A model of weight‐based stigma in health care and utilization outcomes: Evidence from the learning health systems network

Abstract Objective Obesity is stigmatized and people with obesity report experiencing stigmatizing situations when seeking health care. The implications of these experiences are not well understood. This study tests an indirect effects model of negative care experiences as an intermediate variable between obesity and care avoidance/utilization and switching primary care doctors. Methods A survey was completed by 2380 primary care patients in the Learning Health Systems Network (LHSNet) Clinical Data Research Network with a BMI >25 kg/m2. Measures included scales assessing stigmatizing situations, perceived patient‐centered communication, perceived respect, having delayed needed care, and having looked for a new primary doctor in the past 12 months. Sequential and serial indirect effects of care experiences and respect in the association between BMI and care utilization outcomes was modeled. Results The hypothesized model was supported by findings. The associations between BMI and delaying needed care (OR = 1.06, p < 0.001) and attempting to switch primary doctors (OR = 1.02, p = 0.04) was mediated by both stigmatizing situations experienced in a health care context and lower patient‐centered communication. Lower perceived respect mediated the association between care experiences and utilization outcomes. Conclusions People with higher BMIs may avoid care or switch doctors as a result of stigmatizing experiences and poor communication with doctors. These outcomes may contribute to morbidity in people with obesity if they delay or avoid care for health concerns when symptoms first present.


| INTRODUCTION
There is ample evidence that obesity is a stigmatized trait and that health care providers as well as the general public are biased against people with obesity. [1][2][3][4][5][6] Weight stigma consists of negative stereotypes and attitudes about people with overweight or obesity, and discrimination that can result. Commonly held stereotypes among health care providers about people with obesity include that they lack willpower and are non-adherent to medical recommendations, including advice to lose weight. [7][8][9][10] These biases and stereotypes can negatively affect both the interpersonal and technical quality of health care. [11][12][13] People with obesity receive less patient-centered communication, [12][13][14] with health care providers initiating less rapport, 15 making fewer attempts to build relationships, and providing less patient education. 16 Less patient-centered communication is, in turn, associated with lower patient satisfaction 17 and adherence, 18 worse care outcomes, [19][20][21] and less adoption of healthy lifestyle behaviors. 22,23 Technical deficits in care observed during interactions between health care providers and people with obesity include reports of health care providers focusing on weight loss while paying inadequate attention to other complaints and issues 24 and recommending weight loss in place of medications or other therapies for symptoms. 11 Patients with obesity report feeling judged and treated with less respect 13,25,26 than other patients, and subsequently trust their health care providers less. 26 Thus, health care encounters may be experienced as threatening to patients with obesity, possibly leading to avoidance of future health care encounters and delay of needed care. [27][28][29] Because of this, people with obesity may seek care later in the progression of a disease, compared to people with normal BMIs, and thus may present with more advanced and possibly harder to treat conditions. 30 This is one factor that may contribute to the excess morbidity and mortality associated with overweight and obesity. 2,31,32 Studies have found people with obesity are more likely to delay screening for cancer 33 and other health care visits. 34 Additionally, people with obesity may be more likely than others to "doctor shop" or seek a regular health care provider with whom they feel comfortable and well-treated. 35,36 Thus, people with obesity may be less likely to have a regular doctor, a factor associated with better care continuity and health outcomes, 37,38 and may be more likely to receive care at drop-in or urgent care centers, which are less likely to address lifestyle or behavior issues or chronic disease risks. 39 However, the associations among obesity, care experiences, and care utilization have been primarily examined separately, thus, this study tests a model of care experiences as mediating the relationship between obesity and care avoidance/utilization. In the conceptual model ( Figure 1), higher BMI is associated with care-seeking behaviors via two primary pathways. First, (a) higher BMI is associated with more frequent stigmatizing experiences during a health care encounter, such as a health care provider making assumptions about patient behavior based on obesity stereotypes or making embarrassing or cruel remarks, or blaming unrelated symptoms or problems on weight. These experiences may trigger identity threat, or the response to feeling that one's social identity is at risk of being devalued or viewed in terms of a group stereotype. 29 Social identities are the groups and categories that make up the way people see themselves and believe they are seen by others. 40 Identity threat causes a stress response that can reduce a person's ability to perform cognitively taxing activities and lead to future avoidance of similar situations. 41 One's body size is salient in a medical encounter, and people with overweight or obesity who are aware of stereotypes and prejudice against people with larger body sizes may be vigilant for poor treatment or discrimination related to their size. 2 In response to identity threat, the patient may (b) avoid similar situations in the future, and (c) feel disrespected or humiliated, souring the therapeutic relationship with their health care provider and leading them to (h) avoid that provider and/or seek care elsewhere.  43 The detailed description of the study participants has been described elsewhere. 42 Using an algorithm to identify obesity and overweight resulted in a cohort of 2,072,572 adult patients with BMI >25 kg/m 2 across sites. 42

| Survey measures
The survey was developed to measure aspects of the health care experience and health-related behaviors of patients with overweight and obesity. 42,44 The survey was designed to be low-burden and took, on average, 11 min to complete. Measured constructs are described below.
Stigmatizing experiences were measured using a modified Medical Subscale of the Stigmatizing Situations Index. 24 This subscale measures how frequently the respondent experiences each of four stigmatizing situations while seeking health care, including, "Having a doctor make cruel remarks, ridicule you, or call you names," and "A doctor blaming unrelated physical problems on your weight." Participants responded on a 4-point scale of "Never", "Once in your life", "More than once in your life" and "Multiple times." The four items loaded to one underlying factor so an average of the four items was computed (Cronbach's alpha = 0.75).
Patient-centered communication (PCC) was measured using the 6-item scale adapted from the Consumer Assessment of Health Plans survey for use in the National Cancer Institute HINTS survey. 45 Items map onto six dimensions of PCC, and ask how often in the past 12 months a healthcare professional did the following: "Give you the chance to ask all the health-related questions you had"; "Give the attention you needed to your feelings and emotions?"; "Involve you in decisions about your healthcare as much as you wanted?"; "Make sure you understood the things you needed to do to take care of your health?"; "Help you deal with feelings of uncertainty about your health or healthcare?"; and "In the past 12 months, how often did you feel like you could rely on your doctors, nurses, or other health care professionals to take care of your health care needs?" Response options were between "Never" and "Always" on a 5-point scale.
These items loaded to one underlying factor with Cronbach's alpha of 0.93.
Perceived respect was measured using three items developed for this study, "in the last 12 months, how often did you feel your primary health care provider treated you..." "…with respect" and "…as an equal," and "in the last 12 months, did you ever feel that your primary health care provider judged you because of your weight." The first two items were measured on a 5-point scale from "never" to "always" -141 and the third item was measured as "yes" or "no". The 5-point scale items were rescaled to be bounded by 0 and 1 so that the items could be combined. The three items loaded to one factor and a mean score of the items was created (Cronbach's alpha = 0.70).
Delayed health care utilization was measured by a six-item yes/ no question scale. 34 Respondents were asked if in the last 12 months they delayed or avoided getting health care because they…"gained weight", "were told to lose weight", "though you would be weighed", "thought you would discuss weight", "thought you would be asked to undress", or "thought you could get rid of a problem by losing weight". These items were collapsed into one variable representing a "yes" response to any item versus a "no" response to all items.
Doctor shopping was measured using a single yes/no item: "In the last 12 months, have you switched doctors or tried to find a new primary care doctor?" BMI was measured by asking for the respondent's current height and weight without shoes on. Though BMI was available and used to identify the sample, the survey was sent anonymously, so current BMI was collected via the survey.
Patient demographic information was measured using standard survey items to assess age, race, ethnicity, education, and gender. Race was categorized into Black, White, and other race or ethnicity due to a small number of respondents who did not identify as White or Black.
Education was categorized as high school graduate or less, some college, or a 4-year college degree or more.
Respondents were asked what their current weight and height without shoes was, and current BMI was calculated for each respondent. Finally, respondents were asked to indicate which of the following medical conditions they had ever been diagnosed with: arthritis, diabetes mellitus, high cholesterol, high blood pressure, heart disease, obstructive sleep apnea, depression, asthma, or other health problems. A variable represented a count of the number of these conditions indicated was created.

| Analysis
In order to characterize the sample, summary statistics were calculated for all variables with findings presented using mean (SD) for continuous variables and frequency counts and percentages for nominal variables.
Five multivariate linear or logistic regression analyses were used to assess the associations between BMI and the patient experience and patient behavior variables. Covariates included in these analyses included BMI, age, gender, race, education, and the number of comorbidities. Cases with missing data for variables included in a model were excluded for that model. The PROCESS macro for SPSS 46 was used to fit the data to the conceptual path model and test the sequential and parallel mediation (PROCESS model 80) of the relationship between BMI and delayed health care utilization and doctor shopping by perceived stigma, PC communication, and perceived respect. Throughout this paper, the term "mediation" is used to describe an indirect effect of an independent variable on a dependent variable, and is not used to imply causation. Associations where delayed care or doctor shopping was the dependent variable are presented on the log odds scale. Pathway point estimates are the product of each association along the path, and path confidence intervals were calculated by bootstrapping based on selecting 5000 random samples of the data. All data are cross-sectional, and thus this analysis is intended to test fit to a conceptual model, and not to suggest that these data demonstrate causality.

| RESULTS
Sample characteristics are shown in Table 1  respectively. All of the care quality and care utilization variables are associated with BMI in the hypothesized directions. Table 3

Variable and scale range Missing % (n) Mean (SD) or % (n) among non-missing Bivariate association with BMI (B, p-value)
Stigmatizing situations in health care (0-3) 5.  These findings are consistent with evidence that people with higher BMIs are more likely to feel judged and discriminated against by their health care providers, 13,25,26,47 and evidence that physicians and other health care providers engage in less rapport building, offer less education, and spend less time overall with patients with higher BMIs. [12][13][14][15][16]48 The findings also support research that has found that health care providers report having less respect for their patients with obesity. 9,26 The findings extend this work to show that these experi-

CONFLICT OF INTEREST
The authors declare no conflict of interest.