Adiposity throughout adulthood and risk of sudden cardiac death in women

Abstract

Background

Sudden cardiac death (SCD) is often the first manifestation of coronary heart disease (CHD) among women. Data regarding BMI and risk of SCD are limited and conflicting.

Objectives

We examined the association of BMI repeatedly measured over 32 years and BMI during early and mid-adulthood with risk of SCD in the Nurses’ Health Study.

Methods

We prospectively followed 72,484 women free of chronic disease from 1980–2012. We ascertained adult height, current weight, and weight at age 18 at baseline and updated weight biennially. The primary endpoint was SCD (n=445).

Results

When updated biennially, higher BMI was associated with greater SCD risk after adjusting for confounders (p, linear trend: <0.001). Compared to a BMI of 21.0–22.9, the multivariate RR (95%CI) of SCD was 1.46 (1.05, 2.04) for BMI 25.0–29.9, 1.46 (1.00, 2.13) for BMI 30.0–34.9 and 2.18 (1.44, 3.28) for BMI ≥35.0. Among women with a BMI ≥35.0, SCD remained elevated even after adjustment for interim development of CHD and other mediators (RR: 1.72; 95%CI: 1.13, 2.60). In contrast, the association between BMI and fatal CHD risk was completely attenuated after adjustment for mediators. The magnitude of the association between BMI and SCD was greater when BMI was assessed at baseline or at age 18, at which time SCD risk remained significantly elevated at BMI≥30 after adjustment for mediators

Conclusions

Higher BMI was associated with greater risk of SCD, particularly when assessed earlier in adulthood. Strategies to maintain a healthy weight throughout adulthood may minimize SCD incidence.

Introduction

Sudden cardiac death (SCD) accounts for approximately 300,000 deaths in the U.S. annually(1). Clinical guidelines focus prevention efforts on medical therapies in high-risk patients, yet up to 75% of all SCDs occur in patients not classified as high-risk by current guidelines(2). Broader prevention strategies are crucial for reducing the burden of SCD in the general population where the majority of SCDs occur.

Obesity (BMI ≥30 kg/m²⁾ is associated with greater risk of CHD(3), a major risk factor for SCD(4). However, data regarding the association between BMI and SCD have been conflicting. Obesity has been associated with higher risk of SCD in some studies (5–8), but not in others.(9,10) Aging alters body composition (11) and age may modify the relation between BMI and risk of SCD. For example, BMI was linearly associated with risk of SCD in middle-aged persons (7,8). In contrast, BMI was associated with SCD in a U-shaped fashion among older men and women, and the nadir in risk occurred in the overweight range(6).

In this investigation, we quantified the association between BMI repeatedly measured over 32 years and risk of SCD among women in the Nurses’ Health Study (NHS). Additionally we compared the relation between BMI and risk of SCD to the relation of BMI and risk of nonfatal and fatal CHD outcomes. Finally, we quantified the association of BMI and weight gain in early adulthood with risk of SCD.

Methods

Study Population

The NHS began in 1976 when 121,700 female nurses in the US aged 30–55 responded to a mailed questionnaire about demographics, lifestyle and medical history(12). Follow-up questionnaires are administered biennially to update this information and obtain information about newly diagnosed diseases. Diet was assessed initially in 1980 with a semi-quantitative food frequency questionnaire (FFQ). Return of the baseline questionnaire implied informed consent and the institutional review board at Brigham and Women's Hospital approved the study protocol. The overall follow-up rate was 96% through 2013.

The baseline for this analysis was 1980 when information on weight at age 18 and important potential confounders (diet and physical activity) were first reported (N = 92,468 women). We excluded women with a history of CVD and cancer (N = 5076) or missing information on age (N = 46), current weight (N = 560), or diet (N = 439) at baseline. We excluded women who were underweight (BMI <18.5 kg/m2) during follow-up (N = 12,781) or had chronic obstructive pulmonary disease (N = 980), Parkinson’s disease (N = 6), or multiple sclerosis (N = 96) at baseline, to reduce potential reverse causation due to underlying illness. The final analysis included 72,484 women.

Exposure assessment

We calculated BMI as weight in kilograms divided by height in meters squared (kg/m²). Self-reported adult height and weight were ascertained on the 1976 questionnaire. Self-reported weight was highly correlated with directly measured weight in a previous validation study (r= 0.96)(13). Women reported weight at age 18 on the 1980 questionnaire (<1% missing). This Recalled weight was highly correlated with measured weight from physical examination records during that period (r=0.87)(14).

Outcome assessment

The primary study endpoint was SCD and specific details for the classification of SCD in this population have been published previously(5). Deaths were reported by next of kin or postal authorities or identified through a search of the National Death Index (NDI; 98% follow-up rate). For deaths occurring outside of the hospital or in the emergency room, where the death certificate or NDI search indicated possible CVD, we sought further information about the circumstances and timing surrounding the death from medical records or through interviews with the next of kin. We confirmed the endpoint of SCD through review of medical records, autopsy reports, and interviews with family members regarding the circumstances surrounding the death.

A cardiovascular death was considered sudden if the death or cardiac arrest occurred within 1 hour of symptom onset as documented by medical records or through reports from witnesses and next of kin. We excluded women with evidence of circulatory collapse (hypotension, exacerbation of congestive heart failure or neurologic dysfunction) before the disappearance of the pulse to increase the specificity for an “arrhythmic” death(14). SCDs were defined as probable if the death was unwitnessed or occurred during sleep where the participant was documented to be symptom-free when last observed within the preceding 24 hours without obvious extracardiac cause. We included both definite and probable cases in our analysis, as results were not altered when we excluded probable cases.

The secondary endpoints were fatal CHD and nonfatal myocardial infarction (MI). We confirmed fatal CHD events by hospital records or autopsy reports (International Classification of Disease (ICD), 8th and 9th Revision codes 410–412: ICD, 10th Revision codes I21–I22), or if CHD was listed as the underlying and most plausible cause of death on the death certificate and there was prior evidence of CHD. We also included probable fatal CHD events, which included deaths where medical records were unavailable, but CHD was the underlying cause of death on the death certificate or a search of the National Death Index or a family member provided supporting information.

Nonfatal MIs reported on biennial questionnaires were adjudicated by medical records, which were reviewed by study investigators blinded to the participants' risk factor status. MI was defined according to World Health Organization criteria, and when available, cardiac-specific troponin levels(15). MIs that required hospital admission and were verified by letter or telephone interview, but for which medical records or pathology reports were unavailable, were defined as probable cases and included in the analysis. Results were similar if we excluded probable cases.

Statistical analysis

We performed separate analyses for the SCD and CHD endpoints. For the analysis of SCD, women contributed person-time from the return of the baseline questionnaire in 1980 until the date of death, loss to follow-up or date of last available follow-up (December 2012). We used Cox proportional hazards models and modeled the most recently assessed BMI (updated biennially) as a time-varying covariate. In primary models, we adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, menopausal status, use of hormone therapy, aspirin, multivitamins, dietary factors related to SCD, and history of diabetes, hypercholesterolemia and hypertension at baseline. In secondary models, we also adjusted for potential mediators [incident self-reported clinician-diagnosed hypertension, high cholesterol, diabetes, and congestive heart failure (CHF), and CHD] as time-varying covariates. P-values for linear trend were computed by assigning the median value to each exposure variable, and modeling this as a continuous variable. We analyzed BMI using the following categories: 18.5–20.9, 21.0–22.9 (referent), 23.0–24.9, 25.0–29.9 (overweight), 30.0–34.9 (class I obese) and ≥35.0 (class II obese). We used the continuous measure of BMI to fit a restricted cubic spline model to explore a nonlinear relation with risk of SCD(16,17). We used 3 knots to divide continuous BMI into 4 categories(18).

We performed several pre-specified secondary analyses. To minimize potential reverse causation, we applied a 2-year lag between exposure and outcome assessment for other time periods. (19) In this analysis, we also excluded events within the first 2 years of the baseline assessment. For example, BMI in 1980 estimated SCD risk between 1982 and 1984 and so forth. Further, we stratified the population by age (<65 vs. ≥65 years) and by reported incident diagnosis of CHD as time-varying covariates. In the analysis stratified by prior CHD, we included women who reported a CHD event prior to the baseline questionnaire. We tested for interaction with a multiplicative interaction term between continuous BMI and the effect modifier (CHD or age) and compared the model with and without the interaction term using a likelihood ratio test. We attempted to restrict our analysis to never smokers; however few cases (n = 142) across 6 BMI categories led to wide confidence limits and precluded any meaningful interpretation.

Next, we explored the association between BMI updated biennially and risk of nonfatal MI and fatal CHD. Women contributed person-time from the return of the 1980 questionnaire until the date of death, loss to follow-up or end of follow-up (December 2012 for fatal CHD and May 2010 for nonfatal MI). We used time-varying, multivariable Cox proportional hazards models, adjusting for potential confounders. In secondary models, we also adjusted for potential mediators (hypertension, high cholesterol, diabetes, and CHF) as time-varying covariates.

Finally, we examined the relation between BMI at age 18, BMI at study baseline (1980), and weight gain during this timeframe and risk of SCD. In these analyses, we adjusted for covariates from the 1980 questionnaire. In the analysis of early adulthood weight gain, we additionally adjusted for BMI at age 18. We evaluated whether BMI measured at baseline provided information on SCD risk beyond current BMI. We conducted a likelihood ratio test that compared multivariable models of current BMI with and without BMI at baseline.

All statistical analysis was performed using SAS software, version 9 (SAS Institute Inc, Cary NC) and a P-value <0.05 was considered statistically significant.

Results

The prevalence of overweight and obesity increased during follow-up (Figure 1). Only 11% of women had a BMI ≥25 kg/m² at age 18. This proportion increased to 37% in 1980 (mean age 46; range 33–66) and 59% in 2010 (mean age 75; range 63–91). At baseline, women with higher BMI were less likely to smoke, use hormone therapy, exercise or consume alcohol (Table 1). Women with high BMI were older, more likely to use aspirin and more likely to have diabetes, hypertension, hypercholesterolemia, and a family history of CHD. Over 32 years, 445 cases of SCD, 1286 cases of total fatal CHD and 2272 non-fatal MIs occurred. Current BMI and risk of SCD

Figure 1.

Trends in BMI (kg/m²) over time in the Nurses' Health Study, at age 18 and from 1980–2008

Table 1.

Mean characteristics^* of the Nurses’ Health Study by category of BMI (kg/m²) in 1980

	BMI (kg/m2)
	18.5–20.9	21.0–22.9	23.0––24.9	25.0–29.9	30.0–34.9	35.0+
N	12,633	17,881	15,372	17,936	6063	2599
Age, years	44(7)	45(7)	46(7)	47(7)	47(7)	46(7)
BMI(kg/m2)	20.1(0.7)	22.0(0.6)	23.9(0.6)	27.1(1.4)	32.0(1.4)	38.9(3.7)
Current smokers, %	35	31	28	27	22	20
Exercise, hr/wk†	4.4(3)	4.2(3)	4.0(3)	3.7(3)	3.3(3)	2.8(3)
Current hormone therapy in postmenopausal women%	7	7	7	6	4	3
Aspirin use 7+ times/wk, %	12	13	14	15	18	23
Family history of MI, %	17	18	19	20	21	23
Hypertension, %	8	10	13	20	31	45
High Cholesterol, %	3	4	5	6	8	8
Diabetes, %	1	1	1	2	6	9
Alcohol, g/day	7.8(11)	7.4(11)	6.6(11)	5.5(10)	3.9(9)	2.8(8)
Saturated fat, % fat	40(5)	40(4)	40(4)	40(4)	40(4)	40(4)
n-3 PUFA, % fat	1.5(0.4)	1.5(0.4)	1.5(0.3)	1.5(0.4)	1.5(0.3)	1.5(0.4)
n-6 PUFA, % fat	12 (4)	12 (4)	11 (4)	11 (4)	11 (4)	12 (4)
Magnesium, mg	298 (71)	298 (70)	297 (69)	294 (70)	287 (70)	277 (71)

^*Values are means(SD) or percentages and are standardized to the age distribution of the study population;

^†Moderate to vigorous intensity ( activities >4 metabolic equivalents)

When updated biennially, higher BMI was associated with greater risk of SCD risk after adjusting for confounders (p, linear trend: <0.001) (Table 2, Figure 2). Compared to women with a BMI of 21.0–22.9, women with a BMI of 25.0–29.9, 30.0–34.9, and ≥35.0 had a significantly higher risk of SCD within the next two years, after controlling for confounders. The adjustment for potential mediating factors attenuated the magnitude of risk, but this risk remained significantly elevated for women with BMI ≥35.0.

Table 2.

Multivariable hazard ratio (95%CI) of SCD, fatal CHD and nonfatal MI by categories of BMI (kg/m²) updated every 2 years

	BMI (kg/m2)
	18.5–20.9	21.0–22.9	23.0––24.9	25.0–29.9	30.0–34.9	35.0+	P, linear trend
Sudden Cardiac Death
Cases	44	46	56	161	78	60
Frequency, %	10	16	20	33	14	7
Age-adjusted	1.59 (1.05, 2.41)	1.0 (ref)	1.01 (0.68, 1.50)	1.67 (1.20, 2.33)	1.98 (1.37, 2.86)	3.68 (2.49, 5.42)	<0.001
Multivariable model 1*	1.50 (0.99, 2.28)	1.0 (ref)	0.96 (0.65, 1.42)	1.46 (1.05, 2.04)	1.46 (1.00, 2.13)	2.18 (1.44, 3.28)	<0.001
Multivariable model 1 + mediating factors†	1.58 (1.04, 2.40)	1.0 (ref)	0.93 (0.63, 1.38)	1.30 (0.93, 1.82)	1.21 (0.83, 1.77)	1.72 (1.13, 2.60)	0.07
Total fatal CHD
Cases	119	153	180	407	268	159
Age-adjusted	1.26 (0.99, 1.60)	1.0 (ref)	0.96 (0.77, 1.19)	1.25 (1.04, 1.51)	2.10 (1.72, 2.57)	3.15 (2.51, 3.94)	<0.001
Multivariable model 1*	1.22 (0.96, 1.56)	1.0 (ref)	0.92 (0.74, 1.14)	1.02 (0.84, 1.23)	1.27 (1.03, 1.57)	1.39 (1.09, 1.76)	0.002
Multivariable model 1 + mediating factors†	1.23 (0.97, 1.57)	1.0 (ref)	0.86 (0.69, 1.07)	0.86 (0.71, 1.04)	0.98 (0.80, 1.21)	0.99 (0.78, 1.27)	0.83
Nonfatal MI
Cases	165	286	367	810	404	240
Age-adjusted	1.03 (0.85, 1.24)	1.0 (ref)	1.06 (0.91, 1.24)	1.35 (1.17, 1.54)	1.64 (1.41, 1.91)	2.21 (1.86, 2.63)	<0.001
Multivariable model 2‡	0.97 (0.80, 1.18)	1.0 (ref)	1.04 (0.89, 1.21)	1.25 (1.09, 1.43)	1.37 (1.17, 1.60)	1.67 (1.39, 2.00)	<0.001
Multivariable model 2 + mediating factors§	0.99 (0.82, 1.21)	1.0 (ref)	1.00 (0.86, 1.17)	1.14 (0.99, 1.30)	1.13 (0.96, 1.32)	1.26 (1.05, 1.52)	0.004

^*Model 1: adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, current use of hormone therapy, menopausal status, aspirin use, multivitamin, intake of saturated fat, n-3 polyunsaturated fat, n-6 polyunsaturated fat and magnesium (all updated throughout follow-up) and history of diabetes, hypercholesterolemia and hypertension at baseline (1980)

^†Potential mediating factors: incident high cholesterol, hypertension, diabetes, angina, congestive heart failure and nonfatal MI (all updated throughout follow-up)

^‡Model 2: adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, current use of hormone therapy, menopausal status, aspirin use, multivitamin, intake of saturated fat, n-3 polyunsaturated fat, n-6 polyunsaturated fat and magnesium (all updated throughout follow-up), and history of diabetes, hypercholesterolemia and hypertension at baseline (1980)

^§Potential mediating factors: incident high cholesterol, hypertension, diabetes, angina, congestive heart failure and angina (all updated throughout follow-up)

Figure 2.

Multivariable hazard ratio (95%CI) of SCD as a function of BMI (kg/m²) at age 18, at baseline and updated every 2 years. Models adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, current use of hormone therapy, menopausal status, aspirin use, multivitamin, intake of saturated fat, n-3 polyunsaturated fat, n-6 polyunsaturated fat and magnesium use (all updated throughout follow-up) and history of diabetes, hypercholesterolemia and hypertension at baseline. Data were fitted by a restricted cubic spline Cox proportional hazards model. The 95%CI are indicated by the dashed lines. The models were based on 445 cases. The models for BMI at age 18 were based on 339 cases and exclude women with BMI<18.5 and BMI >50.0.

Women with a low BMI (18.5–20.9) also had an elevated risk of SCD in the next 2 years, particularly after adjusting for potential mediating factors (Table 2). However, the J-shaped relation was not statistically significant (p, quadratic trend: 0.13). When we employed a 2-year lag, the risk of SCD in women with a BMI 18.5–20.9 was attenuated and not significantly elevated (RR for BMI 18.5–20.9: 1.41; 95%CI: 0.93, 2.14). In analyses stratified by a prior diagnosis of CHD, the elevated risk associated with obesity was significant only among women without a history of diagnosed CHD. The RR for SCD was 1.62 (95%CI: 1.05 – 2.51) for BMI 30.0–34.9 and 2.19 (95% CI, 1.35–3.55) for BMI ≥35.0 compared to BMI 21.0–22.9 (Supplemental table). Conversely, the elevated risk of SCD at low BMI (18.5–20.9) was significant only among women with clinician-diagnosed CHD (RR: 2.99; 95%CI: 1.43, 6.25; P, interaction, 0.004). When we stratified by age, the elevated risk at BMI ≥30.0 was greater in magnitude among younger (<65 years) compared to older (≥65 years) women. Notably, the interaction was not statistically significant (p, interaction, 0.09) (Supplemental table).

BMI and risk of CHD incidence and mortality

BMI was associated with risk of total fatal CHD in a J-shaped fashion (p, linear trend: 0.002, p, quadratic trend: 0.05) (Table 2). Unlike that observed for SCD, the association between BMI ≥30.0 and risk of total fatal CHD was completely attenuated and no longer significant after adjustment for mediating factors. As with SCD, women with a low BMI (18.5–21.0) had an elevated risk of fatal CHD that was not statistically significant. This association was not appreciably altered when we applied a 2-year lag period (data not shown). When we excluded SCDs from the total fatal CHD endpoint (n=178), the RRs were attenuated. Compared to women with a BMI 21.0–22.9, the multivariate RR (95%CI) were 1.17(0.90, 1.51), 0.87 (0.69, 1.10), 0.96 (0.78, 1.17), 1.21 (0.97, 1.52) and 1.30 (1.00, 1.68) for BMI categories 18.5–20.9, 23.0–24.9, 25.0–29.9, 30.0–34.9, ≥35.0, respectively.

The association between BMI and non-fatal MI was linear (p-linear trend<0.001; p, quadratic trend: 0.55) (Table 2). In multivariable models, women with BMI ≥25 had a significantly elevated risk of nonfatal MI and the risks were attenuated and no longer significant after adjustment for mediating factors, except among women with a BMI ≥35. Women who had low BMI (18.5–20.9) did not have a greater risk of nonfatal MI (RR: 0.88; 95%CI: 0.70, 1.12), unlike SCD and fatal CHD.

BMI earlier in adulthood and risk of SCD

Women with a higher BMI at study baseline had a greater risk of SCD (p, trend <0.001; p, quadratic trend: 0. 27) (Table 3, figure 2). Compared to women with a BMI 21.0–22.9, women with a BMI ≥25 had a significantly elevated risk of SCD, which remained significant after adjusting for potential mediating factors. The association between baseline BMI at baseline and risk of SCD was not explained by current BMI. When included in the same model, baseline BMI provided additional prognostic information beyond current BMI (p, likelihood ratio test =0.003).

Table 3.

Multivariable hazard ratio (95%CI) of SCD by categories of BMI (kg/m²) at study baseline (in 1980) and at age 18

	BMI (kg/m2)
	18.5–20.9	21.0–22.9	23.0––24.9	25.0–29.9	30.0–34.9	35.0+	P, linear trend
BMI (kg/m2) at baseline
Cases	39	70	68	151	75	42
Frequency, %	17	25	22	25	8	3
Age-adjusted	0.96 (0.65, 1.42)	1.0 (ref)	1.01 (0.72, 1.41)	1,83 (1.37, 2.43)	2.87 (2.07, 3.98)	4.23 (2.87, 6.21)	<0.001
Multivariable model*	0.94 (0.64, 1.40)	1.0 (ref)	0.99 (0.71, 1.38)	1.69 (1.26, 2.25)	2.42 (1.72, 3.41)	3.25 (2.16, 4.90)	<0.001
Multivariable model + mediating factors†	1.02 (0.68, 1.51)	1.0 (ref)	0.91 (0.65, 1.28)	1.43 (1.06, 1.91)	1.92 (1.36, 2.72)	2.44 (1.60, 3.71)	<0.001
BMI (kg/m2) at age 18
Cases‡	162	121	84	53	19	6
Frequency, %	46	29	14	9	2	<1
Age-adjusted	0.85 (0.67, 1.07)	1.0 (ref)	1.40 (1.06, 1.86)	1.48 (1.07, 2.04)	2.78 (1.71, 4.54)	5.96 (2.59, 13.7)	<0.001
Multivariable model*	0.88 (0.70, 1.12)	1.0 (ref)	1.33 (1.00, 1.76)	1.33 (0.96, 1.84)	2.09 (1.27, 3.43)	3.92 (1.67, 9.18)	<0.001
Multivariable model + mediating factors†	0.89 (0.70, 1.13)	1.0 (ref)	1.28 (0.96, 1.69)	1.25 (0.90, 1.72)	1.86 (1.13, 3.08)	3.77 (1.61, 8.86)	<0.001

^*Model adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, current use of hormone therapy, menopausal status, aspirin use, multivitamin, intake of n-3 and n-6 polyunsaturated fat, saturated fat and magnesium, and history of diabetes, hypercholesterolemia and hypertension (all at baseline)

^†Potential mediating factors: incident high cholesterol, hypertension, diabetes, angina, congestive heart failure and nonfatal MI (all updated throughout follow-up)

^‡3 cases of SCD occurred among women who were missing BMI at age 18

Elevated BMI at age 18 was also associated with risk of SCD (p, trend <0.001; p, quadratic trend: 0. 63) (Table 3, figure 2). Women with a BMI ≥30 at age 18 had a significantly elevated risk of SCD after adjusting for potential confounders and mediating factors. Further, the magnitude of association was larger when BMI was assessed at baseline or age 18 compared to BMI updated during follow-up (Figure 2). Additionally, weight gain in early-to-mid adulthood (age 18 to baseline; average 27 years) was associated with risk of SCD, independent of BMI at age 18 (p, linear trend<0.001; Table 4). This risk was significantly elevated in women who gained ≥10 kg after adjustment for potential confounders. After we adjusted for potentially mediating CVD risk factors, weight gain of ≥20 kg remained significantly associated with greater risk of SCD.

Table 4.

Multivariable relative risk of SCD according to weight change from age 18 to study baseline (mean age 45 years)

	Change in weight
	>-5 kg	stable	5.0–9.9 kg	10–19.9 kg	20 kg	P, trend
Cases	30	86	114	113	102
% frequency	8	28	30	22	12
Age and BMI-adjusted	0.80 (0.51, 1.24)	1.0 (ref)	1.17 (0.88, 1.55)	1. 50 (1.13, 1.99)	2.45 (1.84, 3.28)	<0.001
Multivariate model*	0.79 (0.50, 1.23)	1.0 (ref)	1.14 (0.86, 1.52)	1. 41 (1.06, 1.88)	2.08 (1.53, 2.83)	<0.001
Multivariate model + potential mediating factors†	0.81 (0.52, 1.27)	1.0 (ref)	1.05 (0.79, 1.40)	1.17 (0.87, 1.56)	1.61 (1.17, 2.20)	<0.001

^*Model adjusted for age, calendar time, smoking, physical activity, alcohol, total energy intake (kilocalories/day), family history of MI, current use of hormone therapy, menopausal status, aspirin use, multivitamin, intake of n-3 and n-6 polyunsaturated fat, saturated fat and magnesium, and history of diabetes, hypercholesterolemia and hypertension (all at baseline)

^†Potential mediating factors: incident high cholesterol, hypertension, diabetes, angina, congestive heart failure and nonfatal MI (all updated throughout follow-up)

Discussion

In this prospective study, women with higher BMI at three periods during adulthood had a greater risk of SCD. Compared to women with a BMI of 21.0–22.9, women with a BMI ≥25.0 had a 1.5–2-fold higher risk of SCD within the next two years after controlling for confounders. When we adjusted for potential mediators, this risk was attenuated but remained significantly elevated at a BMI ≥35.0. BMI was associated with risk of total fatal CHD risk, albeit weaker in magnitude compared with SCD. Further, this association was attenuated completely when we adjusted for potential mediators. The association between BMI and fatal events was J-shaped, with potential elevated risks at low BMI categories. In contrast, BMI was linearly associated with non-fatal MI risk. BMI earlier in adulthood was most strongly associated with SCD. Further, weight gain of ≥20 kg during early to mid-adulthood was associated with a 2-fold greater risk of SCD.

These findings suggest that the timing of BMI assessment plays a critical role in determining its relation to SCD risk and may contribute to the inconsistencies seen in other populations.(5–10) These results are highly consistent with patterns reported for BMI (20,21) and weight gain(22) in early compared with later adulthood and risk of all-cause mortality. Additionally, our results support the hypothesis that obesity may be a stronger risk factor for SCD in middle-aged versus older populations.(23) Notably, BMI measured at baseline was strongly associated with risk of SCD, even after accounting for current BMI. Therefore, excess weight or substantial weight gain may have an early and/or cumulative impact on SCD risk that is not completely negated by weight loss later in life. These findings highlight the necessity of maintaining a healthy weight throughout adulthood to minimize SCD risk.

When we updated BMI throughout follow-up, the risk of SCD was inconsistently elevated in lower BMI categories. Weight loss and low BMI later in life is often a harbinger of pre-clinical disease. The elevated risk of SCD among women with low BMI may be biased by underlying chronic disease and older age (24). Consistent with this hypothesis, women with a BMI in a healthy range of 18.5 – 20.9 had no elevated risk when BMI was assessed earlier in adulthood or among women without clinically diagnosed CHD prior to the SCD. In contrast higher BMI was associated with lower risk of SCD in women with clinically recognized CHD. This obesity paradox in patients with established CVD has been previously noted for other outcomes as well(25).

In previous reports from this population, BMI was positively and linearly associated with BMI and risk of total CHD(19). In the present study, the linear association was limited to non-fatal MI while the association for fatal CHD was J-shaped, similar to SCD. The differential association for non-fatal versus fatal CHD is consistent with results reported in another prospective study of women(3). The excess risk associated with current overweight and obesity for all three endpoints was attenuated when we controlled for intermediary CHD risk factors. However, the magnitude of risk for fatal and nonfatal CHD was attenuated to a greater degree, and the risk of fatal CHD was no longer significant after adjustment for these risk factors. Therefore, excess body weight likely influences both CHD and SCD risk through atherosclerotic pathways and adverse effects on blood pressure, lipids, and insulin resistance(26). Women with BMI ≥35 had a significantly elevated risk of SCD even after adjustment for these risk factors. Therefore, extreme obesity may increase SCD through non-atherosclerotic pathways as well, such as alterations in left ventricular hypertrophy(27,28), autonomic nervous system(28), and ventricular repolarization(29,30). Moreover, excess weight early in adulthood and a greater cumulative exposure to adiposity may lead early alterations in cardiac structure and function, which may serve as substrates for SCD later in life(27,28).

Strengths of this study include a well-defined study population with a large number of rigorously confirmed SCD cases. Further, the repeated measures and wide distribution of BMI allowed us to assess the shape of the association between BMI and SCD at various points in adulthood. Our study has several limitations. We used self-reported measures of anthropometry that have some degree of error. However, these values have been highly correlated with direct measures previously (r= 0.96)(13). Given the prospective nature of this study, such error is likely non-differential with respect to SCD and would underestimate the true association. Also, we lacked direct measures of clinical risk factors. Therefore, we may not have captured the complete mediation effects of these risk factors or removed the potential reverse causation due to underlying illness in the low BMI categories. Although we controlled for numerous clinical and lifestyle factors, there is still the potential for residual confounding. Further, the observational nature of this study prevents us from establishing a causal relation of obesity on SCD. Finally, the generalizability of our findings from a population of educated, primarily Caucasian women to other ethnicities and/or socio-demographic groups is limited. The prevalence of obesity and distribution of BMI is higher in Blacks and Hispanics compared to non-Hispanic Whites in the US(31). Prospective studies within multiethnic populations are needed to determine whether obesity contributes to the excess risk of SCD in Blacks(32) and individuals of low socioeconomic status(33). Future studies should also examine whether the biologic effects of adiposity differs between racial and ethnic groups.

Conclusions

In summary, this study provides new evidence that overweight and obesity throughout adulthood and weight gain in early adulthood are risk factors for SCD. BMI is mostly strongly related to risk when measured earlier in adulthood. Strategies for the maintenance of healthy weight throughout adulthood may provide substantial benefit for SCD prevention in the general population, particularly among women.

Perspectives.

Competency in Medical Knowledge

Overweight and obesity in early and later adulthood are strong risk factors for SCD. Strategies for the prevention of obesity in early adulthood and maintenance of healthy weight throughout adulthood may provide substantial benefit for SCD prevention in the general population, particularly among women.

Translational Outlook

Further research is needed to determine whether overweight and obesity are risk factors in multiethnic populations. Additionally, future research should evaluate whether weight loss during adulthood reduces SCD in the general population.

Abbreviations

BMI

body mass index

CHD

coronary heart disease

CHF

congestive heart failure

CI

confidence interval

CVD

cardiovascular disease

FFQ

food frequency questionnaire

MI

myocardial infarction

NHS

Nurses’ Health Study

RR

Relative risk

SCD

sudden cardiac death