Higher Caloric Exposure in Critically Ill Patients Transiently Accelerates Thyroid Hormone Activation

Liam McKeever; Sarah J. Peterson; Omar Lateef; Sally Freels; Tatiana L. Fonseca; Barbara M. L. C. Bocco; Gustavo W. Fernandes; Kelly Roehl; Kristen Nowak; Marisa Mozer; Antonio C. Bianco; Carol A. Braunschweig


J Clin Endocrinol Metab. 2020;105(2) 

In This Article


Overview and Design

We performed a single-center PRCT to assess the impact of energy intake at 100% vs 40% of ECN on thyroid function tests in patients admitted to the neurosurgical or medical intensive care units (NSICU or MICU) at Rush University Medical Center. Inclusion criteria: Patients were considered for enrollment if they were (a) older than 18 years of age, (b) placed on mechanical ventilation and (c) met the criteria for SIRS (Figure 1). Exclusion criteria: Patients were excluded from enrollment if they were pregnant, severely immunocompromised, had a history of thyroid disease or were receiving thyroid hormone therapy, could not be consented in English, did not require EN, were likely to be extubated within 72 hours, actively dying or brain dead, could not be placed on the study formula, had no family available to provide consent, were in a competing trial, or did not have a white blood cell count within the SIRS range for at least 2 consecutive days. These criteria were established to improve homogeneity in staging of the acute phase response, assuring that patients were profoundly inflamed.

Figure 1.

Criteria for systemic inflammatory response syndrome: SIRS criteria. Abbreviations: WBC, white blood cell count.

Recruitment and Randomization

Within 24 hours of meeting the eligibility criteria, the patient or their surrogate decision maker was approached for participation. A written consent was obtained prior to randomization in the study. Following informed consent, randomization to receive either 100% ECN or 40% ECN for caloric dosing occurred using a computer-generated list of random numbers by a non–hospital-based staff member blinded to all aspects of the participant's profile.

Study Groups

All patients were fasted 7 hours prior to their baseline blood draw and initiation of EN. Participants then received Jevity 1.5 (Nestle Nutrition, 1500 kcals, 63.8 g protein/L), initiated at 20 mL/hour and advanced 20 mL every 4 hours until achievement of a goal hourly rate providing 100% ECN (25–30 kcals/kg) or 40% ECN (10–12 kcal/kg). Feeding interruptions were monitored daily, compensated by rate increases as needed to avoid caloric deficits and returned to the original rate once the correction was achieved. The intervention lasted for 7 days or until intensive care unit (ICU) discharge or death.

Specimen Collection and Assays

An evening blood draw was collected daily for 7 days or until ICU discharge or death. Blood was collected in 2 EDTA tubes, centrifuged within 10 minutes of collection and frozen to −70° C. The plasma samples were later thawed and TSH, total T3, and total T4 were quantified through immunoassay kits purchased from Monobind Inc. in Lake Forest, CA. Reverse T3 was quantified by an immunoassay kit purchased from Lifeome, Oceanside, CA. We are aware of marked discrepancies in thyroid hormone measurement in serum of NTIS patients when reference liquid chromatography with tandem mass spectrometry (LC-MS/MS) is compared to common immunoassay platforms. This is possibly caused by displacement of T4 binding to low-affinity albumin by several drugs. Thus, in the present studies we only measured total thyroid hormone serum concentrations.[26]

Demographic and Clinical Variables

Data were collected from the electronic medical records and confirmed with hospital staff as needed. Basic demographic variables included age, race, and gender. Clinical variables collected included weight, height, nutrition status, complete blood count, blood gases, mean arterial pressures, bilirubin, blood glucose, sodium, potassium, creatinine, vasopressor administration, Glasgow Coma Scale, temperature, pH, heart rate, and respiratory rate. Body mass index (BMI) was calculated as the patient weight at admission divided by the square of patient height (kg/m2). The Acute Physiology and Chronic Health Evaluation (APACHE II) score was calculated at ICU admission; the scale ranges from 0 to 71, with higher scores indicating greater severity of illness.[27] Sequential Organ Failure Assessment (SOFA) scores were taken at baseline and daily thereafter. SOFA scores range from 0 to 24, with higher scores reflecting worse organ dysfunction.[28,29] Nutrition status was quantified through nutrition focused physical examination. Nutrition status was assessed through subjective global assessment (SGA) upon admission.[30]

Outcome Variables

The study was powered to detect a 20% difference in the T3/rT3 ratio between groups. Other primary outcome variables included T3, rT3, TSH, and T4 values. Secondary outcome variables included classic clinical variables such as mortality, ICU/hospital length of stay, infections, and disposition.


Power analysis on the T3/rT3 ratio for this study was carried out using GLIMMPSE software (https://glimmpse.samplesizeshop.org) and data from[25] which provided a median and interquartile range. This was used to estimate a variable variance of 0.38. From this, it was determined a sample size of 40 participants would yield a power greater than 0.8 to detect a 20% increase in T3/rT3 ratio over 7 days. Variables were assessed for normality and transformed as needed. Standard descriptive statistics were utilized to assess baseline characteristics of the 2 groups (mean, median, ranges, standard deviation, and standard errors). Heterogeneity between groups at baseline was explored through χ2 tests (dichotomous variables) or Wilcoxon rank sum and Student t tests (continuous variables) for nonnormally and normally distributed variables, respectively.

The effect of calorie exposure group on the slope of change on outcome measures of plasma T3/rT3 ratio, TSH, T3, T4, and rT3 was explored with mixed linear model regression. This was accomplished by fitting models that explored randomization group interactions with both linear and quadratic versions of the study day variable. Age, sex, BMI, ethnicity, APACHE II, SOFA score, nutrition status, total and daily dose of insulin, number of hyper- and hypoglycemic days, parenteral nutrition (PN), SIRS criteria, number of days overfed, and propofol/amiodarone/pressor administration, were explored as potential confounders and were adjusted for as appropriate. Intercept, study day, and quadratic study day were explored as possible random effects in mixed models. Significance for these random effect variables was determined through likelihood ratio testing. To better understand the driving forces behind our findings for T3/rT3 ratio, an exploratory analysis was performed on all models truncated to the first 4 intervention days. Furthermore, to infer how each randomization group was contributing to the differences between them, group-stratified graphs of the raw thyroid panel data were visually inspected. Paired t tests (one-tailed) were performed for any initial apparent post-intervention increase or decrease. To protect against confounding by dropout or missing datapoints, the paired t tests were restricted to only those participants with complete data from baseline to the study day under inspection. A P value < 0.05 was considered significant for all statistical tests performed. All tests were run in Statistical Analysis Software (SAS 9.4).


Both the study protocol and consent forms were approved by the Institutional Review Boards of both Rush University Medical Center and the University of Illinois at Chicago prior to study initiation.