Obesity Paradox: Laboratory Findings in Uncomplicated Obesity. Is Bias a Plausible Explanation?

Abstract

Objectives

Obesity predisposes to multiple diseases, such as heart disease, diabetes, stroke, arthritis, and malignancy. However, obese patients have better outcomes than normal-weight patients with some of these disorders, including those admitted to critical care units. We compared the results for common laboratory tests in patients with uncomplicated obesity against the findings in normal-weight patients.

Methods

Patients who had a comprehensive metabolic profile test were identified. Patients with acute and/or chronic debilitating disorders were excluded, and the laboratory parameters were compared among 4 groups based on body mass index.

Results

With the exception of elevated triglycerides and lower high-density lipoprotein in obese and morbidly obese patients, laboratory findings were not meaningfully different from those in normal-weight patients.

Conclusions

The obesity paradox of better outcomes in obese patients admitted to critical care units could not be explained on the basis of lower additional disease burden necessitating critical care admission due to abnormal laboratory values at the baseline. It is conceivable that unconscious bias against obese patients, with lower disease burden than normal-weight patients, triggers their admission to critical care, thus creating the appearance of better outcomes.

Obesity paradox, Intensive care, Laboratory parameters

Introduction

IMPACT STATEMENT

The paradox is that obese patients admitted to critical care units have better outcomes than matched lean patients. The evidence presented in this report rules out abnormal baseline laboratory parameters as a potential explanation of the obesity paradox in patients admitted to intensive care units. The findings highlight that uncomplicated obesity does not result in abnormal laboratory findings, with the exception of high serum triglycerides and low HDL. Investigators should continue to search for other explanations for the obesity paradox, including unconscious bias, which could benefit the healthcare experienced by this cohort.

The obesity paradox is a hypothesis that obesity may be protective—the condition is associated with better survival in certain groups of people despite the association of obesity with increases in multiple disorders, such as diabetes, hypertension, ischemic heart disease, stroke, arthritis, and malignancy, to name a few (1–5,). It has also been observed that obese patients tend to have better outcomes when admitted to an intensive care unit when having congestive heart failure or intracerebral hemorrhage or when undergoing surgery and chemotherapy for malignancy and other disorders. The mechanism of this paradox is not known and is the subject of this study (6–15).

We focused on the paradox of better outcomes in obese patients with severe illness requiring admission to intensive care units and considered the following potential explanations for the finding (15–18). First, overweight/obese patients have higher or more abnormal baseline laboratory values. Their laboratory values become abnormal enough to warrant admission to critical care units, with lower levels of additional insult to health. Consequently, they are admitted to intensive care with lower disease burden than lean patients with normal baseline laboratory values. Second, overweight and obese patients have more fat reserves to tide them over the catabolic state of acute illness. Third, overweight and obese patients have higher lean body mass and better cardiorespiratory reserves than lean patients. Fourth, lean patients are not healthy lean individuals but are patients who have lost considerable weight and were “emaciated” on entry to intensive care unit.

We addressed only the first item in this study. Our premise was that “healthy” overweight or obese patients would have higher or more abnormal baseline laboratory values than lean patients, and it would take a smaller pathologic insult and/or additional disease burden for their laboratory values to vault them into a range requiring admission to critical care units.

Methods

This retrospective review of laboratory data was carried out at a tertiary-care, medical-school–affiliated medical center in the southeastern United States. The institution has 480 beds and serves as the level 1 trauma center for the city, offers solid organ and bone marrow transplants, interventional cardiology, open heart surgery, and bariatric surgery, among other tertiary care services. The study was approved by the institutional review board at Augusta University, and requirement for informed consent was waived.

Patients undergoing comprehensive metabolic profile testing in February 2019 were identified from laboratory records. Their laboratory data and clinical records were reviewed. Patients with acute or chronic debilitating illnesses were excluded. Other exclusions included age <18 years; hospital admission within 3 months before or after the testing date; uncontrolled diabetes with hemoglobin (Hb) A_1c >9.0%; chronic kidney disease worse than stage 4; pregnancy at >20 weeks; chronic congestive heart failure; malignant tumor at stage >2; history of organ or bone marrow transplant; paralytic neurologic disorders interfering with any activities of daily living, chronic debilitating illnesses such as Hb SS disease, cirrhosis, and severe chronic obstructive pulmonary disease; and patients hospitalized for any disorder. Patients with >10% weight loss or gain in the prior 3 months were also excluded.

The following demographic and laboratory data were collected: age, race, sex, body mass index (BMI), white blood cell count, Hb, mean corpuscular volume, red cell distribution width, platelet count, sodium, potassium, chloride, carbon dioxide, blood urea nitrogen, creatinine, glucose (random), calcium, total serum protein, serum albumin, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, total bilirubin, total cholesterol, triglycerides, high-density lipoprotein (HDL), calculated low-density lipoprotein, Hb A_1c, and thyroid-stimulating hormone. Clinical notes were reviewed to ensure exclusion of acutely ill, hospitalized patients and those with chronic uncontrolled or debilitating disorders.

Patients were stratified into 4 groups, based on BMI: 18.5 to 25, 25.1 to 30, 30.1 to 35, and >35. Laboratory parameters of the 3 high-BMI groups were compared with the BMI 18.5 to 25 group by t test using a 2-tailed test with unequal variances. The Microsoft Excel program was used for testing. The number of observations and the mean values for each parameter were recorded for all 4 groups. The P value generated by the t test was subjected to Bonferroni correction for the 25 parameters. Bonferroni correction was applied to provide a conservative estimate of the significance, because in the 25 comparisons carried out in the study, at least 1 comparison would be expected to be significant at P = 0.05 by chance alone.

Results

The age, race, and sex information for the 4 groups is given in Table 1. The age for the group with BMI >35 was lower than that for BMI <25 group, but the difference is not significant. There were more women in the population examined, and the difference was especially marked in the BMI >35 group, in which 77% of the patients were women. Similarly, there was a marked preponderance of black participants in the BMI > 35 group compared with other groups.

Table 1

Race and sex distributions among the various groups are significantly different.

	BMIa				χ2	P value
	<25	25.1–30	30.1–35	>35
Race					31.7984	0.000216
White	63	96	72	66
Black	36	49	40	91
Hispanic	4	8	3	4
Other	4	6	1	1
Sex					32.9282	<0.00001
Female	77	81	56	121
Male	30	78	60	41
Average age	54.11	57.80	57.68	49.73
Average BMI	22.07	27.68	32.00	43.23

The laboratory values for the 4 groups are displayed in Table 2. The average value and the number of observations for each analyte in each of the BMI groups are given. The P values of the higher BMI groups, compared with the BMI <25 group, are shown, and the column marked “Corrected P” shows the P value after Bonferroni correction for the 25 comparisons.

Table 2

Laboratory data in different BMI^a (kg/m²) groups.^b

	BMI <25		BMI 25.1-30				BMI 30.1-35				BMI >35
Analyte	BMI <25	n	Mean	n	P vs <25	Corrected P	Mean	n	P vs <25	Corrected P	Mean	n	P vs <25	Corrected P
Sodium, mEq/L	139.98	107	140.12	159	0.913089586	22.82723966	140.09	116	0.76536741	19.13418526	140.11	162	0.716203017	17.90507542
Potassium, mEq/L	4.12	107	4.14	159	0.677786241	16.94465604	4.09	116	0.604929022	15.12322554	4.00	162	0.038744287	0.968607173
Chloride, mEq/L	103.88	107	104.25	159	0.627971901	15.69929753	103.75	116	0.743925054	18.59812636	103.36	162	0.375713918	9.39284796
CO2, mEq/L	27.17	107	26.99	159	0.658543691	16.46359227	26.79	116	0.345417661	8.635441531	26.42	162	0.030216872	0.755421791
BUN, mg/dL	14.90	107	15.84	159	0.271868821	6.796720521	16.41	116	0.064728149	1.618203719	14.29	162	0.44264719	11.06617975
Creatinine, mg/dL	0.88	107	0.98	159	0.005951897	0.148797421	1.01	116	0.000680209	0.017005236	1.46	162	0.291409408	7.285235192
Glucose, mg/dL	101.19	107	104.23	159	0.461032332	11.52580831	111.90	116	0.054977697	1.374442422	113.66	162	0.013054691	0.326367286
Calcium, mg/dL	9.62	107	9.63	159	0.903834538	22.59586345	9.59	116	0.564768636	14.1192159	9.47	162	0.001376631	0.034415772
T-protein, g/dL	7.17	107	7.22	159	0.630197944	15.7549486	7.25	116	0.437082456	10.92706141	7.37	160	0.042884709	1.072117731
Albumin, g/dL	4.23	107	4.24	159	0.864177235	21.60443089	4.26	116	0.609357811	15.23394528	4.10	160	0.035663722	0.891593051
AST, U/L	20.77	107	22.32	159	0.16330473	4.082618259	23.61	116	0.024780619	0.61951547	21.14	160	0.772277119	19.30692796
ALT, U/L	18.61	107	23.17	159	0.013600678	0.340016958	26.97	116	0.000157861	0.003946513	23.39	160	0.004108939	0.102723473
Alkaline phase, U/L	66.86	107	72.90	159	0.096525148	2.413128706	75.36	116	0.004806472	0.120161804	82.31	160	4.00039E-07	1.0001E-05
T-bilirubin, mg/dL	0.51	107	0.58	159	0.059606707	1.49016768	0.50	116	0.853240693	21.33101733	0.46	160	0.118443917	2.961097924
Cholesterol, mg/dL	182.46	61	179.14	99	0.674244985	16.85612462	178.70	79	0.648439284	16.21098209	172.49	109	0.234913477	5.872836913
Triglyceride, mg/dL	90.39	59	121.66	99	0.007890543	0.197263563	173.99	78	9.77806E-07	2.44452E-05	128.34	108	4.61461E-05	0.001153653
HDL, mg/dL	63.08	60	52.80	99	0.000764862	0.01912155	45.57	79	3.25301E-08	8.13252E-07	45.85	108	1.42052E-08	3.5513E-07
LDL, mg/dL	101.30	60	101.72	98	0.944288061	23.60720152	101.09	79	0.974589844	24.36474609	101.32	108	0.997232416	24.93081041
Hb A1c %	5.96	41	6.03	74	0.645352351	16.13380877	6.05	61	0.597362332	14.93405829	6.20	109	0.133965686	3.349142147
TSH, μIU/mL	1.93	72	1.95	94	0.959980276	23.9995069	2.76	64	0.19868716	4.967179011	2.36	112	0.33278858	8.319714499
WBC, 103/μL	6.58	102	6.70	147	0.771371404	19.2842851	6.94	106	0.410443601	10.26109003	7.53	151	0.0288852	0.722129994
Hb, g/dL	13.24	102	13.59	147	0.075554443	1.88886108	13.78	106	0.008960668	0.224016711	13.06	151	0.370790441	9.269761016
MCV, fL	90.14	102	89.41	147	0.378540941	9.463523518	90.75	106	0.534052766	13.35131916	86.34	151	2.07898E-05	0.000519746
RDW	14.08	102	14.30	147	0.452140191	11.30350477	14.15	106	0.813209145	20.33022862	14.96	151	0.002033677	0.050841931
Platelet, 103/μL	277.85	102	250.46	147	0.185748611	4.64371527	238.32	106	0.060826611	1.520665268	274.58	151	0.876051641	21.90129103

a

ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, blood urea nitrogen; Hb, hemoglobin; LDL, low-density lipoprotein; MCV, mean corpuscular volume; RDW, red cell distribution width; TSH, thyroid-stimulating hormone; WBC, white blood cell count.

b

Only triglyceride and HDL levels are significantly different in the obese and morbidly obese patients compared with normal-weight patients. The significant (P < 0.05) P values for the uncorrected t test are highlighted in yellow, and the significant values after Bonferroni correction are highlighted in green. It is remarkable that the only consistent and meaningful difference is higher triglyceride and lower HDL values in obese patients (BMI 30.1–35 and >35), highlighted in blue.

In Table 2, the significant (P < 0.05) P values for the uncorrected t-test are highlighted in yellow, and the significant values after Bonferroni correction are highlighted in green. It is remarkable that the only consistent and meaningful difference is higher triglyceride and lower HDL values in obese patients (BMI 30.1–35 and >35), highlighted in blue.

In brief, the commonly used laboratory test values were not meaningfully different among the various groups with different BMIs.

Discussion

There is general agreement that obesity is associated with increased morbidity and mortality (1–5,). The increase in global obesity is associated with increased prevalence of diabetes, cardiovascular disorders, and cancer. The observation of the obesity paradox—that is, better outcomes in obese patients when admitted to critical care units, compared with normal-weight patients—is still a matter of controversy (17–20,). BMI as the sole indicator of obesity may not be appropriate, and it has been suggested that waist circumference and hip-to-waist ratio be taken into consideration (16). Despite the magnitude of the problem for global health, randomized controlled trials to settle the controversy may not be practical or feasible.

The term obesity paradox is applied differently in different circumstances and by different authors. Some have applied this term to explain better outcomes in overweight and obese patients with cardiovascular disease (17,). Others have used it to point out better outcomes in overweight and obese patients when admitted to critical care units (8). We have only addressed the latter circumstance.

We examined a single issue in this observational, retrospective study: Do “healthy” obese patients have abnormal laboratory values that may result in their entry to an intensive care unit with lower additional disease burden and thus create the appearance of better outcome compared with normal-weight individuals. The lack of abnormal values, with the exception of triglyceride and HDL, effectively rules out this explanation of the obesity paradox.

We did not survey providers for their attitudes toward obese patients; however, obese people tend to face bias in their interactions with society (21). The bias may be unintended or unconscious. A similar, unrealized, or unconscious bias toward obese patients may trigger their admission to critical care units at a disease burden for which normal-weight patients may not be admitted to intensive care units, creating an artifact of better outcomes in obese patients.

The study has important limitations. This retrospective, observational study of normal-weight and overweight or obese individuals are not comparable in terms of age, sex, and race. The patients were also not controlled for the type of pathology resulting in the visit to a tertiary care facility. However, we did try to control for recent weight changes and complications associated with obesity.

In summary, results of common laboratory tests are not significantly different between normal weight and obese, even morbidly obese, patients, except for higher levels of triglycerides and lower levels of HDL in obese and morbidly obese patients. Thus, the hypothesis is unlikely to be tenable that obese patients have better outcomes on admission to intensive care units because they are admitted to intensive care with lower additional disease burden, compared with lean individuals, due the obese patients having more abnormal laboratory values at their baseline and lower additional pathologic insult qualifying them for admission to intensive care units. Although we did not directly address it, a possible explanation for the better outcomes of obese patients on admission to intensive care could still be due to their gaining admission to intensive care with lower disease burden than lean patients. This could be due to a conscious or unconscious bias, by caregivers, against obese patients, assuming them to be sicker than they may be, and the biased assumption that obese patients would be more difficult to care for in routine medical care settings.

Nonstandard Abbreviations: Hb, hemoglobin; BMI, body mass index; HDL, high-density lipoprotein.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

Authors’ Disclosures or Potential Conflicts of Interest: No authors declared any potential conflicts of interest.

Role of Sponsor: No sponsor was declared.