Changes in the Parent Cystic Fibrosis Questionnaire-Revised (CFQ-R) with respiratory symptoms in preschool children with cystic fibrosis

Abstract

Introduction

The Cystic Fibrosis Questionnaire-Revised (CFQ-R) Respiratory score is a validated and widely used patient-reported outcome. This study aimed to establish changes in the score with acute respiratory events in preschool children with CF and to investigate its’ relationship with physiological outcomes.

Methods

The Parent CFQ-R, multiple breath washout test and spirometry were performed at six study visits over one year. The clinical status of participants, stable or symptomatic, was defined by the patient’s physician. Linear regression and distribution-based statistical methods were used to examine the changes in the CFQ-R from the last stable visit and to investigate its relationship with physiological outcomes.

Results

There were 272 stable and 115 symptomatic visits from 78 participants. The mean CFQ-R Respiratory score did not change between consecutive stable visits (−0.73, SD 20.4). The mean (SD) score deteriorated by 15.5 (20.7) points between stable and symptomatic visits and improved by 14.8 (20.1) points between symptomatic and stable follow-up visits. When a clinically important change is defined as 0.5SD change (10-points), the positive predictive value (PPV) was 45% and the negative predictive value (NPV) was 84%. For visits with a 10-point worsening in the CFQ-R Respiratory score and a 15% increase in LCI, the PPV was better (81%) than using either measure alone.

Conclusion

The CFQ-R Respiratory score is responsive to acute respiratory events in preschool children with CF and its utility to monitor individual patients is improved when combined with LCI.

1. Introduction

Patient-reported outcomes (PRO’s) measure symptoms and daily functioning from the patient perspective and are useful outcome measures in young children with respiratory disease [1–3]. The Cystic Fibrosis Questionnaire-Revised (CFQ-R) is the most widely used PRO for cystic fibrosis (CF), especially for interventional trials [4–8]. The CFQ-R has demonstrated reliability and validity [4,5,9], and the CFQ-R Respiratory score, which is the most specific to lung disease, correlates with forced expiratory volume in 1 second (FEV₁), body mass index (BMI), height and weight in older age groups [5,10]. The Parent CFQ-R (ages 6–13) was downwardly extended and modified for infants and preschool children for the Infant Study of Inhaled Saline (ISIS) study [11]. Alpern and colleagues evaluated the reliability of the CFQ-R in preschool children and demonstrated good construct validity and internal consistency [9]. The minimal important difference (MID) of a change in the score, the smallest difference patients perceive as beneficial and which would trigger a change in management, has been reported for patients 6-years and older [4]. However, a meaningful clinical change in the CFQ-R Respiratory score in the preschool population is unknown.

There is a need for practical outcome measures to guide clinical care and for use in interventional trials as new CFTR modulators and other treatment modalities are evaluated in young children, for whom traditional spirometry is challenging and not routinely used. Pulmonary exacerbations, episodes of acute worsening of respiratory symptoms, are challenging to diagnose; young children typically have few or no respiratory symptoms at baseline and increased cough is often the result of viral upper respiratory tract infections. Consequently, treatment decisions vary widely among clinicians. The CFQ-R Respiratory score could be a useful tool to guide treatment decisions [12,13]. The study aimed to evaluate the responsiveness of the CFQ-R Respiratory score to acute respiratory events in preschool children with CF and to establish a clinically important change in the score with increased respiratory symptoms. A secondary objective was to define the score’s relationship with physiological outcomes in this age group. We examined whether combining patient-reported and physiological outcomes resulted in improved identification of respiratory events compared to using any one outcome measure on its own.

2. Methods

2.1. Study Population

This was a secondary analysis of a previously published multicentre 12-month longitudinal observational study in preschool children [14,15]. Briefly, data were collected from participants aged 2.5–6 years with CF from three North American CF centers. Participants attended study visits at enrolment, 1, 3, 6, 9 and 12 months. At each visit a clinical examination, a review of past and current symptoms and treatments, a modified Parent CFQ-R score [9], the multiple breath washout (MBW) test and spirometry were performed.

As previously described [15], documentation for all visits, including both study and clinical notes, were reviewed independently by two physicians to classify visits as symptomatic or stable based on participant/parent-reported symptoms and physical examination findings. Disagreements between reviewers were discussed, and a consensus was achieved in all cases.

Symptomatic visits were defined as those where lower respiratory tract symptoms (cough, wheeze, sputum production or shortness of breath) or signs (work of breathing, hypoxemia, tachpnoea, wheeze, crackles and differential air entry) were judged by the physician to be increased from the participant’s baseline. Symptomatic visits included treated and untreated episodes of increased respiratory symptoms.

Visits were defined as stable if there were no new or increased respiratory symptoms. Follow-up visits were defined as stable if all increased respiratory symptoms had fully resolved and the participant was back to their baseline status as judged by the physician. Respiratory events were defined as pulmonary exacerbations if they were treated with antibiotics (either oral, intravenous or inhaled). Study visits where the participant had upper airway symptoms but without an increase in lower respiratory tract symptoms were not considered symptomatic or stable visits and were excluded.

2.2. Symptoms Questionnaire

The modified version of the Parent CFQ-R (ages 6 to 13 years) was administered at every study visit during the 12-month study period [9]. This modified PRO consists of 26 items over multiple domains. The Respiratory Domain has 6 items and was the primary focus of this analysis; this domain was retained in its entirety in the revised version and no items were reworded. Parents reported on their child’s health over the past two weeks, using a 4-point Likert scale. Responses of the six-item Respiratory Domain were standardized from 0–100, with higher scores indicating better Health-related quality of life (HRQoL) [5].

2.3. Multiple Breath Washout

MBW tests were performed with the Exhalyzer®D (EcoMedics AG, Duernten, Switzerland), with adaptions for preschool children[14]. MBW tests were considered successful if there were at least two technically acceptable trials. Lung clearance index (LCI) was calculated as the cumulative expired volume (CEV) divided by the functional residual capacity (FRC) at 1/40^th of the starting gas concentration.

2.4. Spirometry

FEV₁ was reported if the forced expiratory maneuver was technically acceptable [16] and the forced expiratory time (FET) was at least 1 second, FEV_0.75 was recorded if the FET was at least 0.75 seconds. A spirometry test was considered successful if either FEV_0.75 or FEV₁ results were obtained from technically acceptable forced expiratory maneuvers. Z-scores for either FEV_0.75 or FEV₁ were calculated using the ethnic specific Global Lung Function Initiative Spirometry reference equations [17]. zFEVt was defined as the FEV₁ score, when the z-score for FEV₁ was available, otherwise the z-score for FEV_0.75 was used.

2.5. Statistical analysis

All statistical analyses were performed in Stata Statistical Software Release 15 (StataCorp, College Station, TX). We examined whether baseline CFQ-R Respiratory score at the first stable visit was predictive of the pulmonary exacerbation rate (i.e. treated symptomatic events) during the 12-month study period using Poisson regression analysis. Linear regression was used to investigate whether a lower CFQ-R Respiratory score at the first stable visit was associated with other characteristics at enrolment such as gender, age, race, BMI, height, weight, LCI or zFEVt.

We described the absolute changes in CFQ-R Respiratory score between two consecutive stable visits, between stable and symptomatic visits, and between a symptomatic and a consecutive stable follow-up visit. These changes were compared using linear regression within robust standard errors to account for multiple measurements in the same participant.

This study was not designed to determine the MID for this age-group because it did not incorporate a global rating-of-change questionnaire (GRCQ) at each visit to assess the participant’s perceptions of respiratory symptom change. Instead, we estimated a threshold for a clinically important change in the CFQ-R Respiratory score by investigating how the score deteriorated between a recent stable and a symptomatic visit and improved from a symptomatic to a stable follow-up visit. Distribution-based statistical methods were used, similar to the those used to determine the MID of the CFQ-R in older patients (>6 years) [4,18,19]: 1) 0.5 SD for mean CFQ-R Respiratory score change 2) 1 standard error of the mean (SEM) for baseline CFQ-R Respiratory scores (1SEM = SD√1−α; α = reliability score). We used the α co-efficient value for the Respiratory Domain of the Parent CFQ-R in preschool children as defined by Alpern and colleagues (Cronbach α value of 0.75) [9]. We also investigated the variability of the score between consecutive stable visits by calculating 1.96 and 0.5 standard deviation (SD) for the mean score change.

Linear regression with robust standard errors was used to evaluate the relationship between CFQ-R Respiratory scores – the standardized score (0–100) and individual questions - and physiological outcomes (LCI and z FEVt). We examined whether combining the changes in CFQ-R and lung function resulted in improved identification of symptomatic and stable events compared to using CFQ-R or lung function alone (where LCI or zFEVt and CFQ-R changes were available for the same visit). A 15% relative increase in LCI from a stable visit or a 1.3 z-score decrease in zFEVt was considered significant worsening in lung function [14,15].

3. Results

3.1. Study Population

Over the 12-month study period, there were 470 study visits from 78 participants with CF; 422 visits had usable CFQ-R data of which 272 were stable and 115 symptomatic visits (Figure 1).

Figure 1:

Flow diagram of study population - * excluded from analysis

The baseline demographics of participants are shown in Table 1.

Table 1:

Participant demographics at enrolment

Characteristic	N=78
Age (years), mean (range)	4.3 (12.6 to 5.9)
Gender, female, n (%)	39 (50.0)
White, n (%)	69 (88.5)
Pancreatic insufficient, n (%)	70 (93)
Height-for-age centile mean (range)	44.2 (0.8 to 97.5)
Weight-for-age centile mean (range)	43.1 (2.0 to 94.7)
BMI-for-age centile mean (range)	47.5 (0.99 to 94.9)
CFQR respiratory score, median (range)*	94 (44–100)
LCI (units), mean (range)	8.8 (6.4 to 13.6)
z FEVt, mean (range)	−0.7 (−4.3 to 1.8)
Mutation Class, n (%)
I-III	72 (96)
IV-V	3 (4)

3.2. Changes in the CFQ-R between stable visits

The median CFQ-R Respiratory score at the first stable visit was 94 points (IQR 83–100).

Nearly 20% of all visits (n=76) had a maximum score of 100 points. Overall, the CFQ-R Respiratory score did not change between consecutive stable visits (−0.73, SD 20.4). Applying a 1.96 SD limit for the mean score change between stable visits, 95% of stable visits had a score change within +/− 40 points. However, using a 0.5 SD limit, a relevant change could be considered greater than 10.2 points (Figure 2). On an individual basis, parental reporting of symptoms between stable visits was variable; we found that there was a worsening of the CFQ-R Respiratory score by at least 10 points with 22% (n=43) of stable visits.

Figure 2:

Histograms showing the change in CFQ-R Respiratory score (points) between stable visits. The dashed lines represent ±1.96 SD of the mean and the solid lines represent ±0.5 SD.

3.3. Changes in the CFQ-R with symptomatic visits

On average, each participant had one symptomatic event, whether treated or untreated (range 0–6). Cough and congestion were the most frequently endorsed items at stable visits and were the most responsive to change with symptomatic events. The questions relating to the presence of wheeze, trouble-breathing and nocturnal cough were less frequently endorsed at stable visits and changed little with symptomatic events (Figure 3). There was no association between baseline CFQ-R and other characteristics at enrolment such as gender, age, race, BMI, height, weight, LCI or zFEVt.

Figure 3:

Heat map indicating the frequency that individual questions are endorsed at stable and symptomatic visits.

On average, the mean CFQ-R Respiratory score deteriorated by 15.5 (20.7) points with symptomatic events (13.2 (20.4) points with untreated events and 17.4 (21.0) with treated events) (Table 2 and Figure 4). At consecutive stable follow-up visits, the CFQ-R Respiratory score improved significantly for treated and untreated events, but the magnitude of change was greater for patients treated with antibiotics (Figure 4). The estimated changes (0.5 SD and 1 SEM) in the CFQ-R Respiratory score from a stable to a symptomatic visit were 10.4 and 8.1 points and from a symptomatic to a stable follow-up visit were 10.1 and 7.1 points.

Table 2:

CFQ-R Respiratory scores changes

CFQ-R Respiratory score changes	All symptomatic events Mean (SD)	Untreated Visits Mean (SD)	Treated Visits Mean (SD)	p-value
Stable to symptoms	N=54	N=24	N=30	<0.001*
	−15.5 (20.7)	−13.2 (20.4)	−17.4 (21.0)
Limits of change
0.5 SD	10.4 points	10.2 points	10.5 points
SEM	8.1 points	8.2 points	8.5 points
Symptoms to follow-up	N=44	N=19	N=25	<0.001*
	14.8 (20.1)	10.6 (19.4)	18.9 (20.5)
Limits of change
0.5 SD	10.1 points	9.7 points	10.3 points
SEM	7.1 points	8.2 points	8.5 points

^*Linear regression model- treated and untreated groups are significantly different from stable group

Figure 4:

The CFQ-R Respiratory score deteriorated from stable to symptoms and improved from symptoms to stable follow-up (pattern). Boxplots indicate the median value (centre line); inter-quartile range (box) and minimum and maximum values (error bars). The horizontal dashed lines represent a 10-point change in the CFQ-R Respiratory score.

At an individual level, the smallest possible change in the CFQ-R Respiratory score is a one category change for one question (e.g. worsening in day-time cough report from “not at all” to “a little”). This minimal change corresponds to a 5.6-point change in the scaled score, so a 10-point change corresponds to a two-category change on one question or a one category change on two different questions.

The sensitivity and specificity of a 10-point change to identify a symptomatic event was 65% (95% confidence interval (CI) 51to 77) and 70% (95% CI 62 to 77), respectively. The probability of correctly identifying a symptomatic event (i.e. positive predictive value or PPV) with a 10-point change in the CFQ-R change was 45% (35/78) and negative predictive value (NPV) was 84% (101/120).

3.4. CFQ-R and lung function

Including all stable and symptomatic visits, there was a weak but significant relationship between the relative change in LCI and the CFQ-R Respiratory score (coefficient −0.14; 95% CI −0.26; −0.02; p=0.03). The relationship between the absolute change in zFEVt and CFQ-R Respiratory score was not statistically significant (coefficient 0.01; 95% CI −0.001; 0.01; p=0.10). There was also a relationship between the relative change in LCI and increasing symptom report for daytime cough (coefficient 22.7; 95% CI 13.3; 32.1; p=<0.001), mucus production (coefficient 31.0; 95% CI 18.0; 44.0; p<0.001), wheeze (coefficient 15.8; 95% CI 9.5; 22.1; p<0.001), trouble breathing (coefficient 14.0; 95% CI 7.6; 20.3; p<0.001) and nocturnal cough (coefficient 13.2; 95% CI 2.7; 23.7; p=0.01). Nasal congestion was the only item not associated with LCI changes (coefficient 8.1; 95% CI −2.3; 18.5; p=0.12). For spirometry, lower zFEVt was associated with increasing symptom report of daytime cough (coefficient −0.6; 95% CI −.2; −0.02; p=0.04), mucus production (coefficient −0.47; 95% CI −0.9; −0.05; p=0.03) and nocturnal cough (coefficient −0.98; 95% CI −1.2; −0.8; p<0.001).

A 15% increase (worsening) in LCI on its own had a sensitivity of 43% (17/40) and specificity of 90% (95/105) to identify a symptomatic event. The PPV of the LCI was 63% (17/27) and the negative predictive value (NPV) was 81% (95/118). For zFEVt, a significant worsening (<−1.3 z-scores) had a sensitivity of 8% (3/36) and a specificity of 87%. The PPV of zFEVt was 20% (3/15) and NPV was 71%.

For visits with a 10-point worsening in both the CFQ-R Respiratory score and 15% increase in LCI (n=13), the PPV was better than using either measure alone (81%). The NPV of combining a stable CFQ-R Respiratory score and LCI (n=66) was 86%, similar to using either measure on its own. The PPV of combining a worse CFQ-R Respiratory score and zFEVt (n=3) was low at 43% and the NPV of a stable zFEVt and CFQ-R Respiratory score (n=56) was 81%.

When both lung function measures and CFQ-R Respiratory scores were combined, three visits had a worse zFEVt, LCI and CFQ-R, all of which were classified as symptomatic.

4. Discussion

In this study, we demonstrate that the CFQ-R Respiratory score has discriminant validity in the preschool CF population, deteriorating with acute respiratory events and improving with treatment. We found that a lower CFQ-R Respiratory score at the first stable visit predicted a higher pulmonary exacerbation rate, similar to a previous report [12].

An important clinical change ranged from 7.1 to 10.4 points, which corresponds to a two-category change on one question or a one category change on two different questions. The sensitivity and specificity of a 10-point worsening in the score were low which limits the clinical usefulness of the tool to track individual patients. However, a 10-point worsening in respiratory symptoms combined with a significant worsening in LCI improved the positive predictive value and was better than using either outcome measure alone. The tobramycin inhalation solution (TIS) randomized control trials, which included patients six years and older, identified a larger MID for patients experiencing a pulmonary exacerbation (8–10 points) compared to those in a stable state (4 points) [4]. We report similar changes in the CFQ-R Respiratory score for participants when they are symptomatic (10 points) and when their symptoms resolve (10 points).

Overall, the CFQ-R Respiratory score did not change between consecutive stable visits indicating good reproducibility. However, the variability was high, two standard deviations from the mean represented a 40-point change. A 40-point threshold to define a meaningful change would not have any discriminatory value. Although, a 10-point threshold, estimated from the distribution-based method (0.5 SD), is in keeping with the clinically important change that we identified for symptomatic events.

Other studies have reported similar variability in the CFQ-R Respiratory score and there are several possible sources for this [4,9]. Firstly, respiratory infections within the two-week recall period of the CFQ-R could result in lower scores at stable visits and bigger between visit changes. Secondly, parents rather than the children completed the questionnaire. While there is good agreement between parents and children in the CFQ-R for the 6 to 11 age group, particularly for observable behaviours such as cough frequency, measuring both respondents’ perceptions is important [5]. A pictorial version of the CFQ-R for preschoolers is being developed, which would allow the children to report their own respiratory symptoms [5,20]. This might be less variable and more useful to monitor individuals in clinical practice. Thirdly, physicians categorized study visits as stable or symptomatic but there may be subtle between visit changes that are not captured by this simple classification. Fourthly, we did not exclude outliers from the analysis even though the credibility of some of the extreme scores (e.g. 100 points at symptomatic visits) was questionable, this may have contributed to the observed variability of the measure. Finally, the lack of a standardized definition for pulmonary exacerbations is a major limitation of investigating acute respiratory events in this age-group.

We described clinically important changes in the CFQ-R Respiratory score but did not define the MID. The MID of a PRO is usually investigated with interventional trials but this would have been difficult to do in this population given that they have minimal baseline respiratory symptoms. The SHIP study, a placebo-controlled randomized controlled trial that evaluated 7% hypertonic saline inhalation treatment over 48 weeks in preschool children, identified a significant treatment effect for LCI but did not see any change in CFQ-R Respiratory scores [8]. Secondary analysis of the ISIS trial demonstrated that only the treatment burden score of the CFQ-R changed with 7% hypertonic saline inhalation [9]. However, this is unsurprising given that neither the ISIS or SHIP trials demonstrated a reduction in pulmonary exacerbations with hypertonic saline inhalation.

Cough and congestion were the most frequently endorsed items in the Respiratory Domain and were the most responsive to change with pulmonary symptoms. The questions relating to the presence of wheeze, trouble-breathing and nocturnal cough were the least frequently endorsed questions and changed little with pulmonary symptoms. While the CFQ-R performs well in the preschool population with good construct validity and adequate internal consistency [9], as mentioned above, the Preschool Pictorial CFQ-R would enable children to report on their own respiratory symptoms and this would likely enhance the clinical application of this PRO [20].

There was a weak but significant correlation between the CFQ-R Respiratory score and LCI. The CFQ-R changes with acute respiratory events also mirror the LCI changes previously published from this dataset [15]. Furthermore, the positive predictive value of the CFQ-R to identify symptomatic events was improved when combined with the LCI. Previous studies assessing the relationship between the CFQ-R and LCI reported conflicting results; some interventional trials reported a significant correlation [6,22] while others did not [8,23]. Unlike previous studies in older populations [10], we did not find a significant relationship between CFQ-R and FEVt. This likely reflects the limited sensitivity of FEVt to capture pulmonary exacerbations in subjects with early lung disease.

In addition to the study limitations discussed already, ceiling effects may have impacted the results. The majority of participants were at the higher end of the scaled score, and 18% of all visits had a maximum score of 100 points. However, the proportion of visits with a CFQ-R score of 100 points was similar to that reported by the ISIS study (22%) [9] and 18% is below published lower limits of extremes [5,24,25].

In summary, we report clinical changes in the CFQ-R Respiratory score in preschool children with CF. Overall, the CFQ-R Respiratory score is responsive to acute respiratory events, but, in its current form, it is too variable to be a useful clinical tool in individual patients alone. However, the improved positive predictive value of the CFQ-R with the addition of the LCI supports the concept of using multiple outcome measures, both patient-reported and physiological, to develop a standardized definition of a pulmonary exacerbation to manage these important clinical events [26–28].

Highlights.

• The CFQ-R Respiratory score is responsive to respiratory symptoms in preschool children

• A 10-point worsening in the score from a stable visit represents a clinically important change

• The score is too variable to be used alone to track patients clinically

• The positive predictive value of the score to detect a respiratory event improved when combined with the lung clearance index

Abbreviations:

CF

cystic fibrosis

CFQ-R

Cystic Fibrosis Questionnaire – Revised

PRO

patient-reported outcome

LCI

lung clearance index

MBW

multiple breath washout

ISIS study

Infant Study of Inhaled Saline (ISIS) study