Participant food and activity costs in a translational Diabetes Prevention Program

Participants at elevated risk for acquiring type 2 diabetes were able to adopt healthy lifestyle changes that reduced their diabetes risk factors, without significantly increasing their costs.

Abstract

The Diabetes Prevention Program (DPP) and its translational adaptations have been shown to be effective. However, individual-level economic impacts, such as the out-of-pocket costs borne by participants due to involvement in these programs have not been consistently and thoroughly evaluated. As cost is an important consideration that will impact the willingness of individuals to participate in such programs, this study examined direct monetary costs to participants in the Group Lifestyle Balance (GLB) DPP. Older adults (n = 134, mean age 62.8 years) with body mass index (BMI) ≥24 kg/m² and prediabetes and/or metabolic syndrome participated in this GLB intervention, with two-thirds randomized to begin the intervention immediately and one-third functioning as a control for 6 months before receiving the entire intervention. Food and activity time and costs borne by participants were measured by self-report at baseline and after 6 months. Significant improvements in clinical metabolic measures, weight, and physical activity levels were achieved after 6 months in the intervention group compared both with baseline and the controls. Food costs did not increase among intervention participants. Costs related to physical activity did not change consistently over the course of the intervention. This DPP–GLB lifestyle intervention was effective in reducing risk factors for Type 2 diabetes mellitus among a diverse group of older participants without significantly increasing their out-of-pocket costs for food or physical activity over the course of the intervention. These results should help reduce concerns of individuals who are hesitant to participate in similar programs due to costs. The clinical trial registration number of this study is NCT01050205.

Implications.

Practice: Healthy lifestyle changes to reduce the risk of Type 2 diabetes in individuals at elevated risk need not entail increased out-of-pocket expenses for those individuals.

Policy: Diabetes prevention interventions can be promoted as being cost neutral for participants in terms of out-of-pocket expenses relating to food and participation in physical activity.

Research: Future research should focus on how to effectively overcome the commonly held assumption that healthy lifestyle changes are expensive and, thus, not accessible to all socioeconomic communities.

Introduction

The lifestyle intervention used in the Diabetes Prevention Program (DPP) [1] and translational programs based on this intervention [2–8] have been shown to be effective in reducing risk factors for Type 2 diabetes mellitus (T2DM) and metabolic syndrome. Programs based on the DPP intervention have also been found to be cost effective over the long term [9–13]. However, costs incurred by participating in a clinical trial may differ from the cost of participating in a community-based intervention. Clarity on this topic is important as there is evidence that perceptions about costs may influence decisions as to whether or not individuals engage in health-modifying behavior, including cost-related nonadherence of medication among the elderly [14] and disease management among individuals with diabetes [15].

With Medicare covering the clinical and education costs of CDC-recognized diabetes prevention programming [16], it is likely that interest and participation in DPPs will increase. However, although many individuals will not incur any enrollment costs, they may have concerns regarding additional personal costs they will face when taking part in these healthy lifestyle programs. In economic terms, these costs could include increases in food costs in order to achieve dietary goals, costs for equipment and services related to achieving physical activity goals, travel costs for intervention-related visits, and opportunity costs for time spent in intervention-related activities, such as physical activity or self-monitoring of food and beverage intake [17].

This concern is justified as there is some evidence that healthful food costs more than less healthy alternatives [18,19], and cost has been identified as a possible obstacle for individuals attempting to adopt a healthier diet [18]. In addition, cost and time have been cited as barriers to increasing physical activity [20].

In the DPP clinical trial, food costs were identified as being lower in the intervention group compared to the placebo group [21], with costs related to purchases of physical activity equipment and services not, surprisingly, being higher. In addition, costs related to time spent in the intervention were significantly higher for the intervention group. Although few reports have described the costs incurred by participants in translational community-based DPPs, a report from the Healthy Living Partnerships to Prevent Diabetes (HELP PD) [22] program identified cost results consistent with the DPP findings: slightly lower food costs but higher costs overall in the intervention group when compared to the control, primarily, due to the time spent in self-monitoring, higher costs for physical activity items, and travel costs for the intervention.

Thus, while there is preliminary evidence suggesting that involvement in programs that encourage healthy lifestyle changes can result in some additional costs for participants, this has not been well established in community translation efforts across a variety of populations. Therefore, this investigation reports on individual out-of-pocket costs and time related to participation in the first 6 months of a community-based behavioral lifestyle intervention in older adults in diverse socioeconomic settings, to determine whether costs and experiences differ between the intervention and control groups.

Methods

The Group Lifestyle Balance (GLB) program [2] is a behavioral lifestyle intervention adapted from the DPP, typically delivered by trained health care professionals. The GLB curriculum includes the DPP weight loss goal of 7% and activity goal of 150 min per week of moderately intense physical activity. Meeting weekly for 12 weeks and transitioning to monthly sessions to deliver 22 sessions, the GLB curriculum consists of presentations on dietary strategies to reduce fat and calories, ways to increase physical activity, and social and psychological approaches to establishing and maintaining a healthy lifestyle. The GLB program has been successfully implemented in a variety of settings, including medically underserved communities, a worksite, clinical practices, a fitness center, and churches [2,7,8,23–25].

Population

Potential participants were recruited from three community centers with varying socioeconomic profiles, using community center membership communications, direct mail, and posters, from September to November 2011. While all three centers had large retiree populations, economically and educationally, they ranged from being primarily blue collar to having a significant proportion of participants with graduate degrees. Preliminary telephone or in-person screening preceded an on-site clinical assessment of eligibility. Individuals at least 25 years old, with body mass index (BMI) ≥24 kg/m² for whites and African Americans and ≥22 kg/m² for Asians, were eligible to participate, when meeting the criteria for prediabetes (fasting plasma glucose 100–125 mg/dL) [26] and/or the metabolic syndrome. The metabolic syndrome was defined using the National Cholesterol Education Program Adult Treatment Panel 3 criteria as a clustering of three of the following five conditions: abdominal obesity, atherogenic dyslipidemia, hypertension, and insulin resistance [21]. Individuals previously diagnosed with diabetes, pregnant or lactating women (within past 6 weeks) and individuals planning to leave the study region within 18 months were ineligible to participate.

Study design

Specific details about the study design have been published previously [8]. Briefly, a randomized 6 month delayed control study design was used, with participants randomized to either begin the intervention immediately (IMMEDIATE) or after a 6 month delay (DELAYED) in a 2:1 ratio. A basic sampling procedure was utilized to balance randomization by site and was programmed by a member of the research staff. The participants randomized to the DELAYED control group received only general health information during the first 6 months. After 6 months, the DELAYED control group received an intervention identical to that of the IMMEDIATE group. This design is ideal as it provides a control group during the first 6 months of the study when the IMMEDIATE intervention group is receiving the majority of the intervention sessions while at the same time meeting the ethical imperative to offer an effective risk reduction intervention to all participants, all of whom are at elevated risk for T2DM and/or have the metabolic syndrome.

One hundred and thirty-four participants provided informed consent. Eighty-eight participants were randomized to the IMMEDIATE intervention group, which began meeting weekly, working toward the GLB program weight loss (7% of body weight) and physical activity goals (30 min/day, 5 days/week of moderate aerobic activity). Forty-six were randomized to the DELAYED control group. All randomized participants attended clinical assessment visits at their community enrollment site at baseline and approximately 6 months afterward. Data collection began in January 2012 and concluded in March 2014.

The sample of 134 community participants was not calculated for this cost analysis but rather the sample size necessary to provide sufficient statistical power to answer the primary analyses of the parent study [8]. Cost analysis was a secondary aim of the study and a priori size estimates were not calculated. As such, we used all data available from the community participants to maximize our ability to detect differences related to cost.

Fourteen individuals (seven pairs) were identified as living in the same household and both individuals of the same pair received the same randomized assignment. Clinical variables collected included blood lipids, insulin, fasting plasma glucose, and HbA1c, as well as anthropometric measures. A brief medical history was taken and a series of surveys were given. At baseline, there were no significant differences in clinical or demographic values between the IMMEDIATE intervention and DELAYED control groups (Table 1).

Table 1.

Participant baseline characteristics

	IMMEDIATE intervention Mean (SD) Median (IQR) (n = 88)	DELAYED control Mean (SD) Median (IQR) (n = 46)	p value*
Weight (lbs)	212.4 (46.3) 202.3 (181.3, 241.5)	201.5 (37.1) 195.1 (176.8, 220.4)	.17
BMI (kg/m2)	34.9 (6.7) 33.6 (29.9, 38.7)	33.4 (4.9) 32.8 (30.3, 37.0)	.16
Glucose (mg/dL)	96.0 (10.3) 94.0 (89, 101)	95.9 (13.1) 93.0 (87, 102)	.94
HbA1C (%)	5.80 (0.32) 5.7 (5.6, 5.9)	5.76 (0.33) 5.7 (5.6, 5.9)	.53
Gender (% female, n)	65.9% (58)	69.6% (32)	.67
Age (average years, range)	62.8 (12.1)	61.9 (11.9)	.66
Education (greater than bachelor’s degree)	54.6%	58.7%	.72
Usual number of people in household	2.02 (0.98) 2.0 (1, 2)	1.78 (0.66) 2.0 (1, 2)	.32

BMI body mass index; IQR interquartile range; SD standard deviation.

*p value for comparison between IMMEDIATE intervention and DELAYED control groups.

Survey instruments

To assess the direct, nonmedical costs of participants in this study, the participants completed a 12-page survey at baseline and 6 months (see Supplementary Appendix B) in which they were asked to estimate their purchases and out-of-pocket costs for apparel, equipment, courses, and programs related to physical activity. The survey also asked for estimates of weekly or monthly household food costs, including food purchased to consume at home, from grocery stores, nongrocery stores, restaurants, and takeout and prepared food locations, as well as perceptions of whether and how household food expenditures might have changed over the course of the intervention. All costs were collected and are reported in current U.S. dollars in 2012 and 2013. These survey questions were modeled on the food items from the NHANES Flexible Consumer Behavior Questionnaire [22]. Lastly, the survey also asked the participants for estimates of the average weekly time spent in physical activity, food preparation, and participation in GLB activities.

Analytical strategy

Analyses were conducted on subjects who enrolled in the study and attended the baseline and 6 month clinical visits, with surveys completed at both time points. Changes in costs were examined over this 6 month timeframe. Subjects who responded “I do not know” to food cost questions were excluded from food cost results (Table 2) as this response is neither “missing” nor “zero.” The results were stratified by demographic (age and gender) and goal attainment (weight loss and activity) variables. Grocery and nongrocery store food costs were combined into a Store Food category, reflecting changes in the grocery retailing since the Flexible Consumer Behavior Questionnaire was developed as stores like Walmart and Costco are now major food retailers yet not considered grocery stores by earlier definitions. Due to changing food prices over time, the 6 month food cost data were adjusted for inflation using Consumer Price Index data for the Pittsburgh region during the data collection period [27] and are presented as adjusted and unadjusted results. Adjusted results reflect 2012 U.S. dollars.

Table 2.

Food costs by type of purchase site in DELAYED and IMMEDIATE intervention groups over 6 months

	DELAYED control				IMMEDIATE intervention
Estimated average U.S. dollars spent on food purchases per month in the preceding 6 month period	Baseline Mean (SD) Median (IQR) (n = 31)a	6 months Mean (SD) Median (IQR) (n = 31)	Baseline to 6 month change Mean (SD) Median (IQR) (n = 31)	p valueb	Baseline Mean (SD) Median (IQR) (n = 60)c	6 months Mean (sd) Median (IQR) (n = 60)	Baseline to 6 month change Mean (SD) Median (IQR) (n = 60)	p valueb	p valued
Store food (grocery and nongrocery; $/household)	364.7 (182.6) 360 (240, 430)	455.1 (255.1) 400 (260, 625)	90.4 (226.9) 25 (−40, 150)	0.08	423.9 (300.1) 385 (200, 575)	469.7 (351.0) 400 (222.5, 580)	45.8 (323.8) 40 (−75, 100)	0.21	0.45
Purchased takeout food % (n)	51.6% (16)	41.9% (13)	−9.7% (−3)	0.61	70.0% (42)	55.0% (33)	−15.0% (−9)	0.13
Takeout food per month ($/household)	54.9 (143.6) 10 (0, 60)	25.65 (39.0) 0 (0, 50)	−29.3 (147.8) 0 (−30, 12)	0.35	36.2 (43.1) 24.5 (0, 55)	30.6 (45.5) 6.5 (0, 50)	−5.5 (43.4) 0 (−19, 0)	0.13	0.82
Purchased eat-out % (n)	31	31	0% (0)	1.0	91.6% (55)	85.0% (51)	−6.0%	0.39
Eating out food per month ($/household)	174.7 (184.3) 132 (75, 210)	193.6 (138.9) 150 (100, 240)	18.8 (160.1) 20 (−30, 80)	0.13	115.1 (115.4) 80 (25, 175)	101.9 (97.4) 100 (25, 155)	−13.2 (79.0) 0 (−40, 20)	0.25	0.045
Total of estimated food spending per month ($)	594.4 (434.8) 525 (420, 665)	674.3 (319.9) 635 (430, 785)	79.9 (418.2) 90 (−65, 270)	0.04	575.1 (377.2) 497.5 (275, 780)	602.1 (400.8) 528 (330, 793.5)	27.0 (371.1) 17 (−91.5, 107.5)	0.62	0.14

IQR interquartile range; SD standard deviation.

^aOf 46 participants enrolled at baseline, 42 attended the 6 month clinical visit. Eleven responded “I do not know” to food questions and were excluded.

^b  p value, 6 month change from baseline.

^cOf 88 immediate participants enrolled at baseline, 81 attended the 6 month clinical visit. Twenty-one responded “I do not know” to food questions and were excluded from analyses

^d  p value comparing change from baseline to 6 months between the DELAYED control and IMMEDIATE intervention.

Statistical methods

The Wilcoxon Signed-Rank Sum test was used to test the change of costs within the two study groups from baseline to follow-up clinical visits, and the Wilcoxon Rank Sum test evaluated change between the IMMEDIATE intervention and DELAYED control groups. The Spearman Rank order correlation test was used to test correlations between variables. The Kruskal–Wallis test was used to compare costs between the three community locations. Analyses were carried out using SAS (version 9.3, SAS Institute, Cary, NC).

Results

Clinical results

Overall, the lifestyle intervention was highly successful. Specific clinical and participant results are presented in the primary clinical report on this GLB community intervention [8]. Briefly, during the first 6 months of the intervention, weight for all community sites combined decreased significantly (−5.6%, p < .001) from baseline in the IMMEDIATE intervention group and was significantly greater than the DELAYED control group (p < .001). Additionally, HbA1c and waist circumference were significantly reduced at 6 months from baseline in the IMMEDIATE intervention group and when compared with the DELAYED control group [24]. The median weekly minutes of self-reported physical activity in the IMMEDIATE intervention group increased significantly from baseline to 6 months (180–210, p = .02), and this change was significantly greater than the DELAYED control group (p = .05). This resulted in the IMMEDIATE intervention group increasing physical activity time approximately twofold when compared to the DELAYED control group after 6 months while participating in the intervention [28].

Food costs

Looking at the individual categories of food expenses (store food, takeout food, and eating-out food), there were no significant changes in reported food costs in the IMMEDIATE intervention group when comparing 6 months to baseline. This resulted in total monthly food expenses being unchanged from baseline among participants after 6 months in the lifestyle intervention. Interestingly, there was a significant increase in total monthly food expenses, $594 to $674 (p = .04), in the DELAYED control group over this same time period (Table 2). There was also a significant difference in the change between the DELAYED control and IMMEDIATE intervention groups in the amount of money spent eating at restaurants, cafeterias, and buffets due to the fact that costs in the control group increased, while the intervention group costs decreased (Table 2). Adjusting food cost data for food price inflation did not change the findings (Supplementary Appendix 1).

There were no significant differences in the change in total food costs within the IMMEDIATE intervention group when stratified by the achievement of weight loss goals (any, ≥5%, and ≥7% of baseline body weight) or by education status (data not shown). Among the IMMEDIATE intervention participants, there were no differences in the change in total food expenditures between baseline and 6 months by the site from which subjects were recruited, with the exception of costs for eating out at restaurants and cafeterias (data not shown).

Physical activity costs

Marginal changes in costs related to participating in physical activity were found in both study groups. In the IMMEDIATE intervention group, a decrease in overall activity-related costs from baseline to 6 months was observed ($245 to $180, p = .06) (Table 3), which was largely attributable to a lower expenditure for physical activity services, such as gym memberships and fees ($189 to $134, p = 0.03). No significant changes in activity-related costs were observed in the DELAYED control group during this time period.

Table 3.

Costs related to physical activity

	DELAYED control				IMMEDIATE intervention
Household activity cost categories (U.S. $/6 months)	Baseline Mean (SD) Median (IQR) (n = 42)	6 months Mean (SD) Mdian (IQR) (n = 42)	Change (baseline to 6 months) Mean (SD) Median (IQR) (n = 42)	p valuea	Baseline Mean (SD) Median (IQR) (n = 81)	6 months Mean (SD) Median (IQR) (n = 81)	Change (baseline to 6 months) Mean (SD) Median (IQR) (n = 81)	p valuea	p valueb
Items (eg., shoes, clothing, and bike)	38.9 (55.1) 0 (0, 65)	87.8 (174.9) 39.5 (0, 100)	48.9 (166.0) 0 (−25, 100)	0.06	56.7 (127.6) 0 (0, 75)	46.1 (65.4) 0 (0, 83)	−10.5 (112.3) 0 (−34, 10)	0.40	0.10
Services (eg., classes and gym membership)	161.2 (295.1) 0 (0, 210)	92.8 (158.0) 0 (0, 192)	−68.4 (239.9) 0 (−50, 0)	0.09	188.9 (512.9) 0 (0, 192)	134.1 (382.3) 0 (0, 100)	−54.9 (409.8) 0 (−25, 0)	0.03	0.86
Total	200.1 (300.0) 96 (0, 285)	180.6 (222.2) 95 (0, 300)	−19.5 (279.8) 0 (−55, 100)	1.0	245.6 (535.4) 75 (0, 270)	180.2 (387.3) 60 (0, 199)	−65.4 (421.8) 0 (−76, 20)	0.06	0.51

IQR interquartile range; SD standard deviation.

^a  p value, 6 month change from baseline.

^b  p value comparing change from baseline to 6 months between the DELAYED control and IMMEDIATE intervention.

When stratifying by gender in the IMMEDIATE intervention group, there was a significant decline in costs for physical activity from baseline to 6 months in men ($196 to $103, p = .05) and a similar, although not significant, trend in women ($272 to $221, p = .43). Although participants aged <60 had higher costs related to physical activity than those aged ≥60, change in activity costs over the course of the intervention was not different when stratified by age (data not shown). There were no significant associations between physical activity costs or change in physical activity costs and the monthly frequency or change of frequency of activities as reported on the Modifiable Activity Questionnaire (data not shown). There were no baseline differences in total costs for physical activity between the three community locations, as well as no differences by location in the change of activity costs over 6 months of intervention in the IMMEDIATE intervention participants (data not shown).

Time Spent in Intervention Activities

Physical activity

As expected, participants in the IMMEDIATE intervention experienced meaningful time commitments due to their involvement in the GLB program. These subjects significantly increased time spent in physical activity during the intervention compared with baseline (from averages of 263–417 min/week, p = .006). No change was observed in physical activity time among the DELAYED control group.

Self-monitoring time

The IMMEDIATE intervention group was expected to monitor and record both dietary intake and physical activity during the 6 month intervention period, and this was reflected in the reported self-monitoring time. Average self-monitoring times increased during this time for the IMMEDIATE intervention (13 to 146 min/week, p < .0001) but not for the DELAYED control group (19 to 30 min/week, p = .61). The results for time spent recording activity and monitoring diet did not change when stratified by gender or age (data not shown).

Food preparation time

Average time spent in food preparation decreased from baseline to 6 months in both the IMMEDIATE intervention group (440 to 364 min/week, p = .004) and in the DELAYED control group (384 to 296 min/week, p = .03).

Discussion

The results from this investigation suggest that at-risk participants in a translational T2DM prevention intervention, delivered in a community center, can successfully reduce their diabetes risk factors without substantial increases in out-of-pocket, nonmedical costs, both for food and for the facilitation of physical activity. The only significant increases in costs in the lifestyle intervention group relative to the control group were related to the opportunity costs for the time involved in self-monitoring. The implications of these findings are that nonmedical costs borne by participants in community behavioral-based DPPs need not impose a substantial financial burden on participants, nor should cost related to participation be a barrier to the successful completion of the program.

These results corroborate the findings from the DPP [21] and HELP PD [22] trials in which food costs were not higher, while time spent in physical activity and related dietary activities was higher, in lifestyle intervention programs. This report differs from the DPP results, which suggested that costs related to physical activity did not change over time in the lifestyle intervention group. It should be noted, however, that the DPP costs were estimated and not directly reported by participants as in the current study.

This report did not directly examine the role of cost-related factors in decision-making by the participants. Thus, the finding of no change in food and activity costs over time in the intervention group could be due to changes in purchasing behavior. For example, individuals may act economically and identify lower-cost alternatives to common food or exercise items while in the intervention. Future work should investigate how participants in DPPs integrate economic factors into their decision-making.

Total estimated food costs did increase in the DELAYED control group in this report. However, it was not possible to identify the reasons underlying this finding. Perhaps in anticipation of future deprivation during participation in the intervention, these participants purchased and ate more than they otherwise would have [29]. However, these control subjects volunteered to participate in this study, suggesting that they were motivated to change from the start. Alternatively, it is possible that the control participants might have attempted to improve their diet by increasing purchases of more expensive food items that they considered healthy in an attempt to jump-start healthier lifestyle changes in anticipation of the start of the intervention.

Seasonal factors may contribute to findings in both the intervention and control groups. There was a decrease in cost for activity services in the intervention group. As the baseline assessment visit was in February and March, costs reported for the previous 6 months included fall and early winter months when opportunities for outdoor physical activity are more limited and gym memberships increase. It is possible that increased outdoor physical activity during the spring and summer months, with less time at gyms and fitness centers, could explain the decrease in physical activity service costs from the baseline to August and September, when the 6 month assessment occurred.

The finding of a decrease in food preparation time for the IMMEDIATE intervention group from baseline to 6 months (440 to 364 min/week, p = .004) likely reflects changing eating and food preparation behaviors. It is possible that participants were eating less and more simply during the intervention, guided by ideas and suggestions presented in the GLB curriculum. Further insights on how participant food preparation pattern change during diabetes prevention interventions could be gained by including daily food preparation diaries in future studies.

Additionally, the intended impact of the intervention is reflected in the time reported spent in self-monitoring. There were significant increases in self-monitoring time in the intervention group from baseline to 6 months relative to the control as the GLB lifestyle intervention encourages self-monitoring as a tool to assist in meeting physical activity, calorie, and weight loss goals.

These findings may be limited by the self-report method of data collection, which might not reflect accurate levels of expenditures. However, the use of a food cost questionnaire based on a validated CDC survey and the collection of information within short windows of recall should minimize these concerns. Also, food cost data represent the behavior of the entire household and may be affected by changes in the size of the household over time. However, stratifying the results by household size indicated no significant changes from baseline to 6 months or between groups for total food costs (data not shown). It is possible that recall errors occurred when participants were estimating expenses over a 6 month period as opposed to a shorter period. Although not the explicit goal of the project to recruit a population of older adults, the average age of the participants at 62 would be considered “older.” Further research should investigate issues of participant cost and time among younger adults at risk for T2DM.

In summary, these findings indicate that, as part of a behavior lifestyle intervention, statistically significant and clinically meaningful reductions in T2DM risk factors can be achieved in the absence of significant increases in out-of-pocket, nonmedical costs. These findings might be best applied to community centers, which is of particular relevance given the focus of the Young Men’s Christian Association (YMCA) on diabetes prevention [30], but we are optimistic that other intervention sites should achieve similar results. Such findings should help make participation in a diabetes prevention intervention program more attractive for those at elevated risk for diabetes, who might previously have viewed increased costs as an obstacle to successful program participation.

Funding:

This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases grant 5R18DK081323-04.

Compliance with Ethical Standards

Conflicts of Interest: G.L.S., T.J.S., V.C.A., M.K.K., R.G.M., and A.M.K. declare that they have no conflicts of interest.

Authors’ Contributions: G.L.S. drafted the manuscript, assisted by T.J.S. and A.M.K. Statistical analysis was conducted by G.L.S., with help from T.J.S., V.C.A., and R.G.M. Principal investigator A.M.S. conceived and designed the G.L.B. study, of which these analyses are a part. M.K.K. assisted with study development and implementation, and led data collection. All authors reviewed the manuscript prior to submission.

Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the Institutional Research Board of the University of Pittsburgh. This article does not contain any studies with animals performed by any of the authors.

Informed Consent: Informed consent was obtained from all individual participants included in the study.