Study population
As part of an ongoing QI project, we prospectively collected clinical and demographic data from patients under the age of 25 years who were assessed in the Preventive Cardiology Program at a large regional medical center from September 1, 2010 to March 31, 2016. Patients were eligible for this study if they had an initial visit to the Preventive Cardiology Program during the observation period with a lipid profile that included a total cholesterol, triglyceride level, HDL-C, and LDL-C, regardless of the reason for referral. For example, patients who were referred to the Preventive Cardiology Program for hypertension, but obtained a lipid profile as part of the visit. Patients were excluded if they were prescribed lipid-lowering medications during follow up at any time during the observation period, if they were over the age of 25 years, or if they did not have a baseline visit in the observation period. This study was approved by the research ethics board with a waiver of individual participant consent.
Study design
The QI initiative was implemented for all children and adolescents seen for a lipid disorder in a Preventive Cardiology Program as part of a large, regional medical system starting September 1, 2010. This initiative provided management suggestions and prospectively collected clinical data via the use of a Standardized Clinical Assessment and Management Plan (SCAMP). Providers (physicians, nurse practitioners, and nurses) completed the standardized forms for each patient encounter and when interval laboratory measurements were obtained. In addition to collecting pertinent health information, the SCAMP suggested management by way of treatment algorithms generally consistent with the 2011 US National Heart, Lung, and Blood Institute Pediatric Integrated CVD Guidelines [10]. Patients were counseled on heart healthy lifestyle behaviors based on the 2011 Guidelines; patients visited with a registered dietician and a nurse practitioner or physician. The intervention included nutrition and physical activity counseling tailored for the type of lipid disorder. For example, for those patients with high triglycerides, the focus was on reducing intake of simple sugars and replacing them with vegetables and whole grains. In contrast, for patients with elevated LDL-C, the focus was more on reducing saturated fat and increasing dietary fiber intake in the form of fruits, vegetables and whole grains. Physical activity counseling focused on stepwise increases in activity to target a minimum of 5 h of moderate-to-vigorous physical activity per week. Patients had the opportunity to meet with a registered dietician at each visit. In addition to in-person counselling, patients were provided with printed educational materials. After three visits, additional visits did not necessarily include a visit with a dietician. Clinical information was extracted from this QI dataset and supplemented with information from the electronic health record (EHR).
Payer-type
The participant’s payer-type was determined for the purposes of this investigation from the EHR at the initial clinic visit and was held constant during all longitudinal analyses. Government-sponsored coverage was defined as having any coverage through Medicaid, the State Children’s Health Insurance Program, or Medicare. Private coverage was defined as having health insurance that was not provided through government-sponsored coverage and also included international patients. No patients were identified as uninsured in our study. Out of state patients had insurance coverage by their respective state and were classified as private or government-sponsored in a similar fashion. Payer-type was present on administrative paperwork at each visit.
Anthropometric and clinical assessments
Weight was recorded by a trained clinical provider via a standing scale (Scale-Tronix Stand-on Scale, Scale-Tronix, White Plains, NY) to the nearest 0.1 kg with patients in their own clothing without a jacket or shoes. However, it was not possible to know if a patient’s weight was measured on the same scale at each visit. Height was measured with a vertical stadiometer in patients without shoes to the nearest 0.1 cm. Body mass index (BMI) was calculated as weight in kilograms divided by height in centimeters squared. BMI percentiles and Z-scores were generated using the Centers for Disease Control (CDC) 2000 reference tables, as implemented in CDC EpiInfo version 7.
Laboratory assessments
Lipid measurements were obtained from peripheral blood samples, generally after an 8-h fast. Measurements were made using standard enzymatic assays. Total cholesterol, HDL-C, and triglycerides were measured and LDL-C was calculated according to the Friedewald equation [19]. Non-HDL-C, a surrogate for all atherogenic Apolipoprotein B containing particles, was determines as total cholesterol minus HDL-C. If triglycerides were ≥ 400 mg/dL (4.52 mmol/L), a direct LDL-C was obtained in most cases.
Statistical analysis
We generated summary statistics including counts, percentages, means and medians, as appropriate for distribution, to describe categorical and continuous study variables. Baseline characteristics were compared between payer-type with t-tests or chi-squared tests for continuous and categorical data, respectively. The triglyceride level was natural log transformed due to non-normal distribution. We defined abnormal lipid levels for males and females as an LDL-C greater than or equal to 130 mg/dL (3.3 mmol/L), an HDL-C less than 40 mg/dL (1.03 mmol/L), non-HDL-C greater than 145 mg/dL (3.75 mmol/L), and a triglyceride level greater than or equal to 150 mg/dL (1.69 mmol/L). Race/ethnicity was self-reported and categorized as White, Black, Hispanic, Asian, other, and unknown. BMI was categorized as underweight (BMI <5th percentile), normal (BMI 5th – 84th percentile), overweight (BMI ≥ 85th to 94th percentile), and obese (BMI ≥95th percentile) for age and sex in patients younger than 18 years old. For young adults ages 18–25 years old we used absolute BMI cutoffs of normal (BMI < 25 kg/m2), overweight (25 – < 30 kg/m2), and obese (≥30 kg/m2). Corresponding BMI categories were then combined for those younger and older than 18 years of age.
Baseline analyses
We tested for differences in lipid parameters (LDL-C, HDL-C, non-HDL-C, and natural log-transformed triglyceride level [lnTG]) in relation to payer-type by conducting multivariable adjusted linear regression models with the baseline lipid value as the dependent variable, payer-type as the independent variable of interest and adjusting for age, sex, and race/ethnicity. We ran a secondary model that additionally adjusted for BMI category (in addition to age, sex, and race/ethnicity) to determine if any differences in dyslipidemia based on payer-type could be attributed to differences in adiposity between payer groups. Finally, we ran separate models that additionally included an interaction term for payer-type and one of the covariates (age, sex, race/ethnicity, and BMI category). By adding the interaction term, we tested if the relationship between payer-type and lipid values was modified by the patient’s age, sex, race/ethnicity, or obesity/overweight status.
Changes in lipid parameters during lifestyle counseling
To compare the change in lipid values in response to treatment over time by payer-type, we conducted pairwise multivariable adjusted linear regression models. The dependent variable was the difference in the lipid parameter at baseline to each follow-up visit in separate models (i.e., differences in LDL-C from baseline to first follow up visit). We separately examined change from baseline to first through third visit, as the sample size became too small for comparisons beyond the third follow-up visit. As the time between visits was not consistent for each patient, we adjusted for the length of time between visits. Payer-type was the independent variable of interest, and the model was adjusted for age, sex, race/ethnicity, baseline lipid value, and days between baseline and follow-up visits. The pairwise analyses were restricted to patients with abnormal LDL-C, HDL-C, non-HDL-C, and triglycerides levels at the initial visit for the respective models for that lipid parameter. In the models assessing the change in lipid values between visits, we additionally tested for effect modification by age, sex, and race/ethnicity, and secondary adjustment for BMI category, as described for baseline analysis.
Statistical analyses were conducted with R statistical software (version 3.4.3) and a two-sided p-value < 0.05 was considered significant.