Inhaled corticosteroids are more effective in COPD patients when used with LABA than with SABA

Respiratory Medicine (2005) 99, 1115 1124 Inhaled corticosteroids are more effective in COPD patients when used with LABA than with SABA Victor A. Kiri a,b,, Germano Bettoncelli c, Renato Testi d, Giovanni Viegi e a Worldwide Epidemiology GlaxoSmithKline R&D, Greenford, UK b National Centre for Health Outcomes Research, London School of Hygiene and Tropical Medicine, London, UK c Italian Society of General Medicine (SIMG), Florence, Italy d Medical Department, GlaxoSmithKline, Italy e CNR Institute of Clinical Physiology, Pisa, and Italian Society of Respiratory Medicine, Italy Received 12 November 2004 KEYWORDS COPD; Inhaled corticosteroids; Long-acting betaagonists; Short-acting betaagonists; Re-hospitalization; Mortality Summary Adding long-acting beta agonists (LABA) to inhaled corticosteroids (ICS) has been associated with beneficial effects in COPD patients in randomized controlled trials and observational studies. However, it is not known whether adding short-acting beta agonists (SABA) to ICS instead of LABA will be similarly effective in COPD. We compared the effectiveness of combination therapies involving ICS with LABA versus ICS with SABA in reducing risk of re-hospitalization or death among COPD patients within a year of discharge from a first COPD hospitalization. Using the UK General Practice Research Database, we obtained 437 pairs of patients who either used ICS plus LABA or ICS plus SABA, each pair having been matched on disease severity. We found that 12.1% of patients prescribed ICS with LABA experienced rehospitalization or death within 12 months compared with 18.1% among those given ICS with SABA. In multivariate analyses, we found a 38% risk reduction (Po0:007) among patients given ICS with LABA relative to those given ICS with SABA. Models stratified by SABA use generated a risk reduction of 35% (P ¼ 0:02) among those given ICS and LABA with SABA in the 90-day period, and of 49% (Po0:05) among those given ICS and LABA without any SABA compared with the combination users of ICS and SABA. We conclude that in moderate to severe COPD patients, the combined use of ICS with LABA is more effective in reducing the risk of re-hospitalization or death than the combined use of ICS with SABA. & 2005 Elsevier Ltd. All rights reserved. Corresponding author. 0954-6111/$ - see front matter & 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.rmed.2005.02.018

1116 Introduction Chronic obstructive pulmonary disease (COPD) is a multi-component respiratory disorder characterized by deterioration in lung function in the form of negative changes to expiratory flow. COPD is one of the fastest growing diseases in the world, and although it is currently only the sixth leading cause of death with more than 2.5 million deaths per year worldwide, it is the only major disease with an increasing annual trend of mortality. 1 Indeed, COPD is estimated to be the third most important cause of death by 2020. 2 The projected increase in morbidity and mortality due to COPD is likely to result in substantial increase in the burden of the disease in terms of economic costs. For instance, episodes of exacerbation of COPD, generally associated with progression of the disease, are responsible for many emergency hospital admissions, which in the UK have risen by 50% in the last 5 years. 3 In Italy, COPD is the respiratory disease with the biggest number of hospital discharges, and ranks 7th overall among all diseases. 4,5 In Italy, it has been estimated that mean cost per patient/year for patients diagnosed with COPD between 1999 and 2000 ranged from 1500 to 3912 Euros, depending on illness severity. 6 A study has shown that exacerbations are the key drivers of the costs of COPD in Sweden accounting for between 35 and 45% of the total per capita health-care costs for COPD. 7 A number of studies have found evidence that inhaled corticosteroids (ICS) might have an attenuating effect on airway hyper-responsiveness, in addition to helping to reduce exacerbations among moderate to severe COPD patients. 8 11 The recently produced ATS/ERS COPD treatment guidelines now recommend long-term use of ICS for COPD patients with post-bronchodilator forced expiratory volume in 1 s (FEV 1 ) o50% predicted and repeated exacerbations and as many as 48% of COPD patients in the UK are currently receiving long-term ICS therapy. 12,13 Evidence from a recent clinical trial indicates a combined effect between fluticasone propionate (an inhaled corticosteroid) and salmeterol (a long acting beta-agonist, LABA) in reducing COPD exacerbations and other clinical outcomes. 14 Two recent observational studies of combined inhaled corticosteroids and LABAs, among moderate to severe COPD patients, have reported reduced risk of mortality and rehospitalization. 15,16 Both studies were based on the UK General Practice Research Database (GPRD), an automated database of primary care covering a total population in excess of 3.4 million inhabitants (approximately 5.7% of the population) in the UK. The primary aim of the current study was to explore whether the benefits achieved with combining Short-acting beta agonists (SABA) and ICS are similar to the observed beneficial effects of adding a LABA to ICS in COPD patients. Therefore, using the GPRD, we conducted a study to compare the effectiveness of combination therapies involving inhaled corticosteroids with either long-acting beta-agonists or short-acting beta-agonists in reducing the risk of re-hospitalization or death among COPD patients within a year of discharge from a first COPD hospitalization. Materials and methods Database All data were obtained from the GPRD. The GPRD was originally developed in 1987 with the aim of establishing a longitudinal database of medical records in the UK. The database is managed by the Medicines and Healthcare Products Regulatory Agency (MHRA). It has been described elsewhere, 17 21 and has been used in other published epidemiological studies on COPD. 15,16,21,22 In line with previous GPRD-based studies on COPD, we identified retrospectively all male and female newly physician-diagnosed COPD patients aged 50 years and older from 1990 to 1999. The study protocol was approved by the Scientific and Ethical Advisory Group of the GPRD. Study cohort V. A. Kiri et al. The cohort was similar to those used in two previously published retrospective studies; one based on the Ontario database 10 and the other on the GPRD 16 (Fig. 1). As in the previous GPRD-based study, the cohort was restricted to newly diagnosed COPD patients in the GPRD and cohort entry defined at date of discharge from a first hospitalization for a COPD-related condition. In addition, cohort members were required to have received at least one prescription for inhaled corticosteroids and at least another prescription for either LABAs or SABAs from their general practices (GPs) in the first 90 days after discharge. No restrictions were placed on the use of other respiratory drugs and the criteria for hospitalization for a COPD-related condition included codes for pneumonia and chest infection. However, patients with no record of GP contact within any 12-month period were excluded. To ensure equitable definition of drug exposure, an immortal person time-period of

Inhaled corticosteroids in COPD patients 1117 Incident COPD First hospitalization for COPD Discharge date Respiratory Pharmacology 90 days COPD related re-hospitalization 275 days Death End of respiratory pharmacology period for determining patient s drug group Figure 1 Study design: patients with events in the first 90 days following discharge from hospital are excluded. 90-days was effected, such that patients who died or were re-hospitalized or lost to follow-up within 90 days of discharge from hospital were excluded. 23 25 Drug exposure and outcome Every patient in the study received at least one prescription of ICS in the 90-day period following the patient s discharge from first COPD-related hospitalization. Over the same period, those who also received at least one prescription of a LABA constituted the ICS plus LABA group, whilst the others who received at least one prescription of SABAs made up the ICS plus SABA group. Thus, patients who were given prescriptions of both LABAs and SABAs over the said period were assigned to the ICS plus LABA group. The outcome of interest was the occurrence of a first severe COPD exacerbation, defined here as either re-hospitalization for a COPD-related medical condition or death. Followup was restricted to a maximum period of 365 days from discharge date. The choice of 365 days was to ensure consistency with other previously published studies on the subject matter, and to command sufficient power to analyze the research hypothesis as well as to maintain completeness of the database. Study design We used a propensity scores-matched cohort design. The propensity scores were derived from the patient-level baseline characteristics on asthma status, smoking, age, comorbidities, prior respiratory medications and gender. We then matched users of ICS plus LABA to users of ICS plus SABA with equal probabilities of being prescribed ICS plus LABA. In a randomized trial, the randomization of subjects to different treatment groups guarantees that, on average, there should be no systematic differences (i.e. bias) between treatment groups. In observational studies however, investigators do not assign the treatment as in a randomized trial. Therefore, large differences in observed covariates in the two treatment groups may exist, and these differences could lead to biased estimates of treatment effects. The propensity score is the conditional probability of assignment to a particular treatment given a set of observed covariates and is increasingly being used to adjust for nonrandom treatment assignment. 26 28 Statistical analysis Baseline characteristics of the two drug exposure groups (i.e. ICS plus LABA versus ICS plus SABA) were compared using standard univariate statistical techniques. For age at first COPD-related hospitalization, being a continuous variable, we compared the mean and standard deviation for the groups using Fisher s exact method. For the binary and ordinal variables, w 2 tests were performed. Using standard survival techniques such as Kaplan Meier estimation and the Cox proportional hazards model, we compared the event rates between the two treatment groups. For this purpose, duration of follow-up was defined as the time-period between discharge from first COPDrelated hospitalization and either the next COPDrelated hospitalization, death or censoring (i.e. end of record on database and hence considered lost to follow-up). A number of baseline characteristics (including age at first COPD-related hospitalization)

1118 were treated as potential confounders to obtain adjusted hazard ratios (as relative risks). Confounders Asthma: A reasonable number (more than 40%) of COPD patients in the GPRD have asthma mentioned in their record. Indeed there is evidence that compared with non-asthmatics, asthma patients have 17-times-higher risk of receiving a diagnosis of emphysema and as much as 12.5-times-higher risk of fulfilling COPD diagnostic criteria regardless of smoking history. 29 V. A. Kiri et al. Baseline medications: These were other respiratory medications used in the 90-day therapy definition period and were treated as nine distinct class variables (see Table 1). For instance, use of oral corticosteroids being a potential effect modifier of the association between therapy and mortality was a binary yes/no covariate. 30 Smoking: Information on tobacco use was categorized as either non-smoker (to represent never and ex-smoker), current-smoker or unknown based on patient s records in both the medical and prevention databases. Table 1 Descriptive characteristics of COPD patients with ICS and LABA versus those with ICS and SABA within the first 90 days of discharge from a first COPD-related hospitalization before and after matching by propensity scores (with 90-day event-free restriction). Before matching After matching ICS & SABA ICS & LABA ICS & SABA ICS & LABA Total 2112 All 445 437 All 437 No SABA 102 No SABA 101 With SABA 343 With SABA 336 Mean age (SD) 71.6 (9.0) 68.5 (8.3)* 69.1 (8.6) 68.5 (8.3) Percent Male 50.6 49.4 49.7 49.4 Current smoker 54.9 54.4 59.0 55.2 Asthma recorded previously 69.1 85.2 85.8 84.9 Baseline respiratory drug use in the 90 days prior to first COPD-related hospitalization Inhaled steroids 100.0 100.0 100.0 100.0 Inhaled B2 100.0 100.0 100.0 100.0 Oral B2 4.6 5.2 3.4 5.3 Xanthynes 17.3 23.6* 20.6 23.6 Anticholinergics 22.5 30.3* 28.4 30.0 Oral steroids 35.8 52.6* 51.7 51.7 Non-ICS combination products 2.9 5.2* 3.4 5.3 Oxygen 7.7 8.8 8.2 8.9 Nebulized 2.5 5.2* 3.4 5.0 Nasal corticosteroids 4.2 3.2 4.8 3.0 Antihistamines 2.4 2.7 3.0 2.7 Baseline comorbidity Myocardial infarction 9.9 8.3 7.8 8.5 Congestive heart failure 25.4 17.8* 17.2 17.8 Peripheral vascular disease 2.7 2.9 2.3 5.3* Cerebrovascular disease 3.8 2.7 3.0 2.7 Dementia 0.9 0.2 0.2 0.2 Rheumatologic disease 1.1 1.3 1.1 0.9 Peptic ulcer 7.1 9.7 7.1 9.6 Mild liver disease 0.0 0.0 0.0 0.0 Diabetes 5.2 7.2 5.9 7.1 Diabetes with complication 0.0 0.0 0.0 0.0 Hemiplegia or paraplegia 1.2 0.2 0.9 0.2 Renal disease 1.1 1.3 1.1 1.4 Cancer 0.8 0.4 0.5 0.5 Moderate/severe liver disease 0.2 0.9* 0.2 0.7 AIDS 0.0 0.0 0.0 0.0 *Po0:05 in comparison between groups.

Inhaled corticosteroids in COPD patients 1119 Comorbidities: Comorbidities were identified from patients medical records prior to the end of the therapy definition window and were based on the 15 Charlson comorbidity groups. 31 Year of entry: The year of entry into the drug exposure category was considered as a continuous covariate in the analysis with the first year as 0. Results We identified 2557 patients, each with records of a first COPD-related hospitalization, followed by at least one prescription of ICS and at least another for either LABAs or SABAs or both in the first 90 days of discharge (Table 1). The ICS plus LABA group was comprised of 445 patients, with 343 (77%) having also had prescriptions for SABAs in the 90-day treatment definition period. As per treatment group inclusion criteria, none of the 2112 patients who constituted the ICS plus SABA group received any prescription for (LABA) in the 90-day treatment definition period. The proportions of patients with prescriptions for the various other respiratory medications were general higher in the ICS plus LABA group, possibly indicating difference in the severity of COPD between the two groups. However, the resulting propensity scores-matched pairs of patients, 437 in each group were balanced on the various variables, except on co-existing peripheral vascular disease. The groups were also balanced on quarterly oral corticosteroids prescriptions they indicated a pattern of gradual reduction over the year since discharge from the COPD-related hospitalization (Table 2). Survival Over the 1-year follow-up period, re-hospitalization or death occurred in 132 patients (15.1%) overall, 12.1% among users of ICS plus LABA (regardless of use of SABA) compared with 18.1% among those who used ICS plus SABA (and no LABA) (Po0:05). Re-hospitalization occurred in 6.4% of the users of ICS plus LABA compared with 10.1% among users of ICS plus SABA. Similarly, death occurred in 6.9% of the users of ICS plus LABA compared with 9.6% among the users of ICS plus SABA. The results from a Cox survival analysis with adjustments for possible confounders showed that use of ICS plus LABA was associated with 38% fewer events of re-hospitalization or death compared to ICS plus SABA use (relative risk ¼ 0.62, 95% confidence interval 0.43 0.87). The survival functions for the two treatment groups are presented in Fig. 2. Estimates for the other risk factors for rehospitalization or death are shown in Fig. 3 and include sex, age, comorbid conditions (including asthma), and use of oral corticosteroids. Since almost 70% of the patients in the ICS plus LABA group also received prescriptions for SABA in the 90-day treatment definition period, a stratified Cox analysis was also conducted. The results revealed a risk reduction of 49% among those who used ICS plus LABA and no SABA (relative Table 2 Mean (SD) prescriptions of inhaled corticosteroids, long-acting beta-agonists, short-acting betaagonists, and oral corticosteroids by drug exposure group, per quarter in the propensity scores-matched cohort. 1st Quarter 2nd Quarter 3rd Quarter 4th Quarter Number ICS & SABA 437 437 437 437 ICS & LABA 437 437 437 437 ICS ICS & SABA 2.40 (1.34) 1.83 (1.53) 1.60 (1.55) 1.42 (1.65) ICS & LABA 2.54 (1.45) 2.08 (1.53)* 1.80 (1.58) 1.54 (1.52) LABA ICS & SABA 0.00 (0.00) 0.07 (0.39) 0.11 (0.54) 0.08 (0.42) ICS & LABA 2.36 (1.25)* 2.02 (1.53)* 1.66 (1.49)* 1.40 (1.53)* SABA ICS & SABA 3.41 (2.19) 2.75 (2.33) 2.54 (2.39) 2.18 (2.49) ICS & LABA 2.62 (2.40)* 2.23 (2.21)* 1.96 (2.10)* 1.68 (2.11)* OCS ICS & SABA 1.15 (1.47) 0.79 (1.33) 0.77 (1.29) 0.68 (1.20) ICS & LABA 1.16 (1.62) 0.93 (1.58) 0.86 (1.54) 0.70 (1.45) *Po0:05 in comparison between groups.

1120 risk ¼ 0.51, 95% confidence interval 0.26 0.99), and a risk reduction of 35% among those who used ICS plus LABA and some SABA (relative risk ¼ 0.65, 95% confidence interval 0.44 0.94) relative to those given only ICS and SABA. When re-hospitalization only was considered as the outcome of interest, we found similar evidence of risk reductions in favour of use of ICS plus LABA over use of ICS plus SABA. Figure 4 gives the results from a Cox survival analysis involving adjustments for possible confounders. They showed that use of ICS plus LABA was associated with 42% fewer events of re-hospitalization than use of ICS plus SABA (relative risk ¼ 0.58, 95% confidence interval 0.36 0.95). Furthermore, when death was considered as the outcome of interest, we also found Survival of COPD re-hospitalisation or death 1.0 0.9 0.8 0.7 0.6 0.5 ICS/LABA combination ICS/SABA combination 0 30 60 90 120 150 180 210 240 270 300 330 360 Days after discharge Figure 2 Adjusted Cox survival function of time to rehospitalization or death in COPD patients using inhaled corticosteroids (ICS) with either long-acting beta-agonists (LABA) or short-acting beta-agonists (SABA). similar evidence of risk reductions in favour of use of ICS plus LABA over use of ICS plus SABA. Figure 5 gives the results from a Cox survival analysis involving adjustments for possible confounders. They showed that use of ICS plus LABA was associated with 40% fewer deaths than use of ICS plus SABA (relative risk ¼ 0.60, 95% confidence interval 0.37 0.98). Discussion V. A. Kiri et al. In this study we found that prescriptions of ICS plus LABA versus ICS plus SABA, in the 90 days following discharge for a COPD-related hospitalization, significantly reduced the risk of re-hospitalization or death by as much as 38% (13 57%) for COPD patients in the 12 months following discharge from hospital after adjusting for potential confounding variables. A bigger level of risk reduction was observed among those given ICS plus LABA in the absence of SABA, when compared with those given ICS plus SABA only. When the risk of re-hospitalization alone and death alone were separately assessed, we also found evidence of reduced risk within a year of discharge among those who were given ICS and LABA with or without SABA compared with those who were given ICS and SABA without LABA. The magnitude of the reported risk reductions may have a large impact on the overall burden of the disease in terms of public health, especially in light of the estimation by Murray and Lopez that COPD is expected to become the third leading cause of death worldwide in 2020. 32 Thus, evidence of a better medication regimen capable of reducing COPD-related hospitalization could assist efforts to 4 Sex Year Smoking Comorbidy Respiratory medications 3.5 3 Hazard Ratio 2.5 2 1.5 1 0.5 0 Male year Unknown Current Peripherial Vascular Disease Asthma OCS Non ICS Combined product Nebulized ICS & LABA (no SABA) ICS & LABA & SABA Figure 3 Multivariate Cox survival analysis of time to re-hospitalization or death in COPD patients using inhaled corticosteroids (ICS) with either long-acting beta-agonists (LABA) or short-acting beta-agonists (SABA) without LABA.

Inhaled corticosteroids in COPD patients 1121 Hazard Ratio 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Sex Smoking Comorbidy Respiratory medications Male Unknown Current Peripheral Vascular Dis Asthma OCS Anticho linergics Non ICS Combined product Nebulized ICS & LABA Figure 4 Multivariate Cox survival analysis of time to re-hospitalization in COPD patients using inhaled corticosteroids (ICS) with either long-acting beta-agonists (LABA) or short-acting beta-agonists (SABA) without LABA. Hazard Ratio 6 5.5 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Sex Year Smoking Age Comorbidy Respiratory medications Male Unknown Current Aged60-69 Aged 70-79 Aged 80+ PeripheralVascularDis CVD Rheumotologic Dis Asthma OCS NonICS Combined product Nebulized ICS & LABA Figure 5 Multivariate Cox survival analysis of time to death in COPD patients using inhaled corticosteroids (ICS) with either long-acting beta-agonists (LABA) or short-acting beta-agonists (SABA) without LABA. reduce the burden of the disease in terms of costs to payers. This study is a replication of two other observational studies, both of which reported reductions in the risk of re-hospitalization or death among COPD patients prescribed therapy of ICS with or without LABA. 10,16 However by applying a 90-day immortal time period on a propensity scores-matched cohort using scores derived from a list of possible confounding variables, ours attempted to minimize bias from two possible sources to which the previous studies were susceptible, namely, selection and waiting-time bias. Waiting-time bias can result in studies where the time to initiation of treatment is included in the follow-up analysis. 24 Although a number of studies have failed to show a significant effect of ICS on annual decline in FEV 1, 33 a recent systematic review of eight trials data has reported a slowing of about 8 ml/year in FEV 1 decline with ICS use. 34 The effectiveness of ICS on health status and in reducing the number or severity of exacerbations has also been reported elsewhere. 35,36

1122 Our study confirms the finding of the previous GPRD-based study that combining LABA with ICS enhances the benefit of ICS with or without SABA. In our case, the observed beneficial association of ICS plus LABA with reduction in the risk of rehospitalization is of high statistical and clinical significance. To our knowledge this is the first study that assesses the possible beneficial effect of concomitant ICS and LABA compared with concomitant ICS and SABA on re-hospitalization in COPD patients treated in general practice. ICS and LABAs have a well-established role in the treatment of asthma. 37 By contrast, the position in COPD has been inconclusive. Nevertheless, the findings of the present study support the evidence of improved outcomes from combining ICS and LABA in treating COPD as reported elsewhere in a previous GPRD-based study 16 as well as the evidence of beneficial effect reported for ICS in COPD treatment reported from a study of a different healthcare system. 10 One of the strengths of this study is the threefold approach to guarantee the reliability of physician-diagnosed COPD. We first identified the diagnosis from the computerized general practice record in the GPRD using at least one of the terms for COPD, which have already been reported as valid in distinguishing COPD from asthma in the GPRD. 37 The diagnosis was then subsequently confirmed by a hospital doctor on discharge from hospital for a COPD-related admission and as further safeguard only patients over 50 years old were included in the study. Other demographic features such as smoking history were supportive. Indeed, the high mortality observed in our study population would by itself make frequent misdiagnosis of asthma unlikely. A previous GPRD-based study which used these procedures has reported mortality rate among moderate to severe COPD patients as between 114 398 per 1000 personyears, compared with a mere 52 per 1000 personyears among non-copd patients of similar age. 38 Thus, these procedures and statistics have helped to establish the validity of the diagnosis of COPD in this GPRD cohort. The GPRD is one of the few sources for investigating the implications of therapy on disease management, using data from real-life patients. It is one of the largest population-based medical databases in the world. For instance, it has an advantage over health maintenance organization databases in the USA where elderly people are frequently lost to follow-up. To maintain the GPRD, GPs from participating surgeries enter morbidity events for each patient in the database using a medical coding system (OXMIS 20 and Read codes). V. A. Kiri et al. This is irrespective of whether the event occurred in the surgery, at a visit, or was relayed via telephone. Symptoms, procedures, investigations, diagnoses, prescriptions, referrals, and follow-ups are all recorded. A similar database is managed by the Italian Society of General Medicine (SIMG) and recent publications have also resulted from the database. 39 42 Indeed, a study of the database, involving the Italian Society of Respiratory Medicine, is in progress to assess the effectiveness of ICS plus LABA compared to ICS plus SABA in the treatment of Italian COPD patients on similar outcomes such as mortality and hospitalization. The study will allow us to check in another European country with a different climate, the validity of the findings of the current study. It is important to point out that the latter reflects the implications of treating real-life COPD patients outside the rigorous randomized clinical trial setting where patients are usually highly selected and it is vital to ensure maintenance of the medication regimes under scrutiny. Nevertheless, compared with data from a randomized clinical trial, the GPRD lacks valid, complete information on important variables like respiratory function, body weight, alcohol and tobacco consumption. Differential diagnosis or drug dosages are often difficult to assess. However, the major diagnoses and information on use of health services and medication are registered with a reasonable, acceptable degree of validity. 20 Indeed, the drawbacks of the database are adequately offset by its large sample size and extensive content in terms of information on medical and therapy events, which are available from practices in all parts of the UK. The lack of randomization means that, as with most similar observational studies, bias due to confounding by indication was a potential weakness of our study, namely that treatment for some patients might have been chosen because of the presence or absence of specific features of the disease. 43 We accept that such a bias was potentially capable of reducing or enhancing the apparent treatment effect since severity of the disease or indeed other disease characteristics such as comorbid asthma could affect both choice of treatment and prognosis. However, in our study, the restriction of the age group to 50 years and older would make misdiagnosis of asthma as COPD unlikely, even after allowing for the possibility of late onset of asthma. Furthermore, the results were adjusted for comorbidities to minimize the impact of disease severity and from the prescription records for the two treatment groups and we could find nothing to suggest that the patients in the ICS plus SABA group

Inhaled corticosteroids in COPD patients 1123 received less attention from their GPs. By matching both treatment groups on certain relevant recorded baseline variables, we have also reduced the possibility of selection bias. Indeed, based on the oxygen and nebulized treatments received in the unmatched full cohort, those given ICS plus LABA might be more severely ill than those prescribed ICS and SABA without LABA. 44 The present study has confirmed improved outcomes with the combination therapy of ICS with LABA relative to ICS with SABA, in terms of reduced risks of re-hospitalization or death or both within 12 months among users with moderate to severe COPD treated in the UK general practice setting. If confirmed in randomized controlled trials, the beneficial associations reported in this observational study would clearly bring health benefits to COPD patients, which in turn should mean reductions in both the costs of management of the disease as well as its overall burden on the society. Acknowledgements Victor A. Kiri and Renato Testi are currently employees of GlaxoSmithKline R&D. There is no other conflict of interest in this research. References 1. Murray CJL, Lopez AD, Mathers CD, Stein C. The global burden of disease 2000 project: aims, methods and data sources. Global Programme on Evidence for Health Policy Discussion Paper No. 36, World Health Organization, November 2001. 2. Friedman M, Hilleman DE. Economic burden of chronic obstructive pulmonary disease. Impact of new treatment options. Pharmacoeconomics 2001;19:245 54. 3. LAIA: Lung & Asthma Information Agency. Trends in emergency hospital admissions for lung disease. Factsheet 99/2, St. George s Hospital Medical School. 4. Viegi G, Scognamiglio A, Baldacci S, Pistelli F, Carrozzi L. Epidemiology of chronic obstructive pulmonary disease (COPD). Respiration 2001;68:4 19. 5. Viegi G. Epidemiology of COPD: a European perspective. In: Burney P, Suissa S, Soriano JB, editors. In the pharmacoepidemiology of COPD: recent advances and methodological discussion; Eur Respir J. 2003;22(Suppl. 43):3 7. 6. Dal Negro R, Berto P, Tognella S, Quareni L. Global outcomes in Lung Disease Study Group. Cost-of-illness of lung disease in the TriVeneto Region, Italy: the GOLD Study. Monaldi Arch Chest Dis 2002;57:3 9. 7. Andersson F, Borg S, Jansson SA, et al. The costs of exacerbations in chronic obstructive pulmonary disease (COPD). Respir Med 2002;96:700 8. 8. Burge PS, Calverley PM, Jones PW, Spencer S, Anderson JA, Maslen TK. Randomised, double blind, placebo controlled study of fluticasone propionate in patients with moderate to severe chronic obstructive pulmonary disease: the ISOLDE trial. Br Med J 2000;320:1297 303. 9. The Lung Health Study Research Group. Effect of inhaled triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease. N Engl J Med 2000;343. 10. Sin DD, Tu JV. Inhaled corticosteroids and the risk of mortality and readmission in elderly patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;164:580 4. 11. Sin DD, Man SFP. Inhaled corticosteroids and survival in chronic obstructive pulmonary disease: does the dose matter? Eur Respir J 2003;21:260 6. 12. Celli BR, MacNee W, committee members. Standards for the diagnosis and treatment of patients with COPD: a summary of the ATS/ERS position paper. Eur Respir J 2004;23:932 46. 13. O Brien C, Guest PJ, Hill SL, Stockley RA. Physiological and radiological characterisation of patients diagnosed with chronic obstructive pulmonary disease in primary care. Thorax 2000;55:635 42. 14. Calverley P, Pauwels R, Vesbo J, for the TRISTAN Study Group, et al. Combined salmeterol and fluticasone in the treatment of chronic obstructive pulmonary disease: a randomised controlled trial. Lancet 2003;361:449 56. 15. Soriano JB, Vestbo J, Pride NB, Kiri V, Maden C. Maier WC survival of COPD patients after regular use of fluticasone propionate salmeterol in primary care in the UK. Eur Respir J 2002;20:819 25. 16. Soriano JB, Kiri VA, Pride NB, Vesto J. Inhaled corticosteroids with/without long-acting beta-agonist reduce the risk of rehospitalization and death in COPD patients. Am J Respir Med 2003;2:67 74. 17. Lawson DH, Sherman V, Hollowell J. The general practice research database. Scientific and Ethical Advisory Group. QJ Med 1998;91:445 52. 18. Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resources in the UK. Br Med J 1991;302:766 8. 19. Garcia Rodriguez LA, Perez Gutthann S. Use of the UK General Practice Research Database for pharmacoepidemiology. Br J Clin Pharmacol 1998;45:419 25. 20. Nazareth I, King M, Haines A, Rangel L, Myers S. Accuracy of diagnosis on general practice computer system. Br Med J 1993;307:32 4. 21. Hansell A, Hollowell J, Nichols T, McNiece R, Strachan D. Use of the General Practice Research Database (GPRD) for respiratory epidemiology: a comparison with the 4th Morbidity Survey in General Practice (MSGP4). Thorax 1999;54:413 9. 22. Soriano JB, Maier WC, Egger P, et al. Recent trends of physician-diagnosed COPD in women and men in the UK. Thorax 2000;55:789 94. 23. Rothman KJ, Greenland S. Modern epidemiology. Philadelphia: Lippincott-Raven Publishers; 1998. 24. Suissa S. Effectiveness of inhaled corticosteroids in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2003;168:49 53. 25. Rothman KJ. A potential bias in safety evaluation during open-label extensions of randomized clinical trials. Pharmacoepidemiol Drug Saf 2004;13:295 8. 26. D Agostino RB. Tutorial in biostatistics propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265 81. 27. Morant SV, Pettitt D, MacDonald TM, Burke TA, Goldstein JL. Application of propensity score to adjust for channelling bias with NSAIDS. Pharmacoepidemiology and drug safety 2004;13:345 53.

1124 V. A. Kiri et al. 28. Frolkis JP, Pthier CE, Blackstone EH, Lauder MS. Frequent ventricular ectopy after exercise as a predictor of death. N Engl J Med 2003;348 349. 29. Silva GE, Sherrill DL, Guerra S, Barbee RA. Asthma as a risk factor for COPD in a longitudinal study. Thorax 2004;126: 59 65. 30. Schols AM, Wesseling G, Kester AD, et al. Dose dependent increased mortality risk in COPD patients treated with oral glucocorticoids. Eur Respir J 2001;17:337 42. 31. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613 9. 32. Murray CJL, Lopez AD. Alternative projections of mortality and disability by cause 1990 2020. Lancet 1997;349: 1498 504. 33. Barnes PJ. Inhaled steroids are not beneficial in COPD. Am J Respir Crit Care Med 2000;161:342 4. 34. Sutherland ER, Allmers H, Ayas NT, Venn AJ, Martin RJ. Inhaled corticosteroids reduce the progression of airflow limitation in chronic obstructive pulmonary disease: a metaanalysis. Thorax 2003;58:937 41. 35. Calverley PM. Inhaled steroids are beneficial in COPD. Am J Respir Crit Care Med 2000;161:341 2. 36. Alsaeedi A, Sin DD, McAlister FA. The effects of inhaled corticosteroids in chronic obstructive pulmonary disease: a systematic review of randomized placebo-controlled trials. Am J Med 2002;113:59 65. 37. Barnes PJ. Scientific rationale for inhaled combination therapy with long-acting B2-agonists and corticosteroids. Eur Respir J 2002;19:182 91. 38. Soriano JB, Maier WC, Visick G, Pride N. Validation of general practitioner-diagnosed COPD in the UK General Practice Research Database. Eur J Epidemiol 2001;17: 1075 80. 39. Cricelli C, Mazzaglia G, Samani F, et al. Prevalence estimates fro chronic diseases in Italy: exploring the differences between self-report and primary care databases. J Public Health Med 2003;25(3):254 7. 40. Filippi A, Buda S, Brignoli O, Cricelli C, Esposti ED. Global cardiovascular risk evaluation in Italy: a cross-sectional survey in general practice. Ital Heart J 2004;5(3):223 7. 41. Filippi A, Bignamini AA, Sessa E, Samani F, Mazzaglia G. Secondary prevention of stroke in Italy: a cross-sectional survey in family practice. Stroke 2003;34:1010 4. 42. Mazzaglia G, Caputi AP, Rossi A, Bettoncelli G, Stefanini G, Ventriglia G, Nardi R, Brignoli O, Cricelli C. Exploring patient- and doctor-related variables associated with antibiotic prescribing for respiratory infections in primary case. Eur J Clin Pharmacol 2003;59:651 7. 43. Strom BL. Pharmacoepidemiology, 3rd ed. Chichester: Wiley; 2000. 44. Vesbo J. Another piece of the inhaled corticosteroids in COPD puzzle. Am J Respir Crit Care Med 2001;164: 315 514.