Adrenal function testing in patients with septic shock

cc5077 CCJ Research Adrenal function testing in patients with septic shock Salgado Ribeiro Diamantino d.salgado@globo.com Verdeal Carlos Rosso Juan juanverdeal@alternex.com.br Rocco Rodolfo José jrrocco@globo.com

Intensive Care Unit, Barra Dor Hospital, Avenida Ayrton Senna 2541, Barra da Tijuca, Rio de Janeiro CEP 22775-001, Brazil

Intensive Care Unit, Clementino Fraga Filho University Hospital, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil

Critical Care 1364-8535 2006 10 5 R149 http://ccforum.com/content/10/5/R149 See related commentary by Marik, http://ccforum.com/content/10/6/176 17062164 10.1186/cc5077 27 7 2006 21 8 2006 22 9 2006 25 10 2006 25 10 2006 2006 Salgado et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Introduction

Adrenal failure (AF) is associated with increased mortality in septic patients. Nonetheless, there is no agreement regarding the best diagnostic criteria for AF. We compared the diagnosis of AF considering different baseline total cortisol cutoff values and Δmax values after low (1 μg) and high (249 μg) doses of corticotropin, we analyzed the impact of serum albumin on AF identification and we correlated laboratorial AF with norepinephrine removal.

Methods

A prospective noninterventional study was performed in an intensive care unit from May 2002 to May 2005, including septic shock patients over 18 years old without previous steroid usage. After measurement of serum albumin and baseline total cortisol, the patients were sequentially submitted to 1 μg and 249 μg corticotropin tests with a 60-minute interval between doses. Post-stimuli cortisol levels were drawn 60 minutes after each test (cortisol 60 and cortisol 120). The cortisol 60 and cortisol 120 values minus baseline were called Δmax¹and Δmax²⁴⁹, respectively. Adrenal failure was defined as Δmax²⁴⁹≤ 9 μg/dl or baseline cortisol ≤ 10 μg/dl. Other baseline cortisol cutoff values referred to as AF in other studies (≤15, ≤20, ≤25 and ≤34 μg/dl) were compared with Δmax²⁴⁹≤ 9 μg/dl and serum albumin influence. Norepinephrine removal was compared with the baseline cortisol values and Δmax²⁴⁹values.

Results

We enrolled 102 patients (43 male). AF was diagnosed in 22.5% (23/102). Patients with albumin ≤2.5 g/dl presented a lower baseline total cortisol level (15.5 μg/dl vs 22.4 μg/dl, P = 0.04) and a higher frequency of baseline cortisol ≤25 μg/dl (84% vs 58.3%, P = 0.05) than those with albumin > 2.5 g/dl. The Δmax²⁴⁹levels and Δmax²⁴⁹≤ 9, however, were not affected by serum albumin (14.5 μg/dl vs 18.8 μg/dl, P = 0.48 and 24% vs 25%, P = 1.0). Baseline cortisol ≤ 23.6 μg/dl was the most accurate diagnostic threshold to determine norepinephrine removal according to the receiver operating characteristic curve.

Conclusion

AF was identified in 22.5% of the studied population. Since Δmax²⁴⁹≤ 9 μg/dl results were not affected by serum albumin and since the baseline serum total cortisol varied directly with albumin levels, we propose that Δmax²⁴⁹≤ 9 μg/dl, which means Δmax after high corticotropin dose may be a better option for AF diagnosis whenever measurement of free cortisol is not available. Baseline cortisol ≤23.6 μg/dl was the best value for predicting norepinephrine removal in patients without corticosteroid treatment.

Introduction

Septic shock and multiple organ failure are the most frequent causes of death in medical intensive care units (ICUs), and were responsible for approximately 750,000 admissions and 210,000 deaths throughout 1995 in the USA 1. Although the mortality rate has been declining, the incidence of sepsis has increased, resulting in almost three times the number of sepsis-related deaths in the USA between 1979 and 2000 2. Furthermore, patients who survive sepsis have a worse long-term health-related quality of life 34.

A normally functioning hypothalamic–pituitary–adrenal axis is essential to organ homeostasis during acute illness, especially in severe infections. Patients with adrenal failure (AF) and septic shock are known to present high mortality rates 5. Additionally, stress doses of hydrocortisone have been shown to reduce septic-shock-related mortality 678, mainly in patients with an impaired response to the corticotropin test 910.

The incidence of AF may vary widely, from 0% to 77%, depending on the studied population and the criteria used for the diagnosis 11. Although chronic AF (Addison disease) is rare and relatively simple to identify, the acute presentation is not straightforward and is usually unrecognized. A high suspicion index is necessary since the classical clinical and laboratorial findings, such as hypotension, fever, abdominal pain, hypoglycemia, electrolyte abnormalities and eosinophilia, are easily mistaken for the critically ill condition.

The failure of serum cortisol to rise above 18 or 20 μg/dl at 30 or 60 minutes after administration on 250 μg corticotropin is classically considered diagnostic for adrenal failure in the absence of acute illness 12. There is, however, no consensus on the best diagnostic criteria for AF in acutely ill patients. Some authors recommend the use of baseline cortisol, justifying that acute stress such as severe infections, surgery, shock and hypoxemia are enough stimuli to the adrenal glands 111314. Many cortisol cutoff values have been proposed, ranging from 10 to 34 μg/dl, but most researchers consider a baseline cortisol lower than 15 μg/dl to better identify patients presenting clinical features of corticosteroid insufficiency or those who would benefit from corticosteroid replacement 1516. Conversely, other authors consider that it is possible, by performing a low-dose (1 μg) 10 or mainly a high-dose (250 μg) corticotropin test 917, to identify patients with relative adrenal insufficiency when the difference between the highest serum cortisol value and the baseline cortisol value (Δmax) is lower than 9 μg/dl 1718. Some of these authors suggest these adrenal insufficiency patients could benefit from corticosteroid treatment 91017. Additionally, because more than 90% of circulating cortisol in human serum is bound to proteins (corticosteroid binding protein and albumin), either the baseline cortisol level or the Δmax value may be influenced by serum protein levels, and alterations in the binding proteins could affect the measured concentration of total serum cortisol 192021.

The main aim of the present study was to compare the diagnosis of adrenal failure considering different baseline cortisol cutoff values and the Δmax value after a low dose (1 μg) and a high dose (249 μg) of corticotropin. Secondarily, we analyzed the impact of serum albumin on AF identification and we correlated the laboratorial diagnosis of AF with norepinephrine removal in patients with septic shock.

Patients and methods

Patients

This prospective and noninterventional study was carried out in an ICU of a tertiary private hospital from May 2002 to May 2005. The Institutional Ethics Committee waived the need for informed consent as the test is routinely performed in the unit.

We included patients over 18 years old admitted to the ICU with septic shock, defined by the American College of Chest Physicians/Society of Critical Care Medicine Conference Consensus Committee 22 as the presence of two or more of the following conditions as a result of infection: fever (temperature >38°C) or hypothermia (temperature <36°C); tachycardia (heart rate >90 beats/minute), tachypnoea (respiratory rate >20 breaths/minute) or hyperventilation (partial pressure of arterial blood CO₂<32 mmHg); altered white blood cell count (>12,000 cells/mm³or <4,000 cells/mm³or >10% band forms); and hypotension (mean arterial blood pressure <65 mmHg) with vasopressor drug requirement (norepinephrine at any dose) despite adequate fluid resuscitation.

Exclusion criteria were corticosteroid usage in the past year or its requirement as part of the treatment in the present admission, administration of etomidate, end-stage cancer, AIDS, chronic renal failure requiring hemodialysis support, cirrhosis (Child–Pugh class C), pregnancy, or limitation of lifecare support.

The presence of community-acquired infection was established according to clinical, imaging and microbiological parameters within 72 hours of ICU admission or hospital admission. For nosocomial infection, the diagnosis was based on the Centers for Disease Control criteria 23. The infection sources were classified as lung, urinary, abdominal or pelvic, skin, vascular, and others.

General management

A pulmonary artery catheter or a central venous catheter and an arterial line were placed in all patients. Norepinephrine was used to maintain mean arterial blood pressure higher than 65 mmHg after adequate fluid resuscitation (central venous or pulmonary artery occlusion pressure >12 mmHg) with either crystalloids or colloids. Antibiotics were guided according to institutional recommendations. The mechanical ventilation strategy and the renal replacement therapy (continuous hemofiltration preferably) were in accordance with the Surviving Sepsis Campaign 24. All patients received deep venous thrombosis prophylaxis and stress ulcer prophylaxis. Continuous infusion of midazolam and fentanyl were given when necessary. Hydrocortisone (100 mg intravenously three times per day) therapy was not controlled and was administered at the discretion of the attending physician.

Clinical evaluation

Patients were followed up to hospital discharge. The following variables were recorded: age, gender, admission category (trauma, medical or surgical), source of infection, and dates of hospital admission and ICU admission; the Acute Physiology and Chronic Health Evaluation II score 25 and the Simplified Acute Physiology II score 26 on the first admission day, and the Sequential Organ Failure Assessment 27 on days 1, 3, 5, 7, 10, 14, 18, 21, 24, 28, and 30 of ICU admission; the maximum Sequential Organ Failure Assessment score during the ICU stay 28; the use of norepinephrine during the first five days of hydrocortisone treatment or after the corticotropin test when steroid was not given; and the hospital, ICU, and 28-day mortality rates.

Corticotropin test and laboratory variables

All patients were submitted to a low-dose (1 μg) and a high-dose (249 μg) corticotropin test (Synacthene; Novartis Pharma Stein AG, Stein, Switzerland) within the septic shock period, and analysis of serum cortisol was performed by radioimmunoassay (Immulite; DPC Diagnostic Products, Los Angeles, CA, USA). After measurement of the serum baseline cortisol, the patients were sequentially submitted to 1 μg and 249 μg corticotropin tests with a 60-minute interval between doses. Post-stimuli cortisol levels were drawn 60 minutes after each test (cortisol 60 and cortisol 120). The cortisol 60 and cortisol 120 values minus the baseline cortisol value were called Δmax¹and Δmax²⁴⁹, respectively. Cortisol results were available within 24 hours after testing. All patients were on norepinephrine at the time of the test. Laboratorial adrenal failure was defined as Δmax²⁴⁹≤ 9 μg/dl or baseline cortisol ≤ 10 μg/dl. Clinical adrenal failure was defined as removal of norepinephrine up to 120 hours after hydrocortisone treatment. The different baseline cortisol cutoff values referred to as adrenal failure criteria in other studies (≤15, 20, 25 and 34 μg/dl), as well as Δmax¹≤ 9 μg/dl and Δmax²⁴⁹≤ 9 μg/dl, were analyzed concerning serum albumin levels ≤2.5 g/dl and >2.5 g/dl.

At the onset of septic shock, blood and cultures from the site of infection were obtained. The white blood cell count, the total serum protein and albumin, and the C-reactive protein level were measured on the corticotropin test day.

Statistical analysis

Statistical analyses were conducted using SPSS 10.0 statistical package software (SPSS Incorporation, Chicago, IL, USA). The results of continuous variables are expressed as the median and interquartile range (25–75%). Nonparametric tests were used (Mann–Whitney U test for two independent samples, and Kruskal–Wallis H test for several independent samples) after exclusion of a normal distribution of the data by the Kolmogorov–Smirnov test. The chi-square test or Fisher's exact test was applied for categorical variables when indicated. Norepinephrine removal was estimated by the Kaplan–Meier method and the results were compared between groups with the log-rank test.

The baseline cortisol concentration, the most accurate predictor of the hemodynamic response to corticosteroid treatment, was determined by the area under the receiver operating characteristic curve. Pearson or Spearman's correlation coefficients correlated continuous variables. The sensitivity and specificity of each baseline cortisol cutoff value (≤15, 20, 25 and 34 μg/dl) and Δmax¹were determined using a concentration of Δmax²⁴⁹≤ 9 μg/dl as the reference method. Differences were considered significant at P < 0.05.

Results

Patient characteristics

One hundred and two patients were included (59 female), with a median age of 74 (62–82) years. The demographic and epidemiological data are presented in Table 1. The main types of ICU admission were medical (n = 50) and postoperative (n = 47). Acute respiratory distress syndrome was diagnosed in 52 patients (51%), and the lung was the most frequent source of infection. Despite the clinical septic shock diagnosis, only 69 patients (67.6%) yielded positive microbiological results, with Gram-negative bacillus predominance. Bacteremia was detected in 22 (21.6%) patients, and nosocomial infections were as frequent as community infections. All patients were receiving norepinephrine when the corticotropin test was performed, and hydrocortisone was given to 71 patients during the shock phase for a median of 9 (5–13) days. The 28-day, ICU, and hospital mortality rates were 31.4%, 35.3%, and 45.1%, respectively.

Table 1

Clinical, biological and infection data of patients

Sex

Female

59 (57.8)

Male

43 (42.2)

Age (years)

74 (62–82)

Admission diagnoses

Postoperative

47 (46.1)

Medical

50 (49)

Trauma

5 (4.9)

Acute respiratory distress syndrome

52 (51)

Length of intensive care unit stay (days)

18 (12–27.3)

Length of hospital stay (days)

33 (19–85.3)

Days for corticotropin test^a

2 (1–4)

Total serum protein (g/dl)^b

4.7 (4.1–5.2)

Serum albumin (g/dl)^b

2 (1.7–2.4)

White blood cells (cells/μl)

11,550 (8,275–16,500)

C-reactive protein (mg/dl)

16.3 (10.5–25)

Baseline cortisol levels (μg/dl)

17.2 (10.6–23.8)

Cortisol 60 (μg/dl)

25.4 (18.6–39.2)

Cortisol 120 (μg/dl)

35.5 (24.5–49.9)

Δmax¹(μg/dl)

8.2 (3.5–17.3)

Δmax²⁴⁹(μg/dl)

16.6 (10–26)

Mean daily dose of norepinephrine^c(μg/kg/minute)

0.13 (0.04–0.54)

Sequential Organ Failure Assessment score (corticotropin test day)

9 (7.8–11)

Maximum Sequential Organ Failure Assessment score

10 (9–13)

First-day Sequential Organ Failure Assessment score

7 (4–10)

Acute Physiologic and Chronic Heath Evaluation II score

20 (17–23)

Simplified Acute Physiology II score

43.5 (36–52.3)

28-day mortality rate

32 (31.4)

Intensive care unit mortality rate

36 (35.3)

Hospital mortality rate

46 (45.1)

Site of infection

◦Intra-abdominal

28 (27.5)

◦Soft tissue abscess/cellulitis

3 (2.9)

◦Lung

58 (56.9)

◦Central nervous system

1 (1)

◦Urinary tract

10 (9.8)

◦Endovascular infection

2 (2)

Microorganisms^d

◦Gram-negative bacillus

72 (61.6)

◦Gram-positive coccus

39 (33.3)

◦Fungus

6 (5.1)

Bacteremia

22 (21.6)

Type of infection

◦Community

51 (50)

◦Nosocomial

51 (50)

Data presented as n (%) or median (interquartile range). Cortisol units: 1 μg/dl = 27.6 nmol/l. Cortisol 60, serum cortisol level drawn 60 minutes after stimulus with 1 μg corticotropin; Cortisol 120, serum cortisol level drawn 60 minutes after stimulus with 249 μg corticotropin (120 minutes after serum baseline cortisol); Δmax¹, cortisol 60 value minus the baseline cortisol level; Δmax²⁴⁹, cortisol 120 value minus the baseline cortisol level. ^aDifference between corticotropin test day and intensive care unit admission day. ^bProtein and albumin levels were not measured in 15 patients. ^cMean daily dose of norepinephrine during the first five days after hydrocortisone usage or after the corticotropin test. ^dMicrobiological identification was obtained in 69 patients (67.6%).

Laboratorial evaluation of adrenal function

The median values of the baseline cortisol level, cortisol 60, cortisol 120, Δmax¹, and Δmax²⁴⁹were 17.2 μg/dl, 25.4 μg/dl, 35.5 μg/dl, 8.2 μg/dl, and 16.6 μg/dl, respectively. The median time for the corticotropin test was 2 (1–4) days after ICU admission (Table 1). AF was identified in 23 patients (22.5%) considering either Δmax²⁴⁹≤ 9 μg/dl or baseline cortisol ≤ 10 μg/dl, but only four patients (3.9%) presented both criteria.

There was a weak but a significant correlation between the serum albumin and baseline cortisol values – the lower the albumin levels, the lower the baseline cortisol value (R²= 0.1, P = 0.02). The same finding was observed for cortisol 60 (R²= 0.15, P = 0.002) and for cortisol 120 (R²= 0.1, P = 0.01). Nonetheless, there was no correlation between serum albumin and Δmax²⁴⁹values (R²= 0.02, P = 0.08). AF was identified more frequently in hypoalbuminemic patients than in those with near-normal serum albumin levels when considering the baseline cortisol value as the diagnostic criterion, especially for the cortisol cutoff value ≤ 25 μg/dl (P = 0.05). On the other hand, the incidence of AF according to Δmax¹≤ 9 μg/dl and, mainly, Δmax²⁴⁹≤ 9 μg/dl criteria was not affected by serum albumin levels (Table 2).

Table 2

Adrenal function according to the serum albumin level

Albumin ≤ 2.5 g/dl (n = 75)

Albumin > 2.5 g/dl (n = 12)

P value

Baseline cortisol (μg/dl)

15.5 (10.4–23.4)

22.4 (15.5–26.7)

0.04*

Cortisol 60 (μg/dl)

23.5 (17.2–36.6)

40.3 (27–51.5)

0.009*

Cortisol 120 (μg/dl)

30.4 (23.2–46.7)

41.9 (33.1–54.9)

0.04*

Δmax¹(μg/dl)

6.9 (3.5–13.2)

13.2 (3.7–26.6)

0.24

Δmax²⁴⁹(μg/dl)

14.5 (9.4–23.4)

18.8 (9.9–28.9)

0.48

Baseline cortisol ≤ 10

17 (22.7)

1 (8.3)

0.45

Baseline cortisol ≤ 15

37 (49.3)

3 (25)

0.12

Baseline cortisol ≤ 20

52 (69.3)

6 (50)

0.2

Baseline cortisol ≤ 23.6

58 (77.3)

6 (50)

0.07

Baseline cortisol ≤ 25

63 (84)

7 (58.3)

0.05*

Baseline cortisol ≤ 34

70 (93.3)

11 (91.7)

1.0

Δmax²⁴⁹≤ 9

18 (24)

3 (25)

1.0

Δmax¹≤ 9

46 (61.3)

4 (33.3)

0.07

Data presented as n (%) or median (interquartile range). Protein and albumin levels were not measured in 15 patients. Cortisol units: 1 μg/dl = 27.6 nmol/l. Cortisol 60, serum cortisol level drawn 60 minutes after stimulus with 1 μg corticotropin; Cortisol 120, serum cortisol level drawn 60 minutes after stimulus with 249 μg corticotropin (120 minutes after serum baseline cortisol); Δmax¹, cortisol 60 value minus the baseline cortisol level; Δmax²⁴⁹, cortisol 120 value minus the baseline cortisol level.

The use of different baseline cortisol cutoff values would have shown different AF frequencies according to different cutoff values if they were considered the only criteria, where the higher the cortisol threshold, the higher the AF incidence. AF defined by the baseline cortisol level was highly variable (22.5–92.2%), which was contrary to that determined by the traditional criteria of cortisol < 20 μg/dl after corticotropin stimulus with a low does or a high dose (Figure 1).

Figure 1

Adrenal failure diagnosis according to different criteria

Adrenal failure diagnosis according to different criteria. The number of patients and the frequency of adrenal failure (AF) according to several baseline cortisol cutoff values (≤10, ≤15, ≤20, ≤25, and ≤34 μg/dl), the classical ambulatory criteria considering serum cortisol ≤ 20 μg/dl after stimulus with corticotropin at low dose (cortisol 60) and at high dose (cortisol 120), and the presence of Δmax¹≤ 9 μg/dl and Δmax²⁴⁹≤ 9 μg/dl. It is observed that the higher the baseline cortisol level, the higher the theoretical diagnosis of AF. The low-dose test classified more patients with AF than the high-dose test when we considered the classical criteria of total cortisol ≤ 20 μg/dl after stimulus. This same pattern was observed for Δmax after low (Δmax¹≤ 9 μg/dl) and high (Δmax²⁴⁹≤ 9 μg/dl) corticotropin doses.

The absolute values of Δmax¹and Δmax²⁴⁹were well correlated (R²= 0.42, P = 0.00001). Δmax¹≤ 9 μg/dl was the best criterion to predict the presence of Δmax²⁴⁹≤ 9 μg/dl, presenting an odds ratio of 2.29 (95% confidence interval, 1.74–3; R = 0.45, P < 0.00001).

Clinical evaluation of adrenal function

The dose of norepinephrine was higher in patients treated with hydrocortisone than in nontreated patients on the day before beginning steroid treatment (0.35 μg/kg/minute versus 0.13 μg/kg/minute, P = 0.03). It was impossible to determine a baseline cortisol value, as well as Δmax¹and Δmax²⁴⁹values, which would best define norepinephrine removal in both the general population and in the group of patients treated with hydrocortisone. Nonetheless, in the nontreated group the area under the receiver operating characteristic curve of baseline cortisol was 0.81, and the cortisol value of 23.6 μg/dl was associated with the highest sensitivity (0.83) and the highest specificity (0.75) for norepinephrine removal during the first five days after the corticotropin test. As a general finding, in the nontreated group we observed that the higher the baseline cortisol, the higher the mean dose of norepinephrine and the lower the probability of norepinephrine weaning (Figure 2).

Figure 2

Different baseline cortisol cutoff values and vasopressor dependence

Different baseline cortisol cutoff values and vasopressor dependence. (a) The mean norepinephrine dose and (b) the probability of norepinephrine removal during the five days after the corticotropin test in patients not treated with hydrocortisone.

Discussion

In the present study we analyzed the diagnosis of AF in a heterogeneous group of patients with septic shock. AF failure was identified more reliably by Δmax²⁴⁹≤ 9 μg/dl than by several baseline cortisol cutoff values since serum albumin levels directly affected the measurement of the serum total cortisol at baseline and after low and high doses of corticotropin, as well as the frequency of different baseline total cortisol cutoff values; however, serum albumin levels did not affect the measurement of Δmax levels and the frequency of Δmax²⁴⁹≤ 9 μg/dl. Additionally, in the group of patients not treated with hydrocortisone, the baseline serum total cortisol level was a prognostic marker for vasopressor dependence – the higher the baseline cortisol, the higher the norepinephrine requirement.

Serum albumin levels influenced cortisol measurement at baseline and after low and high corticotropin doses. We observed that the lower the serum albumin, the higher the incidence of AF when considering the baseline cortisol cutoff value as the diagnostic criterion, especially for the value of 25 μg/dl. In the context of reduced serum albumin levels, therefore, the incidence of AF could be overestimated. On the other hand, this effect was not observed for Δmax¹and Δmax²⁴⁹values, and AF diagnosis regarding Δmax²⁴⁹≤ 9 μg/dl was not influenced by serum albumin levels. These findings highlight the importance of measuring the adrenal gland reserve (Δmax) and confirm the results of previous studies that used Δmax²⁴⁹≤ 9 μg/dl as the diagnostic criterion for AF in critically ill patients and hypoproteinemic patients 9101617.

Our data are in accordance with those of Hamrahian and colleagues 20, which also showed significant variation of the baseline total cortisol level, but not of the free cortisol fraction, in patients with near-normal and low serum albumin – suggesting that measurement of free cortisol could be a more reliable method than measurement of total cortisol for diagnosis of AF in critical illness. We were unable to measure the free cortisol level in the present work but we did find, similar to free cortisol, that Δmax²⁴⁹≤ 9 μg/dl was not affected by serum albumin, allowing it to be a diagnostic option in the setting of hypoproteinemia, where baseline total cortisol is susceptible to serum albumin variations and free cortisol measurement is not available.

There are some points to be considered favoring Δmax²⁴⁹≤ 9 μg/dl in comparison with the free cortisol dosage. First, Δmax²⁴⁹≤ 9 μg/dl is easier to measure and produces faster results with lower costs and more widespread availability in the ICU compared with free cortisol measurement. Second, although baseline total cortisol levels may overestimate AF in comparison with the free cortisol fraction dosage, the measurement of free cortisol is still not validated for the prognosis and as a guide for diagnosis and treatment of hypothalamic–pituitary–adrenal axis dysfunction in critically ill septic patients. In a very recent review in the literature about current problems when evaluating hypothalamic–pituitary–adrenal function, Arafah 21 reported significant Δmax ≤ 9 μg/dl variation according to serum albumin. This result was different from that found in our study, and we can speculate that differences in the studied population and the time for performing the corticotropin test might explain the distinct findings.

Several studies have shown the prognostic role of Δmax²⁴⁹≤ 9 μg/dl as it enables the identification of a specific group of patients with low adrenal reserve, also known as 'relative adrenal failure', who could benefit from steroid replacement 59101729. Annane and colleagues 9 demonstrated that patients with septic shock nonresponsive to 250 μg corticotropin (Δmax ≤ 9 μg/dl) had a greater dependence on vasopressor therapy and a higher hospital mortality rate than those responsive to corticotropin stimulation (Δmax > 9 μg/dl), and that a seven day treatment combination of hydrocortisone and fludrocortisone significantly reduced vasopressor dependence and hospital risk of death. The same benefit of steroid treatment, however, was not observed in the responders to the corticotropin test (Δmax > 9 μg/dl). Besides the reliability of Δmax²⁴⁹≤ 9 μg/dl for the diagnosis of adrenal insufficiency, the measurement of Δmax²⁴⁹also enabled the identification of septic shock patients at a high risk of death who should receive hydrocortisone, avoiding excessive use of steroids.

Alongside the problem concerning the direct effect of serum albumin variation on total cortisol levels and consequently AF diagnosis, the definition of AF based on baseline total cortisol levels may be extremely variable according to the threshold established. We found that the higher the cutoff value used, the higher the incidence of adrenal insufficiency. This problem was well explored by Bernard and colleagues 30, who tested different criteria of adrenal insufficiency in a population of patients with recent traumatic brain injury. They showed that the incidence of adrenal insufficiency varied from 22% to 100% according to the diagnostic criteria, which was very similar to our findings. In an attempt to solve the problem, Arafah 21 correlated the levels of baseline serum total cortisol and free cortisol in critically ill patients with low and near-normal serum albumin. The author proposed that in hypoalbuminemic subjects the normal total cortisol level should be ≥9.5 μg/dl at baseline and ≥15.5 μg/dl after the cosyntropin test. These values corresponded to free cortisol concentrations of 1.8 μg/dl and 3.1 μg/dl, respectively, which is considered the normal cortisol range in critical illness by the author. It seems that such an approach is more appropriate and rational than simply trying to establish a fixed diagnostic cutoff value that neglects serum protein variations.

AF was detected in 23 patients (22.5%) when considering a baseline cortisol level ≤ 10 μg/dl or Δmax²⁴⁹≤ 9 μg/dl. Only four of these patients presented both criteria. This is in accordance with the majority of the studies that have reported a wide range of incidence varying from 0 to 77% depending on the studied population and adrenal insufficiency diagnostic criteria 5101113202131323334.

In the more homogeneous group of patients who did not receive hydrocortisone replacement, it was possible to correlate the baseline serum total cortisol levels with vasopressor dependence. Higher norepinephrine doses and greater norepinephrine dependence were observed in patients with higher baseline cortisol levels. The baseline cortisol value of 23.6 μg/dl, which best discriminated individuals who were weaned from norepinephrine, was very close to the value of 23.7 μg/dl described by Marik and Zaloga 13. Other authors have also shown bad prognosis with increased baseline cortisol, correlating the highest cortisol levels with the most severe illness and the highest risk of mortality 51335. Since in our study hydrocortisone therapy was not controlled, patient selection bias might have occurred explaining why it was not possible to determine a value of baseline cortisol, Δmax¹, or Δmax²⁴⁹that could predict norepinephrine removal either in the general population of the study or in the steroid patient group. Additionally, other problems related to the study design, such as the relatively small sample size and the long period of the study, could have contributed to difficult outcome analysis.

Conclusion

Our study showed that the incidence of AF was 22.5% in a heterogeneous group of septic shock patients and that Δmax²⁴⁹≤ 9 μg/dl may be a better option for AF diagnosis whenever free cortisol measurement is not available, since Δmax²⁴⁹≤ 9 μg/dl was not affected by serum albumin while the baseline total cortisol levels and their cutoff values varied directly according to albumin levels. As a potential result, the use of the baseline total cortisol level for AF diagnosis could lead to misleading classification and excessive steroid replacement in the context of hypoproteinemia. In the group of patients without hydrocortisone therapy, we found that a baseline cortisol level ≤ 23.6 μg/dl was the best value to predict norepinephrine removal.

Key messages

• Δmax²⁴⁹≤ 9 μg/dl may be a better option for AF diagnosis whenever free cortisol measurement is not available, since Δmax²⁴⁹≤ 9 μg/dl was not affected by serum albumin while the baseline total cortisol level varied directly according to albumin levels.

• The use of baseline total cortisol for AF diagnosis could lead to misleading classification and excessive steroid replacement in the context of hypoproteinemia.

• In the group of patients without hydrocortisone treatment, a baseline serum total cortisol level higher than 23.6 μg/dl was correlated with greater vasopressor dependence.

Abbreviations

AF = adrenal failure; Δmax = difference between the highest serum cortisol value after corticotropin stimulus and the baseline cortisol value; Δmax¹= difference between serum cortisol value after low (1 μg) corticotropin dose and the baseline cortisol value; Δmax²⁴⁹= difference between serum cortisol value after high (249 μg) corticotropin dose and the baseline cortisol value; ICU = intensive care unit.

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

DRS conceived the design of the study, was responsible for collection of data, participated in the statistical analysis and interpretation of data, and assisted in drafting the manuscript. JCRV participated in the conception and design of the study and helped to draft the manuscript, revising it critically for important intellectual content. JRR participated in the conception and design of the study, and in the statistical analysis, and helped to draft the manuscript, revising it critically for important intellectual content.

Acknowledgements

The authors are grateful to Dr Eliézer Silva and Moira Schottler for their valuable assistance in editing the manuscript. The article-processing charge was covered by institutional funds.

Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care Angus DC Linde-Zwirble WT Lidicker J Clermont G Carcillo J Pinsky MR Crit Care Med 2001 29 1303 1310 10.1097/00003246-200107000-00002 11445675 The epidemiology of sepsis in the United States from 1979 through 2000 Martin GS Mannino DM Eaton S Moss M N Engl J Med 2003 348 1546 1554 10.1056/NEJMoa022139 12700374 Long-term survival and function after suspected gram-negative sepsis Perl TM Dvorak L Hwang T Wenzel RP JAMA 1995 274 338 345 10.1001/jama.274.4.338 7609265 Long-term health-related quality of life in survivors of sepsis: Short Form-36: a valid and reliable measure of health-related quality of life Heyland DK Hopman W Coo H Tranmer J McColl MA Crit Care Med 2000 28 3599 3605 10.1097/00003246-200011000-00006 11098960 A 3-level prognostic classification in septic shock based on cortisol levels and cortisol response to corticotropin Annane D Sebille V Troche G Raphael JC Gajdos P Bellissant E JAMA 2000 283 1038 1045 10.1001/jama.283.8.1038 10697064 Reversal of late septic shock with supraphysiologic doses of hydrocortisone Bollaert PE Charpentier C Levy B Debouverie M Audibert G Larcan A Crit Care Med 1998 26 645 650 10.1097/00003246-199804000-00010 9559600 Stress doses of hydrocortisone reverse hyperdynamic septic shock: a prospective, randomized, double-blind, single-center study Briegel J Forst H Haller M Schelling G Kilger E Kuprat G Hemmer B Hummel T Lenhart A Heyduck M Crit Care Med 1999 27 723 732 10.1097/00003246-199904000-00025 10321661 Hydrocortisone infusion for severe community-acquired pneumonia: a preliminary randomized study Confalonieri M Urbino R Potena A Piattella M Parigi P Puccio G Porta RD Giorgio C Blasi F Umberger R Am J Respir Crit Care Med 2005 171 242 248 10.1164/rccm.200406-808OC 15557131 Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock Annane D Sébille V Charpentier C Bollaert PE François B Korach JM Capellier G Cohen Y Azoulay E Troché G JAMA 2002 288 862 871 10.1001/jama.288.7.862 12186604 Relative adrenal insufficiency in patients with septic shock: comparison of low dose and conventional corticotropin tests Siraux V De Backer D Yalavatti G Mélot C Gervy C Mockel J Vincent JL Crit Care Med 2005 33 2479 2486 10.1097/01.CCM.0000185641.87051.7C 16276169 Adrenal insufficiency in the critically ill: a new look at an old problem Marik PE Zaloga GP Chest 2002 122 1784 1796 10.1378/chest.122.5.1784 12426284 Adrenal insufficiency Oelkers W N Engl J Med 1996 335 1206 1212 10.1056/NEJM199610173351607 8815944 Adrenal insufficiency during septic shock Marik PE Zaloga GP Crit Care Med 2003 31 141 145 10.1097/00003246-200301000-00022 12545007 Adrenal insufficiency in the surgical intensive care unit Barquist E Kirton O J Trauma 1997 42 27 31 9003254 Corticosteroid insufficiency in acutely ill patients Cooper MS Stewart PM N Engl J Med 2003 348 727 734 10.1056/NEJMra020529 12594318 Relative adrenal failure in the ICU: an identifiable problem requiring treatment Gonzalez H Nardi O Annane D Crit Care Clin 2006 22 105 118 16399022 Cortisol response to corticotropin and survival in septic shock Rothwell PM Udwadia ZF Lawler PG Lancet 1991 337 582 583 10.1016/0140-6736(91)91641-7 1671944 Physicians should administer low-dose corticosteroids selectively to septic patients until an ongoing trial is completed [editorial] Luce JM Ann Intern Med 2004 141 70 72 15238374 Binding of cortisol to human albumin and serum: the effect of protein concentration Mueller UW Potter JM Biochem Pharmacol 1981 30 727 733 10.1016/0006-2952(81)90158-1 7247957 Measurements of serum free cortisol in critically ill patients Hamrahian AH Oseni TS Arafah BM N Engl J Med 2004 350 1629 1638 10.1056/NEJMoa020266 15084695 Hypothalamic pituitary adrenal function during critical illness: limitations of current assessment methods Arafah BM J Clin Endocrinol Metab 2006 91 3725 3745 10.1210/jc.2006-0674 16882746 Definitions for sepsis and multiple organ failure, and guidelines for the use of innovative therapies in sepsis American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Crit Care Med 1992 20 864 874 1597042 CDC definitions for nosocomial infections, 1988 Garner JS Jarvis WR Emori TG Horan TC Hughes JM Am J Infect Control 1988 16 128 140 10.1016/0196-6553(88)90053-3 2841893 Surviving sepsis campaign guidelines for management of severe sepsis and septic shock Dellinger P Carlet JM Masur H Gerlach H Calandra T Cohen J Gea-Banacloche J Keh D Marshall JC Parker MM Crit Care Med 2004 32 858 873 10.1097/01.CCM.0000117317.18092.E4 15090974 APACHE II: a severity of disease classification system Knaus WA Draper EA Wagner DP Zimmerman JE Crit Care Med 1985 13 818 829 10.1097/00003246-198510000-00009 3928249 A new simplified acute physiology score (SAP II) based on a European/North American multicenter study Le Gall JR Lemeshow S Saulnier F JAMA 1993 270 2957 2963 10.1001/jama.270.24.2957 8254858 The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine Vincent JL Moreno R Takala J Willatts S De Mendonça A Bruining H Reinhart CK Suter PM Thijs LG Intensive Care Med 1996 22 707 710 8844239 Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: Results of a multicenter, prospective study Vincent JL Mendonça A Cantraine F Moreno R Takala J Suter PM Sprung CL Calardyn F Blecher S Crit Care Med 1998 26 1793 1800 9824069 Low-dose (1 microgram) adrenocorticotropin (ACTH) stimulation as a screening test for impaired hypothalamo–pituitary–adrenal axis function: sensitivity, specificity and accuracy in comparison with the high-dose (250 microgram) test Tordjman K Jaffe A Trostanetsky Y Greenman Y Limor R Stern N Clin Endocrinol (Oxf) 2000 52 633 640 10.1046/j.1365-2265.2000.00984.x 10792344 Incidence of adrenal insufficiency after severe traumatic brain injury varies according to definition used: clinical implications Bernard F Outtrim J Menon DK Matta BF Br J Anaesth 2006 96 72 76 10.1093/bja/aei277 16311283 Plasma cortisol levels in patients with septic shock Schein RMH Sprung CL Marcial E Crit Care Med 1990 18 259 263 10.1097/00003246-199003000-00002 2302948 Adrenal insufficiency occurring during septic shock: incidence, outcome, and relationship to peripheral cytokine levels Soni A Pepper GM Wyrwinski PM Ramirez NE Simon R Pina T Gruenspan H Vaca CE Am J Med 1995 98 266 271 10.1016/S0002-9343(99)80373-8 7872343 A comparison of the adrenocortical response during septic shock and after complete recovery Briegel J Schelling G Haller M Mraz W Forst H Peter K Intensive Care Med 1996 22 894 899 8905423 Adrenal insufficiency in high-risk surgical ICU patients Rivers EP Gaspari M Abi Saad G Mlynarek M Fath J Horst HM Wortsman J Chest 2001 119 889 896 10.1378/chest.119.3.889 11243973 Prediction of outcome in intensive care patients using endocrine parameters Rothwell PM Lawler PG Crit Care Med 1995 23 78 83 10.1097/00003246-199501000-00015 8001391