Acute kidney injury is common, parallels organ dysfunction or failure, and carries appreciable mortality in patients with major burns: a prospective exploratory cohort study

Ringer's acetate was used for fluid resuscitation in volumes according to the Parkland formula (4 mL/kg body weight [BW]/TBSA%) [29‐31], with adjustments for individual variations in hemodynamic variables, aiming at least for a mean arterial pressure of 70 mm Hg and a urine output of 1 mL/kg BW per hour. When fluid volume substitution alone was insufficient to maintain central hemodynamics, adrenergic drugs (dobutamine and norepinephrine) were used. Renal replacement therapy was considered when the plasma creatinine concentration exceeded 300 μmol/L, together with oliguria or anuria.

Excision and grafting operations were done within 24 to 48 hours. Patients who did not seem to have deep burns at primary examination were re-evaluated daily and operated on if full thickness burns (FTBs) were identified. Wounds were covered by autologous grafts when available or, in extensive burns, either by heterologous grafts for temporary cover or (in special cases) by cultured keratinocytes. Operations were repeated when donor sites again became available. Silver sulfadiazine (Flamazine^®; Smith & Nephew, Hull, UK) was applied to both grafted and non-grafted wounds. Infection control was managed in collaboration with the university hospital infection specialists.

Ventilation was pressure-controlled (Siemens 900 C or Siemens 300 A; Siemens, Solna, Sweden) with tidal volumes of 6 to 8 mL/kg BW, a positive end-expiratory pressure of at least 5 cm H₂O, and aiming at low ventilatory plateau pressures (of less than 35 cm H₂O) [32]. Nutrition was provided enterally from day 1, pain was controlled by continuous infusions of opioids, and sedation was carried out by infusion of benzodiazepines.

AKI was classified by a dynamic classification scheme with three levels for acute renal dysfunction: Risk, Injury, and Failure and two clinical outcomes, Loss of kidney function and End-stage kidney disease (RIFLE) [23]. It is based on how the plasma creatinine concentration is increased compared with the baseline value of the individual patient, reduced urinary output, and the need for renal replacement therapy. The earliest available plasma creatinine concentration measurement was used as the baseline. Plasma creatinine concentrations from the first week, and thereafter the highest value weekly, were used for RIFLE classification and assessment of renal recovery.

SOFA score was recorded at admission and at least three times a week. SOFA score is based on the assessment of six organ dimensions: (a) renal: plasma creatinine concentration or urine output, (b) respiratory: arterial partial pressure of oxygen/fraction of inspired oxygen (PaO₂/FiO₂) ratio, (c) cardiovascular: hypotension or need for adrenergic agents, (d) coagulation: platelet count, and (e) hepatic: plasma bilirubin concentration. The neurological part of SOFA (f) was left out because of the difficulties in assessing the Glasgow coma score in sedated patients. Maximum SOFA is the maximum score value from each organ score, regardless of date [27]. For this study, multiple organ failure was defined as 3 to 4 score points in two or more organ dimensions of the SOFA score [26]. Blood samples were drawn at the time of admission and at least three times a week in accordance with the Burn Unit protocol for major burns. Admission values were used to compare baseline values among groups; the worst value of each patient during the first week was used to analyse factors of importance for AKI. The worst overall value was used to analyse factors of importance for mortality, and the worst value of each patient each week was used to calculate the maximum SOFA score and the descriptive figures of the time course. Laboratory variables were analysed by routine methods and data were stored in the countywide database of the laboratory. Sepsis and systemic inflammatory response syndrome (SIRS) were classified according to the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference [33].

All patients were recorded prospectively in the Linköping Burn Unit Database. At admission, extent (TBSA%) and depth (FTB%) of injury were recorded together with patient characteristics such as age and gender [34]. Daily recordings of care and treatment included variables such as requirement for dialysis, mechanical ventilation, and adrenergic agents. Data regarding the giving of nephrotoxic antibiotics (vancomycin, aminoglycosides, and amphotericin B) and exposure to intravenous contrast (computed tomography [CT] scans) were extracted from medication charts.

Data were analysed with STATISTICA 7 (StatSoft, Inc., Tulsa, OK, USA) and presented as mean and 95% confidence interval (CI). The differences in baseline characteristics and outcome among patients with and without AKI and the differences in mean values between those who developed AKI early and late were analysed using Student t test for continuous data and contingency tables with Pearson chi-square test for categorical variables. Analysis of covariance was used adjusting laboratory data for age and TBSA%. The Tukey unequal N HSD (honest significant difference test for unequal sample sizes) was used as a post hoc test. The difference in progress time was analysed by using Student t test for dependent samples. One-way analysis of variance, for continuous data, and contingency tables with Pearson chi-square test for categorical variables were used to analyse the differences in baseline characteristics and outcome among the patients who developed AKI, grouped in RIFLE classes. Continuous variables were arbitrarily categorised when exploring risk factors for the development of early and late AKI with odds ratios: cutoff age of 60 years, FTB% of 25%, TBSA% of 50%, and reaching the level for Risk within the first 7 days for early AKI.

The incidence of AKI among major burns was 0.11 per 100,000 people per year during the study period. (See selection of patients in Figure 1.) For the majority (14 of 17) of the excluded patients who died within 2 days, active burn care was withheld or withdrawn because of extensive and deep burns, and 8 of the 14 patients were more than 70 years old. They were older (71.1 years, 95% CI 63.8 to 78.5) and had more extensive burns (58.6% TBSA%, 95% CI 46.4% to 70.8%) than the 127 patients in the study group (P <0.001). Three patients had renal failure before active treatment was withdrawn, but no renal replacement therapy was started. Eight patients with superficial burns (25.1% TBSA%, 95% CI 20.3% to 29.9%) that did not require operation and who were inpatients for only a short period (5.3 days, 95% CI 0.68 to 9.82) were also excluded.

A total of 127 patients remained in this study (Table 2), of whom 24% developed AKI (11.8% Risk, 7.9% Injury, and 4.7% Failure) and 3% required renal replacement therapy (Table 3). Overall mortality was 14%. Twenty-nine of the 31 patients who developed AKI had flame burns. One of the two remaining patients had an electrical burn, and one had a chemical hot scald burn (industrial); both were classified as Risk. Seven of the 31 patients (1 classified as Risk, 2 as Injury, and 4 as Failure) had previous histories of hypertension, but none had a documented history of renal dysfunction. One of the patients who was classified as Risk had a previous history of taking lithium. No others had histories of diagnoses affecting the kidney before the burn.

Half of the patients who developed AKI (55%, 17 of 31) reached the level for Risk within the first 7 days, and 81% (25 of 31) within 14 days. The progress time from Risk to maximum RIFLE class was 5.2 days (95% CI 2.0 to 8.5) among the 16 patients who reached Injury and Failure, whereas the time from baseline to Risk was 9.4 days (95% CI 5.9 to 13.0) (P = 0.095). Early AKI was arbitrarily defined as when creatinine reached the level for Risk within the first 7 days, late AKI between days 8 and 60 (Table 4). We found a more than twofold higher risk for younger patients (risk ratio 2.35) and for patients with extensive deep burns (risk ratio 2.25) to develop AKI early (Table 5).

Renal function recovered completely during the time of admission to the burn unit among all patients who survived except for two: the dialysed patient who survived, whose renal function partially recovered, and one patient classified as Injury, who was transferred to another hospital before recovery. Mortality increased with increasing AKI class. The 11 patients having AKI and who died all had multiple organ failure. Among the four patients who required renal replacement therapy, plasma creatinine was within the reference range during the first day after injury in all cases but one, whose plasma creatinine was 126 μmol/L. All four had multiple organ failure before dialysis, and two had sepsis before. The remaining two already had SIRS on admission but no sepsis during their stay. The treatment with dialysis started on days 5 to 19 (10 to 15 days of treatment over 13 to 21 days), and the week-maximum plasma creatinine concentration and plasma urea before starting were 392.0 μmol/L (208.5 to 575.5) and 28.9 mmol/L (15.2 to 42.7), respectively. Two of the patients were oliguric the day before, and the patient with early dialysis (day 5) was oliguric 4 days before. Three of the dialysed patients died.

We found AKI to be common, with an incidence of about a quarter of major burns, which is similar to that reported in a recent paper by Coca and colleagues [19] and slightly less than reported in a letter from Lopes and colleagues [20]. Median time to reaching respective RIFLE class was 10 days in our AKI group, which is the same as that reported by Lopes and colleagues [20]. However, unlike Coca and colleagues [19], we did not find a difference in the time of occurrence between RIFLE classes. The requirement of renal replacement therapy among patients with burns who require intensive care seems not to differ from that of general intensive care units (ICUs). The percentage of renal replacement therapy in our study (3.1%) is close to that reported in ICU patients (Hoste and colleagues [2] 4.1%, Bell and colleagues [36] 2.5%, Uchino and colleagues [4] 4.3%, and Dalfino and colleagues [37] 8.1%) and in most studies of patients with burns (Table 1).

All of the surviving patients in the present study recovered renal function (defined according to RIFLE). This is consistent with findings reported by several others [8, 10, 12, 15]. In a multicentre long-term follow-up of patients in intensive care who had required renal replacement while they were in hospital, 3.4% (34 of 998) of those who survived developed late end-stage kidney disease, as identified from a nationwide register for chronic renal disease [38].

Mortality increased with increasing RIFLE class, and in the studies of Coca and colleagues [19] and Lopes and colleagues [20] mortality rates were 60% and 75%, respectively, in the Failure class, whereas the rate was 5 of 6 in our study. Overall mortality rates were 14% in our study, 13% in the study of Coca and colleagues, and 18% in the study of Lopes and colleagues. ICU mortality among RIFLE-Failure-classified patients seems to be somewhat lower than among burned patients who were classified as Failure. Hoste and colleagues [2] reported 26% mortality in the Failure class from a study of critically ill patients, and Lopes and colleagues [39] found a 55% mortality in the Failure class among patients with sepsis.

The reason for AKI among patients with major burns may be multifactorial. We found that the acute increase in plasma creatinine concentration was preceded by the initial inflammatory response (SIRS) and pulmonary dysfunction. Pulmonary dysfunction after trauma has been suggested to promote pathogenic inflammation and the development of multiple organ failure, including renal failure [40]. In our recent study of acute respiratory dysfunction in patients with major burns, we noted that acute respiratory distress syndrome occurs soon after the burn – usually within 3 days – and that renal dysfunction was more common among the patients with the most severe respiratory dysfunction [32]. This, together with the early onset of organ dysfunction, including renal dysfunction, suggests that it is the burn and resuscitation rather than infective complications that are responsible for the failing organs.

Sepsis occurred in 87% of the AKI group, which is of the same magnitude as reported in previous burn studies [11, 19]. We found that severe sepsis was associated with AKI, even if not all episodes of severe sepsis caused renal dysfunction. In four of six cases in which AKI was of latest onset (days 18 to 60 after burn), it was not preceded by sepsis, contradicting the idea that AKI of late onset was associated mainly with sepsis [15, 17]. Chrysopoulo and colleagues [16] found that AKI among survivors was not the result of sepsis since it preceded sepsis in their study. Another interesting finding is that we found sepsis during the renal recovery period without inducing further renal dysfunction, which has not been previously reported. This finding indicates that at least some of the time-associated episodes of sepsis and renal dysfunction may also be just time-related rather than the result of cause and effect – a possibility that is usually not discussed in studies of burned patients where AKI is considered to be strongly associated with sepsis [10, 12, 15, 17, 19].

We found age, TBSA%, and FTB% to be predisposing factors for AKI but were unable to show the corresponding relation for severity of AKI, most probably because of a lack of power. Coca and colleagues [19] also found older patients in the AKI group, whereas others (for example, Holm and colleagues [11] and Kim and colleagues [17]) found a higher TBSA% in the AKI group, but not advanced age. In the study by Kim and colleagues, mean TBSA% was unusually high in the AKI group (80%) whereas age was relatively young (42 years).

It is important to evaluate the characteristics of patients with burns. Effects are seen on incidence of organ dysfunction and on outcome by the number of patients who have treatment withheld or withdrawn. In different studies, the size of this group has been in the range of 5% to 11% [6, 14]. In a number of studies, no such data are presented [8‐13, 15‐17, 19‐21]. We have excluded all patients who died within the first 48 hours, including cases of initial withholding or withdrawal of treatment. The exclusion criteria of 'short hospital stay' has been used by others [7].

The potential selection bias from excluding the patients with the worst (death within 2 days) and the best (short duration of stay) outcomes has probably influenced the incidence of AKI in this study. The finding that young age is a risk factor for early AKI can also be explained by this selection bias since older patients with extensive burns more often have a lethal outcome.

The fluid resuscitation early after burn is a problem when using the RIFLE criteria and not having a true baseline plasma creatinine concentration taken. The initially low concentrations in plasma, however, should be of the same magnitude among burn patients as a group, reflecting a physiological response to the burn and the fluid resuscitation. Hence, using the RIFLE classification may still be reliable for comparing incidences of acute renal dysfunction between studies of burn patients. The same 'misclassification' problem is, however, likely to occur among other patient groups who are subjected to aggressive fluid resuscitation (ICU patients with major trauma or those with severe sepsis) and whose true baseline may be unknown. Whether the RIFLE should be modified for these circumstances needs to be further examined.