Incidence and causes of perioperative mortality after primary surgery for intracranial tumors: a national, population-based study

Clinical article

Full access

Object

Surgical mortality is a frequent outcome measure in studies of volume-outcome relationships, and the Agency for Healthcare Research and Quality has endorsed surgical mortality after craniotomies as an Inpatient Quality Indicator. Still, the frequency and causes of 30-day mortality after neurosurgical procedures have not been much explored. The authors sought to study the frequency and possible causes of death following primary intracranial tumor operations. They also sought to explore a possible predictive value of perioperative mortality rates from neurosurgical centers in relation to long-term survival.

Methods

Using population-based data from the Norwegian cancer registry, the authors identified 15,918 primary operations for primary CNS tumors treated in Norway in the period from August 1955 through December 2008. Patients were followed up until death, emigration, or September 2009. Causes of mortality as indicated on death certificates were studied. Factors associated with an increased risk of perioperative death were identified.

Results

The overall risk of perioperative death after first-time surgery for primary intracranial tumors is currently 2.2% and has decreased over the last decades. An age ≥ 70 years and histopathological entities with poor long-term prognoses are risk factors. Overlapping lesions are also associated with excess risk, indicating that lesion size or multifocality may matter. The overall risk of perioperative death is also higher in biopsy cases than in resection cases. Perioperative mortality rates of the 4 Norwegian neurosurgical centers were not predictive of their respective long-term survival rates.

Conclusions

Although considered surgically related if they occur within the first 30 days of surgery, most early postoperative deaths can happen independent of the handiwork of the operating surgeon or anesthesiologist. Overall prognosis of the disease seems to be a strong predictor of perioperative death—perhaps not surprisingly since the 30-day mortality rate is merely the intonation of the Kaplan-Meier curve. Both referral and treatment policies at a neurosurgical center will therefore markedly affect such early outcomes, but early deaths may not necessarily reflect overall quality of care or long-term results. The low incidence of perioperative death in intracranial tumor surgery also greatly limits the statistical power in comparative analyses, such as between published patient series or between centers and certainly between surgeons. Therefore the authors question the value of perioperative mortality rates as a quality indicator in modern neurosurgery for tumors.

Abstract

Object

Surgical mortality is a frequent outcome measure in studies of volume-outcome relationships, and the Agency for Healthcare Research and Quality has endorsed surgical mortality after craniotomies as an Inpatient Quality Indicator. Still, the frequency and causes of 30-day mortality after neurosurgical procedures have not been much explored. The authors sought to study the frequency and possible causes of death following primary intracranial tumor operations. They also sought to explore a possible predictive value of perioperative mortality rates from neurosurgical centers in relation to long-term survival.

Methods

Using population-based data from the Norwegian cancer registry, the authors identified 15,918 primary operations for primary CNS tumors treated in Norway in the period from August 1955 through December 2008. Patients were followed up until death, emigration, or September 2009. Causes of mortality as indicated on death certificates were studied. Factors associated with an increased risk of perioperative death were identified.

Results

The overall risk of perioperative death after first-time surgery for primary intracranial tumors is currently 2.2% and has decreased over the last decades. An age ≥ 70 years and histopathological entities with poor long-term prognoses are risk factors. Overlapping lesions are also associated with excess risk, indicating that lesion size or multifocality may matter. The overall risk of perioperative death is also higher in biopsy cases than in resection cases. Perioperative mortality rates of the 4 Norwegian neurosurgical centers were not predictive of their respective long-term survival rates.

Conclusions

Although considered surgically related if they occur within the first 30 days of surgery, most early postoperative deaths can happen independent of the handiwork of the operating surgeon or anesthesiologist. Overall prognosis of the disease seems to be a strong predictor of perioperative death—perhaps not surprisingly since the 30-day mortality rate is merely the intonation of the Kaplan-Meier curve. Both referral and treatment policies at a neurosurgical center will therefore markedly affect such early outcomes, but early deaths may not necessarily reflect overall quality of care or long-term results. The low incidence of perioperative death in intracranial tumor surgery also greatly limits the statistical power in comparative analyses, such as between published patient series or between centers and certainly between surgeons. Therefore the authors question the value of perioperative mortality rates as a quality indicator in modern neurosurgery for tumors.

Although sometimes assessed differently, perioperative or surgical mortality is usually defined as death occurring during the first 30 days after an operation. Surgical mortality has historically been a highly relevant outcome measure in neurosurgery, allowing comparisons of outcome between surgeons and different surgical techniques with a hard clinical end point. By 1910, Harvey Cushing had performed 250 brain tumor operations with a surgical mortality rate of 13%. In contrast, operative mortalities of Cushing's contemporaries were approximately 50%.24 Still today, surgical mortality is a frequently studied and published outcome parameter. United States administrative databases have been used extensively to compare low- and high-volume providers of health care, typically with in-house mortality as the primary outcome variable since 30-day mortality is not available. Various neurosurgical populations have also been studied.3–6,8,9,25,28 Results generally indicate lower risks of surgical mortality (measured as in-house mortality) when patients are treated by high-volume providers. However, it is still not known if an inferior perioperative mortality rate at a single center also reflects actual quality of care or longer term treatment results. Nevertheless, surgical mortality has been endorsed as an Inpatient Quality Indicator by the US Agency for Healthcare Research and Quality in 8 surgical procedures for adults, including craniotomies. Surgical mortality rates are also increasingly publicly reported as an indicator of hospital quality (for example, http://health.usnews.com/best-hospitals/rankings), despite often considerable limitations in data concerning differences in referral and case mix.13

The risks of and possible risk factors for perioperative death have not as yet been much explored in large, unselected brain tumor cohorts. Although defined as surgery related if death occurs within 30 days of an operation, little is known about the specific medical causes of death in neurosurgical patients who die perioperatively. Moreover, whether the enormous perceived progress in neurosurgical technique, neuroanesthesia, and neurointensive care has resulted in decreased perioperative death rates over the last decades has not been documented.

Using data from the Cancer Registry of Norway, we assessed perioperative mortality as 30-day mortality after primary brain tumor operations in a population-based setting. The study was based on data from 15,918 first-time operations for primary CNS tumors treated in Norway in the period from August 1955 through December 2008. The aims of the study were to explore the frequency of perioperative death and the reported causes of death after primary intracranial tumor surgery, a possible change in perioperative death rates over the last decades, factors possibly associated with an increased risk of perioperative death after intracranial tumor surgery, and a possible association between perioperative mortality rates and long-term survival rates in treating departments.

Methods

The regional ethics committee approved this study.

Norwegian Health Care and Neurosurgical Centers

Norway has a population of 4.7 million persons and a socialized health care system with quite evenly distributed resources as well as uniform training and licensing for medical professionals. Today, there are 4 neurosurgical centers offering brain tumor surgery, each serving 1 of 4 geographical health regions (southeast, west, middle, and north). Since 2001 patients have been allowed to choose where to undergo surgery; however, very few decide to undergo these procedures in a health region other than their own. Some proportion of patients living in border regions sometimes decide to undergo surgery in a region different from their own because of transportation logistics. No form of brain tumor surgery is currently regionalized, although the management of acoustic schwannomas may soon be. The fact that epilepsy surgery is centralized to the southeastern health region has implications for where some benign lesions are treated. Even so, for the years after 1990, the correlation between geographical region and treating center is almost perfect (Cohen's kappa = 0.920), indicating unselected and population-based referral to all centers. To avoid the possibility of referral bias, data were nevertheless analyzed by geographical health region, that is, by patient address, and not by treating center.

Norwegian Cancer Registry

Reports to the Norwegian cancer registry have been compulsory since 1952. All neoplasms and certain precancerous lesions must be registered. The cancer registry receives data from 3 sources: 1) copies of all pathology and autopsy reports from all Norwegian laboratories, 2) clinical registration forms from treating doctors, and 3) information from Statistics Norway about cause of death as indicated on death certificates as well as the vital status (alive, dead, or emigrated) of all registered persons in the cancer registry. Compliance of the different data sources is generally good. The unique 11-digit personal identification numbers assigned to all Norwegian citizens ensure tracking of patients and limit the risk of duplicate registrations. Since 1998 the cancer registry has also acquired data files with discharge diagnoses (ICD-10 Sections C or D) on all patients treated for neoplastic disease in every Norwegian hospital and outpatient clinic. When C or D diagnoses are used in patients who are not registered in the cancer registry, the clinician will receive reminders to fill out registration forms. If patients undergo repeated surgery or if an autopsy is performed, the database is continuously updated through the submission of additional histological reports. The completeness of patient registration has improved over the years, and there are presumably no systematic biases in the registered data. A study from 2001 to 2005 demonstrated a 93.8% completeness of data for all CNS tumors, including non–histologically verified cases.22 Note also that the Norwegian population is very stable with little migration.

Included Patients and Variables

In this study we sought to include all primary operations for primary intracranial tumors treated in Norway and reported to the Norwegian cancer registry from the first CNS reports in 1955 through December 2008. Only primary operations were included because it was difficult to assess subsequent operations since the quality and method of data coding, especially in the earlier years, made it difficult to convincingly separate histopathological revisions from reoperations. Some reoperations may also not be histologically confirmed. From August 1955 until December 2008, 16,083 histopathologically confirmed primary intracranial CNS tumors were identified, excluding tumors that were only histopathologically confirmed through autopsy. Twelve patients were lost to follow-up before 30 days after surgery because of emigration and thus were excluded from our study. Eighty-nine duplicate reports (difference in histopathology, same person) were excluded because most of them probably represented reoperations and revised or transformed histopathologies. Sixty-four negative follow-ups, indicating wrong surgery date or revisions of histopathology postmortem, were also excluded. Thus, 15,918 primary operations for primary CNS tumors were included in the final analysis.

We assessed perioperative mortality, defined as death within the first 30 days after surgery. Possible risk factors for perioperative death were explored, such as patient sex and age, type of surgery, localization of lesion, and histopathological group. We also investigated causes of death, as reported on death certificates to Statistics Norway. All deaths in Norway are pronounced by a physician. Physicians pronouncing death report time and cause of death in accordance with WHO guidelines and based on the ICD. During the study period, various ICD codes were in use: ICD-6 (1951–1958), ICD-7 (1959–1969), ICD-8 (1969–1985), ICD-9 (1986–1990), and ICD-10 (1996 onwards). The cause-of-death section on a death certificate is divided into 2 parts (Parts I and II). The disease or condition directly leading to death is stated in Part Ia (immediate cause), and the conditions or diseases that started the sequence that eventually led to death are noted in Part Ib or Ic (underlying cause). Other conditions or diseases that have contributed to death can be listed in Part II. Agreement between underlying causes of death in patients with brain tumors as registered by Statistics Norway and a review of hospital records has been reported to be as high as 95%.19 However, in patients with brain tumors, the underlying cause of death (condition starting the sequence leading to death) is usually, by definition, the tumor itself. Death certificates often do not correctly state the sequential causes of death,17,18 so we decided to group all immediate and contributing causes of death together to assess the relationships between conditions and factors, but we did not consider the relative importance or sequence of the various factors in a single patient.

Histopathological Classification

In the Norwegian cancer registry, tumors were classified according to the ICD-7 until 1993. Thereafter, the ICD-O-2 was used. Using the ICD codes, we categorized surgically treated lesions into major groups in accordance with the WHO classification of CNS tumors.26

Statistics

Data were analyzed using SPSS version 15.0 for Windows (SPSS, Inc.). The statistical significance level was set at p ≤ 0.05. All significance tests were 2-sided. Cohen's kappa analyses were used to test concordance between dichotomous variables. The chi-square test, or the Fisher exact test when expected values in any of the cells of a table were below 5, were used to test significance in contingency tables. Binary logistic regression was used to explore possible factors independently associated with excess risk of perioperative mortality.

Two- and 5-year survival rates were assessed in life tables.

Results

Treatment Period and Type of Surgery

During the first 30 days after primary surgery 3.2% (516) of 15,918 patients died from their CNS tumor. As seen in Table 1, overall perioperative mortality rates have improved over time, from 7.6% before 1980 to 2.1% after 2000 (p < 0.001). The overall risk of perioperative death was higher in biopsy cases than in tumor resections (p = 0.003). However, in a large proportion of patients who underwent surgery in the earliest years, the type of surgery performed is unknown.

TABLE 1:

Perioperative mortality in various treatment periods and tumor procedures

Treatment PeriodNo./Total (%)
Tumor ResectionBiopsy OnlyUnknownOverall Perioperative Mortality
prior to 19800/91/3189/2484 (7.6)190/2496 (7.6)
1980s31/1491 (2.1)17/258 (6.6)57/1767 (3.2)105/3516 (3.0)
1990s81/3625 (2.2)22/855 (2.6)7/145 (4.8)110/4625 (2.4)
after 2000102/4887 (2.1)9/324 (2.8)0/70 (0)111/5281 (2.1)
total214/10012 (2.1)49/1440 (3.4)253/4466 (5.7)516/15918 (3.2)

Patient Sex and Age

There was no significant difference in overall perioperative mortality between the sexes (p = 0.686). Age was positively associated with a risk of perioperative mortality, as seen in Table 2. Overall perioperative mortality rates were 2.9% in patients younger than 70 years of age, as compared with 5.7% in patients 70 years or older for the entire study period (p < 0.001). There was no significant overall difference between children (< 16 years) and adults ages 16–69 years (p = 0.678).

TABLE 2:

Perioperative mortality in various age groups

Treatment PeriodAge (no./total [%])
<16 Yrs16–69 Yrs≥70 Yrs
prior to 1990s28/511 (5.5)232/5021 (4.6)35/480 (7.3)
after 1990s10/731 (1.4)127/7561 (1.7)84/1614 (5.2)
total38/1242 (3.1)359/12582 (2.9)119/2094 (5.7)

Tumor Locations

Since 1993 location of the lesion within the brain has been reported to the cancer registry. Perioperative mortality rates associated with different tumor locations are shown in Table 3. As seen, overlapping brain lesions, parietal lesions, unspecified cerebrum location lesions, and unspecified location lesions were associated with an excess risk, whereas pituitary gland lesions were associated with a lower risk.

TABLE 3:

Perioperative mortality associated with different tumor locations, numbers from 1993 through 2008

Distribution of Tumor LocationsPerioperative Mortality (no./total [%])p Value*
frontal lobe13/848 (1.5)0.144
temporal lobe17/743 (2.3)0.921
parietal lobe14/331 (4.2)0.012
occipital lobe2/163 (1.2)0.289
cerebrum, NOS21/449 (4.7)<0.001
cerebellum4/299 (1.3)0.195
brainstem5/127 (3.9)0.155
overlapping brain11/167 (6.6)<0.001
ventricle2/105 (1.9)0.582
meninges43/2346 (1.8)0.121
cranial nerve3/371 (0.8)0.057
pituitary gland5/838 (0.6)<0.001
craniopharyngeal duct0/91 (0)0.126
pineal gland0/29 (0)0.518
unspecified location51/1633 (3.1)0.007
total191/8540 (2.2)

* Compared with other locations. Abbreviation: NOS = not otherwise specified.

† Significant at p ≤ 0.05.

Histopathological Subgroups

As seen in Table 4, perioperative mortality rates vary with tumor type, as the incidence of death is generally higher in patients with high-grade neuroepithelial tumors and lower in those with benign lesions. Comparing results before and after 1990, there has been a general improvement in perioperative mortality for practically all histopathological subgroups, although statistically significant improvement was observed only in low-grade astrocytoma, mixed or high-grade astrocytoma, ependymoma or subependymoma, and pineal parenchymal tumor groups.

TABLE 4:

Perioperative mortality in various histopathological groups, before versus after 1990*

HistopathologyNo./Total (%)p Value1955–2008 (no./total [%])
Before 1990After 1990
low-grade astrocytoma23/652 (3.5)7/884 (0.8)<0.00130/1536 (2.0)
mixed/high-grade astrocytoma165/2268 (7.3)116/2978 (3.9)<0.001281/5246 (5.4)
oligodendroglioma13/250 (5.2)6/302 (2.0)0.06819/552 (3.4)
ependymoma/subependymoma5/108 (4.6)2/276 (0.7)0.0217/384 (1.8)
PNET/medulloblastoma8/116 (6.9)7/169 (4.1)0.30615/285 (5.3)
other glioma1/14 (7.1)13/288 (4.5)0.49314/302 (4.6)
choroid plexus tumor2/15 (13.3)1/37 (2.7)0.1963/52 (5.8)
pineal parenchymal tumor3/14 (21.4)0/31 (0)0.0263/45 (6.7)
nerve sheath tumor8/355 (2.3)3/433 (0.7)0.07311/788 (1.4)
meningioma29/1204 (2.4)45/2717 (1.7)0.11074/3921 (1.9)
pituitary adenoma/carcinoma9/613 (1.5)6/955 (0.6)0.09515/1568 (1.0)
craniopharyngioma3/68 (4.4)0/113 (0)0.0523/181 (1.7)
other lesion26/335 (7.8)15/723 (2.1)<0.00141/1058 (3.9)
total295/6012 (4.9)221/9906 (2.2)<0.001516/15918 (3.2)

* PNET = primitive neuroectodermal tumor.

† Significant at p ≤ 0.05

As seen in Fig. 1, there seems to be an inverse relationship between the risk of perioperative mortality and long-term (5-year) survival among the various tumor types. The relationship may not be naturally linear, but to illustrate the likelihood of a relationship, a gray line was included in the graph and is created by 2 points: 1) the overall perioperative mortality rate of 2.2% and overall 5-year survival of 61% for the entire surgical study population, and 2) crossing of the y-axis by the fact that 100% 5-year survival requires 0% perioperative mortality. Although some histological subgroups are quite small and standard errors are not accounted for, subgroups to the right of the gray line may be associated with a higher risk of perioperative death than expected from long-term survival of the condition. Meningiomas, choroid plexus tumors, and primitive neuroectodermal tumors/medulloblastomas may therefore be associated with higher risks of perioperative mortality than expected from the overall prognosis of these diseases. The figure is based only on operations performed after 1990 to ensure relevance for the situation today.

Fig. 1.
Fig. 1.

Graph showing the risk of perioperative mortality versus 5-year survival for patients with various intracranial lesions, data from 1990. Gray line suggests an overall inverse relationship between the 2 outcome variables.

Regression Analyses: Elderly With Malignant Lesions

When reviewing only the data from 1990 through 2008, to ensure validity for the situation today, we observed that in patients 70 years or older with mixed grade or malignant astrocytomas, perioperative death occurred in 6.6% as compared with 1.1% in patients younger than 70 years with other types of tumors. From Fig. 1, we hypothesized that long-term prognosis as assessed by 5-year survival rates was linked to the risk of perioperative death. Assigning expected 5-year survival rates to all surgically treated lesions, we found that both an age ≥ 70 years and expected 5-year survival rates were independently associated with perioperative mortality, as seen in Table 5. Overlapping lesions (ICD-O-2) were also independently associated with excess risk, indicating that lesion size or multifocality may matter. Type of surgery (that is, resection versus biopsy) did not seem to affect the risk of perioperative death in univariate analyses and was therefore not included in the multivariate model. Further regression analyses to determine other possible independent risk factors of perioperative death were not performed given that there are probably many interactions between the variables (such as location, histopathology, age, and resection/biopsy), violating the assumption of independence of predictors in such analyses.

TABLE 5:

Binary logistic regression analysis for factors independently associated with an increased risk of perioperative death*

VariableUnadjusted OR95% CIp ValueAdjusted OR95% CIp Value
age ≥70 yrs (y/n)3.272.48–4.31<0.0012.901.31–4.75<0.001
5-yr survival rate (%)0.990.98–0.99<0.0010.990.99–0.99<0.001
overlapping lesions (y/n)3.211.71–6.02<0.0012.491.31–4.750.005
biopsy only (y/n)1.210.83–1.780.324

* n = no; y = yes.

Causes of Death

The underlying cause of death, defined as the disease or condition that started the sequence of conditions that led to death, was reported to be the CNS tumor in 91.4% of cases. In 52% of patients, the CNS tumor was the only cause of death reported on the death certificate. Central nervous system tumor as the only reported cause of death was significantly more common in patients with high-grade astrocytomas (p < 0.001). The distribution of reported causes of death is featured in Table 6. As seen, reported causes are quite diverse. On only 4.6% of death certificates did the pronouncing physician note complications of the surgical procedure as a factor. Cerebrovascular causes were reported on 9.8% of certificates, with hemorrhages on 5.8%. Except for cerebral causes, pulmonary causes were most frequently reported. Pulmonary embolism was reported on 3.6% of certificates and deep venous thromboembolism on 1%. Acute myocardial infarction was reported on 2.6% of death certificates. Other cancer, that is, other than CNS cancer, was reported on 2.8%.

TABLE 6:

Frequency of various reported causes of death*

Reported Cause of DeathPercentage
Underlying Cause of DeathImmediate & Contributing CauseSum Cause of Death
cerebral
 CNS tumor91.41.893.2
 cerebrovascular
  hemorrhage0.65.25.8
  infarction0.21.82.0
  unspecified stroke02.02.0
 edema/herniation02.82.8
 epilepsy00.20.2
 meningitis00.80.8
 hydrocephalus00.20.2
pulmonary
 pulmonary embolism03.63.6
 pneumonia0.210.610.8
 chronic bronchitis/fibrosis/COPD01.21.2
 asthma00.20.2
 suffocation due to food aspiration00.20.2
 anoxia00.20.2
 unspecified respiratory failure00.20.2
cardiovascular
 acute myocardial infarction1.21.42.6
 acute cardiopulmonary disease/heart failure02.02.0
 disease of aortic valve00.20.2
 hypertensive heart & renal disease0.20.20.4
 hypertension01.01.0
 atrial fibrillation0.20.20.4
 chronic ischemic heart disease02.22.2
aortic aneurysm/dissection0.400.4
other infection (not pneumonia or meningitis)01.81.8
deep venous thrombosis01.01.0
trauma0.60.41.0
gastric ulcer0.20.60.8
chronic nephritis/kidney failure01.01.0
diverticulitis0.200.2
cholecystitis00.20.2
sudden death, unknown cause01.01.0
cancer other than CNS1.61.22.8
disease of pituitary gland2.602.6
diabetes mellitus0.20.60.8
 hypoglycemia00.20.2
complications of surgical procedure04.64.6
effect of radiation00.20.2
other miscellaneous cause0.22.83.0
none reportedNA60.2NA

* COPD = chronic obstructive pulmonary disease; NA = not applicable.

Perioperative Mortality Rates: A Predictor of Long-Term Outcome?

The possible correlation between perioperative mortality rates and long-term survival in various centers was explored in 5-year intervals for the 4 Norwegian neurosurgical health regions (each served by 1 neurosurgical center) for 1969 through 2008. In lower grade lesions a long follow-up would be of more interest, while in high-grade lesions, such as glioblastomas, 5-year survival rates would have no meaning. We therefore assessed 2-year survival as a compromise to be of relevance for most histopathological entities and because estimated 2-year survival rates were available in recent data. Data were analyzed according to residential address and geographical health region instead of treating health region to eliminate the possibility of referral bias. For the years prior to 1989, only 2 or 3 geographical health regions were assessed because the 2 northernmost neurosurgical centers were established as late as 1986 and 1972. The results in the successive 5-year intervals were dichotomized to better or worse than the national means in the studied time frame, as shown in Table 7. Overall correlation between the perioperative mortality rate and 2-year survival was slight (kappa = 0.12) and nonsignificant (p = 0.547). We decided to explore the possible relationship between short- and long-term survival in all gliomas to study a more homogeneous histopathological patient group (Table 8). The correlation was moderate (kappa = 0.48) but significant (p = 0.013).

TABLE 7:

Comparisons of perioperative mortality rates and 2-year survival rates for all operations in 5-year intervals for 1969–2009*

Treatment PeriodGeographic Health Region (periop mortality/2-yr survival)Range in %
1234Periop Mortality 2-Yr Survival
2004–2008W/BB/BB/WW/W1.1–3.262–73
1999–2003B/WW/BB/WW/B2.3–3.866–69
1994–1998W/WB/BB/BB/W1.4–2.767–75
1989–1993B/BW/WB/WB/B1.9–3.962–68
1984–1988B/WW/WW/W2.8–5.363–66
1979–1983B/BW/BW/W2.0–5.255–64
1974–1978B/BW/WW/W5.9–13.144–52
1969–1973W/BB/W

* Results dichotomized as better (B) or worse (W) than the national mean. Number of cases reflected: 15,127.

TABLE 8:

Comparisons of perioperative mortality rates and 2-year survival rates for glioma operations in 5-year intervals for 1969–2009*

Treatment PeriodGeographic Health Region (periop mortality/2-yr survival)Range in %
1234Periop Mortality2-Yr Survival
2004–2008W/WB/BB/BW/W1.7–3.438–49
1999–2003B/WW/WB/BW/B2.5–6.738–44
1994–1998W/WW/WB/BB/B0.9–4.141–53
1989–1993B/BW/WB/WB/B2.4–5.037–46
1984–1988B/BW/BW/W3.6–7.141–44
1979–1983B/WB/BW/W2.2–7.037–40
1974–1978B/BB/BW/W6.8–15.323–36

* Results dichotomized as better (B) or worse (W) than the national mean. Number of cases reflected: 7881.

Regional Comparisons

In terms of neurosurgical operative volumes, there are significant differences between the 4 centers in Norway given that 54% of the Norwegian population (Statistics Norway, 2010) is served by 1 neurosurgical center in the southeast region. This health region was served by 2 departments prior to a recent fusion of the departments; however, data from this populated southeastern region were analyzed as if from 1 center. When analyzing all treatment periods together, living within the borders of this most populated health region was associated with a lower risk of perioperative death (OR 0.83, 95% CI 0.70–0.99, p = 0.039). Note, however, that the largest center seems to offer surgery to fewer elderly patients with malignant lesions. After adjusting for the identified independent risk factors (overlapping lesions, age ≥ 70 years, and expected 5-year survival), residential address was no longer significant (p = 0.515).

Discussion

In the present study we were able to assess 30-day mortality after primary surgery for intracranial CNS tumors in a population-based setting. The overall incidence of perioperative death after first-time surgery for primary intracranial tumors for the 53-year study period was 3.2%, and the risk is currently around 2.2%. The presented results should be of interest when informing patients. Although possibly of interest for benchmarking purposes, the perioperative mortality rate's value as a quality indicator in modern brain tumor surgery may be questioned.

Surgery-Related Mortality

Surgery-related mortality can be assessed in several ways: as perioperative mortality (defined as death within 2 weeks, 4 weeks, or most frequently 30 days following surgery) or as in-house mortality in publications based on administrative databases since this measure is more readily available. Although 30-day mortality has better face validity, the correlation with in-house mortality rates is generally at least moderate (kappa > 0.40) but with variance across conditions.7 Many variables, including patient selection, may have an effect on surgery-related mortality rate15 but may be difficult to account for in comparative studies. When presented, so-called adjusted rates seldom account for disease-related characteristics but instead only adjust for patient characteristics such as comorbidity, race, social class, and so forth.

In the present population-based dataset from 15,918 primary operations for primary CNS tumors treated over 53 years, we identified several risk factors, most of them disease related. There was no difference between the sexes. Risk was significantly higher in older patients (age ≥ 70 years). Overall risk was also higher in biopsy cases than in tumor resection cases, although not after 1990. Overlapping lesions (ICD-O-2) were also independently associated with excess risk, indicating that lesion size or multifocality may matter. Expected 5-year survival rates of patients with surgically treated histopathological entities were also independently associated with the risk of perioperative mortality. This finding further emphasizes that patient selection is important for such early postoperative results and indicates that comparisons without stratification or adjustment for both patient and disease-related risk factors can be misleading.

Many patients with malignant intracranial lesions that are advanced at the time of surgery will die early during follow-up, and this death will be defined as surgery related. In patients with benign lesions, the rate of tumor-related causes of death is lower according to the shapes of the Kaplan-Meier curves for these conditions. Although avoiding brain tumor surgery in patients with extremely poor prognoses may be advocated for several reasons, active treatment strategies do not necessarily reflect the quality of treatment, but they will increase perioperative mortality rates. Further, although an aggressive surgical approach for various indolent, benign, or incidental lesions may be questioned, such treatment policies will likewise affect perioperative mortality rates. Such case mix and indication differences between centers are almost impossible to control for.

Lack of Power in Many Studies

While pioneers in brain tumor surgery dealt with perioperative deaths quite frequently, early postoperative deaths are fortunately very rare now. As reported in our study, the risk of perioperative death in brain tumor patients younger than 70 years of age without high-grade gliomas is currently only around 1%. This low risk greatly limits the statistical power in comparisons of this measure, even between very large tumor series. For example, if studying a patient selection with an expected perioperative mortality rate of 1%, a sample size of approximately 3500 patients would be needed to detect a 50% difference in the death rate with 80% statistical power. Thus, comparative analyses will often result in an unacceptable chance of false conclusions. The procedures for which surgical mortality has been advocated as a quality indicator are usually not performed frequently enough to judge hospital quality. Only 6% of US neurosurgical centers have craniotomy case loads sufficient to detect a 50% difference in surgery-related mortality over a 3-year period, if compared with expected national benchmark rates.10

Improvement Over Time

As both the completeness and quality of Norwegian cancer registry data have improved over time, results based on more recent data are more reliable. It still seems evident that perioperative mortality rates have declined significantly over time, although they have remained fairly unchanged over the last 2 decades. The improvement may have several and probably a combination of explanations. First, the quality and precision of intracranial tumor surgery has improved much with advances in pre-, intra-, and postoperative imaging; intraoperative visualization; intra- and perioperative surveillance; modern intensive care; and modern neuroanesthesiology. Moreover, given the availability and quality of modern neuroimaging, which have increased the likelihood of detecting subclinical lesions, the incidence of histologically confirmed intracranial tumors is rising. Benign, incidental lesions are often associated with low risks of perioperative mortality, probably affecting recent overall results. In patients with malignant lesions, a proportion will die in the first 30 days, especially if the disease is advanced at the time of surgery. Lead times before surgery are often much shorter today than in the pre-MR imaging era, which also affected results in most histopathological subgroups. Although there has not been a considerable or statistically significant decline in perioperative mortality rates in the last 2 decades, there is still room for improvement. It has been recently demonstrated that simple checklists are associated with a reduction in surgical mortality from 1.5% to 0.8% in adults undergoing noncardiac surgery.16

Causes of Perioperative Death

Evaluating causes of perioperative death is difficult as several risk factors may act together, limiting the possibility of identifying single causes in individual patients. Even the risk of death attributable to general anesthesia is controversial, mostly because of variations in assessment and definitions. The risk of perioperative mortality directly attributable to anesthesia may therefore range from 1 in 6795 to 1 in 200,200 operations.21 The risk of death attributable to anesthesia alone is related to the patient's preoperative physical state, as frequently assessed using the American Society of Anesthesiologists Physical Status Classification System.23 However, few perioperative deaths are caused by anesthesia alone, perhaps accounting for only 15% of 1-day mortality and 5% of 7-day mortality rates.2,20

Although death certificates are not necessarily accurate in the general population, modern diagnostics and imaging often ensure that much is frequently known about causes of death in neurosurgical patients, even without autopsy data. Very few patients tend to die during surgery directly because of the surgical handiwork, for example, from uncontrolled bleedings or other acute complications. Although underreporting of surgery-related complications on death certificates is perhaps likely, the pronouncing physician only noted complications of surgical procedure as a factor in 4.6% of cases. It seems safe to conclude that other and more indirect causes are more frequently encountered. It is known that the total risk of perioperative death is related to patient age, type of surgery, and American Society of Anesthesiologists physical status of the patient.12 Our results indicate that in patients with primary intracranial tumors, the type of lesion is also significantly related to the risk of perioperative death. Technically challenging operations have no clear excess risk, but instead higher risks were associated with histopathological groups with overall poor long-term prognosis. Overlapping lesions were found to be an independent predictor of risk, perhaps implying that the size or multifocality of lesions matters. In malignant lesions, size or multifocality of the lesions may reflect how advanced a disease is at the time of surgery. It was also seen that the risk of perioperative death seems fairly similar in biopsies and open resections in recent years, further emphasizing that the prognosis associated with the histopathology and stage of the disease is of greater importance than the surgical challenge. The cause of early postoperative deaths, at least in malignant brain tumors, is most frequently the tumor itself. The other reported causes of death seem quite diverse and many of them, although perhaps preventable in many cases, may often not directly relate to the skills of the operating surgeon.

Norwegian Results

As mentioned, comparisons with previously published results should be done with much caution given that patient selection and assessments of perioperative mortality can vary and that statistical power can often be limited. Still, a study of surgical procedures performed in 102 hospitals in the Netherlands reported a 7.26% 30-day mortality rate in “63,357 non-vascular brain operations.”27 This rate is considerably higher than we found in our study, but the Dutch study included nontumor surgery, reoperations, and metastasis surgery, which probably accounts for the difference. The risk of perioperative mortality is considerable in metastasis surgery because of the overall grave prognosis of the disease. In a consecutive series of 177 patients surgically treated for brain metastases, perioperative mortality was 9.6%.29 In a study based on administrative data from the US, the in-house mortality rates associated with operations for primary supratentorial brain tumors decreased during the 12-year period from 1988 to 2000, from 4.8% to 1.8%.5 In another article based on US administrative databases, the average mortality rate for craniotomies was 10.7%.10 Because in-house mortality is generally associated with follow-ups shorter than 30-days, the Norwegian results seem better. A study based on administrative databases in the US reported an overall surgical mortality of 3.2% in children who had undergone 101,546 craniotomies, using in-patient mortality as an end point.1 In our dataset, the 30-day mortality rate was 1.4% in children surgically treated since 1990. The overall risk for in-house mortality in operations for acoustic neurinomas was 0.5% in 2643 operations in the US from 1996 to 2000,4 as compared with 0.7% 30-day mortality rate in operations for nerve sheath tumors in Norway since 1990. United States data based on 15,028 meningioma resections between 1988 and 2000 demonstrated an in-house mortality rate of 2.3%,9 as compared with a 30-day mortality rate of 1.9% for meningiomas in our data. United States administrative data from 5497 admissions for transsphenoidal surgery for pituitary tumors between 1996 and 2000 document an overall in-house mortality rate of 0.6%. This rate is the same as the 30-day mortality rate seen after primary operations for pituitary tumors in Norway since 1990.

Predictive Value of a Treatment Center's Perioperative Mortality Rates

Long-term outcomes are seldom assessed in studies on provider volume and quality of care. Instead, perioperative mortality is often assessed as a proxy for overall treatment quality. An extensive systematic review of studies on provider volume in the management of gastrointestinal cancer showed that only 8% of the studies assessed long-term end points, and interestingly 62% of such studies found no apparent effect of operative volumes on outcome.14 Since it is used and widely accepted as a quality indicator in brain tumor surgery, perioperative death rate is often presumed to reflect something about overall treatment results, and not just for the few but unfortunate who die during the perioperative phase. To explore the possible predictive value of 30-day mortality rates on 2-year survival rates, results from the Norwegian centers were dichotomized to better or worse than the national mean for the studied periods. We found no predictive value of 30-day mortality rates in unselected brain tumor patients measured in 5-year intervals in the various centers. However, we observed a correlation between perioperative mortality rates and long-term survival in subgroup analyses of gliomas, albeit statistically moderate.

Volume-Outcome Relationship

A comparison of results between the 4 Norwegian neurosurgical centers was not the aim of this study. Moreover, the low number of centers in Norway perhaps does not allow for any conclusions concerning a possible volume-outcome relationship. Still, we found that patients living within the borders of the most populated health region and served by the neurosurgical center with the highest volume had significantly lower risks of perioperative death, as compared with the 3 lower volume regions. Unless analyzed further, this result could have been interpreted as a reflection of the famous volume-outcome relationship in surgery. However, when adjusting for differences in identified risk factors for perioperative death, patient address was no longer a factor. At least in Norway, centers with lower operative volumes seem to have a more aggressive approach toward elderly patients with malignant lesions. If this is a general trend, it can perhaps be interpreted as an attempt to compensate for lower operative volumes. While it is known that differences in referral (that is, referral bias) may affect volume-outcome relationships, differences in treatment strategies are also possible. Differences in treatment policies and surgical indications are difficult to assess unless referral is population based. Furthermore, adjusting for tumor advancement at the time of surgery is a challenge in neurosurgery, since there is no accepted classification system for brain tumors that assesses the stage of the disease in terms of size, volume, or location. Interestingly, practically all studies on provider volume and outcomes in neurosurgery have used perioperative mortality rates in the form of in-house mortality rates as the key outcome,3–6,8,9,25,28 and adjustment for differences in case mix is for the most part limited to a set of comorbidities.11

Conclusions

The overall risk of perioperative death after first-time surgery for primary intracranial tumors is currently around 2.2%. This low incidence greatly limits the power in comparative analyses, such as between patient series or between centers and certainly between surgeons. Although considered surgically related if it occurs within the first 30 days of surgery, most early postoperative deaths after brain tumor surgery may happen independent of the handiwork of the operating surgeon or anesthesiologist. The overall prognosis of a tumor disease seems to be a strong predictor of perioperative death, perhaps not surprisingly since the 30-day mortality rate is merely the intonation of the Kaplan-Meier curve. Both referral and treatment policies at a neurosurgical center will therefore affect such early outcomes markedly but without necessarily reflecting quality of care. The correlation between a hospital's perioperative mortality rates and long-term survival rates seems weak. Given our findings, we question the value of perioperative mortality rates as a quality indicator in modern neurosurgical tumor operations.

Disclosure

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author contributions to the study and manuscript preparation include the following. Conception and design: Solheim. Acquisition of data: Johannesen. Analysis and interpretation of data: Solheim. Drafting the article: Solheim, Jakola, Gulati. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Solheim. Statistical analysis: Solheim.

This article contains some figures that are displayed in color online but in black and white in the print edition.

References

  • 1

    Abdullah FGabre-Kidan AZhang YSharpe LChang DC: Report of 2,087,915 surgical admissions in U.S. children: inpatient mortality rates by procedure and specialty. World J Surg 33:271427212009

  • 2

    Arbous MSGrobbee DEvan Kleef JWde Lange JJSpoormans HHTouw P: Mortality associated with anaesthesia: a qualitative analysis to identify risk factors. Anaesthesia 56:114111532001

  • 3

    Barker FG II: Craniotomy for the resection of metastatic brain tumors in the U.S., 1988–2000: decreasing mortality and the effect of provider caseload. Cancer 100:99910072004

  • 4

    Barker FG IICarter BSOjemann RGJyung RWPoe DSMcKenna MJ: Surgical excision of acoustic neuroma: patient outcome and provider caseload. Laryngoscope 113:133213432003

  • 5

    Barker FG IICurry WT JrCarter BS: Surgery for primary supratentorial brain tumors in the United States, 1988 to 2000: the effect of provider caseload and centralization of care. Neuro Oncol 7:49632005

  • 6

    Barker FG IIKlibanski ASwearingen B: Transsphenoidal surgery for pituitary tumors in the United States, 1996–2000: mortality, morbidity, and the effects of hospital and surgeon volume. J Clin Endocrinol Metab 88:470947192003

  • 7

    Borzecki AMChristiansen CLChew PLoveland SRosen AK: Comparison of in-hospital versus 30-day mortality assessments for selected medical conditions. Med Care 48:111711212010

  • 8

    Cowan JA JrDimick JBLeveque JCThompson BGUpchurch GR JrHoff JT: The impact of provider volume on mortality after intracranial tumor resection. Neurosurgery 52:48542003

  • 9

    Curry WTMcDermott MWCarter BSBarker FG II: Craniotomy for meningioma in the United States between 1988 and 2000: decreasing rate of mortality and the effect of provider caseload. J Neurosurg 102:9779862005

  • 10

    Dimick JBWelch HGBirkmeyer JD: Surgical mortality as an indicator of hospital quality: the problem with small sample size. JAMA 292:8478512004

  • 11

    Elixhauser ASteiner CHarris DRCoffey RM: Comorbidity measures for use with administrative data. Med Care 36:8271998

  • 12

    Fecho KLunney ATBoysen PGRock PNorfleet EA: Postoperative mortality after inpatient surgery: incidence and risk factors. Ther Clin Risk Manag 4:6816882008

  • 13

    Glance LGOsler TMMukamel DBDick AW: Impact of the present-on-admission indicator on hospital quality measurement: experience with the Agency for Healthcare Research and Quality (AHRQ) Inpatient Quality Indicators. Med Care 46:1121192008

  • 14

    Gruen RLPitt VGreen SParkhill ACampbell DJolley D: The effect of provider case volume on cancer mortality: systematic review and meta-analysis. CA Cancer J Clin 59:1922112009

  • 15

    Hammers RAnzalone SSinacore JOrigitano TC: Neurosurgical mortality rates: what variables affect mortality within a single institution and within a national database? Clinical article. J Neurosurg 112:2572642010

  • 16

    Haynes ABWeiser TGBerry WRLipsitz SRBreizat AHDellinger EP: A surgical safety checklist to reduce morbidity and mortality in a global population. N Engl J Med 360:4914992009

  • 17

    James DSBull AD: Information on death certificates: cause for concern?. J Clin Pathol 49:2132161996

  • 18

    Jansson BJohansson LARosén MSvanström L: National adaptations of the ICD rules for classification—a problem in the evaluation of cause-of-death trends. J Clin Epidemiol 50:3673751997

  • 19

    Johannesen TBLangmark FLote K: Cause of death and long-term survival in patients with neuro-epithelial brain tumours: a population-based study. Eur J Cancer 39:235523632003

  • 20

    Kawashima YTakahashi SSuzuki MMorita KIrita KIwao Y: Anesthesia-related mortality and morbidity over a 5-year period in 2,363,038 patients in Japan. Acta Anaesthesiol Scand 47:8098172003

  • 21

    Lagasse RS: Anesthesia safety: model or myth? A review of the published literature and analysis of current original data. Anesthesiology 97:160916172002

  • 22

    Larsen IKSmåstuen MJohannesen TBLangmark FParkin DMBray F: Data quality at the Cancer Registry of Norway: an overview of comparability, completeness, validity and timeliness. Eur J Cancer 45:121812312009

  • 23

    Lienhart AAuroy YPéquignot FBenhamou DWarszawski JBovet M: Survey of anesthesia-related mortality in France. Anesthesiology 105:108710972006

  • 24

    Liu CYApuzzo ML: The genesis of neurosurgery and the evolution of the neurosurgical operative environment: part Iprehistory to 2003. Neurosurgery 52:3192003

  • 25

    Long DMGordon TBowman HEtzel ABurleyson GBetchen S: Outcome and cost of craniotomy performed to treat tumors in regional academic referral centers. Neurosurgery 52:105610652003

  • 26

    Louis DNOhgaki HWiestler ODCavenee WKBurger PCJouvet A: The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114:971092007

  • 27

    Noordzij PGPoldermans DSchouten OBax JJSchreiner FABoersma E: Postoperative mortality in The Netherlands: a population-based analysis of surgery-specific risk in adults. Anesthesiology 112:110511152010

  • 28

    Smith ERButler WEBarker FG II: Craniotomy for resection of pediatric brain tumors in the United States, 1988 to 2000: effects of provider caseloads and progressive centralization and specialization of care. Neurosurgery 54:5535652004

  • 29

    Stark AMTscheslog HBuhl RHeld-Feindt JMehdorn HM: Surgical treatment for brain metastases: prognostic factors and survival in 177 patients. Neurosurg Rev 28:1151192005

If the inline PDF is not rendering correctly, you can download the PDF file here.

Article Information

Address correspondence to: Ole Solheim, M.D., Department of Neuroscience, Norwegian University of Science and Technology, N-7005 Trondheim, Norway. email: ole.solheim@ntnu.no.

Please include this information when citing this paper: published online January 6, 2012; DOI: 10.3171/2011.12.JNS11339.

© AANS, except where prohibited by US copyright law.

Headings

Figures

  • View in gallery

    Graph showing the risk of perioperative mortality versus 5-year survival for patients with various intracranial lesions, data from 1990. Gray line suggests an overall inverse relationship between the 2 outcome variables.

References

1

Abdullah FGabre-Kidan AZhang YSharpe LChang DC: Report of 2,087,915 surgical admissions in U.S. children: inpatient mortality rates by procedure and specialty. World J Surg 33:271427212009

2

Arbous MSGrobbee DEvan Kleef JWde Lange JJSpoormans HHTouw P: Mortality associated with anaesthesia: a qualitative analysis to identify risk factors. Anaesthesia 56:114111532001

3

Barker FG II: Craniotomy for the resection of metastatic brain tumors in the U.S., 1988–2000: decreasing mortality and the effect of provider caseload. Cancer 100:99910072004

4

Barker FG IICarter BSOjemann RGJyung RWPoe DSMcKenna MJ: Surgical excision of acoustic neuroma: patient outcome and provider caseload. Laryngoscope 113:133213432003

5

Barker FG IICurry WT JrCarter BS: Surgery for primary supratentorial brain tumors in the United States, 1988 to 2000: the effect of provider caseload and centralization of care. Neuro Oncol 7:49632005

6

Barker FG IIKlibanski ASwearingen B: Transsphenoidal surgery for pituitary tumors in the United States, 1996–2000: mortality, morbidity, and the effects of hospital and surgeon volume. J Clin Endocrinol Metab 88:470947192003

7

Borzecki AMChristiansen CLChew PLoveland SRosen AK: Comparison of in-hospital versus 30-day mortality assessments for selected medical conditions. Med Care 48:111711212010

8

Cowan JA JrDimick JBLeveque JCThompson BGUpchurch GR JrHoff JT: The impact of provider volume on mortality after intracranial tumor resection. Neurosurgery 52:48542003

9

Curry WTMcDermott MWCarter BSBarker FG II: Craniotomy for meningioma in the United States between 1988 and 2000: decreasing rate of mortality and the effect of provider caseload. J Neurosurg 102:9779862005

10

Dimick JBWelch HGBirkmeyer JD: Surgical mortality as an indicator of hospital quality: the problem with small sample size. JAMA 292:8478512004

11

Elixhauser ASteiner CHarris DRCoffey RM: Comorbidity measures for use with administrative data. Med Care 36:8271998

12

Fecho KLunney ATBoysen PGRock PNorfleet EA: Postoperative mortality after inpatient surgery: incidence and risk factors. Ther Clin Risk Manag 4:6816882008

13

Glance LGOsler TMMukamel DBDick AW: Impact of the present-on-admission indicator on hospital quality measurement: experience with the Agency for Healthcare Research and Quality (AHRQ) Inpatient Quality Indicators. Med Care 46:1121192008

14

Gruen RLPitt VGreen SParkhill ACampbell DJolley D: The effect of provider case volume on cancer mortality: systematic review and meta-analysis. CA Cancer J Clin 59:1922112009

15

Hammers RAnzalone SSinacore JOrigitano TC: Neurosurgical mortality rates: what variables affect mortality within a single institution and within a national database? Clinical article. J Neurosurg 112:2572642010

16

Haynes ABWeiser TGBerry WRLipsitz SRBreizat AHDellinger EP: A surgical safety checklist to reduce morbidity and mortality in a global population. N Engl J Med 360:4914992009

17

James DSBull AD: Information on death certificates: cause for concern?. J Clin Pathol 49:2132161996

18

Jansson BJohansson LARosén MSvanström L: National adaptations of the ICD rules for classification—a problem in the evaluation of cause-of-death trends. J Clin Epidemiol 50:3673751997

19

Johannesen TBLangmark FLote K: Cause of death and long-term survival in patients with neuro-epithelial brain tumours: a population-based study. Eur J Cancer 39:235523632003

20

Kawashima YTakahashi SSuzuki MMorita KIrita KIwao Y: Anesthesia-related mortality and morbidity over a 5-year period in 2,363,038 patients in Japan. Acta Anaesthesiol Scand 47:8098172003

21

Lagasse RS: Anesthesia safety: model or myth? A review of the published literature and analysis of current original data. Anesthesiology 97:160916172002

22

Larsen IKSmåstuen MJohannesen TBLangmark FParkin DMBray F: Data quality at the Cancer Registry of Norway: an overview of comparability, completeness, validity and timeliness. Eur J Cancer 45:121812312009

23

Lienhart AAuroy YPéquignot FBenhamou DWarszawski JBovet M: Survey of anesthesia-related mortality in France. Anesthesiology 105:108710972006

24

Liu CYApuzzo ML: The genesis of neurosurgery and the evolution of the neurosurgical operative environment: part Iprehistory to 2003. Neurosurgery 52:3192003

25

Long DMGordon TBowman HEtzel ABurleyson GBetchen S: Outcome and cost of craniotomy performed to treat tumors in regional academic referral centers. Neurosurgery 52:105610652003

26

Louis DNOhgaki HWiestler ODCavenee WKBurger PCJouvet A: The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114:971092007

27

Noordzij PGPoldermans DSchouten OBax JJSchreiner FABoersma E: Postoperative mortality in The Netherlands: a population-based analysis of surgery-specific risk in adults. Anesthesiology 112:110511152010

28

Smith ERButler WEBarker FG II: Craniotomy for resection of pediatric brain tumors in the United States, 1988 to 2000: effects of provider caseloads and progressive centralization and specialization of care. Neurosurgery 54:5535652004

29

Stark AMTscheslog HBuhl RHeld-Feindt JMehdorn HM: Surgical treatment for brain metastases: prognostic factors and survival in 177 patients. Neurosurg Rev 28:1151192005

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 34 34 27
Full Text Views 144 144 52
PDF Downloads 116 116 31
EPUB Downloads 0 0 0

PubMed

Google Scholar