Methods
Data Source
The Medical Expenditure Panel Survey (MEPS) is a nationally representative survey of the US civilian, noninstitutionalized population conducted since 1996 by the Agency for Healthcare Research Quality and the National Center for Health Statistics. Details of the MEPS data collection process are available. Briefly, participants in the MEPS are selected from the National Health Interview Survey. We used the Household Component of MEPS (MEPS-HC). MEPS-HC collects data on demographics, health insurance, and other health-related items from household members, who are surveyed during 2 full calendar years. The sample design of MEPS-HC includes stratification, clustering, multiple stages of selection, and disproportionate sampling. MEPS collects supplemental information on responses from the MEPS-HC through a medical provider component, consisting of objective information from hospitals, pharmacies, and medical providers. MEPS maps medical International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes on the basis of medical and pharmacy use and self-report. To represent the noninstitutionalized US population, MEPS uses sample weights to adjust for factors related to survey design and underresponse.
Study Population and Ascertainment of Characteristics
The study sample was 188,006 MEPS respondents from 2002 through 2009. The mean response rate during this period was 60.9% (range, 56.9%–64.7%). MEPS was reviewed and approved by the Westat institutional review board, established under a multiproject assurance (M-1531) granted by the Office for Protection from Research Risks.
All respondents aged 18 years or older were asked if they had ever received low-fat dietary advice from a health professional: "Has a doctor or other health professional ever advised you to eat fewer high-fat or high-cholesterol foods?" We used the term "low-fat dietary advice" to refer to participants' response to this question. We used MEPS data to determine whether a respondent had diabetes or risk factors for developing type 2 diabetes. Respondents who answered yes to the following question on the MEPS-HC were classified as having diabetes: "Have you ever been told by a doctor or health professional that you have diabetes?" Similar to other epidemiologic surveys, MEPS does not differentiate between type 1 and type 2 diabetes. Because the Centers for Disease Control and Prevention estimates that more than 90% of adults who have diabetes have type 2 diabetes, we assumed that most of our sample who had diabetes had type 2 diabetes.
We used the most recent ADA criteria to classify risk factors for type 2 diabetes. The 7 ADA-designated risk factors represented in the MEPS survey were an age of 45 years or older, Hispanic ethnicity or nonwhite race, body mass index (BMI) of 25.0 kg/m2 or more, physical inactivity, hypertension (ICD-9-CM 401), hyperlipidemia (ICD-9-CM 272), and a history of CVD. Age, race, and ethnicity were self-reported. We calculated BMI from self-reported height and weight. Physical inactivity was determined by a negative response to the question, "Do you spend half an hour or more in moderate or vigorous physical activity at least 3 times a week?" Hypertension and hyperlipidemia were designated by the ICD-9-CM code on the MEPS medical provider component. A history of CVD was designated by participants' positive response to the question on whether they had ever been diagnosed with any of the following: coronary heart disease, angina, heart attack, or stroke.
Statistical Analysis
We performed a cross-sectional analysis of data from the 2002–2009 MEPS. To adjust for the complex sample design and ensure nationally representative estimates, we used MEPS person-level and variance-adjustment weights using Stata version 11 (StataCorp LP, College Station, Texas) for all analyses. We conducted χ tests to compare rates of receiving low-fat dietary advice among selected subgroups. Multivariate logistic regression analysis was used to estimate the adjusted odds of receiving low-fat dietary advice by the following factors: demographic characteristics (sex, age, race, ethnicity, geographic region, education, and income), general health-related characteristics (health insurance and smoking status), diabetes status, single risk factors for type 2 diabetes, and cardiometabolic multimorbidity (ie, total number of risk factors for type 2 diabetes). For income, we grouped survey respondent by the federal poverty index developed by MEPS. In separate regression analyses, independent variables included demographic and general health-related characteristics, diabetes status, risk factors for type 2 diabetes, dummy independent variables representing the number of ADA-designated risk factors for type 2 diabetes (range, 0–7), a modified clinical comorbidity index, and dummy variables for each study year, using 2002 as the reference year. The dummy independent variables for the risk factors for type 2 diabetes indicated the total number of risk factors for each respondent (ie, 1 risk factor, 2 risk factors, and so on, up to a maximum of 7 risk factors). The modified clinical comorbidity index represented the number of chronic conditions other than the comorbidities of diabetes or risk factors for diabetes included in the regression model. We also examined trends in the unadjusted rates and adjusted likelihood of receiving advice during the study period. We calculated adjusted odds ratios and 95% confidence intervals (CIs). Significance was set at P < .05.