Many unhealthy dietary and physical activity habits that foster the development of obesity are established by the age of five. Presently, approximately 70 percent of children in the United States are currently enrolled in early childcare facilities, making this an ideal setting to implement and evaluate childhood obesity prevention efforts. We describe here the methods for conducting an obesity prevention randomized trial in the child care setting.
A randomized, controlled obesity prevention trial is currently being conducted over a three year period (2010-present). The sample consists of 28 low-income, ethnically diverse child care centers with 1105 children (sample is 60% Hispanic, 15% Haitian, 12% Black, 2% non-Hispanic White and 71% of caregivers were born outside of the US). The purpose is to test the efficacy of a parent and teacher role-modeling intervention on children’s nutrition and physical activity behaviors. . The Healthy Caregivers-Healthy Children (HC2) intervention arm schools received a combination of (1) implementing a daily curricula for teachers/parents (the nutritional gatekeepers); (2) implementing a daily curricula for children; (3) technical assistance with meal and snack menu modifications such as including more fresh and less canned produce; and (4) creation of a center policy for dietary requirements for meals and snacks, physical activity and screen time. Control arm schools received an attention control safety curriculum. Major outcome measures include pre-post changes in child body mass index percentile and z score, fruit and vegetable and other nutritious food intake, amount of physical activity, and parental nutrition and physical activity knowledge, attitudes, and beliefs, defined by intentions and behaviors. All measures were administered at the beginning and end of the school year for year one and year two of the study for a total of 4 longitudinal time points for assessment.
Although few attempts have been made to prevent obesity during the first years of life, this period may represent the best opportunity for obesity prevention. Findings from this investigation will inform both the fields of childhood obesity prevention and early childhood research about the effects of an obesity prevention program housed in the childcare setting.
Trial registration number: NCT01722032
Keywords:Childhood obesity; Methodology; School-based obesity prevention; Childcare centers; Young children
One in four US children under age five is either overweight (≥85th to <95th percentile for age- and sex-adjusted percentiles for body mass index [BMI]) or obese (≥95th percentile age- and sex-adjusted percentiles for BMI), with ethnic minority groups being disproportionately affected [1,2]. These statistics are of particular concern because obese preschool-aged children are five times more likely to be overweight during adolescence  and more than four times as likely to be obese adults when compared to their normal weight counterparts [3,4]. These results show that contrary to popular belief, children do not “grow out of” their “baby fat.” In fact, evidence indicates that excessive weight gain in the first years of life can alter developing neural, metabolic and behavioral systems in ways that increase the risk for obesity and chronic disease later in life [5,6]. Specifically, childhood obesity is a precursor to type-2 diabetes, cardiovascular disease, hypertension, stroke, osteoarthritis, asthma, and certain cancers [5,6]. As such, many reports have projected that childhood-onset obesity will contribute significantly to increased morbidity and mortality in adulthood , particularly among ethnic minority groups who are disproportionately affected by many of these chronic conditions .
Although few attempts have been made to prevent obesity during the first years of life, this period may represent the best opportunity for obesity prevention. During infancy and early childhood lifestyle behaviors that promote obesity are just being learned, and it is easier to establish new behaviors than to change existing ones. Furthermore, childcare settings offer a potentially very powerful opportunity to implement such efforts because; (1) 70% of preschool-aged children in our nation are enrolled in daily, out-of-home child care ; (2) children from low-income backgrounds consume 50%-100% of their Recommended Dietary Allowances (according to standards established by the Child Care Food Program) in the childcare setting; (3) many children spend the majority of their waking hours out-of-home, and in the childcare setting .
The majority of the childcare setting studies conducted to date focus on individual-level interventions [11-14] with few efforts targeting systemic policy-level changes that include evidence-based curricula and modifications to institutionally-provided meals and snacks. In fact, health and nutrition are often completely overlooked in the childcare setting because administrators and teachers feel they are under-qualified to deliver this information . Additionally, many adults make food choices (purchasing and consumption) couched in personal attitudes and beliefs about food and nutrition, which in turn influence children’s nutritional beliefs, attitudes and behaviors [16,17]. The U.S. Department of Agriculture’s (USDA) concept of the “nutritional gatekeeper” and the “Project M. O. M. : Mother’s & Others & My Pyramid” campaign  suggests that empowering the nutritional gatekeepers in both the home and the child care center will potentiate a lasting and effective impact on the health and nutrition of future adults. For example, using a similar model in the elementary school setting “nutrition education extenders” were found to have a positive effect on child nutrition practices during and after the Healthier Options for Public Schoolchildren (HOPS) Study [19,20].
We describe here the research design and methods for a federally-funded, large-scale comprehensive obesity prevention intervention randomized controlled trial “Healthy Caregivers, Healthy Children HC2” to test (1) the utility of teachers and parents as effective nutritional gatekeeper role models, or change agents; and (2) the efficacy of center-based nutrition and physical activity policies, and how both affect early childhood growth patterns (namely body mass index), nutrition and physical activity practices among preschool-age children.
A randomized, controlled obesity prevention trial in 28 low-income, ethnically diverse child care centers located throughout Miami-Dade County, FL is currently being conducted over two years (2010-present) to test the efficacy of an intervention that poises teachers and parents as lifestyle change agents. Trial methodology, intervention content and outcome measures are described in detail below.
Because it was not feasible to randomize individuals within classrooms to different treatments, we randomly assigned centers to treatment or control conditions. Thus, HC2 is a group-randomized study, often referred to as a cluster randomized trial (CRT). In CRTs, clusters of people, or intact social units (such as childcare centers) rather than individuals are randomized to intervention and control groups and outcomes are measured on individuals within those clusters. CRT designs are used not only to evaluate group interventions but also individual interventions where group level effects are relevant. The University of Miami Institutional Review Board approved all aspects of the study.
Centers must have met the following criteria to be included in the study: (1) > 30 children ages 2-to-5 currently enrolled; (2) Serve low income families who are a part of the USDA food program and SNAP eligible; (3) Reflect the ethnic distribution of the Miami-Dade County Public School System (63 percent Hispanic, 19 percent African American, 18 percent white); and (4) Center director agrees to participate and sign a letter of commitment. Centers were excluded if they did not meet these criteria.
Parents consented to have their child participate in the study. A total of 60% of children were Hispanic, with the most prevalent specific Hispanic ethnicity being ‘Other’ (31%), followed by Cuban (15%), and Mexican (10%). Other specific ethnicities represented were Haitian (15%), non-Hispanic Black (12%) and non-Hispanic White (2%). Table 1 shows demographic characteristics of each study arm. There were 1105 participating children with 726 in the intervention arm and 379 in the control arm.
Table 1. Description of children participating in a childhood obesity prevention intervention trial by randomization arm
The majority of caregivers were born outside of the US (71%). Languages other than English were preferred by 65% of the sample. Low levels of education were prevalent, with 37% completing less than a 12th grade education, 25% completed high school/GED, and 38% completing education beyond high school (Table 2).
Table 2. Description of caregivers of preschool children participating in a childhood obesity prevention intervention trial
Treatment or intervention arm schools received a combination of (1) implementing a daily curricula for teachers/parents (the nutritional gatekeepers); (2) implementing a daily curricula for children; (3) technical assistance with meal and snack menu modifications such as including more fresh and less canned produce, no high fat milk, and less simple carbohydrate items; and (4) creation of a center policy for dietary requirements for meals and snacks, physical activity and screen time. Details of each of these components follows.
Teacher/parent nutritional gatekeeper curriculum
For childhood obesity prevention programs to achieve long-term success the nutritional gatekeeper (i.e. the person buying and supplying food for the children) must engage in healthy lifestyle behaviors and beliefs themselves. For young children the nutritional gatekeeper is most likely the parent or a primary caregiver and for those who attend childcare, the teachers and center staff also. Therefore, Curriculum Specialist (CS) utilized an integrated curriculum with teachers and parents at the child care centers. The teacher-focused curriculum consisted of 29 weekly technical assistance sessions delivered by HC2 curriculum specialists during weekly visits to the childcare centers. Additionally, HC2 program staff conducted joint parent-teacher meetings (providing a substitute so that teachers can attend during parent pickup times) that focused on an evidence-based nutrition and physical activity curriculum. CS taught parents and teachers about (1) Choosing healthy foods, with an emphasis on fruits and vegetables; (2) Making healthy food choices for infant and child development; (3) Preparing nutritious snacks; (4) Preparing new recipes; (5) Tasting new foods; (6) Learning about food safety and storage; (7) Planning meals; (8) Making grocery lists; (9) Shopping wisely for groceries; (10) Budgeting food dollars and food stamps; and, (11) Using food labels to choose best buys. Via role-modeling, parents and teachers were encouraged to implement change at the family/home and child care center level, not child level. Additionally, nutritional professionals served as role models for the teachers and parents and assisted them as nutritional gatekeepers and positive models for the children. Lessons on physical activity for the whole family were also included (Table 3).
Table 3. Nutritional gatekeeper role modeling curriculum lessons, objectives and activities for childcare center teachers and parents
Our framework was comprised of a variety of role modeling components from the USDA’s Project M.O.M [18,21-24]. These components were then adapted and merged into eight sessions that were appropriate for the parents/caregivers and teachers of pre-school aged children. All curriculum materials if not available in Spanish were translated and back translated into Spanish by a team of certified translators at the University of Miami.
Teachers received ongoing technical assistance to support the inclusion of healthy eating and physical fitness concepts in their weekly lessons, without the continued assistance of the CS, and so, during the second year of the study they could lead the lessons independently or without any CS assistance. Currently, in year two, booster trainings are being provided bimonthly in all centers to provide teachers with refresher overviews of all lesson plans.
Parents participated in monthly events such as dinners at the centers where they received the same nutrition and physical activity educational information that formed the content of the teacher curriculum. They also received monthly newsletters that contained information that was covered in that month’s school lessons and an overview of the center’s HC2 activities for that month.
Twenty lesson plans were developed to promote nutrition and physical activity and were supplemented with the OrganWise Guys (OWG). The lesson plans focused on core principles of healthy living (high fiber, low fat, lots of water, exercise) and eating (nutrient-dense foods), nutritious dietary offerings in USDA meals that model classroom-based and parent nutrition education programming via the USDA feeding and nutrition education programs and has been shown to have positive health outcomes among primary school children [19,20].
This classroom-based curriculum consists of 20 weekly sessions that were provided to the teachers by project curriculum specialists (CS) during weekly visits. The teachers then taught the curriculum to the children. The CS would address menu changes and target the cognitive, cultural, and environmental barriers to a low fat, high fiber diet that include more fruits and vegetables. These weekly visits are designed to assist center staff in problem-solving ways to ensure implementation. In addition, the curriculum was translated and subsequently implemented in Spanish where applicable.
Technical assistance and creation of center policies
All center modifications were revenue neutral (do not have an additional cost). The meals, snacks and beverages were provided by caterers who were given technical assistance by the HC2 nutritionist to ensure that the meals followed the policy guidelines yet still met USDA nutrition requirements. Center administrators were given monthly technical assistance by both the project nutritionist and staff on menu development and curriculum enhancements that would include the integration of the nutritional gatekeeper curriculum [3,25-27] critical for intervention maintenance and effective mediators of change.
Menus in intervention centers were modified to include low-fat (1%) or skim milk, water, less juice. Low fat or skim milk was promoted only at lunch, and water was promoted as the primary beverage for staff and children, especially after physical activity.
Menus in intervention centers were modified to include more fresh fruit (or canned fruit in water or 100% juice), fresh vegetables (or low-sodium canned vegetables), and fewer simple carbohydrate snacks (cookies, crackers, etc.). The centers also were encouraged to incorporate fresh fruits and vegetables for snack and meal time as often as possible. In addition, we provided food tastings to the children to expose them to new fruits and vegetables that they may have never tried before (e.g. kiwi, blackberries, rhubarb, edamame, and cauliflower).
Physical activity policy
Physical activity was promoted for at least 60 minutes per day. Lesson plans offered structured physical activities in the classrooms and included music and movement CDs with songs and exercises. Teachers were also provided with rainy day activities that could be used in the classroom and equipment that can be purchased inexpensively at local stores (jump ropes, hula-hoops, balls, etc.) but were encouraged to conduct physical activities outside and incorporate free play as much as possible.
Reduction of screen time policy
The reduction of screen time was addressed by limiting television viewing, watching movies, and playing computer games to 30 minutes or less each day. The amount of time screens was documented.
Those schools randomized to the control arm received a safety curriculum and some child care center locations received an attention control consisting of one visit from the Injury Free Mobile which provided parents and teachers with home, car and child seat safety information. The control group received all the same pre-post measures (see below) as the intervention arms. They also received the same incentives as the intervention arms to foster involvement and ensure retention/reduce loss to follow up.
All measures were administered at the beginning and end of the school year for year one and year two of the study for a total of 4 longitudinal time points for assessment. All measures were included to measure pre-post intervention changes in nutrition and physical activity knowledge, attitudes, and beliefs, defined by intentions and behaviors. Anthropometric measures were also included. Effort was made to maximize assessment sensitivity to functional domains deemed most critical to obesity prevention research and with strong consideration for test validity and reliability, availability of recent multiethnic normative samples, and age ranges avoiding floor/ceiling effects. In addition, each of the instruments selected has either been (a) standardized on culturally, racially and social economically diverse populations or (b) used in research projects involving culturally and racially diverse populations. A summary of all measures are included in Table 4 and their procedures are described briefly below.
Table 4. Primary and secondary study outcomes and accompanying measurement tool for each study variable and participant
Child health measures
Height (by stadiometer), weight (by digital scale) and waist circumference measurements were collected at the beginning and the end of year one and two school year. Children’s shoes and any heavy outer clothing prior to measurement. BMI is converted to an age and sex- adjusted percentile and z-score . According to the CDC, the following percentile categories will be used to define weight groups; underweight (< 5th percentile), normal weight (≥ 5th to < 85th percentile), at risk for overweight (≥ 85th to < 95th percentile ) and overweight (≥ 95th percentile) .
Physical activity level
The Healthy Kids Checklist[30-32] was administered to measure nutrition behaviors and physical activity patterns at home of both parent and child. It also evaluates parental role modeling of healthy lifestyle practices. The parent serves as a reporter for a child’s lifestyle (physical activity, sleep duration, television viewing, and presence of a television in the child’s bedroom) and eating behaviors.
Child nutritional behaviors are assessed with two measures. Healthy Kids Checklist[30-32], which includes 8 determinants: fat, fiber, fruit/vegetable intake, calcium/dairy intake, sweetened beverages, eating in restaurants, breakfast skipping, energy density. This measured (by parent proxy) child nutritional behaviors at home and the Food Frequency Questionnaire (FFQ)  measures nutritional behaviors at school. The FFQ has been adapted from the Willett Semiquantitative Food Frequency Questionnaire  to measure monthly changes in food items served to the children by the childcare centers. Additionally, center directors provided a copy of the invoice of food ordered for the month accompanying the questionnaire.
Parent and childcare provider nutrition and physical activity measures
The following two instruments assessed adult nutritional behaviors; (1) the Food Behavior Checklist is a short, culturally neutral, valid and reliable indicator of fruit and vegetable consumption. The 16-item checklist includes 7 fruit/vegetable, 2 milk/dairy, 1 food security, 3 diet quality and 3 fat/cholesterol items. Convergent validity was reported for 10 selected nutrients and the Healthy Eating Index for the 16-item tool [30,31]; and (2) the Fruit and Vegetable Inventory, an instrument that contains 13 psychosocial items shown to be related to fruit and vegetable intake. The six constructs are perceived control, self-efficacy for eating healthier, readiness to eat more fruit, readiness to eat more vegetables, and perceived diet quality.
Adult health literacy is measured using the Pediatric Health Literacy Activities Test a 10-item instrument validated in English and Spanish among caregivers of young children. It correlates well with lengthier measures of literacy and numeracy. It includes items related to growth and nutrition (e.g., infant formula mixing, nutrition labels), as well as general health items.
A modified (only the 4 items relating to physical activity were administered) version of the CATCH Physical Activity SelfEfficacy Questionnaire measures parent’s self-efficacy to be physically active.
Baseline characteristics for children, parents and childcare providers were collected and included infant feeding history (breastfeeding exclusivity and duration, onset of solid foods, age of weaning from bottle), and parent perception of child weight status, acculturation level, and level of food insecurity in the household.
Time spent in daily physical activity, classroom schedules and lesson plans, and foods and drinks (items, amounts) and monthly behavior observations are recorded by study staff for each classroom.
Parent activity survey
To track the amount of dinners parents attend, activities parents complete, and number of newsletters that are received/read, parents are given an annual survey that includes a satisfaction component inviting parents to give feedback and to rate the helpfulness of each activity.
Teacher process survey
Process data is collected to determine the feasibility of implementing HC2 on a larger scale and to determine the sustainability of the project. The CS complete a weekly process worksheet for each center by interviewing teachers. This information will help answer the larger feasibility questions of: 1) What are the range of common and uncommon barriers to implementing an intervention like HC2? 2) What is the range of creative solutions to overcoming these barriers? 3) What are the unique assets of child-care centers that enable them to adapt such an intervention? 4) What key community assets may be necessary to bring an intervention like HC2 “to scale” across the entire county?
Our power analysis indicates that with 80% power (alpha = 0.05, two tailed) we will need 12 centers per group/cluster (2 groups/clusters) for a total of 24 centers and 30 individuals per center for a 1.10 effect size to give us confidence in our statistical analysis that findings are not due to chance alone.
Our univariate analysis consists of simple descriptive measures such as percentages, medians, ranges, variances, and standard deviations to describe all study participants and sites. Bivariate analysis is used to explore possible confounding variables such as age, gender and ethnicity via simple contingency table analysis and ANOVA. Pearson product moment correlation coefficient matrixes investigate the relationships (either positive or negative) in the percent change in BMI and all predictor variables. We pool data within-cells to account for our randomized group study design that essentially blocks at the center level (PROC DISCRIM procedure in the SAS statistical package). Multivariate. Repeated measures linear regression models are used to determine patterns of progression or improvement in BMI percentile over time, as well as to identify demographic and family characteristics associated with BMI outcomes. Repeated measures analysis of covariance (ANCOVA) determines the relationships between BMI and our environmental and family modification (via the theoretical constructs) covariates. ANCOVA is used to increase power in a one-way or two-way ANOVA by adding a second or third variable as a covariate. ANCOVA was an attractive higher level method of analysis for our design because it is capable of removing the obscuring effects of pre-existing individual differences among subjects while also allowing for compensation for systematic biases among our samples. Cost Analysis. Resource use and cost data for the intervention are being collected across all intervention centers. Key statistics from the cost evaluation will include total annual economic cost for the intervention and weekly economic cost per child.
Healthy People 2020 identifies “Nutrition and Weight Status” as one of their major objectives to “promote health and reduce chronic disease risk through the consumption of healthful diets and achievement and maintenance of healthy body weights.” This objective emphasizes that efforts to change diet and weight should address the policies and environments that support these behaviors in settings such as schools . Although few attempts have been made to prevent obesity during the first years of life, this period may represent the best opportunity for obesity prevention. Findings from this investigation should inform both the fields of childhood obesity prevention and early childhood research about the effects of an obesity prevention program housed in the childcare setting. As childhood obesity continues to be a topic of national interest is in the best interest of health care providers, educators, and parents to begin thinking about this as a priority for ensuring the healthy future of our community’s children.
The Institute of Medicine’s (IOM’s) Standing Committee on Childhood Obesity Prevention recently commenced a study to examine the evidence and provide guidance on obesity prevention policies for children ages birth to five years old. Based on this study’s findings, in 2011 this committee created a set of policy recommendations designed specifically to prevent obesity in early childhood by promoting healthy early environments in settings outside the home where young children spend substantial time, namely childcare settings in particular . Recommendations were formulated using the best evidence available, including both direct (published literature, organizational reports) and indirect (expert input) data about the likely impact of a given policy on reducing childhood obesity. The committee’s overreaching goal was to have the report “find its way to federal, state and local government policy makers who work in areas that impact young children.” However, they acknowledge that “it will be important to act today based on what is known, while also undertaking the necessary research and policy evaluation to ensure better informed and effective actions in the future.” To date, the concept that incorporating these policies in the childcare setting will result in reduced obesity rates, increased access to nutritious foods, increased physical activity and decreased inactivity has been largely based on conjecture, not on scientific evidence produced by adequately controlled trials, making this study all the more timely.
Limitations and strengths of study design
We are limited in our ability to generalize our research findings to children who are older than 5 years and to families that are not from low income backgrounds. Strengths of the study include the design, as randomized controlled trials are considered the “gold standard” in terms of design rigor and ethnic diversity of the sample.
Although few attempts have been made to prevent obesity during the first years of life, this period may represent the best opportunity for obesity prevention. Findings from this investigation should inform both the fields of childhood obesity prevention and early childhood research about the effects of an obesity prevention program housed in the childcare setting.
BMI: Body mass index; CS: Curriculum specialists; HC2: Healthy Caregivers Healthy Children; OWG: Organ Wise Guys; USDA: United States Department of Agriculture.
No conflict of interest to report for all authors.
RN conceived of the study, and participated in its design and coordination and helped to draft the manuscript. SHS assisted with data collection and helped draft the manuscript. SM participated in the design of the study, performed the power analyses, and helped draft the manuscript. MS carried out the data collection. SU assisted with Internal Review Board approval and data collection procedures. AD participated in the design of the study. All authors read and approved the final manuscript.
We would like to acknowledge all center directors, parents, teachers, and children who participated in the study to make it a success. This research is funded by the United States Department of Agriculture NRI/ AFRI Grant # 2009–05065.
Pediatrics 2006, 118:e594-e601.
Erratum in: Pediatrics 2006, 118:2270PubMed Abstract | Publisher Full Text
Obesity among low-income preschool children.
Retrieved May 19, 2012 from http://www.cdc.gov/obesity/downloads/PedNSSFactSheet.pdf webcite
J of Clin Endocrin Metabol 2009, 94:1527-1532. Publisher Full Text
Child Dev Persp 2010, 4:138-143. Publisher Full Text
Federal Interagency Forum on Child and Family Statistics: America’s Children: Key National Indicators of Well-Being. Washington, DC: U.S. Government Printing Office; 2007. PubMed Abstract | Publisher Full Text | PubMed Central Full Text
Fitzgibbon ML, Stolley MR, Schiffer L, Van Horn L, KauferChristoffel K, Dyer A: Two-year follow-up results for Hip-Hop to Health, Jr.: randomized controlled trial for overweight prevention in preschool minority children.
J Ped 2005, 146:618-625. Publisher Full Text
Res Q Exerc Sport 2005, 76:130-139. PubMed Abstract
J Cult Divers 2004, 11:12-17. PubMed Abstract
Retrieved June 3, 2012, from http://www.mypyramid.gov/Challenge/downloads/ProjectMOMTools.pdf webcite
Hollar D, Messiah SE, Lopez-Mitnik G, Hollar TL, Almon M, Agatston AS: Effect of an elementary school-based obesity prevention intervention on weight and academic performance among low income children.
Am J Pub Health 2010, 100:646-653. Publisher Full Text
J Amer Diet Assoc 2010, 110:261-267. Publisher Full Text
Food Qual Pref 2003, 14(4):289-297. Publisher Full Text
Model local schools wellness policies on physical activity and nutrition. 2005.
Institute of Medicine School Food Recommendations [http:/ / www.iom.edu/ Reports/ 2009/ School-Meals-Building-Blocks-for-He althy-Children.aspx] webcite
(accessed June 11, 2012)
American Public Health Association & National Resource Center for Health Safety in Child Care and Early Education: Preventing Childhood Obesity in Early Care and Education Programs: Selected Standards from Caring for Our Children: National Health and Safety Performance Standards. Aurora, CO: NRC; 2010.
Centers for Disease Control and Prevention (CDC): National Center for Health Statistics (NCHS). National Health and Nutrition Examination Anthropometric Procedural Manual. Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention; 2002.
Banna JC, Townsend MS: Assessing factorial and convergent validity and reliability of a food behaviour checklist for Spanish-speaking participants in US Department of Agriculture nutrition education programmes.
J Amer Diet Assoc 2007, 107:2120-2124. Publisher Full Text
Kumar D, Sanders L, Perrin EM, Lokker N, Patterson B, Gunn V, Finkle J, Franco V, Choi L, Rothman RL: Parental understanding of infant health information: health literacy, numeracy, and the Parental Health Literacy Activities Test (PHLAT).
Perry CL, Parcel GS, Stone EJ, Nader P, McKinlay SM, Luepker RV, Webber LS: The Child and Adolescent Trial for Cardiovascular Health (CATCH): overview of the intervention program and evaluation methods.
Healthy People 2020. 2010.
http://www.healthypeople.gov/2020/topicsobjectives2020/objectiveslist.aspx?topicId=33 webcite. (accessed June 11, 2012)
The pre-publication history for this paper can be accessed here: