Assessment of medical resource utilization for Taiwanese children
hospitalized for intracranial injuries
Abstract
Background: Compared to adults, children and adolescents are at greater risk for traumatic brain injury (TBI), often suffering increased severity and prolonged recovery from accidents. It is
a challenge to provide care for those children who are at risk for complications of TBI under health care resources constrain. Aim: To investigate hospitalization among children with intracranial injuries in terms of incidence and factors related to length of stay (LOS) and medical cost. Methods: Data from the National Health Insurance Research Database from 2007-2009 were used. Totally 8632 children aged <=18 years with acute intracranial injuries caused by accidents were discharged from hospitals in Taiwan. The associations between patient and hospital covariates (e.g., age, gender, accreditation level of hospital, surgical intervention, and number of comorbid conditions) and log-transferred hospitalization cost and length of stay (LOS) were examined with multivaria ble regression analysis and mediation analyses. Results: The incidence rate of hospitalization for acute intracranial injury was 63.3/100,000 per year. Motor vehicle accidents and falls accounted for 63.5% and 23.8% of intracranial
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
injuries, respectively. The mean LOS for children was 5.0 days (median, 3 days), incurring a mean direct medical cost of $US 916.70 (median, $356.2). More incidences (64.1%) and greater medicals cost ($965) occurred in boys. Patients with subarachnoid subdural and extradural haemorrhage tended to have a longer LOS and incur greater medical costs. Surgical intervention and type of healthcare institution are also significant predictors for medical costs. Additionally, LOS is the dominant mediator for the
relationship between predictor and medical cost. Conclusions: Acute intracranial injuries among children incur a substantial health care burden. Therefore, health authorities need to optimally allocate medical resources in cares.
Keywords: intracranial injury; medical resource utilization; traumatic brain injury; children; length of stay; mediator
1 2 3 4 5 6 7 8 9 10 11
Introduction
Traumatic brain injury (TBI) is a leading cause of death and disability worldwide. During 1997-2007, an annual average of 53014 deaths (18.4 / 100 , 000; range: 17.8 / 100 , 000 -19.3 / 100 , 000 ) among U.S. residents were associated with TBIs.1
Additionally, an aggregate hospitalized plus fatal TBI incidence rate of about 235
/ 100 , 000 was derived from 23 European countries from 1980 to 2003. During the same period, a n average mortality rate of about 15 / 100 , 000 and case fatality rate of about 11 per 100 were noted.2 Moreover, there are selective age groups at risk for TBI.
This includes the very young, adolescents and young adults, and the elderly.3 For
example, there are 95,000 hospital admissions from childhood head injuries.4 Compared
to adults, children and adolescents are at a greater risk of TBI, often suffering from increased severity and prolonged recovery from accidents.5,6 A previous study showed
that head injuries accounted for 75% of all pediatric trauma-related hospitalizations,7 and
contributed to nearly 80% of trauma-related deaths in children.8 Severe childhood head
injuries carry a mortality rate of 9 to 35%.9
1. The health care burden from i njuries has been substantial. In U.S., it was
estimated annually 235,000 hospitalizations suffer TBI s , resulting in US $56.3 billion in direct and indirect costs from 1995 to 2001.10 An earlier study found that hospital
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
admissions from childhood head injuries cost annually over $1 billion in 1982 . 4 A
retrospective analysis reported that the mean LOS for pediatric TBI-associated hospitalizations was 4.5 days (median: 2 days)and incurred an average of $20325 (median: $8056) in total charges for inpatient care in 2000. Pediatric inpatients accrued more than $1 billion in total charges for TBI-associated hospitalizations. Compared with privately insured children, children covered by Medicaid stayed longer and were
charged more per hospitalization.11 Costs associated with health care services may be
largely dependent upon the severity of TBI. However, causes of TBI (e.g., injuries due to motor vehicles, bicycles, or falls) may still add modestly but significantly to the prediction of cost even after the level of severity had been accounted.12 Some TBIs, such
as intracranial injuries, can affect a child’s subsequent neurological development. Such injuries always involve multiple injuries of varying degrees of severity, and so affect medical cost. However, the occurrence of these injuries, and the medical resources that they consume, are rarely discussed. Given the fact that the management of pediatric head injuries has evolved over the past decades and that the healthcare resources are limited, it has been a challenge to optimally provide care for those injured children who are at risk for complications of TBI under the health care resources constrain.13 Thus, the
detailed information of clinical resources such as medical cost and length of stay according to difference types of TBI and different accreditation levels of medical 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
institutions is crucial in realizing the allocation of clinical resources for childhood TBI. Such information may also be useful in future assessment of whether the medical resources available in treating childhood TBI are optimally allocated. The purpose of the study was to investigate the incidence of hospitalization for intracranial injuries in children, and to assess the impact of these injuries on medical resource utilization. Specifically, this study sought to determine the potential predictors for length of stay (LOS) and medical costs in children with TBI. These potential predictors include demographic characteristics, surgery, co-morbidity, accreditation level of hospital, and injury type. 1 2 3 4 5 6 7 8 9
Materials and Methods
Data sources
In 1995, Taiwan launched a single-payer National Health Insurance (NHI) Program. The NHI system is compulsory for all citizens from birth. As of 2007, 99% of Taiwan’s population of 22.3 million people was enrolled in this program.14 When people fall ill,
the government helps patients pay part of their medical costs to contracted health care institutions. For example, copayments for patients hospitalized in regular wards for fewer than 30 days can range from 5 to 10% of total medical costs. To minimize the financial burden, copayments on acute ward stays of fewer than 30 days are capped and set at about $US 1000 for one day of hospital stay.15 Taiwan’s Bureau of National
Health Insurance (BNHI) created the National Health Insurance Research Database (NHIRD) to collect data from the National Health Program to facilitate healthcare-related research. To ensure the accuracy of claim files, the BNHI conducts quarterly expert reviews of a random sampling of 50 inpatient claims. False diagnostic reports incur severe penalties that are imposed by the BNHI.16 Researchers wishing to access the
NHIRD must sign a written declaration that they do not intend to obtain information that violates the privacy of patients or health care providers. Admissions data analyzed in this study were retrieved from two NHIRD claims files termed ‘monthly claims 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
summary for inpatient claims’, and ‘inpatient expenditure by admissions’. This nationwide population-based study, which was used to analyze NHIRD data, was evaluated and approved by NHRI.
Study subjects
Between January 2007 and December 2009, a total of 9570 patients aged 18 or less were hospitalized due to accidental intracranial injuries. If a patient had more than one
hospitalizations during the three-year enrollment period, we used only his/her first hospitalization as the index hospitalization. Because the importance of skull fractures as predictors for underlying brain injury is still under debate; and previous studies
suggested increased severity and risk of intracranial injury in the presence of a skull fracture,13, 17 we excluded the patients with skull fractures to avoid contaminations. We
excluded patients who encountered in-hospital mortality or had a length of stay (LOS) of one day or less. T o avoid capturing a discharge without well treatment, patients who were transferred to other hospitals or were discharged against medical advice were also excluded , l eading to the final study cohort of 8 , 632 child patients.
Measurements of covariates
Patient age was calculated as the time period between date of birth and date of admission. The primary diagnoses and secondary diagnoses for each admission are 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
coded in the NHIRD database according to the International Classification of Diseases, Ninth Revision, using ICD-9 -CM codes . The primary diagnosis for each of the study patients was classified into four injury types, which were concussion (ICD-9 code: 850), cerebral laceration and contusion (CLC; ICD-9 code: 851), subarachnoid subdural and extradural haemorrhage following injury (SSEH; ICD-9 code: 852), and other
unspecified intracranial haemorrhages or injuries (UIH; ICD-9 code: 853-854) . The external causes, surgical intervention and hospital accreditation level for each admission which coded in the NHIRD database were also be measured. The external causes of accidents can further be categorized into four injury cause s: vehicle or transport accident (E800-848), accidental falls (E880-888), homicide and injury purposely inflicted by other persons (E960-969), and o ther injuries . We used up to four secondary diagnoses recorded in the index hospitalization to determine number of co-morbidity, ranging from 0 to 4. The medical institutions that cared patients with TBI were categorized in to three accreditation levels; namely, region, district hospital, and medical centre. LOS was calculated as the time period between hospitalization and discharge. The hospital charges, used as a proxy for direct medical cost, were derived from the index
hospitalization. The individual charges included expenses for examination, treatment, medication, radiation therapy, and nursing care. In this analysis, the medical cost includes both the reimbursement from the NHI program and the copayment from the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
patients , but did not include other out-of-pocket expenses . The charges were converted into US$ using an exchange rate of 1US$ = 30 New Taiwan Dollars. T hey were not discounted for lower than 1% inflation rate of NIH expenditure during the three-year period.
Statistical analysis
The number and incidence rate of hospital admissions for children over the three-year study period was calculated, according to various injury types. The annualized incidence rate of hospital admission was calculated as the ratio of total number of hospital
admission to the summation of mid-year children population over the three study years. We also described the distribution of four external causes of injury across the four intracranial injury types. Because of the skewed distributions, both LOS and medical cost were log transformed to achieve normality.18 The multiple linear regression models
for which gender, age stratum, surgical intervention, number of co-morbidity, hospital accreditation level, and intracranial injury type were simultaneously entered as potential predictors were used to associate with the two outcome variables (i.e., LOS and medical cost). The multiple linear regression analyses were performed using various categorical variables as the predictors of interest in dummy-variable form where an arbitrarily selected group was used as the reference group. We used the SPSS 18 for Windows software package (SPSS Inc., Chicago, IL) to perform linear regression analyses and to 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
calculate regression coefficients and associated 95% confidence intervals (CIs).
To estimate the mediation effect of LOS on medical costs, we designated LOS as an endogenous variable, and other factors that were significantly associated with medical costs, as exogenous variables. As a categorical measurement, injury type was not included as an exogenous variable but was treated as a stratified analysis instead. In general, a given variable is said to function as a mediator to the extent that it accounts for the relationship between the predictor and outcome. Using the mediation approach presented by Mackinnon,19 we estimated the direct effects of endogenous and exogenous
variables on medical resource utilization using a path standardized regression
coefficient. An indirect effect (i.e., mediated effect), which reflects how much a one-unit change in an exogenous variable affects medical costs through LOS, can be tested for significance using the Sobel method.20When the direct effect of an exogenous variable
on medical costs is much lower than that of an indirect variable, we have evidence for a dominant mediator of LOS. The level of significance was set at an -value of 0.05. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Results
Table 1 shows the incidence and distribution of external causes of accidents among patients with acute intracranial injuries. The annual incidence of child hospitalization for acute intracranial injury patients was 63.3 / 100,000 during the study period. The greatest annual incidence rate (42.1/100 , 000 ) was observed for patients with concussions. The corresponding figures were 8.9/100 , 000 , 14.3/100 , 000 , and 4.7/100 , 000 for CLC, SSEH, and UIH, respectively. Of the 8632 study subjects, 5481 patients (63%) sustained injuries by vehicle or transport accidents. The remaining injury cause s were accidental falls 2072 (24%) , homicide (5%), and ‘other’ accidents (8%). When stratified by injury type, a similar distribution of these four accident categories was observed, except for patients with CLC, who showed a slightly greater incidence of transport accidents (6 5 %) than the other injury types.
Table 2 provides descriptive statistics of factors that affected medical resource utilization. Patients with acute intracranial injuries stayed in the hospital for an average of 5.0 days (median, 3.0 days), and incurred an average medical cost of $916.7 (median, $356.2). In other words, 8,632 injury-related hospitalizations occurred between 2007 and 2009, resulting in more than 42.9 thousand days of hospitalization and $7.9 million of in-patient hospitalization charges. Most patients were boys (64%), and while boys had 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
a similar LOS as girls, their medical costs were greater ($965 vs. $779). In-patients aged 15–18 years old or less than one year of age had a longer mean LOS (5.8 days), and accrued greater costs ($1000-$1500). Nearly half of the patients studied (48.3%) had surgical intervention. These patients stayed in the hospital for an average of 6.4 days at a cost of $1451, which was nearly twice the LOS and four times the cost as those who did not have surgery. An increase in LOS from 3.2 days to 7.3 days elevated medical costs from $444 to $1530, as the number of co-morbidity increased from 0 to 4. Nearly half of all patients stayed in hospitals with a medium accreditation level. Patients who stayed in hospitals with the highest level of accreditation had the longest mean LOS (7.5 days) and the highest mean medical costs ($2060). The most frequently seen injury was concussion (66.7%), which requires little medical resources. Seventeen percents of patients with SSEH had the longest mean LOS (8.9 days) and the highest mean medical costs ($2463).
Table 3 shows the results from the multivariable regression model, which identified significant predictors for medical resource utilization. After adjusting for potential confounders, t here were no significant differences in LOS due to gender or age ,
although i nfants showed the greatest medical cost , as compared to the other age groups. Patients who underwent surgical procedures had 1.37 times (95% CI, 1.33-1.41) longer LOS than those without surgery. Patients with SSEH had the most increased ratio of 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
LOS (1.89 times), as compared to patients diagnosed with concussions. The
corresponding figures for CLC and UIH were 1.25 times and 1.42 times . Patients who were diagnosed with more than one co-morbidity had longer (1.22, 1.37, 1.54, and 1.88 times for 1, 2, 3, and 4 morbidities, respectively) LOS than those without
co-morbidity. Compared to the patients admitted to district hospitals, those who stayed in regional hospitals and medical centres had a significantly longer LOS at 1.15 times and 1.38 times, respectively. Similar results were observed for direct medical costs; patients who had surgical intervention incurred 2.48 times (95% CI, 2.40-2.57) higher cost compared to those without surgery. Patients with SSEH had the most increased ratio of medical costs (2.50 times), as compared to patients diagnosed with concussions. The corresponding figures for CLC and UIH were 1.38 times and 1.68 times . Increasing numbers of co-morbidity were associated with greater costs from 1.25 times (95% CI, 1.18-1.31) to 2.07 times (95% CI, 1.95-2.18). In comparing regional hospitals, district hospitals and medical centre had significant increases in medical cost of 1.28 and 1.78 times, respectively.
The results of mediation analysis of medical resource utilization are shown in Table 4. Among patients with concussions, CLC and UIH, both surgical intervention and high hospital accreditation level had a significantly direct positive effect on medical costs. However, either no association or a less direct association between the two factors and 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
costs were observed in patients with SSEH. The number of co-morbidity had a positive effect on LOS, but only had an indirect effect on medical costs. In comparing direct effects, a larger and significant indirect effect of exogenous factors on medical cost was observed, irrespective of injury type, indicating that these exogenous factors directly affected LOS, and could lead LOS to indirectly influence medical costs.
1 2 3 4 5
Discussion
The present study reports that the annual incidence of child hospitalization for acute intracranial injury was 63.3 / 100 , 000 in Taiwan during the period between 2007 and 2009. Two thirds of these patients were diagnosed with concussions. A recent European study estimated that the incidence of TBI in young people is 113.9 / 100,000 ; 75% of this population had mild TBI.21 Given a confined definition of TBI that excluding skull
fracture, this study showed a lower incidence of child TBI in Taiwanese children, as compared with data from European studies.21 Our results showed that there was a greater
incidence of TBI in boys, and that the leading external causes of accident were motor vehicle and fall-related, which are in accordance with data from previous studies.1,4, 22,23
Our results also showed that adolescents were the main group of hospitalized TBI patients; they also spent more time in the hospital and consumed extensive health
service resources. One earlier study reported that the median cost for TBI acute care and rehabilitation were $11478 . Although rehabilitation care and patients with skull 12
fractures were excluded from our analysis, our results still showed lower medical costs (i.e. average $917 or median $356) for in-patients with intracranial injures. In addition, patients diagnosed with SSEH tended to have a longer LOS and incur greater medical costs. Nonetheless, comparisons of LOS and medical cost between studies conducted in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
different years should proceed with caution as calendar year may post substantial influences on the estimations.
Our results revealed that medical utilization predictors were surgical intervention, co-morbidity, and hospital accreditation level. It is reasonable to assume that surgery will raise the LOS and medical cost. A positive correlation between co-morbidity and medical utilization becomes apparent in the event of disease complications. A study from Taiwan found that the combination of acute spinal trauma with other somatic injuries resulted in greater demands on medical resources, additional LOS, and increased medical costs.24 Our study found that co-morbidity did not directly influence medical
costs, but raised the costs incurred by increased LOS, suggesting that the presence of co-morbidity can lengthen the LOS, but fails to increase daily in-patient costs. I t may reflect an utilization restraint that patient suffered from multiple disorders needs to take a longer stay for waiting procedural diagnosis or treatment. Because it was impossible to distinguish daily description and costs based on the available data, these speculations cannot be empirically substantiated or refuted. Overall, LOS is a critical mediator for the relationship between disease condition or hospital factors and medical costs. Thus, physicians and hospital administrators may claim greater reimbursement from BNHI by retaining intracranial injury patients for a longer period of time before discharge.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Our analysis also revealed that both surgery intervention and hospital accreditation level increase the LOS, and there was a concomitant increase in costs, particularly among patients with mild to moderate intracranial injures, although the relationship was less strong in severe injury cases (i.e., SSEH). Moreover, both surgical intervention and high hospital accreditation level were directly associated with increased medical costs among patients with milder injuries. Although reimbursements can be paid by BNHI to higher-level
hospitals, for the same LOS, there were only small differences in medical resources spent on severely injured patients among differently accredited hospitals. One explanation for this could be that severe TBI patients are usually transported or referred by ambulance to either a higher-level hospital or a lower level brain trauma-specific hospital. Equal medical resource availability for TBI may reduce the fluctuations in medical cost between hospitals of
different levels. On the other hand, because of inexpensive copayments and free access to hospitals in Taiwan, it is possible that some patients without severe injuries (i.e.,
concussions and CLC) may intentionally be admitted to a higher level hospital to obtain better health care, even though the hospital is not close to the accident location. Patients administered a more detailed evaluation and precise equipment (e.g., computer tomography or magnetic resonance imaging), which are lacking in regional level hospitals, may have increases in both LOS and direct medical costs. Thus, although physicians in higher level 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
hospital may observe a large number of patients with head injur ies , they rarely noted patients who have life-threatening intracranial complications after injur y .
According to Taiwan’s National Insurance Act, hospitals can receive the same diagnostic payment irrespective of their equipment cost. Such policy might result in an overuse of medical equipment with lower cost. In order to avoid the unnecessary utilization of medical equipment, decision rules derived to identify patients with head injuries, who require precise scanning, need to be established through hospital administration policy. For example, in a prospective diagnostic cohort study, Dunning et al. derived a clinical decision rule for the management of children presenting with acute head injuries, which can identify children at high risk who will then undergo computed tomography scanning, and allow the remaining patients to be discharged with no further tests.25 Because there are different views
on trauma mechanisms, pathogenesis, and managements in different areas, individualized guidelines for different countries are necessary. An evidence-based, clinical practice guidelines for severe TBIs was implemented in Taiwan in 2009, which selected 9 topics, including emergency room treatment, intracranial pressure monitoring, cerebral perfusion pressure, fluid therapy, use of sedatives, nutrition, intracranial hypertension, seizure
prophylaxis, and second-tier therapy.26 It is hoped that the guidelines may provide clinicians
with concepts and recommendations in order to promote the quality of care for severe TBIs in Taiwan. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
This study was based on a large cohort, allowing us to perform stratified analyses of the relationship between medical utilization and various predictors. The mediation approach allowed us to determine the collinear effects between disease conditions, hospital factors, and LOS on medical costs. However, there were also several limitations to our study. First, excluding chronic hospitalization, subsequent surgical procedural, ambulatory visit, and rehabilitation, the estimated medical utilization cannot be extrapolated to the entire treatment period. Moreover, due to other potential out-of-pocket expenses that cannot be estimated and were not included in the direct medical cost calculation, the noted expenditures may only reflect a part of the medical costs. The NHI dataset provided no direct measures of morbidity severity, which could have confounded the variables used to analyze the relationship between medical resource utilization and risk. Finally, the exclusion of patients transferred to other hospital would result in exclusion of more severely injured patients who were originally admitted to lower accreditation level hospitals. Consequently, those included in the analysis would be less severe cases and therefore incurred lower costs than those admitted to higher accreditation level hospitals and could stay in that hospitals.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Conclusion
Acute intracranial injuries among children incur a substantial health care burden, with annual 14.3 thousand days of hospitalization and annual cost for hospitalization care exceeding $2.6 million in Taiwan. Most patients sustain injuries by vehicle and fall-related accidents, so special attention should be given to preventing these types of accidents in the target population. Surgery, co-morbidity, and hospital level increased the LOS, and thus, are the main contributors to medical costs. Nonetheless,
over-treatment may occur and increase the medical costs for patients with milder intracranial injuries. Health authorities should derive a decision rule for those children who need higher levels of treatment, without wasting health care resources on those in stable condition. 1 2 3 4 5 6 7 8 9 10 11
References
1. Coronado VG, Xu L, Basavaraju SV, McGuire LC, Wald MM, Faul MD, et al. Surveillance for Traumatic Brain Injury–Related Deaths — United States, 1997–2007. MMWR 2011; 60: 1-32.
2. Tagliaferri F, Compagnone C, Korsic M, Servadei F, Kraus J. A systematic review of brain injury epidemiology in Europe. Acta Neurochir (Wien) 2006; 148: 255-68. 3. Bruns J Jr, Hauser WA. The epidemiology of traumatic brain injury: a review.
Epilepsia 2003; 44 Suppl 10: 2-10.
4. Division of Injury Control. Center for Environmental Health and Injury Control, Centers for Disease Control. Childhood injuries in the United States. Am J Dis Child 1990; 144: 627–46.
5. Brookes M, MacMillan R, Cully S, Anderson E, Murray S, Mendelow AD, et al . Head injuries in accident and emergency departments. How different are children from adults? J Epidemiol Community Health 1990; 44: 147–51.
6. McCrory P, Meeuwisse W, Johnston K. Consensus statement on concussion in sport —the 3rd International Conference on Concussion in Sport, held in Zurich, November 2008. J Clin Neuroscience 2009; 16: 755–63.
7. Zuccarello EF, Zampleri P, Zanardi L, Andrioli GC. Severe head injury in children: early prognosis and outcome. Childs Nerv Syst1985; 1: 158–62.
8. Walker ML, Storrs BB, Mayer TA. Head injuries. In: Mayer TA, editor. Emergency 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
management of pediatric trauma. Philadelphia: WB Saunders; 1985, 272–86. 9. Tullous M, Walker ML, Wright LC. Evaluation and treatment of head injuries in
children. In: Fuhrman BP. Pediatric critical care. St. Louis. Mosby; 1992, 1165–82. 10.
Langlois JA, Rutland-Brown W, Thomas KE. Traumatic b rain i njury in the United States: e mergency d epartment v isits, h ospitalizations, and d eaths . Atlanta, GA: Centers for Disease Control and Prevention, National Center for Injury Prevention and
Control; 2004 . 11.
Schneier AJ , Shields BJ , Hostetler SG , Xiang H , Smith GA. Incidence of p ediatric t raumatic b rain i njury and a ssociated h ospital r esource u tilization in the United States. P
ediatrics 2006 ; 118 : 483-92 . 12
. Jaffe KM, Massagli TL, Martin KM, Rivara JB, Fay GC, Polissar NL. Pediatric traumatic brain injury: acute and rehabilitation costs. Arch Phys Med Rehabil 1993; 74: 681-6.
13
. Dias MS. Traumatic brain and spinal cord injury. Pediatric Clinics of North America 2004; 51: 271–303.
14. Wu CY, Chen YJ, Ho HJ, Hsu YC, Kuo KN, Wu MS, et al . Association between nucleoside analogues and risk of hepatitis B virus–related hepatocellular carcinoma recurrence following liver resection. JAMA 2012; 308:1906-14.
15. Cheng TM. Taiwan's new national health insurance program: Genesis and experience so far. Health Affairs. 2003; 22:61-76.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
16. Cheng CL, Kao YH, Lin SJ, Lee CH, Lai ML. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan.
Pharmacoepidemiol Drug Saf 2011; 20: 236-42.
1 7 . Luerssen TG. Head injuries in children. Neurosurg Clin N Am 1991; 2: 399–410. 18. Liu TC, Chen CS, Lin HC. Does elective caesarean section increase utilization of
postpartum maternal medical care? Med Care 2008; 46: 440-3.
1 9 . Mackinnon DP, Mahwah NJ. Introduction to statistical mediation analysis. Erlbaum; 2008, 47-77.
20
. Sobel ME. Asymptotic confidence intervals for indirect effects in structural equation models. Sociological Methodology 1982; 13: 290-312.
21
. van Pelt DE, de Kloet A, Hilberink SR, Lambregts SA, Peeters E, Roebroeck ME, et al
. The incidence of traumatic brain injury in young people in the catchment area of the University Hospital Rotterdam, The Netherlands. Eur J Paediatr Neurol 2011; 15: 519-26.
22
. Kraus JF, Nourjah P. The epidemiology of mild, uncomplicated brain injury. J Trauma 1988; 28: 1637-43.
23
. Tabish A, Lone NA, Afzal, WM, Salam A. The incidence and severity of injury in children hospitalized for traumatic brain injury in Kashmir. Injury 2006; 37 : 410-5. 24
. Chu DC, Lee YH, Lin CH, Chou P, Yang NP. Prevalence of associated injuries of spinal trauma and their effect on medical utilization among hospitalized adult subjects 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
– a nationwide data-based study. BMC Health Serv Res 2009; 9: 1-8. 25
. Dunning J, Patrick Daly J, Lomas JP, Lecky F, Batchelor J, Mackway-Jones K. Derivation of the children’s head injury algorithm for the prediction of important clinical events decision rule for head injury in children. Arch Dis Child 2006; 91: 885-91.
26. Liao KH, Chang CK, Chang HC, Chang KC, Chen CF, Chen TY, et al. Brain Trauma Foundation. Clinical practice guidelines in severe traumatic brain injury in Taiwan. Surg Neurol 2009; 72 Suppl 2: S66-73.
1 2 3 4 5 6 7 8
