The decline in physical activity in aging people is not modified by gender or the presence of cardiovascular disease.

BACKGROUND: A physically active lifestyle decreases the progression of atherosclerosis and consequently reduces cardiovascular mortality. However, activity levels are hampered by aging. The association between aging and physical inactivity might be gender-specific or be modified by the presence of cardiovascular disease (CVD). In this study, we investigated if the association between aging and physical inactivity was different between men and women and between primary and secondary prevention patients.
METHODS: We performed a cross-sectional analysis of three primary care samples including primary and/or secondary prevention patients (total n = 4726). Baseline data for sample 1 were collected in the years 2013-14, for sample 2 in 2009-10 and for sample 3 in 2009. Activity levels were measured by the patient reported Rapid Assessment of Physical Activity questionnaire. A multilevel regression analysis was used to explore the association between aging and physical inactivity, adjusted for confounders. We examined potential effect modification by gender and the presence of CVD on the association between aging and physical inactivity.
RESULTS: Multilevel regression revealed that aging was consistently associated with physical inactivity in three out of three samples. This association was, however, not modified by gender or the presence of CVD.
CONCLUSIONS: Cardiovascular risk management interventions should aim at increasing or maintaining physical activity levels in aging primary care population. This study does not support the notion that extra emphasis should be put in targeting men or women, or people with or without the presence of CVD.

Introduction

A physically active lifestyle decreases the progression of atherosclerosis and consequently reduces cardiovascular mortality. To tailor cardiovascular risk management (CVRM) interventions to individual patients’ needs it is critical to identify patient characteristics that are associated with physical inactivity.

A physically active lifestyle may attenuate the impact of age on mortality. Adherence to a physically active lifestyle, however, becomes more difficult in the aging population, due to a progressive lowering of physical functions and capabilities, especially in frail older people., It is reported that the decline in physical activity along with age is higher in women compared to men. This could be caused by the fact that women have lower self-efficacy and self-management to maintain physical activity levels than men., In addition, women suffer more frequently from depression symptoms and increased prevalence of comorbidity compared to men, and may, therefore, be less physically active than men. To our knowledge, there is no research conducted into gender differences with respect to physical inactivity in an aging primary care CVRM population.

The presence of cardiovascular disease (CVD) may also have a negative impact on activity levels in the aging patients because comorbidity and depressive symptoms are more prevalent resulting in lower activity levels compared to patients at high cardiovascular risk without the presence of CVD (primary prevention patients). Besides, frailty is, independent of age, also more common in patients with the presence of CVD (secondary prevention patients) compared to primary prevention patients.,, On the other hand, it has been reported that aging was inconsistently associated with physical inactivity in a secondary prevention population participating in cardiac rehabilitation.

To further explore the association between aging and physical inactivity, we used baseline data from three large cross-sectional primary care samples including primary and secondary prevention patients. The present study aimed to explore if the association between aging and physical inactivity was different between men and women and between primary and secondary prevention patients. The study investigates whether patient-specific lifestyle interventions to increase physical activity levels should be tailored to either men or women and/or to either primary or secondary prevention patients.

Methods

Study design

A secondary analysis of data from three primary care studies was conducted. Ethical approval was given for the studies from which we used the data. The declaration of Helsinki was followed. In our study, we followed the STrengthening the Reporting of OBservational studies in Epidemiology statement for the cross-sectional cohort design.

Setting

Baseline data of three cross-sectional primary care samples, used in our study, were collected by the general practitioner or a nurse practitioner. Data for sample 1 were collected in the years 2013–14, for sample 2 in 2009–10 and for sample 3 in 2009. Data were used from Dutch primary care general practitioner facilities that offered CVRM.

Participants

Sample 1

Dutch baseline data, of European Tailored Implementation of Cardiovascular risk management (TICD) project were used, consisting of primary (n = 1250) and secondary prevention patients (n = 934). All patients were 18 years or older and were able to provide informed consent. Exclusion criteria were diabetes, pregnancy and lactation, terminal illness, cognitive impairment and poor language skills. Patient data from general practitioner practices that had a nurse practitioner employed with CVRM in the portfolio were included in the TICD project. Primary prevention patients had an estimated 10-year combined cardiovascular risk score of 20% or higher for morbidity and mortality due to CVD, based on the SCORE risk-stratification model used in the Dutch College of General Practitioners Practice Guideline for CVRM. Patient selection was based on the following International Classification of Primary Care (ICPC) codes: K74-K76, K85-K92, K99.1 and T93. Secondary prevention patients had a cardiovascular event, predominantly coronary artery disease or less frequently a stroke, in history or suffered from chronic heart disease or peripheral arterial disease.

Sample 2

This sample consisted of solely secondary prevention patients with angina pectoris, or with a history of acute myocardial infarction, other chronic ischemic heart diseases, transient ischemic attack, ischemic stroke, peripheral arterial disease, or aortic aneurysm. Selection of electronic medical records with these conditions was based on corresponding diagnostic ICPC codes K74, K75, K76, K89, K90.3, K92.1 and K99.1. Patients had to be in treatment for an established CVD for a minimum period of 12 months. Baseline data were derived, based on a sample of secondary prevention patients (n = 1866), who participated in a Cluster Randomized Trial on the effectiveness of improvement plans for primary care practice accreditation.

Sample 3

This sample consists of solely primary prevention patients (n = 676) with an estimated cardiovascular mortality risk of 10% or higher in 10 years, based on the SCORE risk-stratification model. If this was not available, a proxy measure was used: the presence of three out of the following four risk factors: hypertension, hypercholesterolemia, smoking and men over 60 years. Exclusion criteria were established diabetes, terminal illness, cognitive impairment, psychiatric illness and poor language skills. The Dutch sample, from the European Practice Assessment of CVRM (EPA-cardio) project, was included in our study.

Measurements and study size

Data were collected using structured questionnaires. The questionnaires, paper-based versions, were sent by mail to all eligible patients and returned by mail as well. Data collection was performed equally in primary and secondary prevention patients. The total study size consisted of 4726 patients.

Variables

The outcome measure in all samples was the degree of physical activity measured by the patient reported Rapid Assessment of Physical Activity Questionnaire (RAPA). The predictor measure is patient-reported age. Because sample 1 consisted of primary and secondary prevention patients, we were able to include the presence of CVD (i.e. being a secondary prevention patient) as a potential effect modifier of the association between aging and physical inactivity. Gender was considered as a potential effect modifier in all samples.

All included patients filled out the RAPA questionnaire. The RAPA is a valid measure of physical activity for use in clinical practice. Each question has a ‘Yes’ or ‘No’ option. The score ranges from 1 to 7; a score of 6–7 points is considered ‘active’, 4–5 points as ‘suboptimal active’, and ≤3 points is defined as ‘sedentary’ to ‘under-active regular-light activities’. We used the RAPA as a dichotomized and continuous variable. Age was used as a categorical and continuous measure.

Statistical methods

Data analysis was performed using SPSS (version 25, IBM Corp.). Data were described using means, standard deviations, numbers and percentages. Patient characteristics and the outcome of completed questionnaires were presented for the samples separately.

First, independent sample T-tests were used to compare the score for the RAPA and age between men and women, and between primary and secondary prevention patients. Second, the relationship between age and RAPA was checked. We worked with the model that gave the highest explained variance. Thereafter, the assumptions for multilevel regression (i.e. the linear relationship between the independent and dependent variables, multivariate normality, no multicollinearity, and homoscedasticity) were checked. Third, we performed a multilevel regression analysis with random slopes to explore the association between age and physical inactivity per sample, adjusted for confounders. Finally, we explored if this association was modified [95% confidence intervals (CIs)] by gender or the presence of CVD by adding the interaction terms gender*age and primary vs. secondary prevention*age to separate models. Correction for confounding was only applied if the regression coefficient for age, changed relatively by more than 10% when the potential confounding variable was added to the model. We considered the score on the Patient Health Questionaire-9, Patient Activation measure and smoking status, as potential confounders in sample 1. In sample 2, this was subsequently the EuroCol 5 D (EQ-5D) score, education level, marital status, employment status and if hospitalized in the last 3 months and in sample 3, the body mass index, using medication, employment status, smoking and suffering pain. The data were analyzed per sample because the inclusion and exclusion criteria and the measured confounders differed from each other per sample. A P-values <0.05 was considered as statistically significant.

Results

Participants and descriptive data

Figure 1 presents a flowchart of the enrolment of the study population. Table 1 presents descriptive statistics. Figure 2 presents the activity levels (RAPA) for the age-categories; younger than 65, 65–75 years and older than 75 years, for primary and secondary prevention men and women. Of all patients that were included in this study, 56% of the population scored ‘active’ on the RAPA (6–7) in the age-category 65–75 and 44% of the older population (>75 years).

Figure 1

Flowchart for enrolment of study population RAPA: Rapid Assessment of Physical Activity Questionnaire

Table 1

Descriptive statistics per baseline sample

Samples,	Women				Men
N	N, Mean age (±SD)	N, % of study population, RAPA mean (±SD)	RAPA (N, %)		N, Mean age (±SD)	N, % of study population, RAPA mean (±SD)	RAPA (N, %)
			Under-active	Active			Under-active	Active
1, N = 2184
PP, N = 1250	407, 75.74 (7.09)	377 (17.26%), 4.40 (1.62)	226 (59.9%)	151 (40.1%)	843, 73.15 (6.79)	807 (36.95%), 5.01 (1.54)	377 (46.7%)	430 (53.3%)
SP. N = 934	354, 68.86 (13.10)	332 (15.20%), 4.61 (1.55)	195 (58.7%)	137 (41.3%)	580, 68.3 (3.11)	558 (25.55%), 5.02 (1.58)	259 (46.1%)	299 (53.6%)
2, SP, N = 1866	629, 67.19 (11.69)	570 (30.54%), 4.86 (1.55)	283 (49.6%)	287 (50.4%)	1227, 72.11 (10.36)	1176 (63.02%), 5.21 (1.52)	488 (41.5%)	688 (58.5%)
3, PP, N = 676	83, 70.20 (9.10)	79 (14.54%), 4.77 (1.64)	44 (55.7%)	35 (44.3%)	575, 71.29 (6.64)	557 (82.25%), 5.06 (1.58)	253 (45.4%)	304 (54.6%)

Notes: RAPA, Rapid Assessment of Physical Activity Questionnaire, 1–7 points. Under-active, RAPA 1–5 points. Active, RAPA 6–7 points. PP, Primary prevention patients; SP, Secondary prevention patients.

Figure 2

Activity levels per age-category in primary and secondary prevention men and women per sample. SP: secondary prevention, PP: primary; RAPA: Rapid Assessment of Physical Activity Questionnaire. *P = < 0.02

Sample 1 consisted of 2184 patients (1250 primary and 934 secondary prevention patients), of which 761 were female patients and 1423 were male patients. In total were 2074 patients eligible for analysis. The primary prevention population consisted of 33% (mean age 75.7 ± 7.0) women and the secondary prevention group of 38% women (mean age 68.8 ± 13.1). Primary prevention patients were significantly older than secondary prevention patients (73.99 ± 6.99 vs. 68.50 ± 11.73, P < 0.05). Female primary prevention patients scored significantly (P < 0.01) lower (4.40 ± 1.62) on the RAPA than female secondary prevention patients (4.61 ± 1.55). Men in both groups were physically more active compared to women (primary prevention group 5.01 ± 1.54 vs. 4.4 ± 1.62, P < 0.01 and secondary prevention group 5.02 ± 1.58 vs. 4.61 ± 1.55, P < 0.01).

Sample 2 consisted of 1866 secondary prevention patients, 629 (33.6%) women aged 67.19 ± 11.69 and 1227 (64.4%) men aged 72.11 ± 10.36. In total 1734 patients were eligible for analysis. Men were significantly older than women (P < 0.05). Female secondary prevention patients were significantly (P < 0.01) less active than men in this group (4.86 ± 1.55 vs. 5.21 ± 1.52).

Sample 3 consisted of 676 primary prevention patients, from which 636 patients were eligible for analysis. There were 83 (12.3%) women aged 70.20 ± 9.10 and 575 (87.7%) men aged 71.29 ± 6.64. No significant age difference was found between men and women.

Outcome data

The assumptions for multilevel analyses were not violated. The quadratic model was significantly (P < 0.001) the best fitting model for the relationship between age and RAPA in sample 1 (R2 = 0.040) and sample 2 (R2 = 0.017). The R2 was subsequently 0.027 in sample 1 and 0.011 in sample 2 in the linear model. There was no other model with a significantly better fit in sample 3 than the linear model (R2 = 0.006). Multilevel regression analyses revealed a significant (P < 0.05) association between aging and physical inactivity (table 2, model 1). This association was not modified by gender (table 2, model 2), nor by being a primary or secondary prevention patient (table 2, model 3).

Table 2

Multilevel regression presenting if the association between aging and physical activity is modified by gender or the presence of a CVD

		Model 1a				Model 2a				Model 3a
RAPA 1-7			95% CI				95% CI				95% CI
		B	Lower	Upper	P	B	Lower	Upper	P	B	Lower	Upper	P
Sample 1
Linear model	Age (Y)	–0.027b	–0.034	–0.019	<0.001	–0.037	–0.060	–0.013	0.002	–0.040	–0.067	–0.013	0.004
	Gender (F/M)	0.332	0.184	0.480	<0.001	–0.113	–1.133	0.907	0.828	0.327	0.178	0.476	0.000
	Patient group (SP/PP)	–0.009	–0.153	0.136	0.906	–0.008	–0.153	0.137	0.912	–0.551	–1.678	0.576	0.338
	Gender*Age					0.006	–0.008	0.002	0.387
	Patient group*Age									0.008	–0.008	0.023	0.397
Quadratic model	Age (Y)	0.055	–0.007	0.116	0.082	0.046	–0.018	0.109	0.158	0.050	–0.027	0.126	0.202
	Age*Age	–0.001c	–0.001	–0.000	<0.009	–0.001	–0.001	–0.000	0.006	–0.001	–0.001	0.000	0.014
	Gender (F/M)	0.311	0.162	0.460	<0.001	–0.310	–1.338	0.718	0.555	0.310	0.162	0.459	0.000
	Patient group (SP/PP)	0.019	–0.127	0.164	0.803	0.021	–0.125	0.167	0.779	–0.110	–1.290	1.069	0.854
	Gender * Age					0.009	–0.006	0.023	0.231
	Patient group*Age									0.002	–0.014	0.018	0.829
Sample 2
Linear model	Age (Y)	–0.015	–0.024	–0.006	0.001	–0.027	–0.054	–0.000	0.049
	Gender (F/M)	0.244	0.077	0.411	0.004	–0.225	–1.274	0.825	0.675
	Gender*Age					0.007	–0.009	0.023	0.375
Quadratic model	Age (Y)	0.093	0.018	0.168	0.015	0.082	0.003	0.160	0.041
	Age*Age	–0.001	–0.001	–0.000	0.004	–0.001	–0.001	–0.000	0.004
	Gender (F/M)	0.227	0.060	0.394	0.008	–0.298	–1.347	0.750	0.577
	Gender * Age					0.008	–0.008	0.024	0.319
Sample 3
Linear model	Age	–0.021	–0.040	–0.040	0.026	–0.012	–0.052	0.028	0.552
	Gender	0.068	–0.076	0.212	0.352	0.418	–0.915	1.752	0.538
	Gender*Age					–0.005	–0.024	0.014	0.604

Notes: Model 1 exploration of the association between age and physical inactivity per sample, adjusted for confounding. Model 2, the interaction term ‘gender*age’ was added. Model 3, the interaction term ‘primary vs. secondary prevention*age’ was added.

PP, primary prevention; SP, secondary prevention; Y, Years; RAPA, Rapid Assessment of Physical Activity; F, Female; M, Male; Y, years.

Confounders; sample 1: Patient Health Questionaire-9 (0–27), Smoking status (Y/N), Patient Activation Measure (0–13). Sample 2: Patient Experiences Questionnaire (1–5), education level (low-med vs. high), married (Y/N), employed (N/Y), hospitalized the last 3 months (N/Y). Sample 3: BMI, medication use (Y/N), employed (Y/N), Smoking (Y/N) suffering pain (Y/N).

In the linear model, the magnitude of B indicated that a 10-year age increase was associated with 0.27 points decline on the RAPA.

In the quadratic model, a negative B means an acceleration of the decrease in physical activity levels as the patient ages over 70 Years.

Discussion

Key results

The purpose of this study was to explore if the association between aging and physical inactivity was different between men and women and between primary and secondary prevention patients, using data from three cross-sectional primary care samples. We found that aging was consistently associated with physical inactivity in three out of three samples. This association was, however, not modified by gender or the presence of CVD.

Interpretation

Gender and the presence of CVD are not useful predictors for declining physical activity with increasing age in a primary care CVRM population. To our knowledge, this is the first study that investigated the possible differences between the association of aging and physical inactivity in women and men and primary and secondary prevention patients.

In our population only 56% of the patients aged 65–75 years and 44% of the patients aged 75 years and older met recommended activity levels (RAPA 6-7) This is still higher than the earlier reported 33% of the general population between 65 and 75 years and 25% of the patients aged over 75 years that met recommended physical activity levels of at least 150 min/week of moderate intensity activity. Although our population, consisting of primary and secondary prevention patients, was physically more active than people from the general population, it is still of critical importance to stimulate the adoption of an active lifestyle. Women and men aged over 75 who exercised had, for example, similar death rates than women and men aged 65–75 who did not exercise.

In our study, the age differences were inconsistently different between the men and the women, the average age was only higher for women in the primary prevention group in sample 1. Therefore, the difference in physical activity levels that we found cannot be explained by the fact that women experience a CVD 7–10 years later than men, and consequently suffer more frequently from comorbidities, resulting in lower activity levels. We expected that the presence of CVD would modify the association between aging and physical activity because aging secondary prevention patients experience a higher prevalence of frailty, increased levels of depressive symptomatology, lower self-efficacy and self-management levels and increased levels of comorbidity compared to primary prevention patients.,, On the other hand, secondary prevention patients might be more active due to having participated in cardiac rehabilitation programs aiming at a physically active lifestyle in the past. Despite a substantial part of our secondary prevention population suffers from a coronary artery disease, it is unknown if and how long-ago they participated in cardiac rehabilitation. However, other researchers concluded that patients who completed cardiac rehabilitation do not always maintain an active lifestyle in the long term. Finally, our findings do not support those of a review suggesting that aging was not associated with physical inactivity either during center-based cardiac rehabilitation or home-based cardiac rehabilitation, after cardiac rehabilitation completion or in patients not attending cardiac rehabilitation. This could be explained by the fact that we included a primary care CVRM population, that differed from the cardiac rehabilitation population and because we included older patients, who might suffer more comorbidity and therefore are less active.

Limitations and strengths

The inclusion and exclusion criteria for primary and secondary prevention patients were not completely the same in all samples. In sample 1 and 3 for instance, patients with diabetes were excluded, while in sample 2 patients with diabetes were included. In sample 1, primary prevention patients were included based on a 10-year combined cardiovascular risk score of 20% or higher for mortality or morbidity due to CVD, whereas this was respectively 10% or higher for CVD mortality in sample 3. Despite, the difference in in- and exclusion criteria between samples, similar study results were found. Because different patients characteristics (social status, education levels) were measured and different questionnaires (assessing depression levels, smoking status, etc.) were used in each sample, we were not able to adjust for the same confounders. However, by correcting only for potential confounders that changed the regression coefficient for age by more than 10%, we were able to use a uniform statistical method in the same way in all samples. Nevertheless, there may have been other confounders for which we have not adjusted in the analysis, such as personal (scheduling, motivation), social (family support) and environmental (cost, access) barriers.

Possibly, predominantly higher quality practices may have participated in our study. In our included practices in sample 1, a nurse practitioner was employed who had CVRM in her/his portfolio. Therefore, CVRM may have been better implemented in these practices than in other Dutch practices. This nurse practitioner could, for instance, be educated in motivational techniques to enhance patients’ activity levels that aligned to the interest and motives in primary care CVRM women and men.,, Also, general practices in all samples were only included if they were willing to participate in scientific research, and, in addition, the practices included in sample 2 were participating in a practice accreditation program. In sample 134 general practices participated, while 1600 practices were invited. In sample 2336 practices were invited to participate in the study. Eventually, 45 practices were willing to participate and were included in that study. In sample 31 190 practices were invited, while 36 eventually participated.

The selection criteria for patients being included were standardized in all samples, using ICPC diagnoses codes and internationally used cardiovascular risk tables. This is a strength of our study as is the use of the validated RAPA questionnaire. The statistical power was high in our study due to the inclusion of a high number of patients in the samples, showing consistent study results. Finally, the cross-sectional design used in our study lacks the ability to draw causal relationships. For instance, mortality rates are higher in secondary prevention patients than in primary prevention patients and the patients alive may represent a population that have always been physically more active compared with ones who have died.

Generalizability

Nearly all Dutch citizens are registered in a general practice. The general practitioner has a medical record of (almost) all patients, and therefore every patient who is eligible for CVRM within the participating practices is in the original selection from which the samples have been taken. Also, the distribution of one-third female and two-third male patients in sample 1 and 2, is representative of the CVRM population in western primary healthcare facilities. In sample 3, this was 1 : 8. This is less representative of the Dutch population. Only 4% of the patients did not have the Dutch nationality, and therefore the generalizability to people originating from none western countries might be limited. Finally, it must be acknowledged that predominantly general practices from outside the urban area participated, therefore with respect to the urban population, this might hamper the generalization.

Conclusions

CVRM interventions should aim at increasing or maintaining physical activity levels in aging primary care patients. This study does not support the notion that extra emphasis should be put in targeting men or women, or people with or without a history of CVD. Longitudinal research should be performed to re-confirm the present findings. Also, future research should focus on the effect of lifestyle interventions aiming at the increase of physical activity in the aging CVRM population, despite gender difference and the presence of CVD.

Funding

None declared.

Conflicts of interest: None declared.

Aging was associated with physical inactivity in three out of three samples consisting of primary and/or secondary prevention patients that participated in primary care cardiovascular risk management.

The association between aging and physical inactivity was neither different for men and women, nor for patients at high cardiovascular risk with or without the presence of cardiovascular disease.

This study does not support the notion that gender and the presence of cardiovascular disease are useful predictors for declining physical activity with increasing age in a primary care cardiovascular risk management population.