Preoperative Identification of Neurosurgery Patients With A High Risk of In-Hospital Complications: A Prospective Cohort of 418 Consecutive Elective Craniotomy Patients

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

clinical article

J Neurosurg 123:594–604, 2015

Preoperative identification of neurosurgery patients with a


high risk of in-hospital complications: a prospective cohort
of 418 consecutive elective craniotomy patients
Elina Reponen, MD,1 Miikka Korja, MD, PhD,2,3 Tomi Niemi, MD, PhD,1
Marja Silvasti-Lundell, MD, PhD,1 Juha Hernesniemi, MD, PhD,2 and Hanna Tuominen, MD, PhD1
Departments of 1Anaesthesiology and Intensive Care Medicine, and 2Neurosurgery, Helsinki University Central Hospital,
Helsinki, Finland; and 3Australian School of Advanced Medicine, Sydney, New South Wales, Australia

Object  Patients undergoing craniotomy are routinely assessed preoperatively, yet the role of these assessments in
predicting outcome is poorly studied. This study aimed to identify preoperative factors predicting in-hospital outcome
after cranial neurosurgery.
Methods  The study cohort consisted of 418 consecutive adults undergoing elective craniotomy for any intracranial
lesion. Apart from the age criteria (≥ 18 years), almost all patients were considered eligible for the study to increase ex-
ternal validity of the results. The studied preoperative assessments included various patient-related data, routine blood
tests, American Society of Anesthesiologists (ASA) Physical Status Classification system, and a local modification of
the ASA classification (Helsinki ASA classification). Adverse outcomes were in-hospital mortality, in-hospital systemic or
infectious complications, and in-hospital CNS deficits. Resource use was defined as length of stay (LOS) in the intensive
care unit and overall LOS in the hospital.
Results  The in-hospital mortality rate was 1.0%. In-hospital systemic or infectious complications and permanent or
transient CNS deficits occurred in 6.7% and 11.2% of the patients, respectively. Advanced age (≥ 60–65 years), elevated
C-reactive protein level (> 3 mg/L), and high Helsinki ASA score (Class 4) were associated with in-hospital systemic and
infectious complications, and a combination of these could identify one-fourth of the patients with postoperative compli-
cations. Moreover, this combination of preoperative assessment parameters was significantly associated with increased
resource use.
Conclusions  In this first prospective and unselected cohort study of outcome after elective craniotomy, simple pre-
operative assessments identified patients with a high risk of in-hospital systemic or infectious complications as well as
extended resource use. Presented risk assessment methods may be widely applicable, also in low-volume centers, as
they are based on composite predictors and outcome events.
http://thejns.org/doi/abs/10.3171/2014.11.JNS141970
Key Words  craniotomy; elective; outcome; preoperative assessment

D
ueto the increasing awareness and demand for customized risk prediction scores have emerged in a few
improving the value of care—that is, the patient’s subspecialties of major surgery, including cardiac surgery.7
outcome relative to the cost of care—the frequen- The risk predicting the value of the ASA class has not been
cy and predictors of complications have become of sig- confirmed in cranial neurosurgery, and no customized risk
nificant interest to health care providers. The American scores exist.9 A recent large registry-based study with
Society of Anesthesiologists (ASA) Physical Status Clas- more than 38,000 neurosurgical patients (10,041 cranial
sification system is the most well-known preoperative risk neurosurgery patients) examined the frequency and pre-
estimate used for surgical patients worldwide,3,4 but more dictors of complications in the United States from 2006

Abbreviations  ASA = American Society of Anesthesiologists; BMI = body mass index; CRP = C-reactive protein; ICU = intensive care unit; LOS = length of stay; PT =
prothrombin time.
submitted  August 26, 2014.  accepted  November 7, 2014.
include when citing  Published online May 1, 2015; DOI: 10.3171/2014.11.JNS141970.
Disclosure  This work was supported by personal grants awarded by the Sigrid Jusélius Foundation, Biomedicum Helsinki Foundation, Orion-Farmos Research Founda-
tion, Finnish Medical Foundation and Instrumentarium Research Foundation (to M.K.); the Finnish Society of Anaesthesiologists and Finnish Medical Association (to E.R.);
and the Maire Taponen Foundation (to H.T.).

594 J Neurosurg  Volume 123 • September 2015 ©AANS, 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


In-hospital outcome after elective craniotomy

to 2011.10 This retrospective study could not assess the niotomies during the study period were enrolled only once
role of routine preoperative blood tests, ASA class, or at the time of the first craniotomy.
any other preoperative risk score in outcome prediction.
Furthermore, the reliability of predicting outcomes by Data Collection
using registry-based data has been questioned by others5 Preoperative radiological imaging studies as well as
and even by the authors themselves,10 specifically since in consultations with anesthesiologists and neurosurgeons
many large registries detailed preoperative patient data are adhered to the department’s standard practice. Anesthesia,
lacking. Another major problem is that the type and rate surgery, and postoperative care in the neurosurgical inten-
of complications in neurosurgery are variable but limited. sive care unit (ICU) also followed the normal routines and
Thus, single predictors cannot reliably predict single com- protocols of the department. In brief, the overall treatment
plications.5 was unaffected by this study. Since most of the medical
A traditional view suggests that in cranial neurosur- data during the hospital stay are routinely recorded into
gery, primarily the location and nature of the lesion to- electronic hospital databases and patient records, we re-
gether with the surgeon’s expertise determine the risk of trieved all relevant and additional patient data from these
adverse outcome. However, if adverse outcome in cranial sources when necessary.
neurosurgery is seen more as a multifactorial entity, objec-
tive and patient-related parameters of physical status can Preoperative Patient-Related Data
have an effect on the short-term outcome. Evidence sup-
porting the use of any preoperative risk assessment scores Anesthesiologists or neurosurgical nurses recorded
for predicting outcome in a consecutive series of elective data on age, sex, weight, height, body mass index (BMI),
craniotomy patients is lacking.9 In an attempt to bridge systolic blood pressure, diastolic blood pressure, and heart
this gap in knowledge, we aimed to define a composite rate 1–30 days before the operation at the neurosurgical
of prospectively recorded factors, all routinely recorded in ward or at a preoperative outpatient clinic.
preoperative anesthesia assessments, which could be used Preoperative ASA Classes
in predicting short-term outcome and resource use after
elective cranial neurosurgery. During the preoperative assessment, one of the anes-
thesiologist authors determined the ASA class according
to the original ASA criteria and also according to a local
Methods modification of the ASA (Helsinki ASA) criteria (Table
Study Setting 1). The Helsinki ASA classification was implemented into
This study was reviewed and approved by the ethics clinical use at our institution in the 1990s. The neuroanes-
committee of the Hospital District of Helsinki and Uusi- thesiologists considered the modification essential due to
maa. All subjects signed a written informed consent. the clinic’s patient profile with emphasis on intracranial
We collected a consecutive series of patients under- vascular surgery. The Helsinki ASA classification system
going elective craniotomy in the Department of Neuro- has not been validated and was not modified for the pur-
surgery, Helsinki University Central Hospital, between poses of this study.
December 7, 2011, and December 31, 2012. Helsinki Uni-
versity Central Hospital is a public high-volume hospital Preoperative Blood Tests
complex with a neurosurgical department with more than Preoperative blood tests followed the normal routines
3200 operations annually. Finland has a publicly funded and protocols of the unit and included hemoglobin, plate-
health care system, which offers social and health care lets, creatinine, blood glucose (not fasting), C-reactive
for all citizens (http://www.stm.fi/en/social_and_health_ protein (CRP), sodium, potassium, and prothrombin time
services/health_services). In general, all craniotomies are (PT). These blood tests took place within 1 week before
performed in one of the 5 public university hospitals in the clinical preoperative assessment.
Finland. Helsinki University Central Hospital is the larg-
est of these 5 university hospitals, and its catchment area Follow-Up Data
has nearly 2 million people. Before discharge from the study hospital, one of the
authors recorded the in-hospital mortality, systemic and
Study Population infectious complications, and CNS deficits.
Adult (≥ 18 years) patients scheduled for elective cra-
nial operations under general anesthesia were eligible to End Points
participate in the study. Elective craniotomy was defined Adverse outcomes and resource use were chosen as end
as a craniotomy performed no earlier than 7 days after the points. Adverse outcomes included in-hospital mortality,
treatment decision. Exclusion criteria included 1) a lack in-hospital systemic and infectious complications (deep
of fluency in Finnish or Swedish (the official languages vein thrombosis, pulmonary embolism, acute myocar-
in Finland), 2) age younger than 18 years, 3) epilepsy as dial infarction, new ventricular arrhythmias, cardiogenic
an indication for craniotomy (implantations of electrode shock, cardiac arrest, pneumonia, meningitis, sepsis), and
grids for electrocorticographic recordings and subsequent in-hospital CNS deficits (a new or worsened hemiparesis,
resections of epileptogenic zones), or 4) inability to com- stroke [clinical and/or radiological]; transient or perma-
municate because of severe underlying illness or advanced nent). Resource use was defined as length of stay (LOS) in
cognitive dysfunction. Patients with multiple elective cra- the ICU and hospital.

J Neurosurg  Volume 123 • September 2015 595

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


E. Reponen et al.

TABLE 1. American Society of Anesthesiologists Physical Status Classification and its modifications
ASA ASA Classification System
Class 1941* 1962/1963† Today‡ Helsinki ASA
1 “No organic pathology or patients in whom the patho- “A normal healthy patient.” “A normal healthy A previously healthy patient, age
logical process is localized and does not cause any patient” <65 years
systemic disturbance or abnormality.”
2 “A moderate but definite systemic disturbance, caused “A patient with a mild “A patient with mild A previously healthy, age ≥65
either by the condition that is to be treated by surgi- systemic disease.” systemic disease” years
cal intervention or which is caused by other existing A patient with mild systemic
pathological process, forms this group.” disease
3 “Severe systemic disturbance from any cause or “A patient with a severe “A patient with A patient with severe systemic
causes. It is not possible to state an absolute systemic disease that severe systemic disease
measure of severity, as this is a matter of clinical limits activity, but is not disease” A previously healthy patient <65
judgment.” incapacitating.” years, who has a small unrup-
tured intracranial aneurysm
or a small brain tumor with no
symptoms/mild symptoms
4 “Extreme systemic disorders which have already “A patient with an incapac- “A patient with A patient with unbalanced sys-
become an eminent threat to life regardless of the itating systemic disease severe systemic temic disease
type of treatment. Because of their duration or nature that is a constant threat disease that is a A previously healthy patient with
there has already been damage to the organism that to life.” constant threat a clearly symptomatic intracra-
is irreversible. This class is intended to include only to life” nial aneurysm or brain tumor
patients that are in an extremely poor physical state.
There may not be much occasion to use this clas-
sification, but it should serve a purpose in separating
the patient in very poor condition from others.”
5 “Emergencies that would otherwise be graded in Class “A moribund patient not “A moribund A moribund patient who is not
1 or Class 2.” expected to survive 24 patient who is expected to survive without
hours with or without not expected to the operation
operation.” survive without
the operation”
6 “Emergencies that would otherwise be graded in Class “A declared brain-
3 or Class 4” dead patient
whose organs are
being removed for
donor purposes”
E “In the event of emergency
procedure, precede the
number with an E.”
*  From Saklad: Anesthesiology 2:281–284, 1941.
†  From American Society of Anesthesiologists: Anesthesiology 24:111, 1963.
‡  From American Society of Anesthesiologists website (http://www.asahq.org/resources/clinical-information/asa-physical-status-classification-system).

Statistical Analyses test with 2 × 2 contingency tables enabled the calculation


Given the principal idea of defining a new combination of odds ratios with 95% confidence intervals, sensitivity,
of outcome predictors instead of single independent out- specificity, positive predictive value, and negative pre-
come predictors, we categorized continuous variables to dictive value. Even though it was not our goal to identify
quartiles. The Kruskal-Wallis test provided means of com- single independent predictors, we also performed multi-
paring differences between these quartiles. The Mann- variable regression analyses. In all tests, p values < 0.05
Whitney test permitted post hoc comparisons of each pair were considered significant. The statistical analyses were
in groups after significant Kruskal-Wallis tests. We also carried out using IBM SPSS 21.0 statistical software for
used the Kruskal-Wallis test to see if there were differenc- Mac OS X (version 21.0) and for Windows (version 21.0).
es between ASA scores and resource use (LOS), and for
comparing differences in the sum of the ranks of continu-
ous variables between patients with and without adverse Results
outcomes. The Pearson chi-square test or Fisher’s exact A total of 644 patients underwent an elective crani-
test (cell count ≤ 5) was used to used categorical variables otomy during the study period of nearly 13 months. The
related to dichotomized outcome. The Pearson chi-square final study cohort consisted of 418 (75.9%) of 551 eligible

596 J Neurosurg  Volume 123 • September 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


In-hospital outcome after elective craniotomy

patients. The details on patient enrollment are presented in The mean and median general anesthesia times (defined
the study flow chart (Fig. 1). as the time from the beginning of the administration of
the first anesthetic agent until extubation or transfer to the
Patient Characteristics ICU) were 3 hours 16 minutes (SD 93.2 minutes) and 2
hours 42 minutes (range 52–794 minutes), respectively.
The mean and median ages of the 418 patients were After the surgery, 57.3% of the extubations occurred in
56.4 (SD 13.9 years) and 58.0 years (range 18 to 87 years), the operating room, and the remaining extubations took
respectively, and 260 (62.2%) patients were female. In 124 place in the ICU. For logistic reasons, transferring patients
cases (29.7%), patients were 65 years or older.
to the ICU before extubation is common at our clinic even
in the case of uncomplicated surgical care or the absence
Craniotomies of patient-related factors requiring prolonged intubation.
The most common indications were a vascular lesion Of all the patients, 94.7% were extubated within 6 hours
(for example, an aneurysm or arteriovenous malformation) after the end of surgery.
in 138 (33.0%), a benign tumor in 134 (32.1%), and a malig-
nant tumor in 121 (28.9%) of the 418 operations. The mean
and median operation times (defined as the period from End Points
the time of the surgeon’s first intervention, e.g., beginning The in-hospital mortality rate was 1.0% (4 of 418 pa-
of the sterile preparation until completion of the wound tients). In-hospital systemic or infectious complications
dressings) were 2 hours 40 minutes (SD 91.6 minutes) and occurred in 28 (6.7%) of 418 patients. In-hospital perma-
2 hours 8 minutes (range 42–767 minutes), respectively. nent or transient CNS deficits were recorded in 47 (11.2%)

Fig. 1. Flow chart of patient enrollment in the study.

J Neurosurg  Volume 123 • September 2015 597

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


E. Reponen et al.

TABLE 2. LOSs in the ICU and the hospital in all study patients, Morbidity
patients with systemic and infectious complications, and Older age (Tables 3–5), increased CRP (> 3 mg/L)
patients with CNS deficits (Tables 3 and 5), and the Helsinki ASA class (Table 5)
Mean ICU LOS Mean Hospital were associated with in-hospital systemic and infectious
Patients in Days LOS in Days complications. Patients with BMI values between 25.8 and
29.0 had more in-hospital systemic and infectious compli-
All (n = 418) cations than patients in the lowest and highest quartiles
  Mean ± SD 1.3 ± 1.4 5.6 ± 3.5 (Table 4). No preoperative variable correlated with in-
  Median (range) 1.0 (0–13) 5.0 (1–34) hospital permanent or transient CNS deficits (Tables 3–5).
Systemic & infectious complications
  (n = 28) Resource Use
   Mean ± SD 2.9 ± 3.5 9.8 ± 7.8 Age ≥ 65 years and high Helsinki ASA class (> 2) were
   Median (range) 1.0 (0–13) 7.5 (2–34) associated with increased resource use (Table 5). The old-
est age group (66–87 years) stayed on average 1 day longer
CNS deficits (n = 47)
in the hospital than the youngest age group (18–47 years)
  Mean ± SD 2.7 ± 3.5 7.0 ± 3.8 (Table 6). Patients with preoperative blood glucose levels
  Median (range) 1.0 (1–13) 6.0 (2–17) ≥ 6.7 mmol/L had a longer LOS in the ICU than patients
with levels ≤ 5.7 mmol/L (Table 6). Patients with lowest
(2.9–3.7 mmol/L) and highest (4.2–5.6 mmol/L) potassi-
of 418 patients. Table 2 provides the ICU LOS and overall um levels also stayed longer in the ICU than patients with
hospital LOS for all study patients and in the complica- potassium levels 4.0 to 4.1 mmol/L (Table 6). Moreover,
tion groups. Overall, 90.9% of the patients stayed 24 hours patients with the lowest hemoglobin levels (105–133 g/L)
or less in the ICU. Most of the patients (84.7%) were dis- and patients with the highest systolic blood pressure val-
charged within the 1st postoperative week. ues (156–224 mm Hg) seemed to be discharged margin-
ally later than most of the other patients in the same group
Single End Point Predictors in Univariate Analyses (Table 6).
Mortality
Given the limited number of in-hospital deaths (4 pa- Single End Point Predictors in Multivariable Analyses
tients), no reliable predictors for mortality could be cal- Multivariable logistic regression analyses were con-
culated. However, all 4 in-hospital deaths occurred in ducted using variables that were found to be significant in
patients older than 65 years with a preoperative Helsinki the univariate analyses for morbidity (age, CRP, BMI, and
ASA class of 3 or higher. Helsinki ASA class) and resource use (age, Helsinki ASA
TABLE 3. Preoperative patient-related factors, routine blood tests, and in-hospital outcome*

No. of Systemic & Infectious Complications New CNS Deficits


Factor Patients Yes (range) No (range) p Value Yes (range) No (range) p Value
Age (yrs) 418 64 (29–87) 56 (18–85) 0.002 58 (26–80) 56 (18–87) 0.138
Males 418 8.9% 91.1% 13.9% 86.1%
0.168 0.176
Females 418 5.4% 94.6% 9.6% 90.4%
Weight (kg) 417 77 (52–107) 77 (40–164) 0.448 74 (44–164) 77 (40–155) 0.255
Height (cm) 416 172 (150–187) 169 (148–205) 0.176 170 (55–187) 169 (148–205) 0.832
BMI 416 26 (19–33) 27 (16–47) 0.943 26 (17–47) 27 (16–45) 0.140
SBP (mm Hg) 416 140 (104–197) 143 (95–224) 0.400 147 (108–201) 142 (95–224) 0.066
DBP (mm Hg) 416 85 (57–112) 86 (46–137) 1.00 87 (60–112) 86 (46–137) 0.270
HR (per min) 415 71 (47–111) 70 (41–114) 0.722 72 (49–97) 70 (41–114) 0.209
Hb (g/L) 417 137 (105–157) 141 (106–180) 0.509 140 (105–172) 141 (106–180) 0.937
Platelets (109/L) 417 241 (104–543) 246 (32–498) 0.308 247 (82–543) 245 (32–498) 0.994
Crea (µmol/L) 416 74 (43–107) 70 (5–160) 0.102 68 (43–108) 70 (5–160) 0.408
Gluc (mmol/L) 409 6.4 (4.6–10.5) 6.5 (3.9–24.8) 0.393 7.0 (4.5–24.5) 6.4 (3.9–24.8) 0.151
CRP (mg/L) 412 5.5 (3–16) 5.1 (3–77) 0.034 4.4 (3–25) 5.2 (3–77) 0.816
Na (mmol/L) 417 139 (126–146) 138 (123–158) 0.438 138 (127–146) 139 (123–158) 0.912
K (mmol/L) 417 4.1 (3.4–5.1) 4.0 (2.9–5.6) 0.068 4.0 (3.1–4.8) 4.0 (2.9–5.6) 0.423
PT (%) 410 109 (37–152) 112 (9–170) 0.591 111 (37–166) 112 (9–170) 0.965
Crea = plasma creatinine; DBP = diastolic blood pressure; Gluc = blood glucose (not fasting); Hb = plasma hemoglobin; HR = heart rate; K =
plasma potassium; Na = plasma sodium; SBP = systolic blood pressure.
*  Mean values are reported for continuous variables. The Mann-Whitney test was used for all variables except for sex, which was studied using
the Pearson chi-square test. Values in boldface are statistically significant.

598 J Neurosurg  Volume 123 • September 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


In-hospital outcome after elective craniotomy

TABLE 4. Preoperative patient-related factors, routine blood tests, and in-hospital outcome*
Factor No. of Patients Systemic & Infectious Complications (%) p Value CNS Complications (%) p Value
Age (yrs) 418 0.014† 0.343
  18–47 105 2 (1.9) 10 (9.5)
  48–58 110 6 (5.5) 9 (8.2)
  59–65 93 6 (6.5) 11 (11.8)
  66–87 110 14 (12.7) 17 (15.5)
Sex 418 0.168 0.176
  Male 158 14 (8.9) 22 (13.9)
  Female 260 14 (5.4) 25 (9.6)
Weight (kg) 417 0.454 0.596
  40–64 97 5 (5.2) 12 (12.4)
  65–74 113 5 (4.4) 13 (11.5)
  75–86 104 9 (8.7) 14 (13.5)
  87–164 103 9 (8.7) 8 (7.8)
Height (cm) 416 0.105 0.549
  148–162 115 7 (6.1) 15 (13.0)
  163–169 96 2 (2.1) 7 (7.3)
  170–176 102 8 (7.8) 13 (12.7)
  177–205 103 11 (10.7) 12 (11.7)
BMI 416 0.024‡ 0.193
  16.4–23.4 103 4 (3.9) 14 (13.6)
  23.5–25.7 105 8 (7.6) 12 (11.4)
  25.8–29.0 104 13 (12.5) 15 (14.4)
  29.1–47.4 104 3 (2.9) 6 (5.8)
SBP (mm Hg) 416 0.842 0.063
  95–128 106 9 (8.5) 9 (8.5)
  129–141 103 6 (5.8) 9 (8.7)
  142–155 105 6 (5.7) 10 (9.5)
  156–224 102 7 (6.9) 19 (18.6)
DBP (mm Hg) 416 0.615 0.700
  46–78 101 8 (7.9) 10 (9.9)
  79–86 113 5 (4.4) 11 (9.7)
  87–93 105 9 (8.6) 12 (11.4)
  94–137 97 6 (6.2) 14 (14.4)
HR (per min) 415 0.542 0.387
  41–60 106 6 (5.7) 9 (8.5)
  61–68 102 7 (6.9) 11 (10.8)
  69–78 104 10 (9.6) 10 (9.6)
  79–114 103 5 (4.9) 16 (15.5)
Hb (g/L) 417 0.841 0.708
  105–133 108 8 (7.4) 14 (13.0)
  134–140 100 7 (7.0) 8 (8.0)
  141–147 104 5 (4.8) 12 (11.5)
  148–180 105 8 (7.6) 12 (11.4)
Platelets (109/L) 417 0.349 0.400
  32–201 104 6 (5.8) 14 (13.5)
  202–238 105 11 (10.5) 7 (6.7)
  239–285 103 6 (5.8) 13 (12.6)
  286–543 105 5 (4.8) 12 (11.4)
(continued)

J Neurosurg  Volume 123 • September 2015 599

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


E. Reponen et al.

TABLE 4. Preoperative patient-related factors, routine blood tests, and in-hospital outcome* (continued)
Factor No. of Patients Systemic & Infectious Complications (%) p Value CNS Complications (%) p Value
Crea (μmol/L) 416 0.290 0.428
  5–59 104 5 (4.8) 15 (14.4)
  60–68 106 5 (4.7) 10 (9.4)
  69–77 102 7 (6.9) 8 (7.8)
  78–160 104 11 (10.6) 13 (12.5)
Gluc (mmol/L) 409 0.701 0.337
  3.9–5.3 111 5 (4.5) 10 (9.1)
  5.4–5.7 87 6 (6.9) 9 (10.3)
  5.8–6.6 105 7 (6.1) 10 (9.5)
  6.7–24.8 106 9 (9.3) 17 (16.0)
Na (mmol/L) 417 0.077 0.842
  123–136 92 8 (8.7) 10 (10.9)
  137–139 143 4 (2.8) 14 (9.8)
  140 56 3 (5.4) 8 (14.3)
  141–148 126 13 (10.3) 14 (11.1)
K (mmol/L) 417 0.178 0.297
  2.9–3.7 91 5 (5.5) 14 (15.4)
  3.8–3.9 106 5 (4.7) 11 (10.4)
  4.0–4.1 102 5 (4.9) 7 (6.9)
  4.2–5.6 118 13 (11.0) 14 (11.9)
PT (%) 410 0.645 0.715
  9–96 104 8 (7.7) 9 (8.7)
  97–108 103 8 (7.8) 14 (13.6)
  109–128 100 4 (4.0) 12 (12.0)
  129–170 103 8 (7.8) 11 (10.7)
*  Continuous variables were divided into fourths on the basis of quartiles. CRP could not be divided since three-quarters of the patients had
CRP less than 4 mg/L. The Pearson chi-square test was used to test for independence. The post hoc test for the significant results was done on
each pair using the Pearson chi-square 2 × 2 contingency table. Values in boldface are statistically significant.
†  Youngest age group (18­– 47 years) differed from the oldest (66–87 years) age group (p = 0.003).
‡  Lowest (16.4–23.4) and highest (29.1–47.4) BMI groups differed from the 3rd BMI group (25.8–29.0) (p = 0.024 and p = 0.009, respectively).

class, systolic blood pressure, hemoglobin, glucose, and 7). If a patient ≥ 60 years had a preoperative Helsinki ASA
potassium). For systemic and infectious complications, class of 4, the LOS in the ICU was nearly 40% and in hos-
CRP remained significant after bivariate logistic regres- pital approximately 20% longer than in patients without
sion analyses only when treated as a categorized (dichot- this combination of factors (Table 8).
omized) covariate, whereas age retained its significance
whether treated as a categorical or a continuous covariate Discussion
(results not shown). Helsinki ASA classification (categori-
In this first prospective study on outcome predictors in
cal) also remained significant in all possible multivariable
an unselected cohort of patients undergoing cranial neuro-
regression analyses (results not shown). Using linear re-
surgery, higher Helsinki ASA class (> 2), increased age (>
gression analyses for continuous outcome variables, Hel-
60–65 years) and abnormal CRP values (> 3 mg/L) pre-
sinki ASA classification (categorical) was the only vari-
dicted postoperative in-hospital systemic and infectious
able that remained a significant predictor for both ICU
complications. A composite factor of the Helsinki ASA
LOS and hospital LOS (results not shown).
class, age, and CRP value identified patients with up to
nearly 5 times higher odds ratios for postoperative system-
Composite End Point Predictor ic and infectious complications preoperatively (Table 7).
Various combinations of significant end point predic- In other words, it was able to identify almost every fourth
tors were evaluated for the identification of high-risk pa- patient with major postoperative systemic or infectious
tients for complications and increased resource use (results complications (Table 7). The composite factor also pre-
not shown). Almost one-fourth of the patients with Helsin- dicted a prolonged stay in the ICU (Table 8). No preopera-
ki ASA Class 4, age ≥ 60 years, and CRP values > 3 mg/L tive assessment parameter predicted postoperative CNS
had systemic or infectious complications after cranial sur- complications, which probably are more lesion and sur-
gery, with up to nearly 5 times higher odds ratios (Table geon-related complications than systemic and infectious
600 J Neurosurg  Volume 123 • September 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


In-hospital outcome after elective craniotomy

TABLE 5. Major risk predictors and study end points*


Systemic & Infectious
No. of Complications New CNS Deficits ICU LOS Hospital LOS
Factor Patients (%) Yes (%) No (%) p Value Yes (%) No (%) p Value (days) p Value (days) p Value
ASA class 0.213 0.408 0.095 0.389
 1 67 (16.1) 2 (3.0) 65 (97.0) 7 (10.4) 60 (89.6) 1.09 5.12
 2 152 (36.5) 8 (5.3) 144 (94.7) 19 (12.5) 133 (87.5) 1.42 5.53
 3 166 (39.8) 14 (8.4) 152 (91.6) 15 (9.0) 151 (91.0) 1.16 5.64
  4 32 (7.7) 4 (12.5) 28 (87.5) 6 (18.8) 26 (81.3) 1.84 6.34
Helsinki ASA class 0.021† 0.663 0.036‡ 0.007§
 1 4 (1.0) 0 4 (100.0) 0 4 (100.0) 1.0 4.0
 2 94 (22.5) 1 (1.1) 93 (98.9) 9 (9.6) 85 (90.4) 1.12 4.82
 3 251 (60.2) 18 (7.2) 233 (92.8) 28 (11.2) 223 (88.8) 1.22 5.65
  4 68 (16.3) 9 (13.2) 59 (86.8) 10 (14.7) 58 (85.3) 1.87 6.35
Age in yrs 0.004 0.169 0.027 0.008
  ≥65 124 (29.7) 15 (12.1) 109 (87.9) 18 (14.5) 106 (85.5) 1.35 6.06
  <65 294 (70.3) 13 (4.4) 281 (95.6) 29 (9.9) 265 (90.1) 1.28 5.36
CRP in mg/L 0.046 0.835 0.689 0.889
  >3 106 (25.7) 11 (10.4) 95 (89.6) 11 (10.4) 95 (89.6) 1.32 5.60
  ≤3 306 (74.3) 15 (4.9) 291 (95.1) 34 (11.1) 272 (88.9) 1.25 5.34
*  The Pearson chi-square test was used for 2 × 2 and 2 × k contingency tables of categorical variables. The Kruskal-Wallis test was used for studying differences in
LOS. The post hoc Mann-Whitney test was used on each pair of significant groups. Values in boldface are statistically significant.
†  ASA Helsinki Class 2 differed from ASA Helsinki Classes 3 (p = 0.027) and 4 (p = 0.002).
‡  ASA Helsinki Class 4 differed from ASA Helsinki Classes 2 (p = 0.015) and 3 (p = 0.013).
§  ASA Helsinki Class 2 differed from ASA Helsinki Classes 3 (p = 0.022) and 4 (p = 0.001).

complications. Focusing on the preoperative identification modern cranial neurosurgery, identifying single indepen-
of patients susceptible to in-hospital complications was the dent outcome predictors was beyond the scope of our study.
rationale behind the decision not to focus and report on Instead, we aimed to find a preoperative composite factor
numerous other perioperative factors that may contribute that should act as a red flag and suggest a more thorough
to outcome, such as the choice of anesthetic agent, intra- evaluation and individually tailored perioperative plan for
operative temperature control, type of antibiotic prophy- patients at high risk for postoperative systemic and infec-
laxis, or postoperative analgesia. In brief, it is practically tious complications in an unselected cohort of patients
impossible to control all theoretical confounding factors undergoing elective cranial neurosurgery. Identifying a
in similar outcome studies. Furthermore, these results are preoperative composite factor with a reasonable positive
in keeping with the findings based on the recent system- predictive value can be sensible even in low caseload units,
atic review, suggesting that preoperative ASA class can be since it seems unlikely that a single preoperative outcome
useful in predicting other than lesion and surgeon-related predictor for all neurosurgical patients would be found.
complications.9
ASA Class and Composite Preoperative End Point
Role of Study Design and Clinical Applicability Predictor
An era of big data analyses in health care has already The ASA Physical Status Classification system was
started. Despite the potential benefits of the big data rev- first introduced in the 1940s and revised in 1963, and it is
olution, the value of any results relies on validity of the the most widely known and used system for preoperative
data used. When identifying new associations with surgi- risk assessment.1,11 The Helsinki ASA modification imple-
cal outcomes, most large registries and databases cannot mented approximately 2 decades ago was based on clini-
provide us with detailed and useful results. In the recent cal observations and judgments by neuroanesthesiologists.
retrospective study of more than 10,000 cranial neurosur- Ever since, it has been used without a proper validation.
gery patients,10 the incomplete reflection of patient charac- Our study suggests that this call was justified, even though
teristics obtained from the database as well as the lack of the precise criteria for the change are largely unknown.
fit for the complications unique to neurosurgical patients However, the modification takes into account some of the
hindered the identification of useful predictors for adverse characteristics of the unique patient population in cranial
outcome, as discussed by the authors. Furthermore, the re- neurosurgery, and these characteristics are embedded in
liability of evaluating the quality of care based on such the Helsinki ASA class but not in the original ASA class.
registries is compromised by the rarity of outcome events Bearing in mind the original purpose of the ASA classi-
along with the low caseload in many individual hospitals.5 fication, the ASA class should perhaps be seen more as a
Considering the low rates of morbidity and mortality in representative of the burden of systemic disease at a given
J Neurosurg  Volume 123 • September 2015 601

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


E. Reponen et al.

TABLE 6. Preoperative patient-related factors, routine blood TABLE 6. Preoperative patient-related factors, routine blood
tests, and LOS* tests, and LOS* (continued)
No. of LOS in LOS in No. of LOS in LOS in
Factor Patients ICU p Value hospital p Value Factor Patients ICU p Value hospital p Value
Age (yrs) 418 0.303 0.029† Platelets (109/L) 417 0.416 0.500
  18–47 105 1.2 4.9   32–201 104 1.3 5.8
  48–58 110 1.4 5.3   202–238 105 1.2 5.5
  59–65 93 1.4 6.3   239–285 103 1.2 5.4
  66–87 110 1.3 4.9   286–543 105 1.3 5.6
Sex 418 0.861 0.496 Crea (µmol/L) 416 0.143 0.729
  Male 158 1.4 5.6   5–59 104 1.4 5.6
  Female 126 1.3 5.6   60–68 106 1.1 5.7
Weight (kg) 417 0.145 0.191   69–77 102 1.4 5.7
  40–64 97 1.4 6.0   78–160 104 1.3 5.4
  65–74 113 1.2 5.4 Gluc (mmol/L) 409 0.012¶ 0.485
  75–86 104 1.2 5.6   3.9–5.3 111 1.1 5.7
  87–164 103 1.4 5.4   5.4–5.7 87 1.2 5.5
Height (cm) 416 0.667 0.112   5.8–6.6 105 1.4 5.6
  148–162 115 1.4 6.0   6.7–24.8 106 1.5 5.5
  163–169 96 1.2 4.8 Na (mmol/L) 417 0.117 0.295
  170–176 102 1.2 5.6   123–136 92 1.5 6.5
  177–205 103 1.5 5.7   137–139 143 1.2 5.2
BMI 416 0.406 0.545   140 56 1.4 5.0
  16.4–23.4 103 1.4 5.9   141–148 126 1.3 5.6
  23.5–25.7 105 1.4 5.8 K (mmol/L) 417 0.025** 0.826
  25.8–29.0 104 1.2 5.3   2.9–3.7 91 1.3 5.2
  29.1–47.4 104 1.2 5.2   3.8–3.9 106 1.3 6.0
SBP (mm Hg) 416 0.189 0.029‡   4.0–4.1 102 1.2 5.1
  95–128 106 1.1 5.3   4.2–5.6 118 1.4 5.9
  129–141 103 1.3 5.1 PT % 410 0.746 0.293
  142–155 105 1.2 5.7   9–96 104 1.1 4.9
  156–224 102 1.5 6.2   97–108 103 1.5 6.3
DBP (mm Hg) 416 0.556 0.914   109–128 100 1.2 5.5
  46–78 101 1.2 5.4   129–170 103 1.3 5.7
  79–86 113 1.1 5.6 *  Continuous variables were divided into fourths (quartiles). CRP values
  87–93 105 1.4 5.7 could not be divided into fourths since three-fourths of the patients had normal
  94–137 97 1.4 5.6 CRP values (≤ 3 mg/l). The Kruskal-Wallis test was used to study differences
between fourths. The post hoc Mann-Whitney test was used on each pair of
HR (per min) 415 0.343 0.387 significant groups. Values in boldface are statistically significant.
  41–60 106 1.1 5.7 †  Age group 4 (66–87 years) differed from age groups 1 (18–47 years; p =
  61–68 102 1.2 5.6 0.019) and 2 (48–58 years; p = 0.020).
‡  SBP group 2 (129–141 mm Hg) differed from SBP group 4 (156–224 mm Hg;
  69–78 104 1.3 5.7 p = 0.004).
  79–114 103 1.5 5.2 §  Hb group 1 (105–133 g/L) differed from Hb groups 2 (134–140 g/L; p =
Hb (g/L) 417 0.317 0.019§ 0.010) and 4 (148–180 g/L; p = 0.013).
¶  Blood glucose level group 4 (6.7–67.0 mmol/L) differed from blood glucose
  105–133 108 1.4 6.2
level groups 1 (3.9–5.3 mmol/L; p = 0.009) and 2 (5.4–5.7 mmol/L; p = 0.020).
  134–140 100 1.2 5.0 **  Potassium level group 3 (4.0–4.1 mmol/L) differed from groups 1 (2.9–3.7
  141–147 104 1.3 5.6 mmol/L; p = 0.033) and 4 (4.2–5.6 mmol/L; p = 0.003).
  148–180 105 1.3 5.5
(continued) complications but not with new or worsening neurologi-
cal deficits. Quantification of the systemic and infectious
moment, not as a predictor of perioperative in-hospital postoperative risks and making that quantification a part
neurological deficits. Our study results align with this of surgical decision making may guide the selection of the
view, as the preoperative Helsinki ASA score correlated safest treatment modality and resource allocation for an
with postoperatively recorded systemic and infectious individual patient. For resource management, the use of
602 J Neurosurg  Volume 123 • September 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


In-hospital outcome after elective craniotomy

TABLE 7. Various combinations of risk predictors and systemic and infectious complications
No. of p
Combinations Patients OR (95% CI) Value* PPV NPV Spec Sens
Combination 1: Helsinki ASA Class 4 & age ≥60 yrs 44 3.93 (1.62–9.57) 0.001 18.2% 94.7% 90.8% 28.6%
Combination 2: Helsinki ASA Class 4, & age ≥60 yrs, 17 4.82 (1.46–15.91) 0.005 23.5% 94.0% 96.7% 14.3%
  & CRP >3 mg/L
Combination 3: Helsinki ASA Class 4 & CRP >3 mg/L 28 2.54 (0.81–7.90) 0.097 14.3% 93.8% 93.8% 14.3%
NPV = negative predictive value; PPV = positive predictive value; sens = sensitivity; spec = specificity.
*  Boldface indicates statistical significance.

a simple combination of preoperative factors can provide tution-specific phenomenon. Unfortunately, the scope of
a tool to estimate the overall hospital costs for each pa- the study failed to include determining the interrater vari-
tient. For example, the combination of Helsinki ASA Class ability of the original or Helsinki ASA classifications, so
4 and age ≥ 60 years carries an estimated extra cost of we cannot exclude the possibility that the Helsinki ASA
€1500–€2800 per patient during the in-hospital period at classification is subject to variability, or that it is insti-
our institution. The combination of Helsinki ASA class, tutionally biased. However, detailed instructions on de-
age, and CRP in risk prediction needs to be validated in termining both ASA classifications were implemented in
a different cohort of elective craniotomy patients in the the study protocol, and the Helsinki ASA classification is
future. Combining Helsinki ASA Classes 1 and 2 may be comparable to the original ASA classification in terms of
appropriate, since only 4 patients (1%) were classified as its applicability in a clinical scenario. The study can also
Helsinki ASA Class 1 in this study. be criticized for being underpowered and susceptible to
statistical errors. Our data were prospectively collected
Strengths and Weaknesses of the Study and tailored to the specific characteristics of intracranial
The study may have a few advantages. The study was neurosurgery and represent a full year’s case mix at our
based on assessments made by anesthesiologists, who ap- institution. Moreover, the objective was not to evaluate the
parently have no incentives to manipulate any of the study predictive value of single independent factors but rather
measures. In addition to the prospective study design, the to find a combination, i.e., a new widely applicable com-
broad inclusion criteria probably increase the external va- posite factor, which is associated with a greater risk of
lidity of the results: i.e., these results may be applicable to in-hospital complications. We aimed at simplicity in our
many Western neurosurgical units. There are also some statistical approach, but multivariable regression analy-
limitations and drawbacks to this study. Regardless of ses were performed to supplement the analyses. As an-
its simplicity and wide applicability, the original ASA other possible drawback, patients’ comorbidities were
Physical Status Classification has been criticized for its not included as possible contributors to outcome, similar
considerable interrater variability.2,6,8 The Helsinki ASA to the previous registry-based study.10 The main reason
classification may suffer from the same limitation. Most for excluding such data in the analyses was to increase
importantly, the Helsinki ASA system has not been vali- the external validity of the results. In addition to highly
dated previously, and therefore it may represent an insti- varying incidences and prevalence, comorbidities, such
as hypertension, myocardial disease, and renal insuffi-
TABLE 8. Various combinations of risk predictors and ciency, are diagnosed and treated differently in different
resource use countries and institutions, and thus inclusion of these data
would probably decrease the generalizability of the re-
Mean sults. Furthermore, ASA classification indirectly includes
No. of Mean ICU p Hospital p the patient’s comorbidity status. As our main purpose was
Combination* Patients LOS Value† LOS Value† to study the general applicability of various anesthesio-
logical preoperative assessments in predicting in-hospital
Combination 1 0.018 0.004
outcome, the use of perioperative surgical parameters
  Yes   44 1.95 6.86 (such as surgeon’s experience, location of the lesion, pre-
 No 374 1.22 5.41 operative diagnosis, surgical positions, and surgical time)
Combination 2 0.164 0.216 as study inclusion criteria was deemed unnecessary. Fur-
  Yes   17 2.12 6.71 thermore, these factors are absent in previously published
 No 400 1.26 5.52 outcome scores.9 In addition, the presented predictors of
Combination 3 0.077 0.135
outcome (i.e., the Helsinki ASA class, age, and CRP) are
relatively objective and applicable in most Western coun-
  Yes   28 1.82 6.25 tries with life expectancies similar to those in Finland.
 No 389 1.26 5.52 Finally, the low mortality and morbidity rates can perhaps
*  Combination 1: Helsinki ASA Class 4 and age ≥ 60 years; Combination 2: be considered major drawbacks of our study, as statistical
Helsinki ASA = 4, age ≥ 60 years, and CRP > 3 mg/L; and Combination 3: analyses on mortality-associated factors were unreliable,
Helsinki ASA Class 4 and CRP > 3 mg/L. and no detailed subgroup analyses could be performed
†  Boldface indicates statistical significance. on the morbidity-related predictors. This may also have
J Neurosurg  Volume 123 • September 2015 603

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC


E. Reponen et al.

contributed to the fact that none of the factors assessed in   5. Krell RW, Hozain A, Kao LS, Dimick JB: Reliability of
this study predicted new neurological deficits. risk-adjusted outcomes for profiling hospital surgical quality.
JAMA Surg 149:467–474, 2014
  6. Mak PH, Campbell RC, Irwin MG: The ASA Physical Status
Conclusions Classification: Inter-observer consistency. Anaesth Intensive
Our results suggest that using a simple composite factor Care 30:633–640, 2002
  7. Nashef SAM, Roques F, Michel P, Gauducheau E, Lemeshow
as a risk estimate of complications and resource use, health S, Salamon R: European system for cardiac operative risk
care professionals may be able to preoperatively identify evaluation (EuroSCORE). Eur J Cardiothorac Surg 16:9–
patients requiring extra cautions and tailored care. Con- 13, 1999
trary to the traditional view that the location and nature of   8. Ranta S, Hynynen M, Tammisto T: A survey of the ASA
the lesion and the expertise of the surgeon determine the Physical Status Classification: significant variation in alloca-
risk of an adverse outcome in brain surgery, our results tion among Finnish anaesthesiologists. Acta Anaesthesiol
suggest that it is possible to identify patients prone to sys- Scand 41:629–632, 1997
  9. Reponen E, Tuominen H, Korja M: Evidence for the use of
temic and infectious complications with simple preopera- preoperative risk assessment scores in elective cranial neuro-
tive assessments. Even though large registry-based studies surgery: a systematic review of the literature. Anesth Analg
are plausible also in Finland, a customized prospective 119:420–432, 2014
study setting is imperative in establishing the value of 10. Rolston JD, Han SJ, Lau CY, Berger MS, Parsa AT: Frequen-
preoperative factors in predicting postoperative complica- cy and predictors of complications in neurological surgery:
tions. When such prospective studies are piloted in large national trends from 2006 to 2011. J Neurosurg 120:736–
neurosurgical centers, nationwide and international multi- 745, 2014
center studies can be implemented to substantiate the find- 11. Saklad M: Grading of patients for surgical procedures. Anes-
thesiology 2:281–284, 1941
ings. Thus, the results and morbidity rates presented here
may perhaps serve as first objective benchmark numbers
in the modern neurosurgical era and a guide to safeguard-
ing good practice.
Author Contributions
Conception and design: Reponen, Korja, Niemi, Tuominen.
Acknowledgment Acquisition of data: Reponen, Silvasti-Lundell, Tuominen.
We would like to acknowledge Associate Professor Markus Analysis and interpretation of data: Reponen, Korja. Drafting the
Skrifvars for his help in preparing the patient questionnaires and article: Reponen, Tuominen. Critically revising the article: Korja,
his expert advice on IBM SPSS software. Niemi, Silvasti-Lundell, Hernesniemi, Tuominen. Reviewed
submitted version of manuscript: all authors. Approved the final
version of the manuscript on behalf of all authors: Reponen.
Statistical analysis: Reponen, Korja. Administrative/technical/
References material support: Niemi, Hernesniemi. Study supervision: Korja,
  1. American Society of Anesthesiologists: New classification of Niemi, Hernesniemi, Tuominen.
physical status. Anesthesiology 24:111, 1963
  2. Aronson WL, McAuliffe MS, Miller K: Variability in the Supplemental Information
American Society of Anesthesiologists Physical Status Clas-
sification Scale. AANA J 71:265–274, 2003
Previous Presentation
  3. de Cássio Zequi S, de Campos ECR, Guimarães GC, Bache- The results of this study were presented in part in abstract/poster
ga W Jr, da Fonseca FP, Lopes A: The use of the American form at Euroanaesthesia 2013 in Barcelona, Spain, June 4, 2013,
Society of Anesthesiology Classification as a prognostic fac- and at the 32nd Scandinavian Society of Anaesthesiology and
tor in patients with renal cell carcinoma. Urol Int 84:67–72, Intensive Care Medicine Congress in Turku, Finland, August 28,
2010 2013.
  4. Hightower CE, Riedel BJ, Feig BW, Morris GS, Ensor JE
Jr, Woodruff VD, et al: A pilot study evaluating predictors Correspondence
of postoperative outcomes after major abdominal surgery: Elina Reponen, Department of Anaesthesiology and Intesive Care
Physiological capacity compared with the ASA Physical Sta- Medicine, Helsinki University Central Hospital, P.O. Box 900,
tus Classification system. Br J Anaesth 104:465–471, 2010 Helsinki 00029 HUS, Finland. email: elina.reponen@hus.fi.

604 J Neurosurg  Volume 123 • September 2015

Unauthenticated | Downloaded 01/30/21 11:55 PM UTC

You might also like