Octogenarians died mainly of cardiovascular diseases five years after acute coronary syndrome

Abstract

Objective. A substantial part of deaths and readmissions in octogenarians with acute coronary syndrome (ACS) is assumed to be of non-cardiovascular causes. However, limited data on cause-specific long-term mortality and hospital readmissions are available. This study was aimed to investigate 5-year cause-specific deaths and re-hospitalizations as well as their prognostic predictors among octogenarians with ACS managed with percutaneous coronary intervention (PCI). Methods. A total of 181 octogenarians managed with PCI on ACS indication during 2006–2007 at Sahlgrenska University Hospital were included. The time-period was chosen to allow a follow-up period of five years. Results. All-cause 5-year mortality was 46%. Approximately 70% of deaths were cardiovascular. All-cause hospital readmissions were 71%. The majority of readmissions were due to non-cardiovascular diseases, 61% of all readmissions. Cox proportional-hazard regression analyses for cardiovascular mortality identified female sex and culprit lesion in left coronary arteries as independent predictors. Negative binomial regression models showed female sex and complications during index hospitalization as independent predictors of increased cardiovascular re-hospitalizations and prior smoking as independent predictor of increased non-cardiovascular re-hospitalizations. Conclusions. In an octogenarian cohort presented with ACS treated with PCI, cardiovascular diseases were the main causes of deaths, whereas non-cardiovascular diseases were the main causes of re-hospitalizations.

Keywords:

• Cause-specific
• octogenarians
• acute coronary syndrome
• percutaneous coronary intervension
• mortality
• re-hospitalization

Introduction

A substantial part of deaths and readmissions in octogenarians with acute coronary syndrome (ACS) is assumed to be due to non-cardiovascular causes. However, limited data on causes of deaths and readmissions are available in octogenarians with ACS. This is a clinically relevant issue, not only because ACS often occurs in the elderly but also the prognosis of ACS in the elderly might be changed in the era of reperfusion therapy including percutaneous coronary intervention (PCI).[1–6] PCI is without a doubt the most important therapy that affects prognosis in ACS. Old age is associated with higher in-hospital mortality and frequent complications, such as renal failure and bleedings, after PCI.[7,8] However, results from studies on mid- and long-term outcomes of PCI are inconsistent.[9–16]

Since co-morbidities increase with age, it is assumed that non-cardiovascular causes of deaths or readmissions increase among octogenarians. For instance, in elderly patients with heart failure, only 57% of deaths can be attributed to cardiovascular causes.[17] However, the extent to which non-cardiovascular causes contribute to deaths or readmissions in octogenarians with ACS remains unknown, particularly after PCI. Therefore, the purpose of this study was to investigate the causes of deaths, readmissions, and their prognostic predictors in octogenarians with ACS, five years after PCI.

Materials and methods

Study cohort

Among 504 patients aged ≥80 years who presented with acute coronary syndrome (ACS) at Sahlgrenska University Hospital during 2006, only 84 patients were treated with percutaneous coronary intervention (PCI). Patients who did not undergo PCI were older and sicker with a higher mortality rate, 89%. Because the aim of the present study was to study the outcome after ACS in patients treated with PCI, we also included patients from 2007 to get a sufficient number. Altogether, 237 patients aged ≥80 years were treated with PCI with ACS indication during 2006–2007 at Sahlgrenska University Hospital. Fifty-six patients were excluded due to the fact that these patients did not belong to the catchment area of Sahlgrenska University Hospital, since medical records from other hospitals were not accessible. Only patients treated with PCI on ACS indication (n = 181) were included in the present study and retrospectively studied from January 2 to May 30, 2012. Only 8 patients were ≥90 years. Baseline parameters covering social, functional, and medical domains were entered into a database (Table 1). The time-period of 2006–2007 was chosen to allow a follow-up period of five years. PCI procedures for the specified age group and time period were identified from medical records. The study protocol was approved by the Ethical Committee at the University of Gothenburg.

Table 1. Baseline characteristics: comparison between men and women.

Parameters	Total, n = 181	Female, n = 70	Male, n = 111	P
Age, years, (mean ± SD), n = 181	83.7 ± 2.8	83.9 ± 2	83.6 ± 3	.545
Dead (%), n = 181	46	44	47	.649
Cardiovascular mortality (% of total mortality), n = 181	70	80	64	.111
Dependency in ADL (%), n = 181	6	6	6	.871
Current smoker (%), n = 177	5	1	7	.087
Prior smoker (%), n = 130	54	51	56	.599
Body mass index, (kg/m^2), (mean ± SD), n = 166	25 ± 4	25 ± 3	25 ± 4	.458
Clinical findings
STEMI (%)	51	61	45	.028
NSTEMI (%)	35	31	37	.404
Unstable angina pectoris (%)	14	7	18	.041
Heart rate (beats/min) (mean ± SD), n = 170	79 ± 24	78 ± 21	79 ± 26	.790
Systolic BP (mm Hg) (mean ± SD), n = 171	150 ± 27	146 ± 25	153 ± 28	.105
Diastolic BP (mm Hg) (mean ± SD), n = 170	84 ± 16	84 ± 14	85 ± 17	.817
EF % (mean ± SD), n = 162	48 ± 11	47 ± 11	48 ± 11	.401
MR, grade ≥1 (scale 0.5–4) (%), n = 152	33	34	32	.871
AS of any grade (%), n = 143	15	15	16	.861
TR, grade >2 (scale 0.5–4) (%), n = 152	6	8	5	.436
Laboratoy findings
Cholesterol (mmol/L) (mean ± SD)	4.6 ± 1.1	4.8 ± 1.2	4.4 ± 1.1	.140
Hemoglobin (g/L) (mean ± SD), n = 169	133 ± 16	131 ± 17	134 ± 16	.298
eGFR (ml/min) (mean ± SD), n = 164	53 ± 17	53 ± 18	53 ± 16	.810
Comorbidities
Atrial fibrillation (%), n = 175	19	21	19	.518
Hypertension (%), n = 181	65	65	65	.962
Diabetes mellitus (%), n = 167	29	21	21	.936
Malignancies (%), n = 180	13	11	14	.588
Prior myocardial infarction (%), n = 180	30	32	28	.554
Prior CABG (%), n = 180	5	9	4	.147
Prior PCI (%), n = 179	7	7	6	.818
Medications at admission
Aspirin (%), n = 176	48	51	46	.504
Beta-blockers (%), n = 174	55	51	56	.523
ACEI/ARB (%), n = 175	30	33	28	.477

ADL: Activities of daily living; MR: Mitral valve regurgitation; AS: Aortic valve stenosis; TR: Tricuspid valve regurgitation; eGFR: Estimated glomerular filtration rate; STEMI: ST-segment elevation myocardial infarction; NSTEMI: Non-ST-segment elevation myocardial infarction; ACEI: Angiotensin converting enzyme inhibitors; ARB: Angiotensin converting enzyme blockers.

Laboratory analyses

All laboratory variables were analyzed, as routine protocol, by the Clinical Chemistry Laboratory at Sahlgrenska University Hospital. Almost all our patients were of the same ethnicity. We needed only to adjust for age, sex and weight. Therefore, we used the Cockcroft–Gault formula to calculate the estimated glomerular filtration rate (eGFR) in mL/min.

Clinical outcome data

The primary endpoint was cause-specific mortality and hospital readmissions five years following PCI. Data on causes of death were obtained from the death registry of the National Board of Health and Welfare in Sweden. The causes of death were classified according to ICD-10. The Automated Classification of Medical Entities (ACME) program was used to select the underlying cause of death. ACME program selects an underlying cause of death by applying the WHO rules to the ICD-codes. Cardiovascular mortality was defined as the underlying cause of death when categorized as I00-I99. Data on hospital readmissions were obtained from hospital records.

Statistics

The results are presented as percentage and mean ± standard deviation. For continuous variables, statistical analysis was performed using Student’s unpaired t-test or the Mann–Whitney U-test for non-normally distributed variables. For discrete variables, the chi-square test was used. Cox proportional–hazard regression models analyzing time to event were built to analyze for the prediction of cardiovascular mortality. All parameters presented in Table 1 have been analyzed in univariable models. Parameters with p-value ≤.25 from univariable models were included in the final multivariable model. Variables of high clinical relevance, but with p-value >.25 from univariable models, have also been tested in the multivariable model, but withdrawn if resulting in insignificant p-value, ≥.05.

In case of hospital readmissions, due to over-dispersion instead of Poisson regression, negative binomial regression models with an estimated over-dispersion factor were used. By using negative binomial regression models, we could analyze for prediction of the counted outcome, number of re-admissions. The same strategy has also been used in selecting variables for negative binomial regression models.

The final multivariable Cox model included five variables that indicates about 10 events per included variable (Table 2). Negative binomial regression models included 10 and 8 variables indicating 28 and 22 events per included variable for non-cardiovascular and cardiovascular re-hospitalizations, respectively, Table 3.

Table 2. Independent predictors of five-year cardiovascular mortality from multivariable cox proportional-hazard regression analysis.

	Hazard ratio[95% Conferenceinterval]	p-Value
Culprit lesion in left coronary arteries	2.2 [1.0–4.8]	.040
Complications during index hospitalization	1.9 [1.1–3.1]	.018
Female sex	2.2 [1.1–4.6]	.032

Table 3. Independent predictors of re-hospitalizations from negative binomial regression models.

	Hazard ratio[95% Conferenceinterval]	p-Value
Five-year cardiovascular readmissions
Female sex	2.1 [1.4–3.1]	.001
Complications during index hospitalization	1.9 [1.1–3.1]	.018
Medication with beta-blockers	0.4 [0.2–0.8]	.012
Five-year non-cardiovascular readmissions
Prior smoker	2.1 [1.2–3.8]	.012

Cox models were assessed for proportional-hazard assumption for covariates graphically with adjusted log minus log curves. The hazard ratios (HR) and incidence rate ratio (IRR) with confidence intervals (CI) and p-values were presented. The PASW Statistics 18 (USA) statistical package was used for all analyses. p <.05 was regarded as statistically significant.

Results

Cause-specific deaths and hospital readmissions

All-cause mortality 5 years after PCI was 46%. Approximately 70% of causes of deaths were cardiovascular. All-cause hospital readmissions were 71% with a total of 467 readmissions. The majority of readmissions were due to non-cardiovascular diseases (285 events), constituting 61% of all readmissions, Figure 1.

Figure 1. Distribution of causes of deaths and hospital readmissions.

Predictors of cardiovascular deaths and cause-specific re-hospitalizations

Cox proportional-hazard regression analyses for time to event for cardiovascular mortality showed female sex and culprit lesion in left coronary arteries as independent predictors of poor outcome (Table 2). Negative binomial regression models with an estimated over-dispersion factor showed female sex and complications during index hospitalization as independent predictors of an increased rate of cardiovascular re-hospitalizations, but medication with beta-blockers as an independent protective factor (Table 3). In case of non-cardiovascular re-hospitalization prior smoking was the only independent predictor of increased readmission rates (Table 3). In-hospital complications included both specific PCI-related complications and general complications including infections, acute renal failure and bleedings.

Discussion

In our study, the primary causes of deaths were cardiovascular, whereas non-cardiovascular diseases were the main reasons for hospital readmissions. Our findings are highly clinically relevant. Despite PCI, all-cause mortality is still as high as 46% over 5 years. However, according to data from the Swedish Central Bureau of Statistics, the 5-year all-cause mortality in the age group 80–89 years of the normal Swedish population during the same period (2008–2012) was about 42%. Most importantly, ∼70% of causes of deaths were cardiovascular, indicating a high potential for further cardiovascular improvement. In terms of prognostic predictors, female sex and culprit lesion in left coronary arteries were independent predictors of increased cardiovascular deaths, whereas medication with beta-blockers was an independent protective factor. It is well-known that compared to men, women have a worse prognosis after ACS.[18,19] Women had a higher proportion of ST-segment elevation myocardial infarction (STEMI) compared to men. However, the regression models were adjusted for STEMI. Culprit lesion in the left side coronary arteries as a predictor of worse prognosis is not surprising, since these arteries supply the main part of the left ventricle. However, previous studies only showed increased incidence of post myocardial infarction heart failure in proximal culprit lesion compared to distally located lesions.[20] Another possible mechanism is increased risk for arrhythmias.[21] Complications during index hospitalization included both local complications during PCI and systemic complications; bleedings, acute renal failure and infections. These patients might not have received appropriate secondary preventive management due to a higher risk of side effects. In the case of non-cardiovascular re-hospitalizations, prior smoking was the only independent predictor of increased readmissions. Smoking-related conditions are still a major cause of mortality and morbidity.[22] The common diseases which have a well-known association with smoking such as chronic obstructive pulmonary diseases (COPD) and different malignancies were included in the regression analysis. This might indicate that these diseases are under-diagnosed in this patient group, only 12 patients (6%) had COPD and 10 patients (5.5%) had undergone lung function assessment tests. Prediction analysis for non-cardiovascular deaths was impossible due to few events.

Regarding the baseline characteristics and the outcomes, the present study is based on more comprehensive and better validated with less missing data than in any available registry databases. However, the patients in this cohort were selected for PCI on clinical bases on the discretion of the attending physician.

In summary, our study, based on our hospital cohort of ACS patients ≥80 years of age and treated with PCI, demonstrated that cardiovascular diseases were the main causes of deaths, whereas non-cardiovascular diseases were the main causes of hospital readmissions.

Disclosures statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.