Changes in Heart Failure Outcomes After a Province-Wide Change in Health Service Provision A Natural Experiment in Alberta, Canada
2012; Lippincott Williams & Wilkins; Volume: 6; Issue: 1 Linguagem: Inglês
10.1161/circheartfailure.112.971119
ISSN1941-3297
AutoresFinlay A. McAlister, Jeffrey A. Bakal, Padma Kaul, Hude Quan, Robyn Blackadar, David E. Johnstone, Justin A. Ezekowitz,
Tópico(s)Primary Care and Health Outcomes
ResumoHomeCirculation: Heart FailureVol. 6, No. 1Changes in Heart Failure Outcomes After a Province-Wide Change in Health Service Provision A Natural Experiment in Alberta, Canada Free AccessResearch ArticlePDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissionsDownload Articles + Supplements ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toSupplemental MaterialFree AccessResearch ArticlePDF/EPUBChanges in Heart Failure Outcomes After a Province-Wide Change in Health Service Provision A Natural Experiment in Alberta, Canada Finlay A. McAlister, MD, MSc, Jeffrey A. Bakal, PhD, Padma Kaul, PhD, Hude Quan, PhD, Robyn Blackadar, MBA, David Johnstone, MD and Justin Ezekowitz, MB, BCh, MSc Finlay A. McAlisterFinlay A. McAlister From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). , Jeffrey A. BakalJeffrey A. Bakal From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). , Padma KaulPadma Kaul From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). , Hude QuanHude Quan From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). , Robyn BlackadarRobyn Blackadar From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). , David JohnstoneDavid Johnstone From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). and Justin EzekowitzJustin Ezekowitz From the Divisions of General Internal Medicine (F.A.M.) and Cardiology (J.A.B., P.K., D.J., J.E.), University of Alberta, Edmonton, Alberta, Canada; Mazankowski Alberta Heart Institute (F.A.M., J.A.B., P.K., D.J., J.E.), Edmonton, Alberta, Canada; University of Calgary, Calgary, Alberta, Canada (H.Q.); and President and CEO, Alberta Center for Child, Family and Community Research, Edmonton, Alberta, Canada (R.B.). Originally published10 Dec 2012https://doi.org/10.1161/CIRCHEARTFAILURE.112.971119Circulation: Heart Failure. 2013;6:76–82Other version(s) of this articleYou are viewing the most recent version of this article. Previous versions: January 1, 2012: Previous Version 1 AbstractBackground—The Alberta Cardiac Access (ACA) initiative was implemented in the spring of 2008 to increase access to specialized heart failure (HF) clinics after hospital discharge.Methods and Results—We identified all adults hospitalized with a most responsible diagnosis of HF between April 1999 and December 2009. We randomly selected 1 episode of care per patient and evaluated outcomes using interrupted time series: the a priori specified primary outcome was all-cause readmission or death in the first 30 days postdischarge. Between 1999 and 2009, median length of stay increased from 8 days to 10 days (P<0.001), and 30-day mortality increased from 9.1% to 11.5% (P<0.001) in the 37891 HF hospitalizations we examined. However, these temporal changes were attributable to the increasing comorbidity burden over time: the adjusted Risk Ratio for 30-day mortality in 2009 versus 1999 was 0.99, 95% confidence interval, 0.86 to 1.15. After adjusting for secular trends, the ACA initiative was associated with changes in 30-day postdischarge mortality or readmission rates (which were increasing 0.3% per month [0.2%–0.3%] pre-ACA and decreased 1.4% per month [0.3%–2.5%] in the 18 months post-ACA; P=0.008). After roll out of the ACA initiative, patients discharged from vanguard regions (those that had specialized HF clinics) exhibited lower 30-day postdischarge death/readmission rates than those discharged from other areas of the province (18.6% versus 22.2%, adjusted odds ratio 0.83, 95% confidence interval, 0.75–0.93).Conclusions—An initiative which increased specialized HF clinic access was associated with a statistically significant improvement in 30-day postdischarge mortality/readmission rates.IntroductionDespite many advances in diagnosis and therapy during the past 2 decades, heart failure (HF) remains the most common cause of hospitalizations and readmissions in North America and Europe.1–3 Although traditional strategies of knowledge dissemination have minimal effects on physician-prescribing habits in HF,4 involvement of specially trained multidisciplinary teams or specialists in the care of patients with HF has been shown to improve the use of proven efficacious therapies and clinical outcomes.5,6 However, there is still debate about whether wider implementation of specialized HF management programs will yield similar benefits as in randomized trials or whether there will be unanticipated consequences (such as increased hospitalizations or health resource use in other areas because of closer patient follow-up). Unfortunately, only a minority of patients, even in publicly funded health care systems with universal access like Canada, have access to these resources.Clinical Perspective on p 82Between April and May 2008, the Alberta provincial government initiated the Alberta Cardiac Access (ACA; see www.cardiacaccess.ab.ca for full details) initiative to improve access to cardiac care. One area of focus was to enhance access to specialized HF clinics for patients recently discharged after a HF hospitalization. The ACA initiative funded (1) training preceptorships for family physicians, pharmacists, and nurses in HF, (2) the expansion of capacity within the 6 specialized HF clinics, already existing pre-2008, and (3) the establishment of 5 new HF clinics in different regions of the province. Each of the specialized HF clinics implemented or expanded as a result of the ACA initiative were designed as high-intensity clinics that scored maximum points on the HF Disease Management Scoring Instrument7 including: targeted both patients and caregivers, provision of education, and counseling to support self-care, medication regimen management including adherence assessment, delivery by multidisciplinary personnel, face-to-face and telephone communication, and high complexity (multiple contacts for multiple visits).We designed this study to examine secular trends in HF outcomes in Alberta during the past decade, with particular attention to the impact of the ACA initiative on patient-relevant outcomes highlighted by the AHA Get With the Guidelines—Heart Failure project, The Joint Commission, and the Canadian Cardiovascular Society.3,8,9MethodsSettingThe province of Alberta has a single payer, government-funded health care system that provides universal access to more than 3.7 million people for hospital, emergency department, and physician services, and the Alberta Health administrative databases thus capture all interactions with the health care system in the province for conditions such as HF. This study received ethics approval from the Health Ethics Research Board at the University of Alberta.Data SourcesThis study used deidentified linked data from 4 Alberta Health administrative databases: (1) the Discharge Abstract Database, which records the admission and discharge dates, most responsible diagnosis (ie, the diagnosis identified by the attending physician at time of discharge as the primary diagnosis which caused the hospitalization), and up to 25 other diagnoses for all acute care hospitalizations; (2) the Ambulatory Care Database, which records all patient visits to hospital-based specialized clinics or Emergency Departments with coding for up to 10 conditions; (3) the Health Practitioner Claims Database, which tracks all physician claims for services and includes up to 3 diagnoses per outpatient encounter; and (4) the Alberta Health Care Insurance Plan Registry, which tracks vital status of all Albertans.Study CohortWe identified all adult Albertans hospitalized between April 1, 1999, and December 31, 2009, with a most responsible diagnosis of HF (International Classification of Diseases [ICD]-9-CM code 428.x from 1999–2002 or ICD-10 code I50.x after 2002—when Alberta switched from ICD-9 to ICD-10).10,11 These ICD codes in the Alberta Health databases have previously been shown to have high specificity (97%–99%) and positive predictive value (91%–94%) for HF, when validated against chart audit in Alberta in the same time frame.10 In our primary analysis, we treated each calendar year as a new subcohort and thus patients could be included in more than 1 calendar year, if they were hospitalized in different years. We treated each hospitalization as an episode of care and transfers between acute care hospitals were counted as being the same episode with the length of stay incorporating time at both hospitals (but attributed to the hospital the patient was eventually discharged from). We excluded those hospitalizations in which the patient was discharged to a rehabilitation hospital. For those patients with multiple hospitalizations in any 1 calendar year, we randomly selected 1 episode of care in that year for this analysis, as per prior studies in this area.2In a sensitivity analysis, we restricted the analysis to only those patients with an incident (first time) hospitalization for HF and excluded their subsequent hospitalizations, even if they were in different study years.CovariatesComorbidities for each patient were identified using the Discharge Abstract database for the index hospitalization and any hospitalizations in the 12 months prior to their index admission.AnalysesAges and patient comorbidities were calculated per year and compared via linear and Cochran-Armitage trend tests where appropriate. We examined secular trends between 1999 and December 2009 in those outcomes outlined in Table 1. Our a priori specified primary outcome was all-cause readmission or death in the first 30 days postdischarge, which we chose for 4 reasons. First, there are validated risk adjustment models for this 30-day outcome.12–14 Second, this outcome is more closely linked to quality of inpatient care than events during a longer timeframe.14 Third, as our data ended in December 2009, we would have had to exclude any patients hospitalized after December 2008, if we used 12-month outcomes (substantially reducing the power of our interrupted time series analyses, since the ACA initiative occurred in the spring of 2008). Fourth, given the recent proposal within the Affordable Care Act to penalize hospitals with high 30-day readmission rates, there is now substantial interest in evaluating events in this timeframe.Table 1. Baseline Characteristics of Patients Hospitalized With a Most Responsible Diagnosis of Heart Failure by Year1999 (n=3371)2000 (n=3520)2001 (n=3586)2002 (n=3455)2003 (n=3519)2004 (n=3489)2005 (n=3585)2006 (n=3315)2007 (n=3198)2008 (n=3331)2009 (n=3522)P Value for TrendAge, mean (SD)76.2 (11.4)76.6 (11.6)76.8 (11.5)77.1 (11.8)77.2 (11.7)77.4 (11.6)77.5 (11.8)77.8 (11.8)77.7 (11.9)77.6 (12.4)77.5 (12.4)<0.01Men1674 (49.7)1793 (50.9)1773 (49.4)1666 (48.2)1750 (49.7)1721 (49.3)1850 (51.6)1657 (50.0)1578 (49.3)1696 (50.9)1798 (51.1)0.17Charlson comorbidity index score (mean, SD)4.3 (2.1)4.5 (2.2)4.4 (2.1)4.5 (2.2)4.5 (2.3)4.5 (2.2)4.4 (2.2)4.5 (2.3)4.4 (2.1)4.5 (2.2)4.5 (2.2)<0.01Diabetes mellitus1036 (30.7)1185 (33.7)1212 (33.8)1241 (35.9)1316 (37.4)1285 (36.8)1363 (38.0)1327 (40.0)1338 (41.8)1361 (40.9)1420 (40.3)<0.01Hypertension1601 (47.5)1822 (51.8)1905 (53.4)1928 (55.8)2068 (58.8)2053 (58.8)2112 (58.9)2081 (62.8)2051 (64.1)2235 (67.1)2403 (68.2)<0.01Dementia182 (5.4)215 (6.1)220 (6.1)331 (9.6)322 (9.2)391 (11.2)376 (10.5)383 (11.6)320 (10.0)349 (10.5)365 (10.4)<0.01Chronic obstructive pulmonary disease1417 (42.4)1508 (42.8)1555 (43.4)1470 (42.6)1471 (41.8)1453 (41.7)1479 (41.3)1397 (42.1)1306 (40.8)1427 (42.8)1486 (42.2)0.35Anemia988 (29.3)1098 (31.2)1149 (32.0)1050 (30.4)998 (28.4)915 (26.2)935 (26.1)871 (26.3)831 (26.0)910 (27.3)921 (26.2)<0.01Cerebrovascular disease424 (12.6)450 (12.8)462 (12.2)420 (12.2)369 (10.5)339 (9.72)379 (10.6)334 (10.1)321 (10.0)293 (8.8)359 (10.2)<0.01Chronic renal disease564 (16.7)677 (19.2)700 (19.5)848 (24.5)972 (27.6)1019 (29.2)1000 (27.9)1004 (30.3)941 (29.4)1007 (30.2)949 (26.9)<0.01Cancer261 (7.4)285 (8.1)277 (7.7)255 (7.4)285 (8.1)240 (6.9)258 (7.2)247 (7.5)209 (6.5)226 (6.8)250 (7.1)<0.01Peripheral vascular disease458 (13.6)515 (14.6)494 (13.8)472 (13.7)449 (12.8)433 (12.4)391 (10.9)375 (11.3)369 (11.5)363 (10.9)377 (10.7)<0.01Atrial fibrillation1271 (37.7)1410 (40.1)1463 (40.8)1324 (38.3)1472 (41.8)1406 (40.3)1498 (41.8)1516 (45.7)144 (45.3)1575 (47.3)1732 (49.2)<0.01Prior myocardial infarction1032 (30.6)1084 (30.8)1119 (31.2)975 (28.2)913 (25.9)805 (23.1)745 (20.8)670 (20.2)663 (20.7)732 (22.0)726 (20.6)<0.01Values are frequencies (and %) unless otherwise stated.For comparisons of 30-day mortality rates and postdischarge readmission rates over time, we adjusted for covariates included in the Centers for Medicare & Medicaid Services-endorsed models for these outcomes.12,13 For analyses of 30-day postdischarge mortality or readmission rates, we adjusted for covariates included in a recently described risk-prediction model for this outcome (the LACE model), which, importantly, incorporates index hospitalization length of stay.14To examine the impact of the ACA initiative (which was implemented in April and May 2008), we conducted interrupted time series analyses by comparing rates for each outcome in those patients discharged with HF after June 2008 (defined as post-ACA), with those discharged during the baseline (April 1999–March 2008) cohort (defined as pre-ACA) using autoregressive, integrated, moving average models. Autocorrelation, partial autocorrelation, and inverse autocorrelation functions were assessed for model–parameter appropriateness and seasonality. Stationarity was assessed using the autocorrelation function and the augmented Dickey-Fuller test. The presence of white noise was assessed by examining the autocorrelations at various lags, using the Ljung-Box X2 statistic.In a preplanned subgroup analysis, we compared the impact of the ACA initiative in regions identified as ACA vanguard sites (ie, those where access to specialized HF clinics was enhanced via establishment of new clinics or expansion of capacity in existing clinics) compared with others (as defined by the operational team of the ACA, independently of this evaluation, based on distribution of funding and data on patient referrals and follow-up visits after May 2008). Full details of the various operational changes across the province of Alberta are included in the Alberta Cardiac Access Collaborative Final Evaluation Report of May 2009 available at http://www.cardiacaccess.ab.ca/evaluation.aspx. All statistical analyses were done using SAS version 9.3 (Cary, NC) and R version 2.15.1 (Vienna, Austria).ResultsDuring our study timeframe, there were 46 396 adult hospitalizations with HF, as the most responsible diagnosis. After randomly selecting one most responsible diagnosis HF hospitalization per patient, our cohort included 37 891 hospitalizations between 1999 and 2009 (Figure 1)—26 581 of these hospitalizations were incident hospitalizations in patients with a new diagnosis of HF.Download figureDownload PowerPointFigure 1. Flowchart illustrating cohort derivation.Patients hospitalized with a most responsible diagnosis of HF were generally similar across study years, although age and comorbidity burden increased significantly between 1999 and 2009 (Table 1), including in those patients with incident HF hospitalization (Table I in the online-only Data Supplement). In particular, the proportion of HF patients with diabetes (from 30.7% to 40.3%), atrial fibrillation (from 37.7% to 49.2%), chronic renal disease (from 16.7% to 26.9%), or dementia (from 5.4% to 10.4%) increased substantially (all P<0.01), whereas the proportion with prior myocardial infarction decreased substantially (from 30.6% to 20.6%; P<0.01). Of note, the total number of diagnosis codes entered into the Discharge Abstract database for patients hospitalized in Alberta during the past decade increased minimally from a mean of 8.72 (SD 4.17) per hospitalization in 1999 to 9.09 (SD 4.81) in 2009.During the 11 years we studied, we observed several trends in outcomes (Table 2). For example, between 1999 and 2009, median length of stay increased from 8 days to 10 days (Table 2; P<0.01), in-hospital mortality increased from 8.9% to 13.5% (P<0.01), 30-day mortality increased from 9.1% to 11.5% (P<0.001), and 1-year mortality increased from 29.4% to 36.7% (P<0.001). Trends were virtually identical in analyses restricted to incident HF hospitalizations (Table II in the online-only Data Supplement). However, these changes reflected the increasing age and comorbidity burden of patients admitted with HF over time: for example, the adjusted Risk Ratio for 30-day mortality was 0.99, 95% confidence interval (CI) 0.86 to 1.15, in 2009 versus 1999,13 and the adjusted Risk Ratio for all-cause readmission at 30 days was 0.92, 95% CI 0.80 to 1.04, in 2009 versus 1999.12 The rate of our primary composite outcome of death or all-cause readmission within 30 days of discharge was similar throughout the decade (20.1% in 2009 compared with 20.7% in 1999; adjusted Risk Ratio, 0.91; 95% CI, 0.82–1.01, using an adjustment model which included index hospital length of stay),14 with expected seasonal variations in that event rates were higher each winter and lower each summer (Figure 2), except 2009, when the H1N1 epidemic occurred in the summer. Results were very similar in the sensitivity analysis restricted to incident HF hospitalizations: 30-day death or readmission rates decreased minimally from 20.9% in 1999 to 19.9% in 2009, adjusted Risk Ratio 1.06 (95% CI, 0.95–1.18) using the risk-adjustment model that included length of stay during the index hospitalization (Table II in the online-only Data Supplement). Of note, HF was the most responsible diagnosis in less than one-third of readmissions (ranging from 5.3% in 1999 to 5.4% in 2008/2009; Table 2). The majority of patients discharged after a HF hospitalization were reviewed within 30 days by a physician, increasing from 72.1% in 1999 to 78.9% in 2009 (P<0.01).Table 2. Outcomes of Patients Hospitalized With a Most Responsible Diagnosis of Heart Failure, by Year of DischargeOutcomes19992000200120022003200420052006200720082009P Value for TrendN33713520358634553519348935853315319833313522Index hospitalization mortality300 (8.9)423 (12.0)389 (10.9)377 (10.9)418 (11.9)438 (12.6)426 (11.9)373 (11.3)404 (12.6)463 (13.9)475 (13.5)<0.01Median length of stay (IQR)8 (5,14)8 (5,15)8 (5,15)8 (5,15)9 (5,16)9 (5,16)9 (5,17)9 (5,17)10 (6,18)10 (6,18)10 (6,18)<0.01Death within 30 days of admission307 (9.1)421 (12.0)360 (10.0)366 (10.6)401 (11.4)407 (11.7)404 (11.3)334 (10.1)361 (11.3)404 (12.1)404 (11.5)<0.01Death within 365 days of admission990 (29.4)1181 (33.6)1129 (31.5)1113 (32.2)1183 (33.6)1209 (34.7)1197 (33.4)1073 (32.4)1138 (35.6)1222 (36.7)NR<0.001Death or readmission within 365 days of admission2357 (69.9)2486 (70.6)2521 (70.3)2399 (69.4)2518 (71.6)2458 (70.4)2497 (69.7)2300 (69.4)2226 (69.6)2296 (68.9)NR0.29In those patients discharged alive from index hospitalization (n):30713097319730783101305131592942279428683047Median length of stay (IQR)8 (5,13)8 (5,14)8 (5,14)8 (5,15)8 (5,15)9 (5,15)9 (5,16)9 (5,16)9 (6,17)9 (6,17)9 (6,16)<0.0130-day postdischarge readmission (all cause) or death635 (20.7)652 (21.1)647 (20.2)621 (20.2)633 (20.4)617 (20.2)644 (20.4)589 (20.0)534 (19.1)551 (19.2)611 (20.1)<0.01365-day postdischarge readmission (all cause) or death2051 (66.8)2068 (66.8)2135 (66.8)2017 (65.5)2109 (68.0)2023 (66.3)2069 (65.5)1915 (65.1)1815 (65.0)1833 (63.9)NR0.00630-day postdischarge readmission (all cause)575 (18.7)584 (18.9)584 (18.3)557 (18.1)562 (18.1)547 (17.9)572 (18.1)529 (18.0)472 (16.9)487 (17.0)534 (17.5)<0.0130-day postdischarge readmission (with HF listed as most responsible diagnosis)162 (5.3)167 (5.4)164 (5.1)161 (5.2)140 (4.5)152 (5.0)162 (5.1)143 (4.9)129 (4.6)138 (4.8)180 (5.9)0.8730-day postdischarge death104 (3.4)111 (3.6)114 (3.6)99 (3.2)126 (4.1)125 (4.1)124 (3.9)104 (3.5)117 (4.2)127 (4.4)126 (4.1)<0.0130-day postdischarge ED visits (all cause)872 (28.4)916 (29.6)945 (29.6)935 (30.4)945 (30.5)885 (29.0)944 (29.9)864 (29.4)798 (28.6)826 (28.8)893 (29.3)0.6630-day postdischarge ED visits (with HF listed as most responsible diagnosis)205 (6.7)235 (7.6)231 (7.2)241 (7.8)243 (7.8)240 (7.9)268 (8.5)214 (7.3)206 (7.4)202 (7.0)252 (8.3)0.2230-day postdischarge hospital-based specialized clinic visit (all cause)642 (20.9)852 (27.5)940 (29.4)971 (31.6)1176 (37.9)1196 (39.2)1316 (41.7)1241 (42.2)1192 (42.7)1294 (45.1)1412 (46.3)<0.0130-day postdischarge hospital-based specialized clinic (with HF listed as main diagnosis)89 (2.9)82 (2.7)102 (3.2)95 (3.1)112 (3.6)115 (3.8)177 (5.6)156 (5.3)123 (4.4)219 (7.6)331 (10.9)<0.0130-day postdischarge outpatient physician office visit (all cause)2213 (72.1)2257 (72.9)2341 (73.2)2196 (71.4)2276 (73.4)2338 (76.6)2430 (76.9)2288 (77.8)2158 (77.2)2215 (77.2)2404 (78.9)<0.001365-day postdischarge outpatient physician office visit (all cause)2742 (89.3)2777 (89.7)2864 (89.6)2751 (89.4)2816 (90.8)2835 (92.9)2956 (93.6)2776 (94.4)2628 (94.1)2660 (92.7)NR<0.00130-day postdischarge outpatient physician office visit (with HF listed as diagnosis)977 (31.8)905 (29.2)934 (29.2)904 (29.4)914 (29.5)964 (31.6)947 (30.0)929 (31.6)842 (30.1)868 (30.3)989 (32.5)0.08ED indicates emergency department; HF, heart failure; IQR, interquartile range; and NR, not reportable (as we only had follow-up data until December 31, 2009).Values are frequencies of discharges having each event (and %) unless otherwise stated.Download figureDownload PowerPointFigure 2. Death or readmission in the 30 days after discharge from a hospitalization with heart failure (HF) as most responsible diagnosis, broken down by Alberta Cardiac Access (ACA) vanguard site versus others. Shaded area represents timeframe during which the ACA initiative was implemented.After accounting for secular trends using interrupted time series analyses (Table 3), the ACA initiative was not associated with any significant changes in index hospitalization mortality (step change P=0.39 and trend change P=0.30), the proportion of patients seen by an outpatient physician in the first 30 days after discharge (step change P=0.83 and trend change P=0.18), nor the proportion of patients with unplanned emergency room visits in the first 30 days after discharge (step change P=0.75 and trend change P=0.54). However, the ACA initiative was associated with a statistically significant step change (a 10.2% absolute increase; 95% CI, 0.4–20.0%) in specialized clinic visits in the spring of 2008, and a statistically significant trend change in 30-day postdischarge mortality or all-cause readmission rates thereafter (which were increasing 0.3% per month [0.2%–0.3%] pre-ACA and decreased 1.4% per month [0.3%–2.5%] in the 18 months post-ACA; P<0.01).Table 3. Outcomes in Patients With Most Responsible Diagnosis of HF Before and After the Changes to HF Service Delivery in AlbertaOutcomePre-ACA Trend (April 1999 to April 2008)Expressed as Percentage Change per Month (95% CI)Post-ACA Trend (June 2008 to January 2010)Expressed as Percentage Change per Month (95% CI)Death/readmission in the 30 days after discharge0.27 (0.21, 0.32)–1.40 (–2.50, –0.30)Death during index hospitalization0.0004 (–0.014, +0.015)0.026 (–0.1201, +0.172)Death within 30 days of admission0.004 (–0.003, +0.012)–0.079 (–0.18, +0.03)30-day all-cause readmission–0.015 (–0.039, +0.009)0.036 (–0.12, +0.19)30-day readmission with HF listed as most responsible diagnosis–0.005 (–0.012, +0.003)0.107 (0.013, 0.200)30-day death after discharge0.051 (0.046, 0.057)–0.13 (–0.30, +0.04)30-day postdischarge ED visits (all cause)–0.006 (–0.024, +0.012)–0.57 (–1.48, +0.34)30-day postdischarge hospital-based specialized clinic visits (all cause)–0.017 (–0.096, +0.063)0.18 (–0.57, +0.94)30-day postdischarge outpatient physician office visits (all cause)–0.0008 (–0.055, +0.053)0.058 (–0.38, +0.49)ACA indicates Alberta Cardiac Access; ED, emergency department; and HF indicates heart failure.Our subgroup analysis confirmed that patients discharged from ACA vanguard sites (20 162 of the 33 405 hospital discharges in our cohort) did exhibit a statistically significant and substantial step change in specialized clinic visits (absolute increase in the spring of 2008 of +12.2%; 95% CI, 0.6–23.8%), which was not seen in other areas of Alberta. Moreover, wheras ACA vanguard sites exhibited similar trends in 30-day postdischarge death/readmission rates pre-ACA as other sites (increasing 0.24% per month versus 0.30% per month), only the ACA vanguard sites exhibited a statistically significant reversal in that trend after the spring of 2008 (with a 1.6% per month; 95% CI, 0.6–2.6%, decline during the 18 months postimplementation; Figure 2).After roll out of the ACA initiative, patients hospitalized in ACA vanguard regions exhibited similar rates of index hospitalization mortality (odds ratio, 0.98; 95% CI, 0.82–1.17) and 30-day physician follow-up visits (odds ratio, 1.02; 95% CI, 0.87–1.18), as those hospitalized in other regions. However, they were 40% more likely to receive a specialized clinic visit in the first 30 days after discharge (95% CI, 23–59%), and they exhibited better outcomes than those discharged from other areas in the province: 30-day death/readmission rate after discharge 18.6% versus 22.2%, crude odds ratio 0.80 (95% CI, 0.69–0.93), and after adjustment using the model17 incorporating index hospitalization length of stay, adjusted odds ratio 0.83 (95% CI, 0.75–0.93).DiscussionWe found substantial changes in length of stay, mortality, and 30-day all-cause readmission rates for hospitalized HF patients during the past decade; however, after adjustment for temporal shifts in patient demographics and comorbidities, none of these changes were statistically significant. More than three-quarters of the patients surviving a hospitalization for HF in Alberta (which has a publicly funded universal access health care system) saw a physician within 30 days of discharge. A province-wide program designed to enhance access to specialized HF clinics was successful in its goal in that there was a 10% absolute increase in the number of patients seen in specialized HF clinics within 30 days of discharge after program implementation. However, it should be acknowledged that even after roll out of the ACA initiative, only a minority of Albertans discharged from hospital with HF are followed in specialized clinics (http://www.cardiacaccess.ab.ca).The ACA program did not lead to any change in pre-existing secular trends with respect to index hospital mortality or length of stay, and for those patients discharged from hospital there were no appreciable changes in the frequency of outpatient visits with other physicians, emergency departments, or unplanned readmissions for HF pre/post the ACA initiative. However, the ACA initiative was associated with a statistically significant reversal of the upward trend in 30-day postdischarge mortality or all-cause readmission rates observed prior to ACA implementation. This shift in outcomes was only seen in the ACA vanguard regions where access to specialized HF clinics was enhanced. Indeed, patients discharged from vanguard regions exhibited 20% lower rates of death/readmission in the first 30 days after discharge than those discharged from other Alberta regions.Our mortality rate in the first 30 days after admission (11.1%) is similar to recent reports from the American Heart Association Get with the Guidelines-Heart Failure registry (11%).3 Although our 30-day postdischarge readmission rates (18%) are lower than the Get with the Guidelines-Heart Failure registry (24.5%), the length of stay in Alberta was substantially longer than in the United States.3 We were unable to risk ad
Referência(s)