This descriptive cross-sectional study aimed to translate the Revised Illness Perception Questionnaire (IPQ-R) into Arabic and validate the translated questionnaire among Arabic hypertensive patients from the outpatient departments of the Prince Sattam University Hospital and King Khalid Hospital (KKH) in Al-Kharj City in the Kingdom of Saudi Arabia.

We employed an IPQ-R (Hypertension) which was modified from the IPQ-R, a frequently used instrument that assesses patients' perception of their illness. The IPQ-R (Hypertension) comprises three components: illness representation, causal attributions, and identity (20). All questions under illness representation and causal attribution were scored on a five-point Likert scale where 1 = strongly disagree and 5 = strongly agree. Illness representation consisted of 26 questions within 7 subscales pertaining to emotional representations (higher score indicating a stronger emotional impact of the condition), timeline (acute/chronic) (higher score indicating chronicity), Time-cyclical (higher score indicating higher variability), illness coherence (higher score indicating patients' increased awareness of their illness), personal control (higher score indicating patients' perceived efficacy in controlling their illness), treatment control (higher score indicating patients' belief that the treatment is Effective at controlling their illness), consequences (higher scores indicating more severe impact of the illness on patients' lives). Causal attribution components included 18 questions pertaining to 4 subscales psychological attributions, risk factor attributions, immune attributions, and chance attributions, with higher scores indicating a stronger belief that a factor is causally related to their disease. 19 symptoms associated with high blood pressure were included in the identity component. Responses to them were recorded binarily as Yes or No. Only those symptoms identified by patients as related to their hypertension counted toward the identity score.

A bilingual panel consisting of doctors who were proficient in English and Arabic was assembled to translate the IPQ-R into Arabic. Best practices were followed as set out by the Patient-Reported Outcome Consortium's instrument translation process (21). The panel members consisting of 1 general practitioner, 2 family medicine specialists, and 2 cardiologists were familiar with the management of hypertension. An external medical translator was on the panel to review and ascertain the accuracy of the translation and resolve differences in translation. A back translation of the Arabic IPQ-R was performed by another panel of doctors (consisting of 1 general practitioner, 1 family medicine specialist, and 1 cardiologist) and a medical translator who were blind to the original English questionnaire and the aims and objectives of the study to further ensure the equivalence of translation. The back translation was deemed satisfactorily similar to the English IPQ-R. Differences between the Arabic IPQ-R and the back-translated English IPQ-R were resolved after the original panel conferred with the medical translators.

Non-probability convenient sampling was conducted at the general family medicine and cardiology outpatient departments of the Prince Sattam University Hospital and KKH in Al-Kharj City in the Kingdom of Saudi Arabia. Hundred patients were recruited from 1 August to 30 September 2021. Inclusion and exclusion criteria were followed. The inclusion criteria were adult patients of either sex older than 18 years of age, clinically diagnosed with primary hypertension for at least 6 months prior to the study, native Arabic by ethnicity, and fluent and literate in Arabic. The exclusion criteria were patients experiencing acute distressing symptoms or suffering complications of hypertension such as chest pain, hypertensive emergencies, or renal failure; diagnosed with secondary hypertension; diagnosed with other mental or physical comorbid conditions that could influence the perception of illness aside from hypertension such as diabetes, dyslipidemia, anxiety, or depression; patients unable to read the questionnaire such as those who cannot read Arabic or those with severe mental or visual impairment; and patients who did not give consent.

Cognitive debriefing interviews of ten participants were carried out to establish whether the participants interpreted the translated questions as intended and used the response scale appropriately. A standardized approach to cognitive debriefing was employed (22)—a trained investigator performed in-person semi structured interviews and used the think-aloud method. In the think-aloud method the participant completes the questionnaire by reading all items aloud and vocalizing their deciding processes and answers. The fact that the interviewer remains silent during the entire process mitigates interviewer bias. The participants provided their understanding of the questions in the questionnaire, the type of information needed to recall the answers, and the deciding processes undertaken to answer the questions appropriately. The participants were required to provide comments and general feedback about their experience in answering the questionnaire, e.g., if it was too long, if it caused significant mental or emotional burden, etc. The responses were manually transcribed by the interviewers, but were not included in the analysis of the validity of the questionnaire. They were used to revise the questionnaire as necessary.

This study was approved by the Research Ethics Committee in Health and Science Disciplines of the Prince Sattam bin Abdulaziz University. Ethical approval number [REC-HSD-60-2021]. Informed consent was obtained from all the participants. Strict measures were observed to anonymize the personal identifiers of the participants to protect their privacy and confidentiality. Apart from the principal investigator who had access to the complete dataset all members of the research team were given access only to anonymized data.

Participants who met the inclusion and exclusion criteria were given the questionnaire for self-administration after their consultations. The participants completed the questionnaire in a dedicated room to ensure comfort and privacy. They were instructed to ask the investigator for clarifications if they had difficulties with understanding or answering the questions. The participants took an average of 10 min to complete the questionnaire. Demographic data such as age, sex, and socioeconomic status were recorded by the investigator.

The SPSS (Statistical Package Social Science) software version 20 and Python programming language version 3.9 were used for statistical analysis. Descriptive statistics of the participants' characteristics were computed. Descriptive statistics of the IPQ-R domains and the frequency and percentages of the responses were also measured.

In order to measure reliability, test-retest analysis was performed on a random subsample of 10 patients to whom the questionnaire was administered twice with a gap of 6 weeks. Weighted Cohen's kappa with a linear weighting matrix was used for Likert scale items to compensate for the degree of disagreement. Simple Cohen's kappa was applied to binary items. Cronbach's alpha values were calculated for each domain separately and also for the overall Likert scale items for determination of internal consistency. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was applied to check the intercorrelation between each item of scale. KMO values > 0.40 were considered acceptable. Bartlett's Test of Sphericity was also applied to check if there was redundancy between the variables. A 95% confidence interval (CI) was calculated for coefficients.

100 patients participated in this study and completed the IPQ-R. Fifty-seven patients (57%) were male and sixty-five (65%) were older than 40 years. Sixty nine percent of the population had a family history of hypertension and 49% were from the lower socioeconomic status group. Demographic details of the participants are presented in Table 1.

The most frequently reported symptom of hypertension was headache in 65% of the participants, followed by fatigue (53%). Other common reported symptoms were fast heart rate (44%), loss of strength (34%), dizziness (31%), breathlessness (31%), and sleep difficulties as shown in Figure 1.

The mean scores of the patient responses in total and by domain along with the Cronbach's alpha are presented in Table 2. Participants had the highest mean score (3.65, while SD = 0.66) in the domain of personal control which comprised 4 questions. Other domains in which the participants had high mean scores were illness coherence (mean = 3.41, SD = 1.09) and time-cyclical (mean = 3.41 SD = 1.09). The lowest scores were reported in the domains of causes (mean = 2.95, SD = 0.53) and consequences (mean = 2.73, SD = 0.71) which comprised 18 and 5 questions respectively. The domain of disease identity comprising 19 questions which were reported as a binary variable (no = 0 or yes = 1) had a mean of 0.27 and a standard deviation of 0.19.

Except for illness coherence which had only one question, the kappa coefficient was calculated for each domain, including disease identity (0.78; 95% CI: 0.62–0.87), emotional representation (0.49; 95% CI: 0.36–0.63), time (0.49; 95% CI: 0.42–0.55), time-cyclical (0.41; 95% CI: 0.05–0.77), personal control (0.39; 95% CI: 0.31–0.44), treatment (0.22; 95% CI: 0.04–0.34), consequences (0.40; 95% CI: 0.24–0.48), and causes (0.39; 95% CI: 0.24–0.51). The kappa coefficient of the questionnaire was 0.52 (95% CI: 0.44–0.59) overall. Kappa coefficients with 95% confidence interval as determined by the test-retest analysis are depicted in Figure 2.

As described in Table 2, the internal consistency of the questionnaire was determined within each domain and for the whole questionnaire using Cronbach's alpha. Within the domains the Cronbach's alpha levels ranged from 0.34 in the treatment control domain to 0.78 in the disease identity domain. The internal consistency of the whole questionnaire excluding the domain of disease identity was 0.76 indicating satisfactory internal consistency. Given that the disease identity domain was measured on a dichotomous (yes/no) scale compared to the 5-point Likert scale response for the illness representation and the casual attribution domains, it was omitted when computing the Cronbach's alpha for the whole questionnaire.

There were weak to moderate positive linear correlations (r = 0.003–0.561) between the domains of the IPQ–R as shown in Table 3 suggesting a reasonable discriminant validity among the domains. The domain of emotional representation has a weak positive correlation to time (r = 0.258) and a moderate positive correlation to consequences (r = 0.478) (p < 0.01). The time-cyclical domain has a weak positive correlation to personal control (r = 0.265, p < 0.01) and treatment control (r = 0.315, p < 0.01). Personal control has a moderate positive correlation to treatment control (r = 0.561, p < 0.01) and a weak positive correlation to consequences (r = −0.240, p < 0.01) and illness coherence (r = 0.421, p < 0.01). Treatment control and illness coherence have a weak positive correlation (r = 0.260, p < 0.01) as do consequences and causes (r = 0.425, p < 0.01).

The Kaiser-Meyer-Olkin (KMO) test and Bartlett's Test of Sphericity were carried out to test variable collinearity. We obtained a KMO measure of 0.549 and a significance level of < 0.001 for the Bartlett's statistic indicating substantial correlation in the data.

Illness perceptions of hypertension may change with the duration, severity, and complications of hypertension. Since our study included limited patients, further study is required to validate the questionnaire in patients with severe hypertension or those who have experienced complications related to hypertension. Studies at multicenter and multinational levels are also required to ascertain the generalizability of the translated version in different cultural contexts and other Arabic speaking countries. A sample of 100 is smaller than other similar studies (27, 28). The relatively small sample size limited performing the principal component analysis and factor analysis of the Arabic translated questionnaire.

In conclusion, given that Cronbach's alpha values for most of the subscales in the Arabic translated version of the IPQ-R for hypertensive patients were unsatisfactory (< 0.7), this Arabic translation of the IPQ-R did not show sufficient internal consistency to adequately ascertain the illness perceptions of hypertensive patients living in Saudi Arabia. However, the overall questionnaire kappa as ascertained by test-retest analysis was in the moderate range of agreement for instrument reliability. Future studies can employ questionnaires other than the IPQ-R which are validated for hypertension or employ alternative Arabic translations of the IPQ-R to ascertain illness perceptions of Arabic speaking hypertensive patients.