Original Research—Pediatric Otolaryngology
Outcomes of Isolated Basilar Skull Fracture: Readmission, Meningitis, and Cerebrospinal Fluid Leak
Otolaryngology– Head and Neck Surgery 149(6) 931–939 Ó American Academy of Otolaryngology—Head and Neck Surgery Foundation 2013 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0194599813508539 http://otojournal.org
Brandon A. McCutcheon1, Ryan K. Orosco, MD1, David C. Chang, PhD, MPH, MBA2, Francesca R. Salazar2, Mark A. Talamini, MD2, Stephen Maturo, MD3, and Anthony Magit, MD, MPH1
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
Keywords
Abstract
Received July 15, 2013; revised September 17, 2013; accepted September 20, 2013.
Objective. To determine rates of cerebrospinal fluid (CSF) leak, meningitis, and readmission in pediatric and adult patients with isolated basilar skull fracture. Study Design. Cross-sectional analysis of a statewide database. Subjects. Patients with isolated basilar skull fracture (19952010). Methods. Patients were identified within the California Office of Statewide Health Planning and Development database using ICD-9 diagnosis codes. Results. A total of 3563 pediatric and 10,761 adult patients met inclusion criteria. In-hospital rates of meningitis (0.48% and 0.64%, P = .3360) and CSF leak (2.33% and 1.75%, P = .0270) were similar among children and adults, respectively. Rates of 90-day meningitis (0.17% and 0.37%, P = .0714) and CSF leak (0.40% and 0.40%, P = .9823) were also similar. Thirty-day readmission was 4.6% for children compared with 12.4% for adults (P \ .001). For both pediatric and adult patients, extra-axial hematoma (odds ratio [OR] [confidence interval {CI}] 1.65 [1.05-2.59] and 1.61 [1.34-1.95]) and comorbidities (OR [CI] 2.19 [1.11-4.34] and 1.28 [1.041.59]) were associated with significant increases in 30-day readmission. Loss of consciousness greater than 1 hour (OR, 3.05; 95% CI, 1.53-6.08) and CSF leak (OR, 3.28; 95% CI, 1.41-7.64) increased the likelihood of pediatric readmissions. Lack of insurance (OR, 0.67; 95% CI, 0.50-0.90) and female gender (OR 0.83; 95% CI, 0.70-0.99) reduced the likelihood of adult readmission. Conclusion. Meningitis and CSF leak following isolated basilar skull fractures are uncommon. Readmission within 30 days was more common in adults than in children. CSF leak, hematoma, and prolonged loss of consciousness increased the likelihood of readmission in children. Intracranial injury, male gender, having insurance, and comorbidities increased the likelihood of readmission in adults.
basilar skull fracture, outcomes research, 30-day readmission
R
eadmission is an increasingly important clinical outcome associated with additional costs to health systems and psychosocial burdens to patients. Furthermore, the Centers for Medicare and Medicaid Services (CMS) have been mandated by the Affordable Care Act to penalize health institutions with higher-thanexpected rates of readmission.1 As a result, the Hospital Readmissions Reduction Program was started in October 2012 to leverage penalties against health care institutions with elevated rates of 30-day readmission. These penalties will affect nearly two-thirds of hospitals and total more than $280 million in 2013.2 While penalties for excessive readmission have initially focused on medical conditions such as heart failure, pneumonia, and myocardial infarction, penalties may eventually expand to admissions for traumatic injuries and other diagnoses managed by surgical teams. Unfortunately, there has been little research on the risk factors associated with these readmissions. Nor is it known whether these risk factors differ across pediatric and adult populations.
1 Division of Head and Neck Surgery, University of California San Diego, San Diego, California, USA 2 Department of Surgery, University of California San Diego, San Diego, California, USA 3 Department of Otolaryngology, San Antonio Military Medical Center, Fort Sam Houston, Texas, USA
This article was presented at the 2013 AAO-HNSF Annual Meeting & OTO EXPO; September 29–October 3, 2013; Vancouver, BC, Canada. Corresponding Author: Brandon A. McCutcheon, Division of Head and Neck Surgery, University of California San Diego, 4355 Rialto St #2, San Diego, CA 92107, USA. Email: [email protected]
932 Despite a growing body of knowledge pertaining to pediatric and adult head trauma, risk factors for 30-day readmission in patients with basilar skull fractures have not been reported. This patient population is also at risk for cerebrospinal fluid (CSF) leak and meningitis. Reported rates of meningitis and CSF leak following head trauma vary widely across the literature, are often outdated, and commonly arise from single-institution studies. Many studies are also limited in that they do not monitor patients for the development of complications for an extended period of time after discharge. The objective of this study was to analyze an administrative database to evaluate meningitis, CSF leak, and readmission rates in pediatric and adult patients.
Methods Per the University of California, San Diego, Human Research Protection Program, this project was considered exempt from institutional review board review.
Inclusion and Exclusion Criteria A retrospective analysis was conducted using the California Office of Statewide Health Planning and Development (OSHPD) Patient Discharge database from 1995 to 2010. This database captures 100% of patients in California acutecare hospitals. Patients with closed basilar skull fracture were identified using the following ICD-9 diagnosis codes: 801.0 (fracture of base of skull, closed without mention of intracranial injury), 801.1 (fracture of base of skull, closed with cerebral laceration and contusion), 801.2 (fracture of base of skull, closed with subarachnoid, subdural, and extradural hemorrhage), 801.3 (fracture of base of skull, closed with other unspecified intracranial hemorrhage), 801.4 (fracture of base of skull, closed with intracranial injury of other and unspecified nature). The International Classification of Disease–based Injury Severity Score (ICISS), commonly used in trauma, was employed as a measure of trauma severity to allow for the identification of isolated cases of basilar skull fracture. The ICISS uses survival risk ratios for each ICD-9 diagnosis code to assess the effect of all presenting injuries on the likelihood of survival.3 Admissions for basilar skull fracture were considered to be isolated if the patient’s overall survival risk ratio was within 3% of the individual survival risk ratio attributed to the basilar skull fracture diagnosis code alone. Using this methodology, patients were excluded from analysis if they did not have an isolated skull fracture. Patients were also excluded if their admission was recorded as ‘‘scheduled’’ according to the OSHPD database.
Outcomes The primary outcome of interest was 30-day readmission, which was defined as all-cause readmission to any hospital within California within 30 days following discharge. Determinants of readmission were assessed using multivariate logistic regression analysis. Secondary outcome measures included meningitis and CSF leak during the initial
Otolaryngology–Head and Neck Surgery 149(6) Table 1. ICD-9 diagnosis codes used to identify cases of meningitis. ICD-9
Description
322.0 320.9 321.8 321.2 047.0 047.1 047.8
Nonpyogenic meningitis Meningitis due to unspecified bacterium Meningitis due to other nonbacterial organism Meningitis due to viruses not elsewhere classified Meningitis due to enterovirus, Coxsackie virus Meningitis due to enterovirus, Echo virus Meningitis due to enterovirus, other specified viral meningitis Meningitis due to enterovirus, unspecified viral meningitis Meningitis in other fungal diseases Meningitis in other bacterial diseases classified elsewhere Eosinophilic meningitis Meningitis unspecified Hemophilus meningitis Pneumococcal meningitis Streptococcal meningitis Staphylococcal meningitis Meningococcal meningitis Anaerobic meningitis Meningitis due to gram-negative bacteria, not otherwise classified Meningitis due to other specified bacteria Salmonella meningitis Meningitis due to adenovirus Gonococcal meningitis Candidal meningitis Coccidioidal meningitis Herpes simplex meningitis Cryptococcal meningitis Meningitis in other fungal disease
047.9 321.1 320.7 322.1 322.9 320.0 320.1 320.2 320.3 036.0 320.81 320.82 320.89 003.21 049.1 098.82 112.83 114.2 054.72 321.0 321.1
admission and within 90 days after discharge. The presence of CSF leak was assessed using the ICD-9 diagnosis codes 349.81 (cerebrospinal fluid rhinorrhea) and 388.61 (cerebrospinal fluid otorrhea). Instances of CSF leak requiring repair were identified using the ICD-9 procedure code 02.12 (other repair of cerebral meninges). Instances of meningitis were identified using the ICD-9 codes listed in Table 1.
Patient and Hospital Characteristics Information was provided on patient insurance coverage, race, gender, age, and year of admission. Pediatric patients were defined as those younger than 18 years. Insurance status was defined as ‘‘Medicaid’’ if patients were covered by Medi-Cal, ‘‘workers compensation,’’‘‘county indigent program,’’‘‘other government,’’ or ‘‘other indigent’’ insurance status according to the payer variable in OSHPD. The Charlson Index was calculated to assess patient comorbidities using the method established by Romano and colleagues.4
McCutcheon et al Hospitals were classified as teaching or nonteaching according to whether or not they maintained a general surgical residency program. Given previous studies demonstrating increased risks of mortality associated with loss of consciousness,5,6 this variable was used as a surrogate for severity. Patients were labeled as prolonged loss of consciousness (pLOC) if loss of consciousness was greater than 1 hour or without return to baseline status. Patients were classified as brief loss of consciousness (bLOC) if there was no LOC, or LOC was less than 1 hour. Patients were also labeled with 1 of the following 4 designations according to the degree of intracranial injury indicated on the ICD-9 diagnosis code associated with admission: (1) no intracranial injury; (2) contusion or laceration; (3) subarachnoid, subdural, or extradural hematoma; and (4) unspecified intracranial injury or hematoma.
Statistical Analysis Statistical analysis was performed using commercially available software (STATA SE Version 11.2; Stata Corp LP, College Station, TX). Logistic regression models controlled for patient insurance status, race, gender, hospital teaching status, Charlson Comorbidity Index, age, loss of consciousness, intracranial injury, CSF leak, and year. Statistical significance was established at the 5% significance level, and all confidence intervals (CIs) are reported as 95% CI.
Results Patient and Hospital Characteristics A total of 3563 pediatric and 10,761 adults met inclusion criteria (Table 2). The mean age was 8 years for pediatric patients (median, 8 years) and 47.5 years for adults (median, 44 years). Pediatric patients were predominately male (69.8%) and Hispanic (44.3%). Adult patients were mostly male (71.4%) and white (64%). Most children were covered under the Medicaid distinction (56.5%), while most adults carried private insurance or Medicare (59.9%). Most children and adults did not present with comorbidities (93.9% vs 77.9%, P \ .001). Most pediatric patients presented without intracranial injury (49%), while most adults developed a hematoma (46.8%). The proportion of adult patients presenting with a pLOC was also significantly greater than that observed in children (16.5% vs 5.2%, P \ .001).
30-Day Readmission Unadjusted rates of 30-day readmission were significantly higher in adult patients (12.4% vs 4.6%, P \ .001). Figures 1 and 2 display the unadjusted risk of 30-day readmission in subset populations for pediatric and adult patients, respectively. While CSF leak and pLOC were associated with significant increases in the likelihood of readmission for pediatric patients, these risk factors were not significantly associated with readmission in adults. On multivariate logistic regression, several statistically significant determinants of 30-day readmission were observed among pediatric patients (Table 3). CSF leak on
933 index admission was associated with an increased likelihood of readmission (odds ratio [OR], 3.28; CI, 1.41-7.64). Intracranial or extra-axial hematomas were associated with an increase in the likelihood of readmission (OR, 1.65; CI, 1.05-2.59). pLOC also increased the likelihood of readmission (OR, 3.05; CI, 1.53-6.08). The likelihood of readmission was also positively correlated with the number of patient comorbidities. In adults, physiological factors were less significant in their association with readmission, while social factors appeared to exhibit a greater effect (Table 4). Patients with Medicaid (OR, 0.77; CI, 0.63-0.95) and the uninsured (OR, 0.67; CI, 0.50-0.90) showed significant reductions in the likelihood of readmission. CSF leak and loss of consciousness were not significant determinants of readmission. Intracranial injury was significantly associated with elevations in the likelihood of readmission, with intracranial or extra-axial hematoma exhibiting the strongest effect (OR, 1.61; CI, 1.34-1.95). Increases in comorbidities were associated with an increased likelihood of readmission. Patients with 3 or more comorbid conditions were more than twice as likely as those without comorbidities to experience a readmission (OR, 2.08; CI, 1.56-2.76). Females were less likely than their male counterparts to experience a readmission (OR, 0.83; CI, 0.70-0.99). Patients aged 30 to 60 years were also less likely to be readmitted compared with patients aged 18 to 30 years.
CSF Leak In-hospital CSF leak was an uncommon complication for both pediatric and adult patients, with rates of 2.33% and 1.75% (P = .0270), respectively. CSF leak within 90 days was even more rare, and rates were identical between pediatric and adult patients at 0.4% (P = .9823). The rate of CSF leak appeared stable without significant interyear variation over the course of the 13-year period of observation (P = .105, data not shown).
Meningitis Cases of in-hospital and 90-day meningitis were rare, each occurring in less than 1% of both pediatric and adult patients. When stratified, rates of meningitis were 0.4% among patients without a CSF leak, 1.9% in patients with a CSF leak not requiring repair, and 4.3% in patients requiring repair (P \ .001). Rates of 90-day meningitis were also associated with CSF leak on initial admission. Patients without an initial CSF leak had a 90-day meningitis rate of 0.3%. This rate rose to 1.1% of patients with CSF leak without repair, and there were no instances of postdischarge meningitis in patients with CSF leak requiring repair (P = .041).
Discussion This study represents the largest longitudinal populationbased study with 90-day outcomes of patients with basilar skull fracture. Our study identifies predictors of readmission and complications and demonstrates that these factors differ
934
Otolaryngology–Head and Neck Surgery 149(6)
Table 2. Unadjusted outcomes and characteristics of pediatric and adult patients with basilar skull fracture.a
Outcomes, % Readmission In-hospital CSF leak 90-day CSF leak In-hospital meningitis 90-day meningitis Extent of injury, n (%) No injury to brain Laceration or contusion Hematoma Unspecified hematoma Loss of consciousness, n (%) bLOC (0-1 h) pLOC (.1 h) Insurance status, n (%) Medicare and private Medicaid Uninsured Race, n (%) White Black Hispanic Asian or Pacific Islander Native American/other Gender, n (%) Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, n (%) Mean (median) \1 1-4 5-9 10-17 18-30 30-39 40-49 50-59 60-69 70-79 80-89 901 Hospital setting, n (%) Teaching
Pediatric (n = 3563)
Adults (n = 10,761)
P Value
4.6 2.33 0.40 0.48 0.17
12.4 1.75 0.40 0.64 0.37
\.0001 .0270 .9823 .3360 .0714
1744 (49.0) 356 (10.0) 1110 (31.2) 353 (9.9)
3095 (28.8) 1466 (13.6) 5038 (46.8) 1162 (10.8)
\.001
2871 (94.9) 156 (5.2)
6840 (83.5) 1352 (16.5)
\.001
1385 (39.3) 1992 (56.5) 146 (4.1)
6381 (59.9) 3005 (28.2) 1267 (11.9)
\.001
1325 (37.9) 228 (6.5) 1550 (44.3) 181 (5.2) 214 (6.1)
6797 (64.0) 529 (5.0) 2244 (21.1) 701 (6.6) 345 (3.3)
\.001
1077 (30.2)
3077 (28.6)
.063
3346 (93.9) 146 (4.1) 40 (1.1) 31 (0.9)
8383 (77.9) 1317 (12.2) 513 (4.8) 548 (5.1)
\.001
8 (8) 304 (8.5) 997 (28.0) 744 (20.9) 1518 (42.6) — — — — — — — —
47.5 (44) — — — — 2976 (27.7) 1618 (15.0) 1697 (15.8) 1266 (11.8) 879 (8.2) 1075 (10.0) 1006 (9.4) 244 (2.3)
875 (24.6)
2391 (22.2)
—
.004
Abbreviations: bLOC, brief loss of consciousness; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness. a Bold font represents statistical significance.
across pediatric and adult populations. Children with CSF leak, intracranial hematoma, pLOC, or a comorbid condition were at an increased risk of readmission. In contrast, CSF
leak and loss of consciousness were not significant predictors of readmission for adults. Increases in the risk of readmission were associated with intracranial injuries and
McCutcheon et al
935
Figure 1. Pediatric risk-adjusted rates of 30-day readmission in subsets of patients with complications and comorbidities.
Figure 2. Adult risk-adjusted rates of 30-day readmission in subsets of patients with complications and comorbidities.
comorbidities. Reductions in the risk of readmission were noted among patients with Medicaid, the uninsured, females, and patients between the ages of 30 and 60 years. The finding that pediatric patients with hematoma, CSF leak, or pLOC were associated with increased risks of readmission is clinically intuitive. These patients likely suffered from higher-severity injuries, which increases the risk that they will develop a complication. What remains to be seen is whether the increased likelihood of readmission associated with these risk factors could be blunted by additional time in the hospital or further interventions. It is perhaps surprising, however, that CSF leak and pLOC were not significant predictors of readmission in adults. It is unclear whether this difference between pediatric and adult patients is the result of variation in mechanism of injury, underlying
physiology, patient management, diagnosis, or other unmeasured factors. Comorbid conditions were unsurprisingly associated with increases in readmission across both pediatric and adult populations. Patients with disease involving additional organ systems may have delayed recovery and increased risks of complications following an injury. However, patients with comorbid illness also have a higher baseline likelihood of hospitalization, and it is unclear whether a 30-day readmission reflects an event that would not have happened in the absence of the index hospitalization. Regardless, patients with comorbid illnesses may experience reduced readmission rates if hospitalizations for acute injuries are also seen as opportunities to review and optimize treatment of chronic conditions.
936
Otolaryngology–Head and Neck Surgery 149(6)
Table 3. Multivariate logistic regression analysis demonstrating determinants of 30-day readmission in pediatric patients with basilar skull fracture.a
Length of stay 0-25 percentile 25-50 percentile 50-75 percentile .75 percentile CSF leak No leak Leak Extent of injury No injury to brain Laceration or contusion Hematoma Unspecified intracranial injury Loss of consciousness bLOC (0-1 h) pLOC (.1 h) Insurance status Medicare and private Medicaid Uninsured Race White Black Hispanic Asian or Pacific Islander Native American/other Gender Male Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, y \1 1-4 5-9 10-17 Hospital setting Nonteaching Teaching
Coefficient
Standard Error
Wald x2
OR
95% CI
P Value
[Reference] –0.06 –0.05 –0.07
0.26 0.32 0.27
0.05 0.02 0.06
0.94 0.95 0.93
0.56-1.58 0.51-1.77 0.55-1.58
.828 .878 .809
[Reference] 1.19
0.43
7.62
3.28
1.41-7.64
.006
[Reference] 0.37 0.50 0.06
0.32 0.23 0.35
1.39 4.75 0.03
1.46 1.65 1.07
0.78-2.71 1.05-2.59 0.54-2.12
.239 .030 .856
[Reference] 1.12
0.35
3.05
1.53-6.08
.002
[Reference] –0.01 0.57
0.21 0.41
0.00 1.96
0.98 1.77
0.66-1.48 0.79-3.96
.958 .162
[Reference] 0.19 0.11 -0.29 0.48
0.39 0.22 0.54 0.36
0.24 0.25 0.29 1.82
1.20 1.11 0.75 1.62
0.56-2.60 0.72-1.71 0.26-2.15 0.80-3.28
.626 .615 .591 .177
[Reference] 0.12
0.21
0.33
1.12
0.75-1.70
.560
[Reference] 0.78 1.45 0.77
0.35 0.52 0.67
5.15 7.72 1.32
2.19 4.24 2.16
1.11-4.34 1.53-11.75 0.58-8.09
.023 .005 .250
[Reference] 0.06 0.21 0.67
0.41 0.43 0.40
0.03 0.24 2.82
1.06 1.23 1.96
0.47-2.39 0.53-2.87 0.90-4.31
.884 .626 .092
[Reference] 0.02
0.22
0.01
1.02
0.66-1.59
.914
10.0
Abbreviations: bLOC, brief loss of consciousness; CI, confidence interval; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness; OR, odds ratio. a Bold font represents statistical significance.
The observation that patients with Medicaid and the uninsured were less likely to be readmitted may reflect socioeconomic barriers to care. If transportation, absence from work, or other barriers to care are present, the uninsured may delay or avoid seeking follow-up treatment until a higher threshold of concern is reached, thus readmitting at a later date outside of
the 30-day period assessed in this study. This finding speaks to the potential limitation of using 30-day readmission as a quality indicator or outcome metric. The counterintuitive result that patients with Medicaid and the uninsured are less likely to readmit illustrates that this metric may be influenced by a plethora of patient-level
McCutcheon et al
937
Table 4. Multivariate logistic regression analysis demonstrating determinants of 30-day readmission in adult patients with basilar skull fracture.a
Length of stay 0-25 percentile 25-50 percentile 50-75 percentile .75 percentile CSF leak No leak Leak Extent of injury No injury to brain Laceration or contusion Hematoma Unspecified intracranial injury Loss of consciousness bLOC (0-1 h) pLOC (.1 h) Insurance status Medicare and private Medicaid Uninsured Race White Black Hispanic Asian or Pacific Islander Native American/Other Gender Male Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, y 18-30 30-39 40-49 50-59 60-69 70-79 80-89 901 Hospital setting Nonteaching Teaching
Coefficient
Standard Error
Wald x2
OR
95% CI
P Value
[Reference] –0.08 0.00 0.08
0.12 0.10 0.11
0.42 0.00 0.59
0.92 1.00 1.09
0.72-1.18 0.82-1.22 0.88-1.35
.517 .962 .442
[Reference] 0.24
0.28
0.74
1.28
0.74-2.19
.388
[Reference] 0.32 0.48 0.33
0.13 0.10 0.14
6.50 25.00 5.76
1.39 1.61 1.39
1.08-1.79 1.34-1.95 1.06-1.81
.011 \.001 .016
[Reference] 0.17
0.14
1.59
1.18
0.90-1.55
.209
[Reference] –0.25 –0.39
0.10 0.15
5.90 6.81
0.77 0.67
0.63-0.95 0.50-0.90
.015 .009
[Reference] 0.05 –0.05 –0.06 –0.02
0.19 0.10 0.15 0.23
0.08 0.26 0.15 0.00
1.05 0.95 0.94 0.98
0.72-1.52 0.78-1.16 0.71-1.26 0.63-1.55
.782 .611 .698 .945
[Reference] –0.18
0.09
4.20
0.83
0.70-0.99
.041
[Reference] 0.25 0.48 0.73
0.11 0.15 0.14
5.24 9.98 25.50
1.28 1.61 2.08
1.04-1.59 1.20-2.17 1.56-2.76
.022 .002 \.001
[Reference] –1.08 –0.70 –0.49 –0.33 –0.34 –0.09 –0.03
0.24 0.24 0.24 0.24 0.24 0.23 0.23
20.90 8.29 4.28 1.87 1.99 0.17 0.02
0.34 0.50 0.61 0.72 0.71 0.91 0.97
0.21-0.53 0.31-0.80 0.39-0.97 0.45-1.15 0.45-1.14 0.58-1.42 0.61-1.51
\.001 .004 .038 .169 .159 .682 .880
[Reference] 0.18
0.09
3.80
1.21
1.00-1.45
.051
Abbreviations: bLOC, brief loss of consciousness; CI, confidence interval; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness; OR, odds ratio. a Bold font represents statistical significance.
factors unrelated to hospital performance. At the very least, this finding suggests the importance of conducting careful
risk-adjusted analysis prior to leveraging penalties again institutions with observed high levels of readmission.
938 Patients with Medicaid and the uninsured may also be more likely to readmit to a hospital different from the hospital associated with the index admission. While this fact poses a challenge to many single-institution studies, the current study allowed for identification of patients across all hospitals within the state of California. Rates of readmission established in this study are thus more likely to reflect accurately true patient readmission trends. The ability to track readmissions across different hospitals is an important consideration for future studies since CMS establishes penalties based on all readmissions rather than those limited to the original admitting institution. These results provide clinicians and hospital administrators with information that may be useful in identifying patients at an increased risk for readmission. With this information, clinicians may consider additional observation or intervention for patients with risk factors to help reduce readmission rates. This is particularly relevant in the current political landscape in which rates of readmission are receiving increased scrutiny that may eventually expand to surgical specialties. Furthermore, the findings in this study also reinforce the importance of identifying risk factors separately between pediatric and adult patients. Applying readmission reduction interventions to pediatric patients that were developed in adult populations may ultimately prove ineffective. At 2.3% in children and 1.75% in adults, the observed rate of CSF leak is substantially lower in this report than in previous literature. In one single-institution study of 820 temporal bone fractures, CSF leak occurred in 14.9% of cases.7 Other reports observed a CSF leak in 18% to 26% of pediatric temporal bone and basilar skull fractures.8-10 There are several possibilities to explain the differences between this study and previous literature. The figures reported previously may be outdated, as many reports were developed in the 1980s and 1990s. It is possible that safety devices (eg, airbags and helmets) and public awareness have improved, thus reducing the incidence of severe injury. However, improvements would have had to occur prior to 1998 as the rate observed in this study is stable as far back as this date. Previous literature also tends to report numbers from single institutions, which may have a selection bias for higher-severity injury. In contrast, this study gathered data from every inpatient admission across California. Finally, advances in imaging may allow for diagnosis and prompt admission of skull fractures that were previously left untreated. In-hospital and 90-day rates of meningitis following basilar skull fracture were less than 1% for both pediatric and adult patients in this study. While our study could not differentiate between CSF leaks with spontaneous closure versus those with prolonged leaks, repair of CSF leak was used as a surrogate to measure prolonged leaks since surgery is usually delayed. CSF leak repair was given a narrow definition (ICD-9 procedure code 02.12), recognizing that this would not capture all of the potential repair scenarios but knowing that any significant association would likely
Otolaryngology–Head and Neck Surgery 149(6) underestimate the true impact of surgical repair. Despite this limitation, there was an associated increased likelihood of developing meningitis with CSF leak requiring repair. Meningitis occurred at a rate of 0.4% in patients without CSF leak, 1.9% in patients with CSF leak without repair, and 4.3% in patients with CSF leak requiring repair. Previous literature reports meningitis in 1.4% of moderate to severe head trauma. This rate is estimated to reach levels as high as 25% in instances of basilar skull fracture.11,12 However, these numbers are not necessarily consistent across the literature. In one single-institution study of 820 temporal bone fractures, meningitis occurred in 1% of patients without CSF leak, 3% of patients with CSF leak lasting less than 7 days, and 23% of patients with CSF leak greater than 7 days.7 In another study of pediatric patients with basilar skull fracture, meningitis was reported in only 1% of cases.13 Use of antibiotic prophylaxis in patients with skull fracture remains controversial, but differences in management with prophylaxis or improvements in the efficacy of available antibiotics might explain some of the differences between studies. This study has several strengths and limitations. Among its strengths, this study allows for longer-term follow-up of a large number of patients across every inpatient health care institution for a period of greater than 10 years. Therefore, the rates of CSF leak, meningitis, and readmission identified in this study are likely more generalizable. This study is limited in its absence of detailed clinical variables or imaging. As a result, this study cannot comment on the impact of many clinical factors that clinicians use in their management and decision making. Understanding risk factors for developing a readmission or complication following a basilar skull fracture is important for multiple stakeholders involved in patient care. For neurosurgery, trauma, and otolaryngology providers, these risk factors allow for identification and differentiation of patients who may require prolonged hospitalization or close follow-up. For hospital administrators, these risk factors may prompt policies to reduce 30-day readmission in the wake of looming financial penalties. Where postdischarge data regarding meningitis and CSF leak in patients with an isolated basilar skull fracture have been lacking, this study provides insight into 90-day outcomes. Finally, understanding differences between adult and pediatric patients is critical to the surgical community as care for these patients is typically segmented with the existence of pediatric subspecialties. Future work should explore provider and hospitallevel factors that can improve readmission and complication rates in patients with basilar skull fractures. Author Contributions Brandon A. McCutcheon, conceptual design, data analysis, statistical analysis, manuscript writing and editing; Ryan K. Orosco, conceptual design, data analysis, writing and editing; David C. Chang, conceptual design, data acquisition, statistical analysis, manuscript writing and editing, review of final manuscript; Francesca R. Salazar, conceptual design, statistical analysis,
McCutcheon et al manuscript writing and editing; Mark A. Talamini, design of project study, data acquisition, data analysis, manuscript review and writing; Stephen Maturo, conceptual design of project, review of abstract, data and statistical analysis, editing and review of manuscript; Anthony Magit, conceptual design, data acquisition, statistical analysis, manuscript writing and editing.
Disclosures Competing interests: Mark A. Talamini, Leading Biosciences (Board of Directors; Scientific Advisory Board), Guidepoint Global (Consultant). Sponsorships: None. Funding source: This work was supported by an NIH T32 Institutional Research Training Grant (DC000128) to RKO.
References 1. Joynt KE, Jha AK. Thirty-day readmissions: truth and consequences. N Engl J Med. 2012;366:1366-1369. 2. Joynt KE, Jha AK. A path forward on Medicare readmissions. N Engl J Med. 2013;368:1175-1177. 3. Stephenson SC, Langley JD, Civil ID. Comparing measures of injury severity for use with large databases. J Trauma. 2002; 53:326-332. 4. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46:1075, 1079.
939 5. Sanches JE, de Godoy JM, Baitello AL, Chueire AG. Loss of consciousness in injuries of the extremities is an alert to a higher probability of death. Open Orthop J. 2012;6:590-592. 6. Cucchiara BL, Kasner SE, Wolk DA, et al. Early impairment in consciousness predicts mortality after hemispheric ischemic stroke. Crit Care Med. 2004;32:241-245. 7. Brodie HA, Thompson TC. Management of complications from 820 temporal bone fractures. Am J Otol. 1997;18:188-197. 8. McGuirt WF Jr, Stool SE. Cerebrospinal fluid fistula: the identification and management in pediatric temporal bone fractures. Laryngoscope. 1995;105:359-364. 9. Liu-Shindo M, Hawkins DB. Basilar skull fractures in children. Int J Pediatr Otorhinolaryngol. 1989;17:109-117. 10. Shapiro RS. Temporal bone fractures in children. Otolaryngol Head Neck Surg. 1979;87:323-329. 11. van de Beek D, Drake JM, Tunkel AR. Nosocomial bacterial meningitis. N Engl J Med. 2010;362:146-154. 12. Friedman JA, Ebersold MJ, Quast LM. Post-traumatic cerebrospinal fluid leakage. World J Surg. 2001;25:1062-1066. 13. Kadish HA, Schunk JE. Pediatric basilar skull fracture: do children with normal neurologic findings and no intracranial injury require hospitalization?Ann Emerg Med. 1995;26:37-41.
Outcomes of Isolated Basilar Skull Fracture: Readmission, Meningitis, and Cerebrospinal Fluid Leak
Otolaryngology– Head and Neck Surgery 149(6) 931–939 Ó American Academy of Otolaryngology—Head and Neck Surgery Foundation 2013 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0194599813508539 http://otojournal.org
Brandon A. McCutcheon1, Ryan K. Orosco, MD1, David C. Chang, PhD, MPH, MBA2, Francesca R. Salazar2, Mark A. Talamini, MD2, Stephen Maturo, MD3, and Anthony Magit, MD, MPH1
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
Keywords
Abstract
Received July 15, 2013; revised September 17, 2013; accepted September 20, 2013.
Objective. To determine rates of cerebrospinal fluid (CSF) leak, meningitis, and readmission in pediatric and adult patients with isolated basilar skull fracture. Study Design. Cross-sectional analysis of a statewide database. Subjects. Patients with isolated basilar skull fracture (19952010). Methods. Patients were identified within the California Office of Statewide Health Planning and Development database using ICD-9 diagnosis codes. Results. A total of 3563 pediatric and 10,761 adult patients met inclusion criteria. In-hospital rates of meningitis (0.48% and 0.64%, P = .3360) and CSF leak (2.33% and 1.75%, P = .0270) were similar among children and adults, respectively. Rates of 90-day meningitis (0.17% and 0.37%, P = .0714) and CSF leak (0.40% and 0.40%, P = .9823) were also similar. Thirty-day readmission was 4.6% for children compared with 12.4% for adults (P \ .001). For both pediatric and adult patients, extra-axial hematoma (odds ratio [OR] [confidence interval {CI}] 1.65 [1.05-2.59] and 1.61 [1.34-1.95]) and comorbidities (OR [CI] 2.19 [1.11-4.34] and 1.28 [1.041.59]) were associated with significant increases in 30-day readmission. Loss of consciousness greater than 1 hour (OR, 3.05; 95% CI, 1.53-6.08) and CSF leak (OR, 3.28; 95% CI, 1.41-7.64) increased the likelihood of pediatric readmissions. Lack of insurance (OR, 0.67; 95% CI, 0.50-0.90) and female gender (OR 0.83; 95% CI, 0.70-0.99) reduced the likelihood of adult readmission. Conclusion. Meningitis and CSF leak following isolated basilar skull fractures are uncommon. Readmission within 30 days was more common in adults than in children. CSF leak, hematoma, and prolonged loss of consciousness increased the likelihood of readmission in children. Intracranial injury, male gender, having insurance, and comorbidities increased the likelihood of readmission in adults.
basilar skull fracture, outcomes research, 30-day readmission
R
eadmission is an increasingly important clinical outcome associated with additional costs to health systems and psychosocial burdens to patients. Furthermore, the Centers for Medicare and Medicaid Services (CMS) have been mandated by the Affordable Care Act to penalize health institutions with higher-thanexpected rates of readmission.1 As a result, the Hospital Readmissions Reduction Program was started in October 2012 to leverage penalties against health care institutions with elevated rates of 30-day readmission. These penalties will affect nearly two-thirds of hospitals and total more than $280 million in 2013.2 While penalties for excessive readmission have initially focused on medical conditions such as heart failure, pneumonia, and myocardial infarction, penalties may eventually expand to admissions for traumatic injuries and other diagnoses managed by surgical teams. Unfortunately, there has been little research on the risk factors associated with these readmissions. Nor is it known whether these risk factors differ across pediatric and adult populations.
1 Division of Head and Neck Surgery, University of California San Diego, San Diego, California, USA 2 Department of Surgery, University of California San Diego, San Diego, California, USA 3 Department of Otolaryngology, San Antonio Military Medical Center, Fort Sam Houston, Texas, USA
This article was presented at the 2013 AAO-HNSF Annual Meeting & OTO EXPO; September 29–October 3, 2013; Vancouver, BC, Canada. Corresponding Author: Brandon A. McCutcheon, Division of Head and Neck Surgery, University of California San Diego, 4355 Rialto St #2, San Diego, CA 92107, USA. Email: [email protected]
932 Despite a growing body of knowledge pertaining to pediatric and adult head trauma, risk factors for 30-day readmission in patients with basilar skull fractures have not been reported. This patient population is also at risk for cerebrospinal fluid (CSF) leak and meningitis. Reported rates of meningitis and CSF leak following head trauma vary widely across the literature, are often outdated, and commonly arise from single-institution studies. Many studies are also limited in that they do not monitor patients for the development of complications for an extended period of time after discharge. The objective of this study was to analyze an administrative database to evaluate meningitis, CSF leak, and readmission rates in pediatric and adult patients.
Methods Per the University of California, San Diego, Human Research Protection Program, this project was considered exempt from institutional review board review.
Inclusion and Exclusion Criteria A retrospective analysis was conducted using the California Office of Statewide Health Planning and Development (OSHPD) Patient Discharge database from 1995 to 2010. This database captures 100% of patients in California acutecare hospitals. Patients with closed basilar skull fracture were identified using the following ICD-9 diagnosis codes: 801.0 (fracture of base of skull, closed without mention of intracranial injury), 801.1 (fracture of base of skull, closed with cerebral laceration and contusion), 801.2 (fracture of base of skull, closed with subarachnoid, subdural, and extradural hemorrhage), 801.3 (fracture of base of skull, closed with other unspecified intracranial hemorrhage), 801.4 (fracture of base of skull, closed with intracranial injury of other and unspecified nature). The International Classification of Disease–based Injury Severity Score (ICISS), commonly used in trauma, was employed as a measure of trauma severity to allow for the identification of isolated cases of basilar skull fracture. The ICISS uses survival risk ratios for each ICD-9 diagnosis code to assess the effect of all presenting injuries on the likelihood of survival.3 Admissions for basilar skull fracture were considered to be isolated if the patient’s overall survival risk ratio was within 3% of the individual survival risk ratio attributed to the basilar skull fracture diagnosis code alone. Using this methodology, patients were excluded from analysis if they did not have an isolated skull fracture. Patients were also excluded if their admission was recorded as ‘‘scheduled’’ according to the OSHPD database.
Outcomes The primary outcome of interest was 30-day readmission, which was defined as all-cause readmission to any hospital within California within 30 days following discharge. Determinants of readmission were assessed using multivariate logistic regression analysis. Secondary outcome measures included meningitis and CSF leak during the initial
Otolaryngology–Head and Neck Surgery 149(6) Table 1. ICD-9 diagnosis codes used to identify cases of meningitis. ICD-9
Description
322.0 320.9 321.8 321.2 047.0 047.1 047.8
Nonpyogenic meningitis Meningitis due to unspecified bacterium Meningitis due to other nonbacterial organism Meningitis due to viruses not elsewhere classified Meningitis due to enterovirus, Coxsackie virus Meningitis due to enterovirus, Echo virus Meningitis due to enterovirus, other specified viral meningitis Meningitis due to enterovirus, unspecified viral meningitis Meningitis in other fungal diseases Meningitis in other bacterial diseases classified elsewhere Eosinophilic meningitis Meningitis unspecified Hemophilus meningitis Pneumococcal meningitis Streptococcal meningitis Staphylococcal meningitis Meningococcal meningitis Anaerobic meningitis Meningitis due to gram-negative bacteria, not otherwise classified Meningitis due to other specified bacteria Salmonella meningitis Meningitis due to adenovirus Gonococcal meningitis Candidal meningitis Coccidioidal meningitis Herpes simplex meningitis Cryptococcal meningitis Meningitis in other fungal disease
047.9 321.1 320.7 322.1 322.9 320.0 320.1 320.2 320.3 036.0 320.81 320.82 320.89 003.21 049.1 098.82 112.83 114.2 054.72 321.0 321.1
admission and within 90 days after discharge. The presence of CSF leak was assessed using the ICD-9 diagnosis codes 349.81 (cerebrospinal fluid rhinorrhea) and 388.61 (cerebrospinal fluid otorrhea). Instances of CSF leak requiring repair were identified using the ICD-9 procedure code 02.12 (other repair of cerebral meninges). Instances of meningitis were identified using the ICD-9 codes listed in Table 1.
Patient and Hospital Characteristics Information was provided on patient insurance coverage, race, gender, age, and year of admission. Pediatric patients were defined as those younger than 18 years. Insurance status was defined as ‘‘Medicaid’’ if patients were covered by Medi-Cal, ‘‘workers compensation,’’‘‘county indigent program,’’‘‘other government,’’ or ‘‘other indigent’’ insurance status according to the payer variable in OSHPD. The Charlson Index was calculated to assess patient comorbidities using the method established by Romano and colleagues.4
McCutcheon et al Hospitals were classified as teaching or nonteaching according to whether or not they maintained a general surgical residency program. Given previous studies demonstrating increased risks of mortality associated with loss of consciousness,5,6 this variable was used as a surrogate for severity. Patients were labeled as prolonged loss of consciousness (pLOC) if loss of consciousness was greater than 1 hour or without return to baseline status. Patients were classified as brief loss of consciousness (bLOC) if there was no LOC, or LOC was less than 1 hour. Patients were also labeled with 1 of the following 4 designations according to the degree of intracranial injury indicated on the ICD-9 diagnosis code associated with admission: (1) no intracranial injury; (2) contusion or laceration; (3) subarachnoid, subdural, or extradural hematoma; and (4) unspecified intracranial injury or hematoma.
Statistical Analysis Statistical analysis was performed using commercially available software (STATA SE Version 11.2; Stata Corp LP, College Station, TX). Logistic regression models controlled for patient insurance status, race, gender, hospital teaching status, Charlson Comorbidity Index, age, loss of consciousness, intracranial injury, CSF leak, and year. Statistical significance was established at the 5% significance level, and all confidence intervals (CIs) are reported as 95% CI.
Results Patient and Hospital Characteristics A total of 3563 pediatric and 10,761 adults met inclusion criteria (Table 2). The mean age was 8 years for pediatric patients (median, 8 years) and 47.5 years for adults (median, 44 years). Pediatric patients were predominately male (69.8%) and Hispanic (44.3%). Adult patients were mostly male (71.4%) and white (64%). Most children were covered under the Medicaid distinction (56.5%), while most adults carried private insurance or Medicare (59.9%). Most children and adults did not present with comorbidities (93.9% vs 77.9%, P \ .001). Most pediatric patients presented without intracranial injury (49%), while most adults developed a hematoma (46.8%). The proportion of adult patients presenting with a pLOC was also significantly greater than that observed in children (16.5% vs 5.2%, P \ .001).
30-Day Readmission Unadjusted rates of 30-day readmission were significantly higher in adult patients (12.4% vs 4.6%, P \ .001). Figures 1 and 2 display the unadjusted risk of 30-day readmission in subset populations for pediatric and adult patients, respectively. While CSF leak and pLOC were associated with significant increases in the likelihood of readmission for pediatric patients, these risk factors were not significantly associated with readmission in adults. On multivariate logistic regression, several statistically significant determinants of 30-day readmission were observed among pediatric patients (Table 3). CSF leak on
933 index admission was associated with an increased likelihood of readmission (odds ratio [OR], 3.28; CI, 1.41-7.64). Intracranial or extra-axial hematomas were associated with an increase in the likelihood of readmission (OR, 1.65; CI, 1.05-2.59). pLOC also increased the likelihood of readmission (OR, 3.05; CI, 1.53-6.08). The likelihood of readmission was also positively correlated with the number of patient comorbidities. In adults, physiological factors were less significant in their association with readmission, while social factors appeared to exhibit a greater effect (Table 4). Patients with Medicaid (OR, 0.77; CI, 0.63-0.95) and the uninsured (OR, 0.67; CI, 0.50-0.90) showed significant reductions in the likelihood of readmission. CSF leak and loss of consciousness were not significant determinants of readmission. Intracranial injury was significantly associated with elevations in the likelihood of readmission, with intracranial or extra-axial hematoma exhibiting the strongest effect (OR, 1.61; CI, 1.34-1.95). Increases in comorbidities were associated with an increased likelihood of readmission. Patients with 3 or more comorbid conditions were more than twice as likely as those without comorbidities to experience a readmission (OR, 2.08; CI, 1.56-2.76). Females were less likely than their male counterparts to experience a readmission (OR, 0.83; CI, 0.70-0.99). Patients aged 30 to 60 years were also less likely to be readmitted compared with patients aged 18 to 30 years.
CSF Leak In-hospital CSF leak was an uncommon complication for both pediatric and adult patients, with rates of 2.33% and 1.75% (P = .0270), respectively. CSF leak within 90 days was even more rare, and rates were identical between pediatric and adult patients at 0.4% (P = .9823). The rate of CSF leak appeared stable without significant interyear variation over the course of the 13-year period of observation (P = .105, data not shown).
Meningitis Cases of in-hospital and 90-day meningitis were rare, each occurring in less than 1% of both pediatric and adult patients. When stratified, rates of meningitis were 0.4% among patients without a CSF leak, 1.9% in patients with a CSF leak not requiring repair, and 4.3% in patients requiring repair (P \ .001). Rates of 90-day meningitis were also associated with CSF leak on initial admission. Patients without an initial CSF leak had a 90-day meningitis rate of 0.3%. This rate rose to 1.1% of patients with CSF leak without repair, and there were no instances of postdischarge meningitis in patients with CSF leak requiring repair (P = .041).
Discussion This study represents the largest longitudinal populationbased study with 90-day outcomes of patients with basilar skull fracture. Our study identifies predictors of readmission and complications and demonstrates that these factors differ
934
Otolaryngology–Head and Neck Surgery 149(6)
Table 2. Unadjusted outcomes and characteristics of pediatric and adult patients with basilar skull fracture.a
Outcomes, % Readmission In-hospital CSF leak 90-day CSF leak In-hospital meningitis 90-day meningitis Extent of injury, n (%) No injury to brain Laceration or contusion Hematoma Unspecified hematoma Loss of consciousness, n (%) bLOC (0-1 h) pLOC (.1 h) Insurance status, n (%) Medicare and private Medicaid Uninsured Race, n (%) White Black Hispanic Asian or Pacific Islander Native American/other Gender, n (%) Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, n (%) Mean (median) \1 1-4 5-9 10-17 18-30 30-39 40-49 50-59 60-69 70-79 80-89 901 Hospital setting, n (%) Teaching
Pediatric (n = 3563)
Adults (n = 10,761)
P Value
4.6 2.33 0.40 0.48 0.17
12.4 1.75 0.40 0.64 0.37
\.0001 .0270 .9823 .3360 .0714
1744 (49.0) 356 (10.0) 1110 (31.2) 353 (9.9)
3095 (28.8) 1466 (13.6) 5038 (46.8) 1162 (10.8)
\.001
2871 (94.9) 156 (5.2)
6840 (83.5) 1352 (16.5)
\.001
1385 (39.3) 1992 (56.5) 146 (4.1)
6381 (59.9) 3005 (28.2) 1267 (11.9)
\.001
1325 (37.9) 228 (6.5) 1550 (44.3) 181 (5.2) 214 (6.1)
6797 (64.0) 529 (5.0) 2244 (21.1) 701 (6.6) 345 (3.3)
\.001
1077 (30.2)
3077 (28.6)
.063
3346 (93.9) 146 (4.1) 40 (1.1) 31 (0.9)
8383 (77.9) 1317 (12.2) 513 (4.8) 548 (5.1)
\.001
8 (8) 304 (8.5) 997 (28.0) 744 (20.9) 1518 (42.6) — — — — — — — —
47.5 (44) — — — — 2976 (27.7) 1618 (15.0) 1697 (15.8) 1266 (11.8) 879 (8.2) 1075 (10.0) 1006 (9.4) 244 (2.3)
875 (24.6)
2391 (22.2)
—
.004
Abbreviations: bLOC, brief loss of consciousness; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness. a Bold font represents statistical significance.
across pediatric and adult populations. Children with CSF leak, intracranial hematoma, pLOC, or a comorbid condition were at an increased risk of readmission. In contrast, CSF
leak and loss of consciousness were not significant predictors of readmission for adults. Increases in the risk of readmission were associated with intracranial injuries and
McCutcheon et al
935
Figure 1. Pediatric risk-adjusted rates of 30-day readmission in subsets of patients with complications and comorbidities.
Figure 2. Adult risk-adjusted rates of 30-day readmission in subsets of patients with complications and comorbidities.
comorbidities. Reductions in the risk of readmission were noted among patients with Medicaid, the uninsured, females, and patients between the ages of 30 and 60 years. The finding that pediatric patients with hematoma, CSF leak, or pLOC were associated with increased risks of readmission is clinically intuitive. These patients likely suffered from higher-severity injuries, which increases the risk that they will develop a complication. What remains to be seen is whether the increased likelihood of readmission associated with these risk factors could be blunted by additional time in the hospital or further interventions. It is perhaps surprising, however, that CSF leak and pLOC were not significant predictors of readmission in adults. It is unclear whether this difference between pediatric and adult patients is the result of variation in mechanism of injury, underlying
physiology, patient management, diagnosis, or other unmeasured factors. Comorbid conditions were unsurprisingly associated with increases in readmission across both pediatric and adult populations. Patients with disease involving additional organ systems may have delayed recovery and increased risks of complications following an injury. However, patients with comorbid illness also have a higher baseline likelihood of hospitalization, and it is unclear whether a 30-day readmission reflects an event that would not have happened in the absence of the index hospitalization. Regardless, patients with comorbid illnesses may experience reduced readmission rates if hospitalizations for acute injuries are also seen as opportunities to review and optimize treatment of chronic conditions.
936
Otolaryngology–Head and Neck Surgery 149(6)
Table 3. Multivariate logistic regression analysis demonstrating determinants of 30-day readmission in pediatric patients with basilar skull fracture.a
Length of stay 0-25 percentile 25-50 percentile 50-75 percentile .75 percentile CSF leak No leak Leak Extent of injury No injury to brain Laceration or contusion Hematoma Unspecified intracranial injury Loss of consciousness bLOC (0-1 h) pLOC (.1 h) Insurance status Medicare and private Medicaid Uninsured Race White Black Hispanic Asian or Pacific Islander Native American/other Gender Male Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, y \1 1-4 5-9 10-17 Hospital setting Nonteaching Teaching
Coefficient
Standard Error
Wald x2
OR
95% CI
P Value
[Reference] –0.06 –0.05 –0.07
0.26 0.32 0.27
0.05 0.02 0.06
0.94 0.95 0.93
0.56-1.58 0.51-1.77 0.55-1.58
.828 .878 .809
[Reference] 1.19
0.43
7.62
3.28
1.41-7.64
.006
[Reference] 0.37 0.50 0.06
0.32 0.23 0.35
1.39 4.75 0.03
1.46 1.65 1.07
0.78-2.71 1.05-2.59 0.54-2.12
.239 .030 .856
[Reference] 1.12
0.35
3.05
1.53-6.08
.002
[Reference] –0.01 0.57
0.21 0.41
0.00 1.96
0.98 1.77
0.66-1.48 0.79-3.96
.958 .162
[Reference] 0.19 0.11 -0.29 0.48
0.39 0.22 0.54 0.36
0.24 0.25 0.29 1.82
1.20 1.11 0.75 1.62
0.56-2.60 0.72-1.71 0.26-2.15 0.80-3.28
.626 .615 .591 .177
[Reference] 0.12
0.21
0.33
1.12
0.75-1.70
.560
[Reference] 0.78 1.45 0.77
0.35 0.52 0.67
5.15 7.72 1.32
2.19 4.24 2.16
1.11-4.34 1.53-11.75 0.58-8.09
.023 .005 .250
[Reference] 0.06 0.21 0.67
0.41 0.43 0.40
0.03 0.24 2.82
1.06 1.23 1.96
0.47-2.39 0.53-2.87 0.90-4.31
.884 .626 .092
[Reference] 0.02
0.22
0.01
1.02
0.66-1.59
.914
10.0
Abbreviations: bLOC, brief loss of consciousness; CI, confidence interval; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness; OR, odds ratio. a Bold font represents statistical significance.
The observation that patients with Medicaid and the uninsured were less likely to be readmitted may reflect socioeconomic barriers to care. If transportation, absence from work, or other barriers to care are present, the uninsured may delay or avoid seeking follow-up treatment until a higher threshold of concern is reached, thus readmitting at a later date outside of
the 30-day period assessed in this study. This finding speaks to the potential limitation of using 30-day readmission as a quality indicator or outcome metric. The counterintuitive result that patients with Medicaid and the uninsured are less likely to readmit illustrates that this metric may be influenced by a plethora of patient-level
McCutcheon et al
937
Table 4. Multivariate logistic regression analysis demonstrating determinants of 30-day readmission in adult patients with basilar skull fracture.a
Length of stay 0-25 percentile 25-50 percentile 50-75 percentile .75 percentile CSF leak No leak Leak Extent of injury No injury to brain Laceration or contusion Hematoma Unspecified intracranial injury Loss of consciousness bLOC (0-1 h) pLOC (.1 h) Insurance status Medicare and private Medicaid Uninsured Race White Black Hispanic Asian or Pacific Islander Native American/Other Gender Male Female Charlson Index No comorbidities 1 Comorbidity 2 Comorbidities 31 Comorbidities Age, y 18-30 30-39 40-49 50-59 60-69 70-79 80-89 901 Hospital setting Nonteaching Teaching
Coefficient
Standard Error
Wald x2
OR
95% CI
P Value
[Reference] –0.08 0.00 0.08
0.12 0.10 0.11
0.42 0.00 0.59
0.92 1.00 1.09
0.72-1.18 0.82-1.22 0.88-1.35
.517 .962 .442
[Reference] 0.24
0.28
0.74
1.28
0.74-2.19
.388
[Reference] 0.32 0.48 0.33
0.13 0.10 0.14
6.50 25.00 5.76
1.39 1.61 1.39
1.08-1.79 1.34-1.95 1.06-1.81
.011 \.001 .016
[Reference] 0.17
0.14
1.59
1.18
0.90-1.55
.209
[Reference] –0.25 –0.39
0.10 0.15
5.90 6.81
0.77 0.67
0.63-0.95 0.50-0.90
.015 .009
[Reference] 0.05 –0.05 –0.06 –0.02
0.19 0.10 0.15 0.23
0.08 0.26 0.15 0.00
1.05 0.95 0.94 0.98
0.72-1.52 0.78-1.16 0.71-1.26 0.63-1.55
.782 .611 .698 .945
[Reference] –0.18
0.09
4.20
0.83
0.70-0.99
.041
[Reference] 0.25 0.48 0.73
0.11 0.15 0.14
5.24 9.98 25.50
1.28 1.61 2.08
1.04-1.59 1.20-2.17 1.56-2.76
.022 .002 \.001
[Reference] –1.08 –0.70 –0.49 –0.33 –0.34 –0.09 –0.03
0.24 0.24 0.24 0.24 0.24 0.23 0.23
20.90 8.29 4.28 1.87 1.99 0.17 0.02
0.34 0.50 0.61 0.72 0.71 0.91 0.97
0.21-0.53 0.31-0.80 0.39-0.97 0.45-1.15 0.45-1.14 0.58-1.42 0.61-1.51
\.001 .004 .038 .169 .159 .682 .880
[Reference] 0.18
0.09
3.80
1.21
1.00-1.45
.051
Abbreviations: bLOC, brief loss of consciousness; CI, confidence interval; CSF, cerebrospinal fluid; pLOC, prolonged loss of consciousness; OR, odds ratio. a Bold font represents statistical significance.
factors unrelated to hospital performance. At the very least, this finding suggests the importance of conducting careful
risk-adjusted analysis prior to leveraging penalties again institutions with observed high levels of readmission.
938 Patients with Medicaid and the uninsured may also be more likely to readmit to a hospital different from the hospital associated with the index admission. While this fact poses a challenge to many single-institution studies, the current study allowed for identification of patients across all hospitals within the state of California. Rates of readmission established in this study are thus more likely to reflect accurately true patient readmission trends. The ability to track readmissions across different hospitals is an important consideration for future studies since CMS establishes penalties based on all readmissions rather than those limited to the original admitting institution. These results provide clinicians and hospital administrators with information that may be useful in identifying patients at an increased risk for readmission. With this information, clinicians may consider additional observation or intervention for patients with risk factors to help reduce readmission rates. This is particularly relevant in the current political landscape in which rates of readmission are receiving increased scrutiny that may eventually expand to surgical specialties. Furthermore, the findings in this study also reinforce the importance of identifying risk factors separately between pediatric and adult patients. Applying readmission reduction interventions to pediatric patients that were developed in adult populations may ultimately prove ineffective. At 2.3% in children and 1.75% in adults, the observed rate of CSF leak is substantially lower in this report than in previous literature. In one single-institution study of 820 temporal bone fractures, CSF leak occurred in 14.9% of cases.7 Other reports observed a CSF leak in 18% to 26% of pediatric temporal bone and basilar skull fractures.8-10 There are several possibilities to explain the differences between this study and previous literature. The figures reported previously may be outdated, as many reports were developed in the 1980s and 1990s. It is possible that safety devices (eg, airbags and helmets) and public awareness have improved, thus reducing the incidence of severe injury. However, improvements would have had to occur prior to 1998 as the rate observed in this study is stable as far back as this date. Previous literature also tends to report numbers from single institutions, which may have a selection bias for higher-severity injury. In contrast, this study gathered data from every inpatient admission across California. Finally, advances in imaging may allow for diagnosis and prompt admission of skull fractures that were previously left untreated. In-hospital and 90-day rates of meningitis following basilar skull fracture were less than 1% for both pediatric and adult patients in this study. While our study could not differentiate between CSF leaks with spontaneous closure versus those with prolonged leaks, repair of CSF leak was used as a surrogate to measure prolonged leaks since surgery is usually delayed. CSF leak repair was given a narrow definition (ICD-9 procedure code 02.12), recognizing that this would not capture all of the potential repair scenarios but knowing that any significant association would likely
Otolaryngology–Head and Neck Surgery 149(6) underestimate the true impact of surgical repair. Despite this limitation, there was an associated increased likelihood of developing meningitis with CSF leak requiring repair. Meningitis occurred at a rate of 0.4% in patients without CSF leak, 1.9% in patients with CSF leak without repair, and 4.3% in patients with CSF leak requiring repair. Previous literature reports meningitis in 1.4% of moderate to severe head trauma. This rate is estimated to reach levels as high as 25% in instances of basilar skull fracture.11,12 However, these numbers are not necessarily consistent across the literature. In one single-institution study of 820 temporal bone fractures, meningitis occurred in 1% of patients without CSF leak, 3% of patients with CSF leak lasting less than 7 days, and 23% of patients with CSF leak greater than 7 days.7 In another study of pediatric patients with basilar skull fracture, meningitis was reported in only 1% of cases.13 Use of antibiotic prophylaxis in patients with skull fracture remains controversial, but differences in management with prophylaxis or improvements in the efficacy of available antibiotics might explain some of the differences between studies. This study has several strengths and limitations. Among its strengths, this study allows for longer-term follow-up of a large number of patients across every inpatient health care institution for a period of greater than 10 years. Therefore, the rates of CSF leak, meningitis, and readmission identified in this study are likely more generalizable. This study is limited in its absence of detailed clinical variables or imaging. As a result, this study cannot comment on the impact of many clinical factors that clinicians use in their management and decision making. Understanding risk factors for developing a readmission or complication following a basilar skull fracture is important for multiple stakeholders involved in patient care. For neurosurgery, trauma, and otolaryngology providers, these risk factors allow for identification and differentiation of patients who may require prolonged hospitalization or close follow-up. For hospital administrators, these risk factors may prompt policies to reduce 30-day readmission in the wake of looming financial penalties. Where postdischarge data regarding meningitis and CSF leak in patients with an isolated basilar skull fracture have been lacking, this study provides insight into 90-day outcomes. Finally, understanding differences between adult and pediatric patients is critical to the surgical community as care for these patients is typically segmented with the existence of pediatric subspecialties. Future work should explore provider and hospitallevel factors that can improve readmission and complication rates in patients with basilar skull fractures. Author Contributions Brandon A. McCutcheon, conceptual design, data analysis, statistical analysis, manuscript writing and editing; Ryan K. Orosco, conceptual design, data analysis, writing and editing; David C. Chang, conceptual design, data acquisition, statistical analysis, manuscript writing and editing, review of final manuscript; Francesca R. Salazar, conceptual design, statistical analysis,
McCutcheon et al manuscript writing and editing; Mark A. Talamini, design of project study, data acquisition, data analysis, manuscript review and writing; Stephen Maturo, conceptual design of project, review of abstract, data and statistical analysis, editing and review of manuscript; Anthony Magit, conceptual design, data acquisition, statistical analysis, manuscript writing and editing.
Disclosures Competing interests: Mark A. Talamini, Leading Biosciences (Board of Directors; Scientific Advisory Board), Guidepoint Global (Consultant). Sponsorships: None. Funding source: This work was supported by an NIH T32 Institutional Research Training Grant (DC000128) to RKO.
References 1. Joynt KE, Jha AK. Thirty-day readmissions: truth and consequences. N Engl J Med. 2012;366:1366-1369. 2. Joynt KE, Jha AK. A path forward on Medicare readmissions. N Engl J Med. 2013;368:1175-1177. 3. Stephenson SC, Langley JD, Civil ID. Comparing measures of injury severity for use with large databases. J Trauma. 2002; 53:326-332. 4. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46:1075, 1079.
939 5. Sanches JE, de Godoy JM, Baitello AL, Chueire AG. Loss of consciousness in injuries of the extremities is an alert to a higher probability of death. Open Orthop J. 2012;6:590-592. 6. Cucchiara BL, Kasner SE, Wolk DA, et al. Early impairment in consciousness predicts mortality after hemispheric ischemic stroke. Crit Care Med. 2004;32:241-245. 7. Brodie HA, Thompson TC. Management of complications from 820 temporal bone fractures. Am J Otol. 1997;18:188-197. 8. McGuirt WF Jr, Stool SE. Cerebrospinal fluid fistula: the identification and management in pediatric temporal bone fractures. Laryngoscope. 1995;105:359-364. 9. Liu-Shindo M, Hawkins DB. Basilar skull fractures in children. Int J Pediatr Otorhinolaryngol. 1989;17:109-117. 10. Shapiro RS. Temporal bone fractures in children. Otolaryngol Head Neck Surg. 1979;87:323-329. 11. van de Beek D, Drake JM, Tunkel AR. Nosocomial bacterial meningitis. N Engl J Med. 2010;362:146-154. 12. Friedman JA, Ebersold MJ, Quast LM. Post-traumatic cerebrospinal fluid leakage. World J Surg. 2001;25:1062-1066. 13. Kadish HA, Schunk JE. Pediatric basilar skull fracture: do children with normal neurologic findings and no intracranial injury require hospitalization?Ann Emerg Med. 1995;26:37-41.
