Changes in NHS organization of care and management of hospital admissions with COPD exacerbations between the national COPD audits of 2003 and 2008

Q J Med 2011; 104:859 866 doi:10.1093/qjmed/hcr083 Advance Access Publication 26 May 2011 Changes in NHS organization of care and management of hospital admissions with COPD exacerbations between the national COPD audits of 2003 and 2008 P.M. GEORGE 1, R.A. STONE 2,3, R.J. BUCKINGHAM 2, N.A. PURSEY 2, D. LOWE 2 and C.M. ROBERTS 2,4 From the 1 Chest and Allergy Department, St Mary s Hospital, Imperial College Healthcare NHS Trust, London W2 1NY, 2 Clinical Effectiveness and Evaluation unit, Royal College of Physicians, London NW1 4LE, 3 Taunton and Somerset NHS Foundation Trust, Musgrove Park Hospital, Taunton TA1 5DA and 4 Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London E1 2AD, UK Address correspondence to Prof. C.M. Roberts, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London E1 2AD, UK. email: c.m.roberts@qmul.ac.uk Received 12 January 2011 and in revised form 26 April 2011 Summary Background: The 2003 UK Chronic Obstructive Pulmonary Disease (COPD) audit revealed wide variability between hospital units in care delivered. Aims: To assess whether processes of care, patient outcomes and organization of care have improved since 2003. Design: A UK national audit was performed in 2008 to survey the organization and delivery of clinical care provided to patients admitted to hospital with COPD. Methods: All UK acute hospital Trusts (units) were invited to participate. Each unit completed cross-sectional resource and organization questionnaires and a prospective clinical audit comprising up to 60 consecutively admitted cases of COPD exacerbation. Comparison between 2003 and 2008 includes aggregated statistics for units participating in both audit rounds. Results: A total of 192 units participated in both audit rounds (6197 admissions in 2003 and 8170 in 2008). In 2008, patients were older and of a poorer functional class. Overall mortality was unchanged but adjusting for age and performance status, inpatient mortality (P = 0.05) and 90-day mortality (P = 0.001) were both reduced in 2008. More patients were discharged under a respiratory specialist (P < 0.01), treated with non-invasive ventilation if acidotic (P < 0.001) and accepted onto early discharge schemes (P < 0.01) while median length of stay fell from 6 to 5 days (P < 0.001). Within these mean data, however, there remains considerable inter-unit variation in organization, resources and outcomes. Conclusions: Overall improvements in resources and organization are accompanied by reduced mortality, shorter admissions and greater access to specialist services. There remains, however, considerable variation in the quality of secondary care provided between units. Background Chronic Obstructive Pulmonary Disease (COPD) is a condition that carries a high morbidity and mortality for sufferers and places a large burden on the acute care services of the NHS. 1,2 The 2003 UK COPD audit demonstrated significant variability between units in process of care and outcomes. 3,4! The Author 2011. Published by Oxford University Press on behalf of the Association of Physicians. All rights reserved. For Permissions, please email: journals.permissions@oup.com

860 P.M. George et al. Since this audit, data for individual participants in the 2003 audit was reported back to individual units, benchmarked against national and regional performance and new UK management guidelines have been published by the National Institute for Clinical Excellence (NICE). 5 In 2008, the National COPD Audit was repeated providing a range of data gathered from UK hospitals including most that participated in the 2003 round. This article compares the standards of care in the 2008 audit with that of 2003 with particular reference to the resources available, the organization of care and the clinical care of patients admitted with exacerbations of COPD. Methods A UK wide audit of the secondary care of patients admitted to hospital with COPD exacerbations was conducted in both 2003 and 2008. All acute NHS Trusts where patients were admitted with a primary diagnosis of COPD were invited to participate on a voluntary basis. Hospital Trusts were asked to participate as an operational unit and subsequent results refer to these as units. A unit was defined as an operational grouping that functioned as a single respiratory team responsible for the care of COPD patients, meaning that a single NHS Trust might participate either as a single unit or as more than one unit where more than one acute admitting sites were located within a Trust. Units were asked to collect two separate data sets. The first was a single cross-sectional survey of the resources and organization of COPD care in that unit. The specific data items included some that were common to both audits and others that were asked either just for the 2003 or for the 2008 audit that reflected new or changed standards over this time period. The analysis presented here refers only to items common to both audits. The second data set was a record of clinical care of consecutive admissions prospectively identified but audited retrospectively at 90 days after the index admission. In the 2003 audit, units were asked to collect data on up to 40 admissions, but in 2008 this number was increased to 60 cases to provide more meaningful data for individual units. Patients were eligible for the clinical case audit if they had a clinically made diagnosis of COPD confirmed on the post-take ward round. To improve accurate diagnosis of COPD exacerbation, the lead clinician at each unit was encouraged to review the medical notes of audit patients to check for any evidence of misdiagnosis (i.e. diagnosis appeared to be COPD on admission but later deemed incorrect). These patients were excluded, as were patients whose diagnosis was changed to exacerbation of COPD from another presenting condition and no substitute cases were used. For patients having more than one admission within the data collection period, only the index admission was included in the analysis. For the National COPD Audit 2008, an admission was defined as an episode in which a patient with an acute COPD exacerbation was admitted to a ward and stayed in hospital for 4 hours or more prior to discharge or acceptance to an early discharge scheme. This includes Emergency Medicine Centres/Medical Admission Units or similar but excludes Accident and Emergency Units. In 2003, the definition was of a stay in hospital of 12 h or more or acceptance onto an early discharge scheme (EDS). The difference in definitions reflected the change in the NHS standards for wait times in accident and emergency departments before a patient should be admitted or discharged. In the 2008 audit, data were entered into a webbased data collection tool by local staff and overseen by a Respiratory Consultant within each unit. In-built logical checks and prompts helped enhance data quality and completeness. In 2003, the data tool was designed around a Microsoft Access database sent to units as a CD Rom and then returned as either an email file or as a CD copy. Help-notes were provided in both paper and electronic forms. Nineteen units lacking in IT resource made paper returns transcribed into the database by the Royal College of Physicians central administrative staff. The cross-sectional resources and organizational audit was completed between 3rd March and 16th May 2008. Patients for inclusion in the clinical audit of COPD exacerbations were prospectively identified between 3rd March and 16th May 2008. Outcome data at 90 days following the index admission was also collected and all clinical data was submitted by 1st September 2008. In 2003, the resources and organizational audit questionnaire was completed between August and October 2003 and the clinical data collection took place between 1st September and 14th November 2003 with the collection of outcome data at 90 days. This article compares results from the 2008 and 2003 audits. Only units participating in both audits were analysed so as to make time trend analyses relevant. Some units were lost from analysis because of changes either in unit re-configuration due to Trust-level re-organization or changes in how a Trust wished to participate as a unit within the audit. Missing data are reflected in this article by variations in denominators. Results are presented at patient level and at unit level, with unit-level

National COPD audits of 2003 and 2008 861 results being compared between audits using paired tests (Wilcoxon, McNemar), the pairing reflecting unit participation in both audits. The main outcome measures were length of stay, death and readmission rates within 90 days of the index audit admission, with death further analysed as death in hospital, and death within 90 days of the index admission. Length of stay statistics excluded patients who had died within the index admission period. Attempts were made to assess correlation between changes in organization, processes of care and outcomes over the course of the two audits. For numerical data, Spearman s correlation was sought and for categorical data, the Kruskal Wallis test was applied. Both linear regression and median regression methods were used in unit-level analyses to assess changes in mortality between rounds after adjusting for changes in mean age and for changes in the percentage of patients with more severe levels of performance status. This study was approved by the joint University College London/University College London Hospital Committees on the Ethics of Human Research. Results Organizational and resource data were received from 238 units in 2003 and 239 units in 2008 reflecting a 94 and 98% involvement of eligible Trusts, respectively. Clinical data were received from 237 units (7602 patients) in 2003 and from 232 units (9716 patients) in 2008. Organizational, resource and clinical data were submitted by 192 units to both audits. In 2003, there was a median [inter-quartile range (IQR)] of 35 (29 38) patients per unit and a total of 6197 patients while in 2008 there was a median (IQR) of 47 (30 58) patients per unit and a total of 8170 patients. There were no significant differences in clinical outcomes between the 192 units participating in both audits and those units participating in only one round (results not shown). There were key differences in the demographics of the patient groups in the two audits (Table 1). COPD patients in the later audit were older, with 12% aged 85 years or more compared with 7% in 2003. Performance status (PS) was significantly more impaired and patients were more hypoxic on admission. Other admission measurements that reflect illness severity such as serum albumin and blood urea were similar for patients in the two audits, as was admission spirometry with a median Forced Expiratory Volume in 1 second (FEV 1 ) of 0.83 l in 2003 and 0.86 l in 2008. The mean [median (IQR)] ages of 192 units was 71.1 years (69.7 72.8) in 2003 compared with 72.8 years (71.2 74.4) in 2008 (P < 0.001, Wilcoxon), and the median percentage of patients with a PS of 4 5 (i.e. limited self-care or bed/chair bound with no capacity for self-care) was 21% (13 31%) in 2003 and 27% (19 38%) in 2008 (P < 0.001, Wilcoxon). At a unit level, there was a positive correlation between a change in mean age and inpatient mortality (Spearman r = 0.23, P = 0.001) and 90-day mortality (r = 0.24, P = 0.001). There was also an association between change in percentage of patients with a PS of 4 5 and change in median inpatient mortality (r = 0.21, P = 0.003) and 90-day mortality (r = 0.24, P = 0.001). There was no evidence of a reduction in inpatient mortality per se between rounds (Table 2). However, a linear regression model of unit level data adjusting for both the change in mean age and change in the percentage number of patients with a PS of 4 5 revealed a decrease in median inpatient mortality between audits of 1.1% (95% confidence interval 0.01 2.2%, P = 0.05). A similar adjusted analysis of 90-day mortality revealed a decrease in mortality of 3.1% (1.3 4.9%, P = 0.001). Adjusted analyses were also done using median regression (STATA qreg ) that gave similar estimates, a decrease of 0.7% (P = 0.24) in median inpatient mortality and 2.9% (P = 0.02) in median 90-day mortality. Median length of stay fell by 1 day (P < 0.001) in 2008 and there was an increase from 25% to 33% of the proportion of patients staying 3 days or less (Table 2). Median readmission rates, however, increased slightly from 32% to 35% (NS), with the median (IQR) time from admission to readmission falling from 42 (22 66) days in 2003 to 38 (20 66) days in 2008. There was no correlation between age or performance status with readmission rates or LOS and so no adjusted regression analyses were performed. Acceptance onto an EDS of some form was up from 15% to 19%. A weak but significant correlation (Spearman r = 0.23, n = 191, P < 0.001) linked units increasing their percentage of patients in EDS to lower hospital median lengths of stay. The percentage of patients discharged by a respiratory consultant improved from 48% to 53% (P = 0.01) and the percentage seen by a respiratory specialist at some time during admission went up from 73% to 80% (P < 0.001). Table 2 also indicates improvements in regard to documentation in patient notes of respiratory rate and in commenting on chest radiographs. There have been changes in specialty staffing available to meet the needs of COPD patients (Table 3). The median number of respiratory consultants per unit increased from 2.0 to 3.0

862 P.M. George et al. Table 1 Patient characteristics for 192 units participating in both the 2003 and 2008 National COPD clinical audits 2003 audit (N = 6197) 2008 audit (N = 8170) P-value a Male b 52 (3230/6192) 51 (4127/8170) 0.05 Aged < 65 24 (1493/6189) 22 (1785/8170) <0.001 Aged 65 74 36 (2217/6189) 30 (2470/8170) Aged 75 84 33 (2016/6189) 36 (2910/8170) Aged 85+ 7 (463/6189) 12 (1005/8170) PS (if known) b <0.001 Normal activity 12 (640/5470) 9 (660/7381) Strenuous activity limited 19 (1062/5470) 15 (1093/7381) Limited activity but self-care 46 (2527/5470) 48 (3550/7381) Limited self-care 19 (1055/5470) 21 (1573/7381) Bed/chair bound no self-care 3 (186/5470) 7 (505/7381) Current smoker b 41 (2422/5898) 33 (2549/7662) <0.001 Ex-smoker b 55 (3258/5898) 64 (4882/7662) Life-long non-smoker b 4 (218/5898) 3 (231/7662) <20 pack years b 9 (232/2636) 8 (315/4040) 0.13 20 39 pack years b 31 (812/2636) 30 (1194/4040) 40+ pack years b 60 (1592/2636) 63 (2531/4040) Sputum volume increase b 57 (2835/5006) 65 (4185/6412) <0.001 Sputum colour change b 51 (2419/4770) 61 (4028/6595) <0.001 Increased breathlessness b 95 (5716/6009) 98 (7833/8022) <0.001 Peripheral oedema b 32 (1509/4730) 32 (1957/6157) 0.91 Blood ph <7.35 b 22 (1124/5102) 21 (1449/6949) 0.12 PaCO 2 >6.0 kpa b 46 (2336/5120) 44 (3065/6923) 0.15 PaO 2 <7.3 kpa b 963/5113) 21 (1427/6928) 0.02 Age c 72 (65 78) 74 (66 80) <0.001 Respiratory rate c 24 (20 28) 24 (20 28) <0.001 FEV1 c 0.83 (0.61 1.16) 0.86 (0.64 1.20) 0.002 FEV1 %predicted c 37 (28 50) 38 (28 52) 0.09 Serum albumin g/dl c 38 (34 41) 39 (35 42) <0.001 Blood urea mmol/l c 6.0 (4.4 8.3) 6.2 (4.6 8.7) <0.001 Blood ph c 7.40 (7.36 7.44) 7.41 (7.36 7.44) 0.003 BIC c 27 (24 30) 26 (24 30) 0.008 c PCO 2 5.8 (5.0 7.2) 5.8 (4.9 7.2) 0.03 c PO 2 9.2 (7.7 11.9) 8.9 (7.6 11.3) <0.001 a Chi-squared test for categorical data, Mann Whitney test for numerical data. b %(n/n). c Median (IQR). BIC, bicarbonate; PCO 2, partial pressure of carbon dioxide (KPa); PO 2, partial pressure of oxygen (KPa). (P < 0.001) and the percentage of units with specialist respiratory wards increased from 67% to 85% (P < 0.001). However, 30% (58/191) of these units reported current staffing vacancies in the 2008 audit and many units still fall below the staffing levels recommended by The Royal College of Physicians (2008). 6 Despite a considerable rise in the use of specialty triage, almost half (46%) of these units still did not operate this in 2008. Patients gained greater access to specialist nurses (from 74% to 96% of units, P < 0.001) and had greater access to an EDS (47 64%, P < 0.001). More patients with acidotic respiratory failure were treated with non-invasive ventilation (NIV) (Table 2) although access to an intensive care unit (ICU) and mechanical ventilation appears to have become more difficult (Table 3). More units had increased access to pulmonary rehabilitation in 2008 than in 2003 although a supplementary question asked only in 2008 revealed that only 59% (114) of units had rehabilitation available to all suitable patients. It is notable that there are wide IQRs for admissions and staffing (Table 3) reflecting variations in practice between individual hospital units, and wide IQRs for processes of care and outcome

National COPD audits of 2003 and 2008 863 Table 2 Patient process of care and outcome for 192 units participating in both the 2003 and 2008 National COPD clinical audits Summary of results at patient level %(n/n) Summary of results aggregated at unit level, expressed as median percentage unit result and (IQR of unit results) 2003 2008 Change (%) 2003 2008 P-value a Inpatient mortality rate 7.7 (462/6036) 7.7 (629/8167) 0.0 7.0 (2.8-11.5) 7.3 (3.8 10.6) 0.84 Overall mortality rate at 90 days 15.6 (938/5998) 14.0 (1096/7844) 1.6 15.3 (10.0 21.7) 13.7 (8.5 18.3) 0.08 Patients dying of COPD 80 (662/823) 77 (725/946) 3 85 (67 100) 82 (67 100) 0.02 Length of stay from admission to discharge 6 (3 10) b 5 (3 10) b 1 day 6 (5 8) days 6 (4 7) days <0.001 Patients staying 3 days or less 25 (1351/5362) 33 (2516/7538) +8 23 (16 34) 31 (24 41) <0.001 Patients discharged by a Respiratory Consultant 48 (2899/6011) 53 (3918/7430) +5 46 (31 64) 53 (37 69) 0.01 Patients accepted onto EDS 15 (877/5903) 19 (1389/7242) +4 0 (0 22) 6 (0 29) 0.01 Patients readmitted or accepted onto EDS within 90 days of index admission 32 (1707/5402) 35 (2522/7288) +3 32 (23 40) 35 (26 43) 0.01 Performance status known 88 (5470/6197) 90 (7381/8170) +2 93 (83 97) 93 (85 98) 0.45 Peripheral oedema documented in notes 76 (4730/6197) 75 (6157/8170) 1 77 (65 91) 77 (66 87) 0.44 Respiratory rate documented in notes 81 (5000/6197) 93 (7604/8170) +12 84 (70 93) 96 (91 99) <0.001 No comment on CXR appearance in notes 14 (869/6197) 9 (761/8170) 5 13 (5 21) 8 (3 13) <0.001 Patients admitted that were current smokers 41 (2422/5898) 33 (2549/7662) 8 39 (30 46) 32 (28 41) <0.001 Smoking pack years recorded in notes 46 (2636/5680) 54 (4040/7431) +8 46 (29 62) 55 (36 69) 0.002 Spirometry data documented in last 5 years 55 (3429/6197) 56 (4562/8170) +1 55 (40 71) 55 (42 70) 0.95 Patients who had an ABG taken 86 (5164/6017) 87 (7022/8076) +1 88 (80 94) 89 (82 94) 0.22 Patients receiving ventilatory support during admission 10 (610/5965) 12 (1015/8170) +2 8 (3 15) 11 (7 17) 0.001 Acidotic patients (at any time during admission) 32 (446/1406) 45 (845/1860) +13 33 (17 55) 44 (26 58) 0.002 who received NIV Patients who received NIV during admission 9 (519/5965) 11 (935/8170) +2 7 (3 14) 10 (6 16) <0.001 Patients seeing a respiratory specialist during admission 73 (4372/5984) 80 (6436/8082) +7 73 (56 88) 83 (68 94) <0.001 a Wilcoxon matched pairs test. b Median (IQR). CXR, chest X-ray; ABG, arterial blood gas.

864 P.M. George et al. Table 3 Organization and resources for 192 units participating in both the 2003 and 2008 national COPD clinical audits Annual number of admissions per unit, median (IQR); sum (mean) Units 2003 2008 Change (%) P-value 186 486 (331 716); 112 737 (606) 486 (363 792); 125 178 (673) 23 ( 152 to 240); 1241 (67) Consultants per 1000 emergency COPD 187 4.0 (2.3 5.8); 5.2 (3.5 7.6); 0.9 ( 0.6 to 3.5) <0.001 a admissions median (IQR); mean 3.68 4.67 Number of respiratory consultants, 192 2.0 (1.4 3.0) 3.0 (2.0 3.0) 0.8 (0.0 to 1.0) <0.001 a median (IQR); sum (mean) 415.2 (2.2) 585.0 (3.0) 169.8 (0.8) One or no respiratory consultant, % (n) 192 24 (46) 9 (17) 15 <0.001 b Early Warning Scoring systems used, % (n) 190 61 (116) 88 (167) +27 <0.001 b Consultants undertake at least 2 PTWRs 186 61 (113) 80 (149) +19 <0.001 b in 24 h, % (n) Units with Speciality triage, % (n) 192 34 (65) 54 (104) +20 <0.001 b Units with Specialist Respiratory wards, % (n) 192 67 (129) 85 (164) +18 <0.001 b No HDU, % (n) 192 18 (34) 10 (20) 8 0.03 b No ITU beds for COPD patients, % (n) 190 2 (4) 4 (8) +2 0.13 Invasive ventilation not offered to COPD 192 17 (32) 20 (39) +3 0.41 patients, % (n) Availability of NIV throughout units, % (n) 192 96 (185) 98 (189) +2 0.29 Availability of NIV on HDU, % (n) 192 63 (120) 74 (143) +11 0.005 Availability of NIV on ITU, % (n) 192 63 (120) 56 (107) 7 0.18 Formal pulmonary rehabilitation 192 64 (122) 90 (173) +26 <0.001 b programme, % (n) Access to EDS, % (n) 192 47 (91) 64 (123) +17 <0.001 b Access to a respiratory nurse, % (n) 192 74 (142) 96 (184) +22 <0.001 b a Wilcoxon matched pairs test. b McNemar paired test. PTWR, post take ward round; HDU, high dependency unit; ITU, intensive treatment unit; NIV, noninvasive ventilation; EDS, early discharge scheme. 0.07 a (Table 2) reflecting a mix of unit and sampling variation. Discussion This comparative study of the 2003 and 2008 UK COPD Audit data has demonstrated considerable improvement in resources, organization and delivery of acute COPD care across the UK with greater accessibility to COPD-specific services. In absolute terms, there has been little change between 2003 and 2008 in many of the patient outcomes such as inpatient and 90-day mortality. However, there is clear evidence that in 2008 the average UK patient with COPD admitted to hospital with an exacerbation was older, more hypoxic and of a poorer functional class than admitted patients in 2003 (Table 1). We can only hypothesize that this apparent improvement in pre-hospital survival is a result of better long-term condition care but it has great significance to the current data. Both age and in particular PS are important predictors of poorer outcomes 7 and when adjusting for these factors, there was in 2008, a statistically significant reduction in 90-day mortality and a clear trend to reduction in inpatient mortality. There are several areas where resources and organization of care have improved over the 5-year period between the audits. The number of respiratory physicians per unit increased between audits as did the number of consultants expressed per 1000 emergency COPD admission (both P < 0.001). During this period, the percentages of patients discharged under the care of a respiratory specialist or seen by a respiratory nurse or a respiratory specialist during the admission have all increased. The 2003 audit demonstrated that patients under the care of a respiratory physician were more likely to receive or access specialist respiratory services 7,8 and these are known to have implications on the quality of patient care. 9 Such specialty-based improvements in care are observed in the 2008 data where entry to an EDS increased from 15% to 19% (P = 0.01) and treatment with NIV for patients in type II acidotic failure increased by 13% (P = 0.002), both interventions recommended by NICE. 5 The increased uptake of EDS has almost certainly played a part in the significantly decreased median length of stay and has contributed to the

National COPD audits of 2003 and 2008 865 number of patients being rapidly discharged in under 48 h. It has also previously been shown that patients who are appropriately selected for EDS have similar outcomes to those that have longer in-patient stays with similar readmission rates 10 and that reduction in hospital admissions positively impact upon such patients quality of life. 11 Such evidence, combined with that from real life audit, can be used by respiratory teams to support a business case for this service where currently it is not provided. Pulmonary rehabilitation programmes are also more widely used now and are available in some form at 90% of units compared with 64% in 2003. However, only 59% of units were able to provide access to all eligible patients as suggested by NICE 5 with many of the others providing a more sporadic service. A decrease in current smoking habit recorded may be partly related to a change in definition of an ex-smoker, from having stopped at least 3 months before admission (2003) to having stopped before admission (2008). Increased consultant numbers are also associated with prompter senior emergency review. In 2003, 39% of patients were waiting up to 24 h post-admission to be seen by a consultant, this has reduced to 20%, with a significant increase from 61% to 80% (p < 0.001) in the number of units where consultants undertake two post-take ward rounds in 24 h. It is worth noting that where best medical practise is to adopt this standard, 20% of units fail to do so. This and other standards are still variably met across units and there remain unacceptable variations in organization and delivery of care between units. The IQR for patients discharged under a respiratory consultant varied from 37% to 69% in 2008 and acceptance onto an EDS between 0% and 29%. While sample size for an individual unit may account for variations in outcome, this is less an argument for adequate process of care where compliance should be 100% yet, for example, the IQR for documentation of smoking history is 36 69% and recording of spirometry 42 70%. The theme of improvement but variability is demonstrated in other areas where there has been a genuine improvement in specialist service. For example, the provision of specialist respiratory wards has increased from 67% in 2003 to 85% in 2008 and specialty triage has risen from 34% to 54%. Simply increasing the respiratory resource is insufficient to produce meaningful improvements in the absence of a whole systems revision of acute care pathways. The promotion of stroke units for stroke patients has undoubtedly improved outcomes 12 and has changed clinical practice considerably in recent years. There must fall some responsibility on the respiratory community to argue the case for COPD patients to be triaged to receive care in specialty wards by specialists in hospitals where this does not already occur. There is evidence from within the audit that some potentially lifesaving interventions such as NIV are not applied to the standards recommended in national guidelines or are withheld from some patients who might potentially benefit from the treatment. 13 We recognize certain limitations to this study that are acknowledged here. There were some differences in the units included in the two audits. Some merged during this time of NHS reorganization and a small number contributing in 2003 did not do so in 2008. More units overall participated in 2008 than in 2003. We have therefore only included units which participated in both rounds and it is possible that this has excluded some of the less well performing units that participated only once, were subsequently merged, or lost their acute status though a comparative analysis of outcomes did not suggest the presence of selection bias. Units were asked to collect information from consecutive admissions but given the numbers of units involved, it is impossible to ascertain that all units were able to supply a fully consecutive sample. The assessment of PS is a subjective one relying on collateral history, clinical notes, nursing records as well as the patient s own description. However, PS is validated by the correlation with age of the patients and its strong predictive value for all outcomes in these audits. It is challenging to prove causality between advances in processes and resources and clinical outcomes as these are likely to have been multifactorial. We looked for any relationships between organization and outcome and only found weak associations. This is likely to reflect the extremely complex interactions that occur in the relationships between service changes and clinical outcomes. Changes to one part of a complex system are unlikely to produce a defined specific difference in outcome that depends upon a series of additional structures and events. There are other technical limitations of any case note audit that we also acknowledge. Notwithstanding, this is the most comprehensive comparative audit cycle of UK COPD secondary care ever reported comprising over 14 000 COPD admissions and provides as true to life a picture of actual clinical practice that is possible within the limitations of the constantly changing environment of health service reorganization. We conclude that there have been many positive developments in resourcing and organization of COPD care over the period between the two audits that are to be welcomed. At the same time,

866 P.M. George et al. there are significant improvements in the process of care and some reduction in mortality and length of stay although the admitted patients are now older and have a worse PS than their counterparts in 2003. The continued concern however is the extent of the variation that exists between units of organization and resources with resultant variations in care for patients. These data provide the justification for the introduction of a truly national service strategy for COPD with quality indicators and metrics that should underpin the delivery of high-quality services throughout the NHS. Acknowledgements We are grateful to all the respiratory clinicians and audit teams who participated in the programme. Funding This work was funded by the Health Foundation. The Health Foundation had no involvement in the study design, data collection, reporting or the writing of the manuscript. The study was supported by the British Lung Foundation, British Thoracic Society, Royal College of Physicians and guided by the National COPD Resources and Outcomes Project Steering Group. Conflict of interest: None declared. References 1. Britton M. The burden of COPD in the U.K.: results from the Confronting COPD survey. Respir Med 2003; 97(Suppl C):S71 9. 2. British Lung Foundation. Invisible lives: Chronic Obstructive Pulmonary Disease (COPD) Finding the Missing Millions, 2007 [www.lunguk.org/resources/ British%20Lung%20Foundation/Migrated%20Resources/ Documents/I/Invisible%20Lives%20report.pdf]. 3. Price LC, Lowe D, Hosker HS, Anstey K, Pearson MG, Roberts CM. British Thoracic Society and the Royal College of Physicians Clinical Effectiveness Evaluation Unit (CEEu). UK National COPD Audit 2003: impact of hospital resources and organisation of care on patient outcome following admission for acute COPD exacerbation. Thorax 2006; 61:837 42. 4. Hosker H, Anstey K, Lowe D, Pearson M, Roberts CM. Variability in the organisation and management of hospital care for COPD exacerbations in the UK. Respir Med 2007; 101:754 61. 5. National Institute of Clinical Excellence. Chronic obstructive pulmonary disease. Management of chronic obstructive pulmonary disease in adults in primary and secondary care 2004. [http://guidance.nice.org.uk/cg12] Accessed 18 May 2011. 6. Royal College of Physicians. Consultant Physicians Working with Patients: the Duties, Responsibilities and Practice of Physicians in Medicine. 2008. 4th edn. London. RCP Bookshop. ISBN: 9781860163258. 7. Roberts CM, Lowe D, Bucknall CE, Ryland I, Kelly Y, Pearson MG. Clinical audit indicators of outcome following admission to hospital with acute exacerbation of chronic obstructive pulmonary disease. Thorax 2002; 57:137 41. 8. Connolly MJ, Lowe D, Anstey K, Hosker HS, Pearson MG, Roberts CM. British Thoracic Society and the Royal College of Physicians Clinical Effectiveness Evaluation Unit (CEEu). Admissions to hospital with exacerbations of chronic obstructive pulmonary disease: Effect of age related factors and service organisation. Thorax 2006; 61:843 8. 9. Roberts CM, Barnes S, Lowe D, Pearson MG. Evidence for a link between mortality in acute COPD and hospital type and resources. Thorax 2003; 58:947 9. 10. Cotton MM, Bucknall CE, Dagg KD, Johnson MK, MacGregor G, Stewart C, et al. Early discharge for patients with exacerbations of chronic obstructive pulmonary disease: a randomized controlled trial. Thorax 2000; 55:902 6. 11. Niewoehner DE. The impact of severe exacerbations on quality of life and the clinical course of chronic obstructive pulmonary disease. Am J Med 2006; 119(10 Suppl. 1): 38 45. 12. Stroke Units Trialists Collaboration. Collaborative systematic review of the randomised trials of organised inpatient (stroke unit) care after stroke. BMJ 1997; 314:1151 9. 13. Roberts CM, Stone RA, Buckingham RJ, Pursey NA, Lowe D. On behalf of the National Chronic Obstructive Pulmonary Disease Resources and Outcomes Project (NCROP) implementation group. Acidosis, non-invasive ventilation and mortality in hospitalised COPD exacerbations. Thorax 2011; 66:43 8.