Detection of Critical Illness–Related Corticosteroid Insufficiency Using 1 μg Adrenocorticotropic Hormone Test
Burry, Lisa*; Little, Anjuli†; Hallett, David‡; Mehta, Sangeeta†§
*Department of Pharmacy, Mount Sinai Hospital; †Faculty of Medicine, University of Toronto; ‡Institute of Medical Science, University of Toronto; and §Department of Medicine, Mount Sinai Hospital, Toronto, Ontario, Canada
Received 11 Jun 2012; first review completed 2 Jul 2012; accepted in final form 12 Nov 2012
Address reprint requests to Sangeeta Mehta, MD, FRCPC, University of Toronto, Mount Sinai Hospital, 600 University Ave, Room 18-216, Toronto, Ontario, Canada, M5G 1X5. E-mail: firstname.lastname@example.org.
No funding was received for this study.
None of the authors have any conflicts of interest.
ABSTRACT: Our objectives were to determine the incidence of critical illness–related corticosteroid insufficiency (CIRCI) in patients with septic shock using a 1 μg corticotropin (ACTH) test and to describe their clinical outcomes. We retrospectively identified 219 consecutive patients with septic shock assessed for CIRCI with a 1 μg ACTH test. Standardized testing involved plasma cortisol measurements at baseline (T0) and at 30 min (T30) and 60 min (T60) after ACTH administration. The maximal increase in cortisol (Δ max) was calculated as the difference between T0 and the highest cortisol value at T30 or T60. Critical illness–related corticosteroid insufficiency was defined as Δ max less than 9 μg/dL after ACTH administration. The mean age of the cohort was 63.0 ± 15.8 years, mean Acute Physiology and Chronic Health Evaluation II score was 26.3 ± 8.1, 85.6% were mechanically ventilated, and the mean number of organ failures was 3.0 ± 1.2. Critical illness–related corticosteroid insufficiency was diagnosed in 70.8% of patients. Twenty-eight-day mortality was highest in patients with baseline cortisol greater than 65 μg/dL (62.5%) and in those with baseline cortisol 34 μg/dL or greater and Δ max less than 9 μg/dL (50.0%). There was no difference in mortality in patients with and without CIRCI (53.9% vs. 36.4%, P = 0.08). Corticosteroids were administered to 69.4% of patients for 5.3 ± 3.6 days. For patients with CIRCI, intensive care unit mortality was similar for those who received corticosteroids compared with those who did not (46.0% vs. 25.0%, P = 0.166). The incidence of CIRCI based on 1 μg ACTH was high in this septic shock cohort. The highest mortality rates were observed in patients with high baseline cortisol and in those who failed to respond appropriately to ACTH. The administration of corticosteroids was not associated with a reduction in mortality.
Stimulation of the hypothalamic-pituitary-adrenal axis, resulting in elevated levels of corticotropin and cortisol, is an essential hormonal reaction to major stressors such as severe sepsis (1). Although cortisol concentrations are usually elevated in severe illness, in some cases the levels may be inadequately elevated for the degree of stress, and patients may be unable to respond to additional or protracted stress, resulting in poor clinical outcomes (2).
Critical illness–related corticosteroid insufficiency (CIRCI), defined in the literature as inadequate cellular corticosteroid activity for the severity of the patient’s illness, is the result of either a decrease in adrenal steroid production or tissue resistance to glucocorticoids (1, 3–5). Both low plasma cortisol levels and an impaired response to adrenocorticotropic hormone (ACTH) have been observed in the critically ill population. The incidence of CIRCI is estimated to be between 20% and 60% in patients with severe sepsis or septic shock (3). Studies have reported worse outcomes in patients with CIRCI and suggest that even slight impairment of the adrenal response may be associated with increased mortality (2, 6, 7). The diagnosis of CIRCI and treatment with low-dose corticosteroids remain controversial, with trials in the septic shock population showing mixed results (8–11).
Previous critical care trials evaluating the benefits of corticosteroids used 250 μg ACTH to identify CIRCI (8); however, the administration of 250 μg ACTH results in serum levels that are 100 times greater than normal ACTH stress response, potentially resulting in false-positive cortisol responses in some patients (12–15). Siraux and colleagues (14) published data suggesting that 1 μg ACTH identifies a subgroup of patients with septic shock and CIRCI that would have been missed by the high-dose test. The objectives of this study were to determine the incidence of CIRCI in patients with septic shock following 1 μg ACTH and to describe their clinical outcomes.
MATERIALS AND METHODS
The study was conducted in a 16-bed medical-surgical intensive care unit (ICU) of a university-affiliated hospital. We retrospectively identified adult patients with septic shock who had received a 1 μg ACTH test from the computerized laboratory database. Our ICU adopted a standardized protocol to conduct a 1 μg ACTH test in all adult patients (≥18 years) with septic shock in 2002 following the publication by Annane and colleagues (8). The standardized ACTH testing consisted of plasma cortisol levels drawn at baseline (T0) and at 30 (T30) and 60 (T60) min after intravenous injection of 1 μg ACTH. The ACTH solution was prepared locally in the pharmacy department by dilution of 250 μg ACTH ampule (Cortrosyn; Amphastar Pharmaceuticals Inc, Rancho Cucamonga, Calif) in 250 mL of 0.9% NaCl to obtain a concentration of 1 μg/mL. One-milliliter syringes were used to withdraw 1-mL aliquots of this solution, and syringes were refrigerated at 2°C to 8°C. Total plasma cortisol was measured using the Roche Modular Analyser (Hoffmann–La Roche, Basel, Switzerland; total imprecision 3%–4%). If patients had more than one ACTH stimulation test performed for septic shock during their hospital admission, only the first test was included in the analysis. The Institutional Research Ethics Board approved the study protocol and waived the need for informed consent.
Septic shock was defined as (a) two or more systemic inflammatory response syndrome criteria (respiratory rate >20 or PaCO2 <32 mmHg, heart rate >90 beats/min, temperature ≤36°C or ≥38°C, white blood cell count ≥12,000/μL or ≤4,000 cells/μL or >10% immature neutrophils); (b) documented or clinical suspicion of infection; (c) persistent hypotension despite adequate fluid replacement; and (d) need for vasopressor therapy (16). Patients were excluded if they had known disease of the hypothalamic-pituitary-adrenal axis or had received corticosteroids, ketoconazole, or etomidate in the previous 3 months. Therapy with intravenous hydrocortisone with or without enteral fludrocortisone was recommended by the ICU’s septic shock management guideline; however, therapy was initiated and modified at the discretion of the clinical team.
Our ICU standardized protocol used the definitions published by Annane and colleagues (3, 6–8) to classify and manage corticosteroid therapy in patients with septic shock. These investigators found that high basal plasma cortisol levels and weak cortisol response to ACTH were associated with increased mortality. They identified a basal plasma cortisol level of 34 μg/dL and a Δ cortisol of 9 μg/dL as the best cutoffs to discriminate between survivors and nonsurvivors. Based on the basal cortisol level (≤34 or >34 μg/dL) and the highest value of the cortisol response to ACTH (≤9 or >9 μg/dL), these investigators defined three categories of adrenal function, each associated with different clinical outcomes.
The cortisol response to ACTH was calculated as the difference between the baseline concentration and the highest level at T30 or T60 measurements (Δ max), and CIRCI was defined as a Δ max cortisol less than 9 μg/dL (250 nmol/L) following ACTH. Patients were further categorized by baseline cortisol and Δ max as follows (7):
* responders: baseline cortisol ≤34 μg/dL (938 nmol/L) and Δ max greater than 9 μg/dL (250 nmol/L); this group was identified by Annane and colleagues (7) to have the lowest risk of death;
* nonresponders: baseline cortisol 34 μg/dL (938 nmol/L) or less and Δ max 9 μg/dL (250 nmol/L) or less; and
* high nonresponders: baseline cortisol between 34 μg/dL (938 nmol/L) and 64.9 μg/dL (1,800 nmol/L) and a Δ max 9 μg/dL (250 nmol/L) or less; this group was identified by Annane and colleagues (7) to have the highest risk of death;
* as our laboratory could not measure cortisol concentrations more than 65 μg/dL (1,800 nmol/L), we added another category, high basal cortisol, as Δ max following ACTH administration could not be determined in these patients; and
* absolute adrenal insufficiency was defined as a baseline cortisol less than 10 μg/dL (277.7 nmol/L) (1).
Demographics, Acute Physiology and Chronic Health Evaluation II (APACHE II) scores, physiologic data, the source of sepsis, and the use of specific therapies (e.g. renal replacement therapy, activated protein C, intensive insulin therapy) were recorded for all patients. Standard definitions of organ failure were used (16). Treatment with hydrocortisone and fludrocortisone, durations of mechanical ventilation, vasopressor support, and ICU stay, and ICU mortality were also recorded.
Demographic data were expressed as mean ± SDs, medians with corresponding interquartile ranges, or frequencies with percentages where appropriate. To examine differences between groups in age, APACHE II, number of organ failures, and duration of vasopressors, analysis of variance was used. Differences in duration of mechanical ventilation and ICU length of stay were examined with nonparametric analysis of variance. Fisher exact test was used to evaluate differences in sex and mortality. Receiver operating characteristic (ROC) curves were constructed, and area under the curve (AUC) was examined to assess the ability of baseline cortisol and Δ max to predict mortality. An AUC of 0.5 suggests that baseline cortisol and Δ max are no better than chance at predicting ICU mortality. Hypothesis testing to evaluate if the AUC differed significantly from 0.5 was also done with R Software Environment for Statistical Computing and Graphics version 2.4.0 (Palo Alto, Calif). All statistical tests were two-tailed and considered statistically significant at α < 0.05. The SAS System for Windows version 9.1 (SAS Institute, Inc, Cary, NC) was used for all analyses.
We identified 219 patients who met the definition of septic shock and had received an ACTH test with 1 μg. All were admitted and managed between 2002 and 2006; after this time, our institution discontinued ACTH testing, following the release of the CORTICUS data (17). The clinical characteristics of this cohort are presented in Table 1. Mean APACHE II score was 26.3 ± 8.1, and 85.6% of patients required mechanical ventilation. Primary sites of infection were respiratory (48.8%) and abdominal (25.6%), and mean number of organ failures was 3.0 ± 1.2. Critical illness–related corticosteroid insufficiency was identified (Δ max cortisol <9 μg/dL) in 70.8% of the 219 patients.
Baseline cortisol concentrations
Mean baseline cortisol concentration (T0) was 33.2 ± 14.2 μg/dL; no patient met the definition of absolute adrenal insufficiency. One hundred seventeen patients (53.4%) had baseline cortisol 34 μg/dL or less; in 70 patients (32.0%), baseline cortisol was between 34 and 65 μg/dL, and 32 patients (14.6%) had baseline cortisol greater than 65 μg/dL. The area under the ROC curve was 0.568 (P = 0.9643). A baseline cortisol of 30.8 μg/dL was the best predictor of mortality but was associated with poor sensitivity (0.490) and modest specificity (0.658).
Response to low-dose ACTH stimulation testing
Following 1 μg ACTH, the mean increases in plasma cortisol at T30 and T60 were 4.1 ± 5.4 and 2.5 ± 5.9 μg/dL, respectively. Among the 219 patients, 33 (15.0%) were responders, 93 (42.5%) were nonresponders, 62 (28.3%) were high nonresponders, and 31 patients (14.2%) had baseline cortisol greater than 65 μg/dL, and in those, Δ cortisol could not be measured following ACTH. The area under the ROC curve was 0.561 (P = 0.941). The Δ max of 3.99 μg/dL was identified as the best predictor of mortality but associated with poor sensitivity (0.433) and modest specificity (0.684).
Patient characteristics and outcomes by subgroup are summarized in Table 2. There were no significant differences between those with and without CIRCI in APACHE II score, days of vasopressor therapy, durations of mechanical ventilation, or ICU length of stay. However, compared with all other groups, nonresponders were younger; those with baseline cortisol greater than 65 μg/dL had more organ failures. Overall 28-day mortality was 45.7% for the 219 patients. Twenty-eight-day mortality was highest (62.5%) in patients whose baseline cortisol was greater than 65 μg/dL and 50.0% in those with high basal cortisol who did not respond appropriately to ACTH. Twenty-eight-day mortality for nonresponders was 43.5%, compared with 30.3% in responders, although this difference was not significant (P = 0.084).
Use of corticosteroids
Hydrocortisone 50 mg intravenously administered every 6 h was given to 152 patients (69.4%), and 37.5% also received fludrocortisone 50 μg enterally daily. The mean duration of hydrocortisone treatment was 5.3 ± 3.6 days. There was no significant difference in the proportion of nonsurvivors and survivors who were treated with corticosteroids (72.3% vs. 65.0%, respectively). For nonresponders, ICU mortality was similar in those who received corticosteroids compared with those who did not (46.0% vs. 25.0%, P = 0.166) (Fig. 1). In those who responded appropriately, treatment with corticosteroids was associated with a difference in mortality (7.7% vs. 45.0%, P = 0.0495).
In this retrospective study, we identified 219 patients who met the definition of septic shock. All were admitted and received a 1 μg ACTH test between 2002 and 2006; our institution discontinued ACTH testing in 2006 following the release of the CORTICUS trial results (17). Of our 219 subjects, we identified relative adrenal insufficiency in 70.8% following administration of 1 μg ACTH. Overall 28-day mortality was high in this septic shock population, but we did not detect a statistical difference in mortality based on cortisol level. We did not identify CIRCI to be associated with a significantly longer duration of vasopressor support, only more organ failures. Treatment with corticosteroids was not associated with a mortality reduction in either those who did not respond to ACTH or those who did.
The diagnosis of adrenal insufficiency in the critically ill population is difficult because there are no universally accepted diagnostic criteria. Dynamic testing with ACTH has previously been evaluated in the critical care setting, with some data suggesting the degree of change following ACTH is indicative of outcome and response to corticosteroid therapy (3, 6, 7, 18, 19). The administration of 250 μg ACTH results in serum levels that are 100 times greater than a normal ACTH stress response and may lead to a cortisol response in patients who would not normally generate one under stress (12–15); however, limited data suggest that a dosage of 1 μg ACTH is more sensitive to diagnose adrenal insufficiency (20, 21). Our results with the 1-μg dose are consistent with studies that used the 250 μg ACTH. In a retrospective study of 54 patients, simplified acute physiology score (SAPS) and ICU mortality were higher in nonresponders, and mortality was associated with low Δ cortisol (P = 0.020) and higher SAPS and Sequential Organ Failure Assessment scores (P = 0.015) (22). Similarly, another retrospective trial using the same definitions of CIRCI identified 51.8% of 168 patients as nonresponders (low Δ cortisol with high baseline cortisol), 25% as responders, and 23.2% with low baseline cortisol (23). Sequential Organ Failure Assessment and SAPS scores were significantly higher in the nonresponders, as was ICU mortality. These studies illustrate that patients with septic shock have different clinical characteristics and outcomes depending on the degree of adrenal dysfunction defined as either the baseline cortisol or response to ACTH.
An isolated random plasma cortisol level has limited predictive value for patient outcomes except in the setting of very low and very high values (24). Interpretation of cortisol is complicated by the fact that assays measure total cortisol, which includes free and protein-bound cortisol, and it is free cortisol that is responsible for the physiologic function of the hormone at the cellular level (13). In acute illness, protein-binding capacity is reduced by up to 50%, resulting in an increased percentage of free cortisol, which limits the interpretation of baseline total plasma cortisol (25).
Previous trials evaluating corticosteroid administration in critically ill patients with septic shock have shown mixed results. In a prospective, double-blind trial, 300 patients with early septic shock were randomized to hydrocortisone 50 mg intravenously administered every 6 h plus enteral fludrocortisone 50 μg daily or placebo daily for 7 days immediately following a 250 μg ACTH stimulation test. Among those with CIRCI, corticosteroid treatment was associated with a reduction in mortality compared with placebo (53% vs. 63%; hazard ratio, 0.67; 95% confidence interval, 0.47–0.95; P = 0.02) (8). Several smaller ICU randomized studies showed more ventilator- and hospital-free days and increased short-term survival with a 3- to 7-day course of stress-dose corticosteroids (200–350 mg/d) (26, 27). In contrast, the CORTICUS trial (17), a multicenter, 499-patient randomized, double-blind, placebo-controlled trial, showed no difference in mortality between responders and nonresponders to a 250 μg ACTH stimulation test. Furthermore, corticosteroid therapy did not improve survival or hasten reversal of shock. Interpretation of these results is limited by the enrollment of patients with a lower severity of illness (28). Based on these previous trials, the current Surviving Sepsis recommendations are to provide a course of stress-dose corticosteroids for 7 to 14 days with a slow taper in critically ill patients with septic shock, without ACTH testing (29).
The strengths of our study include (a) the large number of consecutive patients with septic shock over a 3-year period, (b) the use of a standardized protocol for the ACTH stimulation test, and (c) the consistency of all cortisol measurements performed in a single laboratory. This study has limitations: (a) the retrospective study design, (b) the potential for selection bias as corticosteroid therapy was not protocolized, and (c) the lack of free cortisol measurements.
In summary, 70.8% of our septic shock cohort met established criteria for CIRCI following the administration of 1 μg ACTH. These findings are similar to other published studies using 250 μg ACTH. Neither the baseline cortisol nor the change in plasma cortisol following ACTH stimulation testing was a useful predictor of mortality, and steroid treatment was not associated with a mortality benefit.
1. Marik PE, Pastores SM, Annane D, Meduri GU, Sprung CL, Arlt W, Keh D, Briegel J, Beishuizen A, Dimopoulou I, et al.: Recommendations for the diagnosis and management of corticosteroid insufficiency in critically ill adult patients: consensus statements from an international task force by the American College of Critical Care Medicine. Crit Care Med
36: 1937–1949, 2008.
2. Ligtenberg J, Girbes A, Beentjes J, Tulleken J, Werf T, Zijlstra J Hormones in the critically ill patient: to intervene or not to intervene? Intensive Care Med
27: 1567–1577, 2001.
3. Annane D, Maxime V, Ibrahim F, Alvarez JC, Abe E, Boudou P Diagnosis of adrenal insufficiency in severe sepsis and septic shock. Am J Respir Crit Care Med
174: 1319–1326, 2006.
4. Bruno JJ, Hernandez M, Ghosh S, Pravinkumar SE: Critical illness–related corticosteroid insufficiency in cancer patients. Support Care Cancer
20 (6): 1159–1167, 2012.
5. Moraes RB, Friedman G, Tonietto T, Saltz H, Czepielewski MComparison of low and high dose cosyntropin stimulation tests in the diagnosis of adrenal insufficiency in septic shock patients. Horm Metab Res
44 (4): 296–301, 2012.
6. Venkatesh B, Mortimer RH, Couchman B, Hall JEvaluation of random plasma cortisol and the low dose corticotrophin test as indicators of adrenal secretory capacity in critically ill patients: a prospective study. Anaesth Intensive Care
33: 201–209, 2005.
7. Annane D, Sebille V, Troche G, Raphael JC, Gajdos P, Bellissant EA 3-level prognostic classification in septic shock based on cortisol levels and cortisol response to corticotropin. JAMA
283 (8): 1038–1045, 2000.
8. Annane D, Sebille V, Charpentier C, Bollaert PE, Francois B, Korach JM, Capellier G, Cohen Y, Azoulay E, Troche G, et al.: Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA
288: 862–871, 2002.
9. Keh D, Weber-Carstens S, Bohnke T, Schulz C, Pettersson M, Ahlers O, Bercker S, Berg A, Risse G, Nordman M, et al.: Effects of stress-dose hydrocortisone therapy in septic shock: influence on hemodynamic stability and plasma nitrite/nitrate levels. Crit Care
3: 51, 1999.
10. Yildiz O, Doganay M, Aygen B, Güven M, Keleçstimur F, Tutuçs APhysiological-dose of steroid therapy in sepsis. Crit Care
6: 251–258, 2002.
11. Annane D, Bellissant E, Bollaert PE, Briegel J, Confalonieri M, De Gaudio R, Keh D, Kupfer Y, Oppert M, Meduri U Corticosteroids in the treatment of severe sepsis and septic shock in adults. JAMA
301 (22): 2362–2375, 2009.
12. Dorin RI, Qualls CR, Crapo LM: Diagnosis of adrenal insufficiency. Ann Intern Med
139: 194–204, 2003.
13. Moran JL, Chapman MJ, O’Fathartaigh MS, O’Fathertaigh MS, Peisach AR, Pannall PR, Leppard P Hydrocortisolaemia and adrenocortical responsiveness at onset of septic shock. Intensive Care Med
20: 489–495, 1994.
14. Siraux V, De Backer D, Yalavatti G, Melot C, Gervy C, Mockel J, Vincent JL. Relative adrenal insufficiency in patients with septic shock: comparison of low-dose and conventional corticotrophin tests. Crit Care Med
33: 2479–2486, 2005.
15. Mayenknecht J, Diederich S, Bahr V, Plockinger U, Oelkers W Comparison of low and high dose corticotropin stimulation tests in patients with pituitary disease. J Clin Endocrinol Metab
83: 1558–1562, 1998.
16. American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: Definitions for sepsis and multiple organ failure, and guidelines for the use of innovative therapies in sepsis. Crit Care Med
20: 864–874, 1992.
17. Sprung CL, Annane D, Keh D, Moreno R, Singer M, Freivogel K, Weiss YG, Benbenishty J, Kalenka A, Forst H, et al.: Hydrocortisone therapy for patients with septic shock. N Engl J Med
358: 111–124, 2008.
18. Shenker Y, Skatrud BJ: Adrenal insufficiency in critically ill patients. Am J Respir Crit Care Med
163: 1520–1523, 2001.
19. Schein RMH, Sprung CL, Marcial E, Napolitano L, Chernow B: Plasma cortisol levels in patients with septic shock. Crit Care Med
18 (3): 259–263, 1990.
20. Dickstein G, Shechner C, Nicholson WE, Rosner I, Shen-Orr Z, Adawi F, Lahav M Adrenocorticotropin stimulation test: effects of basal cortisol level, time of day, and suggested new sensitive low dose test. J Clin Endocrinol Metab
72: 773–778, 1991.
21. Rasmuson S, Olsson T, Hagg E A low-dose ACTH test to assess the function of the hypothalamic-pituitary-adrenal axis. Clin Endocrinol
44: 151–156, 1996.
22. De Jong MFC, Beishuizen A, Van Schijndel RJMS, Girbes ARJ, Groeneveld ABJ Risk factors and outcomes of changes in adrenal response to ACTH in the course of critical illness. J Intensive Care Med
27 (1): 37–44, 2012.
23. Lim SY, Kwon YS, Park MR, Han SG, Jeon K, Um SW, Koh WJ, Chung MP, Kim H, Kwon OJ, et al.: Prognostic significance of different subgroup classifications of critical illness–related corticosteroid insufficiency in patients with septic shock. Shock
36 (4): 345–349, 2011.
24. Rothwell PM, Lawler PG: Prediction of outcome in intensive care patients using endocrine parameters. Crit Care Med
23: 78–83, 1995.
25. Ho JT, Al-Musalhi H, Chapman MJ, Quach T, Thomas PD, Bagley CJ, Lewis JG, Torpy DJ Septic shock and sepsis: a comparison of total and free plasma cortisol levels. J Clin Endocrinol Metab
91: 105–114, 2006.
26. Confalonieri M, Urbino R, Potena A, Paittella M, Parigi P, Puccio G, Della PR, Giorgio C, Blasi F, Umberger R, et al.: Hydrocortisone infusion for severe community-acquired pneumonia: a preliminary randomized study. Am J Respir Crit Care Med
171: 242–248, 2005.
27. Meduri GU, Golden E, Freire AX, Taylor E, Zaman M, Carson SJ, Gibson M, Umberger R, et al.: Methylprednisolone infusion in patients with early severe ARDS: results of a randomized trial. Chest
131: 954–963, 2007.
28. Marik PE, Pastores SM, Kavanaugh BP: Selection bias negates conclusions from the CORTICUS study? N Engl J Med
358: 2069–2070, 2008.
29. Dellinger RP, Levy MM, Carlet JM, Bion J, Parker MM, Jaeschke R, Reinhart K, Angus DC, Brun-Buisson C, Beale R, et al.: Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med
36 (1): 298–327, 2008.
Septic shock; adrenal insufficiency; ACTH; corticosteroids; cortisol
©2013The Shock Society
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.