Analysis of multi-level barriers to physical activity among nursing students using regularized regression

Muge Capan; Lily Bigelow; Yukti Kathuria; Amanda Paluch; Joohyun Chung

doi:10.1371/journal.pone.0304214

Abstract

Physical inactivity is a growing societal concern with significant impact on public health. Identifying barriers to engaging in physical activity (PA) is a critical step to recognize populations who disproportionately experience these barriers. Understanding barriers to PA holds significant importance within patient-facing healthcare professions like nursing. While determinants of PA have been widely studied, connecting individual and social factors to barriers to PA remains an understudied area among nurses. The objectives of this study are to categorize and model factors related to barriers to PA using the National Institute on Minority Health and Health Disparities (NIMHD) Research Framework. The study population includes nursing students at the study institution (N = 163). Methods include a scoring system to quantify the barriers to PA, and regularized regression models that predict this score. Key findings identify intrinsic motivation, social and emotional support, education, and the use of health technologies for tracking and decision-making purposes as significant predictors. Results can help identify future nursing workforce populations at risk of experiencing barriers to PA. Encouraging the development and employment of health-informatics solutions for monitoring, data sharing, and communication is critical to prevent barriers to PA before they become a powerful hindrance to engaging in PA.

Citation: Capan M, Bigelow L, Kathuria Y, Paluch A, Chung J (2024) Analysis of multi-level barriers to physical activity among nursing students using regularized regression. PLoS ONE 19(5): e0304214. https://doi.org/10.1371/journal.pone.0304214

Editor: Hao Wang, JPS Health Network, UNITED STATES

Received: January 25, 2024; Accepted: May 8, 2024; Published: May 24, 2024

Copyright: © 2024 Capan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: De-identified data are uploaded as supplemental material.

Funding: University of Massachusetts Amherst -Elaine Marieb Center for Nursing and Engineering Innovation Faculty Pilot Grant supporting the study entitled "Exploratory Study to Identify Multi-level Factors of Physical Health and Well-being in Nursing Profession" (06/2023 – 06/2024, PI: M. Capan).

Competing interests: The authors have declared that no competing interests exist.

Introduction

Physical inactivity is a growing societal concern with significant impact on disability, burden of non-communicable diseases, mortality, and morbidity [1–3]. National guidelines in the U.S. recommend adults to engage in 150 minutes per week of moderate-intensity or 75 minutes per week of vigorous-intensity aerobic physical activity (PA) and resistance training twice per week [4]. Nurses have a high prevalence of non-adherence to PA recommendations [2]. Low levels of PA are particularly concerning in healthcare professions in direct contact with patients, families, and communities, such as nurses. Nurses represent one of the largest healthcare professions in the U.S. with nearly 4.2 million registered nurses [5]. When nurses’ health and well-being are compromised due to physical inactivity, it can negatively impact safety and quality of care, e.g., increased risk of clinical errors [6].

Understanding the factors acting as barriers to engage in PA in nursing population can improve nurses’ PA. While determinants of PA have been widely studied, the question of how multi-level factors (e.g., individual, intrapersonal, environmental) result in disparities such that some populations experience more barriers to PA than others remains an unresolved question. Existing studies are limited due to low survey response rates from nurses in the workplace [7], and the lack of a comprehensive framework to connect the multi-level factors hindering PA.

The objectives of this study are to categorize multi-level factors associated with barriers to engage in PA according to selected domains and influence levels of NIMHD Research Framework, and analyze the impact of these factors on barriers to PA among nursing students. Specifically, the study questions (Q) explored in this paper are:

Q1: How can we quantify self-reported barriers to engage in PA using a scoring system?
Q2: How can we use individual, interpersonal, behavioral and socio-cultural factors to predict this score in nursing student population?

Below we review relevant studies and their findings regarding: i) barriers to PA in nursing, ii) connecting health disparities and barriers to PA in nursing, and iii) a conceptual framework to categorize and quantify barriers to PA.

Barriers to physical activity in nursing

Nurses self-report levels of moderate- to vigorous-intensity PA below the recommended levels for adults [8–11]. In one study, 60–74% of nurses were classified as being at high risk for poor health outcomes due to low PA levels [2], and 50% of nurses did not meet guidelines for PA [12]. Barriers to PA in the nursing population have been categorized in various ways, e.g., internal (e.g., attitudes reflecting negative attitudes towards PA) or external (e.g., physical or environmental impediments that prevent PA, lack of facilities) [13, 14]. Nurses have largely cited cost, lack of accessible and affordable exercise spaces, fear of inducing adverse health conditions (e.g., injury, fatigue, muscle soreness), physical appearance, lack of social support, lack of time, and stress as barriers to PA [11, 15, 16].

A systematic review identified several barriers to PA in nursing work environments [17]. Some examples include social barriers, such as lack of access to exercise facilities [18], organizational barriers such as long work hours, and individual barriers such as lack of motivation and self-efficacy [7]. Several studies highlighted occupational factors related to physical inactivity in nursing. Nursing profession is reported to be physically demanding [19], and many nurses experience job dissatisfaction, burnout [20], and mental health issues [11]. Nurses affected by these factors are put at risk for poor health behaviors including physical inactivity [11, 19].

Identifying the barriers to PA is particularly important for individuals preparing to enter the nursing workforce. Nursing students are also cited to be physically inactive [21, 22]. In a study of 264 nursing students, 68.6% of the students were not meeting the recommended amounts of PA [23]. A study conducted focus groups with Australian nursing students and identified individual (e.g., lack of motivation, lack of knowledge), environmental (e.g., time, finances, limited access to PA resources), and psychosocial (e.g., competing priorities, increased cognitive load, lack of social interaction and support) barriers to PA [22].

Health disparities and barriers to physical activity

Health disparities represent differences in the incidence and prevalence of health conditions between groups. PA is influenced by health disparities, including but not limited to differences in socioeconomic status, race and ethnicity, and access to PA facilities. Disparities may occur due to some populations disproportionately experiencing barriers to PA. For example, cost- or distance-related factors limiting PA may impact socially vulnerable, minority, and/or rural populations more compared to other populations. In the context of barriers to PA, minority ethnic or racial groups, older individuals, women, low income individuals, and less educated individuals are less likely to engage in sufficient PA [24]. Furthermore, disparities such as rurality and lack of access to exercise facilities have been cited as environmental barriers to PA [24]. Health disparities and barriers to PA are intertwined, and as such, it is necessary to consider both when identifying multi-level factors that influence barriers to PA.

The National Institute on Minority Health and Health Disparities (NIMHD) Research Framework

The National Institute on Minority Health and Health Disparities (NIMHD) framework is a conceptualization of factors impacting minority health and health disparities [25]. The framework identifies five domains of influence (i.e., biological, behavioral, physical/built environment, socio-cultural, and healthcare system) and four levels of influence (i.e., individual, interpersonal, community, and societal) (Fig 1).

Download:

Fig 1. National Institute on Minority Health and Health Disparities (2017) NIMHD Research Framework.

Retrieved from https://nimhd.nih.gov/researchFramework and reproduced with permission from the National Institute on Minority and Health Disparities, National Institute of Health.

https://doi.org/10.1371/journal.pone.0304214.g001

Previous research has utilized various psychosocial models to explain PA engagement, including the Socioecological model, the Theory of Planned Behavior, and the Health Belief Model [26]. Yet, to the best of our knowledge, there are no studies that utilized the NIMHD framework to connect the multi-level factors associated with experiencing barriers to PA.

The remainder of the paper is organized as follows. In Materials and Methods section, we describe the data pre-processing, descriptive, statistical and predictive methods adopted in our research. Specifically, we developed a scoring system based on validated survey tools to quantify barriers to PA, and developed and validated regularized regression models that aim to predict this score. Next, we present the modeling results. Finally, in Discussion section we conclude and discuss methodological and practical contributions, limitations, and future research directions to address disparities in experiencing barriers to PA.

Materials and methods

Study design

This is a cross-sectional study with a set of questionnaire surveys conducted at the University of Massachusetts Amherst, Amherst MA, between September and October 2023. The study was approved by the Institutional Review Board (IRB) of University of Massachusetts Amherst. The recruitment period for this study started on September 15, 2023 and ended on October 31, 2023. Participants provided written informed consent which was documented in Research Electronic Data Capture (REDCap).

Study population

Inclusion criterion of this study is being currently enrolled in a nursing degree program at the study institution. The nursing degree program includes the Bachelors (BS) in Nursing (4-year undergraduate program) and Accelerated BS Nursing program (18-months program for individuals who have completed another BS degree and are pursuing a nursing BS degree). Exclusion criterion is not consenting to participate.

Survey development

Study survey was designed in REDCap using validated questionnaires on demographics, Social Determinants of Health Collection of the PhenX Toolkit, health behaviors (e.g., sleep, diet), physical activity (e.g., type, frequency, duration, and intensity), psychological health (e.g., anxiety, depression using the Patient Health Questionnaire−9), and access to and use of healthcare technology, among other factors. REDCap is a secure web application for building and managing online surveys and databases [27, 28]. The validated survey tools (Table 1) corresponding to domains and levels of influence from the NIMHD Research Framework (Fig 1) are listed below:

International Physical Activity Questionnaire (IPAQ) [29]
Motives for Physical Activity Measure (MPAM-R) [30]
Exercise Benefits/Barriers Scale (EBBS) [31]
PhenX Social Determinants of Health (SDOH) Toolkit [32]
Physical Activity and Social Support Scale (PASSS) [33]

Download:

Table 1. Adapted National Institute on Minority Health and Health Disparities (NIMHD) Framework and corresponding survey tools.

IPAQ stands for International Physical Activity Questionnaire. MPAM-R stands for Motives for Physical Activity Measure—Revised. EBBS stands for Exercise Benefits/Barriers Scale. PASSS stands for hysical Activity and Social Support Scale. SDOH stands for Social Determinants of Health.

https://doi.org/10.1371/journal.pone.0304214.t001

In total, there were 174 questions in multiple-choice, yes/no, and numeric entry formats. The IPAQ includes 27 questions across 4 domains of PA per week (i.e., job, transportation, housework, and leisure. Previous studies have established test-retest reliability for this measure [29, 34]. The MPAM-R is an instrument with 30 questions on a scale of 1–7 that are used to assess the reasons that motivate individuals in the pursuit of physical activities. The scale is based along 5 themes: fitness, appearance, competence/challenge, social and enjoyment and these motives are connected to different outcomes. Reliability and construct validity for the MPAM-R have been previously established [30, 35, 36]. The EBBS has two parts: the 29-item benefits scale and the 14-item barriers scale that measure the perception of individuals towards the benefits and barriers of exercise. The scale ranges from 1–4. Good validity and reliability for this measure has been previously established [31, 37]. The PhenX SDOH Toolkit is comprised of a core and additional specialized collection of questions with individual and structural measurement protocols. The core collection used in this study includes demographic data such as race, age, gender identity, annual family income, access to health services, among other SDOH factors. Protocols from the PhenX SDOH Toolkit are well-established and have demonstrated reliability, reproducibility, and validity [38, 39]. The PASSS consists of 20 items across 5 domains (i.e., emotional support, validation support, informational support, companionship support, instrumental support) and a scale of 1–7 that examines the perception of various support mechanisms related to engaging in PA. The PASSS has been shown to have criterion validity and internal consistency [33, 40].

Each questionnaire described above was matched to the intersection of the two selected domains (behavioral and sociocultural environment) and two selected levels of influence (individual and interpersonal) within the NIMHD framework as shown in Table 1. This matching is a critical step in predictive model development for the prediction of the outcome of interest in our study, which is the Physical Activity Barrier Score (PABS). We operationalized PABS by summing the responses to 14 questions from the EBBS that are designated to measure barriers to exercise. All 14 questions are statements on a Likert-scale on a scale from 1 (never true) to 7 (always true). We developed regression models for each domain-level intersection in Table 1 using participants’ responses to the corresponding questionnaires as independent variables to predict PABS. The model development and evaluation will be further discussed in Methods section. The final version of the REDCap survey is provided in Supplementary Materials (S1 File).

Data collection

Data was collected in four different in-person sessions during September—October 2023. Each session was coordinated and conducted by at least two of the study team members. During each data collection session, participants were informed about human subjects research and IRB. The participants were provided the consent form. Access to the REDCap survey designed for computer, tablet, and phone. Participation took on average 25–35 minutes to complete. To begin the survey, participants clicked on the survey link and/or QR code shown on the screen. They were able to save and continue where they left off if needed. All participants were asked to complete the survey by October 31, 2023. All participants who completed the survey received a $20 gift card for their participation.

Data preprocessing

Informed consent was obtained from all study participants. R statistical software (v4.3.1) was used for data cleaning and preprocessing [41]. One participants observations were excluded for being above possible PA levels achievable in a week [29]. Approximately 14.83% of all data contained missing values, which were accounted for using various imputation methods. Median imputation was used for variables that contained values from a Likert Scale (i.e., variables from the MPAM-R, EBBS, and PASSS). Domain-specific imputation was used for the remaining numerical and categorical variables that contained a majority of missing values, including survey questions that contained branching logic. Missing values for independent variables are discussed in the Supplementary Materials (S1 Table).

Little’s Test [42] was used to determine the state of missingness for our data. Missing data is classified in three ways: Missing Completely at Random (MCAR), Missing at Random (MAR), and Missing Not at Random (MNAR). MCAR is where missing data values are independent of study variables or unobserved data, MAR is missing data values are dependent on study variables, and MNAR is missing data values are dependent on study variables and unobserved data [43, 44]. Under the null hypothesis that missing data is MCAR, Little’s Test calculates Chi-squared statistics based on the differences between observed and expected missingness patterns. Data is MCAR if results are non-significant, failing to reject the null hypothesis. For this study, Little’s Test was performed on variables belonging to each validated tool (e.g., IPAQ, MPAM-R, EBBS, PASSS, PhenX SDOH Toolkit), using R statistical software (v.4.3.1) and package naniar (v.1.0.0) [45] using mcar_test().

Following imputation, categorical independent variables were factorized and variables containing a large number of categories were re-categorized as needed. Factorization refers to the process of grouping distinct values of a categorical variable into discrete levels, allowing qualitative data to be represented numerically. Some of the independent variables were directly derived from survey responses, others were computed using the survey responses. From the IPAQ, four variables describing domains of PA measured in metabolic equivalent (MET)-minutes/week (work, transportation, yard/domestic, and leisure) were calculated according to scoring guidelines [29]. From the MPAM-R, five variables describing types of motives (i.e., fitness, appearance, competence, social, and interest/enjoyment) were calculated by summing scores for questions related to each type of motive [30]. From the PASSS, five variables describing forms of social support (i.e., emotional, informational, instrumental, validation, and companionship) were calculated by summing scores for questions related to each social support [33].

Descriptive and statistical analysis

In addition to questions from survey tools (Table 1), demographic factors consisting of race, age, family income, birthplace, biological sex, level of education, and home language were collected. Summary statistics and visualization of distributions of numerical variables were conducted. Non-parametric statistical tests were used to test whether there was a significant difference regarding PABS between the categories of categorical factors. Age and family income were categorized using median and quantiles of the response distributions prior to testing. The Mann Whitney U Test applies to features that only have two groups whereas the Kruskal Wallis Test is a generalization of the Mann Whitney U Test and is used to test difference with regards to PABS between levels of categorical independent variables with more than two categories.

Predictive analysis

R statistical software (v4.3.1) was used to develop all predictive models and perform all analyses [41]. Outcome of interest was the numerical variable PABS. Predictive models were developed by first performing stepwise regression, and then implementing regularized regression with k-fold cross validation. Stepwise regression is the iterative selection of variables from a model based on the predictors’ level of association with the outcome variable and statistical significance (i.e., contribution to the coefficient of determination, R², or the ability to minimize the sum of squared error) [46, 47]. There are three variable selection approaches that can be used: the iterative addition (forward selection), iterative removal (backward selection), or simultaneous addition and removal of variables (bidirectional elimination). For this study, bidirectional elimination was chosen as the selection approach and was implemented with R package MASS (v7.3.60), using the stepAIC() function [47, 48].

Regularized regression with k-fold cross validation was applied to the best fit models resulting from the stepwise regression. Regularized regression incorporates constraints to minimize the number of insignificant independent variables in a regression model, and includes Ridge, LASSO, and Elastic net regression approaches [49]. Comparatively, ordinary least squares (OLS) regression attempts to minimize the residual sums of squares (RSS) between observed and predicted values [50] and is given by the equation: (1)

In (1), output variables are given by y_i, input variables by x_ij, regression coefficients β_j, and intercept β₀. Regularized regression works similarly in that it attempts to minimize RSS, yet it also applies a penalty term to the objective function that shrinks coefficients of predictors to zero. Ridge regression [51] is expressed as follows: (2)

In (2), λ ≥ 0 is the tuning parameter of the L₂ penalty term, which serves to control the shrinkage of β_j towards zero, but not exactly to zero [49, 50]. LASSO regression [52] uses an L₁ penalty term, which can perform feature selection of variables due to its ability to shrink coefficients exactly to zero [50]. It is expressed as follows: (3)

Elastic net regression [53] includes features of both L₁ and L₂ penalty terms and is expressed as follows: (4)

K-fold cross validation was used to verify predictive accuracy of the regularized models [49]. Root mean squared error (RMSE) was calculated to quantify the test error. The cross validation procedure can be expressed as follows: (5)

For each regularized model that was developed, the optimal model type (Ridge, LASSO, Elastic net) and optimal λ values were selected using a 10-fold cross validation that minimized the RMSE of predictions. Model types are represented by value α, where α = 0 represents Ridge, α = 1 represents LASSO, and 0 < α < 1 represents Elastic net. The optimal λ, which represents regularization strength, determines the overall complexity of each model. R packages glmnet (v4.1.8) [54] and caret (v6.0.94) [55] and bestTune() were used to perform regularized regression and cross validation of models.

Results

Descriptive and statistical analysis results

Table 2 provides a description of the distribution of the study population characteristics. The study participants comprised primarily of female nursing students (85.3%) with a majority of white race group (61.3%). Year of study was spread across multiple categories ranging from freshman to senior and other. Majority of students were born within the United States (90%) and spoke only English at home (70%). For our study, calculated Cronbach’s alpha associated with the IPAQ is 0.66, with the MPAM-R is 0.95, with the EBBS is 0.82, and with the PASSS is 0.92.

Download:

Table 2. Study population characteristics.

https://doi.org/10.1371/journal.pone.0304214.t002

To quantify the outcome of interest, PABS, for each participant, responses from each question in the EBBS related to barriers was summed. These responses are on a Likert scale, ranging from 1 to 4. Resulting PABS ranges from 14 to 49 with a median value of 28 and mean value of 26.5 and a standard deviation of 5.6 in the overall study population. Lower PABS indicates that an individual experiences less barriers, while a higher PABS indicates that the individual experiences more barriers.

Results show that PABS varies between different groups in study population. For example, the participants with a completed Bachelors/Masters or other graduate degree experience more barriers to engage in PA (PABS mean = 27.9, 95% Confidence Interval (CI): [24.89, 28.63]) than those with a high school diploma/GED (PABS mean = 25.3, 95% CI: [23.07, 25.95]). Furthermore, participants who are born in the US experience less barriers to PA (PABS mean = 26.1, 95% CI: [24.1, 26.4]) than those born outside of the US (PABS mean = 29.6, 95% CI: [23.1, 32.4]). Participants who only speak English at home experience less barriers to PA (PABS mean = 26.2, 95% CI: [24.5, 27.0]) than those who speak a language other than English at home (PABS mean = 27.1, 95% CI: [23.0, 27.8]).

Little’s Test for missingness [42] was performed for variables in each validated survey tool and p-values were generated for each test under a 95% confidence interval. For our study, the p-value associated with IPAQ variables is 0.000510, with the MPAM-R is 0.992, with the EBBS is 0.228, with the PASSS is 0.772, with Health data and sharing variables (PhenX SDOH Toolkit) is 0.957, and with demographic variables (PhenX SDOH Toolkit) is 0. Variables belonging to the MPAM-R, EBBS, PASSS, and Health data and sharing tools have non-significant p-values, indicating that this data is MCAR.

Statistical tests were used for testing the differences between the categories of population characteristics as shown in Table 2. Normality assumption was tested via the Shapiro-Wilk test and homogeneity of variances assumption was tested via Levene’s test. While homogeneity of variance was met, normality assumption was violated, warranting the use of non-parametric tests. The Mann-Whitney Test yielded a p-value of 0.113 for biological sex, 0.0045 for age category, 0.0044 for level of education, 0.0088 for participant birthplace, and 0.129 for home language. The Kruskal-Wallis test resulted in p-values of 0.055 for race and 0.093 for year of study. Higher age, higher education level, lower family income, birth place outside of U.S. and home language other than English were statistically significantly associated with higher PABS indicating individuals experiencing more barriers to engage in PA.

Predictive analysis results

For each intersection of NIMHD domain-level of influence combination shown in Table 1, stepwise regression, regularized regression, and k-fold cross validation were performed. After model comparison and validation, four final models were selected to determine the optimal α and λ values that minimized RMSE. Regression coefficients to be included in each final model were selected at the intersection of regression coefficient paths with the optimal λ value (Fig 2). For each model, the outcome variable of interest, y, is defined as the individual-level PABS. Independent variables, given by x_i, are summarized in Table 3. Resulting final model parameters and performance metrics are summarized in Table 4.

Download:

Fig 2. Plots of regression coefficients vs. log(λ) values for each domain-level of influence within the adapted NIMHD Framework.

As log(λ) increases, regression coefficients are shrunk towards 0. Each model’s regression coefficients are selected at the intersections with optimal log(λ) values, shown by red dashed lines. (a) For Individual-Behavioral influences, two coefficients are selected for at log(0.41). (b) For Interpersonal-Behavioral influences, three coefficients are selected for at log(0.28). (c) For Individual-Sociocultural influences, three coefficients are selected for at log(0.26). (d) For Interpersonal-Sociocultural influences, five coefficients are selected for at log(0.16).

https://doi.org/10.1371/journal.pone.0304214.g002

Download:

Table 3. Independent variables used in final models.

https://doi.org/10.1371/journal.pone.0304214.t003

Download:

Table 4. Regression statistics for final models.

https://doi.org/10.1371/journal.pone.0304214.t004

Model 1: Individual-behavioral domain-level of influence intersection.

Model 1 is an Elastic Net with α = 0.55 and λ = 0.41 (Fig 2A). Two independent variables were retained in the final model as shown in Eq (6). (6)

Variable x₁ represents interest as a motivator for PA and takes on values from 7 to 49, with higher values indicating that interest is a greater motivator for PA. Variable x₂ represents total leisure-time PA in MET-minutes/week, and takes on values in . The intercept term is 32.07. Interpreting the coefficients, x₁ is negatively associated with outcome variable PABS: for every unit increase in interest as a motivator, PABS decreases by 0.15. Similarly, x₂ is negatively associated with PABS, and for every unit increase in leisure-time PA, PABS decreases by 3.00 ⋅ 10⁻⁴. Variable x₁ is notably more influential than x₂, where its coefficient β₁ is approximately 3 orders of magnitude higher than β₂. Model 1 explained 22% of the variance in the PABS outcome (R² = 0.22) and had an RMSE value of 5.07.

Model 2: Interpersonal-behavioral domain-level of influence intersection.

Model 2 is an Elastic Net with α = 0.10 and λ = 0.28 (Fig 2B). The final model retains three independent variables, all representing different forms of social support as shown in Eq (7). (7)

The independent variables with higher values indicate that an individual experiences more social support. The intercept of the model is 29.17. Variables x₃ and x₄ are negatively associated with PABS, where experiencing one unit higher emotional support (e.g., advice, positive messages, encouragement) and instrumental social support (e.g., having appropriate gear, financial support) is associated with PABS decreasing by 0.21 and 0.18, respectively. Interestingly, x₅ is positively associated with the PABS outcome, increasing by 0.26 for every unit greater validation support (e.g., seeking out others for social comparison, relative status in a group). This suggests that receiving more validation support increases the real or perceived barriers an individual experiences. Model 2 explained 21% of the variance in the PABS outcome (R² = 0.21) and had an RMSE of 5.07.

Model 3: Individual-sociocultural domain-level of influence intersection.

Model 3 is an Elastic Net with α = 0.10 and λ = 0.26 (Fig 2C). The final model retains 3 independent variables as shown in Eq (8). (8)

Variable x₆ represents income and takes on values in . Variable x₇ represents having attained a bachelor’s degree or higher and x₈ represents having a male biological sex. Both x₇ and x₈ take on binary values of 0 or 1. The intercept is given as 26.10. For every $10000 greater in family income, PABS was 0.0575 less. In contrast, if x₇ and x₈ take on values of 1, PABS increases by 1.94 and 2.46, respectively. These results suggest that having higher levels of education and being of male biological sex are associated with higher barriers to PA, while having a higher income is associated with lower barriers. Model 3 explained 18% of the variance in the PABS outcome (R² = 0.18) and had an RMSE of 5.36.

Model 4: Interpersonal-sociocultural domain-level of influence intersection.

Model 4 is an Elastic Net with α = 0.10 and λ = 0.16 (Fig 2D). The final model retains 5 independent variables as shown in Eq (9). (9)

These variables measure the use of technology as a means to monitor health, and take on binary values of 1 if the statement is true, and 0 otherwise. For example, x₉ takes the value 1 if the statement “Does not make health decisions using a smartphone or tablet” is true for participant, 0 otherwise. Variables x₉ (“Does not make health decisions using a smartphone or tablet”) and x₁₀ (“Does not track health with a wearable device”) are positively associated with barriers to PA, and greater PABS by 2.16 and 1.73, respectively, for every unit increase. On the other hand, reporting yes to x₁₁ (“Does not make health decisions with a wearable device”), x₁₂ (“Does not text a health provider”), and x₁₃ (“Unsure of texting a health provider”) were associated with lower barriers to PA. Model 4 explained 16% of the variance in the PABS outcome (R² = 0.16) and had an RMSE of 5.45.

Comparing all models, Model 1 contained the least number of predictors, the highest alpha value, and highest lambda value. This indicates that, compared to other models, there was a more aggressive shrinkage of regression coefficients to 0. Conversely, Models 2, 3, and 4 all had the same alpha value that leaned towards L₂ regularization, which promoted the inclusion of relevant variables. Model 1 generated the smallest prediction error and had the best predictive accuracy of all models. The reason for developing for separate regression models for each domain-level intersection was the low predictive performance of a model that tried to predict PABS associated with all four NIMHD domain-influence levels at once.

Discussion

In this study, we focused on answering two research questions. To explore the first question (“How can we quantify self-reported barriers to engage in PA using a scoring system?”), we utilized two domains (behavioral and sociocultural environment) and two levels of influence (individual and interpersonal) from the NIMHD framework and matched validated survey questionnaires to these domain-influence intersections. To explore the second study question (“How can we use individual, interpersonal, behavioral and socio-cultural factors to predict this score in nursing student population?”), we developed regularized regression models to predict PABS for each domain-influence intersection.

To the best of our knowledge, this is the first study that utilized the NIMHD framework and regularized regression models to study multi-level factors hindering PA. Previous studies have implemented ordinary least squares (OLS) regression and various logistic regression models to explore the intersection between health disparities and PA, but regularized regression methods have not been explored extensively [13, 21, 56, 57]. When using linear regression, collinearity between parameters and parameters canceling each other, among other reasons, can impact model performance. Our models overcome these challenges by constraining the coefficient parameters by incorporating a penalty term to achieve models with better predictive power.

Key findings from Model 1 show the importance of interest in PA as an intrinsic motivation factor to reduce barriers to PA. Additionally, engaging in leisure time PA is associated with lower barriers score. This is as expected, as adults with lower barriers to PA are more likely to active during their leisure time [3].

Model 2 highlights the importance of social support in reducing barriers to PA with interesting findings. More emotional support (e.g., advice, encouragement) and instrumental support (e.g., having appropriate gear, financial support) are associated with lower PABS. This is in line with previous studies demonstrated that social and environmental supports are key to increasing and maintaining PA [3]. Conversely, individuals reporting higher validation support (e.g., seeking social comparison or holding a relative status in a group) also indicated higher PABS. This unexpected correlation might find explanation in the Social Determination Theory, wherein seeking validation support acts as an external motivator [58]. Consequently, those seeking such support might lack intrinsic motivation, often perceiving more barriers than benefits to PA. In contrast, internally motivated individuals, who tend to focus on the benefits and enjoyment PA, perceive fewer barriers. Our study supports the development of PA interventions focusing on enhancing intrinsic motivation, as proposed in the Self-determination Theory, by addressing facilitating factors.

Model 3 focuses on highest level of education and annual family income. Results show that higher family income is associated with lower PABS, as expected. Interestingly, higher education is associated with higher PABS. This could be related to age being a confounder. PA tends to decline from young to middle age adulthood. This in part, may be due to the life events and transitions (e.g., shifting family and work responsibilities) that may be barriers to being physically active during midlife [59].

Model 4 indicates the importance of decision making and tracking functionalities of health technology that helps reduce the barriers to PA. Wearable technologies and complimentary smartphone and web applications can address barriers and promote PA behavior change techniques such as self monitoring, goal setting, and data sharing for social support [60]. Wearable activity trackers can also encourage lifestyle PA spread throughout the day and reduce the barriers associated with structured forms of PA that often require more time and resources [61].

Practical contributions

There is a growing need for developing health-informatics driven solutions that target future healthcare workforce to identify populations who are at risk of low PA levels. Previous studies have shown the usefulness of influencing PA with health technology, such as smart devices or wearables [62]. Our regression models identified the importance of decision making and tracking functionalities of health technology to reduce the barriers to PA. Wearable technologies and complimentary smartphone and web applications can address barriers and promote PA behavior change techniques such as self-monitoring, goal setting and data sharing for social support [60]. Wearable activity trackers can also encourage lifestyle PA spread throughout the day and reduce barriers associated with structured forms of PA that often require more time and resources [61]. Further, utilizing the NIMHD Research Framework as proposed in this study can inform development of interventions to address disparities in experiencing barriers to PA. Understanding and acknowledging the barriers to PA that stem from socioeconomic factors, like educational background and income accessibility, should be integral in shaping and executing future interventions and policy frameworks. In addition, nurses serve as role models for patients’ health behaviors. Nurses who experience less barriers to engage in PA are better equipped to encourage patients to healthy PA behaviors, promoting a wellness culture in the healthcare system [11, 63].

Limitations and future directions

Our study has several limitations. First, we used survey data from nursing student population at a single academic institution, which may have introduced sampling bias. The conclusions derived from this limited sample may not be broadly applicable and have limited generalizability. To address this, we recommended increasing the sample size and expanding the study to multiple settings in future studies.

Our study has utilized self-reported survey-based data, which is aligned with the PA literature. Studies that investigate PA levels have utilized self-reported questionnaires or devices (e.g., accelerometers, pedometers) [64]. Of these, self-reported questionnaires are common due to their ease of access and cost-effectiveness [65], however, are prone to recall bias. Barriers to PA have been similarly measured with various questionnaires [37, 66, 67]. Our models focused on the individual and interpersonal levels of influence and the behavioral and sociocultural domains.

Future studies can expand the selection of validated survey tools and utilize more domains and levels of influence within the NIMHD Framework to increase the included variables in predictive models. In addition, future studies can incorporate data from wearable devices. This is particularly important to improve the model performance as all predictive models in our study had low adjusted R² values which suggests that there may be other factors impacting barriers to PA that our models have not considered. These potentially missed variables could be captured with wearable device data. Low adjusted R² values could also indicate that linear models are not capturing the complex relationships between multi-level factors associated with barriers to PA. Future studies should consider non-linear models.

Our study highlights that there are significant limitations in the published literature related to populations, particularly in patient-facing healthcare professions like nursing, who may experience higher barriers to engage in PA. This needs to be addressed in future studies by exploring a variety of individual, social and environmental factors specifically targeting the nursing, and future nursing, workforce populations. In conclusion, intrinsic motivation, social and emotional support, and use of health technologies for tracking and decision making purposes were the significant predictors associated with barriers to PA. Developing and employing health informatics solutions to support the monitoring, data sharing, and communication that elicit social support and motivation with complementary health technologies may be critical tactics to prevent barriers and support healthy, active living among the nursing profession.

Supporting information

S1 File. Final version of REDCap survey.

https://doi.org/10.1371/journal.pone.0304214.s001

(PDF)

S1 Table. Missing value percentages for independent variables.

https://doi.org/10.1371/journal.pone.0304214.s002

(PDF)

S1 Data. Participant-level dataset used in descriptive and predictive analyses.

https://doi.org/10.1371/journal.pone.0304214.s003

(XLSX)

Acknowledgments

The authors thank all participants for their participation and nursing faculty at the The Elaine Marieb College of Nursing for dedicating class time for the study data collection.

References

1. Lee DH, Rezende LFM, Joh HK, Keum N, Ferrari G, Rey-Lopez JP, et al. Long-Term Leisure-Time Physical Activity Intensity and All-Cause and Cause-Specific Mortality: A Prospective Cohort of US Adults. Circulation. 2022;146(7):523–534. pmid:35876019
- View Article
- PubMed/NCBI
- Google Scholar
2. Priano SM, Hong OS, Chen JL. Lifestyles and Health-Related Outcomes of U.S. Hospital Nurses: A Systematic Review. Nursing Outlook. 2018;66(1):66–76. pmid:29037501
- View Article
- PubMed/NCBI
- Google Scholar
3. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJ, Martin BW. Correlates of physical activity: why are some people physically active and others not? The Lancet. 2012;380(9838):258–271. pmid:22818938
- View Article
- PubMed/NCBI
- Google Scholar
4. Pronk NP, Kleinman DV, Richmond TS. Healthy People 2030: Moving toward equitable health and well-being in the United States. EClinicalMedicine. 2021;33:100777. pmid:33733077
- View Article
- PubMed/NCBI
- Google Scholar
5. Hassmiller SB, Wakefield MK. The Future of Nursing 2020–2030: Charting a path to achieve health equity. Nursing Outlook. 2022;70(6):S1–S9. pmid:36446536
- View Article
- PubMed/NCBI
- Google Scholar
6. Melnyk BM, Kelly SA, Stephens J, Dhakal K, McGovern C, Tucker S, et al. Interventions to Improve Mental Health, Well-Being, Physical Health, and Lifestyle Behaviors in Physicians and Nurses: A Systematic Review. American Journal of Health Promotion. 2020;34(8):929–941. pmid:32338522
- View Article
- PubMed/NCBI
- Google Scholar
7. Nicholls R, Perry L, Duffield C, Gallagher R, Pierce H. Barriers and facilitators to healthy eating for nurses in the workplace: an integrative review. Journal of Advanced Nursing. 2017;73(5):1051–1065. pmid:27732741
- View Article
- PubMed/NCBI
- Google Scholar
8. Bakhshi S, Sun F, Murrells T, While A. Nurses’ health behaviours and physical activity-related health-promotion practices. British Journal of Community Nursing. 2015;20(6):289–296. pmid:26043015
- View Article
- PubMed/NCBI
- Google Scholar
9. Reed JL, Prince SA, Pipe AL, Attallah S, Adamo KB, Tulloch HE, et al. Influence of the workplace on physical activity and cardiometabolic health: Results of the multi-centre cross-sectional Champlain Nurses’ study. International Journal of Nursing Studies. 2018;81:49–60. pmid:29499395
- View Article
- PubMed/NCBI
- Google Scholar
10. Mc Carthy VJC, Wills T, Crowley S. Nurses, age, job demands and physical activity at work and at leisure: A cross-sectional study. Applied Nursing Research. 2018;40:116–121. pmid:29579485
- View Article
- PubMed/NCBI
- Google Scholar
11. Ross A, Touchton-Leonard K, Perez A, Wehrlen L, Kazmi N, Gibbons S. Factors That Influence Health-Promoting Self-care in Registered Nurses: Barriers and Facilitators. Advances in Nursing Science. 2019;42(4):358–373. pmid:31299691
- View Article
- PubMed/NCBI
- Google Scholar
12. Ross A, Bevans M, Brooks AT, Gibbons S, Wallen GR. Nurses and Health‐Promoting Behaviors: Knowledge May Not Translate Into Self‐Care. AORN Journal. 2017;105(3):267–275. pmid:28241948
- View Article
- PubMed/NCBI
- Google Scholar
13. Bowles HR, Morrow JR, Leonard BL, Hawkins M, Couzelis PM. The Association between Physical Activity Behavior and Commonly Reported Barriers in a Worksite Population. Research Quarterly for Exercise and Sport. 2002;73(4):464–470. pmid:12495249
- View Article
- PubMed/NCBI
- Google Scholar
14. Abedini S, Morowatisharifabad M, Enjezab B, Fallahzade H, Barkhordari A. Perceived barriers to physical activity among nurses in Bandar Abbas, Iran: A qualitative approach;.
15. Albert N, Butler R, Sorrell J. Factors Related to Healthy Diet and Physical Activity in Hospital-Based Clinical Nurses. OJIN: The Online Journal of Issues in Nursing. 2014;19(3). pmid:26824153
- View Article
- PubMed/NCBI
- Google Scholar
16. Nerek A, Wesołowska-Górniak K, Czarkowska-Pączek B. Barriers of physical activity and consequences of inactivity in the population of nurses. A narrative review. Pielegniarstwo XXI wieku / Nursing in the 21st Century. 2021;20(4):256–260.
- View Article
- Google Scholar
17. Stanulewicz N, Knox E, Narayanasamy M, Shivji N, Khunti K, Blake H. Effectiveness of Lifestyle Health Promotion Interventions for Nurses: A Systematic Review. International Journal of Environmental Research and Public Health. 2019;17(1):17. pmid:31861367
- View Article
- PubMed/NCBI
- Google Scholar
18. Al-Tannir MA, Kobrosly SY, Elbakri NK, Abu-Shaheen AK. Prevalence and predictors of physical exercise among nurses. Saudi Medical Journal. 2017;38(2):209–212.
- View Article
- Google Scholar
19. Chin DL, Nam S, Lee SJ. Occupational factors associated with obesity and leisure-time physical activity among nurses: A cross sectional study. International Journal of Nursing Studies. 2016;57:60–69. pmid:27045565
- View Article
- PubMed/NCBI
- Google Scholar
20. Bodenheimer T, Sinsky C. From Triple to Quadruple Aim: Care of the Patient Requires Care of the Provider. The Annals of Family Medicine. 2014;12(6):573–576. pmid:25384822
- View Article
- PubMed/NCBI
- Google Scholar
21. Blake H, Malik S, Mo PKH, Pisano C. ‘Do as I say, but not as I do’: Are next generation nurses role models for health? Perspectives in Public Health. 2011;131(5):231–239. pmid:21999028
- View Article
- PubMed/NCBI
- Google Scholar
22. Thwaite TL, Heidke P, Williams SL, Vandelanotte C, Rebar AL, Khalesi S. Barriers to healthy lifestyle behaviors in Australian nursing students: A qualitative study. Nursing & Health Sciences. 2020;22(4):921–928. pmid:32533602
- View Article
- PubMed/NCBI
- Google Scholar
23. Veseta U, Lagzdiņa R, Rumaka M, Reide L, Arnis V, Kampara M, et al. Physical Activity Habits of Latvian Nursing Students: A Cross-Sectional Study. Nursing Reports. 2022;12(4):922–932. pmid:36548162
- View Article
- PubMed/NCBI
- Google Scholar
24. Jerome GJ, Boyer WR, Bustamante EE, Kariuki J, Lopez-Jimenez F, Paluch AE, et al. Increasing Equity of Physical Activity Promotion for Optimal Cardiovascular Health in Adults: A Scientific Statement From the American Heart Association. Circulation. 2023;147(25):1951–1962. pmid:37222169
- View Article
- PubMed/NCBI
- Google Scholar
25. Alvidrez J, Castille D, Laude-Sharp M, Rosario A, Tabor D. The National Institute on Minority Health and Health Disparities Research Framework. American Journal of Public Health. 2019;109(S1):S16–S20. pmid:30699025
- View Article
- PubMed/NCBI
- Google Scholar
26. Courneya KS, Plotnikoff RC, Hotz SB, Birkett NJ. Social Support and the Theory of Planned Behavior in the Exercise Domain. American Journal of Health Behavior. 2000;24(4):300–308.
- View Article
- Google Scholar
27. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics. 2009;42(2):377–381. pmid:18929686
- View Article
- PubMed/NCBI
- Google Scholar
28. Harris PA, Taylor R, Minor BL, Elliott V, Fernandez M, O’Neal L, et al. The REDCap consortium: Building an international community of software platform partners. Journal of Biomedical Informatics. 2019;95:103208. pmid:31078660
- View Article
- PubMed/NCBI
- Google Scholar
29. Craig CL, Marshall AL, SjöStröm M, Bauman AE, Booth ML, Ainsworth BE, et al. International Physical Activity Questionnaire: 12-Country Reliability and Validity. Medicine & Science in Sports & Exercise. 2003;35(8):1381–1395. pmid:12900694
- View Article
- PubMed/NCBI
- Google Scholar
30. Richard M, Christina MF, Deborah LS, Rubio N, Kennon MS. Intrinsic motivation and exercise adherence. Int J Sport Psychol. 1997;28(4):335–354.
- View Article
- Google Scholar
31. Sechrist KR, Walker SN, Pender NJ. Development and psychometric evaluation of the exercise benefits/barriers scale. Research in nursing & health. 1987;10(6):357–365. Wiley Online Library. pmid:3423307
- View Article
- PubMed/NCBI
- Google Scholar
32. Hamilton CM, Strader LC, Pratt JG, Maiese D, Hendershot T, Kwok RK, et al. The PhenX Toolkit: Get the Most From Your Measures. American Journal of Epidemiology. 2011;174(3):253–260. pmid:21749974
- View Article
- PubMed/NCBI
- Google Scholar
33. Golaszewski NM, Bartholomew JB. The Development of the Physical Activity and Social Support Scale. Journal of Sport and Exercise Psychology. 2019;41(4):215–229. pmid:31461243
- View Article
- PubMed/NCBI
- Google Scholar
34. Wanner M, Probst-Hensch N, Kriemler S, Meier F, Autenrieth C, Martin BW. Validation of the long international physical activity questionnaire: Influence of age and language region. Preventive Medicine Reports. 2016;3:250–256. pmid:27419023
- View Article
- PubMed/NCBI
- Google Scholar
35. Wilson PM, Rodgers WM, Fraser SN. Cross-Validation of the Revised Motivation for Physical Activity Measure in Active Women. Research Quarterly for Exercise and Sport. 2002;73(4):471–477. pmid:12495250
- View Article
- PubMed/NCBI
- Google Scholar
36. Battistelli A, Montani F, Guicciardi M, Bertinato L. Regulation of exercise behaviour and motives for physical activities: The Italian validation of BREQ and MPAM-R questionnaires. Psychologie Française. 2016;61(4):333–348.
- View Article
- Google Scholar
37. Brown SA. Measuring Perceived Benefits and Perceived Barriers for Physical Activity. American Journal of Health Behavior. 2005;29(2):107–116. pmid:15698978
- View Article
- PubMed/NCBI
- Google Scholar
38. Cox LA, Hwang S, Haines J, Ramos EM, McCarty CA, Marazita ML, et al. Using the PhenX Toolkit to Select Standard Measurement Protocols for Your Research Study. Current Protocols. 2021;1(5):e149. pmid:34038028
- View Article
- PubMed/NCBI
- Google Scholar
39. Krzyzanowski MC, Ives CL, Jones NL, Entwisle B, Fernandez A, Cullen TA, et al. The PhenX Toolkit: Measurement Protocols for Assessment of Social Determinants of Health. American Journal of Preventive Medicine. 2023;65(3):534–542. pmid:36935055
- View Article
- PubMed/NCBI
- Google Scholar
40. Chen J, Cheng Z, Herold F, Wang T, Kuang J, Taylor A, et al. Validity and Reliability of the Physical Activity and Social Support Scale among Chinese College Students. 2023;.
41. R Core Team (2023). R: A language and environment for statistical computing R Foundation for Statistical Computing, Vienna, Austria. Available from: https://www.R-project.org/.
42. Little RJA. A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association. 1988;83(404):1198–202.
- View Article
- Google Scholar
43. Agor J, Özaltın OY, Ivy JS, Capan M, Arnold R, Romero S. The value of missing information in severity of illness score development. Journal of Biomedical Informatics. 2019;97:103255. pmid:31349049
- View Article
- PubMed/NCBI
- Google Scholar
44. Gomer B, Yuan KH. A Realistic Evaluation of Methods for Handling Missing Data When There is a Mixture of MCAR, MAR, and MNAR Mechanisms in the Same Dataset. Multivariate Behavioral Research. 2023;58(5):988–1013. pmid:36599049
- View Article
- PubMed/NCBI
- Google Scholar
45. Tierney N, Cook D. Expanding Tidy Data Principles to Facilitate Missing Data Exploration, Visualization, and Assessment of Imputations. Journal of Statistical Software. 2023;105:1–31
- View Article
- Google Scholar
46. Henderson DA, Denison DR. Stepwise Regression in Social and Psychological Research. Psychological Reports. 1989;64(1):251–257.
- View Article
- Google Scholar
47. Wang M, Wright J, Brownlee A, Buswell R. A comparison of approaches to stepwise regression on variables sensitivities in building simulation and analysis. Energy and Buildings. 2016;127:313–326.
- View Article
- Google Scholar
48. Venables WN, Ripley BD. Modern Applied Statistics with S;.
49. Greenwood CJ, Youssef GJ, Letcher P, Macdonald JA, Hagg LJ, Sanson A, et al. A comparison of penalised regression methods for informing the selection of predictive markers. PLOS ONE. 2020;15(11):e0242730. pmid:33216811
- View Article
- PubMed/NCBI
- Google Scholar
50. James G, Witten D, Hastie T, Tibshirani R. An Introduction to Statistical Learning: with Applications in R. Springer Texts in Statistics. New York, NY: Springer US; 2021. Available from: https://link.springer.com/10.1007/978-1-0716-1418-1.
51. Hoerl AE, Kennard RW. Ridge Regression: Biased Estimation for Nonorthogonal Problems. 2023;.
52. Tibshirani R. Regression Shrinkage and Selection Via the Lasso. Journal of the Royal Statistical Society: Series B (Methodological). 1996;58(1):267–288.
- View Article
- Google Scholar
53. Zou H, Hastie T. Regularization and Variable Selection Via the Elastic Net. Journal of the Royal Statistical Society Series B: Statistical Methodology. 2005;67(2):301–320.
- View Article
- Google Scholar
54. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. Journal of Statistical Software. 2010;33(1). pmid:20808728
- View Article
- PubMed/NCBI
- Google Scholar
55. Kuhn M. Building Predictive Models in R Using the caret Package. Journal of Statistical Software. 2008;28(5).
- View Article
- Google Scholar
56. Koh YS, Asharani PV, Devi F, Roystonn K, Wang P, Vaingankar JA, et al. A cross-sectional study on the perceived barriers to physical activity and their associations with domain-specific physical activity and sedentary behaviour. BMC Public Health. 2022;22(1):1051. pmid:35614402
- View Article
- PubMed/NCBI
- Google Scholar
57. Herazo-Beltrán Y, Pinillos Y, Vidarte J, Crissien E, Suarez D, García R. Predictors of perceived barriers to physical activity in the general adult population: a cross-sectional study. Brazilian Journal of Physical Therapy. 2017;21(1):44–50. pmid:28442074
- View Article
- PubMed/NCBI
- Google Scholar
58. Ryan RM, Deci EL. Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being. American Psychologist. 2000;. pmid:11392867
- View Article
- PubMed/NCBI
- Google Scholar
59. Lane-Cordova AD, Jerome GJ, Paluch AE, Bustamante EE, LaMonte MJ, Pate RR, et al. Supporting Physical Activity in Patients and Populations During Life Events and Transitions: A Scientific Statement From the American Heart Association. Circulation. 2022;145(4). pmid:34847691
- View Article
- PubMed/NCBI
- Google Scholar
60. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior Change Techniques Implemented in Electronic Lifestyle Activity Monitors: A Systematic Content Analysis. Journal of Medical Internet Research. 2014;16(8):e192. pmid:25131661
- View Article
- PubMed/NCBI
- Google Scholar
61. Chaddha A, Jackson EA, Richardson CR, Franklin BA. Technology to Help Promote Physical Activity. The American Journal of Cardiology. 2017;119(1):149–152. pmid:27889045
- View Article
- PubMed/NCBI
- Google Scholar
62. Pope Z, Barr-Anderson D, Lewis B, Pereira M, Gao Z. Use of Wearable Technology and Social Media to Improve Physical Activity and Dietary Behaviors among College Students: A 12-Week Randomized Pilot Study. International Journal of Environmental Research and Public Health. 2019;16(19):3579. pmid:31557812
- View Article
- PubMed/NCBI
- Google Scholar
63. Zeng W, Shang S, Fang Q, He S, Li J, Yao Y. Health promoting lifestyle behaviors and associated predictors among clinical nurses in China: a cross-sectional study. BMC Nursing. 2021;20(1):230. pmid:34789261
- View Article
- PubMed/NCBI
- Google Scholar
64. Vanhees L, Lefevre J, Philippaerts R, Martens M, Huygens W, Troosters T, et al. How to assess physical activity? How to assess physical fitness?;.
65. Sylvia LG, Bernstein EE, Hubbard JL, Keating L, Anderson EJ. Practical Guide to Measuring Physical Activity. Journal of the Academy of Nutrition and Dietetics. 2014;114(2):199–208. pmid:24290836
- View Article
- PubMed/NCBI
- Google Scholar
66. Arzu D, Tuzun EH, Eker L. Perceived Barriers to Physical Activity in University Students. Journal of Sports Science & Medicine. 2006;5(4):615–620. pmid:24357957
- View Article
- PubMed/NCBI
- Google Scholar
67. Fox AM, Mann DM, Ramos MA, Kleinman LC, Horowitz CR. Barriers to Physical Activity in East Harlem, New York. Journal of Obesity. 2012;2012:1–8. pmid:22848797
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. Lee DH, Rezende LFM, Joh HK, Keum N, Ferrari G, Rey-Lopez JP, et al. Long-Term Leisure-Time Physical Activity Intensity and All-Cause and Cause-Specific Mortality: A Prospective Cohort of US Adults. Circulation. 2022;146(7):523–534. pmid:35876019
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. Priano SM, Hong OS, Chen JL. Lifestyles and Health-Related Outcomes of U.S. Hospital Nurses: A Systematic Review. Nursing Outlook. 2018;66(1):66–76. pmid:29037501
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Bauman AE, Reis RS, Sallis JF, Wells JC, Loos RJ, Martin BW. Correlates of physical activity: why are some people physically active and others not? The Lancet. 2012;380(9838):258–271. pmid:22818938
View Article
PubMed/NCBI
Google Scholar

[10] View Article

[11] PubMed/NCBI

[12] Google Scholar

[ref4] 4. Pronk NP, Kleinman DV, Richmond TS. Healthy People 2030: Moving toward equitable health and well-being in the United States. EClinicalMedicine. 2021;33:100777. pmid:33733077
View Article
PubMed/NCBI
Google Scholar

[14] View Article

[15] PubMed/NCBI

[16] Google Scholar

[ref5] 5. Hassmiller SB, Wakefield MK. The Future of Nursing 2020–2030: Charting a path to achieve health equity. Nursing Outlook. 2022;70(6):S1–S9. pmid:36446536
View Article
PubMed/NCBI
Google Scholar

[18] View Article

[19] PubMed/NCBI

[20] Google Scholar

[ref6] 6. Melnyk BM, Kelly SA, Stephens J, Dhakal K, McGovern C, Tucker S, et al. Interventions to Improve Mental Health, Well-Being, Physical Health, and Lifestyle Behaviors in Physicians and Nurses: A Systematic Review. American Journal of Health Promotion. 2020;34(8):929–941. pmid:32338522
View Article
PubMed/NCBI
Google Scholar

[22] View Article

[23] PubMed/NCBI

[24] Google Scholar

[ref7] 7. Nicholls R, Perry L, Duffield C, Gallagher R, Pierce H. Barriers and facilitators to healthy eating for nurses in the workplace: an integrative review. Journal of Advanced Nursing. 2017;73(5):1051–1065. pmid:27732741
View Article
PubMed/NCBI
Google Scholar

[26] View Article

[27] PubMed/NCBI

[28] Google Scholar

[ref8] 8. Bakhshi S, Sun F, Murrells T, While A. Nurses’ health behaviours and physical activity-related health-promotion practices. British Journal of Community Nursing. 2015;20(6):289–296. pmid:26043015
View Article
PubMed/NCBI
Google Scholar

[30] View Article

[31] PubMed/NCBI

[32] Google Scholar

[ref9] 9. Reed JL, Prince SA, Pipe AL, Attallah S, Adamo KB, Tulloch HE, et al. Influence of the workplace on physical activity and cardiometabolic health: Results of the multi-centre cross-sectional Champlain Nurses’ study. International Journal of Nursing Studies. 2018;81:49–60. pmid:29499395
View Article
PubMed/NCBI
Google Scholar

[34] View Article

[35] PubMed/NCBI

[36] Google Scholar

[ref10] 10. Mc Carthy VJC, Wills T, Crowley S. Nurses, age, job demands and physical activity at work and at leisure: A cross-sectional study. Applied Nursing Research. 2018;40:116–121. pmid:29579485
View Article
PubMed/NCBI
Google Scholar

[38] View Article

[39] PubMed/NCBI

[40] Google Scholar

[ref11] 11. Ross A, Touchton-Leonard K, Perez A, Wehrlen L, Kazmi N, Gibbons S. Factors That Influence Health-Promoting Self-care in Registered Nurses: Barriers and Facilitators. Advances in Nursing Science. 2019;42(4):358–373. pmid:31299691
View Article
PubMed/NCBI
Google Scholar

[42] View Article

[43] PubMed/NCBI

[44] Google Scholar

[ref12] 12. Ross A, Bevans M, Brooks AT, Gibbons S, Wallen GR. Nurses and Health‐Promoting Behaviors: Knowledge May Not Translate Into Self‐Care. AORN Journal. 2017;105(3):267–275. pmid:28241948
View Article
PubMed/NCBI
Google Scholar

[46] View Article

[47] PubMed/NCBI

[48] Google Scholar

[ref13] 13. Bowles HR, Morrow JR, Leonard BL, Hawkins M, Couzelis PM. The Association between Physical Activity Behavior and Commonly Reported Barriers in a Worksite Population. Research Quarterly for Exercise and Sport. 2002;73(4):464–470. pmid:12495249
View Article
PubMed/NCBI
Google Scholar

[50] View Article

[51] PubMed/NCBI

[52] Google Scholar

[ref14] 14. Abedini S, Morowatisharifabad M, Enjezab B, Fallahzade H, Barkhordari A. Perceived barriers to physical activity among nurses in Bandar Abbas, Iran: A qualitative approach;.

[ref15] 15. Albert N, Butler R, Sorrell J. Factors Related to Healthy Diet and Physical Activity in Hospital-Based Clinical Nurses. OJIN: The Online Journal of Issues in Nursing. 2014;19(3). pmid:26824153
View Article
PubMed/NCBI
Google Scholar

[55] View Article

[56] PubMed/NCBI

[57] Google Scholar

[ref16] 16. Nerek A, Wesołowska-Górniak K, Czarkowska-Pączek B. Barriers of physical activity and consequences of inactivity in the population of nurses. A narrative review. Pielegniarstwo XXI wieku / Nursing in the 21st Century. 2021;20(4):256–260.
View Article
Google Scholar

[59] View Article

[60] Google Scholar

[ref17] 17. Stanulewicz N, Knox E, Narayanasamy M, Shivji N, Khunti K, Blake H. Effectiveness of Lifestyle Health Promotion Interventions for Nurses: A Systematic Review. International Journal of Environmental Research and Public Health. 2019;17(1):17. pmid:31861367
View Article
PubMed/NCBI
Google Scholar

[62] View Article

[63] PubMed/NCBI

[64] Google Scholar

[ref18] 18. Al-Tannir MA, Kobrosly SY, Elbakri NK, Abu-Shaheen AK. Prevalence and predictors of physical exercise among nurses. Saudi Medical Journal. 2017;38(2):209–212.
View Article
Google Scholar

[66] View Article

[67] Google Scholar

[ref19] 19. Chin DL, Nam S, Lee SJ. Occupational factors associated with obesity and leisure-time physical activity among nurses: A cross sectional study. International Journal of Nursing Studies. 2016;57:60–69. pmid:27045565
View Article
PubMed/NCBI
Google Scholar

[69] View Article

[70] PubMed/NCBI

[71] Google Scholar

[ref20] 20. Bodenheimer T, Sinsky C. From Triple to Quadruple Aim: Care of the Patient Requires Care of the Provider. The Annals of Family Medicine. 2014;12(6):573–576. pmid:25384822
View Article
PubMed/NCBI
Google Scholar

[73] View Article

[74] PubMed/NCBI

[75] Google Scholar

[ref21] 21. Blake H, Malik S, Mo PKH, Pisano C. ‘Do as I say, but not as I do’: Are next generation nurses role models for health? Perspectives in Public Health. 2011;131(5):231–239. pmid:21999028
View Article
PubMed/NCBI
Google Scholar

[77] View Article

[78] PubMed/NCBI

[79] Google Scholar

[ref22] 22. Thwaite TL, Heidke P, Williams SL, Vandelanotte C, Rebar AL, Khalesi S. Barriers to healthy lifestyle behaviors in Australian nursing students: A qualitative study. Nursing & Health Sciences. 2020;22(4):921–928. pmid:32533602
View Article
PubMed/NCBI
Google Scholar

[81] View Article

[82] PubMed/NCBI

[83] Google Scholar

[ref23] 23. Veseta U, Lagzdiņa R, Rumaka M, Reide L, Arnis V, Kampara M, et al. Physical Activity Habits of Latvian Nursing Students: A Cross-Sectional Study. Nursing Reports. 2022;12(4):922–932. pmid:36548162
View Article
PubMed/NCBI
Google Scholar

[85] View Article

[86] PubMed/NCBI

[87] Google Scholar

[ref24] 24. Jerome GJ, Boyer WR, Bustamante EE, Kariuki J, Lopez-Jimenez F, Paluch AE, et al. Increasing Equity of Physical Activity Promotion for Optimal Cardiovascular Health in Adults: A Scientific Statement From the American Heart Association. Circulation. 2023;147(25):1951–1962. pmid:37222169
View Article
PubMed/NCBI
Google Scholar

[89] View Article

[90] PubMed/NCBI

[91] Google Scholar

[ref25] 25. Alvidrez J, Castille D, Laude-Sharp M, Rosario A, Tabor D. The National Institute on Minority Health and Health Disparities Research Framework. American Journal of Public Health. 2019;109(S1):S16–S20. pmid:30699025
View Article
PubMed/NCBI
Google Scholar

[93] View Article

[94] PubMed/NCBI

[95] Google Scholar

[ref26] 26. Courneya KS, Plotnikoff RC, Hotz SB, Birkett NJ. Social Support and the Theory of Planned Behavior in the Exercise Domain. American Journal of Health Behavior. 2000;24(4):300–308.
View Article
Google Scholar

[97] View Article

[98] Google Scholar

[ref27] 27. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics. 2009;42(2):377–381. pmid:18929686
View Article
PubMed/NCBI
Google Scholar

[100] View Article

[101] PubMed/NCBI

[102] Google Scholar

[ref28] 28. Harris PA, Taylor R, Minor BL, Elliott V, Fernandez M, O’Neal L, et al. The REDCap consortium: Building an international community of software platform partners. Journal of Biomedical Informatics. 2019;95:103208. pmid:31078660
View Article
PubMed/NCBI
Google Scholar

[104] View Article

[105] PubMed/NCBI

[106] Google Scholar

[ref29] 29. Craig CL, Marshall AL, SjöStröm M, Bauman AE, Booth ML, Ainsworth BE, et al. International Physical Activity Questionnaire: 12-Country Reliability and Validity. Medicine & Science in Sports & Exercise. 2003;35(8):1381–1395. pmid:12900694
View Article
PubMed/NCBI
Google Scholar

[108] View Article

[109] PubMed/NCBI

[110] Google Scholar

[ref30] 30. Richard M, Christina MF, Deborah LS, Rubio N, Kennon MS. Intrinsic motivation and exercise adherence. Int J Sport Psychol. 1997;28(4):335–354.
View Article
Google Scholar

[112] View Article

[113] Google Scholar

[ref31] 31. Sechrist KR, Walker SN, Pender NJ. Development and psychometric evaluation of the exercise benefits/barriers scale. Research in nursing & health. 1987;10(6):357–365. Wiley Online Library. pmid:3423307
View Article
PubMed/NCBI
Google Scholar

[115] View Article

[116] PubMed/NCBI

[117] Google Scholar

[ref32] 32. Hamilton CM, Strader LC, Pratt JG, Maiese D, Hendershot T, Kwok RK, et al. The PhenX Toolkit: Get the Most From Your Measures. American Journal of Epidemiology. 2011;174(3):253–260. pmid:21749974
View Article
PubMed/NCBI
Google Scholar

[119] View Article

[120] PubMed/NCBI

[121] Google Scholar

[ref33] 33. Golaszewski NM, Bartholomew JB. The Development of the Physical Activity and Social Support Scale. Journal of Sport and Exercise Psychology. 2019;41(4):215–229. pmid:31461243
View Article
PubMed/NCBI
Google Scholar

[123] View Article

[124] PubMed/NCBI

[125] Google Scholar

[ref34] 34. Wanner M, Probst-Hensch N, Kriemler S, Meier F, Autenrieth C, Martin BW. Validation of the long international physical activity questionnaire: Influence of age and language region. Preventive Medicine Reports. 2016;3:250–256. pmid:27419023
View Article
PubMed/NCBI
Google Scholar

[127] View Article

[128] PubMed/NCBI

[129] Google Scholar

[ref35] 35. Wilson PM, Rodgers WM, Fraser SN. Cross-Validation of the Revised Motivation for Physical Activity Measure in Active Women. Research Quarterly for Exercise and Sport. 2002;73(4):471–477. pmid:12495250
View Article
PubMed/NCBI
Google Scholar

[131] View Article

[132] PubMed/NCBI

[133] Google Scholar

[ref36] 36. Battistelli A, Montani F, Guicciardi M, Bertinato L. Regulation of exercise behaviour and motives for physical activities: The Italian validation of BREQ and MPAM-R questionnaires. Psychologie Française. 2016;61(4):333–348.
View Article
Google Scholar

[135] View Article

[136] Google Scholar

[ref37] 37. Brown SA. Measuring Perceived Benefits and Perceived Barriers for Physical Activity. American Journal of Health Behavior. 2005;29(2):107–116. pmid:15698978
View Article
PubMed/NCBI
Google Scholar

[138] View Article

[139] PubMed/NCBI

[140] Google Scholar

[ref38] 38. Cox LA, Hwang S, Haines J, Ramos EM, McCarty CA, Marazita ML, et al. Using the PhenX Toolkit to Select Standard Measurement Protocols for Your Research Study. Current Protocols. 2021;1(5):e149. pmid:34038028
View Article
PubMed/NCBI
Google Scholar

[142] View Article

[143] PubMed/NCBI

[144] Google Scholar

[ref39] 39. Krzyzanowski MC, Ives CL, Jones NL, Entwisle B, Fernandez A, Cullen TA, et al. The PhenX Toolkit: Measurement Protocols for Assessment of Social Determinants of Health. American Journal of Preventive Medicine. 2023;65(3):534–542. pmid:36935055
View Article
PubMed/NCBI
Google Scholar

[146] View Article

[147] PubMed/NCBI

[148] Google Scholar

[ref40] 40. Chen J, Cheng Z, Herold F, Wang T, Kuang J, Taylor A, et al. Validity and Reliability of the Physical Activity and Social Support Scale among Chinese College Students. 2023;.

[ref41] 41. R Core Team (2023). R: A language and environment for statistical computing R Foundation for Statistical Computing, Vienna, Austria. Available from: https://www.R-project.org/.

[ref42] 42. Little RJA. A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association. 1988;83(404):1198–202.
View Article
Google Scholar

[152] View Article

[153] Google Scholar

[ref43] 43. Agor J, Özaltın OY, Ivy JS, Capan M, Arnold R, Romero S. The value of missing information in severity of illness score development. Journal of Biomedical Informatics. 2019;97:103255. pmid:31349049
View Article
PubMed/NCBI
Google Scholar

[155] View Article

[156] PubMed/NCBI

[157] Google Scholar

[ref44] 44. Gomer B, Yuan KH. A Realistic Evaluation of Methods for Handling Missing Data When There is a Mixture of MCAR, MAR, and MNAR Mechanisms in the Same Dataset. Multivariate Behavioral Research. 2023;58(5):988–1013. pmid:36599049
View Article
PubMed/NCBI
Google Scholar

[159] View Article

[160] PubMed/NCBI

[161] Google Scholar

[ref45] 45. Tierney N, Cook D. Expanding Tidy Data Principles to Facilitate Missing Data Exploration, Visualization, and Assessment of Imputations. Journal of Statistical Software. 2023;105:1–31
View Article
Google Scholar

[163] View Article

[164] Google Scholar

[ref46] 46. Henderson DA, Denison DR. Stepwise Regression in Social and Psychological Research. Psychological Reports. 1989;64(1):251–257.
View Article
Google Scholar

[166] View Article

[167] Google Scholar

[ref47] 47. Wang M, Wright J, Brownlee A, Buswell R. A comparison of approaches to stepwise regression on variables sensitivities in building simulation and analysis. Energy and Buildings. 2016;127:313–326.
View Article
Google Scholar

[169] View Article

[170] Google Scholar

[ref48] 48. Venables WN, Ripley BD. Modern Applied Statistics with S;.

[ref49] 49. Greenwood CJ, Youssef GJ, Letcher P, Macdonald JA, Hagg LJ, Sanson A, et al. A comparison of penalised regression methods for informing the selection of predictive markers. PLOS ONE. 2020;15(11):e0242730. pmid:33216811
View Article
PubMed/NCBI
Google Scholar

[173] View Article

[174] PubMed/NCBI

[175] Google Scholar

[ref50] 50. James G, Witten D, Hastie T, Tibshirani R. An Introduction to Statistical Learning: with Applications in R. Springer Texts in Statistics. New York, NY: Springer US; 2021. Available from: https://link.springer.com/10.1007/978-1-0716-1418-1.

[ref51] 51. Hoerl AE, Kennard RW. Ridge Regression: Biased Estimation for Nonorthogonal Problems. 2023;.

[ref52] 52. Tibshirani R. Regression Shrinkage and Selection Via the Lasso. Journal of the Royal Statistical Society: Series B (Methodological). 1996;58(1):267–288.
View Article
Google Scholar

[179] View Article

[180] Google Scholar

[ref53] 53. Zou H, Hastie T. Regularization and Variable Selection Via the Elastic Net. Journal of the Royal Statistical Society Series B: Statistical Methodology. 2005;67(2):301–320.
View Article
Google Scholar

[182] View Article

[183] Google Scholar

[ref54] 54. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. Journal of Statistical Software. 2010;33(1). pmid:20808728
View Article
PubMed/NCBI
Google Scholar

[185] View Article

[186] PubMed/NCBI

[187] Google Scholar

[ref55] 55. Kuhn M. Building Predictive Models in R Using the caret Package. Journal of Statistical Software. 2008;28(5).
View Article
Google Scholar

[189] View Article

[190] Google Scholar

[ref56] 56. Koh YS, Asharani PV, Devi F, Roystonn K, Wang P, Vaingankar JA, et al. A cross-sectional study on the perceived barriers to physical activity and their associations with domain-specific physical activity and sedentary behaviour. BMC Public Health. 2022;22(1):1051. pmid:35614402
View Article
PubMed/NCBI
Google Scholar

[192] View Article

[193] PubMed/NCBI

[194] Google Scholar

[ref57] 57. Herazo-Beltrán Y, Pinillos Y, Vidarte J, Crissien E, Suarez D, García R. Predictors of perceived barriers to physical activity in the general adult population: a cross-sectional study. Brazilian Journal of Physical Therapy. 2017;21(1):44–50. pmid:28442074
View Article
PubMed/NCBI
Google Scholar

[196] View Article

[197] PubMed/NCBI

[198] Google Scholar

[ref58] 58. Ryan RM, Deci EL. Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being. American Psychologist. 2000;. pmid:11392867
View Article
PubMed/NCBI
Google Scholar

[200] View Article

[201] PubMed/NCBI

[202] Google Scholar

[ref59] 59. Lane-Cordova AD, Jerome GJ, Paluch AE, Bustamante EE, LaMonte MJ, Pate RR, et al. Supporting Physical Activity in Patients and Populations During Life Events and Transitions: A Scientific Statement From the American Heart Association. Circulation. 2022;145(4). pmid:34847691
View Article
PubMed/NCBI
Google Scholar

[204] View Article

[205] PubMed/NCBI

[206] Google Scholar

[ref60] 60. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior Change Techniques Implemented in Electronic Lifestyle Activity Monitors: A Systematic Content Analysis. Journal of Medical Internet Research. 2014;16(8):e192. pmid:25131661
View Article
PubMed/NCBI
Google Scholar

[208] View Article

[209] PubMed/NCBI

[210] Google Scholar

[ref61] 61. Chaddha A, Jackson EA, Richardson CR, Franklin BA. Technology to Help Promote Physical Activity. The American Journal of Cardiology. 2017;119(1):149–152. pmid:27889045
View Article
PubMed/NCBI
Google Scholar

[212] View Article

[213] PubMed/NCBI

[214] Google Scholar

[ref62] 62. Pope Z, Barr-Anderson D, Lewis B, Pereira M, Gao Z. Use of Wearable Technology and Social Media to Improve Physical Activity and Dietary Behaviors among College Students: A 12-Week Randomized Pilot Study. International Journal of Environmental Research and Public Health. 2019;16(19):3579. pmid:31557812
View Article
PubMed/NCBI
Google Scholar

[216] View Article

[217] PubMed/NCBI

[218] Google Scholar

[ref63] 63. Zeng W, Shang S, Fang Q, He S, Li J, Yao Y. Health promoting lifestyle behaviors and associated predictors among clinical nurses in China: a cross-sectional study. BMC Nursing. 2021;20(1):230. pmid:34789261
View Article
PubMed/NCBI
Google Scholar

[220] View Article

[221] PubMed/NCBI

[222] Google Scholar

[ref64] 64. Vanhees L, Lefevre J, Philippaerts R, Martens M, Huygens W, Troosters T, et al. How to assess physical activity? How to assess physical fitness?;.

[ref65] 65. Sylvia LG, Bernstein EE, Hubbard JL, Keating L, Anderson EJ. Practical Guide to Measuring Physical Activity. Journal of the Academy of Nutrition and Dietetics. 2014;114(2):199–208. pmid:24290836
View Article
PubMed/NCBI
Google Scholar

[225] View Article

[226] PubMed/NCBI

[227] Google Scholar

[ref66] 66. Arzu D, Tuzun EH, Eker L. Perceived Barriers to Physical Activity in University Students. Journal of Sports Science & Medicine. 2006;5(4):615–620. pmid:24357957
View Article
PubMed/NCBI
Google Scholar

[229] View Article

[230] PubMed/NCBI

[231] Google Scholar

[ref67] 67. Fox AM, Mann DM, Ramos MA, Kleinman LC, Horowitz CR. Barriers to Physical Activity in East Harlem, New York. Journal of Obesity. 2012;2012:1–8. pmid:22848797
View Article
PubMed/NCBI
Google Scholar

[233] View Article

[234] PubMed/NCBI

[235] Google Scholar

Figures

Abstract

Introduction

Barriers to physical activity in nursing

Health disparities and barriers to physical activity

The National Institute on Minority Health and Health Disparities (NIMHD) Research Framework

Materials and methods

Study design

Study population

Survey development

Data collection

Data preprocessing

Descriptive and statistical analysis

Predictive analysis

Results

Descriptive and statistical analysis results

Predictive analysis results

Model 1: Individual-behavioral domain-level of influence intersection.

Model 2: Interpersonal-behavioral domain-level of influence intersection.

Model 3: Individual-sociocultural domain-level of influence intersection.

Model 4: Interpersonal-sociocultural domain-level of influence intersection.

Discussion

Practical contributions

Limitations and future directions

Supporting information

S1 File. Final version of REDCap survey.

S1 Table. Missing value percentages for independent variables.

S1 Data. Participant-level dataset used in descriptive and predictive analyses.

Acknowledgments

References