To the best of our knowledge, this is the first ever clinical audit of primary clinical care delivery to COPD patients in Spain. The results of this study indicate that there is considerable variability in clinical performance, not completely attributable to the severity of the disease. Identification of the determinants of this variability will help us understand clinical behavior, and establish strategies to strengthen clinical practice for COPD in primary care.

Measurement of the adequacy of clinical practice in a designated health care setting and temporal context is key for clinical care improvement. In this regard, clinical audits provide an innovative tool to assess the quality of care, playing a significant role in highlighting the gaps between the recommended practices and the care that patients actually receive [1]. Accordingly, audits and feedback have shown to improve health care for different disease conditions [2].

Chronic obstructive pulmonary disease (COPD) is a chronic respiratory disease with extrapulmonary implications that poses a major burden for the patient and the health system [3, 4], with a high prevalence [5], morbidity and mortality [6] rate. Hence, the care for these patients should engage the highest quality standards due to its potential impact on the lives of patients, and relatives, alongside the strain on resources. Therefore, COPD is one of the diseases in which clinical audits are deemed to be of special relevance.

Until the last few years, clinical audits for COPD were not frequently carried out. Over the last several decades, the United Kingdom [7], followed by Spain [8, 9], have been leading the audit process for COPD in Europe. Additionally, several countries have recently started their own audit projects [10–12], like the recent European Clinical COPD Audit in 13 European countries [13, 14]. In Spain, the so-called AUDIPOC network evaluated clinical care in patients hospitalized due to COPD [15] and recent initiatives have explored the clinical performances in specialized respiratory outpatient clinics [16]. These audits have provided valuable information about medical interventions in hospital wards for patients admitted with COPD exacerbation [14], the resources available [17], and the interrelationship between resources and clinical practice [15, 16]. Although not formally labelled as clinical audits, previous analyses have been done describing different aspects of clinical performance in primary care using different methodological approaches [18–20].

The information regarding how COPD patients are treated in the primary care setting is essential and would provide very relevant information on the process of care. It might also reveal the key areas wherein improvements are required in order to complete the picture obtained in secondary care audits. Therefore, formal clinical audits in primary care are emerging to evaluate clinical performance in this setting using a standardized methodology (https://www.rcplondon.ac.uk/COPD). Based on our previous audit experiences, we have developed a clinical audit for COPD patients in the primary care setting in Spain. In this manuscript, we describe the full methodology and main results of this clinical audit regarding guideline adherence in primary care centers (PCC) in Spain. The results of this audit will serve as a basis for an improvement in health care delivery for COPD patients.

The Community Assessment of COPD Health Care (COACH) study was an observational, cross-sectional, multicenter, national, retrospective, and non-interventional clinical audit aimed at evaluating the clinical care delivered to COPD patients in randomly selected primary care centers in Spain. This is the result of a joint project between the Spanish Primary Care Respiratory Group (GRAP) and the Spanish Society of Pneumology and Thoracic Surgery (SEPAR). The enrolment of participants and data collection was performed retrospectively.

During the study, 4307 cases from 63 PCC in 6 regions of the country were finally audited (Fig. 1). The description is summarized in Table 1. This was a cohort of COPD patients, with male gender predominance, in the 7th decade of life, with a considerable number of patients being active smokers and having moderate lung function impairment. Overall, there was a considerable variability among the different regions of the country and PCCs, which was significant for most variables. Interestingly, the variability of forced expiratory volume in 1 s (FEV₁) was not significant.

Variables	Average (patient level)	Inter-center range		Inter-regional range
		Range	P value a	Range	P value a
Male gender (n)	3159 (73.3)	35.7–93.8	<  0.001	67.9–78.1	<  0.001
Age (years)	71.1 (12.7)	61.9–80.2	<  0.001	67.2–73.3	<  0.001
Rural communities	1771 (41.1)	–	–	5.8–74.6	<  0.001
Active smoker (n)	1152 (26.7)	4.8–80.8	<  0.001	20.7–41.9	<  0.001
Ex-smokers (n)	1127 (26.2)	6.1–40.0	<  0.001	0.0–76.5	<  0.001
Life-long never smokers (n)	579 (13.4)	0.0–84.6	<  0.001	8.3–19.0	<  0.001
Previous smoking status unknown (n)	1449 (33.6)	16.8–51.6	<  0.001	2.2–88.1	<  0.001
Tobacco history (pack-years)	44.2 (48.6)	9.0–100	<  0.001	29.9–48.8	0.045
Comorbidities (Charlson index)	2.2 (1.4)	1.5–3.2	<  0.001	1.9–2.4	<  0.001
Comorbidities (COTE index)	1.1 (1.8)	0.5–1.8	<  0.001	0.8–1.4	0.003
Comorbidities: cardiac b (n)	1280 (29.7)	13.3–63.6	<  0.001	26.5–32.5	0.076
Comorbidities: vascular c (n)	913 (21.2)	0–40.0	<  0.001	15.7–24.9	<  0.001
Comorbidities: neoplasms (n)	476 (11.1)	0–31.3	<  0.001	2.9–15.4	<  0.001
Comorbidities: sleep apnea (n)	201 (4.7)	0–20.3	<  0.001	0–7.8	<  0.001
Comorbidities: eyedrops use (n)	54 (1.3)	0–18.2	<  0.001	0.9–2.8	0.018
Comorbidities: psychiatric drugs e (n)	501 (11.6)	0–38.2	<  0.001	6.8–15.8	<  0.001
Comorbidities: prostatic hyperplasia (n)	725 (16.8)	0–33.3	<  0.001	8.8–21.7	<  0.001
Body mass index (kg/m2)	29.2 (5.5)	26.7–37.1	<  0.001	28.4–30.9	0.004
Current FEV1 (mL) d	1690 (694)	810–2720	0.115	1640 – 1794	0.744
Current FEV1 (%) d	64.3 (22.8)	39.0–97.0	0.307	21.0–25.7	0.136
Current FEV1/FVC (%) d	62.3 (13.2)	69.0–51.0	0.741	60.2–65.1	0.249

Data are expressed as mean (standard deviation) or absolute (relative) frequencies, as needed

FEV₁ forced expiratory volume in the first second

^a Calculated by Chi-squared test or ANOVA to test for variability

^b Includes coronary artery disease, heart failure, and atrial fibrillation or flutter

^c Includes coronary artery disease, peripheral vascular disease, and cerebrovascular disease

^d Obtained from post-bronchodilator spirometry, and if not available, from pre-bronchodilator spirometry

^e Includes anxiolytics, antidepressants, and antipsychotics

The most frequently registered non-pharmacological therapies were vaccinations. Interestingly, current smoking habit was registered in 43.5% of audits, similar to the value for registered exercise. Three other key aspects of non-pharmacological evaluation, namely treatment adherence, adverse effects, and inhaler satisfaction, were rarely registered.

Regarding the complementary tests, spirometry was performed in a minority of cases. Alpha 1-antitrypsin determination was the only item showing no significant variability among regions, and this was anecdotal. Other tests and assessments including computed tomography scan and sputum eosinophils, as well as health-related quality of life, were more rarely obtained from the medical records (data not shown).

The use of current therapies for COPD is summarized in Table 3. The most frequently prescribed non-pharmacological treatment was influenza vaccination. Only in 24.4% of audits was it registered that the patients had quitted smoking. The distribution of pharmacological treatment is shown in Fig. 2. Triple therapy, followed by the combined use of an ICS and a LABD, was most commonly used during the audit period. In the whole cohort, 40 (0.9%) cases were receiving non-combined ICS-LAMA therapy. GOLD 2017 patient types were identified in 436 (10.1%) cases, and the distribution of inhaled therapies in this group of patients is shown in Fig. 3. Altogether, the treatments including an ICS by GOLD 2017 patient types were: GOLD A 50 (25.3%), GOLD B 47 (54.7%), GOLD C 35 (53.0%), and GOLD D 57 (66.3%). The percentage of GOLD 2017 patient types who did not receive any maintenance inhaled therapies was: GOLD A 31.3%, GOLD B 8.1%, GOLD C 9,1%, GOLD D 7.0%.

Variables	Average (Patient level)	Inter-center range		Inter-regional range
		Range	P value a	Range	P value a
Non-pharmacological treatments
Recommendations about not smoking	1051 (24.4)	0–76.5	<  0.001	6.1–36.4	<  0.001
Perform some exercise	958 (22.2)	0–84.7	<  0.001	3.2–41.8	<  0.001
Influenza vaccination administered	2869 (66.6)	34.1–95.2	<  0.001	53.4–73.3	<  0.001
Pneumococcal vaccination administered	1620 (37.6)	0–82.3	<  0.001	10.8–65.2	<  0.001
Inhaled maintenance therapies
No inhaled treatment/not available	1446 (33.6)	8.8–54.9	<  0.001	26.5–42.9	<  0.001
One long-acting bronchodilator	653 (15.2)	0–33.3	<  0.001	11.3–18.9	<  0.001
LAMA + LABA b	392 (9.1)	0–35.3	<  0.001	6.9–10.0	0.025
ICS alone	67 (1.6)	0–11.1	0.018	0.6–3.4	0.002
ICS + one long-acting bronchodilator	804 (18.7)	0–38.2	<  0.001	14.8–25.4	<  0.001
Triple therapy	945 (21.9)	0–50.0	<  0.001	18.9–24.9	0.002
ICS-containing regimens	1816 (42.2)	12.5–66.3	<  0.001	38.6–51.1	<  0.001
Incorrect prescription c	224 (5.2)	0–14.7	<  0.001	3.6–6.9	<  0.001
Oral maintenance therapies
Roflumilast	58 (1.3)	0–7.7	<  0.001	1.0–1.7	0.693
Mucolytics	96 (2.2)	0–26.7	<  0.001	0.2–4.0	<  0.001
Antibiotics	29 (0.7)	0–6,7	<  0.001	0.2–1.3	0.022
Methylxanthines	97 (2.3)	0–17.6	<  0.001	0.2–6.2	<  0.001
Home-based therapies
Long-term oxygen therapy	175 (4.1)	0–18.8	<  0.001	2.3–5.4	0.029
Home mechanical ventilation	35 (0.8)	0–10.0	<  0.001	0.1–1.9	<  0.001
Nebulized therapy	121 (2.8)	0–22.4	<  0.001	0.1–6.4	<  0.001

Data are expressed as absolute (relative) frequencies in reference to the whole cohort unless otherwise specified

LABA long-active ß₂ agonist, LAMA long-acting muscarinic antagonist, ICS inhaled corticosteroids

^a Calculated by Chi-squared test or ANOVA to test for variability

^b Combined in one single inhaler or not. Only indacaterol-glycopyrronium combination was available in the country at the time of the audit

^c Per the protocol, the following scenarios were deemed to represent incorrect prescriptions: GOLD 2017 B-D patients without any medication registered, any case of ICS use alone, and those prescriptions that duplicated drugs in combined or single therapies

The results of this study indicate that there is considerable variability in clinical performance that cannot completely be attributed to the severity of the disease. Most evaluated parameters were judged to fall under inadequate performance, except two (i.e., registration of influenza vaccination, and registration of exacerbations in the previous years) that were classified as excellent.

There is increasing evidence and awareness that patients with various health problems do not consistently receive the recommended care despite the multitude of clinical practice guidelines and there is a gap between the clinical care provided and that recommended by practice guidelines [29]. In this regard, clinical audits constitute a valuable tool to assess clinical performance and set the basis for future improvement. The present study represents an innovative initiative to perform a national clinical audit of COPD care in primary care in Spain, as in other countries (https://www.rcplondon.ac.uk/COPD).

The diagnosis of COPD is of utmost importance. The latest version of the GOLD criteria requires three main components for a correct diagnosis including respiratory symptoms, previous exposure to noxious gases and particles, and obstruction on post-bronchodilator spirometry. Our results showed that in the primary care setting, some of these three criteria were missing in a high proportion of patients. Even if older versions of the GOLD criteria are applied (which do not categorically state the requirement of symptoms as a diagnostic sub-criterion) and the results of pre-bronchodilator spirometry in the absence of post-bronchodilator spirometry results are used, the number of acceptably diagnosed cases would still be low, only on the basis of previous exposure and the presence of obstruction. In this context, the availability and use of spirometry in primary care is essential. In the UK, a recent analysis pointed out that although the quality of spirometry procedure undertaken in primary care is high, this does not reflect the quality of interpretation, suggesting an unmet training target in primary care [19]. In Spain, although most centers received training for conducting spirometry tests, this was routinely done in less than 40% of centers and the mean number of spirometries conducted per week in the primary care settings was 5.6 (ranging from 2.0 to 8.8 depending on the region) [30]. In the US, spirometry is infrequently assessed even in those more severe patients receiving triple therapy [31]. Interestingly, the performance is better in other countries [18]. The consequences of this incorrect diagnosis impact several domains, with many patients receiving pharmacological treatment that is not needed and a potential for drug-related adverse effects, giving health services to wrong patients, subjecting patients to additional and often unnecessary tests, labelling them as sick or at-risk, telling them to modify their daily living habits, or insisting on monitoring them regularly. Additionally, it also impacts the health system, leading to potential extra costs [32]. Consequently, in the light of our results, the clinician should be aware of the diagnostic criteria for COPD and consider respiratory symptoms, exposure to risk factors and spirometry results before firmly establishing this diagnosis.

In 33.6% of cases, there was no information on maintenance inhaled therapies. This finding has been previously reported but with a considerable difference. A recent analysis of the medical records of 3376 patients from a general practice in Denmark revealed that 74.4% of them did not receive any maintenance inhaled medication even after the spirometric diagnosis of COPD [33]. In the United Kingdom 28% of the 20,154 patients whose medical records were analyzed [20], received no initial pharmacological treatment. The number was reported to be as low as 20% in another study, which looked into the records of of 29,815 patients [34]. In Sweden, this figure was far below 10% in secondary care [35]. In the United States, it has been reported that 55% of such patients did not receive inhaled maintenance therapy [36]. The reasons for this finding fall outside the scope of the present study and would require a specific debate and an ad-hoc investigation.

Double bronchodilation was probably underestimated during the period of the audit, as double bronchodilator therapy was not widely available at the time. Only one (indacaterol-glycopyrronium) of the 4 double bronchodilator medications later approved in Europe for the treatment of COPD [37] was available in Spain at that time. In spite of this, the number of patients receiving LABA-LAMA combination therapies was slightly higher than those on single therapies, suggesting that the use of LABA-LAMA combinations may have continued to increase in subsequent years. A fact which was confirmed in subsequent audits [16].

Previous studies have reported the overuse of ICS, as well as an increase in their use for all GOLD patient types. Price et al. have reported the use of ICS in 50% of COPD patients in primary care settings in the United Kingdom [38]. In the present study, ICS-containing regimens were used in approximately 42.2% of all patients, ranging from 25.3% in GOLD 2017 A to 66.3% in GOLD 2017 D. Most patients receiving triple therapy were on a LABA-ICS + LAMA combination. Although the efficacies of both forms of triple therapy (LABA-ICS + LAMA vs. LABA-LAMA + ICS) may be similar [39], their appropriateness for use may have changed now in the light of recent studies and recommendations [40, 41]. Interestingly, a recent real-world prescription analysis in the UK, demonstrated inappropriate prescribing of triple therapy and confirmed that starting patients on ICS plus LABA results in the inevitable drift to overuse of triple therapy [42]. Although in the present study only 10% of cases could be classified according to GOLD 2017 patient types, this scenario represents an improvement from previous reports on the use of ICS in COPD. Recent studies have shown that ICS prescription in GOLD A/B patients is common, with significant regional variation independent of lung function impairment [34]. Despite the reported overuse of triple therapy [42, 43], we found an increase in their prescription with the progression of GOLD 2017 patient types. Notably, although not explicitly recommended in the guidelines, some patients were receiving LAMA-ICS combination therapy. Only a few trials have evaluated this combination in COPD patients [44, 45].

Overall, there was a considerable variability among the different regions of the country and PCCs, which was significant for the majority of variables due to the large sample size. Interestingly, the variability of FEV₁ was not significant, suggesting that the variability found is not related to disease severity as measured by this parameter. Previous studies have demonstrated significant variability in the processes and outcomes of COPD care in different settings [46, 47]. It has been shown that this variability in clinical practice is not exclusively influenced by clinical presentation or the resources available, but that there is a so-called cluster effect [15]. The cluster effect indicates that patients with similar characteristics may experience different processes of care and outcomes, depending on the center because they are subject to distinct common contextual influences, beyond resources or standards of care, like the characteristics of the catchment area, such as socioeconomic status, and utilization of health services or COPD specific criteria in a determined clinical situation. In our audit we found an extremely important variability in clinical practice that should be explored in the future. The idea behind quality of care is that all patients with the same condition should be managed similarly. However, this is very complex in clinical practice since the clinical presentation of chronic diseases is variable, the perception by the patients varies and is influenced by many personal aspects, and the response to therapies is also variable. Therefore, describing the variability is essential to really understand its potential implications allowing us to seek specific approaches in the future.

Clinical audits have gained traction in healthcare systems as a way of obtaining information on the clinical care being provided, as shown by a recent qualitative study in a primary care setting [18]. This information is of interest to both funding bodies, who want to ensure that the care they finance is of the highest possible standard, and for patients who hope to receive safe and effective healthcare. However, despite evidence from several studies on audit and feedback, few data are currently available on how to use this information. A recent systematic review assessed the effectiveness of audit and feedback and reported an inconsistent picture; some evaluations obtained positive results, whereas others did not [48]. In this context, feedback from the audited information constitutes a key step in improving clinical practice [49]. Probably the most extended initiative is that of care bundles that have been successfully applied in health care [50]. The applicability of these care bundles or any other initiative aiming to improve primary health care in COPD patients should be assessed next.

The strengths of this project are the potential to raise the profile of COPD, and provide an opportunity to promote respiratory medicine in the community, offer suggestions and recommendations on organization of care for the future COPD management guideline, and develop educational resources to support improved clinical practice in all areas, especially those identified as poor practice. Also, the audit provides a formal documentation of the difference between recommended best practices and the status quo, thereby identifying areas requiring improvement.

To interpret the results correctly, several key aspects of the methodology must be considered. First, the audit assessed interventions that had already been performed, and not the results thereof. For instance, it assessed if the physician in charge registered the degree of dyspnea, rather than the intensity of dyspnea after intervention. Thus, the aim of the audit was not to have an idea of the clinical situation of the patient but to have an idea of the clinical performance of the practitioner. Second, the clinical audit made its evaluation from the contents of medical records. Therefore, what was not documented was not evaluated. This does not imply that no intervention had taken place. Accordingly, the estimates presented here were those inferred from the medical record and we may have underestimated the actual figures. The need to record every aspect of clinical practice in the medical record has been previously emphasized [51], and our results further ascertain such need Third, although medical records are electronic in all participating PCC, the application and interface to access the data were different in different regions of the country. Thus, different investigators interacted with different systems, which may have influenced the availability of the data. To check for the quality of these data, the values that were extreme, missing, or inconsistent were returned to the investigators for correction. Fourth, the concept of incorrect prescription could be questionable or misleading is based on the GOLD document. GOLD patient types constitute a way of categorizing patients to ease the selection of the therapeutic approach [40]. However, the magnitude of the impact of the disease can vary considerably within one type [52]. Also, patients with specific clinical phenotypes or presentations might be classified under the same GOLD type, and therefore warrant a change in therapy [53]. Therefore, to confirm those cases with incorrect prescriptions, we selected those cases with specific criteria showing incorrect treatment strategies. Finally, the present paper describes great variability in clinical practice. The next issue to solve would be to do an analysis of the confounding or explanatory variables for such a wide range of variables, which will be a matter of another analysis.

In summary, our results show areas of greatest concerns in COPD care. Clinicians should be aware that only a correct diagnosis makes it possible to select proper pharmacological treatment for COPD, that COPD cannot be diagnosed without proving irreversible obstructive pattern on post-bronchodilator spirometry, that cigarette smoke exposure and especially duration of smoking in years is an important risk factor for the disease and should be systematically evaluated, that smoking cessation, physical activity, nutrition, vaccinations as well as recognition and treatment of comorbidities are key in the management of the disease beyond the use of bronchodilators, and that the use of the different pharmacological approaches should be patient-tailored, and confined to those who do get the benefit from them with no harm.

In conclusion, the present analysis, to the best of our knowledge, is the first ever clinical audit carried out in SpanishPCC to evaluate clinical performance in stable COPD patients. The results show that there is a considerable variability in clinical performance not fully attributable to the severity of the disease and that the majority of the evaluated parameters fell in the range of inadequate performance. A study of the determinants of this variability will help us to understand clinical behavior, and establish strategies to strengthen the clinical practice of COPD management in primary care.