Treatment resistant obesity among midlife women receiving care from a specialized weight management clinic

Abstract

Background

Obesity is a complex chronic disease conveying significant health risk. Women with obesity have greater risk for certain obesity associated conditions and midlife is when women experience an increase in their overall obesity prevalence and severity. Some midlife women experience treatment resistant obesity (TRO), a phenomenon not previously explored. The purpose of this research was to determine the prevalence of TRO, or failure to achieve ≥5% total body weight loss after one year of medically specialized obesity treatment, among midlife women and to identify factors associated with the phenomenon.

Methods

A retrospective cohort study was completed using information from the electronic health records of 203 midlife women with obesity who received one year of medically specialized obesity care from providers certified in obesity medicine. Quantitative analysis determined the prevalence of TRO among midlife women and found associations between TRO and certain biopsychosocial factors.

Results

More than 27% of participants in this study demonstrated TRO. The use of FDA approved injectable medications and the dietary modification of lowering carbohydrates demonstrated associations with TRO. Participants were more likely to demonstrate TRO if they did not use FDA approved injectable medications (OR = 3.49, p = 0.0002) or follow a diet lower in carbohydrates (OR = 2.18, p = 0.035).

Conclusions

More than a quarter of midlife women with obesity who receive medically specialized obesity care for one year may fail to achieve ≥5% total body weight loss and demonstrate TRO. The combination of lowering dietary carbohydrates and taking FDA approved injectable medications for weight loss in midlife women may reduce their risk of TRO. Incorporation of these two strategies alone does not eliminate the occurrence of TRO among midlife women and therefore additional research into this phenomenon is necessary to determine more efficacious treatment strategies.

Graphical abstract

1. Introduction

Obesity is a complex multifactorial chronic disease conveying significant health risk. Although overall rates of adult obesity in the United States are similar between the sexes, midlife women, or those between the ages of 40–60 years old, have an increase in their prevalence of obesity and have higher rates of severe obesity compared to men [1,2]. These increases in obesity prevalence and severity suggest that for women midlife may present unique obesity-promoting factors. Women with obesity are at greater risk for the development of some obesity related conditions including heart disease, ischemic stroke, and sex-specific cancers [3], [4]. Research has also found a direct relationship between the duration of a woman's obesity and the risk for the development of type 2 diabetes, metabolic syndrome, and certain cancers [[5], [6], [7]]. This suggests that timely efficacious treatment of obesity for midlife women is necessary for the prevention of additional disease.

Clinically meaningful weight loss, consistently defined as 5–10% total body weight loss, is the point at which overall health risk is reduced and is widely accepted as a reasonable initial treatment goal for those with obesity [8,9]. Some people experience treatment resistant obesity (TRO), an unexplored phenomenon that presumably increases risk for poorer health outcomes. Treatment resistant obesity (TRO), a new concept being introduced into the literature, is defined as failure to achieve clinically meaningful weight loss or ≥ 5% total body weight after having received medically specialized obesity management for one year. Medically specialized obesity management refers to that which is provided by those who have received additional education and training to become obesity medicine certified providers. Understanding the need for a comprehensive treatment approach that considers the multiple factors involved in the chronic disease of obesity, these certified providers are trained to offer evidence-based individualized care to patients integrating recommendations for lifestyle modification, behavior change, and obesity medications. Medically specialized obesity management is on the rise as evidenced by growing numbers of physicians obtaining obesity medicine certification annually through the American Board of Obesity Medicine (ABOM). In 2024, over 1800 physicians were certified by the ABOM which was the highest number of certifications issued in any one year bringing the total number of obesity medicine certified providers to 9800 in the United States and Canada [10]. Advanced practice providers, nurse practitioners and physician assistants, can also specialize in obesity medicine and receive a certificate of advanced education through the Obesity Medicine Association by completing the same number of continuing education hours required of physicians pursuing ABOM certification. The number of advanced practice providers who have obtained this certificate has increased more than tenfold over the past five years with approximately 1200 total providers as of October 2025 [A. Madera, personal communication, October 1, 2025].

The concept of TRO is particularly relevant among midlife women, given the increased prevalence, severity, and complications of obesity in this population [[1], [2], [5], [6], [7], [8], [9]]. Although causes of obesity are complex and multifactorial [11,12], much of the prior research surrounding weight loss interventions for midlife women with obesity has focused on lifestyle modifications to diet and exercise. Various dietary interventions for midlife women with obesity have been evaluated including higher amounts of dietary protein [13], time restricted feeding patterns [14], and eating behaviors such as timing of snacking [15]. Research focused on physical activity interventions for midlife women with obesity has included evaluation of the total time per week spent exercising [16], high intensity interval training exercise [17], and different exercise settings such as home based versus group exercise [18]. Other research has evaluated the effects of combining both diet and exercise for weight loss [19,20], but there is very limited research specific to midlife women with obesity and emerging obesity medications. However, one recent study specific to midlife women did find that hormone replacement therapy in conjunction with GLP1 receptor agonist medications demonstrated superior weight loss to those taking GLP1 medications alone [21].

Limited research exists surrounding the weight loss outcomes of those who receive care from obesity medicine certified providers and none of this research has evaluated the effects of this specialized care specifically for midlife women with obesity. Two studies by Srivastava evaluated the weight loss outcomes of participants who received six months of obesity care from certified providers in two different multidisciplinary obesity medicine weight loss centers, one in Florida the other in Tennessee. While the samples in both studies were predominantly women, 70% and 77% respectively, they also included participants not in midlife and men. Findings from the first study found that after three months, 60% of the sample did not achieve ≥5% total body weight loss [22] while the second study found that after six months 45% of the sample did not achieve ≥3% total body weight loss [23]. Another research study evaluated the weight loss outcomes among those participating in shared medical appointments led by a dietitian and an obesity medicine certified provider and found that 48% of the sample did not achieve ≥5% total body weight loss after one year [24]. Other studies have evaluated weight loss outcomes of obesity medicine certified care with similar percentages, 47% and 68%, of the samples not achieving ≥5% total body weight loss after six months to a year [25,26]. It is important to note that the authors of these studies reported their findings as the proportion of their samples that were successful in meeting study weight loss thresholds. The numbers presented above were deduced from their reported findings. Obesity research often focuses on factors or interventions leading to successful weight loss while failing to explore the potential reasons for failure. Without research specifically evaluating those who are unsuccessful with weight loss, assumptions may be made that they are simply doing the opposite of those who experience successful weight loss which may be incorrect given the multifactorial nature of the disease.

This lack of focus on the population of midlife women with obesity who receive medically specialized obesity management and remain unsuccessful in achieving clinically meaningful weight loss highlights a gap in the research. Given that women experience increased risk for the development of serious obesity related conditions and that some are unable to achieve clinically meaningful weight loss of ≥5%, the point at which health risk is reduced [8], increased understanding of TRO is necessary to improve health outcomes. The purpose of this research was to generate knowledge about midlife women who receive medically specialized obesity management for one year and experience TRO. This research study sought to answer the following questions:

• •What is the prevalence of TRO among midlife women who receive medically specialized obesity management over the course of one year?
• •What biopsychosocial factors are associated with the phenomenon of TRO among midlife women?

It was hypothesized that a proportion of midlife women would demonstrate TRO after receiving one year of medically specialized obesity management. It was also hypothesized that biopsychosocial factors that have been found to be associated with the development or persistence of obesity would also be associated with TRO among midlife women.

2. Methods

This pilot study utilized a retrospective observational cohort design. This study was approved by the Institutional Review Board of the authors’ affiliated academic institution and the healthcare organization that owned the specialized clinic from which the data was obtained. This research utilized de-identified patient data and fell under the category of exempt research and therefore informed consent was not required.

2.1. Research setting & sample

Data was collected from the electronic health records of a single metabolic weight management clinic in the mid-Atlantic United States staffed by two obesity medicine certified physicians and one nurse practitioner. All newly established female patients between the ages of 40–60, with a diagnosis of obesity determined by a body mass index (BMI) of ≥30 kg/m², who received care for at least 12 months, and had a minimum of four encounters with an obesity medicine provider were included in the sample.

2.2. Data collection

Data was collected retrospectively from the electronic health records of the metabolic weight management clinic over a three-year period beginning with the clinic inception on October 1, 2021, and ending on September 30, 2024. Necessary clinical data for the research study was collected with the aid of a database analyst employed by the healthcare organization that owned and operated the clinic. The analyst was able to provide de-identified clinical data collected from the electronic health record utilized by the metabolic weight management clinic.

Data for several biopsychosocial variables were collected at different time points throughout the study period (see Supplement). Biopsychosocial factors refer to the biologic, psychologic, and social states that can influence health and wellbeing. The biopsychosocial variables evaluated as part of this research have known associations with the development or persistence of obesity. Data for the variables of age, race and ethnicity, insurance status, body mass index, and body weight were collected from the initial care establishing visit for each included participant. Body weight was collected again from the final visit within the 12-month timeframe for each individual participant so that the change in body weight as a percentage over the one-year study period could be calculated. The total number of billable office visits and all medications prescribed for help with weight loss were collected over the course of a one-year study period for each participant. Other clinical data including self-reported diet modifications, physical activity, sleep, and stress coping strategies were collected from the history of present illness section of the electronic health records. Data for these variables were collected from the final visit within the 12-month study period for each participant.

2.3. Statistical & analysis methods

This pilot study utilized a retrospective observational cohort design to generate knowledge about midlife women who receive medically specialized obesity management for one year and experience TRO. To determine the prevalence of TRO within the sample, the percentage of total body weight lost was calculated by comparing each participant's body weight at the first establishing visit to their weight at their final visit within the 12-month study period. Individuals who achieved <5% total body weight loss were classified as TRO and those who achieved ≥5% total body weight loss were deemed non-TRO.

Unstructured textualized health record data was categorized prior to analysis (see Supplement). Diet modification, physical activity, sleep, and stress coping typically consisted of a few words or short phrases documented by the obesity medicine provider during the participant's final encounter within the 12-month study period. Criteria were determined for each of the independent variables so that they could be categorized into either a “yes” or “no” group based on responses that were consistent with a goal of optimizing weight loss. To ensure the validity and trustworthiness of the categorization process, a second reviewer using set criteria independently categorized 10% of the total data for each of the four variables. When compared to the original reviewer categorizations, total consensus was achieved.

The available retrospective data collected for reported dietary modifications demonstrated several repeated modifications providing and opportunity for more in-depth analysis. The following categories of reported dietary modifications were found within the research sample: lower carbohydrate, calorie reduction, increased fruits and vegetables, decreased sugar, less snacking, intermittent fasting, meal prepping, increased protein, and use of commercial weight loss meal plans. Participant data was dichotomized into a “yes” or “no” response based on these reported dietary modification categories and was analyzed individually for suggestive associations with the outcome of TRO. The total number of concurrent dietary modifications made by each participant was also analyzed for associations with TRO. Likewise, the data pertaining to prescribed medications was amenable to categorization based on medication category providing an opportunity for more in-depth analysis. All medications prescribed for help with obesity management were collected for participants over the 12-month study period. Three relevant medication categories were determined to be applicable to the clinical practice of obesity medicine providers and included injectable medications approved by the Federal Drug Administration (FDA) for either an indication of obesity or type 2 diabetes, oral medications approved by the FDA for the indication of obesity, and medications used off label for weight management. Medication data were dichotomized into “yes” or “no” responses based on patterns of prescribing that suggested medications were regularly used by a participant for help with obesity management. Each individual medication category as well as the total number of unique medication categories that participants were concurrently and regularly prescribed during the study period were analyzed for suggestive associations with TRO. Data for all independent variables were collected for each of the 203 participants except for sleep sufficiency for which 12% of the sample had missing data and for stress coping for which 14% had missing data.

Descriptive statistics were calculated for all variables and compared between the TRO and non-TRO groups using t-tests for continuous measures and either Pearson chi-square or Fisher's exact tests for categorical measures. Logistic regression analysis was performed to examine associations between biopsychosocial variables and the condition of TRO. The Hosmer Lemeshow [27] method of model building was utilized for this multivariable logistic regression analysis. This process started with bivariate analysis using a more generous p-value of ≤0.25 to evaluate for any suggestive associations between each independent variable and the outcome variable of TRO. While building the multivariable logistic regression model, all independent variables with a p-value of (≤0.25), suggesting an association to the outcome, were placed in the final model. Using a stepwise process, independent variables were removed from the multivariable model if they had Wald-type marginal p-values (>0.05) starting with the independent variable having the largest p-value. This process of removing independent variables with p-values >0.05 continued until only variables with p-values <0.05, indicating statistical significance, remained. Odds ratios were used to report the association between the independent variable and the outcome of TRO, with positive values indicating that the outcome, TRO, was more likely. Finally, when evaluating the whole model, the R-Squared value was used to determine how much variability within the outcome the independent variables in the given model explained. Quantitative data analysis was completed utilizing JMP® Pro (version 17) statistical software.

3. Results

3.1. Sample characteristics

From a total patient panel size of more than 1700 unique patients within the identified metabolic weight loss clinic during the study period, a total of 203 participants met inclusion criteria for this research study. Characteristics of the research sample are presented in Table 1.

Table 1.

Characteristics of the sample.

Total Sample (N = 203)	Mean (SD)	CI (95%)	Range
Age	50.90 (5.99)	(50.07, 51.73)	40–60
BMI	43.05 (8.89)	(41.82, 44.29)	30–78
Visits	6.81 (2.23)	(6.50, 7.12)	4–13

Total Sample (N = 203)	n (%)
Race
White	177 (87.1)
Black	18 (8.8)
Asian American/Pacific Islander	2 (0.0)
American Indian/Alaskan Native	1 (0.0)
Other/Not Identified	5 (2.4)
Ethnicity
Non- Hispanic	197 (97.0)
Hispanic	3 (1.4)
Not Identified	3 (1.4)
Insurance
Commercial	145 (71.4)
Government funded	49 (24.1)
Uninsured	9 (4.4)

3.2. Prevalence of TRO

Within the sample (N = 203) the proportion of subjects who demonstrated TRO was nearly 28% (n = 56, 27.5%) with the remaining 72% of the sample (n = 147, 72.4%) not demonstrating the condition and designated as non-TRO. Characteristics of each group, TRO and non-TRO are compared in Table 2.

Table 2.

Comparison of characteristics of TRO and Non-TRO groups.

	TRO (n = 56)		Non-TRO (n = 147)
Total Sample (N = 203)	Mean (SD)	Range	Mean (SD)	Range
Age	50.53 (5.50)	40–60	51.04 (6.18)	40–60
BMI	42.55 (10.47)	30–78	43.25 (8.24)	30–67
Visits	6.44 (2.26)	4–13	6.95 (2.22)	4–13
	n (%)		n (%)
Race
White	48 (85.7)		129 (87.7)
Black	5 (8.9)		13 (8.8)
Asian American/Pacific Islander	0		2 (1.3)
American Indian/Alaskan Native	0		1 (0)
Other/Not Identified	3 (5.3)		2 (1.3)
Ethnicity
Non- Hispanic	54 (96.4)		143 (97.2)
Hispanic	1 (1.7)		2 (1.3)
Not Identified	1 (1.7)		2 (1.3)
Insurance
Commercial	41 (73.2)		104 (70.7)
Government funded	12 (21.4)		37 (25.1)
Uninsured	3 (5.3)		6 (4.0)

3.3. Biopsychosocial variables and their associations with TRO

In the model-building bivariate analyses, number of office visits had significantly different between group means (p = 0.145) and a suggestive association was also found between sleep sufficiency and TRO (p = 0.027) (Table 3). No statistically significant differences were found between TRO and non-TRO for age (p = 0.592), baseline BMI (p = 0.618), insurance status (p = 0.768), race and ethnicity (p = 0.705), physical activity (p = 0.388), or stress coping (p = 0.565). Of the nine categories of reported dietary modifications three demonstrated suggestive relationships to TRO (Table 4), lowering carbohydrates (p = 0.046), calorie reduction (p = 0.149), and less snacking (p = 0.185). The number of concurrent dietary modifications made by each participant were also calculated and nearly half of the sample, n = 93 (45%), reported making two or more dietary modifications, however the number of dietary modifications reported by the participants had no association with the outcome of TRO (p = 0.712).

Table 3.

Bivariate analysis of biopsychosocial independent variables and outcome of TRO.

	TRO (n = 56)	Non-TRO (n = 147)
Total Sample (N = 203)	Mean (SD)	Mean (SD)	T-Ratio (DF)	Prob > [t]
Age	50.53 (5.50)	51.04 (6.18)	−0.535 (201)	0.592
BMI	42.55 (10.47)	43.25 (8.24)	−0.498 (201)	0.618
Visits	6.44 (2.26)	6.95 (2.22)	−1.462 (201)	∗0.145
	n (%)	n (%)	ChiSquare (DF)	Pearson/Fisher's Exact (p-value)
Race & Ethnicity
Non-Hispanic White	47 (87)	129 (89)	0.143 (1)	0.705
Insurance				0.768
Commercial	41 (73.2)	104 (70.7)
Government Funded	12 (21.4)	37 (25.1)
Uninsured	3 (5.3)	6 (4.0)
Diet Modification “yes”	54 (96.4)	142 (96.6)		1.000
Regular Physical Activity “yes”	29 (51.7)	86 (58.5)	0.745 (1)	0.388
Pharmacotherapies “yes”	49 (87.5)	133 (90.4)	0.387 (1)	0.533
	TRO (n = 50) n (%)	Non-TRO (n = 130) n (%)
aSleep Sufficiency “yes” (N = 180)	25 (50.0)	88 (67.6)	4.837 (1)	∗0.027
	TRO (n = 47) n (%)	Non-TRO (n = 128) n (%)
aStress Coping Strategies “yes” (N = 175)	34 (72.3)	98 (76.5)	0.331 (1)	0.565

∗p-value (≤0.25) suggestive of an association with outcome of TRO.

^aMissing data for some participants for these biopsychosocial independent variables.

Table 4.

Contingency table analysis of diet modification categories and TRO.

Diet Modification	TRO n (%)	Non-TRO n (%)	ChiSquare (DF)	Pearson (p-value)	Fisher's Exact Test (p-value)
Calorie Reduction	33 (58.9)	70 (47.6)	2.07 (1)	∗0.149
Lower Carbohydrate	38 (67.8)	119 (80.9)	3.96 (1)	∗0.046
Increased Protein	1 (1.7)	9 (6.1)			0.290
Commercial Meal Plan	0 (0.0)	2 (1.3)			1.00
Decreased Sugar	1 (1.7)	5 (3.4)			1.00
Increased Fruits & Vegetables	2 (3.5)	9 (6.1)			0.730
Intermittent Fasting	3 (5.3)	4 (2.7)			0.397
Less Snacking	2 (3.5)	1 (0.6)			∗0.185
Meal Prepping	0 (0)	1 (0.68)			1.00
Number of dietary modificationsa					0.712
0	3 (5.3)	8 (5.4)
1	27 (48.2)	72 (48.9)
2	25 (44.6)	56 (38.1)
3	1 (1.7)	9 (6.1)
4	0 (0.0)	2 (1.3)

∗p-value (≤0.25) suggestive of an association with the outcome of TRO.

^aNumber of dietary modifications made simultaneously by participants.

Contingency table analysis of the three categories of medication classes used for obesity management revealed all had suggestive associations with TRO, injectable medications approved by the Federal Drug Administration (FDA) for the treatment of type 2 diabetes or obesity (p = 0.0002), FDA approved oral medications for the treatment of obesity (p = 0.037), and off label medications (p = 0.117) (Table 5). In some cases, participants were taking medications from more than one category simultaneously but the number of concurrent medication classes did not suggest an association to the outcome of TRO (p = 0.682).

Table 5.

Contingency table analysis of pharmacotherapy categories and TRO.

Pharmacotherapy Category	TRO n (%)	Non-TRO n (%)	ChiSquare (DF)	Pearson (p-value)
FDA Approved Injectables	19 (33.9)	93 (63.2)	14.11 (1)	∗0.0002
FDA Approved Oral Medications	30 (53.5)	55 (37.4)	4.34 (1)	∗0.037
Off Label Medications	19 (33.9)	34 (23.1)	2.45 (1)	∗0.117
Number of Categoriesa			7.65 (2)	0.682
None (0)	7 (12.5)	14 (9.5)
Low (1)	30 (53.5)	88 (59.8)
High (≥2)	19 (33.9)	45 (30.6)

∗p-value (≤0.25) suggestive of an association with the outcome of TRO.

^aNumber of pharmacotherapy categories taken simultaneously by participants.

The final multivariable logistic regression contained eight biopsychosocial variables with suggestive associations with TRO: three dietary modifications (lower carbohydrate, reduced calorie, less snacking), three classifications of pharmacotherapies (FDA approved injectable, FDA approved oral, and off-label medications), number of office visits, and sleep sufficiency. After stepwise elimination of variables which failed to retain statistical significance at the p = 0.05 level in the multivariable model, The final model included only two biopsychosocial variables, FDA approved injectable medications (OR = 3.49, p = 0.0002) and lower carbohydrate diet modification (OR = 2.18, p = 0.035).

Given that only two predictors remained significant in the final multivariable model, a secondary ad hoc analysis was performed to test the significance of an interaction term between FDA approved injectable medications and a lower carbohydrate diet. The Wald-type marginal p-value for this interaction term in the multivariable logistic regression model was 0.5159, indicating no evidence for a statistically significant interaction.

TRO occurred in 56% of participants who did not use FDA approved injectable medications and who did not use a low carb diet. TRO occurred in 14% of participants who used injectable meds and followed a low carb diet. While the model fit was found to be statistically significant (p = 0.0001), the independent variables of FDA approved injectable medications and low carbohydrate diet modification explain a relatively small amount of the variability of TRO within the model (RSquare = 0.07) (Table 6). This indicates that while the associations between these two biopsychosocial independent variables and the outcome of TRO are statistically significant, the measured predictors in the model do not fully explain the observed variability in TRO status.

Table 6.

Final model analysis predicting TRO.

Independent Variable	Param. Est. (SD) (p-value)	Odds Ratio (95% CI)
Not using FDA Approved Injectable Medications	0.63 (0.168) ∗0.0002	3.49 (1.83–6.86)
Not following Lower Carbohydrate Diet Modification	0.39 (0.186) ∗0.0356	2.18 (1.05–4.55)
Whole Model Test	p-value	ChiSquare (DF)	RSquare (U)
	∗< 0.001	18.52 (2)	0.077

∗Statistically significant p-values <0.05.

4. Discussion

Using a retrospective cohort design, this research generated knowledge about midlife women who receive medically specialized obesity management for one year and experience TRO. In answering the first research question and determining the prevalence of TRO among midlife women midlife who receive medically specialized obesity management over the course of one year, it was discovered that a portion of midlife women with obesity experience this phenomenon. More than a quarter of women within the study sample, 27.6%, demonstrated the presence of TRO, defined as an inability to achieve clinically meaningful weight loss of ≥5% of body weight after one year of medically specialized obesity management. This prevalence was lower than expected given previous research findings that 45–68% of participants receiving medically specialized obesity care did not achieve clinically meaningful weight loss of ≥5% body weight [[22], [23], [24], [25], [26]].

There are several potential reasons that the sample of midlife women with obesity in this study had lower rates of TRO than the previously discussed research studies evaluating weight loss outcomes of those who received medically specialized obesity management including differences in the study samples, length of study, and a focus on factors resulting in successful weight loss. Unlike this study, previous studies included men in their samples and were not specific to midlife women. Although the primary composition of previous research samples were adult women, and while all prior study samples contained only adults, age was not a consideration for inclusion or in the analysis in any of them. Data for this study was collected over the course of one year, which differs from some of the other studies which only collected data for up to six months [22,23]. In this current research, the focus was biopsychosocial factors associated with the population of midlife women with TRO who did not achieve clinically meaningful weight loss after one year of medically specialized obesity care. This differs from the prior research studies which focused on reporting about factors that were associated with those successful in achieving clinically meaningful weight loss with little attention given to those who were unsuccessful. While this difference does not add to the potential explanation of why there is a lower-than-expected rate of TRO within this study, it does highlight the fact that very little is known about this unsuccessful population and the factors that influence the outcome of TRO.

5. Limitations

There were limitations of this research study that should be acknowledged when interpreting these findings. Key limitations in this study were the availability of data in a retrospective design, the measurement of variables, timing of data collection, and the homogenous sample. The data available for collection cannot be predetermined by the researcher when using retrospective health record data. Obesity is multifactorial and complex and there are many variables that can act as barriers to weight loss, exploration of all these factors was not feasible for this study given the data available in the electronic health record. In this research, the available documented data for the biopsychosocial variables of diet modification, physical activity, sleep, and stress coping often lacked specificity and the quantification of these variables to dichotomous categories was a limitation. For example, regarding dietary modifications and physical activity participants did not always report carbohydrate or calorie counts or the amount, timing, or specific types of physical activity they incorporated into their daily routines. One aspect of sleep sufficiency is the number of hours slept per night which could be measured as a continuous variable but the obesity medicine specialists in this clinic only documented the cut point of greater or less than 7 h per night. Given the format of the data available in the electronic health record, there was little choice other than dichotomization of these variables although when possible, such as in the case of dietary modification and pharmacotherapies, refinement to the variable categories was completed. Dichotomization of a variable for measurement often leads to missed data as nuanced information is often lost in that process affecting statistical efficacy [28]. Likewise, data on several variables that may cause inadequate response to obesity treatment (e.g., genetic, environmental, or hormonal factors) were not recorded in the electronic health records and were unable to be assessed as part of this research.

Self-reported data for diet modifications, physical activity, sleep, and stress coping also has recognized limitations including inaccuracy of memory and social desirability response, or the desire to report in a way that is perceived to be more favorable. In research, under reporting of daily food intake [29] and over reporting of physical activity [30] have been noted and occur to a greater extent among those with obesity. Not only could the data collected for these variables be inaccurately reported but it could have been inaccurately documented. Providers within the metabolic weight management clinic can pull documented text from previous health records into the current encounter documentation to improve efficiency. It has been suggested that the ability to clone text from another source could lead to a failure to update information and creation of an inaccurate medical record [31].

Cross-sectional data was collected from only the final encounter within the 12-month study period for several of the biopsychosocial variables of interest including sleep, stress coping strategies, physical activity, and dietary modifications. The decision to limit data collection for each of these variables to only the final visit was primarily based on rationale that changes to these behaviors would require time to incorporate. One year of medically specialized obesity care with a minimum exposure of four visits, was thought adequate for participants to incorporate recommendations made by providers for behavior changes regarding sleep, stress coping, dietary modification, and physical activity that would optimize weight loss. However, having only cross-sectional data poses a limitation that knowledge of when these behavior changes were made is lacking and it cannot be assumed that the participants followed these patterns over the entire study period. Future prospective research could be designed to determine the trajectory of behavior changes related to outcomes in this population.

Finally, the sample for this research study was identified from a single medically specialized obesity management clinic. Its composition was racially and ethnically homogenous and may have represented a population of higher socioeconomic status given the higher-than-average number of participants with commercial insurance which could limit the generalizability of these findings.

Despite the noted limitations, this research study describes a robust measurement of prevalence of clinically meaningful weight loss over one year. This measurement is critical to expanding understanding of a previously unexplored phenomenon of TRO among midlife women with a condition presumed to carry increased health risk. Findings from this study provide an initial baseline epidemiological prevalence for TRO among midlife women which provides not only a starting reference for understanding its frequency but also a foundation on which to build future research, as they report metrics critical for a rigorous power calculation necessary to design a larger research study.

6. Conclusion

This pilot research study was the initial investigation into midlife women with TRO, a newly identified phenomenon that characterizes those who despite receiving medically specialized obesity care do not achieve clinically meaningful weight loss of ≥5%. Prior to this study, very little was known about the phenomenon of TRO because it had not been specifically identified in the obesity research literature which has predominantly focused only on those who are successful in weight loss.

Obesity is a chronic complex multifactorial disease with many contributing and sustaining biopsychosocial factors. This study found that certain medication and dietary changes had associations with the outcome of TRO. Use of FDA approved injectable medications were predictive of TRO in that those who did not use them were three times more likely to experience the phenomenon. Lowering dietary carbohydrates, although measured dichotomously, was another predictor of TRO, given that midlife women who did not follow this dietary modification were at twofold greater risk of demonstrating TRO in this study. There is no single biopsychosocial factor that can adequately explain the difference between midlife women who experience TRO and those who do not. This research demonstrates that although the use of FDA approved injectable medications and lowering dietary carbohydrates are potentially impactful management strategies for midlife women to reduce their chance of TRO, they will likely not fully explain the differences among the population who experiences it. Continued research is necessary to better understand this at-risk population of midlife women with TRO for the development of more efficacious management strategies. Additionally, TRO is a phenomenon that should be explored in men and people in other age groups.

For healthcare providers, this research will offer knowledge that while some women in midlife may be unable to attain clinically meaningful weight loss, this inability does not reflect that they are not following the same recommendations and treatment plans as those who are successful. This research identifies that midlife women with TRO exist and that increased emphasis on a diet lower in carbohydrate and the use of FDA approved injectable medications for weight loss may be appropriate. And while these interventions may improve weight loss outcomes within this population, other biopsychosocial factors will need to be considered for some midlife women to achieve clinically meaningful weight loss.

Takeaway clinical messages (3)

• •More than a quarter of midlife women with obesity may exhibit treatment resistant obesity (TRO) and fail to achieve clinically meaningful weight loss of ≥5% after receiving one year of medically specialized obesity care.
• •The combination of lowering dietary carbohydrates and taking FDA approved injectable medications for weight loss in midlife women should be explored further as this may reduce their risk of TRO.
• •Midlife women with TRO have not been fully explored in research and although this research begins to lay the foundation for future work, greater understanding of this phenomenon will be necessary to develop the most efficacious treatment plans for this at-risk population.

Credit author statement (Dissertation-based manuscript)

Emily C. Stevens: Conceptualization; Methodology; Investigation; Data Curation; Formal Analysis; Writing – original draft; Writing – review & editing; Visualization; Project Administration.

Lisa Shah: Conceptualization; Methodology; Supervision; Writing – review & editing; Validation.

Amanda Elswick Gentry: Methodology; Formal Analysis; Writing – review & editing; Validation.

Jo Robins: Methodology; Writing – review & editing; Validation.

Patricia Kinser: Methodology; Writing – review & editing; Validation.

Disclosures

Amanda Elswick Gentry has contributed in a personal capacity not a representative of BeOne; the contents within this manuscript in no way reflect the beliefs or opinions of BeOne Medicines.

Ethical adherence & ethical review

This research fell under the category of exempt and therefore informed consent was not necessary and was not sought. As per the 2018 Common Rule exemption category 4 (ii), this research study was secondary research using identifiable private information that was de-identified and recorded by the investigator so that the identity of the subjects could not be readily ascertained (Protections (OHRP), 2017). Furthermore, the subjects of this research were not contacted by the investigator nor were the subjects re-identified after study completion. Institutional review board approval was obtained for this research by Virginia Commonwealth University (IRB ID: HM20030564) and the healthcare system which owned the metabolic clinic (IRB # 00002496). Unless otherwise stated, responsibility for editorial decisions and peer review process for this article was delegated to non-author Editors or non-author Associate Editors.

Declaration of artificial intelligence (AI) and AI assisted technology

During the preparation of this manual, generative AI technology was used to create a draft rendering of the graphic abstract submitted. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Source of funding

Beyond payment to the nursing doctoral program faculty by Virginia Commonwealth University, this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Appendix A. Supplementary data

The following is the Supplementary data to this article: