Chengbin Hu

Obese and overweight: problem and how to control? Problem: More than 2 in 3 adults are considered to be overweight or obese. More than 1 in 3 adults are considered to be obese. More than 1 in 20 adults are considered to have extreme obesity. Providers under-identify overweight and obese patients.

Main study question: Does application of EHR-based tools have significant effect on patient weight change? Other questions: Whether HER-based tools affect the proportion of patients who had a diagnosis of overweight or obesity on the electronic health record problem list, and the proportion of who had a follow up appointment about their weight or were prescribed weight loss medication. Solution: Electronic health record-based tools

EHR-based tools 1. Reminders to measure height and weight. 2. An alert asking providers whether they want to add overweight or obesity to the problem list, for patients with BMI 25–29.9 or 30 kg/m2, respectively. 3. Reminders with tailored management recommendations, based on patients’ BMI and other risk factors (e.g. hypertension, hyperlipidemia, type 2 diabetes) included on the problem list or identified from medications or laboratory results. 4. A Weight Management screen

Study design: randomization Three strata: Cluster randomization: clinics as the unit of randomization(why?)

1.Two Phases in this study according to the availability of new HER features. 2.Two arms(control and intervention clinics) 3.Only care providers in clinics need the information of this project, patients do not need to be informed. Study design: Control and Study group Main Endpoint: 12 months follow-up study

Study population, data collection, and outcomes Population: Phase 1: All adult patients(aged 20 or older) who had a visit at one of the intervention or control clinics between 15 December 2011 and 10 June Phase 2: All adult patients who had a visit at one of the intervention or control clinics between 11 June 2012 and 10 December 2012 and had a BMI ≥ 25 kg/m 2. Outcomes: Phase 1: The proportion of patients with a documented BMI in the EHR within 12 months after the initial visit in the study ‘‘enrollment’’ period. Phase 2: The primary outcomes were 6-month and 12-month weight change. Secondary outcome measures for Phase 2 included the proportion of patients with BMI ≥ 25 who had a diagnosis of overweight or obesity on the problem list; the proportion of patients with BMI ≥ 25 who had a subsequent appointment with a provider. The proportion of patients with BMI ≥ 25 who were prescribed weight loss medications. In addition, they assessed providers’ attitudes about management of overweight and obesity using web-based surveys.

Power and sample size Weight change difference: 2.2 to 7.1 pounds. Standard deviation: 9 pounds Method: two-sided Student’s t-test with 90% power and 0.05 alpha. Result: NperGroup = 427 For intra-provider correlation: N 1 : sample size under the assumption of independence. m: average number of patients per provider ρ: intraclass correlation coefficient Required sample size would be 2541 Necessary sample size for the secondary outcome of diagnosis of overweight and obesity among patients 10% increase as clinically meaningful Method: two-sided Chi-square test with 90% power and 0.05 alpha. NperGroup = 347 For intra-provider correlation: Required sample size would be 2065

Substudy Rational: To study detailed information about their experiences with weight management and about discussion of weight management with their primary care providers, as these outcomes are difficult to assess using data from the HER. 200 patients enrolled in this substudy. Power = 86% assuming ICC =0.05 to detect 4 pounds difference.

Data analysis Phase 1 outcomes: Mixed-effects logistic regression models SAS PROC GLIMMIX Phase 2 outcomes: Primary: mixed-effects linear regression models SAS PROC MIXED Secondary: similar as phase 1 No final data reported. Missing Data? Use Markov chain Monte Carlo multiple imputation to predict missing weights based on other covariates SAS PROC MI

Discussion Why is this study important? Providers under- identify the weight problem. Few studies focused on EHS features to assist providers with management of weight problems. Almost no other studies examined effects on patient outcomes.

Limitation EHR architecture and clinical workflow is stable; it is hard to add intervention during this process. It is impossible to completely randomize at the level of individual patient or provider. Cluster randomization cannot balance the patient characteristics. Randomizing clinics in the same practices may result in contamination. EHR data cannot insure visits at regular intervals, resulting in missing data.