https://orcid.org/0000-0002-1281-5296
Abstract
The Global Initiative for Chronic Obstructive Lung Disease (GOLD) recommendations for the initial bronchodilator to use in newly diagnosed chronic obstructive pulmonary disease (COPD) are based on trials of patients with longstanding disease and treatment. We compared the real world effectiveness of initial treatment with long-acting muscarinic agents (LAMA) versus long-acting beta2-agonists (LABA) on the incidence of exacerbations in newly diagnosed patients. We identified a cohort of patients with COPD, new users of a LAMA or LABA (not combined with an inhaled corticosteroid (ICS)) during 2002–2018, age 50 or older, from the UK’s CPRD database, and followed for one year. The hazard ratio (HR) of exacerbation estimated using the Cox regression model, weighted by fine stratification of propensity scores. The cohort included 40,538 initiators of LAMA and 10,680 of LABA. The adjusted hazard ratio (HR) of a first moderate or severe exacerbation comparing LAMA with LABA initiation was 0.96 (95% CI: 0.90–1.02), while for severe exacerbation it was 0.92 (95% CI: 0.75–1.12). The incidence of exacerbation on LAMA was significantly lower than on LABA (HR 0.88; 95% CI: 0.80–0.96) among patients with a prior exacerbation, and the HR of exacerbation increased with percent predicted FEV1. This study in the real world clinical setting of COPD treatment found that using a LAMA or a LABA (no ICS) as the initial bronchodilator is generally as effective at reducing exacerbation incidence and frequency. However, a LAMA may be more effective in patients with prior exacerbations, which supports the GOLD recommendations for newly diagnosed COPD. The role of airway obstruction on the effectiveness of bronchodilators warrants further investigation.
Introduction
Chronic obstructive pulmonary disease (COPD), a leading cause of morbidity and mortality, has become the third leading cause of death in the world [Citation1,Citation2]. The recent Global Initiative for Chronic Obstructive Lung Disease (GOLD) report imparts distinct recommendations for the initial and follow-up pharmacological treatment of COPD [Citation3]. While the GOLD team recognizes the lack of evidence for newly diagnosed COPD patients, the general recommendation is to start with a long-acting bronchodilator, either a long-acting muscarinic agent (LAMA) or a long-acting beta2-agonist (LABA, no inhaled corticosteroid (ICS)) [Citation3].
More specifically, GOLD advises that treatment should be tailored according to patient phenotypes. Thus, patients in GOLD group C (mild dyspnea, multiple exacerbations) should start with a LAMA since it was shown to be “superior to the LABA regarding exacerbation prevention” in two head-to-head comparisons [Citation4,Citation5]. For patients in GOLD group B (significant dyspnea, no/few exacerbations), one could either use a LABA or a LAMA, as there is no evidence to prefer one class over the other. For GOLD group D patients (significant dyspnea, multiple exacerbations), it is recommended to start with a LAMA because of its “effects on both breathlessness and exacerbations”, though no specific evidence was offered for this choice over a LABA.
The two head-to-head trials used as evidence for the recommendations did show that LAMAs were more effective at reducing exacerbations than LABAs, but most patients in the trials had longstanding COPD. Indeed, over 50% and 30% of the patients in the POET trial were using LABAs and LAMAs, respectively, prior to randomization, which had to be discontinued [Citation4]. In the INVIGORATE trial, patients had COPD for 7 years and 72% were using an ICS, while all LABAs and LAMAs had to be discontinued at randomization [Citation5]. It is unclear whether this finding of LAMA superiority among patients with longstanding disease, who survived up to that point, informs the choice of the initial long-acting bronchodilator to be used in stable COPD. The only observational study of treatment initiation with LAMA or LABA, conducted among patients aged 66 years or older, found that those initiating treatment with a LABA had lower incidence of hospitalization for COPD and death over more than 5 years of follow-up [Citation6]. This study, however, included only older patients and did not provide effects for these medications according to patient phenotypes.
In view of this evidence gap, we conducted an observational study in a real world clinical practice setting of treatment-naïve patients diagnosed with COPD to compare the effectiveness of initiating treatment with monotherapy LAMA versus LABA on the incidence of exacerbations, according to patient phenotypes, including exacerbation history, dyspnea and lung function.
Study design and methods
Data source
We used the Clinical Practice Research Datalink (CPRD), a primary care database from the United Kingdom (UK) that contains primary care medical records for over 50 million people enrolled in more than 1800 general practices. Participating general practitioners in the CPRD–GOLD and Aurum networks record medical information as part of the routine care of patients, including demographic data, lifestyle factors, medical diagnoses recorded using Read and SNOMED-CT codes and prescriptions. Over 85% of the CPRD practices can be linked to the Hospital Episodes Statistics (HES) database and the Office for National Statistics (ONS) that provides electronic death certificates. The CPRD population is representative of the overall population and these data sources have been validated [Citation7–9]. In particular, the information on medications and diagnoses has been validated and shown to be of high quality [Citation7,Citation8,Citation10]. These databases have been used extensively for studies of COPD [Citation11–16].
Study design
We used an incident new-user cohort design to compare the initiators of LAMAs with the initiators of LABAs, after weighting by fine stratification weights from probability of treatment propensity scores [Citation17–19]. The study cohort was formed of all subjects with a medical diagnosis of COPD during 1995–2018 who were new users of a LAMA or LABA (not combined with an ICS) from January 2002 (market entry for LAMAs) until November 2018. Only patients from CPRD practices linked to the HES were included. Subjects initiating treatment with a LAMA and LABA on the same date were excluded. To increase the likelihood of identifying COPD patients, patients had to be 50 years of age or older on the date of this initial prescription and patients with a concurrent or prior asthma diagnosis were excluded. All patients required at least one year of medical history prior to cohort entry, defined by the date of the first LAMA or LABA prescription to identify new use. Subjects were followed up for one year from cohort entry, with follow-up ending one year after cohort entry, date of death, November 2018, or the end of coverage in the practice, whichever occurred first.
Outcome events
The primary outcome was the occurrence of a first moderate or severe COPD exacerbation during follow-up. A severe exacerbation was defined as hospitalization with a primary diagnosis of COPD (ICD-10 J41-J44) while a moderate exacerbation by a new prescription for prednisolone. This definition was verified in sensitivity analyses using an alternative validated definition of exacerbation [Citation20]. Secondary outcomes included the number of exacerbations and of severe exacerbations. Prescriptions of prednisolone within 30 days of each other were considered as continuation of the treatment of the same exacerbation, as were hospitalizations within 30 days of each other. The diagnostic codes to identify these outcomes have been shown to have good accuracy and were used in several studies using the CPRD [Citation12–14,Citation20–22].
Covariates
The covariates used to verify balance between the groups included age, sex, body mass index (BMI), smoking status and excessive alcohol consumption, measured at or prior to cohort entry. The severity of COPD at treatment initiation was measured by the number of prior moderate and severe COPD exacerbations, using the same definitions as the outcome events above and the use of other respiratory drugs during the one-year baseline period. In addition, the most recent measures of FEV1, for which the spirometry procedure quality in these general practices was found to be high [Citation23], and of dyspnea, as measured by mMRC, CAT score or the presence of dyspnea symptoms as previously defined [Citation24], prior to cohort entry were identified. The percent predicted FEV1 measurement was calculated from the absolute FEV1 value using age, sex and height, with race classified as “Caucasian” for all [Citation25]. The most recent measure of blood eosinophil count available prior to cohort entry was also identified. Baseline co-morbidity in the year prior to cohort entry was measured using clinical diagnoses for cancer and hospitalizations for heart failure, myocardial infarction and stroke, as well as prescriptions for medications used for conditions often present in subjects with COPD, including mainly cardiovascular diseases, or potentially associated with COPD exacerbations.
Data analysis
We computed propensity scores using all covariates and fine stratification weights were computed using 100 strata defined by the propensity scores to create similar pseudo-populations of LAMA and LABA users with the same covariate distribution as the entire study population [Citation19]. Standardized mean differences of covariates were computed between the two groups to assess the weighting accuracy. Weighted Cox proportional hazard regression with a robust variance estimate was used to compute the adjusted hazard ratio of a first COPD exacerbation comparing LAMA with LABA use during the first year after treatment initiation. An as-treated analysis was used, based on continuing exposure of the initial treatment of LAMA and LABA within the treated groups, measured by successive prescriptions with a maximal 60-day gap between the prescription dates, with current use defined by the 60-day period after the last such prescription date. This model was adjusted further for patient characteristics found to be imbalanced after weighting. The number of exacerbations was analysed using the negative binomial regression model, which accounts for between-subject variability in the exacerbation pattern, with the same covariate adjustment.
The adjusted hazard ratio of a first moderate or severe COPD exacerbation was stratified by the number of prior COPD exacerbations and by the severity of dyspnea during the baseline year. These two characteristics allow the A–D classifications of COPD patients in the GOLD recommendations [Citation3]. In addition, the adjusted hazard ratio of a first moderate or severe COPD exacerbation by continuous values of lung function measured by percent predicted FEV1 was estimated using cubic splines.
Sensitivity analyses first included an intent-to-treat analysis, irrespective of treatment switching or discontinuations during the one-year follow-up. We repeated the primary analysis using an alternative validated definition of exacerbation, namely uses a composite event defined by prescriptions for antibiotics and oral corticosteroids, diagnostic codes for lower respiratory tract infection or acute exacerbation of COPD, as well as a record of two or more treated respiratory symptoms of AECOPD, namely increase in breathlessness or cough or sputum volume and/or purulence [Citation20]. We varied the continuous use definition for the as-treated analysis using 30- and 90-day gaps. All analyses were conducted using SAS version 9.4. The study protocol was approved by the Independent Scientific Advisory Committee of the CPRD (Protocol 18_032MnA) and the Ethics Committee of the Jewish General Hospital (JGH Protocol #18-026), Montreal, Quebec, Canada.
Results
The study cohort included 51,218 patients with COPD aged 50 years or older, of which 40,538 initiated COPD treatment with a LAMA and 10,680 with a LABA (). The baseline characteristics of these patients, comparing the LAMA with the LABA initiators before weighing is shown in Supplemental Table S1. After weighting by fine stratification from the probability of treatment propensity scores, shows good balance between the two groups, except for year of cohort entry which we adjusted for in data analyses. The median time between the diagnosis of COPD and the initial long-acting bronchodilator was 3.4 and 6.4 months for the LAMA and LABA groups, respectively.
Figure 1. Flowchart of cohort selection.
Table 1. Baseline characteristics of the study cohort of initiators of LAMA compared with the initiators of LABA, made comparable by fine stratification weighting from probability of treatment propensity scores, with corresponding standardized mean differences.
During the one-year follow-up, 15% in each group discontinued their bronchodilator and did not restart, while 31% of LAMA and 34% of LABA users discontinued and restarted the same bronchodilator within the one-year follow-up. Moreover, 16% of LAMA and 18% of LABA users were censored because they received an ICS, switched or added the other bronchodilator. The LABAs included salmeterol (83%), formoterol (13%) and indacaterol or olodaterol (4%). The LAMAs included tiotropium (84%), umeclidinium (6%), glycopyrronium (6%) and aclidinium (4%).
The adjusted hazard ratio of a first moderate or severe exacerbation with LAMA relative to LABA is 0.96 (95% CI: 0.90–1.02), while for severe exacerbation the hazard ratio is 0.92 (95% CI: 0.75–1.12) (). The cumulative incidence of a first moderate or severe exacerbation over 1 year is 27.7 for LAMA and 29.3 for LABA (). The mean number of moderate or severe exacerbations over one year was 50.0 per 100 per year with LAMA compared with 54.0 per 100 per year with LABA (). The corresponding adjusted rate ratio of any exacerbation was 0.94 (95% CI: 0.88–1.01), while for severe exacerbations it was 0.88 (95% CI: 0.75–1.04).
Figure 2. Kaplan–Meier curves of the one-year cumulative incidence of the first moderate or severe COPD exacerbation comparing initial treatment with LAMA and LABA, after adjustment by inverse probability of treatment weights.
Table 2. Crude and adjusted hazard ratios of a first COPD exacerbation comparing LAMA initiation with LABA initiation in patients with COPD in the first year after treatment initiation, from the as-treated analysis.
Table 3. Crude and adjusted rate ratios of the number of COPD exacerbations comparing LAMA initiation with LABA initiation in patients with COPD in the first year after treatment initiation, estimated using negative binomial regression from the as-treated analysis.
shows that the comparative incidence of moderate or severe exacerbation is not affected by the severity of dyspnea before initiating treatment with a LAMA or LABA. On the other hand, the incidence of exacerbation on LAMA was lower than on LABA among patients with any prior exacerbations (HR 0.88; 95% CI: 0.80–0.96), but was not different among patients with no prior exacerbations (HR 1.06; 95% CI: 0.96–1.17). shows that the hazard ratio of a moderate or severe exacerbation comparing treatment initiation with LAMA versus LABA rises linearly with the FEV1. It suggests that the incidence of exacerbation is lower with a LAMA among patients with % predicted FEV1 <40%, while it is higher when % predicted FEV1 >75%, compared with LABA.
Figure 3. Hazard ratio of a moderate or severe COPD exacerbation with LAMA versus LABA initiation (and 95% confidence limits), as a function of the baseline FEV1 (%), after adjustment by inverse probability of treatment weights. Computed among the 70% subset of patients with available data on FEV1.
Table 4. Crude and adjusted hazard ratios of a first moderate or severe exacerbation associated with LAMA relative to LABA in patients with COPD, with one-year follow-up, from the as-treated analyses, stratified by severity of dyspnea and by prior exacerbations.
Sensitivity analyses by intent-to-treat, using an alternative definition of exacerbation or altering the gap between prescriptions to define the as-treated analysis did not alter the main results (Supplemental Table S2).
Discussion
This real world observational study suggests that using a LAMA or a LABA (no ICS) as the initial long-acting bronchodilator to treat patients newly diagnosed with COPD results in a similar incidence of exacerbations over the first year of use. However, for patients who had at least one exacerbation in the year prior to treatment initiation, a lower incidence of exacerbation is observed with a LAMA compared with a LABA. This study also suggests that the comparative incidence of exacerbation varies linearly with airways obstruction as measured by percent FEV1 predicted. Thus, our study generally supports the most recent GOLD recommendations for the initial treatment of COPD with long-acting bronchodilators and advocates for further investigations into the role of the severity of airway obstruction on the comparative effectiveness of these long-acting bronchodilators.
These results are the first to support the recent GOLD recommendations specifically for the initial pharmacological treatment of COPD [Citation3]. Indeed, in the absence of studies based on patients newly diagnosed with COPD, these recommendations were based on trials that enrolled patients with longstanding COPD, including the POET and INVIGORATE trials, where most patients had already been treated for a prolonged time [Citation4,Citation5]. The only study comparing treatment initiation with LAMA or LABA was an observational study conducted among patients aged 66 years or older at the time of treatment initiation [Citation6]. That study, however, lacked data on younger patients and did not provide effects for these medications according to patient phenotypes such as exacerbation history, dyspnea and lung function.
Our study is the first to include patients at the time of their first long-acting bronchodilator treatment and include key covariates to apply the GOLD symptom classification, namely measures of dyspnea and of prior exacerbations. We found that for patients with at least one exacerbation, a LAMA should be the preferred initial long-acting bronchodilator. This finding is consistent with the GOLD groups C (mild dyspnea) and D (moderate or higher dyspnea) recommendations of starting with a LAMA, based on the POET and INVIGORATE trials. Our study also supports the GOLD recommendation for group B patients with no prior exacerbations, namely that a LAMA or a LABA provide similar benefit, as we found no difference in the incidence of exacerbation between the two long-acting bronchodilators. The recent American Thoracic Society Clinical Practice Guideline recommend to initiate treatment of symptomatic COPD with a LABA–LAMA combination in patients who complain of dyspnea or exercise intolerance over LABA or LAMA monotherapy as first-line therapy, irrespective of the exacerbation history [Citation26]. However, here again, this recommendation was based on trials that enrolled patients with longstanding COPD and treatment. Nonetheless, this strategy should be the object of further research on the real world effects of treatment initiation options for COPD.
One distinction between our study and the GOLD recommendations is the cut-off for exacerbation history. Whereas GOLD uses two or more exacerbations (or one or more severe exacerbation) to classify groups C and D, we found that any exacerbation is sufficient to influence the initial treatment. This distinction is likely due to the GOLD-cited trials that included patients already treated for a longstanding COPD and required a history of at least one exacerbation. Indeed, both trials required patients to have at least one treated exacerbation in the previous year and thus could not provide evidence for patients with no prior exacerbation. Our study had 72% of the patients with no exacerbation prior to initiating treatment, a probably realistic profile of exacerbation history from the setting of clinical practice, which allows inference on patients with no prior exacerbations. Indeed, this explains our finding of a one-year cumulative incidence of a first exacerbation at around 28%, lower than the 50% incidence seen in trials of patients with a history of exacerbations and longstanding disease.
Another original finding is the linear relation in the comparative incidence of exacerbation with the degree of airway obstruction. Indeed, our study found that treatment initiation with a LAMA results in a lower incidence of exacerbation in patients with a low FEV1 but a higher incidence with high FEV1, when compared with a LABA. This analysis suggests that the degree of airway obstruction should be considered as another potential marker of treatment optimization in selecting the initial bronchodilator in managing COPD. This finding certainly warrants further study.
Our study population is, expectedly, quite different from those enrolled in the POET and INVIGORATE head-to-head trials on which much of the GOLD recommendations for the initial treatment are based. In our study, the mean FEV1 (% predicted) was 59.1 compared with 49.3 in POET and 40.5 in INVIGORATE, differences likely explained by the trials enrolling patients with longstanding COPD. In addition, our study excluded patients previously treated with ICS, not recommended as first-line treatment for COPD. In our study, over 90% of the patients were smokers, compared with 48% in POET and 35% in INVIGORATE, and there was a large sex difference, as over 40% of the patients in our study were women, compared with 25% in POET and 23% in INVIGORATE. Here again, these differences should be considered in the context of patients with longstanding COPD enrolled in the trials. It is also different from the previous observational study from Ontario that included patients aged 66 years or older at the time of treatment initiation [Citation6]. That study included patients with a history of asthma diagnosis (50%) and users of inhaled corticosteroids at cohort entry (51%), both exclusion criteria of our study.
The new-user design that we employed is a strength of our study. By identifying patients initiating treatment with a long-acting bronchodilator, the design reduces the potential for confounding when comparing the two agents head-to-head. Indeed, while we used extensive methods of weighing by fine stratification of propensity scores to make the two groups comparable, we found that the groups were not much different to start with. Second, the simple definition of exacerbation we used for the outcome, as well as the use of hospitalization records which are less affected by misclassification, produced practically identical results to those using a more complex validated definition [Citation11,Citation21]. Finally, this study complements our previous study that compared initial treatment with LABA–ICS versus LAMA [Citation14]. Indeed, that study was based on a smaller version of the CPRD and did not have sufficient patients initiating treatment with a LABA (no ICS), while our current study includes a larger version which allows this important comparison between bronchodilators that do not contain ICS.
Our study design is limited by the exposure measures that can lead to misclassification, as they are based on written prescriptions which may not be dispensed and can accentuate the misclassification by the potentially less than optimal inhaler technique in the real world setting. While the two groups were not much different at the time of treatment initiation, residual confounding cannot be ruled out despite the use of propensity scores and fine stratification. The continuous treatment duration for the as-treated analysis was relatively short because patients either discontinued or added another treatment early, a reflection of real world clinical practice. Our definition of the outcome, namely COPD exacerbation, was based on a hospitalization for COPD (severe) or a new prescription for prednisolone (moderate). The sensitivity analysis using a validated definition of exacerbation [Citation20], did not change the results, with the hazard ratios practically identical at 0.96 and 0.97. Finally, the use of physician diagnoses to identify the study patients with COPD may have misclassified some patients with asthma, though the 50 years of age cut-off and the exclusion of ICS use prior to treatment initiation likely reduced this potential misclassification.
Conclusions
This large real world study in the clinical setting of treatment for COPD generally supports the most recent GOLD recommendations for the initial treatment of COPD. It found that initiating long-acting bronchodilator therapy with a LAMA or a LABA (no ICS) is generally just as effective in preventing COPD exacerbations in most patients. However, a LAMA may be more effective in patients with prior COPD exacerbations. It also suggests that a LAMA may be more effective in patients with more severe airway obstruction while a LABA may be more effective in those with milder airway obstruction.
Supplemental Material
Download PDF (461.5 KB)Acknowledgments
Dr. Ernst participated in study design, data interpretation and writing of the manuscript. Ms. Dell’Aniello participated in study design, data analysis and writing of the manuscript. Dr. Suissa participated in data acquisition, study design, data analysis, data interpretation, writing of the manuscript and acts as guarantor of this manuscript.
Declaration of interest
Ms. Dell’Aniello and Dr. Ernst have no conflict of interest to report. Dr. Suissa attended scientific advisory committee meetings for Atara, Merck, Pfizer and Seqirus, and received speaking fees from Boehringer-Ingelheim and Novartis.
Additional information
Funding
References
- Rabe KF, Hurd S, Anzueto A, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med. 2007;176(6):532–555. DOI:10.1164/rccm.200703-456SO
- Lozano R, Naghavi M, Foreman K, et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012;380(9859):2095–2128. DOI:10.1016/S0140-6736(12)61728-0
- Global strategy for the diagnosis, management, and prevention of Chronic Obstructive Pulmonary Disease. 2019. Report http://goldcopdorg/gold-2019-global-strategy-diagnosis-management-prevention-copd/2019.
- Vogelmeier C, Hederer B, Glaab T, et al. Tiotropium versus salmeterol for the prevention of exacerbations of COPD. N Engl J Med. 2011;364(12):1093–1103. DOI:10.1056/NEJMoa1008378
- Decramer ML, Chapman KR, Dahl R, et al. Once-daily indacaterol versus tiotropium for patients with severe chronic obstructive pulmonary disease (INVIGORATE): a randomised, blinded, parallel-group study. Lancet Respir Med. 2013;1(7):524–533. DOI:10.1016/S2213-2600(13)70158-9
- Gershon A, Croxford R, To T, et al. Comparison of inhaled long-acting β-agonist and anticholinergic effectiveness in older patients with chronic obstructive pulmonary disease: a cohort study. Ann Intern Med. 2011;154(9):583–592. DOI:10.7326/0003-4819-154-9-201105030-00003
- Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44(3):827–836. DOI:10.1093/ije/dyv098
- Herrett E, Thomas SL, Schoonen WM, et al. Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69(1):4–14. DOI:10.1111/j.1365-2125.2009.03537.x
- Wolf A, Dedman D, Campbell J, et al. Data resource profile: Clinical Practice Research Datalink (CPRD) aurum. Int J Epidemiol. 2019;48(6):1740–1740g. DOI:10.1093/ije/dyz034
- Jick SS, Kaye JA, Vasilakis-Scaramozza C, et al. Validity of the general practice research database. Pharmacotherapy. 2003;23(5):686–689. DOI:10.1592/phco.23.5.686.32205
- Quint JK, Mullerova H, DiSantostefano RL, et al. Validation of chronic obstructive pulmonary disease recording in the Clinical Practice Research Datalink (CPRD-GOLD). BMJ Open. 2014;4(7):e005540. DOI:10.1136/bmjopen-2014-005540
- Suissa S, Dell'Aniello S, Ernst P. Concurrent use of long-acting bronchodilators in COPD and the risk of adverse cardiovascular events. Eur Respir J. 2017;49(5):1602245. DOI:10.1183/13993003.02245-2016
- Suissa S, Dell'Aniello S, Ernst P. Long-acting bronchodilator initiation in COPD and the risk of adverse cardiopulmonary events: a population-based comparative safety study. Chest. 2017;151(1):60–67. DOI:10.1016/j.chest.2016.08.001
- Suissa S, Dell'Aniello S, Ernst P. Comparative effectiveness of LABA–ICS versus LAMA as initial treatment in COPD targeted by blood eosinophils: a population-based cohort study. Lancet Respir Med. 2018;6(11):855–862. DOI:10.1016/S2213-2600(18)30368-0
- Suissa S, Dell'Aniello S, Ernst P. Comparative effectiveness and safety of LABA–LAMA vs LABA–ICS treatment of COPD in real-world clinical practice . Chest. 2019;155(6):1158–1165. DOI:10.1016/j.chest.2019.03.005
- Suissa S, Dell'Aniello S, Ernst P. Comparative effects of LAMA–LABA–ICS vs LAMA–LABA for COPD: cohort study in real-world clinical practice. Chest. 2020;157(4):846–855. DOI:10.1016/j.chest.2019.11.007
- Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158(9):915–920. DOI:10.1093/aje/kwg231
- Suissa S, Moodie EE, Dell'Aniello S. Prevalent new-user cohort designs for comparative drug effect studies by time-conditional propensity scores. Pharmacoepidemiol Drug Saf. 2017;26(4):459–468. DOI:10.1002/pds.4107
- Desai RJ, Franklin JM. Alternative approaches for confounding adjustment in observational studies using weighting based on the propensity score: a primer for practitioners. BMJ. 2019;367:l5657. DOI:10.1136/bmj.l5657
- Rothnie KJ, Müllerová H, Hurst JR, et al. Validation of the recording of acute exacerbations of COPD in UK primary care electronic healthcare records. Plos One. 2016;11(3):e0151357. DOI:10.1371/journal.pone.0151357
- Rothnie KJ, Müllerová H, Thomas SL, et al. Recording of hospitalizations for acute exacerbations of COPD in UK electronic health care records. Clin Epidemiol. 2016;8:771–782. DOI:10.2147/CLEP.S117867
- Meropol SB, Metlay JP. Accuracy of pneumonia hospital admissions in a primary care electronic medical record database. Pharmacoepidemiol Drug Saf. 2012;21(6):659–665. DOI:10.1002/pds.3207
- Rothnie KJ, Chandan JS, Goss HG, et al. Validity and interpretation of spirometric recordings to diagnose COPD in UK primary care. Int J Chron Obstruct Pulmon Dis. 2017;12:1663–1668. DOI:10.2147/COPD.S133891
- Rebordosa C, Plana E, Aguado J, et al. GOLD assessment of COPD severity in the Clinical Practice Research Datalink (CPRD). Pharmacoepidemiol Drug Saf. 2019;28(2):126–133. DOI:10.1002/pds.4448
- Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference values for spirometry for the 3–95-yr age range: the global lung function 2012 equations . Eur Respir J. 2012;40(6):1324–1343. DOI:10.1183/09031936.00080312
- Nici L, Mammen MJ, Charbek E, et al. Pharmacologic management of chronic obstructive pulmonary disease. An OFFICIAL American Thoracic Society Clinical Practice Guideline. Am J Respir Crit Care Med. 2020;201(9):e56–e69. DOI:10.1164/rccm.202003-0625ST