of CD caused by chronic corticosteroid excess, as measured by the six scales of the questionnaire.
(Enlarge Image)
Figure 1.
Endpoint model for the treatment of Cushing's disease (Figure based on 'Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims', American Food and Drug Administration).
The acquisition strategy for the initial items of the Tuebingen CD-25 was threefold: First, technical literature was searched to identify relevant publications about the impact of CD on HRQoL. Second, we interviewed 10 CD patients who had already undergone operative treatment about a variety of HRQoL dimensions. They were asked to rate the personal importance of different HRQoL dimensions on separate sheets. Third, experienced endocrinologists, neurosurgeons and a neuropsychologist were asked to enumerate aspects and domains of CD they believed to cause problems in patients' lives. Based on all three approaches, the expert group extensively discussed the impairments that can be caused by hypercortisolism (e.g. osteoporosis with pathological bone fractures, depression, truncal obesity, gonadal dysfunction). The expert group came to a mutual agreement that there are seven different essential domains in HRQoL in CD patients (i.e. body concept, eating behaviour, bodily restrictions, depression, cognition, sexual activity and environment). This a priori determination of the HRQoL domains served primarily to facilitate item generation. Afterwards for each domain an approximately equal amount of items was generated by the authors during an informal meeting that contextually matched with the generic term (e.g. 'I am afraid to fall and suffer a bone fracture' was considered to represent bodily restrictions, respectively the item 'I suffer from concentration disorders' the domain cognition). Initially, 90 items had been assumed by the authors to be relevant in the assessment of HRQoL in CD patients. After a preliminary selection which served to avoid redundant items, 64 HRQoL items were included in the version that was handed out to the patients.
Data were analysed using spss (statistical package for social sciences) version 16·0 for Windows. Descriptive statistics were used to analyse sociodemographic and clinical characteristics (gender, age, level of education, hormone values) using mean and standard deviations (SD) for parametric data and the number of patients was used for categorical variables. Distribution of data was analysed by the Kolmogorov–Smirnov Test. Classical test theory (CTT) included reliability analysis with determination of the internal consistency (IC) (Cronbach's alpha), the discriminatory power, item difficulties and a confirmatory factorial analysis with the Varimax rotation method. Analysis of patterns of correlation (e.g. Item-Subscale, Subscale-Other Subscale, Subscale-Total Score) and validity testing were performed with the Pearson's correlation coefficient. Normally distributed data were compared using Student's t-test and non-normally distributed data with the Mann–Whitney U-Test. The significance level was set at P < 0·05.
To calculate the sample size, and assuming a moderate (i.e. 0·3–0·4) effect size with a level of significance of 0·05 and a statistical power of 0·80, a minimum sample of 59 patients was required. Item reduction included internal consistency testing (Cronbach's alpha), discriminatory power and item difficulty analyses, and confirmatory factorial analysis. Unacceptable discriminatory power was defined as ri < 0·3, and a preferably moderate item easiness index, i.e. between pi 0·4 and 0·6 (according to our answer categories between 1·6 and 2·4) was required as suggested by norm-referenced tests. A confirmatory factorial analysis (Varimax rotation methods with Kaiser normalization) was performed to verify the a priori assumption of the several domains. The number of components was determined by the number of Eigenvalues > 1. Items were assigned to mutually exclusive components based on the rotated factor loadings. Each item was assigned to the component in which it had its highest rotated factor loading, and these became the final subscales. Item consistency testing (Cronbach's alpha) of the defined subscales and analysis of patterns of correlation (e.g. Item-Subscale, Subscale-Other Subscale, Subscale-Total Score) was then performed. Internal consistency values of α > 0·70 were deemed appropriate. In cases where an item of the Tuebingen CD-25 was left blank, the missing entry was imputed with the average of all observed data across the sample.
For the development of a questionnaire, it is a usual approach to provide more than two answer categories to be able to grade the perceived QoL. In cases where answer categories are uneven (for example, three or five answer possibilities), it is recommended to rather provide five to seven answer possibilities. According to these recommendations, the inventory starts with 'Because of my Cushing's disease…' and response options can be given on a five-point Likert scale from 0 (strongly disagree), 1 (somehow agree), 2 (sometimes agree), 3 (mostly agree) and 4 (strongly agree) allowing for interval-scaled data.
Testing the validity of an instrument is often performed with the determination of the 'criterion validity' by correlating the newly developed instrument with an external criterion, i.e. an already established instrument that measures similar constructs. The criterion validity of the Tuebingen CD-25 was examined with the well-established World Health Organization Quality of Life (WHOQoL-BREF) questionnaire. This instrument comprises 26 items and measures the following broad domains: physical health, psychological health, social relationships and environment. The WHOQoL-BREF is a shorter version of the original instrument WHOQoL-100.
Methods
Patients
Inclusion criteria were age above 18 years, an unequivocally established diagnosis of pituitary-dependent CD proved by preoperative endocrinological tests and/or histopathological examination and immunostaining of tumour specimen.
Sixty-three CD patients were eligible for study participation. Thirteen (20·6%) patients were recruited after their admission for operative treatment at the neurosurgical department of the University Hospital in Tuebingen between January 2009 and October 2010. Fifteen newly diagnosed and not yet treated patients (23·8%) with pituitary-dependent CD were recruited with the help of the British Pituitary Foundation (http://www.pituitary.org.uk). The remaining 35 patients (55·6%) had been initially treated between December 2001 and December 2008. These 35 patients were invited for a follow-up visit at our department. The mean time since operative treatment in this group was 2·6 ± 2·0 years (range: 3 months to 9 years). During their visit, they were asked to imagine themselves in their individual pretreatment situation and retrospectively evaluate HRQoL. Additionally, all 35 patients were endocrinologically investigated. Relapse of CD was considered confirmed if serum cortisol levels were not suppressible by a low-dose (2 mg) dexamethasone suppression test in the presence of elevated urinary free cortisol. We found a relapse of their previous disease in ten of these 35 patients (28·6%). In the group that was investigated before operation, 7·7% presented with luteinising hormone/follicle stimulating hormone luteinising hormone/follicle stimulating hormone (LH/FSH) deficiency, 7·7% with growth hormone (GH) deficiency and 15·4% of the patients presented with a panhypopituitarism. In the group that was investigated retrospectively, we found LH/FSH deficiency in 16·1%, GH deficiency in 9·7% and Panhypopituitarism in 3·2%. Persistent hypocortisolism after transsphenoidal tumour removal was encountered in 20·0%.
The 63 patients were divided into six age groups that were similar in size (11 patients aged between 21 and 30 years, 10 between 31 and 40 years, 15 between 41 and 50 years, 11 between 51 and 60 years, 11 between 61 and 70 years and five older than 70 years).
Exclusion criteria comprised major comorbidities known to affect QoL, but not related to CD (e.g. chronic obstructive pulmonary disease, chronic renal failure, liver abnormalities, neurological disorders). Patients' ethnicity, sex, age, level of education, and preoperative serum cortisol and ACTH levels were assessed (Table 1).
To obtain normative data from a healthy population, a large sample of healthy controls was recruited via a German online internet version of the questionnaire (http://www.surveys-engine.com). They participated voluntarily, came from Germany and did not receive any payment. The questionnaire was linked to our department homepage. Authors' affiliations and an introductory explanation about the aim and purpose of the survey were given, and volunteers were invited to fill out the Tuebingen CD-25 and the WHOQoL-BREF. For the healthy subjects, the introductory sentence 'Because of my Cushing's disease…' of the Tuebingen CD-25 was changed into the introductory sentence 'In my daily life…'. The items and the wording of the questions were not changed to allow for a comparison between patients and healthy controls.
Information about age, gender and educational background was assessed online. A total of 1784 healthy controls (1210 women, 574 men) with a mean age of 28·9 ± 12·5 years (range 15–86 years) and a mean educational level of 14·1 ± 2·4 years (range 7–23 years) participated in the study. Volunteers younger than 15 years were excluded from the study. The healthy controls were assigned to seven age groups (i.e. 15–20, 21–30, 31–40, 41–50, 51–60, 61–70 and 70+ years) to obtain age-specific normative data. The majority (90·0%) of the 1210 female subjects respectively of the male subjects (82·4%) were younger than 40 years; however, sample sizes for older persons were also sufficient.
The research protocol was approved by the Ethics Committee of the University of Tuebingen (Tuebingen, Germany) and was performed under the tenets of the Declaration of Helsinki and its subsequent amendments. Written informed consent was given by all patients prior to participating in the study.
The Endpoint Model
On the basis of the Food and Drug Administration (FDA) recommendations for the construction of patient-reported outcome measures (PRO) (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM193282.pdf), the concept and endpoint of the Tuebingen CD-25 are presented in Fig. 1. While the primary endpoint is the successful treatment of the disease, as assessed by the endocrinological investigation, the secondary endpoint is the regression of symptoms and signs of CD caused by chronic corticosteroid excess, as measured by the six scales of the questionnaire.
(Enlarge Image)
Figure 1.
Endpoint model for the treatment of Cushing's disease (Figure based on 'Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims', American Food and Drug Administration).
Conceptual Design of the 64-item Preliminary Instrument
The acquisition strategy for the initial items of the Tuebingen CD-25 was threefold: First, technical literature was searched to identify relevant publications about the impact of CD on HRQoL. Second, we interviewed 10 CD patients who had already undergone operative treatment about a variety of HRQoL dimensions. They were asked to rate the personal importance of different HRQoL dimensions on separate sheets. Third, experienced endocrinologists, neurosurgeons and a neuropsychologist were asked to enumerate aspects and domains of CD they believed to cause problems in patients' lives. Based on all three approaches, the expert group extensively discussed the impairments that can be caused by hypercortisolism (e.g. osteoporosis with pathological bone fractures, depression, truncal obesity, gonadal dysfunction). The expert group came to a mutual agreement that there are seven different essential domains in HRQoL in CD patients (i.e. body concept, eating behaviour, bodily restrictions, depression, cognition, sexual activity and environment). This a priori determination of the HRQoL domains served primarily to facilitate item generation. Afterwards for each domain an approximately equal amount of items was generated by the authors during an informal meeting that contextually matched with the generic term (e.g. 'I am afraid to fall and suffer a bone fracture' was considered to represent bodily restrictions, respectively the item 'I suffer from concentration disorders' the domain cognition). Initially, 90 items had been assumed by the authors to be relevant in the assessment of HRQoL in CD patients. After a preliminary selection which served to avoid redundant items, 64 HRQoL items were included in the version that was handed out to the patients.
Statistical Analysis
Data were analysed using spss (statistical package for social sciences) version 16·0 for Windows. Descriptive statistics were used to analyse sociodemographic and clinical characteristics (gender, age, level of education, hormone values) using mean and standard deviations (SD) for parametric data and the number of patients was used for categorical variables. Distribution of data was analysed by the Kolmogorov–Smirnov Test. Classical test theory (CTT) included reliability analysis with determination of the internal consistency (IC) (Cronbach's alpha), the discriminatory power, item difficulties and a confirmatory factorial analysis with the Varimax rotation method. Analysis of patterns of correlation (e.g. Item-Subscale, Subscale-Other Subscale, Subscale-Total Score) and validity testing were performed with the Pearson's correlation coefficient. Normally distributed data were compared using Student's t-test and non-normally distributed data with the Mann–Whitney U-Test. The significance level was set at P < 0·05.
Feasibility Assessment and Item Reduction
To calculate the sample size, and assuming a moderate (i.e. 0·3–0·4) effect size with a level of significance of 0·05 and a statistical power of 0·80, a minimum sample of 59 patients was required. Item reduction included internal consistency testing (Cronbach's alpha), discriminatory power and item difficulty analyses, and confirmatory factorial analysis. Unacceptable discriminatory power was defined as ri < 0·3, and a preferably moderate item easiness index, i.e. between pi 0·4 and 0·6 (according to our answer categories between 1·6 and 2·4) was required as suggested by norm-referenced tests. A confirmatory factorial analysis (Varimax rotation methods with Kaiser normalization) was performed to verify the a priori assumption of the several domains. The number of components was determined by the number of Eigenvalues > 1. Items were assigned to mutually exclusive components based on the rotated factor loadings. Each item was assigned to the component in which it had its highest rotated factor loading, and these became the final subscales. Item consistency testing (Cronbach's alpha) of the defined subscales and analysis of patterns of correlation (e.g. Item-Subscale, Subscale-Other Subscale, Subscale-Total Score) was then performed. Internal consistency values of α > 0·70 were deemed appropriate. In cases where an item of the Tuebingen CD-25 was left blank, the missing entry was imputed with the average of all observed data across the sample.
Instrument Scoring
For the development of a questionnaire, it is a usual approach to provide more than two answer categories to be able to grade the perceived QoL. In cases where answer categories are uneven (for example, three or five answer possibilities), it is recommended to rather provide five to seven answer possibilities. According to these recommendations, the inventory starts with 'Because of my Cushing's disease…' and response options can be given on a five-point Likert scale from 0 (strongly disagree), 1 (somehow agree), 2 (sometimes agree), 3 (mostly agree) and 4 (strongly agree) allowing for interval-scaled data.
Validity Testing
Testing the validity of an instrument is often performed with the determination of the 'criterion validity' by correlating the newly developed instrument with an external criterion, i.e. an already established instrument that measures similar constructs. The criterion validity of the Tuebingen CD-25 was examined with the well-established World Health Organization Quality of Life (WHOQoL-BREF) questionnaire. This instrument comprises 26 items and measures the following broad domains: physical health, psychological health, social relationships and environment. The WHOQoL-BREF is a shorter version of the original instrument WHOQoL-100.