Introduction

Chronic obstructive pulmonary disease (COPD) represents a global healthcare challenge characterized by airflow limitation, systemic inflammation, and a significant impact on patients, quality of life.^1–4 As part of standard guidance, COPD diagnosis requires an abnormal post-bronchodilator spirometry result with a pre-defined ratio of forced expiratory volume in the first second (FEV₁) over forced vital capacity (FVC) of 0.7.^5–7 It has been suggested that, beyond diagnosis of COPD and other conditions, spirometry testing helps to identify overall lung health.⁸ However, the clinical spectrum of what is diagnosed in primary care as COPD is heterogenous, and some patients present with spirometry results that challenge conventional diagnostic patterns, or receive a diagnosis without any confirmatory spirometry.^9,10 In addition, COPD is often not diagnosed until a patient is hospitalised as a result of worsening of their lung condition encompassing a COPD exacerbation. A UK study identified that up to 85% of opportunities for early diagnosis were missed in Primary Care¹¹ and may include symptomatic cases where spirometry does not meet the standard for diagnosis, yet may meet criteria for preserved ratio impaired spirometry (PRISm).

PRISm, defined as an FEV₁/FVC ratio ≥0.7 combined with a reduced FEV₁ predicted value of <80%, has emerged as a spirometric pattern observed in a subset of patients with clinical features suggestive of future incident COPD.^12,13 PRISm prevalence has been shown to be around 6–20% worldwide, although geographical variation does occur.^14–20 Unlike the more typical pattern of airflow limitation characterized by a reduced FEV₁/FVC ratio, either by a fixed ratio <0.7 or <lower limit of normal (LLN), patients with PRISm present with higher or “preserved” FEV₁/FVC ratio despite demonstrable airflow impairment, raising diagnostic dilemmas and challenging traditional definitions of COPD.¹⁰ PRISm is often associated with younger age, greater BMI, and a greater frequency of comorbidities, presenting as an alternative clinical entity prior to COPD.¹³ The earlier age of appearance could present opportunities for earlier intervention, diagnosis and management.^13,21 Patients with PRISm have also been shown to have greater rates of morbidity and mortality than individuals with normal postbronchodilator spirometry, particularly related to cardiovascular events and risk.^19,22–24 PRISm is not currently recognised within the definition of COPD developed by national and international guidelines, although it has been raised for potential inclusion in future expanded definitions.⁵ This muddying of objective rules for diagnosis underscores the value of clinician decision-making in the diagnostic process. While spirometry is a cornerstone in the evaluation of COPD, additional diagnostic tools, such as lung imaging, diffusion capacity measurements, and clinical assessments, may be required to confirm the diagnosis and rule out alternative respiratory conditions.²⁵

In this study, we aim to determine the frequency and determinants of PRISm within a population of patients with diagnosed COPD served by US primary care providers, and how this subgroup compares to those whose spirometry results meet current COPD diagnostic criteria.

Methods

Design

This is a historical, observational registry study using patient data from the APEX in COPD registry. The APEX in COPD registry contains longitudinal Primary Care EHR data and a subset of patients with linked patient reported outcomes (PRO), spirometric, and CT scan data.^26–28 The study was constructed, performed, and reported in accordance with all relevant regulatory/ethical requirements, namely: the European Network Centres for Pharmacoepidemiology and Pharmacovigilance (study reference number: EUPAS29401), study registration with the European Union Electronic Register of Post-Authorization studies was also undertaken; Approval for this study was granted by the Anonymised Data Ethics Protocols and Transparency (ADEPT) committee (ADEPT0520). Central ethics (Institutional Review Board, IRB) approval was obtained from the American Academy of Family Physicians for most sites (AAFP; IRB reference number: 19–349). Additional ethics approval was obtained from one site that required its own board to approve.

Patients

Patients from participating Primary Care sites were considered eligible if they were aged 35 years or older at time of COPD diagnosis, with the presence of a diagnostic code compatible with COPD (ICD9CM or ICD10CM), or a COPD compatible review code at any time in their EHR. This included bronchitis, emphysema, α₁-antitrypsin deficiency, and mixed COPD/asthma. Patients were to be considered active if they had an appointment within the last two years, and with current COPD with either having (i) been diagnosed/re-coded with COPD in the last year or (ii) had a prescription for a COPD inhaler in the last two years and a COPD diagnosis (including diagnoses with an end date) or (iii) active patient reported COPD symptoms. According to our study protocol, explicitly investigating patients without spirometric confirmation of a COPD diagnosis, no positive result in spirometry (ie, post-BD FEV₁/FVC ratio <0.7) was necessary for inclusion. When assigning groupings for lung-function decline and GOLD status, most patients had recorded values for post-bronchodilator spirometry (n=2415, 84.3%). In instances where post-bronchodilator findings were not available, pre-bronchodilator results (n=451, 15.7%) were used. When investigating longitudinal lung function decline, pre- and post-bronchodilator values were compared against results produced in the same manner. Predicted values were calculated in line with American Thoracic Society reference equations.^29,30

Patients were excluded for any of the following criteria: (i) they were participating in a COPD drug therapy clinical trial at the time of enrollment (ii) had a life expectancy <12 months from the date of enrollment (iii) had an active cancer diagnosis in the last 3 years (excluding non-melanoma skin cancer) or (iv) were receiving palliative care.

Sites

For this study, sites within five healthcare organizations located in five US states, namely Texas, Ohio, Colorado, New York, and North Carolina, provided access to a limited EHR dataset through a data use agreement of all patients with COPD who did not specifically opt out of the project.

Freetext Analysis

Spirometry results were collected from EHR and through manual transcription from digital scans of reports. For CT scan analysis, text was scanned searching for a positive interpretation for emphysema and/or airway wall thickening. Synonyms within the search criteria for emphysema included: overinflation, bulla, low attenuation, and air sac destruction. Airway wall thickening was defined as the presence of a variety of interpretations, including: airway wall thickening, asthma, bronchitis, bronchiectasis, and presence of mucus. Responses were excluded when found to be in proximity to negative connections pre- or post-appearance, eg, “not”, “incorrectly diagnosed”, “exclude”, “nil”, “no”. There was a 5-word distance investigated pre/post-keyword and negative connections.

Statistical Analysis

We endorsed STROBE guidance for observational research, and the STROBE checklist is appended.³¹ STATA version 14 (College Station, TX, US) and R version 3.6 (Vienna, Austria) were used to conduct all statistical analyses and data handling. For the demographic, clinical, and patient reported elements of the study, descriptive statistics were calculated. All available data (non-missing) was summarized. Categorical variables were expressed as counts (%), while numerical variables were expressed as means (standard deviation).

For the exploratory analysis, multivariable logistic regression was performed. GOLD 0 was defined by the presence of symptoms of COPD, an FEV1/FVC ratio ≥0.7, and an FEV1 >80% predicted, and was used as the reference category to examine lung function decline.⁵ A threshold of lung function decline was defined as an annual decrease in FEV₁ of >140 mL. Age at first spirometric reading, gender, smoking status, and baseline FEV₁ were used as covariates. As the number of patients in severe and very severe GOLD airflow limitation were small, both were grouped; GOLD spirometric groups 1 and 2 were also combined. Statistical significance was set at p<0.05.

Results

Patient Population

A total of 2866 patients were identified to be eligible for analysis from the APEX COPD Registry for the purpose of the study (Figure 1). Additional patient-reported data was available for 1354 patients. PRISm was found to be present in 23.6% (678/2866) of patients with available spirometry. Among those meeting COPD obstruction criteria, 638 (22.3%) were categorised as GOLD 0, 211 (7.4%) were GOLD 1, 833 (29.1%) GOLD 2, 420 (14.7%) GOLD 3, and 86 (3.0%) GOLD 4 (Figure 2). In Table 1, those participants with PRISm were compared with those with obstructive COPD (post-BD ratio <0.7), and then with GOLD 0.

Table 1 Demographic Characteristics of PRISm, GOLD 0 and Obstructive COPD Patients in the US APEX COPD Registry*

Figure 1 Flowchart of stages in patient inclusion and exclusion.

Figure 2 Prevalence of PRISm, obstructive COPD (GOLD 1 to 4) and of GOLD 0 groups in the APEX COPD Registry.

Baseline Characteristics

Gender differed between populations, with the PRISm group showing more women (55.9%) compared to the COPD groups overall (46.9%, p<0.001), but fewer women than GOLD 0 population (60.5%, p=0.097) (Table 1). The age distribution was younger within the PRISm 66.3±11.1 years and GOLD 0 groups 66.3±11.5 years than in GOLD 1–4 COPD groups (69.2±10.4 p<0.001). The PRISm population had a higher mean BMI of 33.5±9.2, compared to the Obstructive COPD (GOLD 1 to 4) and GOLD 0 groups with a mean BMI of 27.8±7.2 and 30.6±7.8 respectively (both p<0.001). In examining ethnicity, while all groups had a majority of Caucasian individuals, the PRISm group had a similar percentage of Caucasians (62.8%) compared to the GOLD 0 (63.9%) (p=0.611), and significantly lower to the Obstructive COPD group (73.7%, p<0.001). Finally, patients in the Obstructive COPD group were more likely to be current or ex-smokers (93.0%), with PRISm having a much larger proportion of never smokers (17.0%) (all P<0.001).

Clinical Characteristics

All spirometry comparisons between PRISm and the other groups were highly significant, as grouping was according to spirometry thresholds. Overall, patients with COPD were more likely to have experienced an exacerbation in the previous year (49.8% vs 43.6 in PRISm and 39.5% in GOLD 0, both p<0.001) (Table 2). On symptomology, Obstructive COPD patients were more likely to have the presence of Chronic Bronchitis registered in the EHRs (63.1) than PRISm (56.3%) or GOLD 0 (44.4%) patients (all p<0.001), but interestingly PRISm had similar COPD Assessment Test scores ≥10 to the Obstructive COPD group but a higher proportion of modified Medical Research Council (mMRC) dyspnea scores ≥2 (p<0.001).

Table 2 Clinical Characteristics of PRISm, GOLD 0 and Obstructive COPD Patients in the US APEX COPD Registry*

CT Scans

CT scan data was available for 54.4% of PRISm patients, 53.3% of GOLD 0 patients and 51.4% of Obstructive COPD. Analysis of free text notes in CT scans revealed 70.5% of PRISm patients with available CT scans had a positive interpretation of emphysema in their notes (Table 3). This percentage was lower in the GOLD 0 population, but still substantial, where 64.1% also presented emphysema in some form or severity. Airway wall thickening was present in 34.4% of PRISm patients greater than the 25.9% of GOLD 0 patients, with more frequent bronchiectasis in PRISm group (32.0% vs 22.9%) (p<0.001). The GOLD 0 subgroup had a higher percentage of patients with a CT, but the PRISm subgroup had more CTs per patient (data not shown). Of note, the prevalence of both bronchiectasis and airway thickening was higher in PRISm than GOLD0, and quite similar to those with Obstructive COPD (Table 3), a finding that highlights the clinical relevance of PRISm.

Table 3 Interpretation of CT Scans for Evidence of Lung Structural Change

Lung Function Decline

The risk of decreasing lung function in PRISm and GOLD severity groups was analysed using multivariate logistic regression. PRISm (p=0.792) and other stages were not proven to be significant predictors of >140mL decrease in FEV₁ over time (Table 4).

Table 4 Logistic Regression Model for FEV1 Progression in PRISm and Obstructive COPD Populations

Discussion

Summary of Results

To improve the diagnosis and management of COPD within Primary Care settings, it is essential to gain insights into the patient population investigated. Our study describes the burden of PRISm within a well-defined US cohort of diagnosed COPD patients who had spirometry assessment, with enhanced supplementary data of PRO, longitudinal spirometry, and CT scan imaging with text interpretation. PRISm represents a substantial subgroup (23.6%) of nearly one in four patients who were diagnosed with COPD despite not meeting current COPD diagnostic criteria. (GOLD 2023). While not meeting obstruction criteria, these patients present an opportunity for early recognition of respiratory symptom burden and management potentially at a younger age at presentation) compared to the obstructive COPD populations.

Under international standards for COPD concordant spirometry, those in our study who are allocated to the PRISm subgroup would be considered a COPD misdiagnosis. The differences in smoking status, lung function, and symptomatology are crucial factors in the recognition of early identification of chronic lung disease in the absence of confirmatory spirometry. This is demonstrated in the current study by the differences in symptoms between the groups. These are notable and provide valuable insights into the varying clinical profiles. Individuals with obstructive COPD, particularly in the Obstructive COPD (Ratio <0.7) and those with moderate COPD (Ratio <0.7 and %pred FEV₁ <80%) groups, exhibit a higher prevalence of chronic bronchitis, indicating an advanced disease along the COPD severity progression. However, a greater percentage of individuals in the PRISm groups report high symptom burden, as evidenced by a CAT (COPD Assessment Test) score of ≥10, which represents an effect on activities of daily-life, similar to those with established Obstructive COPD and significantly greater mMRC, and therefore a recommendation for further management. The finding that a large proportion of PRISm patients have an mMRC of ≥2, reflects a high degree of dyspnea and functional impairment. This is compounded in the investigation of CT scans, where ~70% of PRISm and GOLD 0 patients with CTs have shown to have a positive interpretation of emphysema. However, this may be skewed higher due to confirmation bias, where arrangement of tests is clinician-driven, and is a result of recognition when the wider symptomology and supplementary tests suggest the presence of COPD without a confirmatory spirometry. These findings suggest that individuals with obstructive COPD may have reduced lung function compared to those with PRISm. However, they may not experience more severe and impactful respiratory symptoms, possibly due to receiving better management as a result of a confirmed diagnosis of COPD due to impaired airflow limitation. This finding highlights the clinical significance of COPD-related symptoms at the point of clinician-patient interaction. Interestingly, PRISm individuals had more dyspnoea with mMRC≥2 yet with less chronic bronchitis than in obstructive COPD, while the rate of CAT≥10 did not differ (Table 2); this apparent clinical inconsistency might be due to imbalance in age, gender and other respiratory determinants, or because of small sample sizes.

For individuals with PRISm, emphasis might be placed on preventive measures. This could include smoking cessation interventions, vaccinations, and lifestyle modifications to reduce the risk of developing COPD or exacerbations which could lead to further loss of lung function in the future. In contrast, for those with obstructive COPD, clinical decisions would likely revolve around optimizing disease management. Earlier recognition and diagnosis of COPD can allow for increased, personalization of individualized care plans considering comorbidities, age, and overall health status. Regular monitoring of lung function and symptom assessment would be integral to gauge treatment efficacy and adjust therapeutic strategies as needed.

Strengths of this study include its basis in primary care, longitudinal assessment and the single, protocolised assessment of all measurements and relevant PROs, exploring a well-studied COPD population. However, some limitations are worth considering. Overall, there was a lack of consistent, annual spirometry in the registry; however, this is representative of the need for regular scheduled tests and digitization of lung function reports into clinical systems. Similar issues occur with CT scans, where greater insight for management of disease could be provided to clinicians should results be clearly presented in the EHR. Perhaps, the very often used threshold of a lung decline of 140 mL in FEV₁ per year could be challenged as arbitrary for progression within COPD severities; within our multivariate analyses those with moderate COPD were the closest to reach significance, as likely those with severe or very severe airflow limitation have no room to reach such a threshold, and later they are censored or right truncated to perform a valid spirometry. Should this be further investigated, it would be prescient to conduct a sensitivity analysis to explore a wider range of cut-offs to define annual lung function decline. Finally, real-world registry data is inherently lower in validity compared to clinical trial data, which would be more complete as part of the trial design and variable selection. Inherently, patients with higher severity of disease are more likely to have PFTs or imaging performed and recorded due to being further ahead in the progression of their disease. Imaging bias is likely mitigated by the fact that the majority of the CT scans were performed for lung cancer screening and not as part of a respiratory evaluation. To conduct an automatic quantification of emphysema or of airway wall thickening by AI or else was beyond the scope of our research. And we must insist as a strength of our study the free-text review of EHRs and radiologist reports. However, future research might include the search for radiologic abnormalities in PRISm using quantitative or parametric response mapping CTs.³⁰ All of these limitations could be considered representative of real-world evidence in US primary care, collected over numerous sites and regions. The sheer size of the APEX dataset of >17,000 patients lends to its strength of being generalizable to the entire US population.

When spirometry results are inconclusive, clinicians can take a wider and more varied approach to a diagnosis of COPD. This includes considering the patient’s demographic and clinical situation, assessing spirometry metrics beyond just FEV₁/FVC, arranging imaging and interpretation, evaluating symptoms using standardized tools, and inquiring about exacerbation history.

These insights also enable clinicians to provide more informed management decisions, such as lifestyle modifications, pharmacological options, pulmonary rehabilitation, and perhaps earlier vaccination with RSV, pneumococcus and else for younger than 65-yr individuals, to meet the specific needs of each patient. Giving strength to clinician decision-making must be reinforced with further tests, which become crucial for clarifying COPD diagnosis and management in cases where spirometry alone may not provide a clear diagnosis.

Conclusion

This study provides evidence that one in four patients currently diagnosed with COPD in Primary Care, actually fulfil the spirometric criteria of PRISm rather than COPD, with a symptom burden and rates of imaging of emphysema similar to those with established spirometric confirmed COPD.