This study was a psychometric research conducted on 73 patients with a confirmed diagnosis of acromegaly attending a tertiary pituitary clinic for their routine follow-up from 2019 to 2021. The patients were included in the study if they were Iranian and had adequate literacy to fill out the questionnaire. They were excluded from the study if they were reluctant to participate in the research and could not fill out the questionnaire.
Acromegaly is a rare disease, and determining the sample size by applying the general principles of sampling (i.e., respondent-to-item ratio) yields an inappropriately high number of participants. It is recommended that, for each question in the questionnaire, at least 2-3 participants be considered. Thus, an appropriate psychometric tool should be employed for the studies with a low-to-moderate sample size in order to have a precise analysis. Thus, by using the smart Partial Least Square (PLS) software, and convenience sampling, the questionnaires were distributed among 75 patients with a confirmed diagnosis of acromegaly; eventually, 73 questionnaires were completed by the participants (14, 15).
Upon obtaining informed consent, we asked the participants to complete the AcroQoL questionnaire. It is a simple questionnaire designed to be self-administered, but for cases where it cannot be self-administered, it can be completed through an interview (11). The participants completed the questionnaire twice, with a minimum two-week interval. For some participants who had difficulty reading, the questionnaire was read by a researcher and completed through an interview.
The AcroQoL questionnaire has been designed specifically for the evaluation of HRQoL in patients with acromegaly. It consists of 22 items spread across two dimensions: physical (eight items) and psychosocial (14 items). The questions are scored on a five-point Likert scale. The responses are categorized as "always, most of the time, sometimes, rarely, never" where the item measures the frequency of the occurrence, and as "completely agree, moderately agree, neither agree nor disagree, moderately disagree, completely disagree" where the item measures the patient’s degree of agreement (11). Answers are scored from 1 to 5; the response “always” or "completely agree" scores 1, and the response “never” or “completely disagree” scores 5. The higher score the participant achieves, the lower the expected impact of acromegaly on HRQoL. The score ranges from 8 to 40 for the physical dimension, and from 14 to 70 for the psychological dimension, and the global score ranges from 22 (worst HRQoL) to 110 (best HRQoL). To standardize the score for the simplification of interpretation, the items can be scored from 0 (worst HRQoL) to 100 (best HRQoL) by using the following formula, where Y stands for the recalculated score, and X represents the sum of all the item responses within the dimension or study score (min. is the minimum possible score in the studied dimension, and max. is the maximum possible score in the studied dimension).
We used the Persian version of the AcroQoL questionnaire developed through back-translation by two native-like speakers of the English language, whose first language was Persian. One of the translators translated the original questionnaire into Persian, and the other one back-translated it into English, which was then reviewed by a panel of experts.
To quantify the content validity of the AcroQoL, eight expert endocrinologists were asked to examine the necessity/precision of each item for the Iranian culture by using a three-point rating scale (essential, useful but not essential, and not essential) and to rate the items of cultural relevancy, clarity, and simplicity. The content validity ratio (CVR) for every item was calculated by using the formula: [Ne- (N/2)] ÷ (N/2)], where Ne is the number of panelists choosing “essential” for each particular item, and N is the total number of panelists. To calculate the content validity index (CVI), the responses were rated from 1=not relevant, not simple, and not clear, to 4=very relevant, very simple, and very clear. Items with a CVI of >0.78 and CVR of >0.75 were accepted (16) (17).
Construct validity is an important type of validity, showing that the instrument measures what it claims to measure. Using confirmatory factor analysis (CFA), we assessed convergent and divergent validity, representative of the construct validity.
Reliability and internal consistency were evaluated by Cronbach's alpha analysis of the different dimensions of the questionnaire. The time interval for this assessment was at least two weeks. The ICC of > 0.70 for each domain indicates high reliability and internal consistency (18).
Having been approved by the Ethics Committee at Iran University of Medical Science (IUMS) (IR.IUMS.REC.1398.526), the study included 73 patients meeting the inclusion criteria.
Descriptive analysis was applied to analyze the patients’ characteristics. The AcroQol construct was evaluated via CFA in PLS 3. The outer loadings of the measurement indicators (>0.70) and the average variance extracted (AVE) (>0.40) of the model’s constructs were examined to establish convergent validity.
Moreover, the Fornell-Larcker criterion was used to demonstrate divergent validity in the cases when the AVE of a composite construct was higher than the construct’s highest squared correlation with any other composite construct (19). Coefficients of p-value and R square criterion were used to establish the structural model. A greater R square value for the endogenous structural model indicates a better-fitting model. The weak, medium, and strong fitness of the structural model is determined by R square standard and path coefficient. As a rough rule of thumb, R2 values of 0.75, 0.50, and 0.25 can be described as substantial, moderate, and weak, respectively. The direction and significance of the path coefficient will determine whether the structural model is fit (20). A goodness-of-fit (GoF) index was calculated to display whether the model fits the data (21). SPSS 20 was employed for data analysis, and the results are presented as mean ± SD (standard deviation) and frequency (percentage).