Downloaded from http://heart.bmj.com/ on October 19, 2015 - Published by group.bmj.com
Heart failure and cardiomyopathies
ORIGINAL ARTICLE
Relation between process measures and diagnosis-specific readmission rates in patients with heart failure Alex Bottle,1 Rosalind Goudie,1 Martin R Cowie,2 Derek Bell,3 Paul Aylin1 1
Dr Foster Unit, Imperial College, London, UK 2 National Heart and Lung Institute, Imperial College London (Royal Brompton Hospital), London, UK 3 Chelsea and Westminster Hospital, Imperial College London, London, UK Correspondence to Dr Alex Bottle, Dr Foster Unit, Imperial College London, Department of Primary Care and Public Health, School of Public Health, 3 Dorset Rise, London EC4Y 8EN, UK; [email protected] Received 15 December 2014 Revised 18 May 2015 Accepted 24 May 2015 Published Online First 11 June 2015
ABSTRACT Objectives To explore the relations between causespecific readmission rates and National Heart Failure Audit process of care measures in patients admitted for heart failure (HF). Methods Using admissions data for all acute hospitals in England for April 2009–March 2012, we defined an index admission as the first emergency admission with a primary diagnosis of HF for at least three years. We compared risk-adjusted readmission rates for HF and for all non-HF diagnoses combined, risk-adjusted in-hospital mortality rates and performance on six Audit process measures. Results 14.7% of 123 644 patients died during the index admission. Of 105 441 index live discharges, 6853 (6.5%) were readmitted as emergencies within 7 days and 20 144 (19.1%) within 30 days. Index admission mortality rates correlated positively but weakly with nonHF readmission rates but not at all with HF rates. There was modest positive correlation at 7 days between HF and non-HF readmission rates (r=+0.24) but no significant relation at 30 or 365 days. All six process measures ( prescribing of ACE inhibitors and beta-blockers, echocardiogram, cardiology inpatient and follow-up by cardiologist and HF liaison) correlated modestly but significantly with lower HF readmission rates at 7 days (r at most −0.26), only three did at 30 days and only cardiology follow-up did for non-HF at either 7 or 30 days; all associations were diminished at 365 days. Conclusions Hospitals scoring higher on evidencebased HF process measures had lower readmission rates, though the association seems limited to HF readmissions and is modest in strength and duration.
INTRODUCTION
▸ http://dx.doi.org/10.1136/ heartjnl-2015-308185
To cite: Bottle A, Goudie R, Cowie MR, et al. Heart 2015;101:1704–1710. 1704
Heart failure (HF) is one of the commonest causes of hospital admission worldwide and is associated with high mortality and morbidity. Emergency readmission rates are high and can incur financial penalties in some countries. However, our understanding of what drives readmission rates is incomplete. To date, much of the work comes from the USA, particularly the Centres for Medicare and Medicaid Services’ (CMS) data, and it is unclear whether findings translate well to other countries’ systems. Second, the CMS’s pay-for-performance measure and most studies look at all-cause and not HF-specific 30-day readmissions.1 2 Comorbidity is very common in this patient group, and we have found that some comorbidities are associated with readmission either for HF or for other conditions,
but not both.3 Hospitals that manage the HF well might not do so well with patients’ other medical problems, so readmission rates for HF might not correlate with those for other conditions. Third, studies of readmissions necessarily consider only those patients discharged alive from the index admission. US analysis using CMS data has suggested that lower death rates show a modest but significant relation with higher readmission rates for patients admitted for HF though not for acute myocardial infarction or pneumonia.4 Such a relation is plausible if interventions that improve mortality also increase readmission rates by resulting in a higher-risk group being discharged alive and subsequently readmitted. This has led to concern over these publically reported measures. To tackle these shortcomings, we combined national administrative data for all National Health Service (NHS) acute hospitals in England on index admissions with a primary diagnosis of HF with hospital-level information on six evidence-based process measures from the National Heart Failure Audit for England and Wales (NHFA)5 made publically available for the first time. Several studies, mostly from the USA, have investigated the process– outcome association for HF, giving inconsistent results.6–11 We assessed the following correlations: (i) between mortality and readmission rates, (ii) between HF and non-HF readmission rates, and (iii) between the six process measures, mortality, and HF and non-HF readmission rates.
METHODS Data sources England’s national hospital administrative database Hospital Episodes Statistics (HES) comprises records of all admissions to, and outpatient department (OPD) appointments at, all NHS ( public) hospitals in England. Records belonging to the same person were linked using a combination of the patient’s unique NHS number, date of birth, sex and postcode; those with an invalid postcode were excluded (0.1). Adjusted rates for in-hospital mortality showed slight positive correlations with non-HF readmission rates, though these only reached conventional statistical significance for a 7-day time window (r=+0.20, Table 3 Correlations between risk-adjusted death rates and readmission rates by time period and primary diagnosis on readmission, all hospitals
Readmission measure
Correlation coefficient for readmission and in-hospital mortality (p value)
7-day HF 7-day non-HF 30-day HF 30-day non-HF 365-day HF 365-day non-HF
+0.11 (0.178) +0.20 (0.017) +0.03 (0.711) +0.15 (0.069) -0.07 (0.428) +0.13 (0.118)
Pearson’s correlation coefficients are given with p values in parentheses. Bold values indicate p
Heart failure and cardiomyopathies
ORIGINAL ARTICLE
Relation between process measures and diagnosis-specific readmission rates in patients with heart failure Alex Bottle,1 Rosalind Goudie,1 Martin R Cowie,2 Derek Bell,3 Paul Aylin1 1
Dr Foster Unit, Imperial College, London, UK 2 National Heart and Lung Institute, Imperial College London (Royal Brompton Hospital), London, UK 3 Chelsea and Westminster Hospital, Imperial College London, London, UK Correspondence to Dr Alex Bottle, Dr Foster Unit, Imperial College London, Department of Primary Care and Public Health, School of Public Health, 3 Dorset Rise, London EC4Y 8EN, UK; [email protected] Received 15 December 2014 Revised 18 May 2015 Accepted 24 May 2015 Published Online First 11 June 2015
ABSTRACT Objectives To explore the relations between causespecific readmission rates and National Heart Failure Audit process of care measures in patients admitted for heart failure (HF). Methods Using admissions data for all acute hospitals in England for April 2009–March 2012, we defined an index admission as the first emergency admission with a primary diagnosis of HF for at least three years. We compared risk-adjusted readmission rates for HF and for all non-HF diagnoses combined, risk-adjusted in-hospital mortality rates and performance on six Audit process measures. Results 14.7% of 123 644 patients died during the index admission. Of 105 441 index live discharges, 6853 (6.5%) were readmitted as emergencies within 7 days and 20 144 (19.1%) within 30 days. Index admission mortality rates correlated positively but weakly with nonHF readmission rates but not at all with HF rates. There was modest positive correlation at 7 days between HF and non-HF readmission rates (r=+0.24) but no significant relation at 30 or 365 days. All six process measures ( prescribing of ACE inhibitors and beta-blockers, echocardiogram, cardiology inpatient and follow-up by cardiologist and HF liaison) correlated modestly but significantly with lower HF readmission rates at 7 days (r at most −0.26), only three did at 30 days and only cardiology follow-up did for non-HF at either 7 or 30 days; all associations were diminished at 365 days. Conclusions Hospitals scoring higher on evidencebased HF process measures had lower readmission rates, though the association seems limited to HF readmissions and is modest in strength and duration.
INTRODUCTION
▸ http://dx.doi.org/10.1136/ heartjnl-2015-308185
To cite: Bottle A, Goudie R, Cowie MR, et al. Heart 2015;101:1704–1710. 1704
Heart failure (HF) is one of the commonest causes of hospital admission worldwide and is associated with high mortality and morbidity. Emergency readmission rates are high and can incur financial penalties in some countries. However, our understanding of what drives readmission rates is incomplete. To date, much of the work comes from the USA, particularly the Centres for Medicare and Medicaid Services’ (CMS) data, and it is unclear whether findings translate well to other countries’ systems. Second, the CMS’s pay-for-performance measure and most studies look at all-cause and not HF-specific 30-day readmissions.1 2 Comorbidity is very common in this patient group, and we have found that some comorbidities are associated with readmission either for HF or for other conditions,
but not both.3 Hospitals that manage the HF well might not do so well with patients’ other medical problems, so readmission rates for HF might not correlate with those for other conditions. Third, studies of readmissions necessarily consider only those patients discharged alive from the index admission. US analysis using CMS data has suggested that lower death rates show a modest but significant relation with higher readmission rates for patients admitted for HF though not for acute myocardial infarction or pneumonia.4 Such a relation is plausible if interventions that improve mortality also increase readmission rates by resulting in a higher-risk group being discharged alive and subsequently readmitted. This has led to concern over these publically reported measures. To tackle these shortcomings, we combined national administrative data for all National Health Service (NHS) acute hospitals in England on index admissions with a primary diagnosis of HF with hospital-level information on six evidence-based process measures from the National Heart Failure Audit for England and Wales (NHFA)5 made publically available for the first time. Several studies, mostly from the USA, have investigated the process– outcome association for HF, giving inconsistent results.6–11 We assessed the following correlations: (i) between mortality and readmission rates, (ii) between HF and non-HF readmission rates, and (iii) between the six process measures, mortality, and HF and non-HF readmission rates.
METHODS Data sources England’s national hospital administrative database Hospital Episodes Statistics (HES) comprises records of all admissions to, and outpatient department (OPD) appointments at, all NHS ( public) hospitals in England. Records belonging to the same person were linked using a combination of the patient’s unique NHS number, date of birth, sex and postcode; those with an invalid postcode were excluded (0.1). Adjusted rates for in-hospital mortality showed slight positive correlations with non-HF readmission rates, though these only reached conventional statistical significance for a 7-day time window (r=+0.20, Table 3 Correlations between risk-adjusted death rates and readmission rates by time period and primary diagnosis on readmission, all hospitals
Readmission measure
Correlation coefficient for readmission and in-hospital mortality (p value)
7-day HF 7-day non-HF 30-day HF 30-day non-HF 365-day HF 365-day non-HF
+0.11 (0.178) +0.20 (0.017) +0.03 (0.711) +0.15 (0.069) -0.07 (0.428) +0.13 (0.118)
Pearson’s correlation coefficients are given with p values in parentheses. Bold values indicate p
