Temporal Changes in the Association between Modifiable Risk Factors and Coronary Heart Disease Incidence

To the Editor

Diagnosis and control of coronary heart disease (CHD) risk factors have received particular emphasis in the guidelines issued since 1977 (blood pressure) and 1985 (lipids).^1,2 Yet on a population level, little is known about how these efforts have altered CHD incidence and its association with modifiable risk factors. This study explored: 1) how the associations between modifiable risk factors and CHD events changed between 1983–1995 and 1996–2011; and 2) during this timeframe, whether or not the population attributable fractions (PAFs) of CHD due to modifiable risk factors were altered.

Methods

Individual patient-level data from five observational cohort studies (Table) available in the National Heart, Lung, and Blood Institute’s Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC) were pooled.³ Two analytic datasets were created: one with baseline data collected between 1983 and 1990 (earlier era) and one with baseline data collected between 1996 and 2002 (later era). Participants aged 40–79 years free of cardiovascular disease were selected from each period and matched on age (within 2 years), race (black vs. non-black), and sex. Each cohort was followed for up to 12 years (final year 2011) for new onset CHD (i.e., myocardial infarction, coronary death, angina, coronary insufficiency) using outcomes available in BioLINCC. A piecewise constant hazards model adjusted for age, sex, and race was used to estimate the hazard ratios (HRs) of CHD due to systolic blood pressure (SBP), diabetes, smoking (current/within past year), and total-to-high-density lipoprotein cholesterol (total:HDL-C) ratio.⁴ Blood pressure and total cholesterol were adjusted for treatment using a nonparametric approach.^5,6 HRs were compared by testing risk factor-era interaction. Population-attributable fractions were computed and compared using 10-year survival probabilities from the model with confidence intervals (CIs) derived using the delta method and log transformation.⁷ Analysis was approved by the Duke University Institutional Review Board and conducted using SAS version 9.4. Statistical significance used two-sided α=0.05.

Table.

Characteristics of Participants from NHLBI BioLINCC in Earlier and Later Time Periods^a

	Overall	Earlier Era CHD Casesb	No CHDb	Overall	Later Era CHD Casesb	No CHDb
N	14009c	1428	12581	14009c	1317	12692
Age	60.4 (8.2)	64.4 (8.4)	60.0 (8.1)	60.6 (8.1)	65.0 (7.4)	60.1 (8.1)
Sex (female)	56.0	42.2	57.6	56.0	45.6	57.1
Black/AA	15.4	15.9	15.3	15.4	14.7	15.5
SBP (mmHg)	127 (20)	137(21)	126 (20)	125 (19)	134 (21)	125 (19)
SBP ≥140 mmHg	24.1	41.7	22.1	21.0d	35.9	19.5
On BP treatment	27.2	42.6	25.4	31.6d	47.3	30.0
Diabetese	10.4	22.4	9.0	11.0	18.5	10.2
Smokerf	21.8	25.4	21.4	14.3d	18.8	13.8
HDL-C (mg/dL)	53 (17)	49 (15)	54 (17)	52 (16)	49 (15)	52 (16)
HDL-C <50 mg/dL	47.0	58.4	45.7	51.0d	59.1	50.2
Total-C (mg/dL)	217 (40)	220 (42)	217 (40)	200 (36)	206 (39)	200 (36)
Total-C ≥200 mg/dL	65.4	68.3	65.1	48.9d	55.7	48.2
Total:HDL-C ratio	4.4 (1.5)	4.8 (1.5)	4.4 (1.5)	4.2 (1.3)	4.5 (1.4)	4.2 (1.3)
Total:HDL-C ratio ≥4.0	55.8	68.2	54.3	51.4d	61.7	50.3
On LL treatment	2.9	3.4	2.8	11.3d	13.7	11.1

Abbreviations: AA, African American; BioLINCC, Biologic Specimen and Data Repository Information Coordinating Center; CHD, coronary heart disease; HDL-C, high-density lipoprotein cholesterol; NHLBI, National Heart, Lung, and Blood Institute; on BP treatment, on blood pressure medication(s); on LL treatment, on lipid-lowering medication(s); SBP, systolic blood pressure; total-C, total cholesterol; total:HDL-C ratio, total-to-high-density lipoprotein cholesterol ratio

^aMean and standard deviation for continuous variables, % for categorical variables; NHLBI BioLINCC cohorts contributing data include:

Framingham Heart Study Original Cohort (FHS Original, examinations in 1983–1985 and 1997–1999): n=522 earlier, 48 later

Framingham Heart Study Offspring Cohort (FHS Offspring, examinations in 1983–1987 and 1998–2001): n=1935 earlier, 1808 later

Atherosclerosis Risk in Communities (ARIC, examinations in 1987–1989 and 1996–1998): n=8145 earlier, 6696 later

Cardiovascular Health Study (CHS, examinations in 1989–1990 and 1997–1998): n=3407 earlier, 1901 later

Multi-Ethnic Study of Atherosclerosis (MESA, examination in 2000–2002): n=3556 later

^bAll adults free of cardiovascular disease at baseline, these are characteristics of adults who do and do not develop CHD in follow-up.

^c4833 individuals in the earlier and 6080 in the later era were excluded due to no matches

^dComparison of prevalence in the overall group in the earlier and later eras, *p<0.001, otherwise p>0.05

^eDiabetes defined as fasting glucose at or above 126 mg/dL (random glucose above 200 mg/dL in FHS Original Cohort) or on diabetes treatment.

^fSmoker defined as current or within the past year.

Results

The Table shows characteristics of 14,009 pairs of participants in two cohorts. In the later period, the proportions of smoking, high SBP, and dyslipidemia were lower, while diabetes, SBP, and lipid-lowering treatment were higher. The incidence of new-onset CHD declined from 1.18 (95% CI 1.12–1.25; 1,428 events) to 0.98 (95% CI 0.93–1.04; 1,317 events) per 100 person-years (p<0.001).

For the outcome of CHD, the HRs of smoking, SBP, and total:HDL-C were not significantly different between eras (Figure), but diabetes HRs were lower in the later era (2.01, 95% CI 1.76–2.28 vs. 1.49, 95% CI 1.29–1.71; p=0.002). Numerical decreases in the PAFs of CHD due to smoking, systolic hypertension (SBP ≥140mmHg), and dyslipidemia (total:HDL-C ≥4.0) over time were not statistically significant (Figure); however, the PAF of diabetes declined significantly from 9.8% (95% CI 7.7–11.9) to 5.4% (95% CI 3.3–7.6, difference 4.4%, 95% CI 1.4–7.4; p=0.004).

Figure. PAF of CHD.

Displayed is the PAF of CHD for treatment-adjusted SBP ≥140, diabetes, smoking, and treatment-adjusted total:HDL-C ≥4.0 in the early and later eras.

Abbreviations: CHD, coronary heart disease; PAF, population attributable fraction; SBP, systolic blood pressure; total:HDL-C, total-to-high-density lipoprotein cholesterol ratio

Discussion

Examination of adults from five large observational cohort studies led to several findings. First, the incidence of CHD declined almost 20% over time. Second, while the prevalence of diabetes increased, the fraction of CHD attributable to diabetes decreased over time, due to attenuation of the association between diabetes and CHD; this may have resulted from changing definitions and awareness of diabetes, improvements in diabetes treatment and control, and/or better primary prevention. Third, there was no evidence that the strength of the association between smoking, SBP, or dyslipidemia and CHD changed between time periods, nor was there evidence that the proportion of CHD due to these factors changed; this underscores the importance of continued prevention efforts targeting these risk factors. Study limitations include the unknown effect of increasing treatment rates on follow-up, the variability of population-attributable fractions with classification thresholds, and one of the cohorts only contributing to the later period.