Fever is part of the febrile response to inflammation or infection which is an adaptive process with physiologic benefit. Although antipyretics are frequently administered to treat fever in the critically ill some clinicians have advocated against this routine practice and advocate for a careful evaluation of the risks and benefits of treating a fever to established need in each case. In light of recent reports of hypotension following acetaminophen administration, the argument against treating a fever in the critically ill may have further justification. This paper describes a case of possible acetaminophen-induced hypotension in a critically-ill patient and discusses the potential association between this relationship.
Although antipyretics are frequently administered to treat fever in the critically ill, many authors have advocated against this routine practice. Instead, they argue, a careful evaluation of the risks and benefits of treating a fever should be established in each case. In light of recent reports of hypotension following acetaminophen administration, the argument against treating a fever in the critically ill may have further justification.
The thermometric definition of fever is arbitrary and depends upon the purpose for which it is defined. For the purpose of warranting diagnostic investigation, the Society of Critical Care Medicine practice parameters define fever in the intensive care unit (ICU) as a temperature greater than 38.3°C (101°F). Although this definition is not conditional on an anatomic site of temperature measurement, measurement of core temperature is recommended using an intravascular or bladder thermistor, electronic probe in the rectum or the external auditory canal.
Physiologically, fever is one component of the febrile response: a complex physiologic reaction to inflammation or infection involving a cytokine mediated rise in core temperature, generation of acute-phase reactants, and activation of numerous physiologic, endocrinologic and immunologic systems. Fever is an adaptive process where the normal body temperature is set at a higher level in response to circulating pyrogens (1-4,9). This is distinct from hyperthermic states in which temperature may rise to greater than 40°C (104 °F). In hyperthermic states the body is unable to control core body temperature. Consequently, these states should be considered as a separate clinical entity requiring different evaluation and management and will not be considered in this discussion.
Fever is part of the adaptive febrile response that is triggered by tissue injury or infection. In-vitro, temperature elevation has been shown to enhance antibody production, T-cell activation, production of cytokines and enhanced neutrophil and macrophage function. Mammalian models have shown enhanced resistance to infection by increasing body temperature. However, in interpreting these data it is important to consider that temperature elevation per se cannot mimic the effects of fever because fever is just one component of the milieu of reactions taking place in the body that are triggered by inflammation or infection. More convincing evidence comes from experimental models of infection (not just elevated temperature) in reptiles or fish. There is a direct correlation between increased body temperature and survival and an increase in mortality if the febrile response is suppressed with sodium salicylate. Marik noted three studies that provide clinical evidence that fever is beneficial. In a retrospective analysis of 218 patients with Gram-negative bacteremia, a positive correlation was found between maximum temperature on the day of bacteremia and survival. A temperature greater than 38°C was associated with increased survival in patients with spontaneous bacterial peritonitis. In children with chickenpox, treatment with acetaminophen was associated with a longer time to crusting of lesions compared to placebo. Viewed in the context of the physiologic role of fever, these clinical observations support the concept that fever may be beneficial. Conversely, interventions to lower body temperature may have detrimental effects.
Others have summarized the potential benefits of treating fever. These include symptomatic relief and the avoidance of increased oxygen demand imposed by the elevated temperature. With respect to symptomatic relief, it is often not clear exactly which symptoms are being treated and to what extent antipyresis makes the patient feel better. An elevated temperature may be well tolerated if it accords with the thermal set point. Profuse diaphoresis brought on by antipyretics may cause even greater discomfort. Nevertheless if fever is associated with symptoms that could prolong ICU stay (e.g. delirium), a therapeutic trial of antipyresis would seem reasonable. Fever increases oxygen demand, cardiac output, sympathetic tone, respiratory minute volume, and energy expenditure, all of which peak during shivering. Therefore it is reasonable to treat fever, provided shivering is not induced, in patients with a limited cardiorespiratory reserve or in those patients in whom fever is suspected to be causing a metabolic or cardiorespiratory deficit.
In many ICU’s, treating a fever above 38.3°C (101°F) with acetaminophen is almost an automatic reaction. However, this practice could be detrimental for two reasons. First, fever has beneficial aspects as outlined above. Second, acetaminophen, like any drug, is associated with potential adverse effects. Perhaps particularly relevant with respect to even periodic dosing is the observed association between acetaminophen and hypotension in critically-ill patients. There are many causes of hypotension in the ICU such as volume depletion, cytokine-related vasodilation, cardiogenic shock, medications, increased intrathoracic pressure secondary to ventilator setting changes and secondary to a cardiac arrhythmia. Recognition of acetaminophen as a possible contributor to hypotension is important for proper assessment and treatment of hypotension in these patients.
A 38 year old man was previously healthy until 8 days prior to hospital admission was admitted for investigation of shortness of breath, left pleuritic chest pain, fever, chills and myalgias. He was on no medications and had no known allergies. On the medical ward he received 5 doses of enteric coated ASA 325 mg for pleuritic pain then received two doses of acetaminophen 650mg for analgesia. He also received a single 2 g intravenous (IV) dose of ceftriaxone. The patient became increasingly short of breath, tachypneic and diaphoretic and was intubated within 24 hours of admission for hypoxic hypercarbic respiratory failure and admitted to the ICU.
On the first day of ICU admission, the patient was ventilated. Pulmonary embolism was ruled out by CT scan and the presumptive diagnosis was community-acquired pneumonia. The patient was started on erythromycin 1g IV and imipenem 500 mg IV each every six hours and dopamine at 2 mcg/kg/min to maintain the mean arterial pressure (MAP) at 70 mmHg.
Acetaminophen 650mg was given NG for oral temperature 39.3°C (CVP 16 cm H20, MAP 70 mmHg). Within 5 minutes his MAP fell to 66 mmHg and dopamine was increased to 5.9 mcg/kg/min to try to maintain MAP at 70 mmHg but the MAP continued to fall to 64 mmHg resulting in the dopamine infusion rate to be increased over next 30 minutes to 11.9 mcg/kg/min. He received pentaspan 500ml which resulted in an increase in MAP to 70 mmHg 60 minutes after the acetaminophen dose. Six hours later the patient received acetaminophen 650 mg by nasogastric (NG) tube for an oral temperature 39.6°C while dopamine was infusing at 7.1 mcg/kg/min to maintain MAP 70 mmHg. Within 10 minutes dopamine was increased to 10.1 mcg/kg/min to maintain MAP at 70 mmHg; he was given 500 ml human serum albumin, and was noted to be diaphoretic and bronchospastic with a temperature of 38.8°C 30 minutes following administration of acetaminophen. Due to tachycardia of 110-120 beats per minute (bpm), dopamine was changed to norepinephrine at 10.7 mcg/min and heart rate returned to 100 bpm baseline. Neither of these episodes of hypotension was associated with any changes on the ventilator or other drug regimens.
On day two in the ICU the ventilator parameters remained the same and a rectal dose of acetaminophen 650 mg was administered for a temperature 39.5°C. Within 30 minutes of acetaminophen administration, norepinephrine increased from 8.5 to 16 mcg/min to maintain MAP at 70 mmHg and 500 ml of pentaspan was administered. Rocuronium 50 mg IV, morphine 3 mg IV and midazolam 3 mg IV were also given 10 minutes prior to the pentaspan for assisting the ventilator. Over the rest of the day norepinephrine infusion rate was steadily increased to 33.3 mcg/min to maintain MAP of 75 mmHg.
On ICU day three, the ventilator parameters remained unchanged and norepinephrine was increased to 36.3 mcg/min to keep the MAP at 80 mmHg for decreasing urine output. With norepinephrine at 37.3 mcg/min, acetaminophen 650 mg was given NG for pyrexia and norepinephrine dose again required an increase to 49.2 mcg/min for reduction in MAP to 64 mmHg. Morphine and midazolam infusion rates remained unchanged throughout at 5 mg/hour each. No other adjustment in medications or ventilator settings were made.
On ICU day four vasopressin was started at an infusion rate of 0.02 units/min and norepinephrine requirements increased to 64 mcg/min. Pentaspan 250 ml was given, hydrocortisone 100 mg IV every eight hours was started and vasopressin and norepinephrine were titrated up to 0.04 units/min and 70.4 mcg/min, respectively. Because of plateau pressures greater than 35cm H20 with FiO2 60% the patient was changed to pressure control ventilation with PEEP at 12 cm H20 and morphine and midazolam infusions were increased to 16 mg/hour each to suppress the drive to breathe. A cooling blanket was applied for temperature of 41.8°C and two doses of rocuronium 50 mg were given for shivering. Hydrocortisone therapy was initiated. Over the next 9 hours norepinephrine requirements decreased to 5.3 mcg/min and temperature fell to 37.2°C. On ICU day five, vasopressin was discontinued within 10 hours. The patient was discharged from ICU after a 17-day stay with a presumed diagnosis of viral illness complicated by pneumonia, acute tubular necrosis and severe peripheral neuropathy.
The hypotension associated with acetaminophen in this patient paralleled the acute phase of his septic shock and his pyrexial state. This hypotension was not reproducible on day four when his vasopressor requirements were decreasing.
This case report illustrates the possible association between acetaminophen and hypotension. Several alternative explanations exist as well. The dramatic reduction in vasopressor requirements from day 4-5 was attributed to starting hydrocortisone and it is possible that it is the temperature per se and not the direct effect of the drug administration that was the cause of the hypotension. The early onset of hypotension further indicates that acetaminophen is not the cause and that the fever itself may be the cause. However, this patient did experience fever spikes that were not associated with hypotension and received doses of acetaminophen that did not appear to lead to hypotension. Although the patient was receiving large doses of furosemide, episodes of hypotension did not appear to be associated with diuresis or changes in ventilator parameters, heart rate or rhythm. The increasing MAP goals may confound the observed increased requirements of norepinephrine. Also the increased and decreased requirements of vasopressors at least in part could reflect the natural time course of septic shock. Nevertheless, there remains a possibility that hypotension could be due to acetaminophen in this case.
Although hypotension has been reported as a manifestation of anaphylaxis to acetaminophen, the first report of isolated hypotension following acetaminophen administration was in 1996 by Brown where hypotensive reactions associated with the administration of acetaminophen elixir in two patients while in the ICU were described. In the first patient acetaminophen was given for analgesia while the patient was treated for hypoxemic respiratory failure and an erythematous rash. The dose and duration of acetaminophen therapy were not provided. Raw data on the time interval between dose administration and changes in MAP and vasopressor requirements were also not provided but were presented graphically. Unfortunately it is difficult to discern a clear relationship between acetaminophen administration and development of hypotension in this case because of the variation in MAP. Mean arterial pressure fluctuations around 13 doses of acetaminiophen were presented as well as rates of norepinephrine, epinephrine and dopamine infusion. However, the author interpreted these data to reflect reproducible hypotension following acetaminophen. The authors ruled out other causes of hypotension such as a continuous infusion of pancuronium and dehydration. After stopping the acetaminophen no further hypotensive episodes were noted. Because of the concomitant rash, it is possible that this patient’s hypotension could have been part of an undiagnosed anaphylactic reaction.
In the second case a 21 year old man admitted to the ICU with respiratory failure following consolidative chemotherapy for amyeloblastic leukemia. Acetaminophen was administered enterally (exact dose not provided) to reduce oxygen demand during neutropenic fever. Mean arterial pressure fell by as much as 20 mmHg within 2 hours of acetaminophen administration. Eight doses of acetaminophen were administered over 126 hours and vasopressors were given from hours 10-20 only. The effect on temperature was not reported and doses and duration of therapy with acetaminophen were not provided. The authors ruled out hypotension due to diuresis, changes in rates of morphine or lorazepam infusions or septic shock.
After observing an apparent temporal association between acetaminophen administration and the development of hypotension in several of their ICU patients, Boyle and co-workers conducted a prospective observational cohort study in 37 patients to examine this potential relationship. The study population consisted of all patients given acetaminophen for the treatment of fever or pain. Systolic, diastolic and MAP were recorded 30 minutes prior to acetaminophen administration, at the time of administration and then at 15 minute intervals for the first hour and at half-hour intervals for the second hour following administration. Any requirements for fluid administration, commencement or increase rates of inotropic or vasoactive drug infusions in the 120 minute observation period following acetaminophen administration were noted. Ten patients were excluded from the final analysis; three for administration of antihypertensive medications in the hour preceding acetaminophen and seven because acetaminophen was administered rectally. The remaining 27 subjects were included in the final analysis. In 23 of the evaluable patients acetaminophen 1 gram was given to treat fever. Although it was not stated whether any patient was in septic shock, 12 patients were receiving vasoactive infusions while seven were receiving regular anti-hypertensive medications. The changes in systolic blood pressure (SBP) were significant at all observation times after acetaminophen administration and at 30, 45, 60 and 90 minutes for the MAP; mean SBP fell from 133 ± 4.7 mmHg at –30 min to 116 ± 3.8 mmHg at 45 minutes. A maximum individual fall in SBP was by 36% at 30 minutes. Mean arterial pressure fell by 7% from baseline (p<0.0001) from a mean MAP of 84 ± 2.9 mmHg at –30 min to 78 ± 2.5 mmHg at 45 min with a maximum individual fall of 34% occurring in the 30 and 45 minute study periods. No changes in ventilator settings or cardiac rhythm occurred during the study period. Fluid and vasopressor therapy was required in 29.6% of patients. Individual hemodynamic data, suspected sources of fever and effects on temperature were not reported. The authors note that due to the need to prevent reduction in MAP, treatment with fluids or vasoactive drugs were instituted. Therefore the full magnitude of the effect of acetaminophen on reduction in SBP and MAP is likely masked. It was not reported whether the patients were re-challenged with acetaminophen to assess reproducibility. Causality is difficult to ascribe to acetaminophen as individual hemodynamic data is not provided. It remains a possibility that, as in our case report, the hypotension observed may be due to cytokine release around a fever spike and not the acetaminophen per se.
Mackenzie et al. carried out a retrospective review of charts of all patients admitted to their ICU for instances of acetaminophen administration over a 14 month period which was recently published in abstract only. One-hundred-and-eighty-three episodes of acetaminophen administration in 53 patients were analyzed. Each patient served as their own control. Results indicated that MAP fell significantly from baseline at 1, 2 and 3 hours following acetaminophen administration (p< 0.001). As this report is only in abstract form, limited data is available. The mean fall in MAP ± SD is not provided nor were important details regarding this study such as detailed patient demographics, the indication for acetaminophen use and other potential causes of hypotension.
Association vs. Causation
Although these observational reports have limitations they are at least hypothesis-generating. Of course, association does not imply causation. It is possible that the hypotension observed in some patients receiving acetaminophen is due to the underlying fever and is unrelated to acetaminophen use. Theoretically, hypotension could also be the result of an interaction between the drug and high temperature, an interaction not observed when the drug is given to euthermic or patients with mild to moderate temperature. It is also possible that a patient is more at risk of acetaminophen-related hypotension if they are in a certain phase of septic shock or systemic inflammatory response syndrome. In the absence of a randomized, placebo-controlled trial we would require three further pieces of evidence to support the hypothesis that acetaminophen causes hypotension. First, we would require further reports to suggest that hypotension is a consistent finding. Second, the association would be strengthened if in a given patient hypotension was not observed with onset of the febrile episode but only when acetaminophen was given to treat a separate febrile episode. These findings by themselves do not rule out that hypotension is a result of a reaction to the excipients of the formulation as considered by Brown. Third, assuming that it is a direct effect of acetaminophen, we require biologic plausibility: what is the mechanism of this adverse effect? Is it host and time dependent? That is, is the risk of hypotension higher when a patient is in a certain phase of their septic inflammatory response syndrome or sepsis? In the case report by Brown, and our report the patients had received acetaminophen prior to the ICU admission without documented hypotension. Also, it is unknown whether the risk of hypotension would be equally likely if aspirin or other non-steroidal anti-inflammatory drugs (NSAIDs) were given. Experimental evidence that antipyretics work by inhibiting centrally acting prostaglandin E2 is equivocal. Inquiry into the underlying mechanism may further reveal the physiology of fever and the mechanism of acetaminophen’s antipyretic action. Although acetaminophen is known to block IL-1, this should oppose vasodilation as IL-1 is a vasodilator. Perhaps there is a link between generation of nitric oxide by acetaminophen, such speculation may require further consideration. Although acetaminophen-related hypotension could be part of a delayed hypersensitivity reaction, other concomitant symptoms such as skin rash and eospinophila were not present in our patient. Although the patient in our case report was bronchospastic this could be attributed to pulmonary edema.
Fever is part of the febrile response to inflammation or infection which is an adaptive process with physiologic benefit. With the known physiologic benefits of fever and potential adverse effects of treatment it should no longer be routine practice to treat fever (1-4). Antipyretics should be initiated as a therapeutic trial only if there is clinical evidence of, or risk for metabolic or cardiorespiratory compromise, or if the fever may be causing symptoms that could prolong ICU stay (e.g. delerium). Ideally, the initiation of a therapeutic trial with acetaminophen should be done without co-intervention so that any observed benefit can be attributed to antipyresis. Whatever the indication, evidence of benefit should be established for treatment to be continued. In the critically ill, acetaminophen administration may be associated with hypotension. Therefore, clinicians should be vigilant for and report hypotensive episodes that may follow administration of acetaminophen and possibly aspirin or other NSAIDs.
I would like to acknowledge the expert observations of Susanne Yard, RN, who identified the temporal association of acetaminophen administration with hypotension in this case.