Variations between obese latinos and whites in weight-related counseling during preventive clinical visits in the United States


  • Jun Ma,

    Corresponding author
    1. Stanford Prevention Research Center, Stanford University, Stanford, California, USA
    • Department of Health Services Research, Palo Alto Medical Foundation Research Institute, Palo Alto, California, USA
    Search for more papers by this author
  • Lan Xiao,

    1. Department of Health Services Research, Palo Alto Medical Foundation Research Institute, Palo Alto, California, USA
    Search for more papers by this author
  • Veronica Yank

    1. Department of Health Services Research, Palo Alto Medical Foundation Research Institute, Palo Alto, California, USA
    2. Stanford Prevention Research Center, Stanford University, Stanford, California, USA
    Search for more papers by this author

  • Funding agencies: This research was supported by internal funding from the Palo Alto Medical Foundation Research Institute. Disclosure The authors declared no conflict of interest.

    The funding source had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript.

Correspondence: Jun Ma (



To examine rate differences and explanatory factors for lifestyle counseling to obese Latinos versus non-Hispanic whites (NHWs) in U.S. outpatient settings.

Design and Methods

The 2009 National Ambulatory Medical Care Survey data assessed the provision of weight-related lifestyle counseling during general medical exam visits (n = 688) by obese Latino and NHW adults. The Blinder-Oaxaca decomposition technique to identify the fraction of the overall ethnic difference in counseling rate explained by a selection of measured variables based on the Anderson-Newman-Aday behavioral model were used.


Although weight-related lifestyle counseling rates were low in both ethnic groups, the rate among obese Latinos (51.3%) was significantly higher than among NHWs (35.8%) (P = 0.03), with 60% of the difference explained by observed factors. Enabling factors such as provider specialty, metropolitan statistical area, practice type, and provider employment type contributed the most to higher counseling rates among Latinos, whereas geographic region, continuity of care, and health insurance were enabling factors that, along with the predisposing factor of sex, contributed the most in the opposite direction.


Obese Latinos are more likely to receive weight-related counseling during general medical exams than do NHWs, which is partly explained by physician practice and patient factors.


Latinos are the largest and fastest growing minority, currently comprising 16% of the U.S. population. Latino adults have a higher prevalence of overweight and obesity (77%) than non-Hispanic whites (NHWs) (68%) ([1]), and the highest rates of obesity-related risk factors (e.g., metabolic syndrome) and diseases (e.g., type 2 diabetes) ([2]). Concerted, sustainable efforts to prevent and treat obesity and to reduce obesity-related health disparities are federal priorities as emphasized in the U.S. Department of Health and Human Services' Healthy People 2020 initiatives ([3]).

The U.S. Preventive Task Force recommends that clinicians offer intensive lifestyle counseling for obesity ([4]), and recently, coverage of the service provided in primary care settings has been approved for Medicare beneficiaries ([5]). Research has shown that physicians providing counseling on weight and weight-related health behaviors—primarily healthy eating and physical activity—significantly increase awareness of unhealthy weight status, confidence and motivation to initiate lifestyle modification, and successful weight loss attempts among overweight and obese patients ([6]).

However, considerable opportunities are missed in the ascertainment and management of adult obesity in U.S. outpatient settings ([7]). Research on whether primary care-based lifestyle counseling for obesity management varies by patient race and ethnicity is not only limited but has also yielded conflicting results ([8]). The continuing racial and ethnic disparities in the prevalence of obesity and obesity-related comorbidities heighten the urgency of further investigations to elucidate potential gaps in the quality of obesity care by race and ethnicity.

The aims of this study were to ([1]) assess the direction and magnitude of differences between obese Latino and NHW adults in the rates of weight-related lifestyle counseling or referral during general medical exam (GME) visits and ([2]) explicate factors that contributed to the differences, if observed.


Data sources

The National Ambulatory Medical Care Survey (NAMCS) has been administered annually since 1989 by the National Center for Health Statistics (NCHS). We used the data from year 2009, which is the latest released public data at the time of study ([11]).

The NAMCS collects information on utilization and provision of health care during nationally representative samples of patient visits each year to non-federally funded, community, and office-based U.S. physician practices. The three-stage stratified probability sampling design entails sampling based on geographic locations, physician practices (stratified by specialty), and patient visits within the practices during randomly assigned reporting periods. An independent sample of practices is selected for each survey year. The response rate was 62.1% for the year 2009. Sampling weights account for selection probability, nonresponse adjustment, and other adjustments to reflect the universe of office-based U.S. patient visits.

The NCHS conducts quality control using a two-way independent verification procedure for 10% of the sample records in each survey year. Coding error rates for various items ranged from 0.2 to 0.5%, and item nonresponse rates for most data elements were <5.0% in 2009. The NAMCS has been validated against other national ambulatory care data ([12]).


This study focused on GMEs (n = 688) by obese Latinos and NHWs ages 18 years and older to providers primarily engaged in direct patient care. Unlike illness-related visits, GMEs are designed to address preventive health and lifestyle management issues that affect individual disease risk, such as diet, exercise, and obesity. Patients who met any of the following criteria were considered obese: ([1]) BMI ≥ 30 kg/m2, ([2]) visit diagnosis of obesity (ICD-9 code 278.0, excluding overweight 278.02), or ([3]) a check box on the NAMCS standardized encounter form indicating obesity as a current medical problem. Visits by pregnant women were excluded. The GME visits could be performed by health care providers from any specialty (i.e., both primary care clinicians and other types of specialists). We chose this inclusive definition for providers, because specialists of many types, not solely primary care providers, perform GMEs (e.g., endocrinologists following their patients with type 2 diabetes or cardiologists following patients with coronary artery disease) but would also be expected to provide weight-related lifestyle counseling to their patients who are obese at such visits. We also performed sensitivity analyses that restricted provider type to primary care specialties (e.g., general internal medicine and family medicine), which accounted for 83% of the study sample. These analyses had comparable results to those reported below.

The dichotomous primary outcome assessed whether weight-related counseling was addressed at the visit. Counseling was considered to have been addressed through its provision during the visit itself or, by extension, through a referral placed at the time of the visit (for counseling to be delivered by others). The NAMCS form does not distinguish between these two possibilities; rather, it contains a combined check box in which the physician can indicate that he or she “ordered or provided” counseling on specific topics. Of the choices of topic offered, those that met our criteria for weight-related counseling included any of the following: weight reduction, nutrition, or exercise.

As a conceptual framework to guide our choice of explanatory variables, we applied the behavioral model of Andersen, Newman, and Aday ([13, 14]) coupled with considerations of NAMCS data availability. We chose the Andersen-Newman-Aday model because it is one of the most widely applied behavioral models for explaining patient utilization of health care services and has been used to examine other groups of patients with chronic conditions ([15]). We applied the updated version of the model suggested by Phillips et al. that emphasizes inclusion of contextual factors (i.e., setting in which utilization occurs, such as practice location, practice type, or physician characteristics), in addition to factors specific to the individual (i.e., patient characteristics). The model defines three domains that may explain people's use of health services, namely, predisposing (e.g., age and level of health risk taking), enabling (e.g., health insurance status, location in metropolitan vs. other community, and physician characteristics), and need factors (e.g., degree of comorbidity) ([16]). Within the NAMCS data, we identified a priori important patient-, visit-, and practice-level factors that belonged to each of these three domains.

Predisposing factors

Patient demographics included age and sex. Self-reported cigarette smoking status was regarded as an indicator of the health risk-taking predisposition, which could influence health care provider's perception of the patient's willingness and ability to change lifestyle.

Enabling factors

NAMCS measured an array of visit and practice characteristics that within the Andersen-Newman-Aday model are considered to facilitate or impede use of health services. Visit characteristics included health insurance (private, Medicare/Medicaid, or other), continuity of care (established patient with number of visits in past year being 0, 1-2, 3-5, and 6 vs. new patient), time spent with provider (in minutes), type of provider seen (physician only, physician and nurse, or other), and visit disposition (return appointment, referral to other physician, or other). At the practice level, NAMCS provided the geographic region of the practice (Northeast, Midwest, South, or West) and whether the practice was in a metropolitan area. In addition, data were available on provider employment status (owner of the practice vs. employee/contractor), physician specialty group (primary care specialty vs. surgical or medical care specialties), and practice type (private solo practice, private group practice, federally qualified health center/HMO, or other office setting), and whether the practice had an electronic health record (EHR) system. The NAMCS also classified the EHR as basic or fully functional.

Need factor

As a measure of health status and need for medical care, we calculated a modified Charlson comorbidity index for each patient visit. The Charlson comorbidity index counts the presence or absence of 17 major comorbidities (e.g., cardiovascular disease, diabetes, and cancers) ([17]). It is the most widely studied measure of comorbidity and has been validated for use in outpatient populations ([18]). The NAMCS form includes predefined check boxes that overlap with 10 of the 17 Charlson comorbidities as well as write-in spaces for additional or overlapping visit-specific diagnoses that could be matched to the remaining seven Charlson comorbidities. In addition, we expanded the Charlson index to include other common obesity-related comorbidities: hypertension, hyperlipidemia, and osteoporosis. We also examined obesity-related comorbidities as individual factors; these had similar findings to those reported for the overall modified Charlson index.

Statistical analyses

The unit of analysis was the patient visit. Chi-square tests compared frequency distributions of categorical variables between obese Latinos and NHWs, and Student's t-tests evaluated mean differences for continuous variables.

Decomposition model

We used the regression-based Blinder-Oaxaca decomposition technique ([19]) to partition overall ethnic difference in the primary outcome into the unobserved part (part not explained by observed group differences in explanatory variables) and observed part, with the latter further subdivided by explanatory variables. First, we calculated mean values for each measured explanatory variable (called “factors” in the Andersen-Newman-Aday model and below) for obese Latinos and NHWs separately. Next, we estimated separate linear regression models to obtain the vector of coefficients for each ethnic group. The response variable was the binary indicator of whether weight-related counseling was addressed at a given visit. The linear models for each ethnic group included the same predisposing, enabling, and need factors as described above. Finally, the part of the overall difference in counseling rates that was explained by observed factors was calculated by summing the products of the difference in means between obese Latinos and NHWs and the coefficient for obese Latinos across factors included in the linear models. The reported decomposition “score” for each factor indicated the direction and magnitude of contribution of that factor to the observed part of the outcome difference. The larger the absolute factor score, the greater the contribution to the overall observed difference in weight-related counseling rates between obese Latinos and NHWs. Factors with positive scores were those that favored Latinos receiving more counseling versus NHWs, whereas factors with negative scores were those that favored NHWs receiving more counseling versus Latinos. The unobserved part was calculated by summing the products of the difference in coefficients between obese Latinos and NHWs and the mean for obese NHWs across the factors. This portion represented unexplained heterogeneity because differences in observed factors were controlled for by applying the means for obese NHWs.

All analyses were conducted in SAS version 9.2 (SAS Institute, Cary, NC) and accounted for the complex multistage survey design and sample weights of the NAMCS. Statistical significance was defined as P < 0.05 (two-sided).


Ethnic differences in the rate of weight-related counseling

Obese Latinos had higher rates of counseling for each weight-related topic (i.e., weight reduction, nutrition, and exercise) than did obese NHWs, with the differences for both nutrition and exercise reaching statistical significance (P = 0.01 and 0.02, respectively) (Table 1). The percentage of GMEs during which counseling was addressed for any of the three topics was low for both ethnic groups but was 15.5 higher for obese Latinos (51.3%) than obese NHWs (35.8%) (P= 0.03).

Table 1. Weight-related counseling by topic among obese Latinos and non-Hispanic whites during general medical examination visits
 Patient records, No.Estimated visits, No. in millionsWeight-related lifestyle counseling topic
Weight reduction (%)Nutrition (%)Exercise (%)Any of the three (%)
Non-Hispanic White5682119.328.618.635.8
P value  0.330.010.020.03

Ethnic differences in predisposing, enabling, and need factors

Significant ethnic differences were observed for a number of predisposing, enabling, and need factors (Table 2). Regarding predisposing factors, compared with obese NHWs, obese Latinos were comprised of higher percentages of females and young adults. Regarding enabling factors, obese Latinos were more likely than NHWs to live in the Western U.S. region and metropolitan areas, to see physicians who were primary care providers, and to attend practices that were federally qualified health centers or HMOs. On the other hand, obese Latinos were less likely than NHWs to have private health insurance, to see a provider who was owner of the practice, to have continuity of care, or to be seen by both a physician and nurse. For the need factor defined by the modified Charlson comorbidity index, there was no significant difference between ethnic groups in mean score.

Table 2. Comparison of predisposing, enabling, and need factors between obese Latinos and non-Hispanic whites
 Patient records, No.Estimated visits, No. in millionsGeneral medical examsP value
All (%)aLatino (%)aNon-Hispanic White (%)a
  1. EHR, electronic health record.
  2. aColumn percentages.
  3. bIncludes worker's compensation, self-pay, no charge, and payment unspecified.
  4. cIncludes follow-up as needed, no follow-up, telephone follow-up, and disposition unspecified.
  5. dIncludes family practice; family practice, geriatric medicine; sports medicine (family practice); general practice; gynecology; internal medicine; geriatric medicine (internal medicine); internal medicine/pediatrics; maternal and fetal medicine; obstetrics and gynecology; obstetrics; pediatrics; and sports medicine (pediatrics).
  6. eIncludes non-Federal government clinics, urgent care centers, family planning clinic, mental health centers, and faculty practices.
  7. fThe modified index expands the Charlson index to include hypertension, hyperlipidemia, and osteoporosis.
Predisposing factors
Sex     0.04
Age, years (mean, SD)6882552.7 ± 0.945.3 ± 2.054.0 ± 0.9<0.001
Age, years     0.003
Current smoker     0.74
Enabling factors
Health insurance     0.01
Private Insurance36715613865 
Continuity of care     0.04
Established patient, no. visits in past 12 months
New patient913121711 
Visit length (min)6882522.0 ± 0.920.5 ± 1.022.2 ± 0.90.28
Type of provider seen     0.002
Physician only47818718569 
Physician and nurse1987271529 
Other allied health professionals120202 
Visit disposition     0.94
Return appointment48317696669 
Referral to other physician642787 
Geographic region     <0.001
Metropolitan statistical area     0.001
Provider employment     0.003
Provider specialty     0.002
Primary cared53721839481 
Practice type     <0.001
Private solo practice1838343134 
Private group practice37515583562 
Federally qualified health center/HMO11115212 
Other office settinge1913132 
EHR system     0.73
Basic system1165191519 
Fully functional system5829139 
No EHR system51418727272 
Need factor
Modified Charlson comorbidity indexf688251.37 ± 0.081.12 ± 0.181.41 ± 0.090.17

Decomposition of ethnic difference in weight-related counseling

Of the total 15.5 difference between obese Latinos and NHWs in counseling rate during GMEs for any of the three weight-related topics, 9.33 (60.2%) was explained by ethnic differences in the observed predisposing, enabling, and need factors and the remaining 6.18 (39.8%) was unexplained (Table 3). Within the observed part, enabling factors had the greatest proportional impact, with provider specialty, metropolitan statistical area, practice type, and provider employment type contributing the most to higher counseling rates among Latinos. In contrast, geographic region, continuity of care, and health insurance were enabling factors that, along with the predisposing factors of sex and age, contributed the most in the opposite direction (Table 3).

Table 3. Regression-based decomposition of difference between obese Latinos and non-Hispanic whites in the rates of weight-related counseling during general medical examination visits
 Factor domainaDifference in weight-related counselingFactor decomposition scoreb
  1. EHR, electronic health record.
  2. aEach observed factor was categorized into one of three domains of factors (predisposing, enabling, and need) that according to the Anderson–Newman–Aday behavioral model might explain patient use of health services.
  3. bThe factor decomposition score indicates the direction and magnitude of contribution of a given factor to the observed difference in weight-related counseling by ethnicity: the larger the absolute factor score, the greater the contribution. Factors with positive scores are those that favored counseling being given to Latinos, whereas factors with negative scores are those that favored counseling being given to non-Hispanic whites.
  4. cThis is equal to the sum of the factor decomposition scores in the right column.
Total difference between Latinos and non-Hispanic whites 15.51 
Difference due to unobserved factors 6.18 
Difference due to observed factors 9.33c 
Factors that contributed to difference due to observed characteristics
Provider specialtyEnabling 6.96
Metropolitan statistical areaEnabling 6.90
Practice typeEnabling 2.54
Provider employment statusEnabling 2.51
Type of provider seenEnabling 1.17
EHR systemEnabling 1.12
Visit lengthEnabling −0.05
Current smokerPredisposing −0.12
Visit dispositionEnabling −0.26
Modified Charlson comorbidity indexNeed −0.99
AgePredisposing −1.18
SexPredisposing −1.37
Health InsuranceEnabling −1.44
Continuity of careEnabling −2.05
Geographic regionEnabling −4.41


Obesity disproportionately affects racial and ethnic minorities, Latinos in particular ([1, 2]). If unaddressed, it has the potential to sizably worsen existing health disparities. Differential quality of health care in clinical encounters is one recognized source of racial and ethnic disparities in health outcomes ([20]). In this study, we found that the rate of counseling for weight reduction, nutrition, and/or exercise during GMEs was significantly higher among obese Latinos than obese NHWs. On decomposition analysis, practice-level enabling factors (e.g., provider specialty and employment status and practice type) favored Latino patients receiving weight-related counseling compared with NHWs, whereas patient-level enabling factors (e.g., continuity of care and health insurance) along with predisposing factors (e.g., sex and age) favored NHWs. The need factor defined by the modified Charlson comorbidity score contributed proportionally little.

Similar to the findings reported here, several earlier studies showed that racial and ethnic minorities were equally or more likely to receive physician counseling or referral for lifestyle modification than NHWs ([8, 21]). Given the well-documented worse health outcomes among obese Latinos compared with their NHW counterparts ([1, 2]), it is encouraging that we observed higher rates of counseling among this population. These findings may indicate that health care providers are now more aware of the disproportionate burden of obesity-related morbidity and mortality born by Latino patients and so are more activated to provide weight-related counseling to these patients. At the same time, the persistence of ethnic disparities in obesity-related outcomes suggests that more must be done to understand and overcome the disparities. Latino patients may confront greater barriers to self-management of weight than do their NHW counterparts such that, even in the context of higher rates of obesity-related counseling, they have more difficulty in achieving a healthy weight. Communications issues are just the first of many hurdles that may have to be overcome ([22, 23]). Another possible explanation is that the mere provision of counseling (“quantity” of care) is an inadequate marker of effective care having been provided (“quality” of care). For example, it may be that weight-related counseling needs to be culturally tailored to Latinos to have enough of a health impact to reduce existing disparities ([24]). Studies have shown that culturally undifferentiated interventions that originate in mainstream populations may be inefficient or even unintentionally increase health disparities in minority populations ([24, 25]). Finally, there is evidence to suggest that patients experience lifestyle counseling as beneficial when they feel they have a good patient-provider relationship ([26]), which may be more likely in the context of continuity of care. In this study, Latino patients were more likely to receive obesity counseling, but less likely to have continuity of care.

The Blinder-Oaxaca decomposition method ([27]) has previously been used to study racial/ethnic disparities in health insurance coverage and health care access and utilization ([28]). To our knowledge, our study is the first to apply this method to examine ethnic differences in weight-related counseling in the office-based setting. Instead of examining the association between race/ethnicity and the likelihood of receiving obesity counseling using logistic regression as has been done before ([8, 9, 21]), we focused on two questions: ([1]) How much of the total difference in weight-related counseling rates between obese Latinos and NHWs could be explained by observed factors at the patient, visit, and practice levels and ([2]) How each factor's contribution ranks relative to the contributions of other factors and in what direction. Notably, our decomposition analyses identified enabling factors that are contextual (i.e., related to the setting of health care utilization—provider specialty, metropolitan area, and practice type) as being equally influential or even more influential than enabling (e.g., continuity of care and insurance) or predisposing (e.g., sex and age) that are patient-specific. This highlights the importance of including contextual factors in applications of the Anderson-Newman-Aday model ([15]), and the usefulness of the decomposition method in identifying the relative contributions of each factor to observed differences.

In addition, our findings can guide the development of Latino-focused weight management interventions because they provide insight to facilitators and barriers to obese Latinos receiving weight-related counseling. For instance, it is not surprising that continuity of care was a facilitator of higher rates of obesity counseling among NHWs and continuity of care should be promoted as a component of optimized chronic disease management for Latino patients as well, including those with obesity. In addition, the anticipated increased enrollment of Latino patients into private health insurance plans (or health insurance exchanges) as a result of the Patient Protection and Affordable Care Act of 2010 ([29]) should be actively promoted because it might act as an enabling factor for increased rates of obesity counseling, as was seen for NHWs in this study. Other enabling factors that are already present in higher proportion among Latino patients vs. NHWs but should continue to be encouraged on a population level include seeing physicians who are primary care providers and attending practices that are federally qualified health centers or HMOs.

Finally, we found that 40% of the total Latino-NHW difference in counseling rate was not explained by ethnic differences on an array of predisposing, enabling, and need factors characterizing the patient, visit, and practice. We can only speculate, based on the literature, regarding what factors not collected in the NAMCS may differentially influence clinical communication and decision making related to weight-related counseling among obese Latinos and NHWs. These might include factors related to the health system and community at large (e.g., lack of places to refer patients and built environments in which patients live and work), physician (e.g., beliefs about the benefit of lifestyle modification), and patient (e.g., education, income, prior weight loss experiences, and family/social network characteristics). The extent of contribution from these cannot be known unless investigated, but some are likely difficult to measure reliably or are not relevant to all but a very small number of outcomes and, hence, are often unavailable in large-scale surveys such as the NAMCS. Nonetheless, if collected and reported in the future, new factors can be added to subsequent decomposition analyses to determine their relative contributions to observed differences in weight-related counseling.

Our finding of overall low rates of weight-related counseling during GMEs by obese Latinos and NHWs (51 and 36%, respectively) exposes a gap in obesity care that is conspicuously more prominent than any differences in care based on patient ethnicity. Evidence-based practice guidelines consistently recommend routine lifestyle counseling or referral for weight management in clinical practice ([4]). Our reported weight-related counseling rates during 2009 are similar to those from a recently published analysis of 2005 NAMCS data ([8]), indicating a continuing gap. That previous study found comparably low rates of weight-related counseling across many patient subpopulations ([8]). What makes our findings even more striking is that we focused on GMEs, which are intended for preventive care (and thus less likely to be influenced by patients' needs for symptom-driven care), and on patients with clinically indicated needs for obesity care, namely, those with high measured BMIs or a current obesity diagnosis. Given the knowledge of obesity's chronicity and lasting adverse health impact ([30, 31]), evidence supporting the efficacy of lifestyle modification ([32, 33]), and potential effectiveness of physician advice to promote patient engagement in lifestyle modification ([6]), it is reasonable to expect that obese patients routinely receive weight-related counseling or referral at GMEs, whether or not they have previously been counseled. However, our results suggest that one in two obese Latinos and two in three obese NHWs are not counseled on nutrition, exercise, or weight management during their GME visits.

The current primary care delivery model lacks the orientation, organization, and reimbursement structure needed to effectively manage obesity and prevent obesity-related chronic diseases ([34]). Although primary care providers are the designated “gatekeepers” for detection, prevention, and treatment of obesity, they often do not have the training, skills, confidence, and time required to fulfill this role ([35, 36]). These system- and provider-level barriers need to be addressed in order to reduce the substantial quality gap in obesity care exposed in our work here and the work of others.

Several limitations in our study are worth noting. First, the NAMCS' cross-sectional, visit-based design precludes patient-level analyses of lifestyle counseling services received from the same or different providers over time. Yet, population-based national surveys have reported similarly low rates of physician advice about adopting preventive behaviors among individuals at high risk for largely lifestyle-attributable chronic diseases such as type 2 diabetes ([37]). Second, the per-patient visit basis may be associated with a reduced likelihood, at any given visit, of obesity treatment being provided or reported for patients who see physicians more often, because care for their health concerns can be spread across a greater number of visits within a given time frame. Consistent with others reporting that NHWs tend to have more primary care visits than Latinos ([38]), we found that obese NHWs had a higher continuity of care than obese Latinos. This difference, however, favored NHWs as opposed to Latinos receiving counseling in our decomposition analyses. Third, because the NAMCS relies on health care providers and their office staff for completion of the standardized patient encounter forms, it is vulnerable to reporting biases (e.g., underreporting or overreporting of lifestyle counseling) and errors (e.g., misclassification of patient race/ethnicity). Fourth, data available from the NAMCS allowed for only sparse characterizations of predisposing and need factors (three and one factor apiece, respectively) according to the Anderson-Newman-Aday behavioral model. Contributions to the unobserved part of the outcome difference may well belong to these predisposing and need domains, or others that fall outside of the Anderson-Newman-Aday framework. Using a conceptual framework to guide the development of new questions in the future would allow for the data to more uniformly reflect the domains that might be pertinent to health care use. Fifth, the NAMCS does not provide the data needed to evaluate the quality of counseling provided, and in the case of referral services, whether the patient actually received them.

In conclusion, our findings suggest that obese Latinos are more likely to receive weight-related lifestyle counseling during GMEs than obese NHWs. This can be partly explained by a number of measured factors. Further research is warranted to elucidate what and how other factors influence providers' decisions to provide weight-related lifestyle counseling for patients from different racial and ethnic groups who have a clinically indicated need for such service—a need that is likely to grow as the obesity epidemic continues ([39]). Our results also reveal that even during prevention-oriented GMEs, considerable missed opportunities are evident in the management of adult obesity across the examined ethnic groups. The U.S. health care system is undergoing significant changes; the emergence of Patient-Centered Medical Homes, Accountable Care Organizations, and other new health care and payment models has the potential to improve care coordination and promote patient-centered care ([40]). The Patient Protection and Affordable Care Act contains provisions that mandate coverage of recommended preventive services ([29]). The Centers for Medicare and Medicaid Services has recently approved coverage of intensive behavioral therapy for obesity in primary care ([5]). These expanded benefits are expected to help address recognized quality gaps in obesity.