Article Text
PDF
Abstract
Objective To compare two cohorts of middle-aged men from the general population born 30 years apart for incidence and predictors of heart failure (HF).
Methods Two population samples of men, born in 1913 (n=855) and in 1943 (n=797), were examined at 50 years of age and followed up for 21 years (1963–1994 and 1993–2014). Cox regression analysis was used to examine the impact of different factors on the risk of developing HF.
Results Eighty men born in 1913 (9.4%) and 42 men born in 1943 (5.3%) developed HF during follow-up; adjusted HRs comparing the two cohorts (born 1943 vs 1913) were: 0.46 (95% CI 0.28 to 0.74, p=0.002). In both cohorts, higher body mass index, higher diastolic blood pressure, treatment for hypertension, onset of either atrial fibrillation (AF), ischaemic heart disease and diabetes mellitus were associated with higher risk of HF. Higher heart rate was associated with an increased risk only in men born in 1913, whereas higher systolic blood pressure (SBP), smoking, higher glucose, higher cholesterol and physical inactivity were associated with an increased risk in men born in 1943. AF contributed higher risk of incident HF, whereas SBP and physical inactivity contributed lower risk in men born in 1943 compared with men born in 1913.
Conclusions Men born in 1943 had half the risk of HF after their 50s than those born 30 years earlier. AF, obesity, ischaemic heart disease, diabetes and hypertension remain important precursors of HF.
- heart failure
- cardiac risk factors and prevention
Statistics from Altmetric.com
Introduction
Several conditions have been identified as independent predictors of incident heart failure (HF) in different studies,1–5 with hypertension, diabetes mellitus (DM), ischaemic heart disease (IHD), higher heart rate and obesity being consistently associated with a higher risk of HF. The role of tobacco smoking in HF has been more controversial. Of all conditions, hypertension and IHD carry the highest attributable risk of HF incidence.2 5
Significant secular changes regarding cardiovascular risk factors have been reported in studies from Sweden,6 7 Finland8 and the USA.9In 50-year-old men in Gothenburg, Sweden, the mean weight increased by 11.2 kg, and the prevalence of obesity increased from 6% to 20%, while the proportion of smokers decreased from 56% to 12% between 1963 and 2003.7 Reported changes in blood pressure and prevalence of hypertension during this period are not as congruent as those for weight, obesity and smokers. Even so, mean systolic blood pressure (SBP) and diastolic blood pressure (DBP) decreased substantially from 1975 to 2015 in high-income western countries.10
Secular trends in the incidence of HF have been examined in several studies. In the Framingham Heart Study, the incidence remained unchanged for men and declined for women during the period 1950–2000.11 In the Olmsted County Study, the incidence of HF remained unchanged between 1979 and 2000.12 However, data from more recent studies suggest that incidence of HF may have decreased since the mid-1990s.13–15 In Sweden and Denmark, recent studies have documented a decline in the incidence of HF in older individuals, who form the vast majority of patients with HF, but an increase in younger (≤50 years) individuals.15 16
Simultaneously, cardiovascular disease (CVD) mortality has declined in high-income western countries,17 18 which is attributable to changes in risk factor pattern and has improved medical interventions for many CVDs.19 20 In particular, the profile of IHD underwent profound changes during the past few decades with a substantial decrease in mortality.21 22 For instance, Berg et al have shown that the mortality for IHD in Sweden has declined by 70% in adults aged 35–84 years between 1987 and 2009.22
Given the substantial changes in risk factors over time with a general decline in the main precursors for HF (ie, IHD and hypertension), although with an increase in mean body mass index (BMI) in the population, we hypothesised that changes in cardiovascular risk factors and management of CVDs would have resulted in a net decrease in the incidence of HF. Therefore, we set out to compare the cumulative 21-year incidence of HF between two randomly selected cohorts of middle-aged men living in Gothenburg, Sweden, born 30 years apart (1913 and 1943), and to which extent changes in risk factors and comorbidities might have contributed to this potential development.
This study aimed to examine whether and how the association between various characteristics and conditions (lifestyle factors, clinical characteristics and diseases) and the cumulative incidence of HF during a 21-year follow-up has changed during the past three decades. This was done by comparing two randomly selected cohorts of middle-aged men living in Gothenburg, Sweden, born 30 years apart.
Methods
Study design and setting
Two population samples of men born in 1913 and 1943 were first examined at age 50 years in 1963 and 1993, respectively, and followed up longitudinally over 21 years.23 24 In 1963, a sample was drawn from the population register consisting of all men born in 1913 (the 1913 cohort) on a day divisible by 3 (ie, the third, the sixth and ninth day of each month) and living in the city of Gothenburg. These criteria were fulfilled by 973 men, out of whom 855 (88%) agreed to participate. In 1993, the local tax authority in Gothenburg, Sweden, generated a random sample of 50% of all men who were born in 1943 (the 1943 cohort) and who were residents in the city of Gothenburg. These men were invited to participate in the study. Of 1463 men invited, 798 (55%) accepted participation and underwent a health examination at entry.
Health examinations
Data on smoking habits, leisure time physical activity, alcohol consumption, previous diseases and pharmacological treatment were collected using a series of questionnaires. The definitions used for these variables and blood examinations are described in the online supplementary.
Supplemental material
Both cohorts of men born in 1913 and in 1943 were followed up for 21 years from 50 to 71 years of age. The cohort of men born in 1913 were re‐examined at the ages of 54, 60 and 67 years25; the men born in 1943 were re-examined at ages 60 and 71 years.24
The definition of HF used in both cohorts was (1) hospitalisation with a diagnosis of HF, either as a principal or a secondary diagnosis, or (2) death with HF as the primary or contributing cause of death reported on the death registry (International Classification of Diseases, 8th Revision (ICD-8) code 427, International Classification of Diseases, 9th Revision (ICD-9) code 428 and International Classification of Diseases, 10th Revision (ICD-10) code I50).
Data on hospital admissions were obtained from follow-up data in the study and from the National Hospital Discharge Register and covered all hospital admissions in the country. Mortality data were obtained from the National Cause of Death Register and covered the deaths of all Swedish residents domestically or abroad. Mortality and hospital discharge diagnoses were classified according to the ICD. The ICD-8 was used until the end of 1986, the ICD-9 from 1987 to the end of 1996 and the ICD-10 from 1997 onwards.
The definitions of hypertension, DM, AF and IHD are described in the online supplementary.
Statistical analysis
For the baseline characteristics, n (%) is presented for categorical variables and mean with ±SD for continuous variables. For comparison between groups, Fisher’s exact test was used for dichotomous variables, Mantel-Haenszel χ2 test for ordered categorical variables and the Mann-Whitney U-test for continuous variables. Incidence rates were calculated as the number of events divided by the sum of follow‐up years per 1000 person‐years, with exact 95% Poisson confidence limits, and cumulative incidence curves were calculated using the Kaplan-Meier technique. The time-to-incident HF between the two cohorts was compared using Cox regression, first in an unadjusted regression analysis and then adjusting for factors considered to possibly influence HF. These factors included BMI, hypertension, SBP, smoking, cholesterol, physical activity, alcohol, diabetes, AF and IHD (all of these factors as baseline characteristics). Cox regression analysis was performed using both baseline characteristics and time-dependent variables (DM, AF and IHD) as possible predictors and HF as the outcome with HRs as effect measurement. The impact of continuous predictors on outcome was examined both as linear and piecewise linear functions with a median break point; the optimal among those two models was selected using Akaike’s information criterion. HRs per one unit increase in SD for baseline characteristics and time-dependent variables were estimated to rank the importance of risk factors in each cohort. Multivariable models were then built in a stepwise procedure using the results from the univariable analyses (parameters with p<0.05). The proportional hazards assumption was tested by including an interaction term with time in the models. The effect of time-updated variables of AF, IHD and DM on incidence of HF was described in step functions corresponding to Kaplan-Meier estimates when applying left-censoring of the start point for the onset of the studied explanatory event. The CI around the step function used Greenwood’s formula for the SE calculation. The impact of different variables on HF between the two cohorts was studied by including an interaction term between the variable of interest and the cohort.
All tests were two-tailed and the level of significance was set at p<0.05. Statistical analyses were performed with SAS statistical software V.9.4.
Patient and public involvement
The participants were not involved in the design, recruitment and conduct of the study.
Results
One participant with a history of HF from the 1943 cohort was excluded from the analysis, leaving a final sample of 797 patients from this cohort in the analysis. None of the participants from the 1913 cohort had a history of HF at baseline, and thus, all 855 were included in the analysis.
The baseline data in the two cohorts, categorised in groups with or without HF, are presented in table 1. At baseline, and in men with and without subsequent HF, men born in 1913 had lower BMI, higher SBP and DBP and a higher prevalence of smoking compared with men born in 1943. During follow-up, 80 men born in 1913 (9.4%, event rate 4.95 (95% CI 3.98 to 6.16) per 1000 person-years) and 42 men born in 1943 (5.3%, event rate 2.67 (95% CI 1.97 to 3.62) per 1000 person-years) developed HF. Of those men, 40 (50.0%) in the 1913 cohort and 17 (40.5%) in the 1943 cohort died during follow-up. After adjustment for baseline characteristics (BMI, hypertension, SBP, smoking, serum cholesterol, physical activity, alcohol, DM, AF and IHD), men born in 1943 had a 54% lower risk of HF (HR 0.46, 95% CI 0.28 to 0.74, p=0.0016) (figure 1).
Cumulative incidence curves for HF and the unadjusted and adjusted risks (HR) of developing HF for men born in 1943 compared with those born in 1913. HF, heart failure.
Baseline data for 50-year-old men born in 1913 and 1943 classified in groups with or without HF
Baseline characteristics associated with higher HF risk in both cohorts were higher BMI, higher DBP and treatment of hypertension (figure 2). Higher heart rate (>66 beats/min) was associated with an increased risk of HF but only in men born in 1913, whereas higher SBP (>130 mm Hg), smoking, higher glucose levels, higher total cholesterol levels and physical inactivity were associated with higher risk of HF in men born in 1943. Onset of AF, IHD or DM was associated with a higher risk of HF in both cohorts. Of 23 persons in the cohort of men born in 1913 diagnosed with AF during follow-up, 19 (83%) developed subsequent HF. In the cohort of men born in 1943, the corresponding proportion was 20% (15 of 75 persons). For IHD, the proportion of men developing HF in the 1913 cohort was 20% (19 of 95 persons) vs 13% (12 of 96) in the 1943 cohort. Corresponding percentages for DM were 14% (8 of 56) vs 7% (6 of 85 persons). The cumulative incidence of HF for participants with and without each of these three comorbidities is presented in figure 3.
Central illustration: Cox proportional hazard analysis for prediction of heart failure in 1913 and 1943 cohorts and the interaction effect of the predictors and cohorts. BMI, body mass index; DBP, diastolic blood pressure; SBP, systolic blood pressure.
(A) Cumulative incidence of HF in men in the presence or absence of AF. (B) Cumulative incidence of HF in participants in the presence or absence of DM. (C) Cumulative incidence of HF in participants in the presence or absence of IHD. AF, atrial fibrillation; DM, diabetes mellitus; HF, heart failure; IHD, ischaemic heart disease.
Factors with the highest HR per one unit increase in SD for the men born in 1913 were time-updated AF (HR 2.25, 95% CI 1.98 to 2.54), time-updated IHD (HR 1.92, 95% CI 1.61 to 2.29) and a high heart rate (>66 beats/min, HR 1.56, 95% CI 1.20 to 2.04) (table 2). For the men born in 1943, higher SBP (>130 mm Hg) had the highest HR (HR 2.09, 95% CI 1.39 to 3.13), followed by time-updated AF (HR 1.88, 95% CI 1.62 to 2.19), time-updated IHD (HR 1.83, 95% CI 1.48 to 2.26) and higher BMI (HR 1.80, 95% CI 1.42 to 2.30).
Ranking of risk factors for heart failure according to the gradient of risk (HR shown per one unit increase in SD for men born in 1913 and 1943)
Interaction analyses (figure 2) showed that, in comparison with men born in 1913, the relative importance of AF for incident HF has decreased in men born in 1943 (HR 29.17, 95 % CI 17.40 to 48.90 vs HR 13.98, 95% CI 7.43 to 26.32; interaction p value 0.03). In contrast, the relative importance of SBP (>130 mm Hg (HR 1.10, 95 % CI 0.97 to 1.24 vs HR 1.45, 95% CI 1.18 to 1.78; interaction p value 0.02) and of physical inactivity during leisure time (HR 1.02, 95% CI 0.64 to 1.64 vs HR 2.31, 95% CI 1.18 to 4.51; interaction p value 0.048) has increased.
Multivariable models using stepwise regression showed that for men born in 1913, variables independently associated with higher risk of HF were higher BMI (continuous variable), use of antihypertensive medication and a higher heart rate (piecewise linear) at baseline, as well as time-updated AF and IHD during follow-up (table 3). For the 1943 cohort, the independent predictors were time-updated AF, higher glucose levels, time-updated IHD, BMI≥30 kg/m2, higher cholesterol levels and higher SBP (piecewise linear).
Multivariable Cox regression analysis for men born in 1913 and 1943
Discussion
In the present study, we demonstrated that the cumulative 21-year incidence of HF in middle-aged men living in Gothenburg decreased significantly between the two cohorts born 30 years apart. There were also some changes in the risk factor profile for incident HF. The relative importance of AF for incident HF decreased considerably in men born in 1943, whereas SBP and physical inactivity during leisure time proved to be more important in this cohort. The comparison of two cohorts of the same age born 30 years apart and resident in the same area, and applying the same methodology to both cohorts, provides a unique opportunity to examine secular trends in incidence and risk factors for HF.
We found that the incidence of HF decreased from 9.4% in the oldest cohort (born 1913) to 5.4% in the more recent cohort (born 1943), consistent with other registry or hospital cohort studies reporting a decrease in the incidence of HF since the mid-1990s.13–15 This finding might be explained by continuous improvements in both modifications of risk factors and therapy for CVDs. As shown in our study, compared with men born in 1913, men born in 1943 not only had less cardiovascular risk factors (lower SBP, lower cholesterol levels and less smoking) but also had less CVDs (AF, IHD and DM), consequently resulting in less incident HF. Our findings are clinically relevant in that they suggest that the improved primary and secondary cardiovascular preventions have been beneficial for reducing the incidence of HF. Even so, the fact that the association between several conditions and the outcome of HF remains more or less unchanged over time can be considered as a sign of suboptimal prevention with potential for improvement, particularly with respect to increasing obesity rates.
The study also showed that AF has been and still remains a condition strongly associated with HF. What was observed in our study is that both the incidence of HF among individuals with AF and the importance of AF for HF have significantly decreased during the past few decades. A high incidence of HF in patients with AF prior to diagnosis of HF has been reported in previous studies. In the Framingham Heart Study, 22% of both men and women with AF developed HF during a follow-up of 10 years.26 The adverse haemodynamic consequences of AF, as well as the relatively similar risk factor profile may explain the observed association between AF and HF. The improved management of AF in recent years may explain the decreasing trend in HF reported in our study.
Studies have reported that the risk of developing HF after an myocardial infarction (MI) has decreased over the past decades.27 In our study, the impact of IHD for HF decreased numerically for men born in 1943 compared with men born in 1913, although there was no significant interaction between birth year and the effect of IHD. Improved survival of patients with ΙHD might potentially lead to increased life expectancy for these patients and hence increased time to develop HF. Despite the substantial improvement in management reported in the past decades, the association between IHD and HF remains prominent.
For hypertension, which is a well-known risk factor for HF, we examined the effect of SBP, DBP and the use of antihypertensive medication separately to distinguish the association of each separate hypertension parameter with HF. The use of antihypertensive medication and higher DBP were significant predictors in both cohorts without any significant changes over time. SBP over a median of 130 mm Hg was associated with higher risk of HF only in the 1943 cohort. In the gradient of risk analysis, SBP was indeed the parameter with the highest HR per one unit increase in SD, potentially suggesting that a target of SBP of below 130 mm Hg when treating hypertension could be beneficial for the prevention of HF.
We also found that the importance of physical inactivity during leisure time on the risk of HF has changed during the past decades. Unfortunately, data for work-related physical activity are not available and therefore these results should be interpreted with caution. The association of high physical activity level with a low incidence of HF has been demonstrated previously.2 A possible explanation for why physical inactivity during leisure time seems to be of greater importance for developing of HF is that men born in 1943 were probably generally less physically active during working hours than men born 30 years earlier.
Strengths and limitations
One major strength of the present study is that two representative population samples of the same age and from the same geographical area are investigated using the same methodology. Another strength is that data were linked to the Swedish national health data registries. A long follow-up with a minimal loss to follow-up further strengthens the study. Two limitations of the study are the relatively small sample size and the absence of women, which affects the generalisability of the results. Due to limited resources, the inclusion of women born in 1913 was not considered worthwhile at the time because of their anticipated low risk of MI, the main outcome of interest of the study. The fact that the participant rates were 88% in men born in 1913 and 55% in men born in 1943 might also have affected the results. Participants with high interest for their health status may have selectively participated in 1993. This may affect characteristics of participants and the association between characteristics and mortality. Another limitation is that echocardiography, which was not widely available until the 1980s, was not a part of the diagnostic algorithm, potentially leading to risk of misclassification.
Conclusions
This comparison of two population‐based cohorts born 30 years apart demonstrates that the cumulative 21-year incidence of HF in middle-aged men living in Gothenburg has decreased during the past few decades and that there has been consequential changes in the risk profile for incident HF. In comparison with men born in 1913, the relative importance of AF has decreased, whereas the impact of SBP and of physical inactivity has increased in men born in 1943. Obesity was equally important for the development of HF in both cohorts.
Key messages
What is already known on this subject?
A gradual improvement in treatment and prevention of the main precursors of heart failure, for example, ischaemic heart disease and hypertension, has occurred over the past decades, and heart failure hospitalisations may potentially be decreasing as a result of these developments. Therefore, a reasonable question is not only whether the risk of heart failure has decreased but also whether changes have occurred in associations between cardiovascular risk factors and diseases and the incidence of HF comparing two cohorts of men born 30 years apart.
What might this study add?
By examining with the same methodology two representative population samples of the same age and from the same geographical area our results suggest a significant decrease in the incidence of HF, as well as changes in the risk profile of the disease during the past three decades.
How might this impact on clinical practice?
In present study, the incidence of HF decreased from 9.4% in the oldest cohort (born 1913) to 5.4% in the more recent cohort (born 1943). The improvements in both modifications of risk factors and therapy for CVDs seem to be an important factor for the decreasing incidence. However, atrial fibrillation, obesity, ischaemic heart disease, diabetes and hypertension remain important risk factors, and the effective management of these conditions should be a priority for healthcare providers.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors CE, P-OH, KS, AR and MF contributed to the conception and the design of the work. CE, EOT, KS, KC collected the data. CE and AP analysed the data. CE drafted the article. P-OH, KS, AR, EOT, CK, AP and MF revised the article critically for important intellectual content. All authors gave the final approval and agreed to be accountable for all aspects of work, ensuring integrity and accuracy.
Funding This study was supported by grants from the Region Västra Götaland agreement concerning research and education of doctors (ALFGBG-508831, MF; ALFGBG-140341, 447561,726481,824851, KC; ALFGBG-717211, AR), the Health and Medical Care Committee of the Regional Executive Board, Region Västra Götaland (VGFOUREG- 564181) (MF), the Swedish Research Council (grant number 2013-5187 (SIMSAM), 2013-4236, K2012-65X-22036-01-3) and the Swedish Heart and Lung Foundation.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The study complied with the Declaration of Helsinki, except for the screening examination in 1963 where only oral informed consent was given. The study protocol was approved by the ethics committee of Gothenburg University (DNR 157‐93, 0067‐03 and DNR 649‐13).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.