Study design
PRO instrument development employed qualitative data collection and analysis techniques consistent with scientific best practices [16,17,18].
A review of the literature, including evaluation of existing PRO instruments used in CF clinical studies, was conducted to inform instrument development. The review identified 45 different PRO instruments used in published CF studies during the 15-year review period. Only four of the identified instruments were developed as CF-specific assessments, with three of those (the CFQoL, FLZM-CF, and CFQ-R, as detailed above) including coverage of concepts related to the impact of CF on daily life and treatment-related experiences, as opposed to assessment limited to CF symptoms.
The review was followed by qualitative interviews with clinicians and nurses providing care for patients with CF. The study team used input from CF specialists and instrument development experts, as well as the findings from the literature review and clinician interviews, to inform protocol design and to develop semi-structured interview guides for use in qualitative data collection with patients with CF and their caregivers.
Following development of interview guides, in-person qualitative concept elicitation interviews were conducted with pediatric, adolescent, and adult patients with CF to identify symptoms, symptom-related impacts, and treatment experiences relevant and important to patients. Primary caregivers of participating children aged 6–17 years were also interviewed. All interviews were audio-recorded and transcribed for concept coding. Figure 1 provides an overview of the key tasks undertaken in PRO development.
Study population
Eight CF treatment centers across the US recruited patients from their ongoing caseloads to participate in the qualitative interviews. Seven sites enrolled patients for the initial concept elicitation interviews and 4 sites (3 remaining from the concept elicitation phase plus 1 additional clinic) enrolled patients during the cognitive interview phase. A central institutional review board (IRB), Quorum Review IRB (Seattle, WA, USA), approved the study protocol and materials, as did local IRBs where required. All patients (or parents/legal guardians of patients aged < 18 years) provided written informed consent prior to initiation of any study-related procedures. No changes to the participants’ existing care were made, and no clinical interventions were initiated as part of the study protocol.
A purposive, non-probability sample of approximately 40–50 patients with CF was planned for the concept elicitation phase. Eligible patients had a physician-confirmed diagnosis of CF, defined as a sweat chloride value ≥60 mmol/L by quantitative pilocarpine iontophoresis, or 2 documented CF-causing mutations, and either chronic sinopulmonary disease or gastro-intestinal/nutritional abnormalities. Recruitment quotas within 5 age groups (6–8, 9–11, 12–17, 18–24, and ≥ 25 years) were used to capture age-related differences over the course of the disease, as well as in type and intensity of life activities. To reflect the range of comorbidities seen in patients with CF, additional recruitment targets were used to guide the inclusion of patients in 2 key clinical subgroups: 1) patients with comorbid CF-related diabetes were limited to ≤33% of the overall sample; and 2) patients with more severe CF presentations (defined as a history of ≥2 hospitalizations in the previous year for patients aged ≥18 years and ≥ 1 hospitalization in the prior year for patients aged < 18 years) were targeted to represent approximately 33% of the overall sample. Recruitment targets were set for approximately 50% of the sample to be composed of patients receiving CFTR modulator therapy at the time of interview.
During the enrollment process, patients completed a questionnaire to provide demographic characteristics (age, sex, ethnicity, race, educational attainment, employment status) and ratings of their overall health compared with others and the overall severity of their CF symptoms.
Qualitative concept elicitation interviews
Individual concept elicitation interviews were conducted to gather information about key symptoms and disease-related life impacts experienced by patients with CF, as well as functional changes in daily living following treatment. All interviews lasted approximately 60–90 min and used a semi-structured interview guide, starting with general, open-ended questions and a day reconstruction exercise, and continuing with a structured probing system to gather more detailed information about participants’ symptoms, disease impacts, and treatment-related experiences. The probe system was informed by clinician interviews and a systematic review of existing CF literature. Separate, age-appropriate interview guides were used with children (aged 6–11 years), adolescents/adult patients (aged ≥12 years), and the caregivers of participating patients aged < 18 years. Interviewers followed each institution’s infection control protocols (i.e. mask, gloves, gown procedures) during the patient interviews. Young children were interviewed with a parent or caregiver present, if desired, but they provided their own responses to interview questions. In most cases, the caregiver interviews were conducted separately, without the child present in the room at the time the caregiver provided his/her responses.
Analyses
All interviews were audio-recorded and transcribed. Transcript coding was conducted using ATLAS.ti software (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany) [27], and coded expressions were grouped by similarity of content for analysis. Coded transcript data were evaluated for evidence of saturation of concept (defined as the point at which no new concept-level information was forthcoming from the study sample) to confirm the adequacy of the sample size and robustness of the set of concepts elicited. To detect the point of concept saturation, a summary saturation table was used to track the occurrence of novel impact concepts expressed within chronologically ordered sets of interviews. Consistency and reproducibility of concept coding were assessed via evaluation of inter-rater agreement (IRA).
Item generation
The development team met with CF clinical experts and clinical outcome assessment measurement experts to review results of the concept elicitation interviews and identify the concepts most appropriate to include in a CF-impact PRO. The hypothesized relationships between the selected concepts and domains of the draft instrument were modeled in a preliminary conceptual framework. For each selected concept, transcript data were used to develop draft PRO item text (i.e. question and scale) reflecting the choice of words and descriptions used by patients to articulate the concept.
Cognitive interviews
Following the development of the preliminary draft instrument, cognitive interviews were conducted to evaluate concept relevance and understandability, the structure of the draft items and response scales, and the appropriateness of the recall period. Interviews were conducted in 3 iterative waves; the development team reviewed the findings from each wave and implemented refinements to the draft instrument before each subsequent set of interviews. The preliminary conceptual framework for the draft instrument was adjusted based on refinements made during the cognitive interviews.
Cognitive interview participants were selected through the same recruitment process used for qualitative interviews. Cognitive interviews used a semi-structured interview guide, which included broad questions to assess the comprehensiveness and relevance of the concept coverage of the CF-IQ, as well as a “think-aloud” process to ensure the instructions, items, and response options were interpreted as intended for each item of the CF-IQ [18, 28]. When possible, questions were reframed in the terminology used by respondents to aid comprehension.