*Article* **Hunger and Health: Taking a Formative Approach to Build a Health Intervention Focused on Nutrition and Physical Activity Needs as Perceived by Stakeholders**

**Kelsey Fortin \* and Susan Harvey**

Department of Health, Sport and Exercise Sciences, School of Education and Human Sciences, Lawrence Campus, University of Kansas, Lawrence, KS 66045, USA; Suharvey@ku.edu **\*** Correspondence: Kelseyf123@ku.edu

**Abstract:** The intersections between hunger and health are beginning to gain traction. New interventions emphasize collaboration between the health and social service sectors. This study aimed to understand the nutrition and physical activity (PA) needs as perceived by food pantry stakeholders to inform a health intervention approach. The study used formative research incorporating mixed methods through surveying and semi-structured interviews with three food pantry stakeholder groups: Clients (*n* = 30), staff (*n* = 7), and volunteers (*n* = 10). Pantry client participants reported; high rates of both individual (60%, *n* = 18) and household (43%, *n* = 13) disease diagnosis; low consumption (0–1 servings) of fruits (67%, *n* = 20) and vegetables (47%, *n* = 14) per day; and low levels (0–120 min) of PA (67%, *n* = 20) per week. Interviews identified five final convergent major themes across all three stakeholder groups including food and PA barriers, nutrition and PA literacy, health status and lifestyle, current pantry operations and adjustments, and suggestions for health intervention programming. High rates of chronic disease combined with low health literacy among pantry clients demonstrate the need to address health behaviors. Further research piloting the design and implementation of a comprehensive health behavior intervention program in the food pantry setting is needed.

**Keywords:** food insecurity; hunger and health; nutrition; physical activity; health intervention; formative research

#### **1. Introduction**

Food pantries offer important resources in the federal aid system. Food insecurity is defined by the USDA as "limited or uncertain availability of nutritionally adequate and safe foods," and 14 million U.S. households were food insecure in 2018 [1]. Despite the availability of federal nutrition assistance programs (e.g., SNAP, WIC, TANF), there is a gap in services, leaving organizations like Feeding America, a network of 60,000 food pantries and meal programs, still serving roughly 4.3 million meals to hungry people [2]. These emergency food services are reaching the most vulnerable populations needing both food and health services. The number of chronic diseases for adults in households with low food security, is on average, 18 percent higher than those with high food-security [2], and one out of three chronically ill food insecure adults are unable to afford medicine, food, or both [3].

Significant financial constraints leave food insecure individuals frequently limited to food pantry availability and low-cost food items. This translates into coping strategies promoting low nutrient diets high in processed foods [4]. In general, poor dietary intake (e.g., excess saturated or trans-fat intake, a diet low in fruits and vegetables) has been linked to a number of chronic diseases, including cardiovascular disease, Type 2 diabetes, some types of cancer and osteoporosis [5,6]. Overall, those living in food insecure households often have disrupted eating patterns and diets that are inadequate in nutrient-dense

**Citation:** Fortin, K.; Harvey, S. Hunger and Health: Taking a Formative Approach to Build a Health Intervention Focused on Nutrition and Physical Activity Needs as Perceived by Stakeholders. *Nutrients* **2021**, *13*, 1584. https:// doi.org/10.3390/nu13051584

Academic Editor: Carlos Vasconcelos

Received: 24 March 2021 Accepted: 7 May 2021 Published: 10 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

foods, contributing to malnourishment and an increased risk for poor health and chronic disease [7]. Beyond the quality of food, the existence of medical conditions associated with a poor diet can interfere with medication adherence [8]. Clients accessing a mobile food pantry reported that food insecurity impacts medication adherence due to the requirement that some medication be taken with food [9]. Therefore, gaps in the pantry schedule, lack of transportation or conflicting commitments may prevent individuals from accessing the food they need to meet medication recommendations.

Pantry clients have reported similar barriers, such as lack of transportation, inadequate kitchen equipment, lack of nutrition knowledge and skills, and few social support networks impacting their ability to eat healthy [10]. Other than adherence to medication for existing conditions, pantry clients are lacking in access, knowledge, and the to eat healthy diets to prevent disease onset. Begley et al. (2019) postulates that poor food and nutrition literacy behaviors contribute to food insecurity. Behaviors related to food planning and management, shopping, preparation, and cooking all show an association between food literacy behaviors and food security status [11]. In other words, the higher level of food literacy, the more food and nutritional behaviors individuals engage in that are associated with greater food security (e.g., food storage and preparation). Health literacy and self-efficacy have also been found to be predictors of food label use, which positively predict individuals diet quality [12]. As health professionals work to address hunger and health among food insecure populations, issues of food and health literacy are important interventional considerations.

The Department of Health and Human services (DHSS) recommends that American adults engage in a minimum of 150-minutes of moderately intense physical activity (PA) per week to experience health benefits [13]. PA rates among adults are low across the U.S. with nearly 80% of adults not meeting PA-recommended guidelines [13]. Common barriers associated with PA include a lack of confidence performing exercises, lack of time, lack of financial resources, and having diseases that create exercise limitations [13–15]. Food insecurity has demonstrated a significant association with adherence to PA guidelines among both adults and children [16]. Outside of the traditional barriers that prevent adults from engaging in PA, food insecure adults experience higher levels of stress and have poorer health, with a greater number of chronic diseases, creating larger obstacles to engaging in PA [16]. Additionally, food insecurity is associated perceptions, and readiness to engage in PA [17,18]. Within the context of disease, food insecure individuals report physical limitations that may prevent them from activities of daily life, including PA [19]. Connections between food insecurity and PA, particularly among adults, are the areas of hunger and health literature, which merit further research development.

According to the World Health Organization, non-communicable diseases (e.g., diabetes and cardiovascular disease) account for two-thirds of premature deaths worldwide [20]. Food insecure individuals are reporting broader disease prevalence and comorbidities, such as obesity, disability and mental health disorders that warrant the need for a broader approach and multi-sector collaboration among medical providers, public health practitioners, social workers and food banks [21]. Within the space of chronic disease management, health coaching interventions have shown promise in the medical setting [22,23]. Health coaching often makes use of motivational interviewing techniques that promote collaboration, client evocation and autonomy, leading to successful behavior change across a variety of contexts, populations and health behaviors [24]. Health coaching uses a relationship building strategy in health behavior change through activities, such as health education sessions and individual practical support [25]. Health coaching shows positive results when targeting a range of diseases and populations, including diabetes, heart disease, hyperlipidemia and low-income patients [26]. If food pantries can implement a broader intervention design (e.g., health coaching), incorporating more holistic behavioral components (e.g., both nutrition and PA), can be developed that captures a broader range of food insecure individuals with comorbidities, and address lifestyle health behaviors leading to those disease.

The Academy of Nutrition and Dietetics released a position paper stressing the importance for nutrition practitioners to build partnerships with food pantries [27]. Health professionals are beginning to recognize the importance of targeted interventions among food insecure populations [28]. Recent intervention research relating to diabetes, nutrition education, and dietary and food purchasing behaviors within the food pantry setting resulted in positive health outcomes for food insecure pantry clients [29–31]. Among the literature, a recent systematic review of food pantry interventions revealed nutrition literacy and diabetic management interventions have been dominant in the field [32]. The cited studies indicate innovation and promise, yet present gaps in assisting individuals outside of the diabetic and nutrition scope. Only one study in the review of the literature utilized a more holistic health coaching approach within the food pantry setting [23]. Although, the study yielded positive pantry client health outcomes, it was still focused predominantly on nutritional behaviors, disregarding PA as an important disease prevention and management strategy.

As scientists and practitioners develop and implement interventions aimed at food pantry clients, little is known about the design and implementation of health intervention that combine both nutrition and PA health behaviors within a holistic health intervention model. This study uses formative research to understand nutrition and PA needs as perceived by food pantry stakeholders (pantry clients, volunteers, and staff) to inform a health intervention approach at a county-wide Midwest food pantry. This formative approach makes use of a community participatory model [33] to gain buy-in and consultation from the community of interest. The study aims to fill a gap in the literature by; (1) understanding more about PA behaviors and needs among food insecure adult pantry users; and (2) explore the program components of a comprehensive health intervention that incorporates both PA and nutrition as perceived by pantry stakeholders. This study will act as phase one to a multiphase intervention design research project.

#### **2. Materials and Methods**

The Institutional Review Board at a large Midwest research institution approved this study. Formative research using mixed methods incorporated surveys, individual interviews, and one focus group with three stakeholder groups (food pantry staff, volunteers, and clients). All data were collected on site at a local county-wide Midwest food pantry.

#### *2.1. Pantry Context*

The food pantry in the current study is the largest food pantry within the county it's located serving roughly 13,000 residents in 2017 [34]. The county has a food insecurity rate of nearly 17% and overall poverty rate of 19% [34]. The pantry saw a 15% increase in overall client visits between 2017 and 2018 [35], with 51% of clients surveyed reporting having to skip meals between one and three times, on average, per week [31]. Demographically, over half of their clients (65%) identify as white and fall between the 18–64 age group (62%) [35]. Due to chronic disease concerns, with 62% of pantry clients surveyed reporting a household member with type 2 diabetes, the pantry has begun to offer health screenings on-site [35]. Additionally, the pantry offers a variety of nutrition programs including cooking and gardening classes, and an intensive culinary training program to encourage self-sufficiency among pantry clients [35]. The pantry utilizes seven full-time staff members and a fleet of volunteers.

#### *2.2. Sample*

Convenience sampling occurred focusing on three stakeholder groups (*n* = 47) (1) pantry staff (*n* = 7); (2) pantry volunteers (*n* = 10); and (3) pantry clients (*n* = 30). All staff currently employed by the pantry were included in the study, volunteers and pantry client participants were recruited until data saturation occurred. All participants were recruited in-person via direct communication with study staff during regularly scheduled pantry hours (M-F, 9 a.m.–5 p.m.). Inclusion criteria included individuals starting at age 18 to capture those of adult status and ending at age 75. This age range is representative of the majority age range of clients served (18–65) plus an extended age range (65–75) to capture the retired volunteer population. Additionally, stakeholder group classification (pantry staff, volunteer, or client), and ability to speak, read and write English were inclusion requirements. Participants incentives consisted of a "healthy eating goodie bag" containing a reusable grocery tote, cooking oil, one cooking utensil (wooden spoon, fork, or spatula), recipe cards, and informational brochures on various healthy eating topics. Only pantry clients were encouraged and received a "healthy eating goodie bag" upon completion of the study.

#### Demographic Characteristics

Table 1 displays pantry client (*n* = 30) demographics, and individual and household health status information. Majority of clients were Caucasian (80%, *n* = 24) and female (73%, *n* = 22). Disease prevalence was high with 60% (*n* = 18) reporting at least one chronic disease and 37% (*n* = 11) reporting more than one. Additional health status and demographic information is displayed in Tabe 1 below. Staff (*n* = 7) participant age ranged from 23 to 39, with majority of the participants (71%, *n* = 5) identifying as Caucasian/white, and two identified as mixed race. All staff work full-time, with years of experience ranging between one to six years. Lastly, Volunteer (*n* = 10) participants included individuals ages 18 to 79, primarily identifying as Caucasian (90%, *n* = 9), with one identifying as African American. Volunteer employment status ranged from full-time to retired.


**Table 1.** Pantry client demographic Characteristics and health status.


#### **Table 1.** *Cont.*

#### *2.3. Measures*

Primary data collection involved three investigator-designed surveys and corresponding interview guides using a combination of newly developed questions based on the current study's aims, and questions modified based on validated measures previously found in the literature. All survey measures were collected via hard copy, in-person, direct participant response. A researcher was present to answer participant questions.

#### 2.3.1. Client Survey Measures

The client survey included validated measures through questions on self-reported health [36] and the Behavioral Risk Factor Surveillance Survey nutrition and PA module measures [37]. Investigator-designed measures included categorical questions (yes or no) on individual and household chronic disease diagnosis (e.g., diabetes), and barriers to healthy eating (e.g., healthy foods are too expensive) and PA (e.g., I don't know enough about physical activity). Last, the survey asked participants to report individual demographic characteristics (race/ethnicity, gender, annual household income, employment status, and level of education). The survey consisted of 28 questions and the full details can be reviewed under Supplementary File S1.

#### 2.3.2. Client Semi-Structured Interviews

An investigator-designed moderator's guide, which corresponded with survey questions, guided semi-structured interviews. Sample questions included, "What are some of the challenges and barriers to choosing and cooking healthy options?" and "What current health issues are you and/or members of your household facing? Last, questions pertaining to intervention components included "What do you think are some critical characteristics of this program? (Probe: How often meetings are, time of day, days of the week, how long, educator characteristics, location, electronic vs. in-person)?" The moderator's guide consisted of 14 questions and the full details can be reviewed under Supplementary File S2.

#### 2.3.3. Volunteer/Staff Survey Measures

Volunteer and staff measures included an investigator-designed survey informed by the study aims and topics represented in the client survey. Example questions include categorical questions (often, sometimes, never) related to client engagement within the topics of health, nutrition and PA (e.g., "How often do you engage with clients about the cost of food?). Last, the survey asked participants to report individual demographic characteristics (race/ethnicity, gender, annual household income, employment status, level of education and number of years of service at the current food pantry). The survey consisted of 16 questions and the full details can be reviewed under Supplementary File S3.

#### 2.3.4. Volunteer/Staff Semi-Structured Interviews

Interviews consisted of an investigator-designed moderator's guide corresponding to the survey. Sample questions included, "What questions do clients most commonly ask about (a) Food/food products, (b) Nutrition, (c) Physical Activity, (d) Health (e) Programs/resources offered by the pantry" and "What important topics within nutrition, physical activity, and health should be covered in an intervention program?" The moderator's guide consisted of 8 questions and the full details can be reviewed under Supplementary File S4.

#### *2.4. Data Collection*

Participants completed a written informed consent prior to data collection. Collection occurred through in-person hard copy surveys completed by participants, individual semi-structured interviews with pantry clients and volunteers, and one focus group with pantry staff.

#### 2.4.1. Participant Surveys

Survey responses were collected from all three-stakeholder groups (staff, volunteers, and clients) immediately before conducting interview questions. Surveys were administered in hard copy using paper and pencil, and were completed independently by study participants. Study staff were available for participant support.

#### 2.4.2. Participant Interviews and Focus Group

Volunteer and client groups participated in follow-up individual semi-structured interviews, while staff participated in a single focus group during a routine staff meeting. All interviews and focus groups were semi-structured, immediately followed survey completion, and were located in a secure private room on-site at the food pantry. All correspondence was audio recorded with sessions lasting between roughly 30 to 60 min in length. A single investigator (the PI) with training and experience in qualitative methods and the interview protocol conducted interviews and took field notes. Member checks and debriefings occurred during interviews to ensure accuracy of participant statements and to increase trustworthiness [38].

#### *2.5. Data Analysis*

All data were reviewed and analyzed separately, then brought back together to find convergent themes across all sources and stakeholder groups. All survey responses were input into IBM SPSS Statistics 26 software for descriptive data analysis.

#### Interview/Focus group Analysis

All interviews were audio recorded and transcribed verbatim by the PI of the study. Once transcribed, a priori categories, based on categories within the semi-structured interview guide, directed the initial coding process and were combined with exploratory findings to generate final themes [39]. Last, data triangulation occured between the existing literature, stakeholder surveys and stakeholder interviews/focus group to informed research findings [39,40]. This process included two co-investigators of the research team.

#### **3. Results**

This section will provide a detailed description of each stakeholder group's results separately, followed by a joining of the data generating final convergent themes. Final convergent major themes include food and PA barriers, nutrition and PA literacy, health status and lifestyle, current pantry operations and adjustments and suggestions for health intervention programming.

#### *3.1. Client Results*

Client survey responses revealed low consumption of fruits and vegetables with over half (67%, *n* = 20) reporting zero to one servings of fruits per day, and 47% (*n* = 14) reporting zero to one servings of vegetables per day. Commonly reported healthy eating barriers include: healthy food being too expensive (40%, *n* = 12), not knowing enough about healthy cooking (37%, *n* = 11), not knowing enough about general nutrition to make healthy meals (30%, *n* = 9), and not knowing how to choose and store fresh produce (27%, *n* = 8). A high rate of participants (67%, *n* = 20) reported low PA between zero to 120 min per week. Common barriers preventing participants from engaging in regular PA, included having health conditions that restrict activity (30, *n* = 9), lack of enjoyment for PA (27%, *n* = 8), lack of access to a facility to engage in PA (23%, *n* = 7), and having a job that is physically demanding (20%, *n* = 6).

During client interviews, four themes emerged, including Food and PA barriers, Nutrition and PA literacy, Health Status and Lifestyle, and Suggestions for Health Intervention Programming. In the first major theme, participants reported things such as cost, and food preparation restrictions as leading roadblocks to improving nutrition. One participant reported, "Right now, I live in a camper out in the park, and I don't have electricity in it, so mostly it's the food banks, or going to get something that's cooked in the store. Unless I can get a fire going, so that limits me and what I can do." Barriers to PA, included mental and physical limitations and occupational restrictions. Occupational restrictions include sedentary jobs and lack of time due to multiple jobs.

The second major theme, Nutrition and PA Literacy is associated with general nutrition and PA education. Within nutrition, categories, such as cooking, specialty diets, and produce storage and preparation were noted. Within PA most feedback was focused on general strength exercises, and exercises for physical limitations. Participants also mentioned a desire for weight management education with statements like, "Losing weight. What I would need to do to really lose some weight. And not just do strange starving eating type of things. The healthy way to do it."

Third, the Health Status and Lifestyle theme corresponded with the depth of physical and mental illness across participants with one participant sharing "I have anxiety, depression, migraines, frontal lobe seizures, turrets, treated for blood clots, get treated for low vitamin B, Arthritis." Additionally, there were reports on impacts to lifestyle due to disease. These related to both positive impacts, such as disease translating to improvements in health behaviors, and negative impacts with connections to disease affecting quality of life in examples like "She took me off work for two months to see if we could get it under control [high blood pressure], so hopefully."

In the fourth major theme, Suggestions for Health Intervention Programming, pertained to programmatic and structural recommendations from pantry clients. Structurally, participants were interested in both electronic services and face-to-face services, as well as group and individual formats. They reflected on the idea of social support from both the health educator or coach, and other pantry client intervention participants in a group setting. Recommendations for intervention content included statements such as "It should be the holistic approach. Teaching people to eat better sooner, instead of waiting until the point of diabetes or the health issues." Last, participants demonstrated support and excitement for a health intervention program by stating comments, such as "I think this is a fabulous idea I think it is doable with a lot of your hard work and I look forward to you moving forward and changes ahead."

#### *3.2. Staff Results*

Staff survey responses demonstrated that within the category of health, staff reported often engaging with clients about health insurance (29%, *n* = 2) and local health services (29%, *n* = 2). Within nutrition, staff reported often engaging with clients about cost of food (86%, *n* = 6), quick meal options (57%, *n* = 4), and food restrictions (57%, *n* = 4). Within PA, staff indicated often engaging with clients about physical limitations (57%, *n* = 4).

Staff participated in a follow-up focus group instead of interviews to capture collaborative staff ideas as a part of a monthly staff meeting. Three themes emerged. Themes included Specialty Diet Questions, Pantry Operations, and Client Education. Specialty Diet questions included clients coming in with specific recommendations from medical providers with one staff member reporting, "I am finding more hyper specificity. [clients reporting] This is my diet, I have talked to my doctor, and they say I need to be eating these specific items, do you have any of those?"

Within the theme of Pantry Operations, staff proposed a variety of pantry operational changes that may assist clients with questions and food choice. This theme included creating general handouts, nutritional nudges, and increased meal kit options. Last, the Client Education theme, informed by direct client experience and observations, led to recommending general nutrition and PA education. For example, one staff member said the following: "Helping people understand how to be more realistic [portion size], my immediate thought goes to My Healthy Plate campaign."

#### *3.3. Volunteer Results*

Volunteer survey responses demonstrated that within the category of health, volunteers reported often communicating with clients about high blood pressure (30%, *n* = 3) and local health services (30%, *n* = 3). Volunteers reported sometimes engaging about unusual food items (50%, *n* = 5), building healthy meals (*n* = 4), and food storage (*n* = 4). Volunteers indicated never engaging with clients about PA in nearly all categories.

During semi-structured interviews, four themes emerged including Pantry Questions, Pantry Shopping Adjustments, Client Education, and Volunteer Training. Within the theme of Pantry Questions, volunteers highlighted frequent client questions related to either food products or preparation. One volunteer indicated: "Sometimes people will ask about what would be a good way to prepare this vegetable or meat," or pantry logistics "not too many questions other than how many points is this [food item]".

Volunteers offered recommendations for Pantry Shopping Adjustments addressing the topics of food products/preparation and pantry logistics. Recommendations included adding information for use and preparation of unusual produce and including simple recipes directly with these items. Major topics highlighted within the Client Education theme included general nutrition and PA guidelines with comments like, "Most people don't have a general understanding of nutrition," and shopping strategies, "Educating on how to effectively use their points. Some people only have 10 points and they get 4 sandwiches and that is going to last you a max of 2 days."

The Volunteer Training theme emphasized conflicting opinions. Regarding volunteer training, some volunteers indicated interest in receiving training related to "Food stamp options. How or where; opportunities to talk about options for food," with other volunteers indicated a lack of interest in further training with rationales like, "A lot of us are retired and not wanting to fill that role [health specific volunteer role]."

#### *3.4. Final Convergent Major Themes*

There were five identified final convergent major themes including Food and PA Barriers, Nutrition and PA Literacy, Health Status and lifestyle, Current Pantry Operations and Adjustments, and Suggestions for Health Intervention Programming.

Food and PA Barriers, include identification of life circumstance that make healthy eating and PA difficult among pantry clients. Barriers that were reported included cost of food, produce storage and self-life, physical limitations to exercise, and the perception that PA is a privilege based on social status.

Nutrition and PA Literacy, the second theme, pertains to gaps in knowledge about healthy eating, selection and preparation of foods, PA recommendations based on limitations both identified by the clients through personal experience, and volunteers and staff based on client interactions and questions.

Similarly, clients' personal reports, and volunteer and staff interactions with clients demonstrate how food insecurity and limitations due to disease influence clients lives under the Health Status and Lifestyle major theme. This included reporting on how much disease clients were experiencing daily and coping strategies such as seeking out dietary recommendations from staff at the pantry.

The fourth major theme, Current Pantry Operations and Adjustments, relates to volunteer and staff experience with the current climate within the pantry associated with nutrition and PA among clients, and ideas for adjustments to create a more informed and positive experience. This included ideas for inclusion of nutrient information in meal kits and throughout the pantry, as well as guidance on how to use their pantry points and potential training opportunities for volunteers.

Suggestions for Health Intervention Programming highlights the perspectives from all three stakeholder groups related to intervention program components consisting of nonjudgmental, supportive, coaching, with the inclusion of PA and nutrition education, and support for hosting such a health intervention program in the pantry setting. A summary of these final convergent themes and corresponding client, volunteer, and staff quotes can be found in Table 2.

**Table 2.** Final convergent Major Themes and Quotes.



**Table 2.** *Cont.*

Note: C = Client quote, S = Staff quote, V = Volunteer quote.

#### **4. Discussion**

Consistent with previous research, pantry clients reported high levels of individual and household chronic diseases [40,41], which are compounded by client reported gaps in doctors' visits and health insurance coverage [41]. As more research connects the dots between food insecurity and insufficient medical care, organizations work to provide solutions in both pantry and clinical settings. Within clinical settings, screenings, referrals, and connecting patients with emergency food services is becoming a more common practice [19]. Additionally, interventions in the form of food pharmacy programs are connecting patients with food and nutrition resources within medical facilities [42,43]. Medical

interventions are surfacing and have shown promise in food pantry settings [44]. Within the pantry setting, disease specific interventions (e.g., diabetes management interventions) have shown success among pantry clients [22,45]. However, disease specific interventions leave an unmet gap in serving pantry clinics with co-morbidities outside of the scope of that intervention. Additionally, little is known about targeting nutrition and PA behaviors in a holistic health intervention framework to address chronic disease among food pantry users. Health coaching frameworks with the use of motivational interviewing techniques have demonstrated effectiveness in chronic illness management [46]. Only one study was found with the employment of health coaching as a component of a more comprehensive intervention model within the food pantry [28]. By providing interventions around a health coaching framework, using a combination of health education and motivational interviewing, health coaches can address a broader range of clients providing clients with both nutrition and physical activity education, and social support, thereby increasing self-efficacy [47].

All three-stakeholder groups identified poor nutrition and PA literacy as a contributor to poor health outcomes. Research has shown low food and nutrition literacy may contribute to food insecurity in developing countries [15], while health literacy and selfefficacy have been found to predict food label use, which is positively related with diet quality [16]. As health education contributes to relationship building between health coaches and patients [30], further education through health intervention programming using these program components within the pantry setting could lead to improvements in food security status and diet quality [15,16]. The lack of skills in preparing fresh produce and irregularity of food supply have been noted in the literature as pantry client barriers to utilizing fresh produce [48]. The current study found consistencies with all three-stakeholder groups reporting barriers in using and preparing unusual produce. Interventions targeting weekly cooking classes within a six-week format have been shown to improve diet quality and decrease food cost within the pantry setting [21] by teaching food preparation skills. Little is known about using a similar program structure targeting PA, and further a holistic program targeting both, PA and nutrition as a comprehensive chronic disease health intervention program.

Staff report more "hyper specificity" in the types of foods clients are requesting due to doctor recommendations through food prescriptions, yet neither staff nor volunteers have the expertise to address these client needs. Thus, trained health educators and/or health coaches could help fill this void [49]. Health professionals could provide services such as pantry shopping assistance, food item identification, recipes, and food skills training that match specific client needs [50]. Due to this gap in expertise among current volunteers and staff, health intervention programming within the food pantry setting would require, either a hired staff member, additional recruitment and training of volunteers, and/or a partnership with local health organizations.

Nutrition and PA knowledge gaps across a diverse range of categories were recognized between all three stakeholder groups. This ranged from healthy cooking on a budget to exercising with limitations, giving direction to health content as an educational component to health intervention programming. Clients advocated for a positive, non-judgmental climate, entailing goal setting and accountability components. This is consistent with elements used within health coaching models that are linked to improvements in health lifestyle behaviors [49]. Health coaching can combine traditional health education strategies with motivational interviewing techniques to increase knowledge, skills, individual motivation, autonomy, and self-efficacy, promoting changes in health behaviors [29]. Last, support for health intervention programing was generated by all three stakeholder groups, particularly among the priority population. By using a formative community participatory approach [33] to gain support and develop intervention components, there will be a greater chance for intervention success and adoption by pantry clients during implementation.

#### *Study Limitations*

The current study only included one county-wide Midwest food pantry with a small sample of the key stakeholders creating generalization limitations. Additionally, the tools included in the study were designed by an investigator and were not first tested for reliability or validity.

#### **5. Conclusions**

High rates of chronic disease combined with low nutrition and PA literacy among pantry clients demonstrates the need to address health behaviors. In this study, each stakeholder group provided program component recommendations and indicated support for a health intervention program within the food pantry setting. Further research piloting the design and implementation of such a program in the pantry setting is needed. More specifically, design and implementation of a more holistic approach incorporating both nutrition and PA aimed at individual needs and disease prevention. The results will be used to prepare phase two, design and implement a health intervention program within a county-wide Midwest food pantry. Furthermore, key highlights from this research work that could be transferable into the field include:


**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/nu13051584/s1, File S1: Shaping a Food Pantry Health Intervention-Client Survey, File S2: Shaping a Food Pantry Health Intervention-Client Interview Questions, File S3: Shaping a Food Pantry Health Intervention-Staff/volunteer Survey, File S4: Shaping a Food Pantry Health Intervention-Staff/volunteer Interview Questions.

**Author Contributions:** K.F., primary investigator (PI), co-designed instruments, collected and analyzed data, and co-wrote manuscript; S.H. advisor to the project, co-PI, co-designed instruments, and co-wrote manuscript. Both authors have read and agreed to the published version of the manuscript.

**Funding:** This research work was supported by the University of Kansas office of Graduate Studies Summer Research Scholarship. The article processing charges related to the publication of this article were supported by The University of Kansas (KU) One University Open Access Author Fund sponsored jointly by the KU Provost, KU Vice Chancellor for Research & Graduate Studies, and KUMC Vice Chancellor for Research and managed jointly by the Libraries at the Medical Center and KU-Lawrence.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of The University of Kansas (STUDY00144140, approved 11 June 2019).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank all of the individuals that were willing to participate and tell their stories. Additionally, we would like to thank Elizabeth Keever and the staff at the Just Food food bank for providing secondary data and hosting the project work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Jae Hyun Lee 1,†, Ae Wha Ha 2,†, Woo Kyoung Kim <sup>2</sup> and Sun Hyo Kim 3,\***


**Abstract:** The purpose of this study was to examine the combined effects of milk intake and physical activity on bone mineral density in adolescents. This study was conducted using data from the 2009–2011 Korea National Health and Nutrition Examination Survey (KNHANES), which provided measurements of bone mineral density (BMD) in addition to basic health-related data. This study included 1061 adolescents aged 13 to 18 years (557 males and 504 females) whose data on milk intake and participation time in moderate to vigorous physical activity were available. BMD was measured by dual-energy X-ray absorptiometry (DXA). Milk intake was assessed using the 24-h recall method, and the levels of physical activity were examined using a questionnaire. The physical activity questions of 2009–2011 KNHANES were based on the Korean version of the International Physical Activity Questionnaire (IPAQ) short form. The subjects were classified into four groups according to milk intake and physical activity level: no milk intake + low-level physical activity group (MnoPlow), no milk intake + high-level physical activity group (MnoPhigh), milk intake + low-level physical activity group (MyesPlow), and milk intake + high-level physical activity group (MyesPhigh). The results of partial correlation controlling for age, body mass index (BMI), and energy intake showed that the BMD variables were associated significantly with physical activity in both males and females. Among males, the MnoPlow group had the lowest BMD in all BMD variables, showing a significant difference from the high-level physical activity groups (MnoPhigh, MyesPhigh) by multiple logistic regression analysis. Among females, the MyesPhigh group showed a significantly higher lumbar BMD value than the other groups. The MnoPlow group had approximately 0.3 to 0.5 times lower odds ratio for median or higher BMD values, compared to MyesPhigh group. These results show that milk intake and physical activity have a combined effect on BMD, and suggest that to achieve healthy bone growth, it is important to encourage both moderate to vigorous physical activity and milk intake during adolescence.

**Keywords:** bone mineral density; milk intake; physical activity; adolescence

#### **1. Introduction**

Bones are major organs that determine the body's physique and perform various functions, such as protection of internal organs, mineral storage, and blood cell formation. Bone ossification begins in the prenatal period and almost reaches the total peak bone mass by the end of teenage growth [1,2]. During puberty, the bone mineral accrual rate reaches a peak, and approximately one quarter of the total bone minerals of adults are accumulated within two years at this time [3]. In Koreans, the peak bone mass of the femoral neck and total hip is achieved around the age of 20, and the greatest increase in

**Citation:** Lee, J.H.; Ha, A.W.; Kim, W.K.; Kim, S.H. The Combined Effects of Milk Intake and Physical Activity on Bone Mineral Density in Korean Adolescents. *Nutrients* **2021**, *13*, 731. https://doi.org/10.3390/nu13030731

Academic Editor: Carlos Vasconcelos

Received: 29 January 2021 Accepted: 19 February 2021 Published: 25 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

lumbar bone mineral density (BMD) occurs between 11–13 years of age in females and 12–14 years of age in males [4,5]. Hence, adolescence is a very important period of life for the formation of healthy bones. People who fail to achieve optimal peak bone mass and strength during childhood and adolescence have been reported to be more likely to develop osteoporosis later in life [6,7], and low BMD is associated with a higher risk of fractures even in healthy children and adolescents, just as it is a risk factor for fracture in adults with osteoporosis [8,9]. Therefore, to obtain the benefits of healthy bones for life, appropriate interventions are required to help children and adolescents build healthy and strong bones during the growth period.

Peak bone mass, which means the maximum accumulation of bone mineral content, is determined by genetic and environmental factors. Environmental factors include physical activity, sedentary lifestyle, and dietary factors such as milk intake [10–14]. The consumption of milk and dairy products helps maximize the bone mineral content during puberty, which is the second period of the growth spurt [15,16]. Milk has a high calcium content, and calcium in milk has high digestibility and bioavailability [17]. This is because milk contains lactose, vitamin D, and peptides promoting calcium absorption, which help the body to absorb calcium, and contains calcium and phosphorus in an appropriate ratio that increases the rate of calcium absorption [18]. The consumption of milk and dairy products during the growth period can be a good source of calcium as well as energy, macronutrients, and micronutrients important for the growth and development of children and adolescents [13,14,17–20]. A four-year follow-up study of 19,991 children in eight European countries reported that the consumption of milk and dairy products (yogurt and cheese) as snacks was associated with better diet quality [21]. Therefore, the daily consumption of milk and dairy products for children and adolescents can be a good strategy for maintaining a balanced diet during the growth period.

Mechanical stimulation is an important determinant of bone growth and formation. Exercises that provide physical and physiological stimulation improve muscular strength, cartilage preservation, and bone remodeling [22,23], and they have a positive effect on increasing BMD [24,25]. Most studies on the effects of weight-bearing exercises on the accumulation of bone mineral content during childhood and adolescence reported that such exercises have positive effects, and this phenomenon is particularly pronounced in early puberty [26]. The performance of the activities of high intensity or impact and participation in sports activities have also found to have a positive effect on the BMD or cortical bone size [25,27,28].

As described above, various studies have been conducted on the effects of milk intake or physical activity alone on BMD during the growth period. Limited studies suggested an important interaction between physical activity and the intake of dietary calcium, not milk intake, to increase bone mass. When physical activity and calcium intake were combined, bone density formation was greater than either physical activity or calcium intake alone [29–31]. In addition, those studies have been conducted in preschool or school children. Therefore, this study aims to evaluate the combined effects of milk intake and physical activity on BMD during adolescence. We hypothesized that adolescents who had a high level of physical activity and consumed milk would have higher BMD than those who had a low level of physical activity and did not consume milk.

#### **2. Methods**

#### *2.1. Data Collection*

This study examined the relationship of BMD with milk intake and physical activity using 2009 to 2011 data from the fourth (2007–2009) and fifth (2010–2012) Korea National Health and Nutrition Examination Survey (KNHANES). The KNHANES survey began in 1998 and has been conducted annually, with BMD measurements conducted from July 2008 to May 2011. Data of 1731 people aged 13–18 (1198 males and 812 females) who underwent BMD measurements using dual energy X-ray absorptiometry (DXA) were collected. Subjects with missing data regarding milk intake or physical activity and those whose data showing extreme outliers were excluded. Ultimately, the data of 1061 people (557 males and 504 females) were included in the final analysis.

This study used the data from the KNHANES approved by the Institutional Review Board of the Korea Centers for Disease Control and Prevention (2009-01CON-03-2C, 2010- 02CON-21-C, 2011-02CON-06-C), which was conducted after receiving an exempt determination from the Institutional Review Board of Kongju National University (KNU\_IRB\_2020-65).

#### *2.2. Milk Intake*

The analysis of milk intake was conducted using data from a dietary intake survey by the 24-h recall method among the raw data sets of the KNHANES. According to the food-group classification standard codes presented in the guidelines on the use of the KNHANES data, the food name of 'milk' among the secondary food names was first classified. The type of milk consumed was then examined using the primary food names, and the participant was classified as a person consuming milk when the type of milk consumed was white milk.

#### *2.3. Physical Activity*

The level of physical activity was calculated by the time of moderate or vigorous physical activity performed per week (number of days per week (days/week) × activity time (minutes/day)). The questions on moderate and vigorous physical activity in KNHANES were as follows:

	- # On how many days in the past week did you perform moderate physical activity that made you feel slightly more tired than usual, or during which you felt a little short of breath for at least 10 min?
	- # On the days when you performed moderate physical activities, how many minutes per day did you usually perform them?

Examples of moderate physical activities: vocational and physical activities, such as slow swimming, doubles tennis, volleyball, badminton, table tennis, moving, or carrying light items.

	- # On how many days in the past week did you perform vigorous physical activity that made you feel much more exhausted than usual, or during which you felt very short of breath?
	- # On the days when you performed vigorous physical activities, how many minutes per day did you usually perform them?

Examples of vigorous activities: vocational and physical activities, such as jogging or running, mountain climbing, fast cycling, fast swimming, soccer, basketball, jumping rope, squash, singles tennis, moving or carrying heavy objects.

In this study, based on the guidelines on physical activity presented by the Ministry of Health and Welfare for calculating weekly physical activity time, it was assumed that one minute of vigorous physical activity is equal to two minutes of moderate physical activity [32]. Using this guideline, the total physical activity time was calculated by converting vigorous physical activity time to moderate physical activity time. The physical activity questions of the 2009–2011 KNHANES were based on the Korean version of the International Physical Activity Questionnaire (IPAQ) short form.

#### *2.4. Subject Grouping*

The subjects were divided into the milk intake group (Myes group: milk intake >0 g/day) and the no milk intake group (Mno group: milk intake = 0 g/day). For physical activity grouping, the median of the weekly participation time of moderate-to-vigorous physical activity was calculated by converting vigorous physical activity times to moderate physical activity time. Subjects with a value below the median were classified as the

low-level physical activity (Plow) group. Those with a value equivalent or higher than median were classified as the high-level physical activity (Phigh) group. Groups can also be classified according to the satisfaction of the physical activity guidelines of 60 min of moderate-to-vigorous activities every day. However, only 5.1% of men and 1.9% of women actually meet these criteria (420 min per week), making it impossible to compare the groups using statistical analysis. Therefore, in this study, groups were classified using the median of converted physical activity time per week. The physical activity questions of the 2009–2011 KNHANES were based on the Korean version of the International Physical Activity Questionnaire (IPAQ) short form.

By combining these two classifications, the subjects were finally classified into four groups according to milk intake and physical activity level: no milk intake + low-level physical activity group (MnoPlow), no milk intake + high-level physical activity group (MnoPhigh), milk intake + low-level physical activity group (MyesPlow), and milk intake + high-level physical activity group (MyesPhigh).

#### *2.5. Bone Mineral Density*

BMD was measured using dual-energy X-ray absorptiometry (DXA; DISCOVERY-W fan-beam densitometer Hologic Inc., Bedford, MA, USA) and each subject's whole body, lumbar spine, and femur were scanned. When measuring the lumbar spine, a lumbar positioner was used to reduce spinal lordosis, and the lumbar spine was positioned straight so as to be in line with the vertical central axis of the image. The image included the midsection of T12 and L5, and to determine whether the lumbar spine was correctly positioned, it was checked whether the 12th rib and iliac crest were visible in the image, and whether the intervertebral disc of L4–L5 passed in line with the iliac crest. When measuring the femur, the angle of the leg was adjusted so that the femoral shaft was positioned straight in line with the vertical central axis of the image. When measuring DXA, it was checked if there were any artifacts such as coins or keys, buttons, wires, jewelry, or metal objects in the pocket. Among the various DXA measurement indices, total body, femur, femur neck, and lumbar spine (L1–4) BMD were analyzed statistically, and total body BMD was calculated using the BMD values of the whole body except for head BMD.

#### *2.6. Statistical Analysis*

The data of the KNHANES were collected not by simple random sampling but by stratified multistage probability sampling. Hence, the weight, strata (KSTRATA), and cluster (primary sampling unit, PSU) were included in the analysis. The sociodemographic characteristics of the subjects were expressed as frequency and percentage, and differences in distribution between the groups were compared using PROC SURVEYFREQ (chi-squared test). For the continuous variables, descriptive statistical analysis was performed to calculate the mean and standard error. Partial correlation analysis was performed to identify the relationship of BMD with physical activity and milk intake while controlling for age, body mass index (BMI), and energy intake. The differences in explanatory variables between the four groups (MnoPlow, MnoPhigh, MyesPlow, and MyesPhigh groups) were analyzed by PROC SURVEYREG analysis after adjusting for age, BMI, and energy intake. For a post-hoc test of the differences among the groups, the *p*-values were assessed using a Bonferroni test considering the design effect of complex sampling design. The PROC SURVEYLOGISTIC analysis was performed (after adjusting for age, BMI, and energy intake) to calculate the risk ratio of each BMD index of the three groups compared to the reference group (the MyesPhigh group). The analysis results were expressed as an odds ratio (OR) and 95% confidence interval (CI).

All statistical analyses were conducted using SAS version 9.4 (Statistical Analysis System, SAS Institute, Cary, NC, USA), and *p* values <0.05 were considered significant.

#### **3. Results**

Table 1 lists the sociodemographic characteristics of the subjects. Significant differences in school year and gender were observed among the four groups. Of the 1061 subjects, high-school students (57.0%) comprised a larger proportion than middle-school students (43.0%), and the difference in the percentage between middle school and high school was the largest in the MnoPlow group. The subjects consisted of 557 males (52.5%) and 504 females (47.5%), and the difference in the percentage between males and females was the largest in the MyesPhigh group (68.0% in males vs. 32.0% in females). Therefore, the analysis was conducted separately for males and females, and data analysis was conducted by controlling for age. There were no significant differences in the distribution of income levels or residential areas.


**Table 1.** Sociodemographic characteristics of the subjects.

<sup>1</sup> MnoPlow: no milk intake + low physical activity; MnoPhigh: no milk intake + high physical activity; MyesPlow: milk intake + low physical activity; MyesPhigh: milk intake + high physical activity (Plow: physical activity less than 50th percentile; Phigh: physical activity of 50th percentile or more); <sup>2</sup> *p*-value by chi-square test. <sup>3</sup> *n* (%).

> Regarding the distribution of daily milk intake among subjects, the milk intake ranged from 0 to 1484 mL/day among males and from 0 to 848 mL/day among females. Approximately 55.4% of males and 62.6% of females did not consume milk, and in both males and females, the proportion of people drinking 200–400 mL/day was highest, accounting for 24.2% and 20.9%, respectively. According to the dietary reference intakes for Koreans (KDRIs), it is recommended that adolescents drink two glasses (400 mL) of milk a day [33], and the percentage of adolescents consuming the recommended amount or more of milk was 14.7% in males and 8.1% in females; females tended to drink less milk than males (Figure 1).

**Figure 1.** Distribution of daily milk intake.

The converted time of physical activity ranged from 0 to 780 min/week among males and 0 to 600 min/week among females. The weekly participation time of moderate to vigorous physical activity except for walking was 0 min in 31.2% of males and 49.7% of females. For both males and females, the proportion of adolescents showing a converted physical activity time of 60–120 min per week was highest, accounting for 14.6% and 16.3%, respectively. The proportion of those participating in physical activity for 300 min or more per week was 12.9% in males and 4.9% in females. Hence, the level of participation in physical activity was significantly lower among females than among males (Figure 2).

Table 2 lists the milk intake and physical activity time of each group. Because the subjects were classified according to whether they consumed milk or not, the daily milk intake of the no milk intake groups (MnoPlow, MnoPhigh) was 0 mL. In the milk intake groups, the milk intake levels for males in the MyesPlow and MyesPhigh groups were 360.1 mL and 349.0 mL, respectively. For females, the milk intake levels in the MyesPlow and MyesPhigh were 280.8 mL and 278.8 mL, respectively. There was a large difference in the physical activity time between the high-level physical activity groups (MnoPhigh, MyesPhigh) and low-level physical activity groups (MnoPlow, MyesPlow). For groups with high physical activity, weekly physical activity time among males was 227.3 min for MnoPhigh and 230 min for MyesPlow. Among females, the physical activity time in the MnoPhigh and MyesPhigh groups was 130.7 min and 175.9 min per week, respectively. The weekly physical activity time among males was 11.8 ± 1.9 min for MnoPlow and 26.3 ± 3.7 min for MyesPlow.



<sup>1</sup> MnoPlow: no milk intake + low physical activity; MnoPhigh: no milk intake + high physical activity; MyesPlow: milk intake + low physical activity; MyesPhigh: milk intake + high physical activity (Plow: physical activity less than 50th percentile; Phigh: physical activity of 50th percentile or more); <sup>2</sup> *<sup>p</sup>*-value by PROC SURVEYREG adjusted for age, body mass index, and energy intake; <sup>3</sup> Mean <sup>±</sup> SE; 4 abc: values with different alphabets in the same row are significantly different at *p* = 0.05 by a Bonferroni test.

Table 3 lists the general characteristics of each of the four groups classified according to milk intake and the level of physical activity. In both males and females, the mean age was highest in the MnoPlow group and lowest in the MyesPhigh. For BMI, there was a significant difference only in females, showing that the MnoPlow and MyesPlow groups with low levels of physical activity had a significantly lower mean BMI than the groups with high levels of physical activity. However, the mean BMI of these four groups were not largely different from 20.9 kg/m <sup>2</sup> , the median BMI (50th percentile) of 15.4-year-old boys in the 2017 Korean National Growth Charts for Children and Adolescents published by the Ministry of Health and Welfare [34]. For reference, the BMI corresponding to overweight (from the 85th percentile to less than the 95th percentile) for a 15.4-year-old Korean girl is 23.7~25.5 kg/m <sup>2</sup> , and the BMI corresponding to obesity (95th percentile or more) is 25. 5 kg/m <sup>2</sup> or more [34].

**Gender Variables MnoPlow <sup>1</sup> MnoPhigh MyesPlow MyesPhigh** *p***-Value <sup>2</sup> Total** Male Age (year) 15.9 <sup>±</sup> 0.2 3 a 4 15.6 <sup>±</sup> 0.2 ab 15.3 <sup>±</sup> 0.2 <sup>b</sup> 15.1 <sup>±</sup> 0.2 <sup>b</sup> 0.005 15.5 <sup>±</sup> 0.1 Height (cm) 171.0 <sup>±</sup> 0.9 172.0 <sup>±</sup> 0.8 169.9 <sup>±</sup> 0.9 170.4 <sup>±</sup> 0.7 NS 5 0.416 171.2 <sup>±</sup> 0.4 Weight (kg) 61.3 <sup>±</sup> 1.2 64.1 <sup>±</sup> 1.3 62.5 <sup>±</sup> 1.7 62.3 <sup>±</sup> 1.4 NS 0.745 62.5 <sup>±</sup> 0.6 BMI (kg/m<sup>2</sup> ) <sup>6</sup> 20.9 <sup>±</sup> 0.4 21.8 <sup>±</sup> 0.4 21.1 <sup>±</sup> 0.5 21.4 <sup>±</sup> 0.4 NS 0.307 21.2 <sup>±</sup> 0.2 %Fat (%) 20.2 <sup>±</sup> 0.7 21.3 <sup>±</sup> 0.7 20.0 <sup>±</sup> 0.8 22.2 <sup>±</sup> 0.9 NS 0.425 20.8 <sup>±</sup> 0.4 Female Age (year) 15.9 <sup>±</sup> 0.2 <sup>a</sup> 15.3 <sup>±</sup> 0.2 bc 15.3 <sup>±</sup> 0.2 bc 14.8 <sup>±</sup> 0.2 <sup>b</sup> <0.001 15.4 <sup>±</sup> 0.1 Height (cm) 159.7 <sup>±</sup> 0.6 160.4 <sup>±</sup> 0.5 160.4 <sup>±</sup> 0.7 160.0 <sup>±</sup> 0.6 NS 0.551 160.2 <sup>±</sup> 0.3 Weight (kg) 53.0 <sup>±</sup> 0.8 56.2 <sup>±</sup> 1.1 52.5 <sup>±</sup> 1.1 55.3 <sup>±</sup> 1.7 NS 0.890 54.1 <sup>±</sup> 0.6 BMI (kg/m<sup>2</sup> ) 20.8 <sup>±</sup> 0.3 <sup>a</sup> 21.8 <sup>±</sup> 0.3 <sup>b</sup> 20.3 <sup>±</sup> 0.3 <sup>a</sup> 21.6 <sup>±</sup> 0.6 <sup>b</sup> 0.036 21.0 <sup>±</sup> 0.2

**Table 3.** Physical characteristics of subjects.

<sup>1</sup> MnoPlow: no milk intake + low physical activity; MnoPhigh: no milk intake + high physical activity; MyesPlow: milk intake + low physical activity; MyesPhigh: milk intake + high physical activity (Plow: physical activity less than 50th percentile; Phigh: physical activity of 50th percentile or more); <sup>2</sup> *<sup>p</sup>*-value by PROC SURVEYREG adjusted for age, body mass index (BMI) and energy intake; <sup>3</sup> Mean <sup>±</sup> SE; 4 abc: values with different alphabets in the same row are significantly different at *p* = 0.05 by Bonferroni test; 5 NS: not significant; <sup>6</sup> BMI = weight(kg)/height(m<sup>2</sup> ).

%Fat (%) 31.9 <sup>±</sup> 0.5 33.6 <sup>±</sup> 0.6 31.8 <sup>±</sup> 0.6 33.0 <sup>±</sup> 1.0 NS 0.195 32.7 <sup>±</sup> 0.4

There was no difference in body fat (%) between groups in both male and females. Regarding the percentage of body fat in each group, the lowest and highest mean values were 20.0% and 22.2% among males and 31.8% and 33.6% among females. For reference, the mean body fat percentages of the male groups correspond to the 50–75th percentile of the percent body fat of Korean male adolescents, and the mean body fat percentages of female groups correspond to the 25–75th percentile of Korean female adolescents [35].

In order to identify the association of physical activity and milk intake with BMD, a partial correlation analysis for each gender group was conducted while controlling for age, BMI, and energy intake (Table 4). The results of this analysis showed that milk intake had no significant correlation with BMD. On the other hand, physical activity was found to have a weak but significant correlation with total body, femur, femur neck, and lumbar BMD.

Table 5 lists the results of comparative analysis of BMD among the four groups. Among males, there was a significant difference among the groups in all BMD variables, and the MnoPlow group, the group of adolescents who did not consume milk and had a low level of physical activity, had a significantly lower BMD than the MnoPhigh and MyesPhigh groups, which had a high level of physical activity. The BMD values of the MnoPlow group were lower than the median BMD value among 15-year-old Korean boys and higher than the 10th percentile [4]. In the case of females, there was a significant difference among the groups only in lumbar BMD. The MyesPhigh group, the group of females who consumed milk and had a high level of physical activity, showed a significantly higher lumbar BMD value of 0.931 (g/cm <sup>2</sup> ) than the other groups (MnoPlow: 0.902, MnoPhigh: 0.900, MyesPlow: 0.898). For reference, the median lumbar BMD value among 15-year-old Korean girls was 0.875 g/cm <sup>2</sup> [4].


**Table 4.** Relationships of bone mineral density with milk intake and physical activity.

<sup>1</sup> BMD: bone mineral density; <sup>2</sup> *p*-value by partial correlation controlled by age, body mass index, and energy intake.



<sup>1</sup> BMD: bone mineral density; <sup>2</sup> MnoPlow: no milk intake + low physical activity; MnoPhigh: no milk intake + high physical activity; MyesPlow: milk intake + low physical activity; MyesPhigh: milk intake + high physical activity (Plow: physical activity less than 50th percentile, Phigh: physical activity of 50th percentile or more); <sup>3</sup> *p*-value by PROC SURVEYREG adjusted for age, body mass index, and energy intake; <sup>4</sup> Mean <sup>±</sup> SE; 5 abc: Values with different alphabets in the same row are significantly different at *<sup>p</sup>* = 0.05 by Bonferroni test; 6 NS: not significant.

> Table 6 lists the odds ratio and confidence interval (CI) for the 50th or higher percentile of the BMD value in each BMD variable for each group compared to the MyesPhigh group. Among males, the MnoPlow group had significantly lower odds ratio for the 50th percentile or higher of the BMD value than the MyesPhigh group in all BMD variables. More specifically, the MnoPlow group was 0.317 times less likely to have the 50th or higher percentile of total body BMD value than the MyesPhigh group. For femur, femur neck, and lumbar BMD, the MnoPlow group had 0.289, 0.512, and 0.493 times lower odds ratio for the 50th or higher percentile of the BMD compared to the MyesPhigh group. In other words, the ratio of individuals with a median or higher BMD was significantly lower among the males who did not drink milk and had a low level of physical activity than the males who consumed milk and had a high level of physical activity. Among the females, the MnoPlow group and MyesPlow group were 0.433 and 0.434 times less likely, respectively, to have the 50th or higher percentile of lumbar BMD than the MyesPhigh group.


**Table 6.** Odds ratios on the bone mineral density according to the combination of milk intake and physical activity.

<sup>1</sup> CI: confidence interval; <sup>2</sup> BMD: bone mineral density; <sup>3</sup> MnoPlow: no milk intake + low physical activity; MnoPhigh: no milk intake + high physical activity; MyesPlow: milk intake + low physical activity; MyesPhigh: milk intake + high physical activity (Plow: physical activity less than the 50th percentile; Phigh: physical activity of the 50th percentile or more); <sup>4</sup> \*: *p*<0.05 by PROC SURVEYLOGISTIC.

#### **4. Discussion**

Adolescence is a very important period for lifelong bone health. Several studies have reported that the factors that positively affect the increase in bone mineral content and density have greater effects during this period than in adulthood, and that the effects of such factors continue into adulthood [2,3,15,16,36,37].

The two methods for building strong bones or improving bone strength are ingesting sufficient nutrients related to the bone matrix or bone metabolism and applying appropriate mechanical stimulation to the bones. Typically, when the former method is used, people consume milk, which has a high calcium content and high digestibility and bioavailability of calcium. Weight-bearing physical activities are performed when the latter method is used. Consequently, the study was designed to examine the combined effects of milk intake and physical activity on BMD.

In a partial correlation analysis controlling for age, BMI, and energy intake, physical activity had a significant positive correlation with total, femur, femur neck, and lumbar BMD in both males and females. Physical activity has beneficial effects on bone health in all age groups, including adolescents. In particular, bone mineral content is higher among children and adolescents participating in activities involving the exertion of high impact force than among those who participate in non-weight bearing exercises, such as swimming, or in low-impact activities, such as walking [38,39]. Therefore, activities involving high ground-reaction forces, such as jumping, skipping, and running, are recommended as exercises for strengthening the bones during the growth period [40,41]. A cross-sectional analysis of the relationship between physical activity and hip BMD in 724 adolescents found that high impact (>4.2 g) activities, such as jumping and running (speeds>10 km/h), were associated with hip BMD, but moderate impact activity, such as jogging, had little effect [25]. However, the physical activity variable analyzed in this study was the time of participation in moderate to vigorous physical activities. The physical activities with moderate intensity examined in this study included sports, such as slow swimming, doubles tennis, badminton, and table tennis. Walking was excluded from the analysis because it was examined separately with a different format. Given these facts, it seems that the low correlation between physical activity and BMD might be related to the type and intensity of physical activity analyzed in this study. Nevertheless, physical activity was consistently related to the BMD variable in both males and females.

On the other hand, milk intake and BMD had no significant correlation, which is inconsistent with previous studies reporting a quantitative relationship between the intake of milk and dairy products and bone mineralization. Several studies on the relationship between the intake of calcium, vitamin D, and dairy products and bone frailty during growth have reported conflicting or inconsistent results [42]. In this study, calcium and vitamin D intakes were not included as control variables. The reason is that it was confirmed that the calcium intake of Korean adolescents was very low, and milk was the major source of calcium. Besides, when the bone mineral density variable was analyzed using calcium as a parameter, the same result was obtained as from using milk intake. Also, vitamin D intake was excluded from the control variable as there was no significant difference between groups. There might be a threshold in the expression of the effect of calcium intake. Eating above a certain level of calcium does not affect the bone mass significantly but eating less than this can lead to an inadequate balance [1]. In addition, the effect of nutritional intake may vary depending on the nutritional status of the subjects. In cases where the intake of minerals or high-quality protein may be insufficient, the subjects may show a distinct increase in bone growth after the supply of dairy products [1,43]. On the other hand, Ren et al. reported that children with a good nutritional status did not show a clear positive correlation between major bone nutrients and bone outcomes compared to children with nutritional deficiencies [44]. In considering the results of this study that concern the relationship between milk intake and BMD, it is also necessary to consider that the overall milk intake level of the subjects was low, with an average daily milk intake of less than one glass (200 mL), and that 58.8% of subjects did not drink any milk. Therefore, additional studies will be needed to investigate the relationship between milk intake and BMD considering the distribution of milk intake and the basic nutritional status of subjects.

No linear relationship was observed between milk intake and BMD, but physical activity and milk intake had a statistically significant combined effect on BMD. Among males, the MnoPlow group had the lowest BMD in all BMD variables, showing the statistical difference from the groups with a high level of physical activity, the MnoPhigh group and the MyesPhigh group. Among females, the MyesPhigh group had a significantly higher lumbar BMD than the other groups.

In particular, an analysis of the odds ratios of male subjects showed that those who did not consume milk and had a low level of physical activity (MnoPlow) were significantly less likely to have a high BMD than those who consumed milk and had a high level of physical activity (MyesPhigh). Specifically, the MnoPlow group was approximately 0.5 times less likely to have a high femur neck BMD and lumbar BMD and was approximately 0.3 times less likely to have high BMD for the total body and the femur than the MyesPhigh group.

These results suggest that milk intake and physical activity have combined effects in strengthening bones. Branca et al. (2001) reported that bone anabolism could be increased by weight-bearing exercise during adolescence, and adequate calcium intake is necessary for exercises to have a bone stimulating effect [36]. In a review study on the interactions between physical activity and nutrients in children and adolescents, Julián-Almárcegui (2015) reported that the combined effects of exercise and calcium intake were greater than the effects of exercise or calcium intake alone, and physical activity required calcium intake to have a positive effect on bones [45]. According to a clinical report of the American Academy of Pediatrics, routine calcium supplementation is not required for healthy children and adolescents for bone health, and it is necessary to increase the supply of calcium through dietary intake to meet daily recommended levels [46]. Therefore, drinking milk, a major source of calcium, combined with moderate to vigorous physical activity that provides mechanical stimulation to the bones during growth, is considered an effective strategy to maximize bone growth potential.

In females, the effect of milk intake and physical activity was found only in lumbar BMD. The positive effect of physical activity or physical activity combined with nutrients on BMD was relatively insignificant in females, because the overall physical activity of females was low. The converted weekly moderate physical activity time was only 70.4 ± 47.9 min for females, compared to 120.7 ± 7.9 min for males. Probably due to the fact that a mechanical load has an impact on the bones in a region-specific and tissue-specific manner [44,47,48], there was no difference in the BMD of the femur and femur neck among groups, which are the areas where the mechanical impact is applied more directly during physical activity. In addition, the increase in the total body BMD and leg BMD slows down in females after the age of 13 years, whereas the lumbar BMD shows a relatively continuous increase during adolescence [4]. Therefore, the lumbar BMD of females seemed to reflect the effects of lifestyle more sensitively during adolescence.

The results of this study showed that a combination of moderate to vigorous physical activity with milk intake during adolescence, which is a very important period for laying the foundation for lifelong bone health, is an effective strategy for maximizing the growth potential of BMD. Nevertheless, there is a need to consider the following limitations when interpreting and applying the findings of this study. First, this study was a cross-sectional survey study. A longitudinal study will be needed to elucidate and verify the causal relationships among physical activity, milk intake, and BMD with respect to the combined effects of the two factors on BMD. Second, the level of physical activity was assessed based on the participation time of moderate to vigorous physical activity, but the time spent walking was not included. The intensity of walking can vary from low to moderate. Although there was a separate questionnaire item on walking, it was excluded from the analysis because it did not quantify the intensity of walking. Third, the KNHANES used in this study does not investigate type of physical activity. Therefore, non-weight bearing physical activity participation time, such as slow swimming, included as an example of moderate intensity activity, could not be considered separately to be analyzed, possibly reducing the correlation between physical activity and BMD. Fourth, in this study, the level of physical activity was examined through a questionnaire survey. Hence, this study has inherent limitations regarding the objectivity and reliability of the self-report measures of physical activity, compared to objective, direct measures of physical activity. Fifth, when carrying out subject grouping, in terms of physical activity, the subjects were divided into high- and low-level physical activity groups. In the case of milk intake, however, because a considerable proportion of people did not consume milk, the subjects were classified into two milk groups: those who did not drink milk at all or those who did. In interpreting the results, it will be necessary to consider these differences in the criteria for evaluating impact of milk intake and physical activity. Sixth, due to dietary variation within the individual, there is a limit to grasp accurately the usual intake status with a single-day survey through the 24-h recall method.

In conclusion, adolescents who did not drink milk and had a low level of physical activity were less likely to have a high BMD than those who drank milk and had a high level of physical activity. These results show that there is a synergistic effect of physical activity and milk intake on BMD, suggesting that practicing both moderate to vigorous physical activity and milk consumption in adolescence is an effective way to build healthy bones. The findings of the present study are expected to be useful as empirical data for establishing strategies for promoting healthy bone growth during adolescence.

**Author Contributions:** Conceptualization by J.H.L., A.W.H., S.H.K. and W.K.K., Formal analysis by A.W.H. and J.H.L., Funding acquisition by S.H.K. Investigation by J.H.L. and A.W.H., Methodology by A.W.H. and J.H.L., Project administration by S.H.K. and W.K.K., Resources by S.H.K. and W.K.K., Supervision by S.H.K. and W.K.K., Writing original draft by J.H.L., Writing, review & editing by S.H.K., W.K.K. and A.W.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by a 2017 grant from the Korea Dairy & Beef Farmers Association and the Korea Milk Marketing Board (2017-0243-01).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Kongju National University (protocol code: KNU\_IRB\_2020-65, date of approval: 21 August 2020).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data were obtained from the Korean National Health and Nutrition Examination Survey (KNHANES) and are available from the KNHANES website (at http://knhanes. cdc.go.kr (accessed on 26 June 2017)).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Nutrients* Editorial Office E-mail: nutrients@mdpi.com www.mdpi.com/journal/nutrients

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34

www.mdpi.com

ISBN 978-3-0365-6271-1