PDF
Citation
Print
Introduction
Background
Hypertension and type 2 diabetes (T2D) are well-known cardiovascular risk
factors. Additionally, obesity, smoking, alcohol consumption, and decreased
physical activity are established risk factors associated with hypertension or
T2D [1,2]. Previous studies have highlighted that lifestyle modifications
can significantly reduce mortality risks in patients with hypertension or T2D
[3,4]. However, once patients are diagnosed with hypertension or
diabetes, they often begin treatment with medication, which means that analyses
on the effect of lifestyle modifications are very limited. Previous research has
primarily focused on how one specific lifestyle change affects hypertension or
T2D. Moreover, studies examining whether lifestyle modifications have a similar
clinical effect on cardiovascular events or mortality in the prehypertension or
prediabetes stages are somewhat scarce. For example, a weight loss of
approximately 10 kg may reduce systolic blood pressure by between 5 and 20 mmHg
[5]. Alcohol consumption is directly
correlated with elevated blood pressure; thus, reducing alcohol intake is
associated with lower blood pressure [6].
However, regardless of weight loss, engaging in at least 150 minutes of physical
activity per week has been shown to reduce the incidence of T2D by 44% [7]. Furthermore, there are correlations
among certain lifestyles, suggesting that individuals who engage in intensive
exercise may be more inclined to quit smoking or drinking. Therefore, it is
necessary to dynamically analyze the changes in various types of lifestyle
modifications rather than focusing on a single type. Furthermore, lifestyle
habits such as smoking, drinking, and exercise are influenced by gender.
Typically, men are more likely to smoke and drink, leading to the hypothesis
that the impact of rigorous management of unhealthy lifestyles on the occurrence
of cardiovascular events varies between genders. The frequency and intensity of
exercise also differ by gender.
Objectives
Thus, the present study investigated (1) how dynamic changes in lifestyle can
affect cardiovascular events; (2) how various lifestyles change organically; (3)
whether gender-specific lifestyle changes can affect the occurrence of
cardiovascular events, specifically in relation to prehypertension or
prediabetes.
Methods
Ethics statement
The Institutional Review Board (IRB) at Ewha Womans Medical College Mokdong
Hospital (IRB no. EUMC-2021-11-029) and the NHIS Big Data Steering Department
(NHIS-2022-2-197) granted approval for this study. As the NHIS data used were
completely anonymous and handled in accordance with the Personal Data Protection
Act, obtaining written consent from subjects was not required.
Study design
This is a nationwide, population-based cohort study. It has been described in
accordance with the STrengthening the Reporting of OBservational studies in
Epidemiology (STROBE) statement, which is available at: https://www.strobe-statement.org/.
Setting
In December 2022, the authors conducted a search on prediabetes and
prehypertension within the National Health Insurance Service–Health
Screening Cohort (NHIS-HEALS) database. Prediabetes is defined by the presence
of impaired fasting glucose, impaired glucose tolerance, or a hemoglobin A1c
level of 5.7%–6.4%, while prehypertension is characterized by a systolic
pressure of 120 to 139 mmHg or a diastolic pressure of 80 to 89 mmHg at the
initial health screening. The NHIS-HEALS database was established from a cohort
of 514,866 Koreans, aged 40 to 79 in 2002, who were randomly selected to
represent 10% of the national health screening subjects from 2002 to 2003.
Data source and study cohort
The source of data for this study was the Korean NHIS-HEALS, which is a
nationwide population-based cohort. Detailed information on the cohort has
already been published elsewhere [8].
Additional details on the study cohort can be found in the Supplement 1.
Participants (study population)
For this study, we constructed a sub-cohort based on the research hypothesis.
Initially, we excluded subjects with inconsistent examination dates, resulting
in a cohort of 514,795 individuals who had undergone at least one health
screening between 2002 and 2003. Of these, 334,937 subjects who also
participated in a health screening during the second period (2004–2005)
were included. We designated the date of the second health screening as the
index date (Fig. 1C). Subsequently, we
excluded individuals with a history of specific conditions prior to the index
date, as defined by the International Classification of Diseases, Tenth Revision
(ICD-10): heart failure (ICD-10: I50, n=668); cerebrovascular accident (I60-I69,
n=13,187); cardiovascular disease (CVD; I20-I25, I71, I72, n=25,528); or cancer
(C00-C96, n=3,921). Additionally, we excluded individuals with missing
lifestyle-related data (n=13; n=297,010).
Fig. 1.
Flow diagram of selection of the study population from the National Health Insurance Service database. (A) Prehypertension subgroup; (B) prediabetes subgroup; (C) overall schema of the study. NHIS, National Health Insurance Service; HF, heart failure; CVD, cardiovascular disease; CVA, cerebrovascular accident; HTN, hypertension; DM, diabetes mellitus; BMI, body mass index.
Next, we defined cohorts for prediabetes and prehypertension. Detailed criteria
used to define these cohorts are available in the Supplement 1. The
prehypertension cohort included 95,152 subjects (60,084 men and 35,068 women),
while the prediabetes cohort comprised 57,865 subjects (37,836 men and 20,029
women). The flow diagram illustrating the selection of the study population is
displayed in Fig. 1A, B.
Outcome variables
The primary outcome of the study was major adverse cardiovascular events (MACE),
which were defined as death, non-fatal myocardial infarction (MI), or non-fatal
stroke resulting from CVD. Hospital admission data were used to identify
instances of CVD, including MI (ICD-10: I21), stroke (ICD-10: I60–I69),
and cardiovascular death (I20–I25, I71–I72, and I60–I69).
The secondary outcome of the study was all-cause mortality. The date of death
was obtained from the claims database, and this information was used to
determine all-cause mortality. The follow-up period began on the index date and
concluded either on the date of the first occurrence of the primary outcome or
on the last follow-up date (12/31/2015).
Measurement (assessment of lifestyle changes)
Regarding lifestyle factors, we considered body mass index (BMI), current smoking
status, alcohol intake categorized by weekly frequency (never; 2–3 times
a month; 1–2 times a week; 3–4 times a week; ≥5 times a
week), and physical activity also categorized by weekly frequency (never;
1–2 times a week; 3–4 times a week; 5–6 times a week; every
day). Changes in these lifestyle factors were assessed using health checkup data
from the first observation period (2002–2003) to the second
(2004–2005) as shown in Fig. 1C.
Detailed methods for evaluating changes in lifestyle factors are available in
the Supplement 1.
Bias
Since subjects were selected from the cohort database according to the inclusion
criteria and disease criteria, selection bias was not a concern.
Statistical analysis
Summary statistics for baseline characteristics were presented as means with SDs
for numerical data and as counts with percentages for categorical data. To
explore the impact of disease incidence and lifestyle behaviors according to
gender, we performed gender-stratified analyses. The incidence rate (per 10,000
person-years) of outcomes was estimated using a Poisson regression model. To
evaluate the influence of lifestyle changes on disease incidence, we calculated
hazard ratios (HRs) with 95% CIs using the Cox proportional hazards regression
model. Detailed descriptions of the covariates used for adjusting the HRs, as
well as the methodology for the sensitivity analysis, are available in the Supplement 1.
All statistical analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC,
USA). Statistical significance was set to P<0.05.
Results
Males constituted 65.2% of the prehypertension cohort (n=60,084). Additionally, women
were found to consume less alcohol (32.0% vs. 80.7% for never drinking) and to
exercise less frequently (46.4% vs. 65.0% for never exercising) compared to men.
Among the men, 44.1% were current smokers (n=25,201), whereas only 1.7% of women
smoked currently (n=577). Similarly, in the prediabetes cohort, men made up 65.4% of
the subjects (n=37,836). The percentage of women who neither exercised (47.2% vs.
67.7%) nor consumed alcohol (29.8% vs. 80.9%) was lower compared to men. In this
cohort, 41.8% of men (n=15,144) were current smokers, while only 2.2% of women
(n=432) were smokers. Furthermore, the prevalence of BMI over 25.0 kg/m2
was lower in women than in men in both cohorts (32.3% vs. 28.5% in the
prehypertension cohort and 38.1% vs. 36.7% in the prediabetes cohort). The general
characteristics of subjects according to group are summarized in Table 1.
Table 1.
Participants’ characteristics based on the index examination of National Health Insurance Service National Health Screening
Table 2 presents the incidence of diseases
during follow-up in subjects with prehypertension and prediabetes, categorized by
gender. In the prehypertension cohort, there were 3,979 MACE cases in men and 2,290
in women. The MACE incidence rates were similar between genders (62.6/10,000 PY for
men vs. 62.1/10,000 PY for women; P=0.724). However, a significant difference was
observed in all-cause mortality rates, with men showing a higher rate than women
(47.1/10,000 PY vs. 24.3/10,000 PY; P<0.001). In the prediabetes cohort,
3,435 MACE cases were recorded in men and 1,904 in women. The incidence rate of MACE
in women was slightly higher than in men, but this difference was not statistically
significant (88.0/10,000 PY for women vs. 92.8/10,000 PY for men; P=0.063). However,
there was a significant difference in all-cause mortality between men and women,
with men experiencing a higher rate (66.7/10,000 PY vs. 41.4/10,000 PY;
P<0.001).
Table 2.
Disease incidence rate between two biennial health screening periods
| Prehypertension group | Prediabetes group | |||||
|---|---|---|---|---|---|---|
| Outcome | Men | Women | P for gender difference | Men | Women | P for gender difference |
| Incidence rate (per 10,000 PY), [95% CI] | Incidence rate (per 10,000 PY), [95% CI] | Incidence rate (per 10,000 PY), [95% CI] | Incidence rate (per 10,000 PY), [95% CI] | |||
| MACE1) | 62.63 [60.71–64.61] | 62.05 [59.56–64.65] | 0.724 | 87.98 [85.09–90.97] | 92.78 [88.70–97.04] | 0.063 |
| Non-fatal MI | 18.52 [17.50–19.60] | 12.14 [11.08–13.30] | <0.001 | 23.13 [21.69–24.67] | 19.11 [17.34–21.06] | 0.001 |
| Non-fatal stroke | 44.21 [42.61–45.87] | 50.31 [48.07–52.64] | <0.001 | 64.04 [61.59–66.58] | 74.51 [70.88–78.32] | <0.001 |
| CVD death | 3.85 [3.41–4.35] | 3.20 [2.69–3.82] | 0.093 | 7.47 [6.68–8.34] | 6.55 [5.56–7.71] | 0.195 |
| All-cause death | 47.13 [45.49–48.82] | 24.27 [22.76–25.89] | <0.001 | 66.73 [64.26–69.29] | 41.38 [38.75–44.20] | <0.001 |
As previously mentioned, lifestyle factors were defined as BMI, current smoking,
drinking, and physical activity. In the prehypertension and prediabetes cohorts,
26.6% and 26.1% of subjects, respectively, experienced a worsening in one or more
lifestyle factors. Conversely, 34.2% and 34.8% of subjects in these cohorts improved
in one or more lifestyle factors. Overall, fewer than 10% of subjects experienced a
decline or improvement in two or more lifestyle factors (Supplement 2).
By controlling for lifestyle and clinical factors at the initial health screening, we
evaluated the effects of lifestyle changes on major outcomes using a multivariate
model (Tables 3, 4). In the prehypertension group, 7,376 patients (7.75%) had
already been diagnosed with T2D, and 2,601 (2.73%) were taking anti-hyperglycemic
medication. Similarly, in the prediabetes group, 11,577 patients (20.01%) had been
diagnosed with HTN, and 8,962 (15.49%) were on antihypertensive medication (Table 1). To isolate the effects of medication,
the multivariate analysis accounted for the impact of antihypertensive drugs,
anti-hyperglycemic drugs, and aspirin. In men with prehypertension, the risk of MACE
increased if their lifestyle worsened (HR, 1.13; 95% CI, 1.04–1.23, P=0.004),
particularly if they gained weight (HR, 1.15; 95% CI, 1.03–1.28, P=0.010) or
started smoking (HR, 1.34; 95% CI, 1.17–1.55, P<0.001). For women with
prehypertension, the risk of MACE was higher for those who started smoking (HR,
1.69; 95% CI, 1.15–2.49, P=0.008) or reduced their physical activity (HR,
1.25; 95% CI, 1.06–1.47, P=0.010). Conversely, in men with prehypertension,
improving lifestyle factors reduced the risk of MACE (HR, 0.91; 95% CI,
0.84–0.99, P=0.025), particularly through smoking cessation (HR, 0.79; 95%
CI, 0.70–0.89, P<0.001), drinking less (HR, 1.09; 95% CI,
1.00–1.20, P=0.048), or increasing physical activity (HR, 0.91; 95% CI,
0.84–0.99, P=0.027). In men with prediabetes, those whose lifestyle factors
worsened had a 23% higher risk of MACE compared to those with no lifestyle changes
(HR, 1.23; 95% CI, 1.12–1.35, P<0.001). An increased risk of MACE was
also observed in those who gained weight (HR, 1.19; 95% CI, 1.06–1.33,
P=0.003), started smoking (HR, 1.41; 95% CI, 1.22–1.64, P<0.001), or
decreased their physical activity (HR, 1.21; 95% CI, 1.09–1.35,
P<0.001). Additionally, in men with prediabetes, a reduction in alcohol
consumption was linked to a higher risk of MACE (HR, 1.17; 95% CI, 1.07–1.29,
P=0.001). In women with prediabetes, the risk of MACE was 1.24 times higher for
those who gained weight compared to those with no change in BMI levels (HR, 1.24;
95% CI, 1.06–1.45, P=0.006). As weight change can be a consequence of
lifestyle changes, the association between unhealthy lifestyles, excluding BMI, and
MACE was evaluated. Among pre-hypertensive men, those whose lifestyles worsened had
a higher risk of MACE (HR, 1.10; 95% CI, 1.02–1.20, P=0.022). Among
pre-hypertensive women, those whose lifestyles improved tended to have a lower MACE
risk, although this association was not statistically significant (HR, 0.91; 95% CI,
0.81–1.01, P=0.072) (Table 3, Supplement 3). In the
prediabetes group, men whose lifestyles worsened showed a significantly higher risk
of MACE (HR, 1.23; 95% CI, 1.12–1.35, P<0.001), while there was no
significant difference in MACE risk among comparative female subjects (Table 4, Supplement 4).
Table 3.
Multivariate analysis of total cardiovascular disease events and mortality associated with lifestyle changes between two biennial health screening periods in the prehypertension group
| Outcome | Parameter | Change | Men | Women | ||
|---|---|---|---|---|---|---|
| Adjusted hazard ratio | P-value | Adjusted hazard ratio | P-value | |||
| [95% CI] | [95% CI] | |||||
| MACE1) | Unhealthy lifestyle | Worsened | 1.13 [1.04–1.23] | 0.004 | 1.10 [0.98–1.23] | 0.121 |
| Improved | 0.91 [0.84–0.99] | 0.025 | 0.94 [0.85–1.04] | 0.243 | ||
| Unhealthy lifestyle (exclude BMI) | Worsened | 1.10 [1.02–1.20] | 0.022 | 1.03 [0.90–1.17] | 0.718 | |
| Improved | 0.92 [0.85–1.00] | 0.052 | 0.91 [0.81–1.01] | 0.072 | ||
| BMI | Worsened | 1.15 [1.03–1.28] | 0.010 | 1.09 [0.95–1.25] | 0.220 | |
| Improved | 0.98 [0.88–1.09] | 0.683 | 0.97 [0.85–1.11] | 0.642 | ||
| Currently smoking | Worsened | 1.34 [1.17–1.55] | <0.001 | 1.69 [1.15–2.49] | 0.008 | |
| Improved | 0.79 [0.70–0.89] | <0.001 | 0.76 [0.48–1.21] | 0.247 | ||
| Frequency of drinking | Worsened | 1.08 [1.00–1.18] | 0.061 | 0.92 [0.78–1.08] | 0.307 | |
| Improved | 1.09 [1.00–1.20] | 0.048 | 1.10 [0.88–1.38] | 0.413 | ||
| Frequency of physical activity | Worsened | 1.09 [0.99–1.21] | 0.086 | 1.25 [1.06–1.47] | 0.010 | |
| Improved | 0.91 [0.84–0.99] | 0.027 | 0.93 [0.83–1.03] | 0.166 | ||
Adjusted covariates include age, income level (quantiles), current smoking status, alcohol consumption, physical activity, CCI score, BMI, systolic blood pressure, total cholesterol, fasting serum glucose level at the first health screening, and usage of statin medication prior to the index date.
Table 4.
Multivariate analysis of total cardiovascular disease events and mortality associated with lifestyle changes between two biennial health screening periods in the prediabetes group
| Outcome | Parameter | Change | Men | Women | ||
|---|---|---|---|---|---|---|
| Adjusted hazard ratio | P-value | Adjusted hazard ratio | P-value | |||
| [95% CI] | [95% CI] | |||||
| MACE1) | Unhealthy lifestyle | Worsened | 1.23 [1.12–1.35] | <0.001 | 1.09 [0.96–1.24] | 0.199 |
| Improved | 1.05 [0.96–1.14] | 0.304 | 0.96 [0.85–1.07] | 0.425 | ||
| Unhealthy lifestyle (exclude BMI) | Worsened | 1.23 [1.12–1.35] | <0.001 | 0.97 [0.83–1.13] | 0.671 | |
| Improved | 1.08 [0.99–1.17] | 0.084 | 0.97 [0.86–1.09] | 0.64 | ||
| BMI | Worsened | 1.19 [1.06–1.33] | 0.003 | 1.24 [1.06–1.45] | 0.006 | |
| Improved | 0.93 [0.83–1.04] | 0.214 | 0.93 [0.80–1.07] | 0.315 | ||
| Currently smoking | Worsened | 1.41 [1.22–1.64] | <0.001 | 0.88 [0.46–1.69] | 0.693 | |
| Improved | 0.91 [0.80–1.03] | 0.138 | 1.08 [0.69–1.70] | 0.741 | ||
| Frequency of drinking | Worsened | 1.08 [0.99–1.18] | 0.102 | 0.96 [0.81–1.14] | 0.65 | |
| Improved | 1.17 [1.07–1.29] | 0.001 | 1.06 [0.82–1.38] | 0.659 | ||
| Frequency of physical activity | Worsened | 1.21 [1.09–1.35] | <0.001 | 1.05 [0.86–1.28] | 0.661 | |
| Improved | 0.98 [0.90–1.07] | 0.632 | 0.99 [0.88–1.12] | 0.912 | ||
Adjusted covariates include age, income level (quantiles), current smoking status, alcohol consumption, physical activity, CCI score, BMI, systolic blood pressure, total cholesterol, fasting serum glucose level at the first health screening, and usage of statin medication prior to the index date. MACE, major adverse cardiovascular events; BMI, body mass index; MI, myocardial infarction; CVD, cardiovascular disease.
To mitigate the risk of reverse causality, a sensitivity analysis was conducted by
excluding cardiovascular events that occurred within two years following the
observation of lifestyle changes. In the prehypertension male group, the risk of
MACE increased among subjects who experienced a decline in lifestyle quality, gained
weight, decreased their physical activity frequency, or began smoking between
biennial screenings. Conversely, the risk decreased in those who improved their
lifestyles or quit smoking. In the prehypertension female group, an increase in MACE
risk was observed in subjects who started smoking (Fig. 2A). In the prediabetes group, the MACE risk escalated in men who
worsened their lifestyle, gained weight, reduced their physical activity frequency,
or started smoking between the biennial screenings. In women, the risk increased
among those who gained weight (Fig. 2B).
Fig. 2.
Risk of cardiovascular disease events and mortality due to lifestyle changes. (A) The prehypertension group; (B) the prediabetes group. For the sensitivity analysis, events occurring 2 years after the observation of the lifestyle changes were excluded and evaluated. MACE, major adverse cardiovascular events; BMI, body mass index.
Discussion
Key results
This study highlights two significant findings: First, the risk of MACE increased
in men with prediabetes or prehypertension whose lifestyle factors deteriorated,
and this risk escalated even with the worsening of just one parameter. Second,
the impact on MACE risk varied slightly depending on lifestyle changes, with
smoking being strongly linked to an increased risk of MACE in both prediabetes
and prehypertension conditions, irrespective of gender.
Interpretation/comparison with previous studies
Smoking is a well-known risk factor for CVD risk. However, the precise level of
risk that smoking presents to CVD in the prehypertensive population has not been
clearly defined. The incidence of CVD is higher in people with a blood pressure
of 120–129/80–84 mmHg and 130–139/85–89 mmHg than in
normal blood pressure group in Europe and the United States [9]. Previous studies have demonstrated that
smoking significantly influences the progression from prehypertension to
hypertension by stiffening the arteries [10]. Furthermore, smoking increases the risk of developing
hypertension in a dose-response manner over long-term follow-up [11]. This study confirmed that MACE
increased as smoking habits worsened in both women and men. In prehypertensive
patients, smoking not only contributes to the progression to hypertension but
also independently elevates the risk of CVD. These findings underscore the
importance of smoking cessation in patient education for individuals with
prehypertension.
In a previous study, smoking was strongly associated with prediabetes in healthy
young individuals, showing a linear risk gradient with increased cumulative
smoking exposure [12]. Additionally,
prior epidemiological studies have indicated that smoking may be an independent
risk factor for T2D and is significantly associated with a higher risk of
coronary heart disease [13–16]. A meta-analysis of prospective cohort
studies revealed a relative risk of 1.55 (95% CI 1.46–1.64) for total
mortality and 1.29 (95% CI 1.29–1.71) for cardiovascular mortality,
assessing the relationship between smoking and mortality risk in diabetes
patients [17]. In this study, men with
prediabetes who started smoking showed an increased risk of MACE, while smoking
cessation decreased CVD death.
In this study, we also analyzed changes in the risk of MACE, specifically
excluding BMI to assess whether its inclusion as a factor influenced the
outcomes. Males in the prehypertension and prediabetes groups who experienced a
deterioration in their lifestyle had an increased risk of MACE, a trend that
persisted even when BMI was excluded. However, among women in both groups, the
risk of MACE did not show significant differences when unhealthy lifestyle
parameters were considered, with or without BMI (Tables 3, 4). Previous
research has demonstrated a U-shaped relationship between obesity and all-cause
mortality and a linear association between BMI and cardiovascular events [18–20]. In this context, changes in an individual's BMI category
do not directly correlate with an increase or decrease in CVD risk. For
instance, overweight patients with stable coronary heart disease exhibited lower
mortality compared to those of normal weight, and overweight patients with acute
coronary syndrome showed significantly lower in-hospital and 12-month mortality
[21–23]. Furthermore, BMI does not accurately reflect body
composition, such as muscle mass or fat distribution. This phenomenon, known as
'reverse causality,' suggests that BMI is often considered a
result influenced by other confounding factors rather than a standalone factor
in previous studies [24]. In particular,
the BMI values in this dataset do not indicate any intention to lose weight.
Additionally, a decrease in BMI could also result in a reduction in muscle mass.
Given these considerations, although the primary outcome based on lifestyle
changes, excluding BMI, revealed significant differences in MACE risk among male
groups, an analysis of the impact of BMI parameters on the detailed elements of
MACE yielded conflicting results (Supplementals 5, 6).
Previous studies have shown that women generally have lower rates of smoking and
obesity than men, while men are more likely to engage in regular and sustained
physical activity [25,26]. When comparing men and women, the
increase in MACE among men whose unhealthy lifestyles worsened could be linked
to differences in lifestyle patterns between the genders. Indeed, data from the
NHIS indicate gender-specific lifestyle patterns; men are more likely to smoke,
be obese, and consume more alcohol. A previous study demonstrated that moderate
to high physical activity during leisure time was associated with a lower risk
of MACE [27]. Specifically, Fig. 2 illustrates that a decreased frequency
of physical activity was associated with an increased risk of MACE in both the
prehypertension and prediabetes male groups.
Limitations
This study may have several limitations. First, there is a potential for
misclassification bias, as the diagnoses of prediabetes and prehypertension were
based on the recorded history in the NHIS and single measurements of blood
pressure or serum glucose levels. Second, the data on lifestyle patterns were
derived from self-reported questionnaires, which may have introduced some
misclassification. Third, further research is necessary to determine if similar
trends are observable in multi-racial populations beyond Asians. Fourth, the
study assessed changes in lifestyle by analyzing data collected at two distinct
time points, specifically between the index examination and the second
examination. However, it is possible that any lifestyle changes occurring after
the second examination were not captured.
Conclusion
In this retrospective cohort study, a worsening smoking habit was associated with
an increased incidence of MACE in populations with prehypertension and men with
prediabetes. This finding remained consistent after sensitivity analysis.
Specifically, a decline in healthy lifestyle habits among men significantly
increased the risk of MACE. Therefore, lifestyle modifications are crucial even
before the diagnosis of hypertension or T2D, as they can significantly reduce
cardiovascular events and mortality, particularly in men.
XML Download