Bolstering the Prognostic Utility of Coronary Risk Assessments with PAI: A Physical Activity Metric

ABSTRACT

Purpose

Personal activity intelligence (PAI) translates heart rate during physical activity (PA) into a weekly score, which credits vigorous over low- and moderate-intensity PA. We prospectively investigated the association between PAI and fatal and nonfatal coronary heart disease (CHD) in self-reported healthy participants from Norway, with specific reference to improving the accuracy of conventional coronary risk assessment.

Methods

We studied 40,961 healthy adults (56% women) from the population-based Trøndelag Health Study (the HUNT study). Individual data were linked to hospital and cause of death registries. The weekly PAI score of each participant was divided into four groups (PAI scores of 0, ≤50, 51–99, or ≥100). Adjusted hazard ratios and 95% confidence intervals for fatal and nonfatal CHD related to PAI were estimated using Cox proportional hazard regression analyses.

Results

During a median follow-up period of 13.1 yr (interquartile range, 12.7–13.6), 3303 (3109 nonfatal, 194 fatal) CHD events occurred. Compared with the inactive group (0 PAI), weekly PAI scores at baseline of 51–99 and ≥100 were associated with a lower risk of CHD [0.80 (0.71–0.91) and 0.86 (0.78–0.95), respectively]. By adding PAI to traditional risk factors, the net reclassification improvement of CHD was 0.472 (P < 0.001).

Conclusions

PAI was inversely associated with CHD risk among healthy participants at baseline, and its cardioprotective effect persisted across diverse risk factor profiles. A PAI score >50 was substantially associated with a reduced risk of CHD. These findings have implications for improving the accuracy of conventional coronary risk assessments with PAI.

Coronary heart disease (CHD) remains the leading cause of cardiovascular disease (CVD) death worldwide, accounting for >9 million deaths and 185 million disability-adjusted life years in 2021 (1,2). Although the global burden of CHD increased from <100 million in 1990 to 258 million in 2021 (1,3), the age-standardized prevalence of CHD during this period was relatively unchanged, suggesting that the rise in CHD worldwide is attributed to population growth and its associated aging. The increasing number of absolute CHD cases suggests that countries and health systems need to place greater emphasis on “proactive” healthcare, with a stronger focus on preventing diseases and enhancing overall well-being (4). Nevertheless, this approach will not be achieved without paying greater attention to underserved, less educated, lower socioeconomic population subsets (5).

Regular physical activity (PA) is inversely associated with all-cause and CVD mortality, and sufficient levels of PA serve as an evidence-based, inexpensive, readily accessible approach to prevent and treat chronic diseases (6,7). Despite the recognized benefits of regular PA, a pooled analysis of 358 population-based surveys including 1.9 million participants reported that ~23% of men and ~32% of women worldwide did not meet the recommended levels of weekly PA (i.e., a minimum of 150 min of moderate-intensity PA, or a minimum of 75 min of vigorous-intensity PA, or combinations thereof) for health benefits (8). Accordingly, an estimated 0.4 million CVD deaths globally were attributed to inadequate PA levels in 2021 (2). The 2022 Global Status Report on Physical Activity by the World Health Organization (7) introduced a comprehensive set of strategies, along with assessment and monitoring frameworks, to enhance PA. A primary focus of these recommendations is the potential use of mobile and wearable technologies as tools for guiding and supporting individuals to improve their participation in regular PA. These technologies have the potential to transform healthcare services by efficiently reaching escalating numbers of people at minimal cost (7). Indeed, the results of a recent systematic review of systematic reviews and meta-analyses showed that PA wearable technologies were effective at increasing PA (∼40 min more walking per day) in both clinical and general populations and across various age groups. The results of the review further emphasized that these benefits of improved PA are sustained over time (9).

Personal activity intelligence (PAI) is a metabolic metric that was developed to provide an objective measure of individual’s PA levels and has since been incorporated into self-assessment heart rate devices and health apps. PAI incorporates an individual’s age, sex, resting and maximal heart rate, and transforms the cumulative fluctuations in heart rate and associated energy expenditure during PA over the last 7 d into a straightforward and user-friendly score, where 0 PAI signifies inactivity, and 100 PAI represents health-promoting levels of PA (10,11). Objective PAI measurements through wearable heart monitors have shown to improve exercise capacity and sleep duration in individuals with type 2 diabetes (12) and increased PA among cardiac patients (13). Similarly, PAI estimations based on self-reported data in large healthy cohorts have shown that a weekly PAI score ≥100 was associated with a lower risk of death from all-causes, CVD, and dementia (10,14–18). To our knowledge, only two previous investigations, one conducted in the United States using the Aerobics Center Longitudinal Study database (16) and another from the China Kadoorie Biobank Study (19), examined the association between PAI and CHD showing that high weekly PAI scores were associated with a lower risk of CHD mortality. However, data regarding the prospective association of PAI with CHD in a European population with varying demographic and PA profiles are lacking.

Epidemiological and biological plausibility studies support a cause-and-effect relationship between increased levels of PA, cardiorespiratory fitness, or both and reduced CHD events (20). Moreover, inactive and low-fit individuals are two to three times more likely to die prematurely from CHD than their active, fitter counterparts when matched for risk factor profile (6,21). Accordingly, it remains unclear why conventional coronary risk assessments have not included indices of fitness or PA (22). The present study was designed to investigate the association between PAI and fatal and nonfatal CHD and to determine if the addition of PAI to a conventional coronary risk assessment could enhance the prediction of future events in an apparently healthy cohort of men and women from Norway. We hypothesize that higher levels of weekly PAI are associated with lower risk of CHD events, and the integration of PAI into coronary risk assessment could enhance prognostic accuracy relative to identifying subsequent fatal and nonfatal CHD events.

METHODS

Study participants

The Trøndelag Health Study (the HUNT study) is a population-based health study conducted in the Trøndelag county located in central Norway. The HUNT study comprises four surveys, conducted at different time points: HUNT1 (1984–1986), HUNT2 (1995–1997), HUNT3 (2006–2008), and HUNT4 (2017–2019). These surveys involve the longitudinal follow-up of participants and are supplemented with data from comprehensive national health registries. All adults 20 yr old and older in the northern region of county (Nord-Trøndelag) were invited to participate in these surveys. During each visit, individuals completed questionnaires regarding their health status and lifestyle and underwent varied health assessments. Detailed accounts of the HUNT surveys have been previously described (23,24). In the present study, we used data from HUNT3 (recruitment: October 2006 to June 2008). Of the 50,800 participating individuals ≥20 yr old, we excluded those with a self-reported history of myocardial infarction (MI), angina, stroke, heart failure or other CVD (n = 6210), those with missing data on PA (n = 1380), and those with incomplete information on selected potential confounders (n = 2249). A total of 40,961 individuals (22,869 women and 18,092 men) were included in our current analyses (see Supplemental Fig. 1, Supplemental Digital Content, showing the flow of the participants in the study, http://links.lww.com/MSS/D116). All individuals provided written informed consent before enrolling in and completing HUNT surveys. The study was approved by the Data Inspectorate and the Regional Committee on Medical and Health Research Ethics of Norway (2019/REK Midt 67711).

Clinical and questionnaire-based information

Trained health professionals assessed body weight, height, and resting blood pressure (BP) with calibrated instruments using standardized methods (23,24). BP was determined using a Dinamap 845XT (Critikon), and height and weight were measured with participants wearing light clothes and no shoes. Participants completed questionnaires providing information on a wide range of variables including PA levels, smoking status, history of chronic disease including diabetes or CVDs, marital status, alcohol consumption, and use of BP medications. Venous blood sampling was done following a standardized protocol, and serum concentrations of total cholesterol (enzymatic cholesterol esterase methodology) and serum glucose (Hexokinase/G-6-PDH methodology) were analyzed (Clinical Chemistry, Abbott, Abbott Park, IL) (23,24).

PAI

An individual’s PAI score was estimated using the responses to self-reported PA questions about frequency, duration, and intensity. Frequency of PA was assessed by the query “How often do you exercise? (on average)” with options: “Never,” “Less than once a week,” “Once a week,” “2–3 times a week,” “Nearly every day,” and were translated to 0, 0.5, 1, 2.5, and 5 weekly days, respectively. The duration of PA was assessed with “For how long do you exercise each time?” with options: “Less than 15 minutes,” “16–30 minutes,” “30 minutes–1 hour,” “More than 1 hour,” and were translated to 7.5, 22.5, 45, and 60 minutes, respectively. Exercise intensity was assessed by asking “How hard do you exercise?” with three options: “no sweating or heavy breathing,” “heavy breathing and sweating,” and “push myself to exhaustion.” According to the PAI algorithm previously described (10,14), the strenuousness of PA was expressed as a relative exercise intensity, with ~44%, 73%, and 83% of heart rate reserve signifying low-, moderate-, and vigorous-intensity activities, respectively. Weekly minutes of PA were computed by multiplying the average frequency and duration. These exercise volumes were then combined with relative intensity calculations to estimate the weekly PAI scores (10,14).

End points and follow-up

The HUNT3 baseline data were linked to a local, validated hospital database (Nord-Trøndelag Hospital Trust Myocardial Infarction Registry) and the Norwegian Cause of Death Registry. We used 10th International Classification of Diseases (ICD10) codes to define CHD end points. The primary end point in the analyses was defined as diagnosis of, or death from CHD (ICD10: I20–I25), and the secondary outcomes were diagnosis of, or death from MI (ICD10: I21–I23); diagnosis of, or death from chronic CHD (ICD10: I25); or death from CVDs (ICD: I00–I99). Study participants were followed from baseline to date of event, or to January 1, 2021, whichever came first.

Statistical analysis

Baseline characteristics of the participants are presented as number (%) for categorical variables and mean (SD) for continuous variables. Participants were categorized into four groups according to their weekly PAI score: 0 (inactive), ≤50, 51–99, or ≥100 (10,14,19). The inactive group (0 PAI) was used as a reference cohort. Further, we categorized individuals into weekly PAI scores <100 and ≥100. We used Cox proportional hazard models to investigate the associations between PAI and varied end points. Basic models were stratified by sex and age (5-yr age-at-risk interval, which is a time-varying covariate determining the current risk rather than the age when the risk factor was assessed) (14). The final multivariable-adjusted model further included body mass index (18.5–24.9, 25–29.9, or ≥30 kg·m⁻²), smoking status (never, former, current, or occasionally), marital status (married, unmarried, divorced/separated, or widow/widower), alcohol consumption over a 2-wk period (0 to ≤7 drinks, >7 to ≤14 drinks, or >14 drinks), hypertension (yes [systolic BP ≥140 mm Hg or diastolic BP ≥90 mm Hg or taking BP medications], or no), diabetes status (yes [self-reported history of diabetes or nonfasting serum glucose >11.1 mmol·L⁻¹], or no), elevated serum cholesterol (yes [age-based serum cholesterol levels: >6.1 mmol·L⁻¹ for those <30 yr, >6.9 mmol·L⁻¹ for those between the ages of 30 and 49, >7.8 mmol·L⁻¹ for those ≥50 yr], or no), occupation (armed forces, legislators/officials, professional, technicians, clerks, service/shop/market sales, skilled agriculture/fishery, craft and related trades, machine and plants operators, elementary occupations or other/not stated), and family history of CVD (yes, no). The proportional hazard assumption was examined and satisfied with the use of Schoenfeld residuals, and results are reported as adjusted hazard ratios (aHR) and 95% confidence intervals (CI) to indicate the precision of estimates. We also investigated the associations of PAI with CHD in prespecified subgroups of participants based on baseline age, and those with known CHD risk factors, such as smoking, hypertension, diabetes, or overweight/obesity.

We also estimated the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) by adding PAI to a Norwegian CHD risk prediction model (NORRISK2) (25). The risk factors included in the NORRISK2 were age, systolic BP, BP medications, serum total cholesterol, high-density lipoprotein (HDL) cholesterol, smoking status, and family history of CVD. The 10-yr risk of CHD was estimated, separately for men and women, based on the risk algorithm provided in the NORRISK2 (25). Next, we generated a reclassification table to assess the improvement in the prediction performance when adding PAI to the NORRISK2 (26,27). We calculated net correct reclassification separately for those with CHD events, and those who survived until the end of follow-up (nonevents), and total NRI was estimated based on the proportion of participants moving up or down in terms of predicted risk category (28).

To assess the robustness of our findings, we conducted sensitivity analyses by excluding the first 3 yr of follow-up to minimize the likelihood of bias due to reverse causality. In another sensitivity analysis, we adjusted for additional variables such as serum creatinine, C-reactive protein, triglycerides, HDL cholesterol, calcium, serum phosphate, thyroid stimulating hormones, serum urea, estimated glomerular filtration rate, and presence of common chronic disorders, including ankylosing spondylitis, sarcoidosis, osteoporosis, osteoarthritis, and fibromyalgia. In a separate sensitivity analysis, we adjusted for fruits and berries consumption, high-fat fish (salmon, trout, herring, mackerel, or haddock) intake, dietary supplements of omega-3 or vitamins or minerals, whole milk, soda/squashes, or daily coffee ingestion in addition to the characteristics and descriptors that were already included in the multivariable models. All statistical tests were two-sided, and P < 0.05 was considered significant. The statistical analyses were performed using Stata for Windows (version 16, StataCorp LLC, College Station, TX).

RESULTS

The characteristics of the study participants stratified by sex and, according to their baseline weekly PAI scores and CHD status, are presented in Table 1 and Supplemental Table 1 (Supplemental Digital Content http://links.lww.com/MSS/D116). Overall, the proportion of women with overweight and obesity, hypertension, and diabetes was lower than men. However, women had higher prevalence of current smoking and hyperlipidemia, and the proportion of women at baseline with ≥100 weekly PAI was greater than men. Furthermore, participants with high weekly PAI scores had a lower prevalence of obesity, hypertension, hyperlipidemia, current smoking, and diabetes, regardless of the event status at follow-up (see Supplemental Table 1, Supplemental Digital Content, http://links.lww.com/MSS/D116).

TABLE 1.

Characteristics of study participants.

	Women (n = 22,869)	Men (n = 18,092)
Age, mean (SD), yr	50.5 (15.4)	51.0 (14.7)
PAI, n (%)
Inactive	3982 (17.4)	4984 (27.6)
≤50	7623 (33.3)	4436 (24.5)
51–99	2755 (12.1)	2389 (13.2)
≥100	8509 (37.2)	6283 (34.7)
Body mass index, n (%), kg·m−2
<18.5	191 (0.8)	54 (0.3)
18.5–24.9	9049 (39.6)	4651 (25.7)
25.0–29.9	8557 (37.4)	9531 (52.7)
≥30.0	5072 (22.2)	3856 (21.3)
Systolic BP, mean (SD), mm Hg	126.8 (18.9)	133.4 (16.4)
Diastolic BP, mean (SD), mm Hg	70.9 (10.6)	76.6 (11.0)
Hypertension status, n (%)
No	15,671 (68.5)	11,040 (61.0)
Yes	7198 (31.5)	7052 (39.0)
Hyperlipidemia status, n (%)
No	21,893 (95.7)	17,386 (96.1)
Yes	976 (4.3)	706 (3.9)
Smoking status, n (%)
Never	10,134 (44.3)	7884 (43.6)
Former	6685 (29.2)	5960 (32.9)
Current	4494 (19.7)	2720 (15.0)
Occasionally	1556 (6.8)	1528 (8.5)
Diabetes status, n (%)
No	22,125 (96.7)	17,328 (95.8)
Yes	744 (3.3)	764 (4.2)
Alcohol intake, n (%)a
0 to <7	19,487 (85.2)	11,932 (65.9)
7 to ≤14	2893 (12.7)	4622 (25.6)
>14	489 (2.1)	1538 (8.5)
Family history of CVD, n (%)
No	17,827 (77.9)	14,672 (81.1)
Yes	5042 (22.1)	3420 (18.9)

^a Average number of drinks over last 2 wk.

During a median follow-up period of 13.1 yr (interquartile range, 12.7–13.6, 507,168 person-years), a total of 3303 primary end point events of either diagnosis (n = 3109) or death (n = 194) from CHD occurred. High weekly PAI levels were associated with higher event-free survival from the primary end point as shown in Figure 1. The absolute risks of CHD per 1000 person-years were 8.04, 7.91, 5.28, and 4.96 for inactive and those with weekly PAI scores of ≤50, 51–99, and ≥100, respectively. Multivariable-adjusted analyses showed that healthy individuals at baseline with high weekly PAI levels also had lower relative risks of CHD. Compared with the inactive (0 PAI) cohort, aHR associated with a weekly PAI score of 51–99 and ≥100 were 0.80 (95% CI = 0.71–0.91) and 0.86 (95% CI = 0.78–0.95), respectively (Table 2). Using participants with <100 weekly PAI as the reference cohort, aHR for the primary end point associated with ≥100 weekly PAI was 0.87 (95% CI = 0.81–0.94).

FIGURE 1.

Kaplan–Meier survival estimates for primary end point according to PAI.

TABLE 2.

Hazard ratios for primary^a end point by PAI.

PAI	Person-Years	Events	Rateb	HR	(95% CI)c	HR	(95% CI)d
Inactive	108,829	876	8.04	1.00	(Ref.)	1.00	(Ref.)
≤50	145,363	1151	7.91	0.98	(0.90–1.08)	1.06	(0.97–1.17)
51–99	65,307	345	5.28	0.72	(0.63–0.81)	0.80	(0.71–0.91)
≥100	187,669	931	4.96	0.70	(0.64–0.77)	0.86	(0.78–0.95)
<100	319,499	2372	7.42	1.00	(Ref.)	1.00	(Ref.)
≥100	187,669	931	4.96	0.75	(0.70–0.81)	0.87	(0.81–0.94)

^a Diagnosis or death from CHD.

^b Rate per 1000 person-years.

^c Adjusted for age and sex.

^d Adjusted for age, sex, BMI (<18.5, 18.5–24.9, 25–29.9, or ≥30 kg·m⁻²), smoking (never, previous, current or occasionally smokers), marital status (single, married, widow/er, or divorced/separated), alcoholic drink consumption over a 2-wk period (abstainers, 0– ≤ 7, >7– ≤ 14, or >14), occupation (armed forces, legislators/officials, professional, technicians, clerks, service/shop/market sales, skilled agriculture/fishery, craft and related trades, machine and plants operators, elementary occupations or other/not stated), hypertension (yes, no), diabetes status (yes, no), hyperlipidemia (yes, no), and family history of cardiovascular disease.

HR, hazard ratio.

The results of secondary end points of MI, chronic CHD, and CVD mortality were compatible with the findings in the main analyses (Fig. 2). Compared with inactive participants, aHR values associated with ≥100 weekly PAI were 0.75 (95% CI = 0.65–0.86) for MI, 0.65 (95% CI = 0.52–0.82) for chronic CHD, and 0.54 (95% CI = 0.45–0.64) for CVD mortality.

FIGURE 2.

Hazard ratio for secondary end points by PAI. HR, hazard ratio; CVD, cardiovascular disease. Diamonds represent the HR, and horizontal bars represent the 95% CI. ^aAdjusted for age, sex, BMI (<18.5, 18.5–24.9, 25–29.9, or ≥30 kg·m⁻²), smoking (never, previous, current, or occasionally smokers), marital status (single, married, widow/er, or divorced/separated), alcoholic drink consumption over a 2-wk period (abstainers, 0– ≤7, >7–≤ 14, or >14), occupation (armed forces, legislators/officials, professional, technicians, clerks, service/shop/market sales, skilled agriculture/fishery, craft and related trades, machine and plants operators, elementary occupations or other/not stated), hypertension (yes, no), diabetes status (yes, no), hyperlipidemia (yes, no), and family history of CVD.

In prespecified subgroups of participants, a weekly PAI score ≥100 was associated with a lower risk of CHD (see Supplemental Table 2, Supplemental Digital Content, http://links.lww.com/MSS/D116). For example, aHR values associated with ≥100 weekly PAI were 0.90 (95% CI = 0.80–1.00) in men, 0.68 (95% CI = 0.57–0.80) in women, 0.74 (95% CI = 0.65–0.84) in participants ≥60 yr at baseline, 0.76 (95% CI = 0.61–0.94) in current smokers, 0.81 (95% CI = 0.72–0.92) in hypertensive patients, 0.81 (95% CI = 0.58–1.12) in diabetics, and 0.86 (95% CI = 0.77–0.96) in overweight/obese participants as compared with the inactive reference group (Supplemental Table 2, Supplemental Digital Content, http://links.lww.com/MSS/D116).

The addition of PAI to the NORRISK2 model classified more participants in a lower risk classification than a higher category, giving a total NRI of 0.472 (P < 0.001) (Table 3). The enhanced model with PAI correctly reclassified 0.33% of participants with CHD, and among those who survived, more people were correctly classified to a lower risk category giving a net correct reclassification of nonevents of 46.9%. The estimated IDI was 0.065 (P < 0.001).

TABLE 3.

Reclassification of CHD risk after addition of PAI to traditional risk factors (NORRISK2).

		NORRISK2 + PAI
		<5%	5%–7.5%	>7.5%	Total	Net Correctly Reclassified
NORRISK2	CHD
	<5%	7	7	17	31	0.33%
	5%–7.5%	142	157	601	900
	>7.5%	169	303	1900	2372
	Total	318	467	2518	3303
	No CHD
	<5%	683	83	171	937	46.86%
	5%–7.5%	7497	1957	3470	12,924
	>7.5%	10,390	3485	9922	23,797
	Total	18,570	5525	13,563	37,658
						NRI: 0.472 (P < 0.001)
						IDI: 0.065 (P < 0.001)

NRI, net reclassification improvement; IDI, integrated discrimination improvement.

In the sensitivity analyses, after additional adjustments for potential confounders were considered, the results remained like those findings of the main analyses (see Supplemental Table 3, Supplemental Digital Content, http://links.lww.com/MSS/D116). After adjusting for additional clinical variables and the presence of common chronic disorders, aHR for primary end point associated with ≥100 weekly PAI was 0.82 (95% CI = 0.71–0.94), and aHR was 0.87 (95% CI = 0.76–0.99) when further adjusted for fruits and berries consumption, high-fat fish intake, or chronic ingestion of dietary supplements (e.g., omega-3) or vitamins or minerals, whole milk, soda/squashes, or daily coffee intake. (see Supplemental Table 3, Supplemental Digital Content, http://links.lww.com/MSS/D116). There were 2624 primary end point events after excluding the first 3 yr of follow-up, and the results were not substantially altered [aHR, 0.88 (95% CI = 0.79–0.98, associated with ≥100 weekly PAI)] from the main findings (see Supplemental Table 4, Supplemental Digital Content, http://links.lww.com/MSS/D116).

DISCUSSION

In a cohort of self-reported “healthy” participants at baseline, we found that high weekly PAI scores were associated with a lower risk of CHD. Moreover, the independent and additive cardioprotective effect of PAI persisted across diverse risk factor profiles. Previous studies have reported that physically active and fit individuals with any combination of risk factors had lower CHD adjusted death rates than inactive individuals with none of these characteristics (29–33). Collectively, these data raise the question as to why conventional CHD risk assessments have failed to include metrics related to PA.

The main findings of our study are consistent with and extend earlier observations showing a lower risk of CHD associated with a high weekly PAI score (16,19). In a study of 443,792 healthy Chinese (19), a weekly PAI score of ≥100 was associated with 9% lower risk of future CHD, and the results of Aerobics Center Longitudinal study (16) showed a 30% lower risk of CHD mortality associated with a weekly PAI score ≥100, compared with an inactive control group. Interestingly, the magnitude of risk reduction (14%) associated with ≥100 PAI observed in our study is comparable with the CVD risk reduction associated with a one MET (metabolic equivalent of the task: 1 MET = 3.5 mL·kg⁻¹⋅min⁻¹) increment in cardiorespiratory fitness (CRF) (34), which may represent the single best predictor of overall and cardiovascular health (20).

Our results show that high weekly PAI scores are associated with a lower incidence of MI and risk of chronic CHD. A previous study examining the relation between PAI and MI showed that high weekly PAI scores were associated with a lower risk in participants 60 yr and older (19). For chronic CHD, to our knowledge, we are the first to report a 35% lower risk associated with a weekly PAI score ≥100.

We also observed that weekly PAI scores >50 and <100 were associated with a lower risk of adverse cardiovascular health outcomes. For the primary end point of CHD, we observed a 20% lower risk, whereas for MI, chronic CHD, and overall CVD mortality, the risk reductions were 25%, 31%, and 49%, respectively, associated with a 51–99 weekly PAI score. These findings support the results of previous meta-analyses showing that “some PA is better than none” for the association between PA and risk of CHD (35), and a transition from an inactive state to regularly engaging in moderate PA was associated with a 20% decrease in the risk of both incident cases and mortality from CHD, and low to moderate levels of PA (<11.5 MET·h·wk⁻¹) were associated with favorable outcomes (36). Additional findings from a general population cohort further confirmed that a PA of at least 9.4 MET·h·wk⁻¹ was associated with a 34% lower risk of MI and 51% lower risk of chronic CHD (37). In this context, while striving toward evidence-based optimal goal (i.e., PAI score ≥100) should be considered as the ultimate objective for individuals, pragmatic increases in PA, coupled with transitioning away from a sedentary lifestyle, are highly beneficial. As such, a “move more sit less” message should be considered an overarching theme in PA promotion and counseling (38).

In subgroups of participants stratified according to hypertension, diabetes, body mass index, and smoking status, we observed a 14% to 37% lower risk of CHD associated with a high weekly PAI score. These findings suggest that specific “at risk” patient subsets might lower their risk of CHD by attaining higher weekly PAI scores. In prior studies involving these patient subgroups, a favorable impact of increased PA on CVD mortality has been observed. For example, moderate to vigorous levels of PA were correlated with a reduced incidence of CVD events among hypertensive (pre-, stage 1, or stage 2) and diabetic patients (39,40). We observed sex differences for CHD risk associated with weekly PAI score, where women appeared to benefit more at equivalent PAI levels. Our results are consistent with previous findings showing that women derive greater gains in mortality reductions compared with men in relations of PA (35,41). For example, women had a twofold lower risk of CHD mortality compared with men at similar PA levels (35), and active women had a 36% lower risk of CVD mortality whereas active men had 14% risk reduction compared with their inactive counterparts (41). The observed sex differences in mortality reductions associated with PA maybe attributable to underlying biological factors, and varying physiological response to exercise (42,43), which may make women more responsive to the protective effect of PA compared with men (41).

The addition of PAI to NORRISK2 traditional risk factors, which included age, sex, resting systolic BP and use of BP medications, total and HDL cholesterol, smoking status, and family history of CVD (25), correctly reclassified 18.92% of CHD cases into a higher risk category but incorrectly reclassified 18.6% of cases into a lower risk category, with a corresponding net correct reclassification of 0.33% among those with a CHD event. In nonevents, a larger proportion of survivors were correctly reclassified to be in a lower risk category (56.8%), and a positive total NRI (0.472, P < 0.001) suggests that the independent and additive inclusion of PAI scores into the NORRISK2 model may improve the accuracy of CHD risk stratification. The findings of the previous studies incorporating PA or fitness into established risk prediction tools suggest that PA or fitness may provide additional prognostic value for CVD and CHD risk assessment and can enhance the accuracy of risk prediction beyond the traditional risk factors (22,44,45). However, none of the risk prediction models incorporate PA or fitness or PAI, despite their well-documented and consistent association with the CVD risk.

Recent studies have focused on the reductions in overall CVD and all-cause mortality with achieving even less than 10,000 steps per day (46–48). However, all steps are not created equally, and some faster steps are helpful to increase CRF and to elevate heart rates to achieve higher PAI scores. We previously reported that those who did not meet PA recommendations but who had high PAI scores had CVD—and mortality reductions, whereas those who met PA recommendations but who had low PAI scores did not have significant CVD—and mortality reductions (10). Therefore, PAI appears to have greater prognostic significance than meeting PA recommendations for the prevention of CHD, CVD, and mortality. This is likely attributed, at least in part, to its reliance on the individual’s heart rate modulations to approximate the relative intensity (mild, moderate, vigorous, high) of PA.

Globally, the burden of CHD has continued to increase and accounted for ~43% of all CVD mortality in Western Europe in 2021 (1). The corresponding estimates were ~49% in Central Europe, ~54% in high-income North America and South Asia, and ~34% in Western Sub-Saharan Africa (1). On the other hand, more than a quarter of the world’s population (27.5%, 1.4 billion adults) do not meet the recommended level of PA, and the prevalence of insufficient PA has remained largely unchanged over the years (7,8). If this pattern persists, the global target of achieving a 15% relative reduction in physical inactivity by 2030 among the adult population will not be attained (7). In contrast to other report-based or objective measures of PA that typically rely on absolute measures of intensity, the PAI emphasizes on relative intensity of exercise (10). This approach is particularly advantageous as it aligns with health benefits regardless of an individual’s fitness level (49) and may be more intuitive and applicable for individuals who find it challenging to engage in PA or meet recommended PA guidelines (50). By focusing on relative intensity, PAI may offer a more accessible and personalized PA assessment for diverse populations, potentially improving PA engagement and adherence.

The main strengths of our study include a relatively large sample size of healthy men and women at baseline, a comprehensive source of information on possible confounders, and linkage to national registries ensuring the acquisition of high-quality outcome data. However, because of the observational nature of our study, the findings are not necessarily causal. The PAI estimation was based on the self-reported leisure time PA data, which are susceptible to classification bias. Nonetheless, in prospective studies, measurement errors, and resulting misclassification tend to be nondifferential regarding future disease, thus leading to probable underestimation of actual effects. Although we adjusted for possible confounders in our analyses, and excluded individuals with comorbid conditions at baseline, specifically those with MI, angina, stroke, or heart failure, our estimates may have been influenced by residual unmeasured or unknown variables such as prescribed medications, presence of arrythmias or occupational PA, as well as other unaccounted for biological and social factors. Interestingly, a large cohort study has shown an increased mortality risk associated with higher levels of occupational PA (51). Further, we observed no meaningful changes in the estimates after excluding events that occurred during the first 3 yr of follow-up, thereby reducing the likelihood of reverse causation bias. The HUNT study population is predominantly Caucasian and generally presents a more favorable baseline health profile than many other population cohorts. However, other prospective studies (16,19) utilizing the same PAI cutoff values as ours have shown an inverse association between PAI and CHD. These unique observations underscore the robustness and accuracy of the PAI metric across diverse ethnic backgrounds and socioeconomic groups, as well as its utility in U.S., Chinese, and European populations (Norway).

CONCLUSIONS

In this large, prospective study of relatively healthy individuals, PAI was inversely associated with the risk of CHD. Moreover, the addition of PAI to traditional risk factors improved the CHD risk prediction. The present findings highlight the significance of PAI as an adjunct or complement to conventional coronary risk assessments, which have traditionally ignored the prognostic impact that including indices of PA may provide.