Improving children's obesityrelated health care ... - Wiley Online Library

Obesity

Brief Cutting Edge Report PEDIATRIC OBESITY

Improving Children’s Obesity-Related Health Care Quality: Process Outcomes of a Cluster-Randomized Controlled Trial Elsie M. Taveras1,2, Richard Marshall3, Christine M. Horan2, Matthew W. Gillman2, Karen Hacker4, Ken P. Kleinman2, Renata Koziol2, Sarah Price2, Sheryl L. Rifas-Shiman2 and Steven R. Simon5,6

Objective: To examine the extent to which an intervention using electronic decision support delivered to pediatricians at the point-of-care of obese children, with or without direct-to-parent outreach, improved health care quality measures for child obesity. Design and Methods: Process outcomes from a three-arm, cluster-randomized trial from 14 pediatric practices in Massachusetts were reported. Participants were 549 children aged 6-12 years with body mass index (BMI) 95th percentile. In five practices (Intervention-1), pediatricians receive electronic decision support at the point-of-care. In five other practices (Intervention-2), pediatricians receive pointof-care decision support and parents receive information about their child’s prior BMI before their scheduled visit. Four practices receive usual care. The main outcomes were Healthcare Effectiveness Data and Information Set (HEDIS) performance measures for child obesity: documentation of BMI percentile and use of counseling codes for nutrition or physical activity. Results: Compared to the usual care condition, participants in Intervention-2, but not Intervention-1, had substantially higher odds of use of HEDIS codes for BMI percentile documentation (adjusted OR: 3.97; 95% CI: 1.92, 8.23) and higher prevalence of use of HEDIS codes for counseling for nutrition or physical activity (adjusted predicted prevalence 20.3% [95% CI 8.5, 41.2] for Intervention 22 vs. 0.0% [0.0, 2.0] for usual care). Conclusion: An intervention that included both decision support for clinicians and outreach to parents resulted in improved health care quality measures for child obesity. Obesity (2014) 22, 27-31. doi:10.1002/oby.20612

Introduction In recognition of the public health importance of childhood obesity, in 2009, the National Committee for Quality Assurance released new nationally standardized performance measures (Healthcare Effectiveness Data and Information Set [HEDIS]) focused on improving quality of care for obesity in children (1). More recently, in 2010, The United States Preventive Service Task Force recommended screening for obesity in children as young as 6 years of age (2). Despite availability of HEDIS quality measures for obesity and obesity management guidelines for nearly a decade, providers have

been slow to adopt recommended practices (3). Although providers often cite barriers such as limited time, skill, and resources, a frequently overlooked barrier is the lack of data systems to improve quality of care. The use of electronic health records (EHRs) offers potential for improving the quality of care for obese children and for accelerating the adoption of research evidence regarding childhood obesity screening and management. It also holds promise for establishing treatment benchmarks and for supporting patients and their clinical teams in care improvement (4,5). Incorporation of point-of-care health information technologies may be especially effective if augmented by outreach to parents and

1

Division of General Pediatrics, Department of Pediatrics, Massachusetts General Hospital for Children, Boston, Massachusetts, USA. Correspondence: Elsie M. Taveras ([email protected]) 2 Obesity Prevention Program, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, Massachusetts, USA 3 Department of Pediatrics, Harvard Vanguard Medical Associates, Boston, Massachusetts, USA 4 Institute for Community Health, Department of Medicine, Cambridge Health Alliance, Harvard Medical School, Cambridge, Massachusetts, USA 5 Section of General Internal Medicine, Veterans Affairs (VA) Boston Healthcare System, Boston, Massachusetts, USA 6 Division of General Internal Medicine, Brigham and Women’s Hospital, Boston, Massachusetts, USA Additional Supporting Information may be found in the online version of this article. Funding agencies: American Recovery and Reinvestment Act (Award #R18 AE000026). Disclosures: The authors declare no conflict of interest. Author Contributions: Study concept and design: Taveras EM, Kleinman K, Gillman MW, Hacker K, Marshall R, Simon SR; Acquisition of data: Horan, CM, Koziol R. Analysis and interpretation of data: Taveras EM, Rifas-Shiman SL, Kleinman K; Drafting of manuscript: Taveras EM, Rifas-Shiman SL; Critical revision of the manuscript for important intellectual content: Taveras EM, Rifas-Shiman SL, Kleinman K, Gillman MW, Hacker K, Marshall R, Horan CM, Price S, Koziol R, Simon SR; Statistical analysis: Rifas-Shiman SL, Kleinman KP; Obtained funding: Taveras EM; Marshall R; Administrative, technical, or material support: Horan CM, Price S, Koziol R; Study supervision: Taveras EM, Marshall R. Received: 1 February 2013; Accepted: 16 August 2013; Published online 26 August 2013. doi:10.1002/oby.20612

www.obesityjournal.org

Obesity | VOLUME 22 | NUMBER 1 | JANUARY 2014

27

Obesity

Improving Children’s Obesity-Related Health Care Quality Taveras et al.

children to create “informed, activated families” defined in the widely adopted Chronic Care Model as patients who have the motivation, knowledge, skills, and confidence to make effective decisions to manage their health (6,7). The purpose of this study is to report process outcomes of an ongoing randomized controlled trial to accelerate the adoption of effective childhood obesity management strategies, including improving the use of health care quality measures, for children aged 6-12 years (8).

Methods Study design, setting, and randomization The Study of Technology to Accelerate Research (STAR) is a cluster-randomized controlled trial being conducted within 14 pediatric offices of Harvard Vanguard Medical Associates (HVMA), a multi-specialty group practice in Massachusetts. The overarching model for the STAR intervention is the Chronic Care Model (9). We randomly assigned each practice to one of three intervention arms: (i) electronic point-of-care decision support to pediatric primary care providers (Intervention-1; five practices); (ii) electronic decision support plus direct-to-parent outreach relating to their child’s BMI (Intervention-2; five practices); and (iii) usual care (control; four practices). We did not have sufficient practices/clusters to allow a fourth arm of direct-to-parent outreach only. The primary outcomes of the trial will be child BMI and obesity-related behaviors over a 1-year intervention period. This article reports the results of point-of-care process outcomes after the initial study visit. We made three strata of practices based on patient volume. A blinded biostatistician used a pseudo-random number generator in SAS to assign practices to one of the two intervention arms or to the control condition.

Eligibility and recruitment Details of the study design, eligibility, and recruitment have been published (10). Eligibility for STAR includes (i) child is 6.0-12.9 years old, (ii) child’s BMI 95th percentile for age and sex at the

initial well child visit, (iii) child has received well child care at HVMA within the past 15 months, and (iv) at least one parent can communicate in English. Recruitment was from October 2011 through June 2012. After receiving permission from pediatricians to contact eligible patients, study staff sent each family a letter 1 month prior to the child’s scheduled well child visit. The letter included an opt-out phone number to call if parents did not want to be contacted. Research assistants who were blinded to intervention groups called families and established eligibility, obtained verbal consent, and completed a survey over the phone. We obtained written informed consent via mail. All study activities were approved by the Human Subjects Committee at Harvard Pilgrim Health Care.

Intervention arms Intervention. Electronic point-of-care alerts. In the 10 practices randomized to the two intervention arms, we modified the existing EHR to deploy a computerized, point-of-care alert to pediatricians at the time of a well child care visit for a child with a BMI 95th percentile (Figure 1). We designed the alert to trigger as a new window “in front of” the clinician’s screen. The alert contains links to the Centers for Disease Control and Prevention (CDC) growth charts, links to existing childhood obesity comparative effectiveness research (CER) evidence, and a link to a pre-populated, standardized note specific for obesity that includes links for (i) documenting and coding for BMI percentile and diagnosis of obesity, (ii) documenting and coding for nutrition and physical activity counseling, (iii) placing referrals for weight management programs, (iv) placing orders for obesity-related laboratory studies if appropriate, and (v) printing patient education information.

Direct to parent outreach. In Intervention-2, practices have letters sent to their enrolled families in addition to the electronic decision support tools available for their clinicians. The letter was written at a sixth grade literacy level. Prior to the initial study well child visit, we mail parents a letter that provides an explanation of their child’s BMI from their well child care visit 1 year prior to enrollment (pre-intervention visit). The letter encourages parents

FIGURE 1 Computerized, point-of-care alert developed for the Study of Technology to Accelerate Research (STAR) Intervention, to alert pediatric clinicians at the time of a well child care visit of a child between the ages of 6 and 12 years with a BMI 95th percentile. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

28



Brief Cutting Edge Report

Obesity

PEDIATRIC OBESITY

FIGURE 2 Letter sent to parents in Intervention-2 practices prior to their children’s upcoming well child care visit with information related to their children’s most recent body mass index within the last 12 months. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]



29

Obesity

Improving Children’s Obesity-Related Health Care Quality Taveras et al.

to discuss BMI with their doctor at their child’s upcoming visit (Figure 2).

Usual Care. Participants at practices randomized to the control group received the current standard of care offered by their pediatric office. No electronic decision support tools for obesity were made available to the four usual care practices.

Outcome measures We collect outcome measures using the child’s EHR from well child care visits. The main outcomes for this analysis were HEDIS measures of International Statistical Classification of Diseases (ICD) ninth edition diagnostic coding for (i) BMI percentile (ICD-9 code V85.5) and (ii) counseling for nutrition (ICD-9 code V65.3) or physical activity (ICD-9 code V65.41) (1). Visits that contained specific documentation and use of each ICD-9 diagnosis code were considered meeting the HEDIS requirements.

percentile documentation, the odds ratio can be a poor analogue for the risk ratio. Thus, we present predicted prevalence in addition to odds ratios to provide a more accurate assessment of the effects. Using parameter estimates from the adjusted GLIMMIX model, we estimated the predicted prevalence (95% confidence interval [CI]) of having nutrition or physical activity counseling codes for Intervention-1 and Intervention-2 participants. We chose fixed covariate values representing an “average” participant, that is, the mean value for parent age (40.6 years) and distributions of the categorical covariates in our sample (marital status, parent place of birth, child’s race/ethnicity). At the pre-intervention visit, no participant in the study had documentation of HEDIS counseling codes for nutrition or physical activity, so OR could not be calculated. For usual care participants, we assumed a 95% CI of 0.0-2.0% (11). All analyses were performed in SAS 9.3 (SAS Institute, Cary, NC, USA).

Statistical analysis

Result

Our intent-to-treat analyses used unadjusted and adjusted generalized linear mixed models, with repeated measures, accounting for clustering by practice. We examined differences in the odds and predicted prevalence of BMI percentile documentation from the preintervention to initial study well child care visits between each intervention arm and usual care group. We included a term for time (pre-intervention or initial study visit), status (Intervention-1, Intervention-2, or usual care), and interaction (time 3 status). In settings where the outcome prevalence is high, as it was for BMI

Using the EHR, we identified and attempted contact with 2242 children who had a BMI 90th percentile sometime within the year prior to their initial study well child care visit. We pre-enrolled 817 children, of whom 549 met final enrollment eligibility (CONSORTtype flow chart available as Supporting Information). At the initial study visit, mean (SD) age of the children was 9.8 (1.9) years; mean (SD) BMI was 25.8 (4.3) kg/m2; 68% lived in households with annual incomes >$50,000; and 51% were White, 21% were Black, 14% were Hispanic, and 14% were of multiracial or other race/

TABLE 1 Multivariable adjusted odds and predicted prevalence of documentation of HEDIS performance measures for childhood obesity

Odds ratio (95% confidence interval) HEDIS performance measures for childhood obesity

Unadjusted

Adjustedb

BMI percentile documentationa Intervention-1 vs. usual care Intervention-2 vs. usual care

1.40 (0.73, 2.68) 3.80 (1.86, 7.77)

1.46 (0.75, 2.84) 3.97 (1.92, 8.23)

Predicted prevalence (95% confidence interval) Adjustedb

Unadjusted

BMI percentile documentation Intervention-1 Intervention-2 Usual care Nutrition or physical activity counseling documentationc Intervention-1 Intervention-2 Usual care

Pre-intervention

Initial study visit

Pre-intervention

Initial study visit

46.3% (27.0, 66.8) 66.5% (45.4, 82.6) 66.2% (43.7, 83.2)

57.6% (36.7, 76.0) 89.5% (77.1, 95.6) 68.6% (46.4, 84.7)

46.9% (27.8, 66.9) 65.7% (44.9, 81.8) 65.4% (43.4, 82.4)

57.9% (37.5, 75.9) 89.1% (76.6, 95.3) 66.7% (44.7, 83.2)

0.0% 0.0% 0.0%

6.4% (0.4, 51.3) 23.7% (10.6, 44.9) 0.0% (0.0, 2.0)

0.0% 0.0% 0.0%

3.5% (0.1, 52.9) 20.3% (8.5, 41.2) 0.0% (0.0, 2.0)

a

Repeated measures analysis among 486 participants who had a BMI 95th percentile at their pre-intervention visit. Adjusted for differences in parent age, marital status, US born, and child race/ethnicity. c Predicted prevalence based on an “average” participant, that is, mean value for parent age (40.6 years) and distributions of the categorical covariates in our sample (marital status, parent place of birth, child’s race/ethnicity). b

30



Brief Cutting Edge Report

Obesity

PEDIATRIC OBESITY

ethnicity. The parents were a mean (SD) age of 40.6 (6.5) years; 74% were married or cohabitating; 75% were born in the United States (Supporting Information table). Table 1 shows multivariable adjusted odds and predicted prevalence of HEDIS performance measures at initial study well child care visits. Compared to the usual care condition, participants in Intervention-2 (adjusted OR 3.97 [95% CI 1.92, 8.23]), but not Intervention-1 (1.46 [0.75, 2.84]), had substantially higher odds and predicted prevalence of BMI percentile documentation from the pre-intervention to initial study visits. Predicted prevalence (95% CI) of having counseling codes for nutrition or physical activity for Intervention-2 was 20.3% (95% CI 8.5%, 41.2%) and for Intervention-1 was 3.5% [95% CI 0.1%, 52.9%]. The CI for Intervention-2, but not Intervention-1, excluded the CI for the usual care group (95% CI 0.0%, 2.0%).

Discussion In this analysis of process outcomes of a cluster-randomized controlled trial, we found that an intervention that included both decision support for clinicians and outreach to parents appeared to be effective in improving health care quality measures for child obesity. Solely providing electronic decision support tools to clinicians without also “activating” parents prior to their children’s pediatric visits was not associated with the use of obesity-related quality measures. It is possible that clinician behavior change in the practices that only received the electronic decision support is lagging behind that of the clinicians in the practices receiving both the EHR alert and parent activation. Thus, we will evaluate the effects of the intervention after the planned 1-year intervention Our findings lend support to a growing body of evidence that computerized decision support tools can improve clinician performance (12-15). But they also suggest that for childhood obesity, outreach to families prior to their visits may be needed to augment point-ofcare health information technologies. Thus, these findings highlight the value of both informed, activated families, and prepared, proactive practice teams, which are central tenets of the Chronic Care Model (8). As in any study, this one is subject to potential limitations. One is generalizability. Much pediatric primary care is currently provided in settings unlike HVMA, that is, smaller practices without an EHR. However, as a relatively large medical group, HVMA is a typical primary care setting for many children, and EHRs are increasingly penetrating even small practices. Thus, the STAR intervention is likely to generalize to more and more pediatric settings in the future. Furthermore, our study was conducted among families with relatively high socioeconomic backgrounds and with good access to health care. It is possible that intervention effects could differ among families in low resource settings.


Studying the effects of STAR on individual-level weight and behavior change will be critical next steps. Interventions that take advantage of efficient health information technologies and parent outreach have the potential to provide sustainable, low-cost, high-reach approaches for accelerating adoption of comparative effectiveness evidence for childhood obesity, for improving quality of care for childhood obesity in pediatric primary care, and for effectively supporting patients and families in improving obesity-related behaviors. O

Acknowledgments Dr. Taveras had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

C 2013 The Obesity Society V

References 1. National Committee for Quality Assurance. Body mass index (BMI) percentile assessment and counseling for nutrition and physical activity (for children/ adolescents) (BCA), 2009 proposed Healthcare Effectiveness Data and Information Set (HEDIS) measures 2008; http://www.ncqa.org/Portals/0/PublicComment/ HEDIS2009/BCA_Specs_and_Workup_PDF.pdf. 2. US Preventive Services Task Force. Screening for obesity in children and adolescents: US preventive services task force recommendation statement. Pediatrics. 2010;125:361-367. 3. Huang TT, Borowski LA, Liu B, et al. Pediatricians’ and family physicians’ weight-related care of children in the US. Am J Prev Med 2011;41:24-32. 4. Masica AL, Ballard DJ. The protean role of health care delivery organizations in comparative effectiveness research. Mayo Clin Proc 2009;84:1062-1064. 5. Conway PH, Clancy C. Transformation of health care at the front line. JAMA 2009; 301:763-765. 6. Nihiser AJ, Lee SM, Wechsler H, et al. Body mass index measurement in schools. J Sch Health 2007;77:651-671; quiz 722-654. 7. Thompson JW, Card-Higginson P. Arkansas’ experience: statewide surveillance and parental information on the child obesity epidemic. Pediatrics 2009;124:S73-S82. 8. Fiks AG, Grundmeier RW, Margolis B, et al. Comparative effectiveness research using the electronic medical record: an emerging area of investigation in pediatric primary care. J Pediatr 2012;160:719-724. 9. Wagner EH. Chronic disease management: what will it take to improve care for chronic illness? Eff Clin Pract 1998;1 2-4. 10. Taveras EM, Marshall R, Horan CM, et al. Rationale and design of the STAR randomized controlled trial to accelerate adoption of childhood obesity comparative effectiveness research. Contemp Clin Trials 2012;34:101-108. 11. Ulm K. A simple method to calculate the confidence interval of a standardized mortality ratio (SMR). Am J Epidemiol 1990;131:373-375. 12. Garg AX, Adhikari NK, McDonald H, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA 2005;293:1223-1238. 13. Bell LM, Grundmeier R, Localio R, et al. Electronic health record-based decision support to improve asthma care: a cluster-randomized trial. Pediatrics 2010;125: e770-e777. 14. Fiks AG, Grundmeier RW, Biggs LM, Localio AR, Alessandrini EA. Impact of clinical alerts within an electronic health record on routine childhood immunization in an urban pediatric population. Pediatrics 2007;120:707-714. 15. Keehbauch J, Miguel GS, Drapiza L, Pepe J, Bogue R, Smith-Dixon A. Increased documentation and management of pediatric obesity following implementation of an EMR upgrade and education. Clin Pediatr 2012;51:31-38.


31