The use of furosemide in critically ill trauma patients: A retrospective review.

INTRODUCTION: Excessive fluid administration in critically ill post-traumatic patients is common and is associated with poorer outcomes. Once resuscitation is complete; however, assisted diuresis with furosemide is not an option commonly exercised. We hypothesize that diuresis with furosemide in hemodynamically stable, critically ill trauma patients is safe and effective in promoting diuresis.
MATERIALS AND METHODS: In this retrospective chart review, all injured patients admitted to the trauma ICU between March 2007 and June 2009 were identified. Data collection included demographic data, traumatic mechanism, physiologic data, laboratory data, medications, complications, ventilator days, ICU and hospital length of stay. Statistical analyses using two-sample t tests, Wilcoxon rank sum tests, chi-square tests, paired t-tests, and one-sample signed rank tests were performed.
RESULTS: Of 162 screened patients, 85 were identified as eligible. Twenty-seven patients (31.8%) received furosemide within the first 14 ICU days, and there were no significant differences in age, ISS, gender, blunt mechanism, co-morbid conditions, overall complications, or mortality when compared to patients who did not receive diuresis. Furosemide administration resulted in a median of 45% increased 24 h urine output and a median of 82% less 24 h net fluid gain without any significant change in HR, MAP, CVP, Hct, creatinine, or potassium.
CONCLUSIONS: Administration of furosemide in stable, significantly fluid positive critically ill trauma patients results in significantly increased urine output and significantly less net fluid gain with no detrimental effect on hemodynamic parameters or laboratory values.

INTRODUCTION

The initial resuscitation of an unstable trauma patient may require massive amounts of crystalloids, colloids, and blood products to maintain perfusion to critical organs and preserve life. Once hemostasis is achieved, however, additional net fluid gain may be detrimental. Redistribution of crystalloids to non-vascular compartments occurs relatively quickly in the absence of ongoing hemorrhage.[1]

Excessive fluid administration in post-operative and post-traumatic patients is a common occurrence.[2] We are beginning to recognize that excessive fluid administration may not simply be a marker for serious injury or illness, but rather a cause of prolonged mechanical ventilation, prolonged ICU/hospital stay, ileus, anastomotic leak, and possibly mortality. In animal studies, excessive intraoperative fluid administration results in decreased bursting pressure and decreased collagen content at the intestinal anastomosis[3] as well as increased re-bleeding and mortality in hemorrhagic shock models.[4] Studies in healthy human volunteers report detrimental effects on pulmonary function (FVC, FEV1, and peak flow) lasting eight hours after a bolus of Ringer's lactate.[5] Drummer et al. demonstrated that under ideal laboratory conditions, a healthy human requires almost two days to recover from an acute saline infusion.[6]

Judicious fluid administration has been practiced for many years in thoracic surgery and is currently well-accepted.[7] In trauma, however, assisted diuresis with furosemide is an option not universally exercised. Common reservations include concerns about the effect of furosemide on hemodynamic parameters such as heart rate (HR), mean arterial pressure (MAP), and central venous pressure (CVP) or laboratory values such as potassium, blood urea nitrogen (BUN), hematocrit (Hct), or base deficit (BD). Additionally, many surgeons rely upon the patient's own homeostasis to “auto-diurese,” the excess fluid. This auto-diuresis does not always occur and is often insufficient and delayed. In this study, we hypothesize that administration of furosemide in hemodynamically stable, critically ill trauma patients is effective in promoting diuresis and reversing net fluid gain.

MATERIALS AND METHODS

The Human Research Protection Program Committee on Human Research approved this study for expedited review. The need for informed consent was waived as the research involved no more than minimal risk to the subjects. After receiving Institutional Review Board approval, all injured patients admitted to the trauma intensive care unit (ICU) between March 2007 and June 2009 were identified by database. Exclusion criteria included age < 18, pregnancy, burn injury, pre-existing renal insufficiency (defined as baseline creatinine > 1.5 mg/dL), traumatic brain injury, spinal cord injury, delay in presentation > 6 hours, transfer to or from an outside hospital, and ICU stay < 72 hours (including death within the first 72 hours). If the patient was admitted to the ICU multiple times, only the first ICU admission was included in the analysis. Our definition of “adequately resuscitated” required a BD <-6, and no vasopressor requirements.

Using the medical chart, all fluids received by the patient were recorded, including those received in the emergency department, the operating room, and the ICU. Data collection from chart review included demographic data (age, gender, co-morbid medical conditions, traumatic mechanism of injuries (blunt vs. penetrating), injury severity score (ISS), operative vs. non-operative, type of operation); physiologic data (weight, daily fluid balance, cumulative ICU fluid balance, diuretic administration, crystalloids vs. colloids vs. blood components, vital signs); laboratory data; radiologic reports; medications (diuretics, vasopressor agents, antibiotics); complications; ICU days, and hospital length of stay. These were recorded for the first 14 ICU days, or until the patient was discharged from the ICU.

Statistical analysis

Data were summarized using mean ± SD or median with interquartiles, whichever was more appropriate. For the comparisons between those who received furosemide and those who did not, two-sample t-tests or Wilcoxon rank sum tests were used for continuous variables, and Fisher's exact tests were used for categorical variables. For the comparisons between the day of diuresis and the day before, paired t-tests and one-sample signed rank tests were used for continuous variables. The two patients who received diuresis on day 1 were excluded from this analysis since there were no data available on the “prior day.” All analyses were conducted using SAS version 9.3 (SAS Institute, Inc., Cary, NC). Two-sided P < 0.05 was considered as statistically significant.

RESULTS

Of 162 screened patients, 85 were identified by inclusion criteria as eligible. Of these patients, 27 (31.8%) patients received furosemide within the first 14 ICU days. When compared to those patients who did not receive furosemide, those who did had significantly more ventilator days (7 [2-20] vs. 2 [1-5], P = 0.005), ICU length of stay (9 [6-22] vs. 5 [3-8], P < 0.001), and hospital length of stay (28 [10-42] vs. 16.5 [8-23], P = 0.02). [Table 1] There were no significant differences between these two groups in age, ISS, gender, blunt mechanism, co-morbid conditions, ED fluids, OR fluid balance, ICU admission weight, cumulative ICU fluid balance (at ICU discharge or day 14, whichever came first), or mortality. [Tables 1 and 2] Patients who received diuretics had significantly more ventilator-associated pneumonia (30% vs. 3%, P = 0.001); all other complications were comparable between the two cohorts.

Table 1

Demographic characteristics and outcomes

Table 2

Complications

The subsequent data analysis focuses on the 27 patients who received furosemide. The mean age was 42.4 ± 19.6 years (range 19-89), and the mean ISS was 21 ± 10.5 (range 4-41). The cohort was 85.2% men. Blunt injury was the mechanism in 59.3%. The median fluids (crystalloids, colloids, and blood products) received in the ED was 2000 [1000-2500] mL. The median day of first diuresis was ICU day 3 [2-6] [Figure 1], and the median cumulative crystalloid fluid balance on the day of first diuresis was 16,203 [11,638-24,965] mL (range 235 mL-44,880 mL). Patients most commonly received 10 mg as a starting dose. The median cumulative dose of furosemide was 60 mg (range 20-610 mg). Average daily furosemide dose is shown in Figure 2. Compared to the prior day, on the day the patients first received furosemide, median urine output significantly increased by 45% [2-85%] (P = 0.001) and median net fluid gain in 24 h significantly decreased from 2353 mL to 482 mL (P <.0001), a decrease of 83% [Figures 3 and 4]. After initiating furosemide treatment, there was no significant change in HR, MAP, CVP, creatinine, potassium, or PaO2/FiO2 (P/F) ratio [Table 3]. Even though the change in CVPmin, Hgb, Hct, and HCO3 were statistically significantly different from 0, these changes did not reach clinically meaningful magnitude.

Figure 1

Initiation of diuresis after ICU admission

Figure 2

Average Daily Furosemide Dose

Figure 3

Median daily urine output after initiation of diuresis

Figure 4

Net fluid balance after initiation of diuresis

Table 3

Vital signs and laboratory values

CONCLUSIONS

In this study, we demonstrate that administration of furosemide in adequately resuscitated, significantly fluid positive critically ill trauma patients is utilized in less than a third of eligible patients. Comparing those who did to those who did not receive furosemide, there were no significant differences in age, ISS, gender, blunt mechanism, co-morbid conditions, ED IV fluids, OR fluid balance, ICU admission weight, or mortality. Aside from VAP, there was no significant difference between groups regarding complications. It is unlikely that this finding is a result of diuretic administration, and more likely reflects unaccounted confounding factors. Patients who received diuresis had significantly more ventilator days, longer ICU stays, and hospital days. The administration of furosemide was associated with an average of 45% increase in urine output and 83% decrease in net fluid gain. There was no apparent detrimental effect on hemodynamic parameters or laboratory values.

Retrospective studies of ICU patients have shown a significant association between overly excessive fluid gain and mortality.[891011] Additionally, negative cumulative fluid balance has been shown to predict successful extubation in critically ill patients.[121314] Prospective trials in elective general and vascular surgery have shown that post-operative weight gain of as little as 3 kg are associated with delayed recovery of gastrointestinal function, increased complications, and increased hospital stay.[15161718] In a multicenter blinded study of elective colorectal surgery, Brandstrup et al. reported a dose-response relation between complications and increasing volumes of fluids independent of allocation group. The patients in the restricted fluids group had significantly less complications than those randomized to the standard care group (33% vs. 51%, P = 0.002).[19] A large trial conducted by the ARDSnet trialists involving 20 North American centers randomized more than 1,000 patients to a strategy of conservative or liberal fluid administration.[20] The conservative group had improved lung function, more ventilator-free days, and shorter ICU stay; there was, however, no difference in mortality.

Recently, Schnuriger et al. reported an association between increased use of crystalloids after primary colonic anastomosis at initial trauma laparotomy and anastomotic leakage. The authors identified a threshold of 10.5 L of crystalloid infused over the first 72 hours as associated with a 5-fold increased risk of suture line failure.[21]

Many questions remain unanswered. When is it safe to begin assisted diuresis? What is the endpoint? Which patient populations benefit most? What is the best dose and delivery mechanisms (continuous vs. intermittent)? Additionally, the best metric for fluid gain remains to be determined. Past investigators have used absolute volume administered, absolute mass gained, and percent change in body weight.

Our study has several limitations, which must be acknowledged. First and foremost, the single center retrospective design subjects our conclusions to all attendant biases, most importantly, selection bias and confounding by unidentified factors. We were unable to determine pre-injury diuretic use. However, the average age of patients receiving diuresis was 42 years, and the majorities were free of co-morbidities. Only four patients had hypertension, and no patient had congestive heart failure. Therefore, it is reasonable to assume that most patients in this study were naïve to furosemide. Patients who received diuresis had significantly more ventilator days, longer ICU stays, and hospital stays than those that did not receive diuresis, suggesting either a harmful effect of furosemide or, more likely, a selection bias. Additionally, we could not establish the clinical rationale for furosemide administration in these patients using our retrospective study design. Although we chose reasonable parameters to designate a patient as “adequately resuscitated” (base deficit better than -6, no vasopressor requirements), we fully acknowledge that these criteria are somewhat arbitrary and do not fully describe the clinical acumen needed to deem a patient “adequately resuscitated.” Secondly, the small sample size diminishes the statistical power of our conclusions. Lack of statistical significance is likely due to a Type 2 error.

In conclusion, our data suggests that the administration of furosemide in hemodynamically stable, critically ill trauma patients is effective in limiting excessive fluid gain without causing detrimental effects in heart rate, blood pressure, or renal function. This knowledge may assist surgeons when weighing the risks and benefits of diuresis with furosemide. Prospective trials are lacking in trauma patients and are warranted given the growing body of literature suggesting increased complications accompanying overaggressive fluid management.