- Email: [email protected]
The impact of oxandrolone on length of stay following major burn injury: A clinical practice evaluation
burns 39 (2013) 1374–1379
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/burns
The impact of oxandrolone on length of stay following major burn injury: A clinical practice evaluation Amalia Cochran a,*, Wiley Thuet b, Brennen Holt c, Iris Faraklas c, Randall J. Smout d, Susan D. Horn e a
University of Utah, Department of Surgery, Burn Center at the University of Utah, United States Wake Forest University, Department of Emergency Medicine, United States c University of Utah, Department of Surgery, United States d ISIS/ICOR, United States e ISIS/ICOR, University of Utah, Department of Biomedical Informatics, United States b
article info
abstract
Article history:
Introduction: The anabolic agent oxandrolone (OX) has been found to decrease length of stay
Accepted 1 April 2013
(LOS) following 20–60% total body surface area (TBSA) burn injury. This study uses the Comprehensive Severity Index (CSI) to control for severity of illness and explores the
Keywords:
relationship between OX and LOS in a more broadly selected sample of burn patients
Oxandrolone
and a natural practice setting.
Severity of illness
Methods: A practice-based evidence study was conducted at a single regional burn center.
Clinical practice improvement
Maximum severity of illness (MCSIC) was measured using a burn-specific version of CSI.
Burns
Data on 167 consecutive surviving patients with TBSA 15% were analyzed using case-
Length of stay
control matching for MCSIC, TBSA, and age. Thirty-eight patients received OX. Results: Median patient age for the entire patient sample was 42.7 years. Using a 1:1 match based upon MCSIC, TBSA, then age, mean LOS for patients who received OX was 33.6 days, as opposed to 43.4 days for those who were not managed with OX ( p = 0.03). If patients were matched >1:1 for controls: cases, mean LOS was 40.9 days (controls) versus 31.6 days (cases). Conclusions: OX is associated with shorter LOS after controlling for MCSIC, TBSA, and age. Future comparative effectiveness studies should better define which patients derive the greatest benefits from receipt of OX during their recovery from major burn injury. # 2013 Elsevier Ltd and ISBI. All rights reserved.
1.
Introduction
Severe burns lead to a physiologic hypermetabolic response. This response is associated with increased energy expenditure, increased protein consumption, and ultimately catabolic lean muscle mass wasting and decreased wound healing that may persist for a year or more following major burn injury [1]. Numerous studies have aimed to determine pharmacologic
methods of counteracting this hypermetabolism in adult burn patients. Most notably the use of oxandrolone (OX) has come into favor as an oral synthetic testosterone analog to decrease the catabolic loss of lean body mass and speed up wound healing. In turn, it is expected that these improvements in catabolism and wound healing will result in a decreased length of stay (LOS) associated with severe burns. Two studies have supported this hypothesis, demonstrating that LOS was significantly shorter with the administration of OX in the
* Corresponding author at: 30 North 1900 East, SOM 3B313, Salt Lake City, UT 84132, United States. Tel.: +1 801 581 7508; fax: +1 801 585 6005. E-mail address: [email protected] (A. Cochran). 0305-4179/$36.00 # 2013 Elsevier Ltd and ISBI. All rights reserved. http://dx.doi.org/10.1016/j.burns.2013.04.002
burns 39 (2013) 1374–1379
setting of acute burns [2,3]. More recently it has come into question as to whether OX truly decreases LOS in severely burned patients. In one multicenter study that identified multiple benefits associated with OX use in burn patients including decreased mortality, LOS remained unchanged [4]. Because of the unclear influence of oxandrolone on LOS following major burn injury, our goal was to evaluate its impact in a natural practice setting. The aim of this study was to clarify the relationship between LOS in severely burned patients with the use of OX in the acute stages of burn. We conducted a retrospective study that utilized a practice-based evidence (PBE) research method to control for severity of illness and explore the relationship between oxandrolone and LOS in adults with major burn injury.
2.
Methods
A retrospective practice-based evidence (PBE) study with the goal of identifying process measures associated with shorter LOS following burn injury was conducted at a regional burn center located within an academic health science center. The goal of PBE studies is to optimize both outcomes and costs of care; the formal definition states that a PBE study is ‘‘an analysis of the content and timing of the individual steps of a medical care process to produce superior medical outcomes for the least necessary cost over the continuum of a patient’s care’’ [5]. The design of PBE studies enables the clinician to establish if differences in outcomes are due to differences in care process or differences in severity of illness, and does so in a natural practice environment. This stands in contrast to randomized controlled trials, which often have strict enrollment criteria that are not always consistent with the characteristics of the ‘‘typical’’ patient. The incorporation of Comprehensive Severity Index (CSI) into a PBE study protocol provides a covariate that balances clinical conditions besides those under study. CSI is a physiologic indexing system that is age- and disease-specific and can be used to control for differences in acuity at the individual patient level. CSI has been validated in many care settings over the last 25 years, and has been used in two major general surgery PBE studies and other clinical areas such as stroke, spinal cord, and traumatic brain injury, and nursing homes [6–8]. Most importantly, CSI accounts for the complex interaction between all diseases that are present, their severity, and the relationships between the diseases; use of CSI in this study accounts for differences in TBSA and severity of illness. A burn-specific CSI was used in this analysis as a covariate for balancing the impact of co-morbid and co-occurring conditions on severity of illness in burn patients (Table 1); CSI stratifies patient severity of illness as mild, moderate, severe, or catastrophic. This use of CSI allows for detection of differences that might otherwise be hidden or ‘‘washed out’’ by the effect of overall injury severity. Clinical use of oxandrolone was at practitioner discretion. Patients were considered candidates for oxandrolone if they had total body surface area (TBSA) burn injury 15%. Oxandrolone could be started at any time during the hospital course and was never started prior to completion of initial
1375
burn shock resuscitation. Oxandrolone was only stopped during hospitalization if a patient had a worsening hepatic function profile or if their volume status management became unusually complicated. Using previously described techniques the clinical components of the CSI module data were used to calculate severity scores for each patient [9,10]. Severity scores were captured at three different time windows during patient hospitalization: within 24 h of admission, within 24 h of discharge, and maximum (any time during admission with highest scores). In most medical conditions, maximum CSI (MCSIC) > 60 indicates that the patient is very severely ill, catastrophic, or life threatening. Although the intent of the CSI system as developed was not to explain specific outcomes such as costs, length of stay, or mortality, CSI scores were highly correlated with these outcomes in multiple studies [11–16]. All adult patients with 15% TBSA burn injury admitted as inpatients to a regional ABA-verified burn center between January 1, 2005 and April 30, 2009 were included. Patients meeting inclusion criteria were identified using the center’s ABA/TRACSTM registry. Data collection from chart review included demographics, calculation of the Baux index, and abstraction of data for the burn CSI data module. Trained personnel (WT, BH) abstracted data from the patient’s medical chart after hospital discharge. Accuracy of data abstraction was verified after each data abstractor entered data from five patients.
2.1.
Statistical analysis
The initial plan for data analysis was to use least squares regression to identify patient and treatment variables associated with LOS. However, careful review of those patients who received oxandrolone versus those who did not clearly demonstrated that the treatment was not evenly distributed between these two groups; propensity scoring could not be used because the OX patients had much longer LOS, meaning many non-OX patients with short LOS could not be matched. Because of the bias at our center for treatment of the sickest patients with OX, Student’s t-test was used to compare LOS between treatment groups using MCSIC, TBSA, and age, in that order for manual matching. Coarsened matching was used to select MCSIC within 10, age within 10 years, and TBSA within 10% TBSA. While most matches were 1:1 matches control: case, if more than one control matched a case then the multiple controls were included in the analysis. SAS 9.2 was used for all analyses.
3.
Results
3.1.
Patients demographics and injury characteristics
Between January 1, 2005 and April 30, 2009, 198 patients with 15% TBSA injury were admitted to the regional burn center. Of these patients, 55 had burn size between 15 and 20% TBSA, 93 were between 20 and 40% TBSA, 50 were >40% TBSA, and 8 of these had >60% TBSA. The mortality rate in this patient cohort was 15% (N = 29). Two patients were excluded due to insufficient data in their medical chart; therefore data from
1376
burns 39 (2013) 1374–1379
Table 1 – CSI matrix. Category Cardiovascular
Respiratory
Digestive
Indicator
Level 1
Level 2
Level 3
Pulse characteristics
Bounding peripheral pulses; no absent pulses
Pulses absent or thready in 1 or more limbs
All peripheral pulses thready
All peripheral pulses absent
Lowest MAP
70 mmHg
60–69 mmHg
51–59 mmHg
50 mmHg
Dyspnea
No dyspnea
Dyspnea on exertion; breathing difficulties NOS
Dyspnea at rest
Apnea
PaO2/FiO2 ratio
>300
200–299
100–199
99
Tube feeding tolerance
Full feeds, at goal
50% of goal
Unable to tolerate enteral feeds at any rate for 24 h
Unable to tolerate enteral feeds at any rate for >24 h
Abdominal compartment syndrome
IAP 15 mmHg
IAP 16–20 mmHg
IAP 21–25 mmHg
IAP 25 mmHg
Oliguria with increasing peak airway pressures
No oliguria or elevated peak airway pressures
Nephrology
Neurology
Skin
Level 4
Highest serum creatinine (mg/dL) Urinary output
1.2
1.3–2.5
2.6–3.9
>4
0.5 mL/kg/h
0.5 mL/kg/h for 6 h
<0.5 mL/kg/h for 12 h
Anuria 12 h OR <0.3 mL/kg/h for 12 h
Dysphagia
Dysphagia NOS
Neurological status
Normal swallowing No confusion or unresponsiveness
Percent and depth of burns
No burns present
Unable to swallow liquids or solids Chronic confusion (dementia)
Acute confusion (delirium)
Unresponsive, excluding medication-induced
Burn wound with cellulitis or infection OR <10% 3rd degree OR 10–20% 2nd degree
Burns of face, hands, feet, and/or perineum OR 10–20% 3rd degree burns OR 21–49% 2nd degree OR Compartment syndrome in 1 extermity
50% 2nd degree OR >21% 3rd degree OR Visible subcutaneous tissue or muscle and/or bone at burn site OR Compartment syndrome in >1 extremity OR Thoracic compartment syndrome
Burns <10% 2nd degree Lab-arterial blood gases
Lab-chemistry
Vitals
Highest pH
7.45
7.46–7.50
7.51–7.60
7.61
Lowest pO2 Lowest pH Highest carboxyhemoglobin
61 mmHg 7.35 10%
7.25–7.34 11–20%
51–60 mmHg 7.10–7.24 21–30%
50 mmHg 7.09 31%
Highest lactate (mMol/L) Base deficit (mMol/L)
2.0
2.1–3.5
2.0
3.6–5.0 or 3 serial measurements 2.1 3.6 to 7.0 or 3 serial measurements 3.5
>5.0 or 3 serial measurements 3.6 7.1 or 3 serial measurements 7.0
Lowest pulse rate Highest temperature
51 BPM 38.5 8C
31–40 BPM 39.0–39.9 8C
30 BPM 40 8C
2.1 to
3.5
41–50 BPM 38.5–38.9 8C
1377
burns 39 (2013) 1374–1379
Table 1 (Continued ) Indicator
Category
Level 1
Lowest temperature Highest pulse rate EKG rhythm
Level 2 36.0–36.4 8C 100–129 BPM Bigeminy OR Trigeminy OR Quadrigeminy OR Atrial fibrillation OR PACs/PVCs NOS
36.5 8C 99 BPM EKG ectopy NOS OR No EKG ectopy OR Non-sustained ventricular tachycardia
167 surviving patients was used for this analysis of factors impacting length of stay. The median age of included patients was 42.7 years (IQR 29.3–51.7 years). Seventy-seven percent of the sample was male and the Median TBSA injury was 23% (IQR 18–35%). Median LOS was 21 days (IQR 14–41 days); when standardized for TBSA injury, median LOS/TBSA was 0.89 days/%TBSA (IQR 0.64–1.33 days/%TBSA). Severity of illness as measured by CSI was calculated within 24 h of admission, maximum during hospital stay, and within 24 h of discharge. Median admission CSI was 35 (IQR 17–64). MCSIC had a median of 72 (IQR 29–121). Median CSI at discharge was 13 (IQR 9–17).
3.2.
Factors influencing LOS
The initial sample of 167 patients yielded 22 patients in each group with 1:1 matching. Characteristics for this group are shown in Table 2. With 1:1 matching, the mean MCSIC was 111 for cases and 114 controls (t-test, p = 0.81). Mean TBSA was 32 for each group (t-test, p = 0.94). LOS differed significantly between the two groups, with a mean LOS for OX patients of 33.6 days and mean LOS for untreated patients of 43.4 days (ttest, p = 0.03). With >1:1 matching of controls to cases, 29 patients were included in each arm of the analysis; characteristics of patients included in this analysis are shown in Table 3. Findings in this analysis did not change significantly; mean MCSIC was 106 for cases and 108 for controls (t-test, p = 0.87), mean TBSA was 32 for each group (t-test, p = 0.96), and the mean LOS was 31.6 days for OX patients and 40.9 days for controls (t-test, p = 0.03).
4.
Discussion
Increased energy expenditure, hypermetabolism, delayed wound healing, insulin resistance, and a severe lean tissue catabolism are substantial problems in severe burns and these physiologic processes are well documented in the literature
Level 3
Level 4
35.5–35.9 8C 130 BPM 6 PVCs per minute OR SCT OR Junctional ectopic tachycardia
35.4 8C Runs of ventricular tachycardia
[1,17,18]. OX has been shown to be beneficial to burn patients in order to counteract some of these natural detrimental sequelae. However, it is still not fully understood for which patients OX is most appropriate. We conducted a single burn center retrospective analysis including 167 adult burn patients with TBSA > 15%. We then utilized CSI to standardize for severity of illness, and also matched based upon TBSA and age, known independent risks for mortality. What we found was that in our center there was a selection bias for use of oxandrolone in the sickest patients; however, when matching was used to control for MCSIC, OX patients had significantly shorter LOS than did their control counterparts. These findings both reinforce the benefit in the use of oxandrolone in major burn injury and the relevance of burn-specific CSI as a measure of severity of illness in burns. Prior studies have been done in this area, most notably Wolf et al.’s double-blind randomized controlled trial published in 2006 examining the effects of OX on LOS in patients with 20–60% TBSA burns [3]. In this study the authors found that LOS significantly decreased so long as patients were compared based on burn size. One limitation, however, is that the study was not powered for subgroup analysis to evaluate which patients would benefit most from OX. Our study builds upon the findings of the Wolf study in that we quantified other conditions besides TBSA that can contribute to an increased severity of illness by controlling for CSI. In addition, our inclusion of patients with 15–20% TBSA and high severity of illness, as well as those with >60% TBSA, within our study cohort of patients receiving oxandrolone differs from the inclusion criteria used for the Wolf study. Pham et al. published a cohort study in 2008 looking at the effects of OX use in severely burned patients. In this study OX appeared to be beneficial in terms of survival but it seemed that LOS remained unchanged and the authors ultimately decided that further study was necessary to fully characterize the effects of OX in burn patients [4]. Our study is similar to Pham’s in that we conducted it in a natural practice setting. Again, our findings are complementary, because we were able to identify a significant decrease in LOS when TBSA and severity of illness were controlled.
Table 2 – Characteristics of patients with 1:1 match by MCSIC and TBSA.
Oxandrolone (n = 22) No oxandrolone (n = 22)
Age mean years (SD)
Gender N (% male)
TBSA mean SD
Max CSI mean score (SD)
47.5 (19.3) 45.3 (17.5)
18 (81.8) 11 (72.7)
32.3 (12.9) 32.1 (14.9)
111.1 (42.1) 114.4 (49.2)
1378
burns 39 (2013) 1374–1379
Table 3 – Characteristics of patients with >1:1 match for MCSIC and TBSA.
Oxandrolone (n = 29) No oxandrolone (n = 29)
Age mean years (SD)
Gender N (% male)
TBSA mean SD
Max CSI mean score (SD)
46.7 (18.0) 42.9 (17.2)
25 (86.2) 23 (79.3)
31.5 (14.8) 31.7 (12.6)
106.3 (43.8) 108.3 (49.1)
One novel feature of this study is the demonstration of the utility of CSI as a measure of severity of illness in burns. Unlike traditional measures such as Baux or the Abbreviated Burn Severity Index (ABSI), CSI captures overall illness severity, not simply those driven by the burn injury itself. Secondary analysis compared the use of MCSIC with the use of age and TBSA as a determinant of LOS in our study patient population; this analysis demonstrated that MCSIC was more predictive of LOS than either TBSA or age. The adjusted R [2] for MCSIC explaining LOS was .49 for the >1:1 matched control group. Early analyses with the full dataset demonstrated that use of Baux and ABSI were not as predictive as MCSIC as well, reinforcing the idea that the primary determinant of LOS in acute burn patients is how sick the patient is. Further, this is the first use of CSI in burns, building upon and complementing previous work evaluation CSI as a determinant of LOS [12,14,15,19]. It is important to acknowledge the limitations of our study. Most importantly, the administration of OX was not on a protocol. It was left to individual practitioner discretion as to if and when OX was used. Based upon our attempt at propensity matching, we were able to confirm that the lack of a protocol did result in a selection bias, with oxandrolone use favoring sicker patients with much greater CSI and TBSA. This selection bias was secondarily confirmed by the fact that the LOS for the entire sample of patients was much lower than either the cases or the controls considered as individual groups. The presence of this selection bias was the basis for our decision to analyze using a matched cohort. The second, and most notable, limitation was our study methodology. Although we were able to achieve adequate power for the study, this analysis was conducted at a single burn center, one in which LOS may be profoundly affected by a unique geographic setting. Data from other centers, or from centers that have standardized indications for OX administration, might result in different findings. A final limitation may be present in the methodology used because of the possibility of missing predictors of severity of illness in this model. However, based upon the complexity of the CSI and the conduct of an analysis using more traditional measures of burn severity make this limitation unlikely in our analysis. Our study findings, which are based on actual practice of care, illustrate that OX is associated with a decreased LOS when both MCSIC and TBSA are controlled. These data complement previous studies in this area by identifying those patients who will benefit most and by supporting the idea that OX should be considered when treating patients with major burns who have higher clinical severity. While this study does not provide adequate evidence to set a much-needed standard for OX treatment in severely burned patients, it does indicate an important question within burn care that would benefit from larger scale comparative effectiveness research. In addition, our novel use of the CSI to capture severity of illness
following burn injury merits further exploration in other facets of burn care.
Conflict of interest No extramural support was involved in this work and none of the authors have a conflict of interest to report.
references
[1] Hart DW, Wolf SE, Mlcak R, et al. Persistence of muscle catabolism after severe burn. Surgery 2000;128: 312–9. [2] Jeschke MG, Finnerty CC, Suman OE, Kulp G, Mlcak RP, Herndon DN. The effect of oxandrolone on the endocrinologic, inflammatory, and hypermetabolic responses during the acute phase postburn. Ann Surg 2007;246:351–60. discussion 60-2. [3] Wolf SE, Edelman LS, Kemalyan N, et al. Effects of oxandrolone on outcome measures in the severely burned: a multicenter prospective randomized double-blind trial. J Burn Care Res 2006;27:131–9. discussion 40-1. [4] Pham TN, Klein MB, Gibran NS, et al. Impact of oxandrolone treatment on acute outcomes after severe burn injury. J Burn Care Res 2008;29:902–6. [5] Horn S, Hopkins D. Clinical practice improvement: a new technology for developing cost-effective quality health care. New York: Faulkner and Gray; 1994. [6] Deutscher D, Horn SD, Dickstein R, et al. Associations between treatment processes, patient characteristics, and outcomes in outpatient physical therapy practice. Arch Phys Med Rehabil 2009;90:1349–63. [7] Horn SD, Wright HL, Couperus JJ, et al. Association between patient-controlled analgesia pump use and postoperative surgical site infection in intestinal surgery patients. Surg Infect (Larchmt) 2002;3:109–18. [8] Neumayer LA, Smout RJ, Horn HG, Horn SD. Early and sufficient feeding reduces length of stay and charges in surgical patients. J Surg Res 2001;95:73–7. [9] Horn SD, Horn RA. Reliability and validity of the severity of illness index. Med Care 1986;24:159–78. [10] Horn SD, Horn RA, Sharkey PD, Chambers AF. Severity of illness within DRGs. Homogeneity study. Med Care 1986;24:225–35. [11] Alemi F, Rice J, Hankins R. Predicting in-hospital survival of myocardial infarction. A comparative study of various severity measures. Med Care 1990;28:762–75. [12] Averill RF, McGuire TE, Manning BE, et al. A study of the relationship between severity of illness and hospital cost in New Jersey hospitals. Health Serv Res 1992;27:587–606. discussion 7-12. [13] Horn S, editor. Clinical practice improvement methodology: implementation and evaluation. New York: Faulkner and Gray; 1997. [14] Horn SD, Sharkey PD, Buckle JM, Backofen JE, Averill RF, Horn RA. The relationship between severity of illness and
burns 39 (2013) 1374–1379
hospital length of stay and mortality. Med Care 1991;29:305–17. [15] Iezzoni L. Risk adjustment for measuring health care outcomes. Ann Arbor, MI: Health Administration Press; 1994. [16] Thomas JW, Ashcraft ML. Measuring severity of illness: six severity systems and their ability to explain cost variations. Inquiry 1991;28:39–55.
1379
[17] Newsome TW, Mason Jr AD, Pruitt Jr BA. Weight loss following thermal injury. Ann Surg 1973;178:215–7. [18] Wilmore DW, Aulick LH. Metabolic changes in burned patients. Surg Clin North Am 1978;58:1173–87. [19] Gassaway J, Horn SD, G.D., Smout RJ, Clark C, James R. Applying the clinical practice improvement approach to stroke rehabilitation: methods used and baseline results. Arch Phys Med Rehabil 2005;86:S16–33.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/burns
The impact of oxandrolone on length of stay following major burn injury: A clinical practice evaluation Amalia Cochran a,*, Wiley Thuet b, Brennen Holt c, Iris Faraklas c, Randall J. Smout d, Susan D. Horn e a
University of Utah, Department of Surgery, Burn Center at the University of Utah, United States Wake Forest University, Department of Emergency Medicine, United States c University of Utah, Department of Surgery, United States d ISIS/ICOR, United States e ISIS/ICOR, University of Utah, Department of Biomedical Informatics, United States b
article info
abstract
Article history:
Introduction: The anabolic agent oxandrolone (OX) has been found to decrease length of stay
Accepted 1 April 2013
(LOS) following 20–60% total body surface area (TBSA) burn injury. This study uses the Comprehensive Severity Index (CSI) to control for severity of illness and explores the
Keywords:
relationship between OX and LOS in a more broadly selected sample of burn patients
Oxandrolone
and a natural practice setting.
Severity of illness
Methods: A practice-based evidence study was conducted at a single regional burn center.
Clinical practice improvement
Maximum severity of illness (MCSIC) was measured using a burn-specific version of CSI.
Burns
Data on 167 consecutive surviving patients with TBSA 15% were analyzed using case-
Length of stay
control matching for MCSIC, TBSA, and age. Thirty-eight patients received OX. Results: Median patient age for the entire patient sample was 42.7 years. Using a 1:1 match based upon MCSIC, TBSA, then age, mean LOS for patients who received OX was 33.6 days, as opposed to 43.4 days for those who were not managed with OX ( p = 0.03). If patients were matched >1:1 for controls: cases, mean LOS was 40.9 days (controls) versus 31.6 days (cases). Conclusions: OX is associated with shorter LOS after controlling for MCSIC, TBSA, and age. Future comparative effectiveness studies should better define which patients derive the greatest benefits from receipt of OX during their recovery from major burn injury. # 2013 Elsevier Ltd and ISBI. All rights reserved.
1.
Introduction
Severe burns lead to a physiologic hypermetabolic response. This response is associated with increased energy expenditure, increased protein consumption, and ultimately catabolic lean muscle mass wasting and decreased wound healing that may persist for a year or more following major burn injury [1]. Numerous studies have aimed to determine pharmacologic
methods of counteracting this hypermetabolism in adult burn patients. Most notably the use of oxandrolone (OX) has come into favor as an oral synthetic testosterone analog to decrease the catabolic loss of lean body mass and speed up wound healing. In turn, it is expected that these improvements in catabolism and wound healing will result in a decreased length of stay (LOS) associated with severe burns. Two studies have supported this hypothesis, demonstrating that LOS was significantly shorter with the administration of OX in the
* Corresponding author at: 30 North 1900 East, SOM 3B313, Salt Lake City, UT 84132, United States. Tel.: +1 801 581 7508; fax: +1 801 585 6005. E-mail address: [email protected] (A. Cochran). 0305-4179/$36.00 # 2013 Elsevier Ltd and ISBI. All rights reserved. http://dx.doi.org/10.1016/j.burns.2013.04.002
burns 39 (2013) 1374–1379
setting of acute burns [2,3]. More recently it has come into question as to whether OX truly decreases LOS in severely burned patients. In one multicenter study that identified multiple benefits associated with OX use in burn patients including decreased mortality, LOS remained unchanged [4]. Because of the unclear influence of oxandrolone on LOS following major burn injury, our goal was to evaluate its impact in a natural practice setting. The aim of this study was to clarify the relationship between LOS in severely burned patients with the use of OX in the acute stages of burn. We conducted a retrospective study that utilized a practice-based evidence (PBE) research method to control for severity of illness and explore the relationship between oxandrolone and LOS in adults with major burn injury.
2.
Methods
A retrospective practice-based evidence (PBE) study with the goal of identifying process measures associated with shorter LOS following burn injury was conducted at a regional burn center located within an academic health science center. The goal of PBE studies is to optimize both outcomes and costs of care; the formal definition states that a PBE study is ‘‘an analysis of the content and timing of the individual steps of a medical care process to produce superior medical outcomes for the least necessary cost over the continuum of a patient’s care’’ [5]. The design of PBE studies enables the clinician to establish if differences in outcomes are due to differences in care process or differences in severity of illness, and does so in a natural practice environment. This stands in contrast to randomized controlled trials, which often have strict enrollment criteria that are not always consistent with the characteristics of the ‘‘typical’’ patient. The incorporation of Comprehensive Severity Index (CSI) into a PBE study protocol provides a covariate that balances clinical conditions besides those under study. CSI is a physiologic indexing system that is age- and disease-specific and can be used to control for differences in acuity at the individual patient level. CSI has been validated in many care settings over the last 25 years, and has been used in two major general surgery PBE studies and other clinical areas such as stroke, spinal cord, and traumatic brain injury, and nursing homes [6–8]. Most importantly, CSI accounts for the complex interaction between all diseases that are present, their severity, and the relationships between the diseases; use of CSI in this study accounts for differences in TBSA and severity of illness. A burn-specific CSI was used in this analysis as a covariate for balancing the impact of co-morbid and co-occurring conditions on severity of illness in burn patients (Table 1); CSI stratifies patient severity of illness as mild, moderate, severe, or catastrophic. This use of CSI allows for detection of differences that might otherwise be hidden or ‘‘washed out’’ by the effect of overall injury severity. Clinical use of oxandrolone was at practitioner discretion. Patients were considered candidates for oxandrolone if they had total body surface area (TBSA) burn injury 15%. Oxandrolone could be started at any time during the hospital course and was never started prior to completion of initial
1375
burn shock resuscitation. Oxandrolone was only stopped during hospitalization if a patient had a worsening hepatic function profile or if their volume status management became unusually complicated. Using previously described techniques the clinical components of the CSI module data were used to calculate severity scores for each patient [9,10]. Severity scores were captured at three different time windows during patient hospitalization: within 24 h of admission, within 24 h of discharge, and maximum (any time during admission with highest scores). In most medical conditions, maximum CSI (MCSIC) > 60 indicates that the patient is very severely ill, catastrophic, or life threatening. Although the intent of the CSI system as developed was not to explain specific outcomes such as costs, length of stay, or mortality, CSI scores were highly correlated with these outcomes in multiple studies [11–16]. All adult patients with 15% TBSA burn injury admitted as inpatients to a regional ABA-verified burn center between January 1, 2005 and April 30, 2009 were included. Patients meeting inclusion criteria were identified using the center’s ABA/TRACSTM registry. Data collection from chart review included demographics, calculation of the Baux index, and abstraction of data for the burn CSI data module. Trained personnel (WT, BH) abstracted data from the patient’s medical chart after hospital discharge. Accuracy of data abstraction was verified after each data abstractor entered data from five patients.
2.1.
Statistical analysis
The initial plan for data analysis was to use least squares regression to identify patient and treatment variables associated with LOS. However, careful review of those patients who received oxandrolone versus those who did not clearly demonstrated that the treatment was not evenly distributed between these two groups; propensity scoring could not be used because the OX patients had much longer LOS, meaning many non-OX patients with short LOS could not be matched. Because of the bias at our center for treatment of the sickest patients with OX, Student’s t-test was used to compare LOS between treatment groups using MCSIC, TBSA, and age, in that order for manual matching. Coarsened matching was used to select MCSIC within 10, age within 10 years, and TBSA within 10% TBSA. While most matches were 1:1 matches control: case, if more than one control matched a case then the multiple controls were included in the analysis. SAS 9.2 was used for all analyses.
3.
Results
3.1.
Patients demographics and injury characteristics
Between January 1, 2005 and April 30, 2009, 198 patients with 15% TBSA injury were admitted to the regional burn center. Of these patients, 55 had burn size between 15 and 20% TBSA, 93 were between 20 and 40% TBSA, 50 were >40% TBSA, and 8 of these had >60% TBSA. The mortality rate in this patient cohort was 15% (N = 29). Two patients were excluded due to insufficient data in their medical chart; therefore data from
1376
burns 39 (2013) 1374–1379
Table 1 – CSI matrix. Category Cardiovascular
Respiratory
Digestive
Indicator
Level 1
Level 2
Level 3
Pulse characteristics
Bounding peripheral pulses; no absent pulses
Pulses absent or thready in 1 or more limbs
All peripheral pulses thready
All peripheral pulses absent
Lowest MAP
70 mmHg
60–69 mmHg
51–59 mmHg
50 mmHg
Dyspnea
No dyspnea
Dyspnea on exertion; breathing difficulties NOS
Dyspnea at rest
Apnea
PaO2/FiO2 ratio
>300
200–299
100–199
99
Tube feeding tolerance
Full feeds, at goal
50% of goal
Unable to tolerate enteral feeds at any rate for 24 h
Unable to tolerate enteral feeds at any rate for >24 h
Abdominal compartment syndrome
IAP 15 mmHg
IAP 16–20 mmHg
IAP 21–25 mmHg
IAP 25 mmHg
Oliguria with increasing peak airway pressures
No oliguria or elevated peak airway pressures
Nephrology
Neurology
Skin
Level 4
Highest serum creatinine (mg/dL) Urinary output
1.2
1.3–2.5
2.6–3.9
>4
0.5 mL/kg/h
0.5 mL/kg/h for 6 h
<0.5 mL/kg/h for 12 h
Anuria 12 h OR <0.3 mL/kg/h for 12 h
Dysphagia
Dysphagia NOS
Neurological status
Normal swallowing No confusion or unresponsiveness
Percent and depth of burns
No burns present
Unable to swallow liquids or solids Chronic confusion (dementia)
Acute confusion (delirium)
Unresponsive, excluding medication-induced
Burn wound with cellulitis or infection OR <10% 3rd degree OR 10–20% 2nd degree
Burns of face, hands, feet, and/or perineum OR 10–20% 3rd degree burns OR 21–49% 2nd degree OR Compartment syndrome in 1 extermity
50% 2nd degree OR >21% 3rd degree OR Visible subcutaneous tissue or muscle and/or bone at burn site OR Compartment syndrome in >1 extremity OR Thoracic compartment syndrome
Burns <10% 2nd degree Lab-arterial blood gases
Lab-chemistry
Vitals
Highest pH
7.45
7.46–7.50
7.51–7.60
7.61
Lowest pO2 Lowest pH Highest carboxyhemoglobin
61 mmHg 7.35 10%
7.25–7.34 11–20%
51–60 mmHg 7.10–7.24 21–30%
50 mmHg 7.09 31%
Highest lactate (mMol/L) Base deficit (mMol/L)
2.0
2.1–3.5
2.0
3.6–5.0 or 3 serial measurements 2.1 3.6 to 7.0 or 3 serial measurements 3.5
>5.0 or 3 serial measurements 3.6 7.1 or 3 serial measurements 7.0
Lowest pulse rate Highest temperature
51 BPM 38.5 8C
31–40 BPM 39.0–39.9 8C
30 BPM 40 8C
2.1 to
3.5
41–50 BPM 38.5–38.9 8C
1377
burns 39 (2013) 1374–1379
Table 1 (Continued ) Indicator
Category
Level 1
Lowest temperature Highest pulse rate EKG rhythm
Level 2 36.0–36.4 8C 100–129 BPM Bigeminy OR Trigeminy OR Quadrigeminy OR Atrial fibrillation OR PACs/PVCs NOS
36.5 8C 99 BPM EKG ectopy NOS OR No EKG ectopy OR Non-sustained ventricular tachycardia
167 surviving patients was used for this analysis of factors impacting length of stay. The median age of included patients was 42.7 years (IQR 29.3–51.7 years). Seventy-seven percent of the sample was male and the Median TBSA injury was 23% (IQR 18–35%). Median LOS was 21 days (IQR 14–41 days); when standardized for TBSA injury, median LOS/TBSA was 0.89 days/%TBSA (IQR 0.64–1.33 days/%TBSA). Severity of illness as measured by CSI was calculated within 24 h of admission, maximum during hospital stay, and within 24 h of discharge. Median admission CSI was 35 (IQR 17–64). MCSIC had a median of 72 (IQR 29–121). Median CSI at discharge was 13 (IQR 9–17).
3.2.
Factors influencing LOS
The initial sample of 167 patients yielded 22 patients in each group with 1:1 matching. Characteristics for this group are shown in Table 2. With 1:1 matching, the mean MCSIC was 111 for cases and 114 controls (t-test, p = 0.81). Mean TBSA was 32 for each group (t-test, p = 0.94). LOS differed significantly between the two groups, with a mean LOS for OX patients of 33.6 days and mean LOS for untreated patients of 43.4 days (ttest, p = 0.03). With >1:1 matching of controls to cases, 29 patients were included in each arm of the analysis; characteristics of patients included in this analysis are shown in Table 3. Findings in this analysis did not change significantly; mean MCSIC was 106 for cases and 108 for controls (t-test, p = 0.87), mean TBSA was 32 for each group (t-test, p = 0.96), and the mean LOS was 31.6 days for OX patients and 40.9 days for controls (t-test, p = 0.03).
4.
Discussion
Increased energy expenditure, hypermetabolism, delayed wound healing, insulin resistance, and a severe lean tissue catabolism are substantial problems in severe burns and these physiologic processes are well documented in the literature
Level 3
Level 4
35.5–35.9 8C 130 BPM 6 PVCs per minute OR SCT OR Junctional ectopic tachycardia
35.4 8C Runs of ventricular tachycardia
[1,17,18]. OX has been shown to be beneficial to burn patients in order to counteract some of these natural detrimental sequelae. However, it is still not fully understood for which patients OX is most appropriate. We conducted a single burn center retrospective analysis including 167 adult burn patients with TBSA > 15%. We then utilized CSI to standardize for severity of illness, and also matched based upon TBSA and age, known independent risks for mortality. What we found was that in our center there was a selection bias for use of oxandrolone in the sickest patients; however, when matching was used to control for MCSIC, OX patients had significantly shorter LOS than did their control counterparts. These findings both reinforce the benefit in the use of oxandrolone in major burn injury and the relevance of burn-specific CSI as a measure of severity of illness in burns. Prior studies have been done in this area, most notably Wolf et al.’s double-blind randomized controlled trial published in 2006 examining the effects of OX on LOS in patients with 20–60% TBSA burns [3]. In this study the authors found that LOS significantly decreased so long as patients were compared based on burn size. One limitation, however, is that the study was not powered for subgroup analysis to evaluate which patients would benefit most from OX. Our study builds upon the findings of the Wolf study in that we quantified other conditions besides TBSA that can contribute to an increased severity of illness by controlling for CSI. In addition, our inclusion of patients with 15–20% TBSA and high severity of illness, as well as those with >60% TBSA, within our study cohort of patients receiving oxandrolone differs from the inclusion criteria used for the Wolf study. Pham et al. published a cohort study in 2008 looking at the effects of OX use in severely burned patients. In this study OX appeared to be beneficial in terms of survival but it seemed that LOS remained unchanged and the authors ultimately decided that further study was necessary to fully characterize the effects of OX in burn patients [4]. Our study is similar to Pham’s in that we conducted it in a natural practice setting. Again, our findings are complementary, because we were able to identify a significant decrease in LOS when TBSA and severity of illness were controlled.
Table 2 – Characteristics of patients with 1:1 match by MCSIC and TBSA.
Oxandrolone (n = 22) No oxandrolone (n = 22)
Age mean years (SD)
Gender N (% male)
TBSA mean SD
Max CSI mean score (SD)
47.5 (19.3) 45.3 (17.5)
18 (81.8) 11 (72.7)
32.3 (12.9) 32.1 (14.9)
111.1 (42.1) 114.4 (49.2)
1378
burns 39 (2013) 1374–1379
Table 3 – Characteristics of patients with >1:1 match for MCSIC and TBSA.
Oxandrolone (n = 29) No oxandrolone (n = 29)
Age mean years (SD)
Gender N (% male)
TBSA mean SD
Max CSI mean score (SD)
46.7 (18.0) 42.9 (17.2)
25 (86.2) 23 (79.3)
31.5 (14.8) 31.7 (12.6)
106.3 (43.8) 108.3 (49.1)
One novel feature of this study is the demonstration of the utility of CSI as a measure of severity of illness in burns. Unlike traditional measures such as Baux or the Abbreviated Burn Severity Index (ABSI), CSI captures overall illness severity, not simply those driven by the burn injury itself. Secondary analysis compared the use of MCSIC with the use of age and TBSA as a determinant of LOS in our study patient population; this analysis demonstrated that MCSIC was more predictive of LOS than either TBSA or age. The adjusted R [2] for MCSIC explaining LOS was .49 for the >1:1 matched control group. Early analyses with the full dataset demonstrated that use of Baux and ABSI were not as predictive as MCSIC as well, reinforcing the idea that the primary determinant of LOS in acute burn patients is how sick the patient is. Further, this is the first use of CSI in burns, building upon and complementing previous work evaluation CSI as a determinant of LOS [12,14,15,19]. It is important to acknowledge the limitations of our study. Most importantly, the administration of OX was not on a protocol. It was left to individual practitioner discretion as to if and when OX was used. Based upon our attempt at propensity matching, we were able to confirm that the lack of a protocol did result in a selection bias, with oxandrolone use favoring sicker patients with much greater CSI and TBSA. This selection bias was secondarily confirmed by the fact that the LOS for the entire sample of patients was much lower than either the cases or the controls considered as individual groups. The presence of this selection bias was the basis for our decision to analyze using a matched cohort. The second, and most notable, limitation was our study methodology. Although we were able to achieve adequate power for the study, this analysis was conducted at a single burn center, one in which LOS may be profoundly affected by a unique geographic setting. Data from other centers, or from centers that have standardized indications for OX administration, might result in different findings. A final limitation may be present in the methodology used because of the possibility of missing predictors of severity of illness in this model. However, based upon the complexity of the CSI and the conduct of an analysis using more traditional measures of burn severity make this limitation unlikely in our analysis. Our study findings, which are based on actual practice of care, illustrate that OX is associated with a decreased LOS when both MCSIC and TBSA are controlled. These data complement previous studies in this area by identifying those patients who will benefit most and by supporting the idea that OX should be considered when treating patients with major burns who have higher clinical severity. While this study does not provide adequate evidence to set a much-needed standard for OX treatment in severely burned patients, it does indicate an important question within burn care that would benefit from larger scale comparative effectiveness research. In addition, our novel use of the CSI to capture severity of illness
following burn injury merits further exploration in other facets of burn care.
Conflict of interest No extramural support was involved in this work and none of the authors have a conflict of interest to report.
references
[1] Hart DW, Wolf SE, Mlcak R, et al. Persistence of muscle catabolism after severe burn. Surgery 2000;128: 312–9. [2] Jeschke MG, Finnerty CC, Suman OE, Kulp G, Mlcak RP, Herndon DN. The effect of oxandrolone on the endocrinologic, inflammatory, and hypermetabolic responses during the acute phase postburn. Ann Surg 2007;246:351–60. discussion 60-2. [3] Wolf SE, Edelman LS, Kemalyan N, et al. Effects of oxandrolone on outcome measures in the severely burned: a multicenter prospective randomized double-blind trial. J Burn Care Res 2006;27:131–9. discussion 40-1. [4] Pham TN, Klein MB, Gibran NS, et al. Impact of oxandrolone treatment on acute outcomes after severe burn injury. J Burn Care Res 2008;29:902–6. [5] Horn S, Hopkins D. Clinical practice improvement: a new technology for developing cost-effective quality health care. New York: Faulkner and Gray; 1994. [6] Deutscher D, Horn SD, Dickstein R, et al. Associations between treatment processes, patient characteristics, and outcomes in outpatient physical therapy practice. Arch Phys Med Rehabil 2009;90:1349–63. [7] Horn SD, Wright HL, Couperus JJ, et al. Association between patient-controlled analgesia pump use and postoperative surgical site infection in intestinal surgery patients. Surg Infect (Larchmt) 2002;3:109–18. [8] Neumayer LA, Smout RJ, Horn HG, Horn SD. Early and sufficient feeding reduces length of stay and charges in surgical patients. J Surg Res 2001;95:73–7. [9] Horn SD, Horn RA. Reliability and validity of the severity of illness index. Med Care 1986;24:159–78. [10] Horn SD, Horn RA, Sharkey PD, Chambers AF. Severity of illness within DRGs. Homogeneity study. Med Care 1986;24:225–35. [11] Alemi F, Rice J, Hankins R. Predicting in-hospital survival of myocardial infarction. A comparative study of various severity measures. Med Care 1990;28:762–75. [12] Averill RF, McGuire TE, Manning BE, et al. A study of the relationship between severity of illness and hospital cost in New Jersey hospitals. Health Serv Res 1992;27:587–606. discussion 7-12. [13] Horn S, editor. Clinical practice improvement methodology: implementation and evaluation. New York: Faulkner and Gray; 1997. [14] Horn SD, Sharkey PD, Buckle JM, Backofen JE, Averill RF, Horn RA. The relationship between severity of illness and
burns 39 (2013) 1374–1379
hospital length of stay and mortality. Med Care 1991;29:305–17. [15] Iezzoni L. Risk adjustment for measuring health care outcomes. Ann Arbor, MI: Health Administration Press; 1994. [16] Thomas JW, Ashcraft ML. Measuring severity of illness: six severity systems and their ability to explain cost variations. Inquiry 1991;28:39–55.
1379
[17] Newsome TW, Mason Jr AD, Pruitt Jr BA. Weight loss following thermal injury. Ann Surg 1973;178:215–7. [18] Wilmore DW, Aulick LH. Metabolic changes in burned patients. Surg Clin North Am 1978;58:1173–87. [19] Gassaway J, Horn SD, G.D., Smout RJ, Clark C, James R. Applying the clinical practice improvement approach to stroke rehabilitation: methods used and baseline results. Arch Phys Med Rehabil 2005;86:S16–33.