This was a mixed-method multicentre study conducted in five facilities in the Czech Republic (one outpatient renal clinic and four dialysis centres). The study was approved by the Ethical Committee of Faculty Hospital Kralovske Vinohrady [EK-VP/I1101202] and the Ethical Committee of Fresenius Medical Care [ekfmc_301/20].

When preparing the study design, we followed the COSMIN checklist for evaluating the methodological quality of studies on outcome measurement [15].

Concept analysis

The first step was a brief literature review of all concepts used in the IPOS-r followed by the translation and cultural adaptation of the measure.

Translation and cultural adaptation

This phase was based on guidelines for translation and cultural adaptation of the IPOS family instruments, available on the POS web page [16].

These guidelines are based on International Society for Pharmacoeconomics and Outcomes Research (ISPOR) guidelines [17] and are included in the Mapi Research Trust library specialising in Patient-Centred Outcomes.

Forward translation of IPOS-r was made by two translators with Czech as their first language: one was a health care worker, and the other was a professional translator. Their translations were merged by the research team, and the version for cognitive interviews was created. The Czech version of IPOS-r was then translated back into English by two translators with English as their first language, one with and one without a health care background, and both versions were sent to the POS team in the United Kingdom for the final check.

The final corrected version was used afterwards for cognitive interviews. We performed in-depth qualitative interviews to check the views of patients and staff on the outcome measure.

We reviewed ten patients with advanced kidney disease (three were on conservative management and seven were on haemodialysis) and ten members of the health care team (three physicians, six nurses, and one social worker). We used a convenience sample of respondents who were available and willing to participate in the renal clinic and two dialysis centres at two timepoints. Here is a brief guide to the cognitive pre-testing.

1. Patient/staff completed the IPOS-r. 2. We asked them how they understood the questions and the answers and how they chose from them. 3. We assessed how well they understood the measure and compared their assessment with their answers. In the case of misunderstandings, we asked them what was confusing, and then reformulated the wording. 4. For every item, we asked if was relevant for them. 5. Ultimately, we asked if the length of the measure was acceptable and if the recall period was optimal. 6. We asked if there were any questions that caused discomfort. 7. All the answers and comments on the measure were written down on the table, which was prepared for this purpose.

Content analysis of the answers and comments was performed, and the final IPOS-r version was created using patients’ and staff’s views on the measure. We used one-to-one interviews in which verbalization was used to access the thoughts and feelings, and to understand the ideas and interpretations, of respondents who are being asked to process information [18]. We used ‘think-aloud’ technique which was used retrospectively (once a measure was completed).

Validation

The validation phase was conducted in one outpatient clinic (Faculty Hospital Kralovske Vinohrady in Prague) and four dialysis centres across the Czech Republic (BBraun Avitum Ohradni in Prague, Fresenius Medical Care in Melnik, Fresenius Medical Care in Louny and Fresenius Medical Care in Slany). Data were collected by physicians, nurses, and social workers during regular patient encounters, or patients sent the completed measure by post. We included a convenience sample of adult patients with advanced kidney disease (eGFR < 15 ml/min/1.73m2) who were treated with haemodialysis, home haemodialysis, peritoneal dialysis, or conservative management. We excluded those who were cognitively impaired, did not have the Czech language as their mother tongue or were too unwell to participate. Patients were asked to participate by the health care professionals who were involved in the patient’s care. Participants completed the Czech IPOS-r independently or with help from their families or health care provider. Doctors, nurses, or social workers completed their version on the same day independently from the patients.

Measurement data were collected at three time points. Different instruments were used at each time point. At the first time point (T1), patients completed the Czech IPOS-r patient version and the Czech KDQOL-SF 1.2, and health care staff independently completed the Czech IPOS-r staff version. At the second time point (T2), patients completed the Czech IPOS-r 3 days after the first questionnaire had been completed. At the third time point (T3), the Czech IPOS-r was completed 1 month after the first questionnaire, and the patients answered an item asking if their situation had changed since their last completed the IPOS-r. The answer options for this external change criterion were “no”, “yes, negative change” or “yes, positive change”. A negative change meant deterioration of the patient’s overall condition, a positive change denoted an improvement in the patient’s overall condition. It was hypothesized that an improvement in the patient’s overall condition would be associated with a lowering in IPOS-r scores between the time points; deterioration in the patient’s overall condition would be associated with an increase in IPOS-r scores. During the third assessment, patients also completed the time needed to complete the IPOS-r.

Statistical analysis

Demographic data were reported using descriptive statistics. Patients who had any missing values in the IPOS-r were excluded from the analysis. A significant p value was set at 5%, and all analyses were conducted using SPSS version 28.01. We tested the item analysis, reliability, and validity of the Czech version of the IPOS-r as follows:

Item analysis

For every item of the IPOS-r, we computed the mean and standard deviation. We also computed item difficulty via the individual item‘s mean score and converted it to an interval (0;1) using the formula mean-scale min/(scale max-scale min). Correlations with the total score without a particular item were also provided. Item analysis provides information about the variance of scores and is also used for content validity [19]. Exploratory factor analysis was not done due to the small sample size.

Internal consistency

The internal consistency was determined via Cronbach’s α for the total score of the IPOS-r.

Reliability

Two types of reliability were computed. Test-retest reliability was determined based on the first and second assessments of the IPOS-r. We computed the level of perfect agreement for each item with quadratic weighted kappa. The test-retest reliability of the IPOS-r total score was assessed with intraclass correlation coefficient correlations (ICCs). ICCs of 0.7 were considered acceptable, but values > 0.8 indicated high test-retest reliability [20]. Interrater reliability was determined for patient and staff ratings at the first time point using weighted kappa, level of agreement for every item, and ICCs for the total score. The level of kappa from 0.41 to 0.60 was considered moderate, 0.61–0.80 as substantial, and 0.81–1 as almost perfect [21, 22].

Sensitivity to change

We also assessed the sensitivity to change in our sample using a distribution-based approach [23]. We compared mean changes based on the global change rating, which was assessed by patients during the third assessment after 1 month. Patients were divided into three groups: positive change, negative change, and no change according to their own assessment. The comparison was performed only using descriptive statistics, i.e., the mean change in T1 and T3.

Validity

To assess the convergent validity of the IPOS-r, we used the KDQOL-SF 1.2, which is the only validated measure that is used in the Czech Republic for assessing the symptom burden and concerns of patients with renal disease. We expected a high correlation (r > 0.70) for items related to the physical status of patients who had similar or identical items in KDQOL and a mid-range correlation (0.5–0.7) between items related to psychological and information needs from IPOS and KDQOL. There was a whole team consensus on the selected items using content analysis. If there were no questions assessing the same concept, we chose those assessing the most similar items; however, some concepts in the IPOS-r were missing in KDQOL (constipation, diarrhoea, sore or dry mouth). To assess validity, we used nonparametric Spearman correlations.