ORIGINAL PAPER

Predictors of poor outcome of decompressive craniectomyin pediatric patients with severe traumatic brain injury:a retrospective single center study from Pakistan

Saad Akhtar Khan & Hussain Shallwani & Muhammad Shahzad Shamim &

Ghulam Murtaza & Syed Ather Enam & Reema Obaid Qureshi &Muhammad Zubair Tahir

Received: 22 October 2012 /Accepted: 2 July 2013# Springer-Verlag Berlin Heidelberg 2013

AbstractObjective This study aimed to determine the risk factorsassociated with poor outcome of decompressive craniectomy(DC) for severe traumatic brain injury (TBI) in pediatricpatients.Methods This retrospective study is conducted on pediatricpopulation (age 1–15 years) presenting with TBI whounderwent DC at our institute between January 2000 and2010. Based on Glasgow outcome score (GOS) at a mini-mum follow-up of 5 months, patients were divided into twogroups, namely poor outcome (GOS 1, 2, and 3) and goodoutcome (GOS 4 and 5). Records were reviewed and ana-lyzed for preoperative and intraoperative predictors.Results We found 25 patients who were eligible as per se-lection criteria. Mean age at presentation was 6±4 years andthere was male preponderance (84 %). Fall (60 %) was themost common mechanism of injury followed by gunshotsand road traffic accident. On univariate analysis, presentingGCS ≤5 (p value=0.009), delay in presentation of more than150 min (p value=0.010), DC performed after more than 4 hof arrival in hospital (p value=0.042), and intraoperativeblood loss exceeding 300 ml (p value=0.001) were signifi-cant predictors of poor outcome.Conclusion Our study suggests that DC in children is notonly a life-saving procedure, but also leads to a good func-tional outcome after severe injury. However, patient selec-tion still remains an important aspect, and the above-mentioned factors should be considered while deciding forDC to improve survival. Further prospective studies on larg-er sample size are warranted to validate our results.

Keywords Traumatic brain injury . Decompressivecraniectomy . Pediatric population

Introduction

Reported incidence of severe traumatic brain injury (TBI) is5 % of all TBI, and mortality in severe TBI ranges from 34 to60 % [1]. TBI in pediatric patients differs from that in adultsdue to certain anatomical and biomechanical factors. Thepediatric cranium is relatively soft and thus pediatric patientstend to have more white matter shear injuries and fewer masslesions. Moreover, cerebral edema is found twice more com-monly in pediatric patients due to cerebral hyperemia [5, 25].

The role of decompressive craniectomy (DC) in pediatricTBI has remained a subject of debate, even though the firstever randomized controlled trial on DC and its role in TBIwas actually published on pediatric patients [7, 20]. This trialshowed favorable outcome with DC when compared withconservative management, although the numbers were smalland the results were not statistically significant [20]. Sincethen, several trials have been conducted, or are in process, toevaluate the role of DC in adults with stroke and TBI [11,21]; however, little more than case series has been added tothe literature on the role of DC in pediatric victims of TBI.We have undertaken this study to determine the risk factorsof poor outcomes in pediatric patients undergoing DC at atertiary care hospital of a developing country, Pakistan.

Methods

This retrospective study was conducted at our universityhospital over a period of 10 years (January 2000–January2010). We included pediatric patients of age 1–15 years, who

S. A. Khan :H. Shallwani :M. S. Shamim (*) :G. Murtaza :S. A. Enam :R. O. Qureshi :M. Z. TahirSection of Neurosurgery, Department of Surgery, Aga KhanUniversity Hospital, Karachi, Pakistane-mail: shahzad.shamim@aku.edu

Childs Nerv SystDOI 10.1007/s00381-013-2225-2

presented with isolated TBI and underwent DC. Patients withsignificant systemic injuries (that required any intervention orhad potential to alter the outcome of patients), or those whorequired a form of craniotomy but had their bone flaps imme-diately restored (such as patients with a depressed skull frac-ture or an extradural hematoma), were not included. Recordswere retrieved using ICD-9 coding. Seven hundred forty-threepatients had presented with TBI; 109 of them had undergone acranial procedure other than DC and only 25 patients hadundergone DC. Outcome was defined on the basis of theGlasgow outcome score (GOS) and final outcome wasassessed at 5 months follow-up for survivors; patients weredivided into two groups, namely poor outcome group (GOS 1,2, and 3) and good outcome group (GOS 4 and 5).

As an institutional protocol, all patients presenting withTBI in the emergency department with GCS of 8 or less areintubated and managed in the Pediatric Intensive Care Unit(PICU), after getting an immediate CT scan of brain. We donot routinely perform intracranial pressure monitoring forpatients with TBI due to financial constraints and non-availability of equipment, although very occasionally, anexternal ventricular drain has been used. Only two patientsin our sample received an external ventricular drain. Thedecision for surgery is based on the patient’s hemodynamicstatus, pre-intubation neurological examination, and radio-logical findings on neuroimaging. The decision is taken bythe on-call senior neurosurgical resident in his final years oftraining, in consultation with the neurosurgical fellow, andalways confirmed and consented by the attending physician.Patients who require immediate surgery are taken to theoperating room at the earliest; the rest are managed medical-ly, guided by serial CT scans of brain. The DC procedure isreserved for patients with diffuse brain swelling, with clinicaland radiological signs of herniation. Nine patients underwentDC based on this initial assessment. Patients with mass lesionssuch as intracranial hematomas causing significant midlineshift may also undergo DC if the intraoperative impression ofthe surgeon suggests massive swelling of brain where replac-ing the bone flap is generally not advisable. Fifteen patientsunderwent DC based on intraoperative findings. Patients with-out obvious mass lesions but with intracranial hypertension(based on radiological findings of loss of sulci, cisterns) andpersistently low GCS scores are managed in the ICU andtreated on the basis of Lund protocol, hypertonic saline, serialneurological examinations, and serial neuroimaging. Thesepatients may later undergo DC in case of medically refractoryintracranial hypertension, worsening neurological status suchas new onset pupillary abnormalities, or development of amass lesion such as flared contusions or expanding hematomacausing midline shift on neuroimaging. Only one such patientwas included in our study.

In the case of unilateral mass lesion causing midline shift, alarge fronto-temporo-parietal flap is raised. Diffuse swelling

or bilateral injury is dealt with large bifrontal craniectomies.Dura mater is opened in all patients in a stellate fashion, andexpansile duraplasty is done using pericranium or temporalisfascia. All bone flaps are cryopreserved. Postoperatively pa-tients are kept intubated and managed in the PICU. Patientsrequiring ventilation for more than 72 h undergo tracheostomyby pediatric general surgeons. After extubation or tracheosto-my, patients are shifted to high dependency care unit (HDU),and once stable, patients are shifted out of the HDU andmanaged accordingly.

Being a retrospective review, the study protocol wasexempted from ethical review committee approval. Data werecollected by a neurosurgery resident on a pro forma thatincluded demographic, pre-/intra-/postoperative variablesand outcomes. The data regarding demographics were takenfrom the phase sheet of each patient, preoperative variablesfrom trauma form note of on-call neurosurgery resident,intraoperative variables from operative note, and postopera-tive variables from daily progress notes and clinic follow-ups.

Statistical analysis

Data was analyzed using SPSS 19.0. Continuous variableswith normal distribution are presented as mean ± standarddeviation, while those with skewed distribution are presentedas median and interquartile ranges. All continuous variableswere converted into categories while looking at frequencydistribution or as reported in literature, and data were ana-lyzed using chi-square or Fisher’s exact test, whicheverapplicable. A p value <0.05 was taken as significant.

Results

We found 25 patients whowere eligible as per selection criteria.Mean age at presentation was 6±4 years and there was malepreponderance (84 %). Fifteen out of 25 patients had a historyof fall (60%)making it the most commonmechanism of injury,followed by gunshots and road traffic accident. At presentation,mean GCS was 6 (±3 SD) and median time from event toarrival in our emergency room was 100 (15–440)min. Cerebraledema (80%), fractures (60%), and subdural hematoma (24%)were the common pathologies on neuroimaging. Median timefrom arrival to DC was 3 h (1–25).

Outcome assessed at a mean follow-up of 5 (±2 SD)-months showed that nine patients (36 %) died (GOS 1) post-operatively, while all remaining patients showed good func-tional outcome (64 %). Of the survivors, good recovery (GOS5) was observed in ten patients (40 %), and moderate disabil-ity (GOS 4) in six patients (24 %). No surviving patient hadvegetative state (GOS 2) or severe disability (GOS 3).

Comparison of poor outcome group (n=9) with the goodoutcome group (n=16) revealed that presenting GCS of 5 or

Childs Nerv Syst

less (p value=0.009), delay in presentation of more than150 min (p value=0.010), DC performed after more than4 h of arrival in hospital (p value=0.042), and intraoperativeblood loss exceeding 300 ml (p value=0.001) were signifi-cant predictors of poor outcome (Tables 1 and 2). The pres-ence of a single skull fracture showed tendency as a predictorof poor outcome when compared with multiple fractures,which had better outcome (p value=0.069).

Discussion

In this retrospective study, we found GCS of ≤5, delay inpresentation of more than 150 min, DC performed after morethan 4 h of arrival in hospital, and intraoperative blood lossexceeding 300 ml as the significant predictors of poor

outcome. None of the patients had vegetative state, signify-ing that the outcomes of DC are good if it is undertakenwithin due course of time after injury; otherwise, severe TBIin pediatric population is fatal.

DC was first described by Kocher in 1901 and isperformed as a life-saving measure to protect the injuredbrain from the damaging effects of propagating edemaand intracranial hypertension failing to respond to opti-mal medical management [22]. Sustained intracranial hy-pertension carries a poor prognosis with reported mortal-ity exceeding 80 % in some series [3]. DC has emergedas a useful option in the treatment of such cases, andthere is growing evidence of its utility in the manage-ment of several pathologies including TBI and strokes,although majority of literature has addressed the problemin adult patients [8, 11, 21].

Table 1 Comparison of preop-erative variables between thepoor outcome group and goodoutcome group

Variable Total (n=25) Poor outcomegroup (n=9)

Good outcomegroup (n=16)

p value

Age (mean) 6.76±4.25 6.33±4.9 7.01±3.9 0.2591–5 years 12 6 6

6–10 years 6 0 6

11–15 years 7 3 4

GCS at presentation (mean) 6.20±2.739 4.22±1.64 7.31±2.62 0.0093–5 11 8 3

6–8 10 1 9

9–11 3 0 3

≥12 1 0 1

Delay in presentation (mean) 100 (60–137.50) 155 (110–175) 85 (52.50–100) 0.0100–60 min 9 2 7

61–100 min 7 1 6

101–150 min 4 1 3

>151 min 5 5 0

Delay to OR (mean) 3 (2–4.5) 4.04 (3–7) 2.15 (1.47–3.0) 0.0420–2 h 4 0 4

2.1–3 h 5 0 5

3.1–4 h 8 4 4

>4 h 8 5 3

Mass present 12 4 8 0.560

Fracture

Single 7 5 2 0.069Multiple 8 2 6

None 10 2 8

Marshall grading

I 6 0 6 0.172II 4 4 0

III 8 1 7

IV 6 4 2

Evacuated mass 1 0 1

Intubated (on arrival) 17 6 11 0.106

Anisocoria (on arrival) 6 3 3 0.063

Childs Nerv Syst

The role of DC in pediatric patients with TBI is underinvestigated. Whereas several trials for adult patients havebeen published or are ongoing, there is only one randomizedtrial of 27 patients by Taylor et al. for pediatric patients, whichinterestingly showed favorable outcome after DC [20]. Polinet al. also suggested that DCmight bemore useful for properlyselected pediatric patients, with a reported favorable outcomein 80 % [17]. However, most published DC studies includingTaylor’s trial have been limited by small sample size, variabletreatment algorithm, and short follow-up [13]. With at least5 months of follow-up, favorable outcome was observed in64 % of our patients. Our study, to the best of our knowledge,is one of the largest series on the role of DC in pediatric TBI;the largest being recently published series by Guresir et al. [9]consisted of 34 patients. Despite this, the study sample size isinadequate to perform regression analysis to determine riskfactors after adjusting for confounders.

Twenty-one of our 25 patients were male, which is con-sistent with other reports published on TBI, especially fromdeveloping countries [19]. Fall was the commonest mecha-nism of injury, which is again a finding more often reportedfrom developing countries; in contrast, developed countrieshave road traffic injuries as the commonest mechanism ofinjury. We also found an alarming number of patientsbrought with gunshot injuries, which were mostly from straybullets [12]. Our mortality rate was 36 %, and the rest ofpatients showed good recovery or moderate disability, withno patient with severe disability or in vegetative state atfollow-up of at least 5 months. Out of the 16 patients whosurvived, 11 underwent cranioplasties as indicated. No com-plications of infection, osteolysis, or others were observedduring the mean follow-up time. Long-term follow-up forthese patients was not available as is the case with most ofthe literature from developing countries. Some of these pa-tients come from remote villages with no addresses or phonenumbers and return to these villages after recovering, whichmake follow-ups impossible.

A low presenting GCS is a well-known predictor of pooroutcome after TBI and this trend was observed in our study

too [15]. Although, previous reports have suggested age as asignificant predictor, with younger patients showing betteroutcomes after TBI, the same could not be validated fromour data [4, 6, 10, 16]. We also found that patients who werebrought to the emergency room after a delay exceeding150 min also showed poor outcome. This is understandableas a large body of literature exists supporting the benefits ofearly resuscitation of trauma patients [23]. In a city such asours, where pre-hospital care is nonexistent, such a findingwas not unexpected [19]. The timing of DC has also beenstudied previously and most papers, including Taylor’s land-mark trial, have suggested good outcome correlating withearly surgery [14, 17, 20]. Our study supports this notion thatas patients who were operated after 4 h of arrival had pooroutcome.

Although pupillary abnormalities (unilateral or bilateraldilatation and abnormalities with reaction) on presentationafter TBI have been shown to be associated with unfavorableoutcome in several studies, we did not find it to be a signif-icant predictor of outcome in our study [2, 18]. This isperhaps because a few of our patients, despite presentingwith abnormal pupillary response, did reasonably well aftersurgery. Even though the cons of intraoperative blood lossare well accepted especially in pediatric patients, to the bestof our knowledge, this is the first study to documentintraoperative blood loss of more than 300 ml as an indepen-dent predictor of poor outcome after DC. We also found thepresence of single cranial fractures to be associated withrelatively poor outcome when compared to multiple frac-tures. We believe that this might be because after multiplefractures, the cranium no longer remains a fixed volumecompartment and the brain is allowed more space to swell,translating to a lower ICP and better outcome.

This study proposes that DC is not only a life-savingprocedure in majority of patients, but it also improves func-tional outcome [24]. Our study showed that nine patients(36 %) died postoperatively, while all surviving patientsshowed good functional outcome with good recovery in tenpatients (40 %), moderate disability in six (24 %), and nopatients with severe disability or vegetative state on mean afollow-up of 5 months. It was beyond the objective of thisstudy to evaluate the outcome of patients not undergoing DCafter traumatic brain injury. A fair comparison of outcomesof various therapies after traumatic brain injury would bebest carried out with a prospective study design.

Conclusion

Our study corroborates the assumption that DC in children isa life-saving procedure and should be considered in severetraumatic brain injury. It also leads to a good functionaloutcome after severe injury. However, patient selection still

Table 2 Comparison of intraoperative predictors between the pooroutcome group and good outcome group

Variable Total(n=25)

Poor outcomegroup (n=9)

Good outcomegroup (n=16)

p value

Operative blood loss (ml)

0–200 8 1 7 0.001201–300 7 0 7

>300 10 8 2

Flap

Unilateral/right 11 3 8 0.723Unilateral/left 7 3 4

Bilateral 7 3 4

Childs Nerv Syst

remains an important aspect, and the above-mentioned fac-tors should be considered while deciding for DC to improvesurvival. Larger sample size, preferably multicenter studiesand clinical trials, are warranted to assess the validity ofabove-mentioned factors that influence outcome and to ad-dress other potential factors which did not reach statisticalsignificance in our study due to low power. This wouldfurther help in patient selection for this procedure in thepediatric subgroup.

Conflict of interest None

References

1. Aldrich EF, Eisenberg HM, Saydjari C, Luerssen TG, Foulkes MA,Jane JA, Marshall LF, Marmarou A, Young HF (1992) Diffuse brainswelling in severely head-injured children. A report from the NIHTraumatic Coma Data Bank. J Neurosurg 76:450–454

2. Andrews BT, Pitts LH (1991) Functional recovery after traumatictranstentorial herniation. Neurosurgery 29:227–231

3. Balestreri M, Czosnyka M, Hutchinson P, Steiner LA, Hiler M,Smielewski P, Pickard JD (2006) Impact of intracranial pressureand cerebral perfusion pressure on severe disability and mortalityafter head injury. Neurocrit Care 4:8–13

4. Bruce DA, Schut L, Bruno LA, Wood JH, Sutton LN (1978)Outcome following severe head injuries in children. J Neurosurg48:679–688

5. Bruce DA (1990) Head injuries in the pediatric population. CurrProbl Pediatr 20:61–107

6. Bullock MR, Chesnut R, Ghajar J, Gordon D, Hartl R, Newell DW,Servadei F, Walters BC, Wilberger JE (2006) Surgical managementof acute subdural hematomas. Neurosurgery 58:S16–S24, discus-sion Si-iv

7. Dam Hieu P, Sizun J, Person H, Besson G (1996) The place ofdecompressive surgery in the treatment of uncontrollable post-traumatic intracranial hypertension in children. Childs Nerv Syst12:270–275

8. Guerra WK, Gaab MR, Dietz H, Mueller JU, Piek J, Fritsch MJ(1999) Surgical decompression for traumatic brain swelling: indi-cations and results. J Neurosurg 90:187–196

9. Guresir E, Schuss P, Seifert V, Vatter H (2012) Decompressivecraniectomy in children: single-center series and systematic review.Neurosurgery 70:881–888, discussion 888–889

10. Howard JL, Cipolle MD, Anderson M, Sabella V, Shollenberger D,Li PM, Pasquale MD (2008) Outcome after decompressivecraniectomy for the treatment of severe traumatic brain injury. JTrauma 65:380–385, discussion 385–386

11. Hutchinson PJ, Corteen E, Czosnyka M, Mendelow AD, MenonDK, Mitchell P, Murray G, Pickard JD, Rickels E, Sahuquillo J,

Servadei F, Teasdale GM, Timofeev I, Unterberg A, Kirkpatrick PJ(2006) Decompressive craniectomy in traumatic brain injury: therandomized multicenter RESCUEicp study (www.RESCUEicp.com). Acta Neurochir Suppl 96:17–20

12. Irfan FB, Hassan RU, Kumar R, Bhutta ZA, Bari E (2010)Craniocerebral gunshot injuries in preschoolers. Childs Nerv Syst26:61–66

13. Jagannathan J, Okonkwo DO, Dumont AS, Ahmed H, Bahari A,Prevedello DM, Jane JA Sr, Jane JA Jr (2007) Outcome followingdecompressive craniectomy in children with severe traumatic braininjury: a 10-year single-center experience with long-term followup. J Neurosurg 106:268–275

14. Johnson DL, Krishnamurthy S (1998) Severe pediatric head injury:myth, magic, and actual fact. Pediatr Neurosurg 28:167–172

15. Josan VA, Sgouros S (2006) Early decompressive craniectomy maybe effective in the treatment of refractory intracranial hypertensionafter traumatic brain injury. Childs Nerv Syst 22:1268–1274

16. Kim KT, Park JK, Kang SG, Cho KS, Yoo DS, Jang DK, Huh PW,Kim DS (2009) Comparison of the effect of decompressivecraniectomy on different neurosurgical diseases. Acta Neurochir(Wien) 151:21–30

17. Polin RS, Shaffrey ME, Bogaev CA, Tisdale N, Germanson T,Bocchicchio B, Jane JA (1997) Decompressive bifrontal craniectomyin the treatment of severe refractory posttraumatic cerebral edema.Neurosurgery 41:84–92, discussion 92–84

18. Prasad MR, Ewing-Cobbs L, Swank PR, Kramer L (2002)Predictors of outcome following traumatic brain injury in youngchildren. Pediatr Neurosurg 36:64–74

19. Shamim S, Razzak JA, Jooma R, Khan U (2011) Initial results ofPakistan’s first road traffic injury surveillance project. Int J InjControl Saf Promot 18:213–217

20. Taylor A, Butt W, Rosenfeld J, Shann F, Ditchfield M, Lewis E,Klug G, Wallace D, Henning R, Tibballs J (2001) A randomizedtrial of very early decompressive craniectomy in children withtraumatic brain injury and sustained intracranial hypertension.Childs Nerv Syst 17:154–162

21. Vahedi K, Vicaut E, Mateo J, Kurtz A, Orabi M, Guichard JP,Boutron C, Couvreur G, Rouanet F, Touze E, Guillon B,Carpentier A, Yelnik A, George B, Payen D, Bousser MG (2007)Sequential-design, multicenter, randomized, controlled trial of ear-ly decompressive craniectomy in malignant middle cerebral arteryinfarction (DECIMAL Trial). Stroke 38:2506–2517

22. Veen E, Aerdts S, Brink W (2006) Decompressive (hemi)craniectomyfor refractory intracranial hypertension after traumatic brain injury.Critical Care 10:1–1, C7-P458

23. von Elm E, Schoettker P, Henzi I, Osterwalder J, Walder B (2009)Pre-hospital tracheal intubation in patients with traumatic braininjury: systematic review of current evidence. Br J Anaesth103:371–386

24. Williams RF, Magnotti LJ, Croce MA, Hargraves BB, Fischer PE,Schroeppel TJ, Zarzaur BL, Muhlbauer M, Timmons SD, FabianTC (2009) Impact of decompressive craniectomy on functionaloutcome after severe traumatic brain injury. J Trauma 66:1570–1574, discussion 1574–1576

25. Zimmerman RA, Bilaniuk LT (1981) Computed tomography inpediatric head trauma. J Neuroradiol 8:257–271

Childs Nerv Syst