Home spirometry was implemented at our center in 2020. Spirometry data were extracted from the AirNext portable home spirometer (NuvoAir). As per our local clinical care pathway, we routinely review people with CF at least every 3 months, and they are expected to perform spirometry prior to clinic review and during periods of increasing clinical symptoms. Hence, we expect them to perform around 4 to 6 spirometry sessions per year. The home spirometry system comprises 3 components: the handheld AirNext device, a patient-facing smartphone app, and a clinician portal that provides centralized access to spirometry data. The device connects to the patient’s smartphone via Bluetooth, with all spirometry data automatically uploaded to both the app and clinician portal.

All people with CF received initial training during rollout and were instructed to perform at least 3 maneuvers per session. The NuvoAir algorithm selects the highest-quality reading in accordance with American Thoracic Society and European Respiratory Society guidelines and uploads this result to the platform [18]. Patients are expected to perform spirometry prior to all clinic visits (virtual or in-person). If unavailable, spirometry is either performed during the clinic or requested afterward. However, no formal follow-up is undertaken unless there is clinical concern. Only spirometry sessions conducted in 2024 were included, either performed at home or as part of routine clinic reviews. Sessions performed in the context of clinical trials, nebulized antibiotic or mucolytic challenges were excluded. Serial spirometry sessions performed during inpatient admissions, home intravenous antibiotic therapy, or both, were counted as a single encounter to prevent possible bias due to an awareness of individuals who more frequently exacerbate. Hospital records for all people with CF included in the study were cross-referenced with spirometry records to ensure these criteria were met.

All statistical analyses and visualizations were performed using RStudio (version 2025.05.0+496; Posit PBC). Bias was defined as the difference between the estimated and actual number of spirometry sessions (estimate−actual). A positive bias was reported as “overestimate,” that is, the estimated number of spirometry sessions was higher than the actual number; a zero bias was reported as “correct,” that is, the estimated number of spirometry sessions equaled the actual number; and a negative bias was reported as “underestimate,” that is, the estimated number of spirometry sessions was less than the actual number. While bias indicates the direction (negative or positive), MAE indicates the precision of the estimates. MAE is defined as the average of the absolute differences between estimates and actual values. Comparative analyses were conducted by HCP role (physicians, nurses, and physiotherapists) and familiarity level (level 2 vs level 3). Statistical tests included the chi-square test of independence (HCP role vs categorical estimation of spirometry use), Wilcoxon rank-sum test (level of familiarity vs mean bias or MAE), Kruskal-Wallis test (HCP role vs mean bias or MAE), and Spearman rank correlation (numerical estimation of spirometry use vs actual spirometry use and CF years of experience vs mean bias or MAE). No weighting was applied to account for differences in the numbers of respondents across HCP groups, as between-group comparisons were based on standardized metrics (proportions, mean bias, and MAE) that inherently account for differences in group size.

Ethics approval was obtained from the North West–Greater Manchester East Research Ethics Committee (IRAS 340274). The survey participants were health care providers from our CF center, and participation in the study was voluntary, with no compensation provided. Participants were informed that completion and return of the paper survey would be taken as implied consent to participate, while nonreturn of the survey was considered a decision not to participate. Apart from their clinical role, no other identifiable data were collected from the participants. Regarding patient data, this was a noninterventional study, and the authors were part of the direct clinical care team for the patients. Therefore, additional patient consent was not sought. All patient data were obtained from routine clinical records and pseudonymized prior to analysis.

The survey achieved a response rate of 96.6%, including 28 of the 29 HCPs invited to participate. Of the 28 respondents, 10 (100%) were physicians, 6 (85.7%) were nurses, and 12 (100%) were physiotherapists. Analyses were restricted to responses where participants rated their familiarity with the patient as level 2 or level 3, indicating sufficient clinical knowledge to recall patient history. Physicians reported familiarity with 16.7% to 100% of patients, nurses with 12.5% to 77.1%, and physiotherapists with 10.4% to 79.2%. All 48 patients were familiar to at least 1 HCP.

A total of 790 categorical responses were analyzed. HCPs frequently incorrectly estimated spirometry use among people with CF: 332 (42%) responses underestimated use, 96 (12.2%) responses overestimated, and 362 (45.8%) responses were correct. HCPs who rated themselves as “very familiar” with the individual (level 3) were no more likely to estimate correctly than those who were “somewhat familiar” (level 2; P=.77). There was also little difference in accuracy between the different HCP roles, as shown in Figure 1. Nurses had the highest rate of correct estimation (67/136, 49.3%), while physiotherapists had the lowest (144/336, 42.9%), along with the highest rates of both underestimation (146/336, 43.5%) and overestimation (46/336, 13.7%). There was no statistically significant association between HCP role and direction of estimation (P=.61). Table 1 shows the detailed breakdown by use group (infrequent, expected, or highly frequent users) against the estimation accuracy (underestimation, correct estimation, and overestimation).

There were 794 completed responses for the numerical estimation of spirometry frequency. We assessed accuracy using mean bias and MAE. Overall, the mean bias was −4.91 (SD 8.9) and the MAE was 6.32 (SD 8.1), indicating a consistent trend toward underestimation and moderate variability in responses. Figure 2 shows a boxplot comparing the mean biases across HCP roles, ranging from −4.50 (SD 8.6) to −5.13 (SD 8.9). MAEs were similar across roles, ranging from 6.13 (SD 7.5) to 6.47 (SD 8.6). This suggested that all HCP roles underestimated spirometry use with similar imprecision, with no significant differences between roles (P=.86 for mean bias; P=.78 for MAE). We observed that the numerical estimation accuracy (defined by bias rather than exact estimate) declined with increasing actual spirometry use, with a strong negative Spearman rank correlation (ρ=−0.88; P<.001; Figure 3). Additionally, HCPs tended to cluster their estimates within a narrow range, with a median of 4 (IQR 1-5) spirometry sessions per year across all people with CF, and similar medians were observed across all HCP roles (Figure 4). Similar to categorical estimation, there were no significant differences in bias or MAE by familiarity level (P=.89 and P=.30, respectively).

The mean duration of HCPs’ CF experience, expressed as years of working at the center, was 15.7 (SD 8.2) years, ranging from 2 to 31 years (mean 14.8, SD 7.8 years among physicians, mean 17.9, SD 7.2 years among nurses, and mean 15.7, SD 8.7 years among physiotherapists). There was no significant association between years of CF experience and mean bias or MAE (P=.58 and P=.81, respectively).

Our findings demonstrate that CF HCPs were generally poor at accurately predicting home spirometry use among people with CF. First, our study highlights a tendency to underestimate the number of spirometry sessions performed by people with CF attending the center. This underestimation occurred across all roles and regardless of HCPs’ self-reported familiarity with people with CF attending the center. Second, CF experience, defined as the number of years an HCP has worked at the CF center, did not show any strong signal in predicting home spirometry use. Both findings suggest that perceived clinical insight may not reliably translate to accurate evaluation of remote monitoring behavior. This is an important observation, particularly as home monitoring becomes more integrated into clinical care and models of care continue to evolve.

To our knowledge, no previous study has evaluated whether CF HCPs can accurately estimate people with CF’s engagement with home spirometry using objective data. Previous studies involving clinician-reported estimates have primarily focused on treatment adherence rather than disease monitoring. For example, Daniels et al [19] reported that HCPs tended to overestimate adherence to nebulized therapy when compared with electronic monitoring data. Although home spirometry and nebulized treatment represent distinct health behaviors, both are performed unsupervised at home and depend on patient-initiated engagement. It is therefore interesting that previous work reported clinician overestimation of nebulized treatment adherence, whereas our study found underestimation of home spirometry use. This contrast should be interpreted cautiously, but it suggests that clinicians may perceive treatment adherence and disease monitoring differently.

In our center, we specifically evaluated CF HCPs who were directly involved in home spirometry, namely physicians, nurses, and physiotherapists. This role may differ from other centers. While we acknowledge that the nature and extent of input on home spirometry may vary among HCPs, it is important to note that all HCPs in our study showed a similar directional pattern of underestimation in their predictions. This consistency highlights the need to improve understanding of home spirometry across all professional roles.

Our findings suggest that HCPs are better at identifying people with CF who underuse home spirometry than those who overuse it. Figure 3 shows the inverse relationship between the actual number of spirometry sessions and estimation accuracy. The more frequently an individual performed spirometry, the more likely CF caregivers were to underestimate it. While recognizing people with CF who are not engaging with home spirometry is important, it is equally crucial to identify those who tend to use it more frequently and to understand the underlying drivers of this behavior. Anxiety around home spirometry may be a contributing factor. In a survey of 90 people with CF, 83% reported experiencing at least mild anxiety related to home spirometry [20]. It is possible that HCPs’ perceptions of spirometry as an anxiety-inducing procedure contribute to their tendency to underestimate use among people with CF, although this remains speculative. Interpretation of home spirometry during periods of clinical stability can also be challenging, as lung function may vary substantially (around 15% FEV₁) without leading to pulmonary exacerbation [21]. This creates uncertainty for some people with CF about how to interpret and respond to home spirometry results. In a mixed methods study of home monitoring in adults with CF in the United Kingdom, although many participants found monitoring reassuring and empowering, a minority found frequent monitoring more challenging, describing feeling “a bit more anxious or a bit more worried,” “a bit more paranoid,” or that spirometry became “like an obsession that you want to know all the time” [8]. These observations suggest that high-frequency home spirometry use should not simply be viewed as “good adherence” but may represent different monitoring phenotypes. This concept of different monitoring phenotypes is supported by a mixed methods study by Oppelaar et al [6], where they described several patterns of engagement with remote monitoring: people with CF who never initiated monitoring, those who discontinued because of technical or psychosocial barriers, those who used monitoring only when clinically indicated, and those who engaged regularly with the program. We similarly observed marked heterogeneity in spirometry behaviors within our cohort despite protocolized recommendations at our center. The rationale for use may be a contributing factor to the observed heterogeneity. Tan et al [15] found that 48% of children with CF performed home spirometry as part of routine follow-up, whereas 33% performed spirometry in response to illness-related telephone contact with their CF team.

Exploring the reasons for high-frequency spirometry use was not part of the aim of this study. This may reflect proactive self-management or fluctuating respiratory symptoms, which carry a psychological burden associated with symptom monitoring. Understanding these factors is important to mitigate potential unintended risks of spirometry monitoring. Further mixed methods or qualitative research is needed to explore these behaviors in greater detail.

We recognize several limitations in this study. The survey design may have inadvertently introduced anchoring bias. Participants were asked to provide both a categorical and a numerical estimate of spirometry use on the same form. This may have led some HCPs to default to the values provided in the survey (ie, 0, 1, 4, 6, or 10). Furthermore, the clustering of estimates around 4 sessions per year, with a narrow IQR of 1 to 5, suggests a central tendency bias, where HCPs tended to avoid selecting extreme values. This pattern may reflect normative expectations overriding clinical recall, indicating that estimates may be influenced more by perceived care-pathway policy than by recollection of individual patient behavior. The categorical question presented HCPs with 3 options: infrequent users (0‐1 sessions/year), expected users (4‐6 sessions/year), and highly frequent users (≥10 sessions/year). By design, this limited their ability to underestimate infrequent users or overestimate highly frequent users. To address this, we included an open-ended numerical estimation section, allowing HCPs to provide a more nuanced answer regarding the spirometry frequency estimates. Grouping patients into 3 categories (infrequent, expected, and highly frequent users) was a deliberate simplification to facilitate clinical recall and does not reflect the true distribution of spirometry use within our cohort. For context, 15.6% (73/469) of patients completed 0‐1 sessions, 27.1% (127/469) completed 4‐6 sessions, and 21.3% (100/469) completed ≥10 sessions in 2024 (these figures represent raw spirometry frequency and were not adjusted to our study definition of a single spirometry session—refer to the Methods section). This sampling strategy may overrepresent behavioral extremes and exaggerate bias. Results should therefore be interpreted as HCPs’ ability to discriminate between predefined groups rather than as estimates of population-level predictive accuracy. Clinical recall and familiarity with an individual with CF are inherently subjective and can vary significantly. An HCP may be familiar with certain aspects of their care but not with their home spirometry behavior. Additionally, people with CF at either end of the clinic attendance spectrum (those who attend very frequently or very rarely) may stand out more in an HCP’s memory, potentially influencing estimation accuracy. HCPs in our study also work across varied clinical settings, including outpatient, inpatient, and home-based care. HCP familiarity with a given individual is often shaped by their clinical needs. For example, those working primarily on the wards may be more familiar with people with CF requiring frequent admissions. To provide a fair representation of clinical familiarity, we included only people with CF for whom the HCP reported a familiarity score of 2 or higher on the survey’s familiarity scale. Finally, each HCP provided multiple estimates across patients, and these were analyzed independently. Our data were not adjusted to account for within-HCP clustering. The results should be interpreted as exploratory and descriptive.

As CF care increasingly incorporates remote monitoring into everyday practice, it is essential that we understand how people with CF are actually using these tools. At its core, home spirometry promotes autonomy and self-directed care in people with CF, but it carries the risk of shifting the burden from the disease itself to the act of monitoring. Therefore, understanding individual behavior around remote monitoring and being able to identify both underusers and overusers, along with the triggers behind these patterns, are crucial. This study identified that HCPs’ estimation of home spirometry use was exceptionally poor in people with CF who undertook frequent spirometry, that is, more than expected as per our local standards. Exploring the reasons for high-frequency spirometry use was not part of the aim of this study. This may reflect proactive self-management or fluctuating respiratory symptoms, which carry a psychological burden associated with symptom monitoring. Understanding these factors is important to mitigate potential unintended risks of spirometry monitoring. Further mixed methods or qualitative research is needed to explore these behaviors in greater detail.

Errors in estimating spirometry frequency may have important clinical implications. Inaccurate assumptions about spirometry use could inadvertently affect prioritization of health care resources, introduce cognitive biases in the interpretation of health behaviors, and influence how clinical conversations are framed, as well as subsequent decisions regarding patient follow-up. Given the inaccuracy of clinical estimates, more robust system-wide approaches to presenting objective adherence data should be adopted, such as automated spirometry alerts integrated into health care records.

This study highlights a systematic underestimation of spirometry use by CF HCPs in people with CF, regardless of their professional role, perceived familiarity with the individual, or level of CF experience (measured by years working at a CF center). Our findings suggest that clinical intuition alone may not reliably reflect how adults with CF engage with home spirometry in real-world settings. As hybrid models of care increasingly incorporate remote monitoring and virtual consultations, an accurate understanding of patient engagement with home spirometry becomes increasingly important. The discrepancy between HCPs’ perception and objectively recorded spirometry behavior highlights an important gap in our understanding of remote monitoring engagement. Further qualitative and mixed methods research involving both people with CF and HCPs is needed to better understand the motivations, barriers, and different phenotypes of home spirometry use and how these behaviors can be appropriately identified and supported.