ABSTRACT
In moderate–severe chronic obstructive pulmonary disease (COPD), long-acting bronchodilators (LBs) are recommended to improve the quality of life. The aims of this study were to measure adherence to LBs after discharge for COPD, identify determinants of adherence, and compare amounts of variation attributable to hospitals of discharge and primary care providers, i.e. local health districts (LHDs) and general practitioners (GPs). This cohort study was based on the Lazio region population, Italy. Patients discharged in 2007–2011 for COPD were followed up for 2 years. Adherence was defined as a medication possession ratio >80%. Cross-classified models were performed to analyse variation. Variances were expressed as median odds ratios (MORs). An MOR of 1.00 stands for no variation, a large MOR indicates considerable variation. We enrolled 13,178 patients. About 29% of patients were adherent to LBs. Adherence was higher for patients discharged from pneumology wards and for patients with GPs working in group practice. A relevant variation between LHDs (MOR = 1.21, p = 0.001) and GPs (MOR = 1.28, p = 0.035) was detected. When introducing the hospital of discharge in the model, the MOR related to LHDs decreased to 1.05 (p = 0.345), MOR related to GPs dropped to 1.22 (p = 0.086), whereas MOR associated with hospitals of discharge was 1.38 (p < 0.001). Treatments with proven benefit for COPD were underused. Moreover, a relevant geographic variation was observed. This heterogeneity raises equity concerns in access to optimal care. The reduction of variability among LHDs and GPs after entering the hospital level proved that differences we observe in primary care partially ‘reflect’ the clinical approach of hospitals of discharge.
Background
Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of death in the world. By 2020, COPD is expected to become the third leading cause of death, accounting for 5 million deaths per year, and the fifth leading causing of disability worldwide Citation(1). Care of patients with COPD has changed radically over the past two decades and novel therapies can improve patients' health status Citation(2). In the case of moderate-to-severe COPD, exacerbations or persistent symptoms, regular treatment with long-acting bronchodilators (LBs; long-acting beta-agonists and/or long-acting muscarinic antagonists) is recommended to control the symptoms, to reduce the occurrence of acute exacerbations and to improve the quality of life Citation(2). However, observational studies reported poor adherence to LBs, even after hospitalisation for COPD exacerbation Citation(3–6). Non-adherence to treatment reduces the clinical benefit of therapy and can account for many of the observed differences between efficacy and effectiveness of the drug treatment Citation(7). Moreover, substantial geographic variation exists in the treatment of patients with COPD Citation(8). This unwarranted variability has important consequences in terms of equity in access to optimal care Citation(8). Unfortunately, from the current scientific evidence it is not possible to quantify how much of the gap between clinical guidelines and practice is attributable to the patient behaviour, to the clinical and organisational processes within hospitals, or to the primary care providers, such as local health districts (LHDs) or general practitioners (GPs). The LHD is a body delegated by the National Health System to provide health care to a specific area. Each LHD is composed of a group of GPs sharing the same clinical guidelines and participating in the same educational interventions, coordinated by a district director. Within this framework, the analysis of ‘components of variation’ may be a useful tool to define areas for targeted interventions aimed at improving adherence to guidelines and equity in health care. In fact, where there is large variability between health care providers there is scope for intervention. The main goal for health policies should be to shrink the variability upwards, decreasing health inequalities and increasing medication adherence. This topic can now be addressed thanks to the multilevel methodology. Multilevel models analyse the health care system as if it were a ‘hydraulic system’. If there are some ‘leaks’, this approach allows us to locate where they are and prioritise interventions. Moreover, it is possible to apply multilevel methodology to data that are not purely hierarchical, properly handling levels that are ‘crossed’ with each other, such as hospitals of discharge and primary care providers. The application of these flexible statistical methods to real practice data collected on very large population is the innovative aspect of our research.
The objectives of this study are to measure the adherence to LBs after discharge for COPD, to quantify and compare the amounts of variation attributable to the hospital of discharge and to the primary care providers, and to identify determinants of adherence to LBs.
Materials and methods
Data sources
Our department has access to the Lazio regional health information systems that contain mortality, hospital admission, and drug claims data. The details of the individual information systems are reported in the Appendix (supplementary data).
Setting and study cohort
This population-based, observational study was based on the population living in the Lazio region of Italy, which comprises approximately 5 million persons. Using data from the regional Hospital Information System (HIS), the study included all patients discharged from hospital between January 1st 2007 and June 30th 2011 with a recorded diagnosis of COPD. Specific criteria for cohort inclusion were main diagnosis of COPD (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM] codes: 490, 491, 492, 494, 496) or main diagnosis of COPD-related causes along with a secondary diagnosis of COPD. COPD-related causes included respiratory failure (ICD-9-CM: 518.81–518.84), dyspnoea and other respiratory distress (786.0), cough (786.2), and abnormal sputum (786.4). The first hospitalisation fulfilling the selection criteria during the enrolment period was considered as the index admission. Recorded diagnoses in HIS had been validated in a previous evaluation of clinical records Citation(9). About 94% of the reviewed cases were confirmed as being COPD cases.
Exclusion criteria
Patients not alive at discharge, aged less than 45 years, with secondary diagnosis of major trauma (ICD-9-CM: 484–487) or major surgeries (surgical diagnosis-related group, excluding 482 and 483) during the index admission were excluded. Finally, patients who received an outpatient regimen for less than 15 days were excluded, in order to allow for enough time for consistently estimating the adherence to LBs.
Follow-up
Individual follow-up for measuring drug use was considered to start on the first day after discharge from the index admission. The end of the observation period was defined as either the end of 2-year follow-up, time of death or the date of any hospitalisation following discharge from the index admission, whichever occurred first. The last censoring criterion allows to measure the net impact of the hospital that has discharged the patient without the potential interference of subsequent hospitalisations.
Measures of adherence to medication
Drug use information was collected from the regional registry of all drugs dispensed by public and private pharmacies. All drugs in this study were included in the patients' health care plans and were equally available to all residents, in accordance with the universal health care coverage provided in Italy. Moreover, in the Lazio region, inhaled respiratory drugs are dispensed completely free of charge to low-income patients. Information about prescriptions of long-acting beta-agonists (LABA), long-acting muscarinic antagonists (LAMA), inhaled corticosteroids (ICS) and short-acting beta-agonists (SABA) were retrieved for all patients. The degree of drug treatment coverage was measured through the medication possession ratio (MPR), calculated as the number of days of medication supplied during the follow-up on the basis of the defined daily doses (DDD) Citation(10) divided by the number of calendar days in the follow-up. Patients were defined as adherent when more than 80% Citation(11) of their individual follow-up was covered by treatment (MPR > 80%). The threshold of 75% (MPR > 75%) was also used as a sensitivity analysis. Moreover, we evaluated medication use ‘immediately’ after the acute event, by analysing the prescription patterns during the 6 months following discharge from the index admission.
Statistical analysis
Continuous variables were presented as mean value ± standard deviation and/or median value. A map of the Lazio region was produced in order to show and compare the proportions of adherent patients by LHD. The classes used in the maps were calculated applying the Jenks natural breaks optimisation algorithm Citation(12), which reduces the variance within classes and maximises the variance between classes.
It is important to note that the data structure is not purely hierarchical. In fact, patients are nested within LHDs and within hospitals of discharge. However, the nesting structure may be less clear when we consider health districts and hospitals of discharge. In other words, patients are nested within the ‘cross-classification of health districts and hospitals’ (). Therefore, cross-classified logistic multilevel models Citation(13) were performed in order to analyse geographic variation, by measuring and comparing the amounts of variability attributable to the hospital of discharge and to the primary care providers. The standard multilevel regression including only the primary care levels (i.e. LHD and GP) was compared with the cross-classified model including both primary care levels and the hospital of discharge level. The Akaike Information Criterion (AIC) was used to determine the model that provided the best account of the data. In fact, AIC deals with the trade-off between the goodness-of-fit and the complexity of the model. The ‘best’ model is the one with minimum AIC value Citation(14). The variance components were expressed in terms of median odds ratios (MORs). The MOR quantifies the variation between clusters. This measure is always greater than or equal to 1.00. If the MOR equals 1.00, there is no variation between clusters. If there is considerable between-cluster variation, the MOR will be large Citation(15). The MORs were estimated controlling for patient characteristics. In fact, explanatory variables that are divided very selectively across groups can often explain a fair amount of group level variance. The interpretation would generally be that this does not reflect a real contextual effect, but rather the unequal composition of the groups Citation(13). We did not control for GP characteristics in order to measure and emphasise the source of variability attributable to primary care features.
Figure 1. The cross-classified structure.
A cross-classified model was also applied to identify determinants of adherence to LBs, properly taking into account the correlation within the specified clusters Citation(13). In this case, both patient and GP characteristics were included. Determinants of adherence were identified in two steps. First, the following factors were selected based on a priori knowledge Citation(3): patient's socio-demographic factors (age, gender and educational level), discharge ward, COPD severity, concomitant respiratory diseases and co-morbidities (see Appendix for details); gender, age and organisational arrangement Citation(16) (none, association, network, group practice) of the GP. Second, the potential determinants were further selected using a bootstrap stepwise procedure to determine which factors were actually associated with the outcome of interest Citation(17). Using this approach, 1,000 replicated bootstrap samples were selected from the original cohort. A bootstrap sample is a sample of the same size as the original dataset chosen with replacement. Thus, a given subject in the original cohort may be selected multiple times, only once, or not at all, in a specific bootstrap sample. A stepwise procedure, using thresholds of p = 0.05 for variable selection and elimination, was applied to each replicated sample, and only factors selected in at least 50% of the procedures were included in the final cross-classified multilevel model. Odds ratios (ORs), 95% confidence intervals (95% CIs) and p-values were reported.
Results
From the initial number of 14,769 patients discharged from hospital with a diagnosis of COPD, 13,178 patients were enrolled in the cohort (). The mean follow-up time was 554 days (95% CI: 550–557). The median follow-up time was 671 days. About 21% of patients were discharged from pneumology wards, 60% from internal medicine wards and 19% from other wards. The ‘hierarchical’ health care system was composed as follows: 3,852 GPs, 55 LHDs and 135 cross-classified hospitals of discharge. A total of 7,180 patients (54%) were men. The mean age was 73 ± 9 years for men and 75 ± 10 years for women. shows the prevalence of the proxies of COPD severity, concomitant respiratory diseases and previous use of drugs, by age group. Respiratory failure (46%), pulmonary infections (15%) and acute pulmonary symptoms (9%) were the most common conditions. It is worth noting that about 84% of patients were treated with cardiovascular drugs in the 12 months preceding the index admission for COPD. The prevalence of co-morbidities by age group is shown in . Hypertension (52%), diabetes (22%), ischaemic heart disease (21%) and arrhythmia (18%) were the most common co-morbidities. Overall, more than 85% of patients had at least one co-morbid condition, excluding concomitant respiratory diseases. The prevalence of co-morbidity progressively increased with age, ranging from 71% in the ‘45–54’ age group up to 90% in the ‘85+’ age group.
Figure 2. The exclusion criteria flowchart.
Table 1. Prevalence of proxies of COPD severity, concomitant respiratory diseases and previous use of drugs, by age group.
Table 2. Prevalence of co-morbidities, by age group.
shows the prescription patterns during the 6 months after discharge. The majority of patients (31%) received triple therapy (i.e. LABA + ICS + LAMA), 19% were treated with LABA + ICS and 27% received prescriptions for SABA. A relevant variation in respiratory prescription patterns was observed among LHDs. The use of triple therapy ranged from 15% to 54%, LABA + ICS use from 8% to 29%, whereas the proportion of patients receiving SABA prescriptions ranged from 8% to 48%.
Table 3. Prescription patterns during the 6 months following discharge.
Overall, in the Lazio region about 29% of patients were adherent to LB therapy during the first 2 years following hospitalisation for acute exacerbation of COPD. The proportion of adherent patients increased to 31% when considering an MPR > 75%.
The cross-classified logistic multilevel model () showed that the probability of adherence to LB therapy after hospitalisation for acute exacerbation of COPD was strongly influenced by the patient and GP characteristics. Female patients had a lower probability of adherence. The effect of age was not linear: with respect to the reference category (45–54 years), the probability of adherence significantly increased in the age classes 55–64 years (OR = 1.63, p < 0.001); 65–74 years (OR = 2.11, p < 0.001); 75–84 years (OR = 1.87, p < 0.001), and decreased in the older age group (age greater than or equal to 85 years; OR = 1.10, p = 0.464). The effect of the educational level was not significant. Moreover, adherence was significantly higher for patients discharged from pneumology wards, for patients with a more severe COPD, and for patients with concomitant respiratory diseases. On the contrary, all co-morbidities were associated with lower adherence to LBs. As regards the GP characteristics, adherence was higher for patients with GPs working in group practice, i.e. sharing facilities, electronic patient records, administrative and clinical staff. In addition, a high geographic variation was observed between the LHDs of the Lazio region. The percentages of adherence to LBs ranged from 15% to 40% (). In , the amounts of variation attributable to the hospital of discharge and to the primary care providers were measured and compared. When analysing the variation among primary care providers, after controlling for patient characteristics, a relevant variation between LHDs (MOR = 1.21, p = 0.001) and between GPs working in the same LHD (MOR = 1.28, p = 0.035) was detected. However, when introducing the hospital of discharge as a cross-classified level, the variation between LHDs substantially decreased (MOR = 1.05, p = 0.345) as well as the variation between GPs working in the same district (MOR = 1.22, p = 0.086). When introducing the hospital level, the variation between primary care providers can be seen as the variability between LHDs and GPs as if all patients were discharged from the same hospital. Therefore, a portion of the variability in primary care is attributable to the hospital that has discharged the patient. Moreover, the variability in patient adherence attributable to the hospital of discharge was statistically significant (p < 0.001) and substantially higher with respect to the other sources of variability. In fact, the MOR associated with the hospital of discharge was 1.38, the MOR associated with the LHD was 1.05, and the MOR associated with the GP was 1.22. The sensitivity analysis produced stable results ().
Table 4. Determinants of adherence to long-acting bronchodilators (MPR > 80%).
Figure 3. Percentages of adherence to long-acting bronchodilator therapy by local health district.
Table 5. The trade-off between hospital of discharge and primary care providers.
Table 6. Sensitivity analysis (MPR > 75%): The trade-off between hospital of discharge and primary care providers.
Discussion
In a study of 13,178 patients, we found that after hospital discharge for acute exacerbation of COPD, less than 30% of patients were adherent to LBs in the following 2 years. Treatments with proven benefit for COPD are underused, despite strong evidence that their use will result in better patient outcomes Citation(2). This result is even more alarming if we consider that when using the less restrictive definition of adherence (MPR > 75%) the adherence ‘rate’ increased by only two percentage points. Our findings are consistent with the results of other studies, which reported unsatisfactory prescribing rates of LBs in different time periods Citation(3) and in different countries (Citation4–6). Among patient determinants of adherence, we found that co-morbidities played an important, negative role. A hypothesis for this finding may be related to the cumbersomeness of therapy, which increases with age and number of co-morbidities. The longer and more complex is the list of drugs prescribed, the lower is the medication adherence Citation(18). The impact of the type of discharge ward was very impressive and consistent with the results from a previous investigation Citation(19): patients discharged from pneumology wards were much more likely to be adherent to evidence-based medications. However, COPD undertreatment is not entirely due to patient characteristics and appropriateness of hospital management. In fact, survey results suggest that primary care clinicians may have major gaps in their knowledge of COPD management. Several studies demonstrated that GP's adherence to clinical guidelines is low and COPD is often misdiagnosed or treated inappropriately (Citation20–22). In our study adherence to LBs was higher with physicians working in teams, sharing facilities, electronic patient records and clinical staff.
A relevant and unwarranted geographic variation in adherence to guidelines was observed between the LHDs of the Lazio region. Spatial heterogeneity raises equity concerns in access to optimal care. This study focuses on the ‘trade-off’ between hospital and primary care in determining variation. The MORs estimated by the cross-classified multilevel models are very interesting. They allow to measure and compare the amount of variation attributable to the ‘discharge phase’ and to the following ‘primary care phase’. The remarkable reduction of the variability among LHDs and GPs after entering the hospital level proved that differences in adherence we observe in primary care partially ‘reflect’ the clinical and organisational approach of the hospital, whose aims are both the correct setting of drug therapy, and the planning of subsequent visits for patient monitoring. Actually, adherence to LBs after hospital discharge for COPD exacerbation was influenced more by the hospital that discharged the patient (MOR = 1.38) than by the primary care providers (MORs = 1.05 and 1.22 for LHD and GP, respectively). According to the study results, it is possible to formulate hypotheses about the potential ‘plans of action’ for health policies aimed at improving adherence to LBs. They include the following: 1) to organise prescribing upskilling sessions for GPs, focusing on the most recent clinical guidelines, shared decision-making, and practical concerns about drug interactions and side effects; 2) to promote education on doctor–patient relationships, underlining the effectiveness of systematic motivational support; 3) to stimulate association for primary care physicians, to improve continuity of care; and 4) to improve the organisational processes within the hospital to discharge patients from specialist wards, implement regular training courses for hospital doctors, provide advice on the correct use of devices, write clear and accurate hospital discharge letters, and plan the subsequent visits for patient monitoring.
There are some study limitations to be considered. First, data on lung function that would allow for a spirometric staging of the disease were not available. However, we can assume that patients hospitalised for an acute exacerbation of COPD are probably affected by moderate or more severe stages of the disease. In this regard, recorded diagnoses in the HIS had been validated in a previous re-abstract study: the majority (94%) of reviewed cases were confirmed as being cases of COPD Citation(9). Second, while restricting the study to patients discharged from hospital can be considered a strength of the study since it increases the reliability of the COPD diagnosis, it also restricts the generalizability of results. Third, our drug claims database does not contain information on the prescribed daily doses and adherence to drug treatment was estimated on the basis of the DDDs. Although this is a useful instrument for comparing the results from different studies Citation(10), misclassification of drug utilisation may have occurred. Finally, there are some things to keep in mind concerning the use of stepwise procedures for selecting determinants of adherence to LBs. The original list of potential determinants was defined on the basis of a priori knowledge Citation(3). Bootstrap stepwise is a tool to improve the efficiency of the statistical model. In fact, this procedure allowed to identify which of the a priori potential determinants were actually associated with adherence to LBs in the specific context of our data. This avoids overparameterisation and improves estimator efficiency Citation(17).
Conclusions
In clinical practice, pharmacotherapy after hospital discharge for acute exacerbation of COPD is not consistent with clinical guidelines. Moreover, the relevant geographic variation in adherence to LBs raises concerns about equity in access to optimal care. Finally, adherence was influenced more by the hospital that discharged the patient than by primary care providers. Cross-classified models proved to be a useful tool for identifying the priority lines of action to improve adherence and define areas for targeted health care interventions.
Declaration of interest
The authors declare that they have no conflict of interest. This study was carried out in full compliance with current privacy laws. The Department of Epidemiology is legitimised by the Lazio Region Committee in managing and analysing data retrieved from the regional health information systems for epidemiological purposes.
Notes
1 Long-acting bronchodilators available in the Italian market during the study period.
References
- Murray CJ, Lopez AD. Alternative projections of mortality and disability by cause 1990–2020: Global Burden of Disease Study. Lancet 1997; 349(9064):1498–1504.
- Global Initiative for Chronic Obstructive Lung Disease (GOLD). Global Strategy for the Diagnosis, Management, and Prevention of COPD. 2014. Available from: http://www.goldcopd.org/
- Di Martino M, Agabiti N, Bauleo L, Kirchmayer U, Cascini S, Pistelli R, et al. Use patterns of long-acting bronchodilators in routine COPD care: the OUTPUL study. COPD 2014; 11(4):414–423.
- Krigsman K, Nilsson JL, Ring L. Refill adherence for patients with asthma and COPD: comparison of a pharmacy record database with manually collected repeat prescriptions. Pharmacoepidemiol Drug Saf 2007; 16:441–448.
- Dolce JJ, Crisp C, Manzella B, Richards JM, Hardin JM, Bailey WC. Medication adherence patterns in chronic obstructive pulmonary disease. Chest 1991; 99:837–841.
- Bosley CM, Parry DT, Cochrane GM. Patient compliance with inhaled medication. Does combining beta agonists with corticosteroids improve compliance? Eur Respir J 1994; 7:504–509.
- Revicki DA, Frank L. Pharmacoeconomic evaluation in the real world: effectiveness versus efficacy studies. Pharmacoeconomics 1999; 15:423–434.
- Mularski RA, Asch SM, Shrank WH, Kerr EA, Setodji CM, Adams JL, et al. The quality of obstructive lung disease care for adults in the United States as measured by adherence to recommended processes. Chest 2006; 130:1844–1850.
- Fano V, D'Ovidio M, del Zio K, Renzi D, Tariciotti D, Agabiti N, et al. The role of the quality of hospital discharge records on the comparative evaluation of outcomes: the example of chronic obstructive pulmonary disease. Epidemiol Prev 2012; 36(3–4):172–179.
- World Health Organization. Collaborating centre for drug statistics methodology, guidelines for ATC classification and DDD assignment 2013 (16th edition). Oslo, Norway: Norwegian Institute of Public Health, 2012.
- Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf 2006; 15:565–574.
- Jenks GF. The data model concept in statistical mapping. Int Yearb Cartogr 1967; 7:186–190.
- Hox J. Multilevel Analysis: Techniques and Applications. Mahwah, NJ: Lawrence Erlbaum Associates, 2002, p. 123.
- Bozdogan H. Akaike's information criterion and recent developments in information complexity. J Math Psychol 2000; 44:62–91.
- Larsen K, Merlo J. Appropriate assessment of neighborhood effects on individual health: integrating random and fixed effects in multilevel logistic regression. Am J Epidemiol 2005; 161:81–88.
- Fantini MP, Compagni A, Rucci P, Mimmi S, Longo F. General practitioners' adherence to evidence-based guidelines: a multilevel analysis. Health Care Manage Rev 2012; 37(1):67–76.
- Austin PC, Tu JV. Automated variable selection methods for logistic regression produced unstable models for predicting acute myocardial infarction mortality. J Clin Epidemiol 2004; 57(11):1138–1146.
- Vik SA, Maxwell CJ, Hogan DB. Measurement, correlates, and health outcomes of medication adherence among seniors. Ann. Pharmacother 2004; 38(2):303–312.
- Faustini A, Marino C, D'Ippoliti D, Forastiere F, Belleudi V, Perucci CA. The impact on risk-factor analysis of different mortality outcomes in COPD patients. Eur Respir J 2008; 32(3):629–636.
- Jochmann A, Neubauer F, Miedinger D, Schafroth S, Tamm M, Leuppi JD. General practitioner's adherence to the COPD GOLD guidelines: baseline data of the Swiss COPD Cohort Study. Swiss Med Wkly 2010; www.smw.ch (Early Online Publication).
- Rutschmann OT, Janssens JP, Vermeulen B, Sarasin FP. Knowledge of guidelines for the management of COPD: a survey of primary care physicians. Respir Med 2004; 98(10):932–937.
- Foster JA, Yawn BP, Maziar A, Jenkins T, Rennard SI, Casebeer L. Enhancing COPD management in primary care settings. Med Gen Med 2007; 9(3):24.
Appendix
Data sources
Hospital information system (HIS)
The HIS includes the patients' characteristics (the patients' identifier, gender, date and place of birth, and place of residence); admission and discharge dates; discharge diagnoses (up to 6); procedure codes (up to 6) according to the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM); hospital ward(s); date(s) of in-hospital transfer; and a regional code that corresponds to the admitting facility.
Mortality information system
The MIS includes the patients' demographic characteristics (the patient's identifier, age, gender, place and date of birth, residence, marital status, and occupation), as well as the date, place, and cause of death (codified by ICD-9 codes).
Drug claims registry
Pharm registry comprises individual records for each medical prescription dispensed in public and private pharmacies within the territory of the local health authorities for the resident population. The registry is limited to those drugs prescribed for outpatient use that are reimbursed by the health care system. All the drugs in the study are included in the Pharm registry. In this registry, the drugs are identified by the national drug registry code, which refers to the international ATC classification and allows for the exact quantification of the dispensed drug. Individual patient data (the patient's identifier) and the date the drug is dispensed are reported for every prescription.
Proxy of COPD severity (retrieved in the 12 months preceding the index admission)
| • | hospitalizations for COPD; | ||||
| • | diagnosis of respiratory failure (ICD-9-CM: 518.81–518.84); | ||||
| • | invasive respiratory procedures: temporary tracheostomy (ICD-9-CM: 31.1), permanent tracheostomy (31.2), continuous invasive mechanical ventilation (96.7), tracheostomy status (V44.0); | ||||
| • | staying in intensive care unit during a COPD hospitalization; | ||||
| • | emergency visits for COPD. | ||||
Concomitant respiratory diseases (retrieved in the 24 months preceding the index admission and in the index admission)
| • | asthma (ICD-9-CM: 49.3); | ||||
| • | chronic respiratory diseases other than COPD (ICD-9-CM: 135, 495, 500–505, 515–517, 519, 508.1, 518.1–518.3); | ||||
| • | pulmonary infections (ICD-9-CM: 011, 480–487.0, 510, 511, 513); | ||||
| • | acute pulmonary symptoms (ICD-9-CM: 512, 415, 786.0, 518.0); | ||||
| • | sleep apnoea syndrome (ICD-9-CM: 780.51, 780.53, 780.57). | ||||
Previous use of drugs (at least one prescription retrieved in the 12 months preceding the index admission)
| • | long-acting bronchodilators: LABA (ATC codes: R03AC12, R03AC13, R03AC18), LAMA (R03BB04) and fixed combinations LABA/ICS (R03AK06, R03AK07)Footnote1*; | ||||
| • | oral corticosteroids (ATC code: H02AB); | ||||
| • | antibacterials (ATC code: J01); | ||||
| • | cardiovascular drugs: cardiac therapies (ATC code: C01), antidiabetic drugs (A10), antiplatelets (B01AC04, B01AC05, B01AC06), antihypertensive drugs (C02, C03, C07, C08, C09) and statins (C10AA). | ||||