International Journal of Cardiology 190 (2015) 42–46
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Life expectancy after implantation of a first cardiac permanent pacemaker (1995–2008): A population-based study☆,☆☆ Pamela J. Bradshaw a,⁎, Paul Stobie b, Matthew W. Knuiman a, Thomas G. Briffa a, Michael S.T. Hobbs a a b
School of Population Health, The University of Western Australia, Australia Department of Cardiovascular Medicine, Sir Charles Gairdner Hospital, Australia
a r t i c l e
i n f o
Article history: Received 24 March 2015 Accepted 14 April 2015 Available online 15 April 2015 Keywords: Cardiac pacemaker Life expectancy Comorbidity Survival
a b s t r a c t Introduction: Research suggests that survival among the recipients of a cardiac permanent pacemaker (PPM) matches the age- and sex-matched general population in the absence of cardiovascular disease. We used linked administrative data to examine life expectancy-based outcomes for adults requiring a cardiac PPM. Methods: Population-level hospital admissions data were used to identify all recipients of an initial PPM during 1995–2008. Expected years of additional life remaining at the time of implantation were calculated for each patient from population life tables. Observed years were calculated using linked mortality data to end 2011. Cox regression was used to determine demographic and clinical predictors of survival. Results: In 8757 patients age-adjusted risk of death to 5 years was associated with male sex, higher Charlson Comorbidity Index score (excluding cardiac disease), a history of heart failure, cardiomyopathy or atrial fibrillation and emergency admission. Coronary revascularisation surgery reduced long-term risk. The observed/expected ratio of additional years of life was 0.80 for men and 0.84 for women overall, varying from 0.92 for women without significant comorbidity to 0.40 for patients with the highest Charlson score and cardiomyopathy. The oldest patients (80–99 years) did relatively well, probably reflecting patient selection. Heart disease was the most frequent cause of death. Conclusions: Life expectancy among PPM recipients without significant comorbidity approached that of the general population. Greater non-cardiac comorbidity, heart failure, atrial fibrillation and, in particular, cardiomyopathy, contributed most to the loss of expected years of life in all age groups. The oldest patients and women did relatively well. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Background While long-term survival for adult recipients of a cardiac permanent pacemaker (PPM) shows the expected age-related decline [1] there is little information on what this means in terms of life expectancy (LE) compared to the general population. A single centre study from England of 803 patients with PPMs implanted in the early 1990s, with follow-up to seven years, reported overall survival fell well below that of the ageand sex-matched population [2]. The cohort included patients with a range of cardiac and non-cardiac comorbidities. More recently, a ☆ Each of the authors takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation and has given final approval to the submitted manuscript. ☆☆ Funding sources: This work was supported by a project grant from The Heart Foundation of Australia (Grant No. G 08A 3656) obtained initially by the late Professor Konrad Jamrozik. The Heart Foundation of Australia had no involvement in the conduct of the study. ⁎ Corresponding author at: School of Population Health, M431 The University of Western Australia, Stirling Highway, Nedlands WA 6009, Australia. E-mail address: [email protected] (P.J. Bradshaw).
http://dx.doi.org/10.1016/j.ijcard.2015.04.099 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.
nation-wide follow-up of 1517 patients from 2003–2007 in the Netherlands found survival to a minimum of 3.3 years significantly worse for the more than 50% of patients with concomitant cardiovascular disease (CVD) than the age- and sex-matched general population. However, survival for recipients without CVD matched that of the general population. This was also the case for the oldest old (octogenarians and nonagenarians) [3,4]. Similar findings for elderly patients without symptomatic heart disease were reported in earlier decades from a small study from Olmsted County, Minnesota [5], and survival for age– sex- and comorbidity-matched Taiwanese nonagenarians implanted with PPMs from 2001–2012 was no worse than for controls [6]. Commonly identified factors associated with increased risk for death after PPM implantation include older age, CVD, heart failure (HF) and cardiomyopathy [1–6]. ‘Years of potential life lost’ up to age 65 or 75 years is a frequently used population health measure as a reference against which to estimate premature deaths [7] but is not useful for studies of PPM outcomes as the average age of patients requiring a first PPM exceeds 75 years in developed countries [3,8]. We selected ‘expected (remaining) years of life lost’ (EYLL), based on the complete expectation of life, to measure
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
the gap between estimated LE and actual age at death [9], providing a more clinically meaningful measure of outcomes for recipients of a PPM in ageing populations. To determine whether life expectancy after initial implantation of a PPM meets that expected for the patients' age and era we calculated and compared the observed and expected years of life up to 31 December 2011 for groups of PPM recipients who received their first PPM during 1995–2008 in the state of Western Australia (WA). 2. Methods Thirty years of linked hospital and mortality data from WA were used in a retrospective cohort study to examine survival and life expectancy outcomes among a population of patients who had a first PPM implanted between 1995 and 2008. 2.1. Data sources and patient selection The data sources for the study cohort have been described previously [10]. Briefly, the study used de-identified, linked data prepared by the Data Linkage Branch of the Health Department of WA. The linked file available for this study included all admissions and deaths for all patients hospitalised in WA for CVD from 1980 to 2011. This file was used to identify PPM admissions, defined as an admission with a World Health Organisation's International Classification of Diseases (ICD-9CM and ICD-10) diagnosis or procedure for insertion or management of a cardiac pacing device (available from author) for the years 1980–2009. Admissions with a procedure code for an implantable cardioverter defibrillator alone, without an additional code for pacing, were excluded. Using 15-year look back to identify incident cases, a cohort of patients who had their first PPM admission during 1995 to 2008 was identified. As the indications for pacing differ between children and adults, we excluded patients younger than 30 years and those with congenital heart disease (n = 97), as well as five people aged ≥100 years, for whom LE data are not available/reliable. 2.2. Study variables Coded cause of death was available to the end of 2010. Cardiac conditions were identified from the ‘principal’ and 19 additional ‘diagnosis’ fields. The major indications for PPM insertion were categorised according to those used in the world surveys of pacing [11], and in a hierarchical fashion (major indication in principal diagnosis, secondary diagnosis etc.). Pacemaker type (single, dual, triple chamber or ‘not otherwise specified’) was identified from the 11 ‘procedure’ fields. The Charlson Comorbidity Index, a widely-used tool used to predict mortality by weighting a range of comorbid conditions, such as heart disease or cancer (a total of 17 conditions), was calculated. We used the method of Quan and others [12] for administrative data, based on the ICD-9CM and ICD-10 codes recorded in the index admission and all admissions in the previous five years. Acute myocardial infarction (AMI) and HF were excluded as comorbidities in the Charlson score as they were frequently the primary disorder and were considered as separate variables as were diagnoses of atrial fibrillation/flutter (AF), cardiomyopathy or cardiac valve disease recorded in the index admission. The final Charlson score was categorised as zero (no comorbid non-cardiac conditions recorded), a score of 1–2, 3–4, or of 5 or more. The ‘complete expectation of life at exact age’ was calculated for each patient using the life-tables for WA [13–15] and based on the patient's age in the year of PPM implantation. For example, a female who had a first pacemaker implanted in 2001 at the age of 72 years could expect to live another 15.5 years, to year 2016. The total ‘expected’ years of remaining life for the whole cohort was calculated from the date of index PPM hospital discharge to the date of LE or the study end date (31st December, 2011), whichever was first,
43
and total ‘observed’ years of life to the date of death or the end date (to a maximum of 16 years), whichever came first. The EYLL up to the study end date was calculated by subtracting the total ‘observed’ years of life from total ‘expected’ years of life. Coded ‘underlying cause of death’ was available for deaths up to end of 2010 and was grouped into system-related deaths using the tenth revision, Australian Modification, of ICD-10 e.g. Chapter 9: Diseases of the Circulatory System (I00–I99). Specific conditions such as ischaemic heart disease (I20–I25) and Alzheimer's disease (G30) were identified for comparison with national age-specific causes of death [16]. 2.3. Study outcomes Long-term survival, survival compared to the general population (the ratio of observed to expected years of additional life) and the EYLL for those discharged from hospital alive after initial implantation of a PPM. 2.4. Ethics The study was approved by the Human Research Ethics Committees at the University of Western Australia and Health Department of Western Australia. 2.5. Statistical analysis The demographic and clinical characteristics of the cohort are described using the mean and standard deviation (SD) and percentages. The cohort was grouped by age at PPM implantation into 30–59, 60–70 and 80–99 years. Age at death was grouped as 30–44, 45–54, 55–64, 65–74, 75–84 and ≥ 85 years for comparison with Australian national ‘cause of death’ statistics. Groups were defined by sex, age-group at implantation, era (i.e. calendar period), cardiac history, diagnosis, Charlson score, and emergency or elective hospital admission. Differences between groups were tested using the chi-squared test for categorical factors and the Student's ‘t-test’ or one-way ANOVA for quantitative variables. Observed and expected remaining survivals were compared across groups using two statistics: the ratio of the total observed to total expected years for each group, and the mean EYLL for each group. Two time-to-death (from all causes) outcomes were analysed using Cox regression models. The first (one year mortality) restricted followup to one year and the second (5-year mortality) used all available follow-up up to 5 years for each patient, censored at 31st December, 2011. Each variable was initially tested separately for associations with one-year and 5-year all-cause mortality in a Cox model that adjusted for age group as a covariate (age-adjusted univariate associations). Variables significantly associated with each mortality outcome, after adjustment for age, were then included in a multivariable Cox model to identify the independent predictors of death. The assumption of proportionality of hazards in the Cox model was tested using an interaction between follow-up time and the Variable, and in all cases the assumption was met. A ‘p’ value of b0.05 was considered statistically significant. Analyses were carried out in IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY: IBM Corp. SPSS Version 21. 3. Results 3.1. Patient characteristics There were 8847 incident PPM cases aged 30–99 years during the period 1995–2008, with 90 deaths in hospital. The characteristics of the cohort of 8757 patients discharged from hospital are shown in Table 1 by study period; 59% were male, 39% were aged 80–99 years, and about half were emergency admissions.
44
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
Table 1 Characteristics of 8757 patients discharged from hospital after implantation of a cardiac PPM, 1995–2008. Patient characteristics n (%)
1995–1999 n = 2199
2000–2004 n = 3108
2005–2008 n = 3450
Overall N = 8757
Men Age group (years) 30–59 60–79 80–99 Admission — elective Emergency Charlson score⁎ Zero 1–2 3–4 ≥5 AMI principal diagnosis AMI within last 5 years CHF index admission CHF within last 5 years Cardiomyopathy — index Atrial fibrillation — index Indications for PPM High-level block Bundle-branch block SA-node dysfunction Atrial fibrillation/flutter Other dysrhythmia⁎⁎ CSS/NCGS Ablation Cardiomyopathy HF None⁎⁎⁎
1301 (59)
1831 (59)
2068 (60)
5200 (59)
230 (10) 1217 (55) 752 (34) 1100 (50) 1099 (50)
253 (8) 1585 (51) 1270 (41) 1568 (50) 1540 (50)
320 (9) 1711 (50) 1419 (41) 1938 (56) 1512 (44)
803 (9) 4513 (52) 3441 (39) 4606 (53) 4151 (47)
536 (24) 995 (45) 450 (20) 218 (10) 37 (2) 510 (23) 374 (17) 930 (42) 89 (4) 647 (29)
880 (28) 1364 (44) 585 (19) 279 (9) 61 (2) 667 (22) 319 (10) 1131 (36) 87 (3) 975 (31)
945 (27) 1397 (40) 728 (21) 380 (11) 69 (2) 777 (22) 424 (12) 1355 (39) 158 (5) 1016 (29)
2361 (27) 3756 (43) 1763 (20) 877 (10) 167 (2) 1954 (22) 1117 (13) 3416 (39) 334 (4) 2638 (30)
934 (42) 226 (10) 365 (17) 272 (12) 221 (10) 38 (2) 89 (4) 22 (1) 6 (b1) 26 (1)
1177 (38) 258 (8) 854 (28) 409 (13) 167 (5) 136 (4) 55 (2) 17 (b1) 18 (b1) 17 (b1)
1217 (35) 350 (10) 804 (24) 464 (13) 257 (7) 123 (4) 66 (2) 62 (2) 78 (2) 29 (b1)
3328 (38) 834 (10) 2023 (23) 1145 (13) 645 (7) 297 (3) 210 (2) 101 (1) 102 (1) 72 (b1)
644 (29) 1445 (66) 0 110 (5) 211 (10)
822 (26) 2257 (73) 21 (b1) 8 (b1) 287 (9)
693 (20) 2067 (60) 296 (9) 394 (11) 261 (8)
2159 (25) 5769 (66) 317 (4) 512 (6) 759 (9)
Pacemaker type Single Dual Triple Not specified All-cause mortality 1 year
AMI = acute myocardial infarction, HF = heart failure, VF = ventricular fibrillation, SA = sino-atrial, AF = atrial fibrillation, CSS/NCGS = carotid sinus syncope/neurocardiogenic syncope. ⁎ Charlson Comorbidity Index score excluding AMI and HF. ⁎⁎ Includes paroxysmal tachycardia, ventricular fibrillation and flutter, sinus bradycardia and other specified and unspecified rhythm disorders. ⁎⁎⁎ No major indication coded — includes ‘cardiac arrest’ or other ‘cardiac complications’, myocarditis, pericarditis, hypertensive heart disease and ‘other-ill-defined heart disease’.
Women were older than men (mean 76.3, SD 10.8 vs 74.3, SD11.1 years, p b 0.001) and more likely to present with sick sinus syndrome (27% vs 20%, b0.001). There was little difference in AMI or acute coronary syndrome as a principal diagnosis in the index admission, but women were less likely to have ischaemic heart disease (IHD) recorded as a new or existing condition (33% vs 38%, χ2 = 25.4, p = b 0.001). Emergency admission (≈90% having a principal diagnosis of a conduction disorder or dysrhythmia) was more likely for patients with a higher Charlson score, being 40% of those with no cardiac comorbidity rising to 59% among those with the highest Charlson score (χ2trend = 134.9, df = 3, p = b0.001). Single chamber PPMs were more frequently implanted in patients aged ≥80 years (35% vs b20% in younger patients), with dual-chamber pacing the most frequent in each age-group (58% in ≥80 years, N70% in younger patients).
3.2. Mortality There were 759 deaths (9%) within one year of hospital discharge. Crude all-cause 1-year mortality decreased across the study (χ2 trend = 8.9, 2p = 0.01) (Table 1). Mortality was highest among those aged 80–99 years (13.5%) being more than double those aged 60–79 (6%) and 30–59 years (3.5%, χ2 trend = 173.2, 2p = b0.001). Crude mortality
at one year was greater for men (10%) than women (6.7%, χ2 = 27.9, 2p = b0.001). Table 2 shows the estimated hazard ratios from the multivariable Cox models for one-year mortality and for survival to 5-years from hospital discharge. Age and male sex were significant predictors of both 1-year mortality and poorer survival to five years, as were emergency admission, higher non-cardiac Charlson score, HF, cardiomyopathy, and AF. Coronary artery bypass graft surgery in the index admission was associated with better survival to five years, but not death within the first year. Era of implantation, and clinical factors including the major indication for pacing, initial pacing mode, admission for AMI or acute coronary syndrome, were not independent predictors of mortality after adjustment for age. 3.3. Comparison of observed and expected years of remaining life Among the 8757 PPM patients discharged from hospital, a total of 3804 (43%) died before attaining the date of their calculated LE, 1366 patients (16%) met or exceeded their LE by the end of 2011; the remaining 3587 (41%) patients were alive and had an expectation of life extending beyond the study end date. Among the 3478 patients (40%) whose LE date was before the December 31st 2011, 1366 (39%) met or exceeded their LE. There were no patients younger than 60 years in this group; the proportion of those who met their LE was 37% for those aged 60–79 and 41% for those 80–99 years. Almost half (48%) of all patients with a zero comorbidity score met or exceeded their LE, while this proportion fell to 28% for those with the highest Charlson scores. The ratio of total observed to total expected years of remaining life (to a maximum of 16 years) was 0.80 for men and 0.84 for women overall. The ratio declined with increasing Charlson score, with the poorest comparative survival being among patients with the highest score and HF or cardiomyopathy recorded in the index admission (Table 3). The mean EYLL was higher among men (1.41 years, 95% CI 1.33– 1.50) than women (1.19 years, 95% CI 1.09–1.30 p = 0.001) and increased with increasing non-cardiac Charlson score, with the greatest EYLL being among those patients with the highest non-cardiac comorbidity score and HF or cardiomyopathy recorded in the index admission
Table 2 Estimated hazard ratios from multivariable Cox models showing associations between demographic and clinical factors and risk for all-cause mortality at 1 year and survival to 5 years among 8757 patients after hospital discharge for initial PPM implantation 1995–2008. Factor
One year HR (95% CI) p-value
Men 1.00 Women 0.62 (0.53–0.72) b 0.001 Age group — years 30–59 1.00 60–79 1.69 (1.14–2.50) b 0.01 80–99 3.94 (2.65–5.80) b 0.001 Elective 1.00 Emergency 1.61 (1.38–1.87) b 0.001 Charlson score (non-cardiac morbidity) Zero 1.00 1–2 1.33 (1.06–1.66) 0.01 3–4 1.81 (1.42–2.30) b 0.001 ≥5 2.52 (1.95–3.25) b 0.001 Heart failure⁎ 2.07 (1.74–2.45) b 0.001 Cardiomyopathy⁎ 1.65 (1.23–2.21) 0.001 Atrial fibrillation⁎ 1.24 (1.07–1.44) b 0.01 CABG⁎ NA
Five years HR (95% CI) p-value 1.00 0.73 (0.67–0.77) b0.001 1.00 2.21 (1.84–2.64) b0.001 4.73 (3.95–5.66) b0.001 1.00 1.21 (1.13–1.28) b0.001 1.00 1.10 (1.02–1.18) 0.02 1.27 (1.16–1.39) b 0.001 1.62 (1.46–1.80) b 0.001 1.66 (1.53–1.80) b 0.001 1.23 (1.06–1.44) b 0.01 1.12 (1.05–1.19) b 0.001 0.68 (0.54–0.88) b 0.01
Age and any variable that was significantly associated with all-cause mortality after adjustment for age is included in the model. Other variables in the model included era, indication for pacing, initial pacing mode and admission for AMI or acute coronary syndrome, CABG = coronary artery bypass surgery in index admission. ⁎ Condition/procedure recorded in index admission.
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46 Table 3 Comparative survival among 8757 patients after initial PPM implantation 1995–2008 versus the general population: ratio of observed/expected years of remaining life up to 31st December 2011, by Charlson Comorbidity Index score. Ratio of observed to expected years of additional life Charlson score⁎
Zero
1–2
3–4
≥5
Men Women Age group (years) 30–59 60–79 80–99 Elective Emergency No heart failure Heart failure⁎⁎
0.88 0.92
0.82 0.85
0.73 0.76
0.64 0.71
0.95 0.87 0.93 0.91 0.88 0.90 – 0.90 – 0.91 0.86
0.90 0.80 0.86 0.86 0.79 0.85 0.68 0.83 0.74 0.84 0.81
0.86 0.72 0.74 0.80 0.70 0.79 0.59 0.75 0.67 0.77 0.68
0.70 0.64 0.70 0.75 0.61 0.73 0.49 0.69 0.40 0.69 0.63
No cardiomyopathy Cardiomyopathy⁎⁎ No Atrial fibrillation Atrial fibrillation⁎⁎
⁎ Charlson Comorbidity Index score excluding cardiac comorbidity. ⁎⁎ Condition recorded in index admission.
(Table 4). All these patients suffered additional comorbid conditions including diabetes, renal, pulmonary and peripheral vascular diseases. 3.4. Cause of death Cause of death was coded for 4163 (90%) of 4600 deaths. ‘Cardiac causes’ consisting of IHD and ‘other cardiac deaths’ were the most frequently recorded, being 43% for both men and women and ≥ 40% of deaths in each age group, with IHD being more than 50% of those deaths, except among those aged b 44 years. IHD as cause of death exceeded the proportion in the Australian general population; being double that for the general population for those aged 45–54 years (30% vs 14%) and N1.5 for those aged 55–75 years. Deaths from non-cardiac causes were lower than expected when compared with national statistics. This included cancer, the peak among PPM patients being 25% in those aged 65–74 years at death (vs 41% in that age-group in the general population). Cerebrovascular disease, ‘dementia’ and Alzheimer's disease increased with age among PPM patients but were lower than national statistics. 4. Discussion Patients without significant cardiac or non-cardiac comorbidity achieved around 90% of the total expected additional years of life to
45
the study end. This was the case for around one quarter of the cohort, for the majority the ratio worsened with increasing comorbidity, and with additional cardiac comorbidity. Among those patients whose total expectation of life fell wholly within the study period, survival to LE or beyond was almost 50% (e.g. equivalent to the general population), in the absence of comorbidities. This finding is supported by similar results from the Netherlands, where survival to a minimum of 3.3 years for patients who did not have CVD was estimated to be equivalent to age- and sex-matched controls, (patients with serious conditions such as cancer, and immune and degenerative diseases at implantation having been excluded from the analysis) [3]. The ratio of observed to expected additional years of life was lower for those aged 60–79 years than for those aged 80–99 years in our study, regardless of comorbidity score. Similarly favourable outcomes were reported for the eighty- and ninety-year olds in the Dutch study, with survival for elderly patients, including those with CVD, equivalent to population controls [3]. The relatively good survival among elderly patients in our study may represent greater selection for PPM among the oldest patients, who are already ‘healthy survivors’. Despite possible selection, it is evident that age was no barrier to PPM use, with those aged eighty years or more being 40% of all adult recipients, a proportion found in other developed countries such as France, Italy, Denmark, Sweden, the United Kingdom and Israel in 2005 [11]. At the time of PPM implantation 52% of men and 40% of women in the study cohort had exceeded the life expectancy at birth for men (75.2 years) and women (81.0 years) in WA born at the start of the study (1996) [17]. The outcome was more favourable for women compared to men, as observed in other studies [1–3], and this benefit may be associated with the lower prevalence of IHD among women, and, being older, they may also have been a more select population. Both cardiac and non-cardiac comorbidities were associated with the loss of expected years of life; the EYLL being greatest among those with the highest non-cardiac Charlson score and cardiomyopathy (almost 5 years) or HF (3.5 years). The significant contribution of increasing non-cardiac comorbidity to reduced LE is evident from this study. Among the few PPM studies to employ a measure of comorbidity, the large Healthcare Cost and Utilization Project-National Inpatient Sample from the United States found the Charlson score a stronger predictor than age of in-hospital mortality, complications, length of stay and costs [18]. Emergency admission has been previously identified as a risk for inhospital death; in the same 2004–2008 National Inpatient Samplebased study non-elective admissions incurred three times the risk for
Table 4 Mean expected years of remaining life lost to 31st December 2011 among patients after initial PPM implantation 1995–2009, by Charlson Comorbidity Index score. Mean expected years of life lost (95% CI) Charlson score⁎
Zero
1–2
3–4
≥5
Men Women Age group 30–59 60–79 80–99 Elective Emergency No HF HF⁎⁎
0.83 (0.69–0.98) 0.58 (0.40–0.75)
1.32 (1.19–1.46) 1.14 (0.99–1.29)
1.87 (1.67–2.07) 1.73 (1.49–1.98)
2.38 (2.07–2.69) 2.13 (1.75–2.51)
0.45 (0.20–0.70) 1.03 (0.87–1.19) 0.38 (0.20–0.56) 0.65 (0.51–0.79) 0.84 (0.66–1.02) 0.73 (0.60–0.86) – 0.73 (0.60–0.86) – 0.61 (0.48–0.74) 1.04 (0.81–1.27)
0.90 (0.62–1.19) 1.68 (1.53–1.82) 0.76 (0.62–0.91) 1.03 (0.90–1.16) 1.50 (1.34–1.66) 1.09 (0.98–1.19) 2.28 (1.97–2.58) 1.21 (1.11–1.31) 2.02 (1.48–2.57) 1.18 (1.06–1.30) 1.40 (1.21–1.58)
1.26 (0.76–1.76) 2.24 (2.01–2.47) 1.37 (1.16–1.58) 1.41 (1.21–1.61) 2.16 (1.93–2.39) 1.49 (1.33–1.66) 2.95 (2.57–3.33) 1.77 (1.61–1.93) 2.52 (1.76–3.28) 1.60 (1.42–1.78) 2.29 (2.00–2.58)
2.58 (1.50–3.66) 2.90 (2.52–3.28) 1.56 (1.27–1.84) 1.81 (1.44–2.19) 2.62 (2.31–2.93) 1.87 (1.61–2.13) 3.54 (3.01–4.08) 2.13 (1.90–2.37) 4.90 (3.49–6.30) 2.13 (1.82–2.43) 2.57 (2.18–2.97)
No cardiomyopathy Cardiomyopathy⁎⁎ No AF AF⁎⁎
AF = atrial fibrillation, HF = heart failure. ⁎ Charlson Comorbidity Index score excluding cardiac comorbidity. ⁎⁎ Medical condition recorded in index admission.
46
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
in-hospital mortality among those aged 70 years or more [18]. In our cohort this disadvantage, independent of age and comorbidity, persisted at one year and to 5 years. The lack of a significant association between survival and pacemaker type or index arrhythmia has been noted elsewhere for the 1990s, unlike previous decades [1], and among those aged 80 years or more [4]. Compared with the few studies to report cause-specific death, the proportion of cardiac deaths in the WA population (43%) was somewhat higher than the 33% reported after a mean of 5.8 years in a population study from The Netherlands [3], and at a median follow-up of 33 months in the Mode Selection Trial (MOST) where 35% of deaths were attributed to cardiac causes, 49% to non-cardiac causes, and 16% unknown [19]. COD coding may account for some of the difference, as ‘cardiac deaths’ in the Dutch study were restricted to AMI, HF, PPM-related, rhythm disturbances and ‘resuscitation for unknown causes’. Differences in population demographics, the health care environments, and length of follow-up can also influence the reporting of cause of death. In this population-based cohort it seems likely that those selected for treatment by implantation of a PPM, especially among the very old, are those who have a relatively good likelihood of survival. This would account for the lower proportion of age-stratified deaths from cancer, stroke and dementia than observed at the national level, albeit on low numbers of patients. Although medical interventions become more complicated and expensive in patients with limited LE due to advanced age [8,18], the oldest patients, in particular, may benefit from improvements in mobility, cognitive function, and activities of daily living [20,21]. Increasing LE and longer follow-up, even for older patients, has been associated with relatively low costs [22], making advanced age alone no argument for not implanting a PPM if indicated. 5. Strengths and limitations This strength of this study is access to 30 years of routinely collected hospital admission and separation data, providing a ‘real world’, population-based cohort of incident PPM implantations. The WA Linked Data are a highly accurate compilation of person-based records from both public and private hospitals within WA. The validity of the matching for the WA hospital morbidity data has been assessed, with invalid links (false positives) and missed links (false negatives) estimated at 0.11% each [23]. A validation study of the linked data against PPM procedures recorded in the catheter laboratory register at a single tertiary centre found less than 2% discordance [24]. No adjustment to EYLL for disability was required, as implantation of a pacemaker is generally associated with improved quality of life and health values [20,21]. We were not, however, able to describe the LE outcomes across all age-groups within the 16 years of the study period. For most of patients younger than 60 years at implantation the expectation of life was years, or decades, in the future. The use of administrative data does not permit the precision in determining the indication for pacing of clinical data. Fortunately, indication was not significantly associated with the outcomes. Similarly, other than the principal diagnosis, it is not always possible to distinguish between acute and chronic comorbid medical conditions (such as HF) recorded during the index admission. 6. Conclusions Life expectancy among PPM recipients without significant comorbidity approaches that of the general population. Greater non-cardiac comorbidity, HF, AF and, in particular, cardiomyopathy, contribute most to the loss of expected years of life in all age groups. The oldest patients are probably a relatively healthy and select population and older women, in particular, have several years of life remaining.
Disclosures None of the authors has any conflicts of interest. Dr Paul Stobie has participated in Advisory Board meetings for Medtronic, Inc. and St Jude Medical Inc.
Acknowledgements The authors thank the data custodians, data linkage and client services staff of Department of Health WA for assistance with gaining approvals and the linkage, extraction and checking of the linked data.
References [1] M. Brunner, M. Olschewski, A. Geibel, C. Bode, M. Zehender, Long-term survival after pacemaker implantation. Prognostic importance of gender and baseline patient characteristics, Eur. Heart J. 25 (2004) 88–95. [2] J. Pyatt, J. Somauroo, M. Jackson, et al., Long-term survival after permanent pacemaker implantation: analysis of predictors for increased mortality, Europace 4 (2002) 113–119. [3] E. Udo, N. van Hemel, N. Zuithoff, P. Doevendans, K. Moons, Prognosis of bradycardia pacemaker recipient assessed at first implantation: a nationwide cohort study, Heart 99 (21) (Nov 2013) 1573–1578. [4] E. Udo, N. van Hemel, N. Zuithoff, et al., Long-term outcomes of cardiac pacing in octogenarians and nonagenarians, Europace 14 (2012) 502–508. [5] W-K. Shen, D. Hayes, S. Hammill, et al., Survival and functional independence after implantation of a permanent pacemaker in octogenarians and nonagenarians, Ann. Intern. Med. 125 (1996) 476–480. [6] T-F. Chao, C-J. Liu, T-C. Tuan, et al., Long-term prognosis of patients older than ninety years after permanent pacemaker implantation: does the procedure save the patients? Can. J. Cardiol. 30 (10) (Oct 2014) 1196–1201. [7] Australian Institute of Health and Welfare, Premature mortality from chronic disease, Bulletin no. 84. Cat. no. AUS 133, AIHW, Canberra, 2010. [8] A. Greenspon, J. Patel, E. Lau, et al., Trends in permanent pacemaker implantation in the United States from 1993 to 2009: increasing complexity of patients and procedures, J. Am. Coll. Cardiol. 60 (2012) 1540–1545. [9] T. Aragon, D. Lichtensztajn, R. Katcher, R. Reiter, M. Katz, Calculating expected years of life lost for assessing local ethnic disparities in causes of premature death, BMC Public Health 8 (2008) 116. [10] P. Bradshaw, P. Stobie, T. Briffa, M.S. Hobbs, Use and long-term outcomes of implantable cardioverter-defibrillators, 1990 to 2009, Am. Heart J. 165 (2013) 816–822. [11] H. Mond, M. Irwin, H. Ector, A. Proclemer, The world survey of cardiac pacing and cardioverter-defibrillators: calendar year 2005, PACE 31 (2008) 1202–1212. [12] H. Quan, V. Sundararajan, P. Halfon, et al., Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data, Med. Care 43 (2005) 1130–1139. [13] Australian Bureau of Statistics, Cat no. 3312.5 — Deaths, Western Australia, http:// www.abs.gov.au/ausstats/[email protected]/mf/3101.01993. [14] Australian Bureau of Statistics, Cat no. 3311.5 — Demography, Western Australia, http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3311.520012001. [15] Australian Bureau of Statistics, Cat no. 3302.5.55.001 — Life Tables, Western Australia, 2003–2005 and 2007–2009, http://abs.gov.au/AUSSTATS/[email protected]/ DetailsPage/3302.5.55.0012007-2009?OpenDocument. [16] Australian Bureau of Statistics, Causes of Death, Australia, 2000. Canberra. ABS Cat no: 3303.0, http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3303.02000? OpenDocument. [17] J.M. Katzenellenbogen, P. Somerford, S. Serafino, Western Australian Burden of Disease Study: Mortality 2000, Department of Health, Perth, Western Australia, July 2003. [18] A. Mandawat, J. Curtis, A. Mandawat, V. Njike, R. Lampert, Safety of pacemaker implantation in nonagenarians, Circulation 127 (2013) 1453–1465. [19] G. Flaker, A. Greenspon, B. Tardiff, E. Schron, L. Goldman, A. Hellkamp, K. Lee, G. Lamas, Mode Selection Trial (MOST) Investigators, Death in patients with permanent pacemakers for sick sinus syndrome, Am. Heart J. 146 (2003) 887–893. [20] F. Lopez-Jimenez, L. Goldman, E.J. Orav, et al., Health values before and after pacemaker implantation, Am. Heart J. 144 (2002) 687–692. [21] E. Udo, N. van Hemel, N. Zuithoff, et al., Long term quality-of-life in patients with bradycardia pacemaker implantation, Int. J. Cardiol. 168 (2013) 2159–2163. [22] B. Schmidt, M. Brunner, M. Olschewski, et al., Pacemaker therapy in very elderly patients: long-term survival and prognostic parameters, Am. Heart J. 146 (2003) 908–913. [23] C. Holman, A. Bass, I. Rouse, M. Hobbs, Population-based linkage of health records in Western Australia: development of a health services research linked database, Aust. N. Z. J. Public Health 23 (1999) 453–459. [24] P.J. Bradshaw, P. Stobie, M.W. Knuiman, T.G. Briffa, M.S.T. Hobbs, Trends in the incidence and prevalence of cardiac pacemaker insertions in an ageing population, Open Heart 1 (1) (Dec 10 2014) e000177.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Life expectancy after implantation of a first cardiac permanent pacemaker (1995–2008): A population-based study☆,☆☆ Pamela J. Bradshaw a,⁎, Paul Stobie b, Matthew W. Knuiman a, Thomas G. Briffa a, Michael S.T. Hobbs a a b
School of Population Health, The University of Western Australia, Australia Department of Cardiovascular Medicine, Sir Charles Gairdner Hospital, Australia
a r t i c l e
i n f o
Article history: Received 24 March 2015 Accepted 14 April 2015 Available online 15 April 2015 Keywords: Cardiac pacemaker Life expectancy Comorbidity Survival
a b s t r a c t Introduction: Research suggests that survival among the recipients of a cardiac permanent pacemaker (PPM) matches the age- and sex-matched general population in the absence of cardiovascular disease. We used linked administrative data to examine life expectancy-based outcomes for adults requiring a cardiac PPM. Methods: Population-level hospital admissions data were used to identify all recipients of an initial PPM during 1995–2008. Expected years of additional life remaining at the time of implantation were calculated for each patient from population life tables. Observed years were calculated using linked mortality data to end 2011. Cox regression was used to determine demographic and clinical predictors of survival. Results: In 8757 patients age-adjusted risk of death to 5 years was associated with male sex, higher Charlson Comorbidity Index score (excluding cardiac disease), a history of heart failure, cardiomyopathy or atrial fibrillation and emergency admission. Coronary revascularisation surgery reduced long-term risk. The observed/expected ratio of additional years of life was 0.80 for men and 0.84 for women overall, varying from 0.92 for women without significant comorbidity to 0.40 for patients with the highest Charlson score and cardiomyopathy. The oldest patients (80–99 years) did relatively well, probably reflecting patient selection. Heart disease was the most frequent cause of death. Conclusions: Life expectancy among PPM recipients without significant comorbidity approached that of the general population. Greater non-cardiac comorbidity, heart failure, atrial fibrillation and, in particular, cardiomyopathy, contributed most to the loss of expected years of life in all age groups. The oldest patients and women did relatively well. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Background While long-term survival for adult recipients of a cardiac permanent pacemaker (PPM) shows the expected age-related decline [1] there is little information on what this means in terms of life expectancy (LE) compared to the general population. A single centre study from England of 803 patients with PPMs implanted in the early 1990s, with follow-up to seven years, reported overall survival fell well below that of the ageand sex-matched population [2]. The cohort included patients with a range of cardiac and non-cardiac comorbidities. More recently, a ☆ Each of the authors takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation and has given final approval to the submitted manuscript. ☆☆ Funding sources: This work was supported by a project grant from The Heart Foundation of Australia (Grant No. G 08A 3656) obtained initially by the late Professor Konrad Jamrozik. The Heart Foundation of Australia had no involvement in the conduct of the study. ⁎ Corresponding author at: School of Population Health, M431 The University of Western Australia, Stirling Highway, Nedlands WA 6009, Australia. E-mail address: [email protected] (P.J. Bradshaw).
http://dx.doi.org/10.1016/j.ijcard.2015.04.099 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.
nation-wide follow-up of 1517 patients from 2003–2007 in the Netherlands found survival to a minimum of 3.3 years significantly worse for the more than 50% of patients with concomitant cardiovascular disease (CVD) than the age- and sex-matched general population. However, survival for recipients without CVD matched that of the general population. This was also the case for the oldest old (octogenarians and nonagenarians) [3,4]. Similar findings for elderly patients without symptomatic heart disease were reported in earlier decades from a small study from Olmsted County, Minnesota [5], and survival for age– sex- and comorbidity-matched Taiwanese nonagenarians implanted with PPMs from 2001–2012 was no worse than for controls [6]. Commonly identified factors associated with increased risk for death after PPM implantation include older age, CVD, heart failure (HF) and cardiomyopathy [1–6]. ‘Years of potential life lost’ up to age 65 or 75 years is a frequently used population health measure as a reference against which to estimate premature deaths [7] but is not useful for studies of PPM outcomes as the average age of patients requiring a first PPM exceeds 75 years in developed countries [3,8]. We selected ‘expected (remaining) years of life lost’ (EYLL), based on the complete expectation of life, to measure
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
the gap between estimated LE and actual age at death [9], providing a more clinically meaningful measure of outcomes for recipients of a PPM in ageing populations. To determine whether life expectancy after initial implantation of a PPM meets that expected for the patients' age and era we calculated and compared the observed and expected years of life up to 31 December 2011 for groups of PPM recipients who received their first PPM during 1995–2008 in the state of Western Australia (WA). 2. Methods Thirty years of linked hospital and mortality data from WA were used in a retrospective cohort study to examine survival and life expectancy outcomes among a population of patients who had a first PPM implanted between 1995 and 2008. 2.1. Data sources and patient selection The data sources for the study cohort have been described previously [10]. Briefly, the study used de-identified, linked data prepared by the Data Linkage Branch of the Health Department of WA. The linked file available for this study included all admissions and deaths for all patients hospitalised in WA for CVD from 1980 to 2011. This file was used to identify PPM admissions, defined as an admission with a World Health Organisation's International Classification of Diseases (ICD-9CM and ICD-10) diagnosis or procedure for insertion or management of a cardiac pacing device (available from author) for the years 1980–2009. Admissions with a procedure code for an implantable cardioverter defibrillator alone, without an additional code for pacing, were excluded. Using 15-year look back to identify incident cases, a cohort of patients who had their first PPM admission during 1995 to 2008 was identified. As the indications for pacing differ between children and adults, we excluded patients younger than 30 years and those with congenital heart disease (n = 97), as well as five people aged ≥100 years, for whom LE data are not available/reliable. 2.2. Study variables Coded cause of death was available to the end of 2010. Cardiac conditions were identified from the ‘principal’ and 19 additional ‘diagnosis’ fields. The major indications for PPM insertion were categorised according to those used in the world surveys of pacing [11], and in a hierarchical fashion (major indication in principal diagnosis, secondary diagnosis etc.). Pacemaker type (single, dual, triple chamber or ‘not otherwise specified’) was identified from the 11 ‘procedure’ fields. The Charlson Comorbidity Index, a widely-used tool used to predict mortality by weighting a range of comorbid conditions, such as heart disease or cancer (a total of 17 conditions), was calculated. We used the method of Quan and others [12] for administrative data, based on the ICD-9CM and ICD-10 codes recorded in the index admission and all admissions in the previous five years. Acute myocardial infarction (AMI) and HF were excluded as comorbidities in the Charlson score as they were frequently the primary disorder and were considered as separate variables as were diagnoses of atrial fibrillation/flutter (AF), cardiomyopathy or cardiac valve disease recorded in the index admission. The final Charlson score was categorised as zero (no comorbid non-cardiac conditions recorded), a score of 1–2, 3–4, or of 5 or more. The ‘complete expectation of life at exact age’ was calculated for each patient using the life-tables for WA [13–15] and based on the patient's age in the year of PPM implantation. For example, a female who had a first pacemaker implanted in 2001 at the age of 72 years could expect to live another 15.5 years, to year 2016. The total ‘expected’ years of remaining life for the whole cohort was calculated from the date of index PPM hospital discharge to the date of LE or the study end date (31st December, 2011), whichever was first,
43
and total ‘observed’ years of life to the date of death or the end date (to a maximum of 16 years), whichever came first. The EYLL up to the study end date was calculated by subtracting the total ‘observed’ years of life from total ‘expected’ years of life. Coded ‘underlying cause of death’ was available for deaths up to end of 2010 and was grouped into system-related deaths using the tenth revision, Australian Modification, of ICD-10 e.g. Chapter 9: Diseases of the Circulatory System (I00–I99). Specific conditions such as ischaemic heart disease (I20–I25) and Alzheimer's disease (G30) were identified for comparison with national age-specific causes of death [16]. 2.3. Study outcomes Long-term survival, survival compared to the general population (the ratio of observed to expected years of additional life) and the EYLL for those discharged from hospital alive after initial implantation of a PPM. 2.4. Ethics The study was approved by the Human Research Ethics Committees at the University of Western Australia and Health Department of Western Australia. 2.5. Statistical analysis The demographic and clinical characteristics of the cohort are described using the mean and standard deviation (SD) and percentages. The cohort was grouped by age at PPM implantation into 30–59, 60–70 and 80–99 years. Age at death was grouped as 30–44, 45–54, 55–64, 65–74, 75–84 and ≥ 85 years for comparison with Australian national ‘cause of death’ statistics. Groups were defined by sex, age-group at implantation, era (i.e. calendar period), cardiac history, diagnosis, Charlson score, and emergency or elective hospital admission. Differences between groups were tested using the chi-squared test for categorical factors and the Student's ‘t-test’ or one-way ANOVA for quantitative variables. Observed and expected remaining survivals were compared across groups using two statistics: the ratio of the total observed to total expected years for each group, and the mean EYLL for each group. Two time-to-death (from all causes) outcomes were analysed using Cox regression models. The first (one year mortality) restricted followup to one year and the second (5-year mortality) used all available follow-up up to 5 years for each patient, censored at 31st December, 2011. Each variable was initially tested separately for associations with one-year and 5-year all-cause mortality in a Cox model that adjusted for age group as a covariate (age-adjusted univariate associations). Variables significantly associated with each mortality outcome, after adjustment for age, were then included in a multivariable Cox model to identify the independent predictors of death. The assumption of proportionality of hazards in the Cox model was tested using an interaction between follow-up time and the Variable, and in all cases the assumption was met. A ‘p’ value of b0.05 was considered statistically significant. Analyses were carried out in IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY: IBM Corp. SPSS Version 21. 3. Results 3.1. Patient characteristics There were 8847 incident PPM cases aged 30–99 years during the period 1995–2008, with 90 deaths in hospital. The characteristics of the cohort of 8757 patients discharged from hospital are shown in Table 1 by study period; 59% were male, 39% were aged 80–99 years, and about half were emergency admissions.
44
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
Table 1 Characteristics of 8757 patients discharged from hospital after implantation of a cardiac PPM, 1995–2008. Patient characteristics n (%)
1995–1999 n = 2199
2000–2004 n = 3108
2005–2008 n = 3450
Overall N = 8757
Men Age group (years) 30–59 60–79 80–99 Admission — elective Emergency Charlson score⁎ Zero 1–2 3–4 ≥5 AMI principal diagnosis AMI within last 5 years CHF index admission CHF within last 5 years Cardiomyopathy — index Atrial fibrillation — index Indications for PPM High-level block Bundle-branch block SA-node dysfunction Atrial fibrillation/flutter Other dysrhythmia⁎⁎ CSS/NCGS Ablation Cardiomyopathy HF None⁎⁎⁎
1301 (59)
1831 (59)
2068 (60)
5200 (59)
230 (10) 1217 (55) 752 (34) 1100 (50) 1099 (50)
253 (8) 1585 (51) 1270 (41) 1568 (50) 1540 (50)
320 (9) 1711 (50) 1419 (41) 1938 (56) 1512 (44)
803 (9) 4513 (52) 3441 (39) 4606 (53) 4151 (47)
536 (24) 995 (45) 450 (20) 218 (10) 37 (2) 510 (23) 374 (17) 930 (42) 89 (4) 647 (29)
880 (28) 1364 (44) 585 (19) 279 (9) 61 (2) 667 (22) 319 (10) 1131 (36) 87 (3) 975 (31)
945 (27) 1397 (40) 728 (21) 380 (11) 69 (2) 777 (22) 424 (12) 1355 (39) 158 (5) 1016 (29)
2361 (27) 3756 (43) 1763 (20) 877 (10) 167 (2) 1954 (22) 1117 (13) 3416 (39) 334 (4) 2638 (30)
934 (42) 226 (10) 365 (17) 272 (12) 221 (10) 38 (2) 89 (4) 22 (1) 6 (b1) 26 (1)
1177 (38) 258 (8) 854 (28) 409 (13) 167 (5) 136 (4) 55 (2) 17 (b1) 18 (b1) 17 (b1)
1217 (35) 350 (10) 804 (24) 464 (13) 257 (7) 123 (4) 66 (2) 62 (2) 78 (2) 29 (b1)
3328 (38) 834 (10) 2023 (23) 1145 (13) 645 (7) 297 (3) 210 (2) 101 (1) 102 (1) 72 (b1)
644 (29) 1445 (66) 0 110 (5) 211 (10)
822 (26) 2257 (73) 21 (b1) 8 (b1) 287 (9)
693 (20) 2067 (60) 296 (9) 394 (11) 261 (8)
2159 (25) 5769 (66) 317 (4) 512 (6) 759 (9)
Pacemaker type Single Dual Triple Not specified All-cause mortality 1 year
AMI = acute myocardial infarction, HF = heart failure, VF = ventricular fibrillation, SA = sino-atrial, AF = atrial fibrillation, CSS/NCGS = carotid sinus syncope/neurocardiogenic syncope. ⁎ Charlson Comorbidity Index score excluding AMI and HF. ⁎⁎ Includes paroxysmal tachycardia, ventricular fibrillation and flutter, sinus bradycardia and other specified and unspecified rhythm disorders. ⁎⁎⁎ No major indication coded — includes ‘cardiac arrest’ or other ‘cardiac complications’, myocarditis, pericarditis, hypertensive heart disease and ‘other-ill-defined heart disease’.
Women were older than men (mean 76.3, SD 10.8 vs 74.3, SD11.1 years, p b 0.001) and more likely to present with sick sinus syndrome (27% vs 20%, b0.001). There was little difference in AMI or acute coronary syndrome as a principal diagnosis in the index admission, but women were less likely to have ischaemic heart disease (IHD) recorded as a new or existing condition (33% vs 38%, χ2 = 25.4, p = b 0.001). Emergency admission (≈90% having a principal diagnosis of a conduction disorder or dysrhythmia) was more likely for patients with a higher Charlson score, being 40% of those with no cardiac comorbidity rising to 59% among those with the highest Charlson score (χ2trend = 134.9, df = 3, p = b0.001). Single chamber PPMs were more frequently implanted in patients aged ≥80 years (35% vs b20% in younger patients), with dual-chamber pacing the most frequent in each age-group (58% in ≥80 years, N70% in younger patients).
3.2. Mortality There were 759 deaths (9%) within one year of hospital discharge. Crude all-cause 1-year mortality decreased across the study (χ2 trend = 8.9, 2p = 0.01) (Table 1). Mortality was highest among those aged 80–99 years (13.5%) being more than double those aged 60–79 (6%) and 30–59 years (3.5%, χ2 trend = 173.2, 2p = b0.001). Crude mortality
at one year was greater for men (10%) than women (6.7%, χ2 = 27.9, 2p = b0.001). Table 2 shows the estimated hazard ratios from the multivariable Cox models for one-year mortality and for survival to 5-years from hospital discharge. Age and male sex were significant predictors of both 1-year mortality and poorer survival to five years, as were emergency admission, higher non-cardiac Charlson score, HF, cardiomyopathy, and AF. Coronary artery bypass graft surgery in the index admission was associated with better survival to five years, but not death within the first year. Era of implantation, and clinical factors including the major indication for pacing, initial pacing mode, admission for AMI or acute coronary syndrome, were not independent predictors of mortality after adjustment for age. 3.3. Comparison of observed and expected years of remaining life Among the 8757 PPM patients discharged from hospital, a total of 3804 (43%) died before attaining the date of their calculated LE, 1366 patients (16%) met or exceeded their LE by the end of 2011; the remaining 3587 (41%) patients were alive and had an expectation of life extending beyond the study end date. Among the 3478 patients (40%) whose LE date was before the December 31st 2011, 1366 (39%) met or exceeded their LE. There were no patients younger than 60 years in this group; the proportion of those who met their LE was 37% for those aged 60–79 and 41% for those 80–99 years. Almost half (48%) of all patients with a zero comorbidity score met or exceeded their LE, while this proportion fell to 28% for those with the highest Charlson scores. The ratio of total observed to total expected years of remaining life (to a maximum of 16 years) was 0.80 for men and 0.84 for women overall. The ratio declined with increasing Charlson score, with the poorest comparative survival being among patients with the highest score and HF or cardiomyopathy recorded in the index admission (Table 3). The mean EYLL was higher among men (1.41 years, 95% CI 1.33– 1.50) than women (1.19 years, 95% CI 1.09–1.30 p = 0.001) and increased with increasing non-cardiac Charlson score, with the greatest EYLL being among those patients with the highest non-cardiac comorbidity score and HF or cardiomyopathy recorded in the index admission
Table 2 Estimated hazard ratios from multivariable Cox models showing associations between demographic and clinical factors and risk for all-cause mortality at 1 year and survival to 5 years among 8757 patients after hospital discharge for initial PPM implantation 1995–2008. Factor
One year HR (95% CI) p-value
Men 1.00 Women 0.62 (0.53–0.72) b 0.001 Age group — years 30–59 1.00 60–79 1.69 (1.14–2.50) b 0.01 80–99 3.94 (2.65–5.80) b 0.001 Elective 1.00 Emergency 1.61 (1.38–1.87) b 0.001 Charlson score (non-cardiac morbidity) Zero 1.00 1–2 1.33 (1.06–1.66) 0.01 3–4 1.81 (1.42–2.30) b 0.001 ≥5 2.52 (1.95–3.25) b 0.001 Heart failure⁎ 2.07 (1.74–2.45) b 0.001 Cardiomyopathy⁎ 1.65 (1.23–2.21) 0.001 Atrial fibrillation⁎ 1.24 (1.07–1.44) b 0.01 CABG⁎ NA
Five years HR (95% CI) p-value 1.00 0.73 (0.67–0.77) b0.001 1.00 2.21 (1.84–2.64) b0.001 4.73 (3.95–5.66) b0.001 1.00 1.21 (1.13–1.28) b0.001 1.00 1.10 (1.02–1.18) 0.02 1.27 (1.16–1.39) b 0.001 1.62 (1.46–1.80) b 0.001 1.66 (1.53–1.80) b 0.001 1.23 (1.06–1.44) b 0.01 1.12 (1.05–1.19) b 0.001 0.68 (0.54–0.88) b 0.01
Age and any variable that was significantly associated with all-cause mortality after adjustment for age is included in the model. Other variables in the model included era, indication for pacing, initial pacing mode and admission for AMI or acute coronary syndrome, CABG = coronary artery bypass surgery in index admission. ⁎ Condition/procedure recorded in index admission.
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46 Table 3 Comparative survival among 8757 patients after initial PPM implantation 1995–2008 versus the general population: ratio of observed/expected years of remaining life up to 31st December 2011, by Charlson Comorbidity Index score. Ratio of observed to expected years of additional life Charlson score⁎
Zero
1–2
3–4
≥5
Men Women Age group (years) 30–59 60–79 80–99 Elective Emergency No heart failure Heart failure⁎⁎
0.88 0.92
0.82 0.85
0.73 0.76
0.64 0.71
0.95 0.87 0.93 0.91 0.88 0.90 – 0.90 – 0.91 0.86
0.90 0.80 0.86 0.86 0.79 0.85 0.68 0.83 0.74 0.84 0.81
0.86 0.72 0.74 0.80 0.70 0.79 0.59 0.75 0.67 0.77 0.68
0.70 0.64 0.70 0.75 0.61 0.73 0.49 0.69 0.40 0.69 0.63
No cardiomyopathy Cardiomyopathy⁎⁎ No Atrial fibrillation Atrial fibrillation⁎⁎
⁎ Charlson Comorbidity Index score excluding cardiac comorbidity. ⁎⁎ Condition recorded in index admission.
(Table 4). All these patients suffered additional comorbid conditions including diabetes, renal, pulmonary and peripheral vascular diseases. 3.4. Cause of death Cause of death was coded for 4163 (90%) of 4600 deaths. ‘Cardiac causes’ consisting of IHD and ‘other cardiac deaths’ were the most frequently recorded, being 43% for both men and women and ≥ 40% of deaths in each age group, with IHD being more than 50% of those deaths, except among those aged b 44 years. IHD as cause of death exceeded the proportion in the Australian general population; being double that for the general population for those aged 45–54 years (30% vs 14%) and N1.5 for those aged 55–75 years. Deaths from non-cardiac causes were lower than expected when compared with national statistics. This included cancer, the peak among PPM patients being 25% in those aged 65–74 years at death (vs 41% in that age-group in the general population). Cerebrovascular disease, ‘dementia’ and Alzheimer's disease increased with age among PPM patients but were lower than national statistics. 4. Discussion Patients without significant cardiac or non-cardiac comorbidity achieved around 90% of the total expected additional years of life to
45
the study end. This was the case for around one quarter of the cohort, for the majority the ratio worsened with increasing comorbidity, and with additional cardiac comorbidity. Among those patients whose total expectation of life fell wholly within the study period, survival to LE or beyond was almost 50% (e.g. equivalent to the general population), in the absence of comorbidities. This finding is supported by similar results from the Netherlands, where survival to a minimum of 3.3 years for patients who did not have CVD was estimated to be equivalent to age- and sex-matched controls, (patients with serious conditions such as cancer, and immune and degenerative diseases at implantation having been excluded from the analysis) [3]. The ratio of observed to expected additional years of life was lower for those aged 60–79 years than for those aged 80–99 years in our study, regardless of comorbidity score. Similarly favourable outcomes were reported for the eighty- and ninety-year olds in the Dutch study, with survival for elderly patients, including those with CVD, equivalent to population controls [3]. The relatively good survival among elderly patients in our study may represent greater selection for PPM among the oldest patients, who are already ‘healthy survivors’. Despite possible selection, it is evident that age was no barrier to PPM use, with those aged eighty years or more being 40% of all adult recipients, a proportion found in other developed countries such as France, Italy, Denmark, Sweden, the United Kingdom and Israel in 2005 [11]. At the time of PPM implantation 52% of men and 40% of women in the study cohort had exceeded the life expectancy at birth for men (75.2 years) and women (81.0 years) in WA born at the start of the study (1996) [17]. The outcome was more favourable for women compared to men, as observed in other studies [1–3], and this benefit may be associated with the lower prevalence of IHD among women, and, being older, they may also have been a more select population. Both cardiac and non-cardiac comorbidities were associated with the loss of expected years of life; the EYLL being greatest among those with the highest non-cardiac Charlson score and cardiomyopathy (almost 5 years) or HF (3.5 years). The significant contribution of increasing non-cardiac comorbidity to reduced LE is evident from this study. Among the few PPM studies to employ a measure of comorbidity, the large Healthcare Cost and Utilization Project-National Inpatient Sample from the United States found the Charlson score a stronger predictor than age of in-hospital mortality, complications, length of stay and costs [18]. Emergency admission has been previously identified as a risk for inhospital death; in the same 2004–2008 National Inpatient Samplebased study non-elective admissions incurred three times the risk for
Table 4 Mean expected years of remaining life lost to 31st December 2011 among patients after initial PPM implantation 1995–2009, by Charlson Comorbidity Index score. Mean expected years of life lost (95% CI) Charlson score⁎
Zero
1–2
3–4
≥5
Men Women Age group 30–59 60–79 80–99 Elective Emergency No HF HF⁎⁎
0.83 (0.69–0.98) 0.58 (0.40–0.75)
1.32 (1.19–1.46) 1.14 (0.99–1.29)
1.87 (1.67–2.07) 1.73 (1.49–1.98)
2.38 (2.07–2.69) 2.13 (1.75–2.51)
0.45 (0.20–0.70) 1.03 (0.87–1.19) 0.38 (0.20–0.56) 0.65 (0.51–0.79) 0.84 (0.66–1.02) 0.73 (0.60–0.86) – 0.73 (0.60–0.86) – 0.61 (0.48–0.74) 1.04 (0.81–1.27)
0.90 (0.62–1.19) 1.68 (1.53–1.82) 0.76 (0.62–0.91) 1.03 (0.90–1.16) 1.50 (1.34–1.66) 1.09 (0.98–1.19) 2.28 (1.97–2.58) 1.21 (1.11–1.31) 2.02 (1.48–2.57) 1.18 (1.06–1.30) 1.40 (1.21–1.58)
1.26 (0.76–1.76) 2.24 (2.01–2.47) 1.37 (1.16–1.58) 1.41 (1.21–1.61) 2.16 (1.93–2.39) 1.49 (1.33–1.66) 2.95 (2.57–3.33) 1.77 (1.61–1.93) 2.52 (1.76–3.28) 1.60 (1.42–1.78) 2.29 (2.00–2.58)
2.58 (1.50–3.66) 2.90 (2.52–3.28) 1.56 (1.27–1.84) 1.81 (1.44–2.19) 2.62 (2.31–2.93) 1.87 (1.61–2.13) 3.54 (3.01–4.08) 2.13 (1.90–2.37) 4.90 (3.49–6.30) 2.13 (1.82–2.43) 2.57 (2.18–2.97)
No cardiomyopathy Cardiomyopathy⁎⁎ No AF AF⁎⁎
AF = atrial fibrillation, HF = heart failure. ⁎ Charlson Comorbidity Index score excluding cardiac comorbidity. ⁎⁎ Medical condition recorded in index admission.
46
P.J. Bradshaw et al. / International Journal of Cardiology 190 (2015) 42–46
in-hospital mortality among those aged 70 years or more [18]. In our cohort this disadvantage, independent of age and comorbidity, persisted at one year and to 5 years. The lack of a significant association between survival and pacemaker type or index arrhythmia has been noted elsewhere for the 1990s, unlike previous decades [1], and among those aged 80 years or more [4]. Compared with the few studies to report cause-specific death, the proportion of cardiac deaths in the WA population (43%) was somewhat higher than the 33% reported after a mean of 5.8 years in a population study from The Netherlands [3], and at a median follow-up of 33 months in the Mode Selection Trial (MOST) where 35% of deaths were attributed to cardiac causes, 49% to non-cardiac causes, and 16% unknown [19]. COD coding may account for some of the difference, as ‘cardiac deaths’ in the Dutch study were restricted to AMI, HF, PPM-related, rhythm disturbances and ‘resuscitation for unknown causes’. Differences in population demographics, the health care environments, and length of follow-up can also influence the reporting of cause of death. In this population-based cohort it seems likely that those selected for treatment by implantation of a PPM, especially among the very old, are those who have a relatively good likelihood of survival. This would account for the lower proportion of age-stratified deaths from cancer, stroke and dementia than observed at the national level, albeit on low numbers of patients. Although medical interventions become more complicated and expensive in patients with limited LE due to advanced age [8,18], the oldest patients, in particular, may benefit from improvements in mobility, cognitive function, and activities of daily living [20,21]. Increasing LE and longer follow-up, even for older patients, has been associated with relatively low costs [22], making advanced age alone no argument for not implanting a PPM if indicated. 5. Strengths and limitations This strength of this study is access to 30 years of routinely collected hospital admission and separation data, providing a ‘real world’, population-based cohort of incident PPM implantations. The WA Linked Data are a highly accurate compilation of person-based records from both public and private hospitals within WA. The validity of the matching for the WA hospital morbidity data has been assessed, with invalid links (false positives) and missed links (false negatives) estimated at 0.11% each [23]. A validation study of the linked data against PPM procedures recorded in the catheter laboratory register at a single tertiary centre found less than 2% discordance [24]. No adjustment to EYLL for disability was required, as implantation of a pacemaker is generally associated with improved quality of life and health values [20,21]. We were not, however, able to describe the LE outcomes across all age-groups within the 16 years of the study period. For most of patients younger than 60 years at implantation the expectation of life was years, or decades, in the future. The use of administrative data does not permit the precision in determining the indication for pacing of clinical data. Fortunately, indication was not significantly associated with the outcomes. Similarly, other than the principal diagnosis, it is not always possible to distinguish between acute and chronic comorbid medical conditions (such as HF) recorded during the index admission. 6. Conclusions Life expectancy among PPM recipients without significant comorbidity approaches that of the general population. Greater non-cardiac comorbidity, HF, AF and, in particular, cardiomyopathy, contribute most to the loss of expected years of life in all age groups. The oldest patients are probably a relatively healthy and select population and older women, in particular, have several years of life remaining.
Disclosures None of the authors has any conflicts of interest. Dr Paul Stobie has participated in Advisory Board meetings for Medtronic, Inc. and St Jude Medical Inc.
Acknowledgements The authors thank the data custodians, data linkage and client services staff of Department of Health WA for assistance with gaining approvals and the linkage, extraction and checking of the linked data.
References [1] M. Brunner, M. Olschewski, A. Geibel, C. Bode, M. Zehender, Long-term survival after pacemaker implantation. Prognostic importance of gender and baseline patient characteristics, Eur. Heart J. 25 (2004) 88–95. [2] J. Pyatt, J. Somauroo, M. Jackson, et al., Long-term survival after permanent pacemaker implantation: analysis of predictors for increased mortality, Europace 4 (2002) 113–119. [3] E. Udo, N. van Hemel, N. Zuithoff, P. Doevendans, K. Moons, Prognosis of bradycardia pacemaker recipient assessed at first implantation: a nationwide cohort study, Heart 99 (21) (Nov 2013) 1573–1578. [4] E. Udo, N. van Hemel, N. Zuithoff, et al., Long-term outcomes of cardiac pacing in octogenarians and nonagenarians, Europace 14 (2012) 502–508. [5] W-K. Shen, D. Hayes, S. Hammill, et al., Survival and functional independence after implantation of a permanent pacemaker in octogenarians and nonagenarians, Ann. Intern. Med. 125 (1996) 476–480. [6] T-F. Chao, C-J. Liu, T-C. Tuan, et al., Long-term prognosis of patients older than ninety years after permanent pacemaker implantation: does the procedure save the patients? Can. J. Cardiol. 30 (10) (Oct 2014) 1196–1201. [7] Australian Institute of Health and Welfare, Premature mortality from chronic disease, Bulletin no. 84. Cat. no. AUS 133, AIHW, Canberra, 2010. [8] A. Greenspon, J. Patel, E. Lau, et al., Trends in permanent pacemaker implantation in the United States from 1993 to 2009: increasing complexity of patients and procedures, J. Am. Coll. Cardiol. 60 (2012) 1540–1545. [9] T. Aragon, D. Lichtensztajn, R. Katcher, R. Reiter, M. Katz, Calculating expected years of life lost for assessing local ethnic disparities in causes of premature death, BMC Public Health 8 (2008) 116. [10] P. Bradshaw, P. Stobie, T. Briffa, M.S. Hobbs, Use and long-term outcomes of implantable cardioverter-defibrillators, 1990 to 2009, Am. Heart J. 165 (2013) 816–822. [11] H. Mond, M. Irwin, H. Ector, A. Proclemer, The world survey of cardiac pacing and cardioverter-defibrillators: calendar year 2005, PACE 31 (2008) 1202–1212. [12] H. Quan, V. Sundararajan, P. Halfon, et al., Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data, Med. Care 43 (2005) 1130–1139. [13] Australian Bureau of Statistics, Cat no. 3312.5 — Deaths, Western Australia, http:// www.abs.gov.au/ausstats/[email protected]/mf/3101.01993. [14] Australian Bureau of Statistics, Cat no. 3311.5 — Demography, Western Australia, http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3311.520012001. [15] Australian Bureau of Statistics, Cat no. 3302.5.55.001 — Life Tables, Western Australia, 2003–2005 and 2007–2009, http://abs.gov.au/AUSSTATS/[email protected]/ DetailsPage/3302.5.55.0012007-2009?OpenDocument. [16] Australian Bureau of Statistics, Causes of Death, Australia, 2000. Canberra. ABS Cat no: 3303.0, http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3303.02000? OpenDocument. [17] J.M. Katzenellenbogen, P. Somerford, S. Serafino, Western Australian Burden of Disease Study: Mortality 2000, Department of Health, Perth, Western Australia, July 2003. [18] A. Mandawat, J. Curtis, A. Mandawat, V. Njike, R. Lampert, Safety of pacemaker implantation in nonagenarians, Circulation 127 (2013) 1453–1465. [19] G. Flaker, A. Greenspon, B. Tardiff, E. Schron, L. Goldman, A. Hellkamp, K. Lee, G. Lamas, Mode Selection Trial (MOST) Investigators, Death in patients with permanent pacemakers for sick sinus syndrome, Am. Heart J. 146 (2003) 887–893. [20] F. Lopez-Jimenez, L. Goldman, E.J. Orav, et al., Health values before and after pacemaker implantation, Am. Heart J. 144 (2002) 687–692. [21] E. Udo, N. van Hemel, N. Zuithoff, et al., Long term quality-of-life in patients with bradycardia pacemaker implantation, Int. J. Cardiol. 168 (2013) 2159–2163. [22] B. Schmidt, M. Brunner, M. Olschewski, et al., Pacemaker therapy in very elderly patients: long-term survival and prognostic parameters, Am. Heart J. 146 (2003) 908–913. [23] C. Holman, A. Bass, I. Rouse, M. Hobbs, Population-based linkage of health records in Western Australia: development of a health services research linked database, Aust. N. Z. J. Public Health 23 (1999) 453–459. [24] P.J. Bradshaw, P. Stobie, M.W. Knuiman, T.G. Briffa, M.S.T. Hobbs, Trends in the incidence and prevalence of cardiac pacemaker insertions in an ageing population, Open Heart 1 (1) (Dec 10 2014) e000177.
