A quality-of-life measure for adults with primary ciliary dyskinesia: QOL–PCD

Literature review and expert panel

First, a systematic literature review was conducted to identify key symptoms and effects of PCD on patient functioning. MEDLINE and EMBASE were searched, and additional references were sought through citations in the identified studies. Abstracts were reviewed and manuscripts sourced for research investigating the effects of PCD on adults.

In the next step, expert clinicians, allied health professionals and researchers met to discuss their own perceptions of the impact of PCD on adults, based on their clinical experiences. These sources of information contributed to a long list of items patients rated for relevance. These items also informed the development of the open-ended interview guide.

Participants

In the UK, participants for the open-ended and cognitive interviews were recruited from PCD clinics and from an advert through the PCD Family Support Group UK. A list of potential items was sent to the Family Support Group in the UK to rate their relevance. Participants in North America were recruited from a cohort of PCD patients evaluated at the University of North Carolina, as well as from the US PCD Foundation's registry of patients. Expert opinion was also sought during the item generation and item reduction phases from members of the European PCD Group.

Criteria for participation in the open-ended and cognitive interviews included age ≥18 years with a diagnosis of PCD. Patients were recruited from English-speaking countries: UK, USA and Canada. UK participants had an existing diagnosis from one of the English diagnostic centres [1, 24] based on clinical phenotype plus high-speed video analysis of ciliary function and/or assessment of ciliary ultrastructure by electron microscopy. North American participants were diagnosed at a specialised PCD research center, based on: a compatible clinical phenotype plus defect in ciliary ultrastructure and/or identification of biallellic disease-causing mutations in one of the PCD genes.

Open-ended interviews

In-depth interviews were conducted either in-person or by phone to elicit the effects of PCD from the patients' perspective. Interviews were conducted in the UK by L. Behan and in USA and Canada by A.L. Quittner, A. Alpern and A.M. Morris. All participants were fluent in English. We attempted to interview patients who were geographically representative and clinically stable. The audio of all interviews was recorded and transcribed for content analysis using either NVivo (version 8 2008; QSR International Pty Ltd, Daresbury, UK) in the UK or Atlas.ti (Version 7.0; Scientific Software Development, Corvallis, OR, USA) in the USA. Thematic coding was used to identify key symptoms and psychosocial impacts. Two members of each research team coded the interview transcripts using consensus coding. When there was a discrepancy, this was discussed and resolved within the pair of coders. We did not calculate percentage agreement since we used a consensus coding process. These data were then analysed to identify critical items based on frequency of endorsement. Content analysis of these transcripts yielded saturation and indicated that the measure was comprehensive and that all relevant items were included.

Item generation

Questions from the literature review, expert panel and open-ended interviews were also sent by post to respondents of an advert circulated to members of the PCD Support Group and to adult patients at PCD clinics in the UK. A pre-paid envelope and covering letter was included. Participants rated each item using a 5-point Likert scale (1: “not relevant”; to 5: “highly relevant”).

Items for the initial QOL–PCD measure were based primarily on the patient-based content analysis. We discussed these results in a series of teleconferences chaired by J.S. Lucas. Each meeting included clinicians, psychologists and interviewers from Ireland, the UK and the USA. Decisions made using a modified Delphi approach guided by two main principles. The primary criterion for including an item was its impact on HRQoL, measured by the frequency with which items were mentioned across patients and in relation to other items mentioned. There was no pre-determined frequency required for inclusion of an item, but the researchers considered the frequency each item was mentioned relative to the other items in that domain. They also considered the importance interviewees placed on these items. Secondly, patients' ratings of relevance from the UK survey were considered. For each item, decisions were made to include an item based on our discussion. When there was initial disagreement, the chair invited each individual to explain their rationale. Unanimous agreement was achieved within two rounds of discussion. Finally, we determined that all of the content had been identified based on saturation matrices in the UK and US, which showed that no new content emerged after six patients, on average, in each country. We included at least four items on each subscale to ensure adequate internal consistency.

Construct of prototype questionnaire

Agreement on item selection and wording was achieved during multi-disciplinary, multinational conference calls. Selected items were written using patient language obtained in the qualitative interviews and then combined into scales (e.g. frequency and severity of respiratory symptoms, perceptions of treatment burden). Where appropriate, the items were formulated into questions based on the Cystic Fibrosis Questionnaire-Revised (CFQ-R), which has been well-validated [12, 25]. Some example CFQ-R items that were adapted to the PCD HRQoL measure include: “Did you cough during the day?” and “How often does CF get in the way of meeting your school, work, or personal goals?” Items were written to ensure conceptual, cultural and linguistic equivalence for North America, the UK and Ireland, by researchers from the USA, England and Ireland. We also adhered to both the FDA and EMEA Guidance [13, 22, 23]. For example, as recommended, we asked patients to answer questions based on their symptoms and impact over the past week.

Cognitive interviews

Cognitive interviews were conducted prior to formal psychometric validation of questionnaires to evaluate how respondents process the question and rating options cognitively; i.e. What “meaning” do these items have for respondents? Is this the same meaning we intended? What were they considering when rating frequency or impact? Specifically, we wanted to identify any problems with the instructions, organisation of the questionnaire, item interpretation, memory retrieval, decision-making processes and response selection. A “think aloud” procedure was used to investigate the participants' comprehension of the instructions, items and rating scales. They were asked clarifying questions, such as: “What were you thinking of when answering that question?” and “What does X word mean to you?” For example, “What does feeling ‘well’ mean to you?”, “What would have made you endorse a higher/lower frequency for that question?”, “How relevant/important is this question for you?”, “Are the rating options clear to you?” and “Are they easy to use?”

Participants were first asked to complete the prototype questionnaire independently. Next, they were interviewed using specific cognitive probes, which focused on the clarity of the question, its meaning, relevance and importance, and what would have shifted their response to an adjacent answer. The audio of all interviews was recorded and transcribed. The results were discussed during a series of teleconferences to determine whether revisions were required for the format, instructions or items. The measure was refined based on the cognitive interviews and then finalised.