Non traumatic coma
http://www.100md.com
《美国医学杂志》
1 Departments of Pediatrics, Advanced Pediatrics Center, Post Graduate Institute of Medical Education and Research Center, Chandigarh, India
2 Departments of Radiodiagnosis, Advanced Pediatrics Center, Post Graduate Institute of Medical Education and Research Center, Chandigarh, India
Abstract
OBJECTIVE: To study the etiology and clinical profile of non-traumatic coma in children and to determine the clinical signs predictive of outcome. METHODS: 100 consecutive cases of non- traumatic coma between 2 months to 12 years. Clinical signs studied were temperature, pulse, heart rate, blood pressure, coma severity by Glasgow coma scale (GCS), respiratory pattern, pupillary and corneal reflex, extra ocular movements, motor patterns, seizure types and fundus picture. These were recoded at admission and after 48 hours of hospital stay. Etiology of coma was determined on basis of clinical history, examination and relevant laboratory investigations by the treating physician. The outcome was recorded as survived or died, and among those who survived as normal, mild, moderate, or severe disability. Chi-square test and logistic regression analysis were done to determine predictors of outcome. RESULTS: Etiology of coma in 60% cases was CNS infection (tubercular meningitis- 19, encephalitis- 18, bacterial meningitis- 16, others- 7); other causes were toxic-metabolic conditions (19%), status epilepticus (10%), intracranial bleed (7%), and miscellaneous (4%). 65 children survived, 11 were normal, 14 had mild disability, 21 had moderate disability and 14 were severely disabled and dependent. Survival was significantly better in patients with CNS infection (63%) as compared to those with toxic-metabolic causes (27%) and intracranial bleed (43%, P < 0.05). On bivariate analysis age ￡3 years, poor pulse volume, abnormal respiratory pattern and apnoea, abnormal pupillary size and reaction, abnormal extra ocular movements, absent corneal reflex, abnormal motor muscle tone at admission or 48 hours correlated significantly with mortality. Survival was better with increasing GCS (Spearman rho = .32, P < 0.001). On logistic regression age < 3 years, poor pulse volume, absent extraocular movements and papilloedema at admission and 48 hours after admission were independent significant predictors of death. CONCLUSION: CNS infections were the most common cause of non-traumatic coma in childhood. Simple clinical signs were good predictors of outcome.
Keywords: Non-traumatic coma; Outcome variables; Glasgow coma scale
Non-traumatic coma in childhood is an important pediatric emergency. It can result from a wide range of primary etiologies. Neurologic outcome is often of foremost concern to parents and physicians. It may range from absence of impairment to severe disability or death. Etiology of coma and clinical status at the time of presentation are likely predictors of outcome. A better understanding of causes and outcome is essential to help improve the approach and to plan rational management of non-traumatic coma.
Literature on pediatric non-traumatic coma is rather inconclusive, as there are few systematic studies, and most of these are retrospective. Very little information is available particularly so from developing countries including India.[1],[2],[3] In a prospective study the authors have therefore examined the etiology, clinical signs and severity of non-traumatic coma in children, with a view to define predictors of outcome.
Materials And Methods
This prospective observational study was conducted in the pediatric emergency unit of a tertiary care teaching and referral hospital over a period of 10 months. The study was approved by Institutional Ethics committee. Informed written consent was obtained from the parents /guardian of the subjects.
Inclusion and exclusion criteria: All the children between 2 months to 12 years of age, presenting with coma were eligible for inclusion in the study. Those with history of trauma were excluded.
Demographic and clinical data was recorded at admission. Patients were re-examined at 48 hours to record clinical data and at discharge to record the outcome. The clinical variables recorded were heart rate, respiratory rate and pattern, blood pressure (average of three recordings, using mercury sphygmomanometer, by auscultatory method), temperature, coma severity (using modified Glasgow Coma Scale), pupillary size and response to light, extra ocular movements, corneal reflex, posture, motor pattern (recorded subjectively by assessing the passive tone), seizure if any, type of seizure, involuntary movements and fundus picture.
The etiology of coma was determined on the basis of history, clinical examination and relevant laboratory investigations. The investigations, such as lumbar puncture, CT scan and metabolic work-up, depended on the clinical presentation and were determined by the consultant in-charge. Etiology was classified into infectious, toxic-metabolic, including hypoxic-ischemic, post-status epilepticus, intracranial bleed and miscellaneous.
Bacterial meningitis was defined as acute febrile encephalopathy with identification of microorganisms from the CSF culture or latex agglutinins, or presence of 3 or more of the following abnormalities in CSF - (i) polymorphonuclear leucocytosis 3 100 cells/mm3 (ii) glucose ~ 40 mg/dl or 50% of blood sugar (iii) elevated proteins > 40 mg/dl (iv) micro-organisms seen by Gram staining. Diagnosis of tuberculous meningitis was based on the criteria by Ahuja et al.[4] Encephalitis was defined as acute febrile encephalopathy with CSF pleocytosis with lymphocyte predominance (>5 cells/mm3) and absence of bacteria on direct microscopy, culture or latex agglutination and where no other alternative diagnosis was identifiable. Hypertensive encephalopathy was diagnosed when coma of acute onset was associated with blood pressure more than 95th percentile for age and sex, with or without retinal changes. When there was metabolic derangement commensurating with the clinical picture or toxic ingestion was confirmed, a label of toxic-metabolic coma was used. Coma following hypoxic cerebral injury such as after cardio-respiratory compromise or shock was considered to be hypoxic-ischemic. Children with coma with evidence of bleed on radio imaging of head were labeled as having intracranial bleed.
Clinical variables that were found significant on chi-square test were considered for logistic regression analysis. Etiology was not included because a definitive diagnosis was not often made at the initial examination.
Definitions of study variables were as follows
Coma : A state of unresponsiveness without evidence of awareness of self or environment, a state from which the patient cannot be aroused by vocal or sensory stimuli.
Tachycardia : Heart rate above the upper limit for that age.
Bradycardia : Heart rate less than sixty per minute.
Hypertension : Blood pressure more than 95th centile for age and sex.
Hypotension : Blood pressure below 5th centile for the age and sex.
Hyperthermia : Axillary temperature above 380C.
Hypothermia : Temperature below 350C.
Coma severity : Based on score obtained on the modified Glasgow coma scale5.
Pupils : (i) normal - both pupils equal in size, 2-3 mm in diameter and reactive to light, (ii) abnormal-pupils small (= 1mm), or dilated (= 4 mm), unequal or non reactive to light.
Extra ocular movements (EOM): (i) normal - no impairment of movement in any direction, (ii) abnormal - if lateral, medial, upward, downward or all movements of eyeballs were absent.
Corneal reflex : absent or present.
Motor patterns were recorded as normal pattern or abnormal if there was diffuse decrease in tone (flaccidity), increase in tone (hypertonia) or decerbrate or decorticate posture6
Respiratory pattern was said to be abnormal if the breathing was central neurogenic hyperventilation, apneustic, ataxic or apnoeic.6
Outcome variables: Outcome was recorded as survived and died. Among those who survived it was further graded as normal, or those having mild, moderate or severe disability, defined as:
Normal : no motor deficit, ataxia, cranial nerve palsy, and functional level back to pre-illness state.
Mild disability : minimal alterations of tone/deep tendon reflexes, isolated cranial nerve palsy and weakness of grade 4 or ataxia.
Moderate disability : moderate weakness (grade 3) or ataxia, behaviour disturbance and multiple cranial nerve involvement.
Severe disability: severe weakness (grade 3) or ataxia and quadriplegia.
Statistical Analysis: Descriptive statistics (frequency, percentages) were calculated. The study variables were analyzed for their association with the outcome by applying the Chi-square test, and calculation of Odd's ratio and relative risk (RR) with 95% confidence interval (95% CI). Clinical variables that were found significant on Chi-square test were further analysed using logistic regression analysis. SPSS 10.0 and Epinfo 1.1 statistical package were used.
Results
A total of 100 comatose children (65 boys, 35 girls) were included in the study. Fifteen patients were below the age of 1 yr, 25 were 1-3 yr, 34 were 4-5 yr and 26 were 6 to 12 ys old.
Central nervous system (CNS) infections accounted for 60% of the cases table1. Among the 19 children classified as toxic-metabolic, 6 fulfilled the Center for Disease Control's criteria for Reye's syndrome, three gave history of toxic ingestion and four children had hypoxic ischaemic encephalopathy (HIE) secondary to shock caused by sepsis (n-2) or acute diarrhoea (n-2). Seven children had intracranial bleed because of non-accidental injury, which was not suspected at the time of admission.
Outcome : Thirty-five patients died and 65 survived. Of the 65 survivors, 5 left against medical advice before the recovery from primary illness, 10 were normal (without any deficit), 13 had mild disability, 22 were moderately disabled and 15 were severely disabled and dependent table3
Mortality rate was 48 % (7 of 15) among infants, 44% (11 of 25) among toddlers 1-3 years old, 74% (8 of 34) among preschool children, and 27.7% among children 6 -12 years. None of the infants had a normal outcome, whereas 10% of toddlers, 12% of preschool and 16% of school age children had no deficits at discharge. Mortality was similar between the two sexes. Severe disability was higher in male children (10 of 33, 30%) compared to female (5 of 27, 18.5%) whereas intermediate level of disability was more in female children.
Mortality with various CNS infections was similar table1 but as a group CNS infections had significantly better survival rate as compared to toxic- metabolic group (RR = 0.5; 95% CI - 0.3 to 0.9; P = 0.04).
Predictors of outcome at admission
On bivariate analysis younger age (￡ 3 years), poor pulse volume, bradycardia, hypotension, modified GCS ￡ 5, abnormal respiratory pattern and apnoea, non-reactive pupil, impaired or absent extra ocular movements, absence of corneal reflex and abnormal motor muscle tone at admission correlated significantly with mortality. Hypothermia / hyperthermia did not show significant association table2. Bradycardia had significant correlation with mortality (odds ratio 13.2; 95% CI 1.6 - 305, P = 0.0037) as also the hypotension (odds ratio 5.1, 95% CI - 1.5 - 17.7; P = 0.002). Poor pulse volume was associated with death in 63.6% (21 of 33) patients (RR -2.2; 95% C.I. 1.4 - 3.5, P = 0.00003) while 79% (19 of 25) children who had non-reacting pupil died table2. Significantly increased risk of mortality was seen in 25 patients who had non-reactive pupils as compared to 75 patients with reactive pupils (odds ratio 11.7, 95% CI 3.6 - 39.7, P = 0.000008). Absent corneal reflex and apnoea each was associated with death in 9 of 10 patients with the sign (RR - 7.1; 95% C.I. - 1.1 to 45.9; P = 0.0005). Survival rate and Glasgow coma scores at admission were directly proportional table2 and had a significant correlation. (Spearman rho = 0.32, P < 0.001).
On logistic regression age < 3 years, low pulse volume, absent oculocephalic reflex and presence of papilloedema at presentation were the significant independent predictors of death table3.
Predictors of outcome at 48 hours
Of 35 deaths, 21 occurred after 48 hours of admission. The significant predictors of death after 48 hours stay in hospital on bivariate analysis were similar to the predictors at admission. These were poor pulse volume, abnormal respiratory pattern / apnoea, non-reactive pupil, abnormal extra ocular movements, absence of corneal reflex, and abnormal motor tone. In addition, tachycardia and presence of papilloedema correlated significantly with mortality at 48 hours table2. The assessment of GCS at 48 hours of admission was confounded by various factors. Several children were under sedation and some were also paralyzed for mechanical ventilation. Use of diazepam infusion for status epilepticus had also made assessment of GCS difficult. GCS score could be obtained at 48 hrs of admission in only 51 patients.
Of 16 patients with poor pulse volume at 48 hours 14 (87%) died; this accounted for two-third of all deaths after 48 hours (RR - 7.2; 95% CI 2.0 to 26; P = <0.001). Apnoea (RR 5.6; 95 CI - 0.9 to 34.6; P = < 0.001), absent corneal reflex (RR = 8.4; 95 CI - 1.3 to 53.8; P = <0.001), absent EOM (RR =6.1; 95% CI - 1.7 to 22.0; P < 0.001), non-reacting pupils (RR - 5.5; 95% CI - 1.9 to 15.3; P < 0.001), and flaccid motor tone (RR = 1.7; 95% CI - 1.2 to 2.4; P = 0.0001) were other signs associated with very high risk of mortality.
On logistic regression age < then 3 years, poor pulse and abnormal extra ocular movements at 48 hours after admission were the most significant predictors of death table3.
Predictors of neurodeficits at admission and at 48 hours
Pulse volume at admission as well as at 48 hours had significant association with neurological outcome table3. Poor pulse volume at admission was associated with mortality in about two third of patients; and of those who survived, 81.8% had moderate or severe disability. Neurodeficits amongst survivors had significant correlation with low GCS, both at admission (OR - 0.554, 95% CI - 0.355 - 0.864, P = 0.009) and at 48 hours (OR - 0.742, 95% CI - 0.583 - 0.945, P = 0.016).
Discussion
Pediatric coma has been for long an enigma with only few studies and inconclusive information. This study fills in the gap of a prospective observational study from India. In addition to demographic variables and etiology, two sets of clinical signs as prognostic indicators of coma were studied. First set included vital signs and second set consisted of specific neurological signs that assessed integrity of cerebral hemispheres and brain stem.
In comatose patients, evidence of widespread damage of brain stem or cerebral hemispheres at onset usually predicts death or severe disability. Therefore clinical signs that reflect extent and severity of dysfunction of cerebral hemispheres and/or brain stem were studied. Glasgow Coma Scale (GCS) Score reflects the integrity of cerebral functions. Eye movements (in response to vestibular stimulation), pupillary responses, and corneal responses primarily express functions regulated by the brain stem. Breathing pattern and signs of abnormal tone of limbs and posturing indicate the extent and severity of cerebral hemisphere as well as brain stem damage.
It was observed that CNS infections were the commonest cause of non-traumatic coma. This is also supported by other studies,[7], [8] wherein infections of the CNS were found to be the leading causes of non-traumatic coma in children table4. However, the type of infection seems to vary in different regions. Cerebral malaria was common in Africa[9], whereas dengue haemorrhagic fever was an important cause of coma in South-East Asia.[2] The importance of infective etiologies in children is in sharp contrast to adult-hospital based series where degenerative and cerebrovascular pathologies predominate.[10] Among the non-infectious causes, toxic-metabolic causes were the commonest and were also comparable in frequency with other studies table1. In series from U.K.[7] and Nigeria[8] status epilepticus was the second commonest cause contributing to more than one fourth of cases of non-traumatic coma; in contrast, it was seen in only 10% of cases in the present study.
The overall mortality of 35% was slightly higher as compared to other pediatric hospital based series, (26.7%) from Nigeria[8] and Canada.[1] This was possibly due to presentation later in the natural history of disease. However, mortality was considerably lower than that reported in adults; their mortality rates being 60% and neurological intact survival rates 10%.[9]
Mortality rate among children under 3 yrs was significantly higher in the present study. It was related to higher frequency of toxic-metabolic causes, intracranial bleed and higher mortality with CNS infections. The incidence and outcome of coma was not associated with gender. Seshia and Seshia[1] also did not observe any significant difference in the incidence of coma between the two sexes. Earlier studies[11] had shown a greater mortality in male (42%) compared to female children (20%).
CNS infection group fared significantly better than the toxic/metabolic group. In the study by Seshia et al,[11] these two groups had a similar outcome with four deaths. Among survivors of CNS infections in the present study, only one third had a normal outcome or mild disability. Severe disability was seen in about 23% of patients. This is in contrast to other published studies where 59% and 74% of the infectious group had a good outcome compared with 36% and 47% respectively in the metabolic group.[12], [13]
It is believed that prognosis in coma depends on its severity but there is rather inconclusive data on the use of GCS score and its predictive value in pediatric non-traumatic coma. In the present study the modified GCS recorded at admission had significant association with outcome; mortality rates progressively increased with decreasing GCS.
Contrary to common belief, fever at admission or its continuing presence at 48 hours after hospitalization was not associated with higher risk of mortality or poor neurological outcome. On the other hand hypothermia spelled poor prognosis, but the number was too small for any meaningful conclusion. In the series by Johnson and Seshia,[11] all the 13 hypothermic children died. Hypothermia, regardless of etiology causes diminished cerebral metabolism and very low temperature may result in an isoelectric electroencephalogram.[14]
All the signs of circulatory instability predicted higher mortality and poor neurological outcome. Poor pulse volume at admission and at 48 hours predicted two-thirds of all deaths. Further, 81% of 11 survivors who had a poor pulse at admission had moderate or severe neurodeficits. Bradycardia had a very high association with mortality; six of seven patients with bradycardia at admission died. Tachycardia was also associated with increased risk of death and poor neurological outcome when compared with a normal heart rate. No study reporting on predictive value of heart rate in the prognosis of a non-traumatic coma in children.
Hypotension was a poor prognostic sign; two-third of 18 children who were hypotensive at admission died. Further, all the survivors of hypotension except one, had moderate or severe neurological deficit. This is similar to the study by Johnston and Seshia,[11] wherein 14 of 15 hypotensive children died. Interestingly, the risk of mortality and neurological deficits was similar in hypertensive and normotensive patients. This data supports the view that acute hypertension in comatose patients may be neuro protective and should not be brought down.
Various abnormalities in respiratory pattern may be seen in coma depending on the region of brain involved. It was therefore expected that an abnormal breathing pattern would predict a poorer outcome. The prognosis was best with a normal respiratory pattern while apnea at admission had highest risk of mortality. This is similar to a study from Canada[11] in which about 60% of apneic children died. Abnormal breathing pattern including apneustic, ataxic, Cheyne-Stokes respiration More Details or neurogenic hyperventilation, also had significant association with increased mortality and moderate to severe neurological disability.
Pupillary signs were very good predictors of survival and neurologic outcome. Non-reactive pupils at admission as well as at 48 hours were strong predictors of a fatal outcome. In the study by Seshia et al 68% of children with fixed dilated pupils for more than 2 hr died[1]. Ogunmekan made similar observations in a large retrospective study from Nigeria.[8] However, it should be appreciated that about one-fourth of children with non-reactive pupils may survive.
Neurological outcome and disability also correlated with abnormality of pupillary size. Severe deficits were seen in 28% of those with unequal reacting pupils, and in 18% of those with equal reacting but dilated pupils.
Presence of the corneal reflex indicates functional interconnections in the pons. Absent corneal reflexes in children with deep coma prognosticated a poor outcome.
Presence of doll's eye movements suggests intact interconnections between cranial nerve nuclei III, IV and VI via the medial longitudinal fasciculus and intact vestibular input to this system. Asymmetrical or partial absence of eye movements therefore, generally indicates asymmetrical brain stem lesion in mid brain or pons while complete absence of doll's eye movements suggests bilateral structural brainstem abnormality or severe metabolic-toxic encephalopathies.
In the present study, 80% of children, with preserved EOM at presentation, survived whereas, 54% of children with absent EOM died. None of the children with impaired eye movements had normal neurological outcomes; all of them had moderate to severe disability and were not capable of self-care. This goes with the earlier studies that suggest absent or impaired EOM as a sensitive index for prognosticating the outcome. Preserved EOM have been associated with favourable outcome in previous studies.[8], [10] Seshia et al[11] observed that 67% of children who had normal EOM recovered and 16% died; in contrast, all those with absent EOM died and needed assisted ventilation.
In the present study, abnormality of motor pattern was predictive of higher risk of death. Interestingly, it was flaccidity that was associated with higher risk as compared to hypertonia. This was similar to findings of Seshia et al[1] who observed that 84% of those with a normal motor pattern had a normal outcome whereas 82% of those who were flaccid died.
The present study shows clinical signs such as poor pulse volume, absent EOM or non-reactive pupils were strong predictors of death or severe handicap following severe non-traumatic coma. The study also reaffirms that clinical variables and GCS score remain the most readily available tools for assessment of non-traumatic coma, to identify those who are likely to die and those having the greatest potential for recovery. This is particularly helpful in resource-limited countries for directing the limited resources for maximal benefit.
References
1. Seshia SS, Seshia MMK, Sachdeva RK. Coma in childhood. Dev Med Child Neurol 1977; 19: 614-628.
2. Sofiah A, Hussain HM. Childhood non-traumatic coma in Kuala Lumpur, Malaysia. Ann Trop Pediatr 1997; 17: 327-331.
3. Nayana Prabha PC, Nalini P, Serane VT. Role of Glasgow Coma Scale in pediatric nontraumatic coma. Indian Pediatr 2003; 40: 620-625.
4. Ahuja GK, Mohan KK, Prasad K, Behari M. Diagnostic criteria for tuberculous meningitis and their validation. Tuber Lung Dis 1994; 75: 149-152.
5. Kirkham FJ. Non-traumatic coma in children. Arch Dis Child 2001; 85: 303-312.
6. Plum F, Posner JB. The Diagnosis of Stupor and Coma. 3rd edn. Philadelphia; FA Davis CC; 1980.
7. Vijayakumar K, Knight R, Prabhakar P, Murphy PJ, Sharples PM. Neurological outcome in children with non-traumatic coma admitted to a regional paediatric intensive care unit. Arch Dis Child 2003; 88: A30-32.
8. Ogunmekan AO. Non-traumatic coma in childhood: etiology, clinical findings, morbidity, prognosis and mortality. J Trop Pediatr 1983; 29: 230 - 232.
9. Matuja WB, Matekere NJ. Causes and early prognosis of non-traumatic coma in Tanzania. Muhimbili Medical Centre experience. Trop Geogr Med 1987; 39: 330-335.
10. Bates D, Caronna JJ, Cartlidge NE, Knill -Jones RP, Levy DE, Shaw DA, Plum F. A prospective study of nontraumatic coma: methods and results in 310 patients. Ann Neurol 1977; 2: 211- 220.
11. Johnston B, Seshia SS. Prediction of outcome in non-traumatic coma in childhood. Acta Neurol Scand 1984; 69: 417-427.
12. Tasker RC, Matthew JD, Helms P, Dinwiddie R, Boyd S. Monitoring in non-traumatic coma. Part I: invasive intracranial measurements. Arch Dis Child 1988; 63: 888-894.
13. Ishikawa T, Asano Y, Morishima T, Nagashima M, Sobue G, Watanabe K, Yamaguchi H. Epidemiology of acute childhood encephalitis. Aichi Prefecture, Japan, 1984-90. Brain Dev 1993; 15: 192-197.
14. Holbrook PR, Fields IA. Drowning and near drowning. In Holbrook PR, ed. Textbook of Pediatric Critical Care . WB Saunders Co., Philadelphia, PA, 1993; 91-93.(Bansal Arun, Singhi Sunit)
2 Departments of Radiodiagnosis, Advanced Pediatrics Center, Post Graduate Institute of Medical Education and Research Center, Chandigarh, India
Abstract
OBJECTIVE: To study the etiology and clinical profile of non-traumatic coma in children and to determine the clinical signs predictive of outcome. METHODS: 100 consecutive cases of non- traumatic coma between 2 months to 12 years. Clinical signs studied were temperature, pulse, heart rate, blood pressure, coma severity by Glasgow coma scale (GCS), respiratory pattern, pupillary and corneal reflex, extra ocular movements, motor patterns, seizure types and fundus picture. These were recoded at admission and after 48 hours of hospital stay. Etiology of coma was determined on basis of clinical history, examination and relevant laboratory investigations by the treating physician. The outcome was recorded as survived or died, and among those who survived as normal, mild, moderate, or severe disability. Chi-square test and logistic regression analysis were done to determine predictors of outcome. RESULTS: Etiology of coma in 60% cases was CNS infection (tubercular meningitis- 19, encephalitis- 18, bacterial meningitis- 16, others- 7); other causes were toxic-metabolic conditions (19%), status epilepticus (10%), intracranial bleed (7%), and miscellaneous (4%). 65 children survived, 11 were normal, 14 had mild disability, 21 had moderate disability and 14 were severely disabled and dependent. Survival was significantly better in patients with CNS infection (63%) as compared to those with toxic-metabolic causes (27%) and intracranial bleed (43%, P < 0.05). On bivariate analysis age ￡3 years, poor pulse volume, abnormal respiratory pattern and apnoea, abnormal pupillary size and reaction, abnormal extra ocular movements, absent corneal reflex, abnormal motor muscle tone at admission or 48 hours correlated significantly with mortality. Survival was better with increasing GCS (Spearman rho = .32, P < 0.001). On logistic regression age < 3 years, poor pulse volume, absent extraocular movements and papilloedema at admission and 48 hours after admission were independent significant predictors of death. CONCLUSION: CNS infections were the most common cause of non-traumatic coma in childhood. Simple clinical signs were good predictors of outcome.
Keywords: Non-traumatic coma; Outcome variables; Glasgow coma scale
Non-traumatic coma in childhood is an important pediatric emergency. It can result from a wide range of primary etiologies. Neurologic outcome is often of foremost concern to parents and physicians. It may range from absence of impairment to severe disability or death. Etiology of coma and clinical status at the time of presentation are likely predictors of outcome. A better understanding of causes and outcome is essential to help improve the approach and to plan rational management of non-traumatic coma.
Literature on pediatric non-traumatic coma is rather inconclusive, as there are few systematic studies, and most of these are retrospective. Very little information is available particularly so from developing countries including India.[1],[2],[3] In a prospective study the authors have therefore examined the etiology, clinical signs and severity of non-traumatic coma in children, with a view to define predictors of outcome.
Materials And Methods
This prospective observational study was conducted in the pediatric emergency unit of a tertiary care teaching and referral hospital over a period of 10 months. The study was approved by Institutional Ethics committee. Informed written consent was obtained from the parents /guardian of the subjects.
Inclusion and exclusion criteria: All the children between 2 months to 12 years of age, presenting with coma were eligible for inclusion in the study. Those with history of trauma were excluded.
Demographic and clinical data was recorded at admission. Patients were re-examined at 48 hours to record clinical data and at discharge to record the outcome. The clinical variables recorded were heart rate, respiratory rate and pattern, blood pressure (average of three recordings, using mercury sphygmomanometer, by auscultatory method), temperature, coma severity (using modified Glasgow Coma Scale), pupillary size and response to light, extra ocular movements, corneal reflex, posture, motor pattern (recorded subjectively by assessing the passive tone), seizure if any, type of seizure, involuntary movements and fundus picture.
The etiology of coma was determined on the basis of history, clinical examination and relevant laboratory investigations. The investigations, such as lumbar puncture, CT scan and metabolic work-up, depended on the clinical presentation and were determined by the consultant in-charge. Etiology was classified into infectious, toxic-metabolic, including hypoxic-ischemic, post-status epilepticus, intracranial bleed and miscellaneous.
Bacterial meningitis was defined as acute febrile encephalopathy with identification of microorganisms from the CSF culture or latex agglutinins, or presence of 3 or more of the following abnormalities in CSF - (i) polymorphonuclear leucocytosis 3 100 cells/mm3 (ii) glucose ~ 40 mg/dl or 50% of blood sugar (iii) elevated proteins > 40 mg/dl (iv) micro-organisms seen by Gram staining. Diagnosis of tuberculous meningitis was based on the criteria by Ahuja et al.[4] Encephalitis was defined as acute febrile encephalopathy with CSF pleocytosis with lymphocyte predominance (>5 cells/mm3) and absence of bacteria on direct microscopy, culture or latex agglutination and where no other alternative diagnosis was identifiable. Hypertensive encephalopathy was diagnosed when coma of acute onset was associated with blood pressure more than 95th percentile for age and sex, with or without retinal changes. When there was metabolic derangement commensurating with the clinical picture or toxic ingestion was confirmed, a label of toxic-metabolic coma was used. Coma following hypoxic cerebral injury such as after cardio-respiratory compromise or shock was considered to be hypoxic-ischemic. Children with coma with evidence of bleed on radio imaging of head were labeled as having intracranial bleed.
Clinical variables that were found significant on chi-square test were considered for logistic regression analysis. Etiology was not included because a definitive diagnosis was not often made at the initial examination.
Definitions of study variables were as follows
Coma : A state of unresponsiveness without evidence of awareness of self or environment, a state from which the patient cannot be aroused by vocal or sensory stimuli.
Tachycardia : Heart rate above the upper limit for that age.
Bradycardia : Heart rate less than sixty per minute.
Hypertension : Blood pressure more than 95th centile for age and sex.
Hypotension : Blood pressure below 5th centile for the age and sex.
Hyperthermia : Axillary temperature above 380C.
Hypothermia : Temperature below 350C.
Coma severity : Based on score obtained on the modified Glasgow coma scale5.
Pupils : (i) normal - both pupils equal in size, 2-3 mm in diameter and reactive to light, (ii) abnormal-pupils small (= 1mm), or dilated (= 4 mm), unequal or non reactive to light.
Extra ocular movements (EOM): (i) normal - no impairment of movement in any direction, (ii) abnormal - if lateral, medial, upward, downward or all movements of eyeballs were absent.
Corneal reflex : absent or present.
Motor patterns were recorded as normal pattern or abnormal if there was diffuse decrease in tone (flaccidity), increase in tone (hypertonia) or decerbrate or decorticate posture6
Respiratory pattern was said to be abnormal if the breathing was central neurogenic hyperventilation, apneustic, ataxic or apnoeic.6
Outcome variables: Outcome was recorded as survived and died. Among those who survived it was further graded as normal, or those having mild, moderate or severe disability, defined as:
Normal : no motor deficit, ataxia, cranial nerve palsy, and functional level back to pre-illness state.
Mild disability : minimal alterations of tone/deep tendon reflexes, isolated cranial nerve palsy and weakness of grade 4 or ataxia.
Moderate disability : moderate weakness (grade 3) or ataxia, behaviour disturbance and multiple cranial nerve involvement.
Severe disability: severe weakness (grade 3) or ataxia and quadriplegia.
Statistical Analysis: Descriptive statistics (frequency, percentages) were calculated. The study variables were analyzed for their association with the outcome by applying the Chi-square test, and calculation of Odd's ratio and relative risk (RR) with 95% confidence interval (95% CI). Clinical variables that were found significant on Chi-square test were further analysed using logistic regression analysis. SPSS 10.0 and Epinfo 1.1 statistical package were used.
Results
A total of 100 comatose children (65 boys, 35 girls) were included in the study. Fifteen patients were below the age of 1 yr, 25 were 1-3 yr, 34 were 4-5 yr and 26 were 6 to 12 ys old.
Central nervous system (CNS) infections accounted for 60% of the cases table1. Among the 19 children classified as toxic-metabolic, 6 fulfilled the Center for Disease Control's criteria for Reye's syndrome, three gave history of toxic ingestion and four children had hypoxic ischaemic encephalopathy (HIE) secondary to shock caused by sepsis (n-2) or acute diarrhoea (n-2). Seven children had intracranial bleed because of non-accidental injury, which was not suspected at the time of admission.
Outcome : Thirty-five patients died and 65 survived. Of the 65 survivors, 5 left against medical advice before the recovery from primary illness, 10 were normal (without any deficit), 13 had mild disability, 22 were moderately disabled and 15 were severely disabled and dependent table3
Mortality rate was 48 % (7 of 15) among infants, 44% (11 of 25) among toddlers 1-3 years old, 74% (8 of 34) among preschool children, and 27.7% among children 6 -12 years. None of the infants had a normal outcome, whereas 10% of toddlers, 12% of preschool and 16% of school age children had no deficits at discharge. Mortality was similar between the two sexes. Severe disability was higher in male children (10 of 33, 30%) compared to female (5 of 27, 18.5%) whereas intermediate level of disability was more in female children.
Mortality with various CNS infections was similar table1 but as a group CNS infections had significantly better survival rate as compared to toxic- metabolic group (RR = 0.5; 95% CI - 0.3 to 0.9; P = 0.04).
Predictors of outcome at admission
On bivariate analysis younger age (￡ 3 years), poor pulse volume, bradycardia, hypotension, modified GCS ￡ 5, abnormal respiratory pattern and apnoea, non-reactive pupil, impaired or absent extra ocular movements, absence of corneal reflex and abnormal motor muscle tone at admission correlated significantly with mortality. Hypothermia / hyperthermia did not show significant association table2. Bradycardia had significant correlation with mortality (odds ratio 13.2; 95% CI 1.6 - 305, P = 0.0037) as also the hypotension (odds ratio 5.1, 95% CI - 1.5 - 17.7; P = 0.002). Poor pulse volume was associated with death in 63.6% (21 of 33) patients (RR -2.2; 95% C.I. 1.4 - 3.5, P = 0.00003) while 79% (19 of 25) children who had non-reacting pupil died table2. Significantly increased risk of mortality was seen in 25 patients who had non-reactive pupils as compared to 75 patients with reactive pupils (odds ratio 11.7, 95% CI 3.6 - 39.7, P = 0.000008). Absent corneal reflex and apnoea each was associated with death in 9 of 10 patients with the sign (RR - 7.1; 95% C.I. - 1.1 to 45.9; P = 0.0005). Survival rate and Glasgow coma scores at admission were directly proportional table2 and had a significant correlation. (Spearman rho = 0.32, P < 0.001).
On logistic regression age < 3 years, low pulse volume, absent oculocephalic reflex and presence of papilloedema at presentation were the significant independent predictors of death table3.
Predictors of outcome at 48 hours
Of 35 deaths, 21 occurred after 48 hours of admission. The significant predictors of death after 48 hours stay in hospital on bivariate analysis were similar to the predictors at admission. These were poor pulse volume, abnormal respiratory pattern / apnoea, non-reactive pupil, abnormal extra ocular movements, absence of corneal reflex, and abnormal motor tone. In addition, tachycardia and presence of papilloedema correlated significantly with mortality at 48 hours table2. The assessment of GCS at 48 hours of admission was confounded by various factors. Several children were under sedation and some were also paralyzed for mechanical ventilation. Use of diazepam infusion for status epilepticus had also made assessment of GCS difficult. GCS score could be obtained at 48 hrs of admission in only 51 patients.
Of 16 patients with poor pulse volume at 48 hours 14 (87%) died; this accounted for two-third of all deaths after 48 hours (RR - 7.2; 95% CI 2.0 to 26; P = <0.001). Apnoea (RR 5.6; 95 CI - 0.9 to 34.6; P = < 0.001), absent corneal reflex (RR = 8.4; 95 CI - 1.3 to 53.8; P = <0.001), absent EOM (RR =6.1; 95% CI - 1.7 to 22.0; P < 0.001), non-reacting pupils (RR - 5.5; 95% CI - 1.9 to 15.3; P < 0.001), and flaccid motor tone (RR = 1.7; 95% CI - 1.2 to 2.4; P = 0.0001) were other signs associated with very high risk of mortality.
On logistic regression age < then 3 years, poor pulse and abnormal extra ocular movements at 48 hours after admission were the most significant predictors of death table3.
Predictors of neurodeficits at admission and at 48 hours
Pulse volume at admission as well as at 48 hours had significant association with neurological outcome table3. Poor pulse volume at admission was associated with mortality in about two third of patients; and of those who survived, 81.8% had moderate or severe disability. Neurodeficits amongst survivors had significant correlation with low GCS, both at admission (OR - 0.554, 95% CI - 0.355 - 0.864, P = 0.009) and at 48 hours (OR - 0.742, 95% CI - 0.583 - 0.945, P = 0.016).
Discussion
Pediatric coma has been for long an enigma with only few studies and inconclusive information. This study fills in the gap of a prospective observational study from India. In addition to demographic variables and etiology, two sets of clinical signs as prognostic indicators of coma were studied. First set included vital signs and second set consisted of specific neurological signs that assessed integrity of cerebral hemispheres and brain stem.
In comatose patients, evidence of widespread damage of brain stem or cerebral hemispheres at onset usually predicts death or severe disability. Therefore clinical signs that reflect extent and severity of dysfunction of cerebral hemispheres and/or brain stem were studied. Glasgow Coma Scale (GCS) Score reflects the integrity of cerebral functions. Eye movements (in response to vestibular stimulation), pupillary responses, and corneal responses primarily express functions regulated by the brain stem. Breathing pattern and signs of abnormal tone of limbs and posturing indicate the extent and severity of cerebral hemisphere as well as brain stem damage.
It was observed that CNS infections were the commonest cause of non-traumatic coma. This is also supported by other studies,[7], [8] wherein infections of the CNS were found to be the leading causes of non-traumatic coma in children table4. However, the type of infection seems to vary in different regions. Cerebral malaria was common in Africa[9], whereas dengue haemorrhagic fever was an important cause of coma in South-East Asia.[2] The importance of infective etiologies in children is in sharp contrast to adult-hospital based series where degenerative and cerebrovascular pathologies predominate.[10] Among the non-infectious causes, toxic-metabolic causes were the commonest and were also comparable in frequency with other studies table1. In series from U.K.[7] and Nigeria[8] status epilepticus was the second commonest cause contributing to more than one fourth of cases of non-traumatic coma; in contrast, it was seen in only 10% of cases in the present study.
The overall mortality of 35% was slightly higher as compared to other pediatric hospital based series, (26.7%) from Nigeria[8] and Canada.[1] This was possibly due to presentation later in the natural history of disease. However, mortality was considerably lower than that reported in adults; their mortality rates being 60% and neurological intact survival rates 10%.[9]
Mortality rate among children under 3 yrs was significantly higher in the present study. It was related to higher frequency of toxic-metabolic causes, intracranial bleed and higher mortality with CNS infections. The incidence and outcome of coma was not associated with gender. Seshia and Seshia[1] also did not observe any significant difference in the incidence of coma between the two sexes. Earlier studies[11] had shown a greater mortality in male (42%) compared to female children (20%).
CNS infection group fared significantly better than the toxic/metabolic group. In the study by Seshia et al,[11] these two groups had a similar outcome with four deaths. Among survivors of CNS infections in the present study, only one third had a normal outcome or mild disability. Severe disability was seen in about 23% of patients. This is in contrast to other published studies where 59% and 74% of the infectious group had a good outcome compared with 36% and 47% respectively in the metabolic group.[12], [13]
It is believed that prognosis in coma depends on its severity but there is rather inconclusive data on the use of GCS score and its predictive value in pediatric non-traumatic coma. In the present study the modified GCS recorded at admission had significant association with outcome; mortality rates progressively increased with decreasing GCS.
Contrary to common belief, fever at admission or its continuing presence at 48 hours after hospitalization was not associated with higher risk of mortality or poor neurological outcome. On the other hand hypothermia spelled poor prognosis, but the number was too small for any meaningful conclusion. In the series by Johnson and Seshia,[11] all the 13 hypothermic children died. Hypothermia, regardless of etiology causes diminished cerebral metabolism and very low temperature may result in an isoelectric electroencephalogram.[14]
All the signs of circulatory instability predicted higher mortality and poor neurological outcome. Poor pulse volume at admission and at 48 hours predicted two-thirds of all deaths. Further, 81% of 11 survivors who had a poor pulse at admission had moderate or severe neurodeficits. Bradycardia had a very high association with mortality; six of seven patients with bradycardia at admission died. Tachycardia was also associated with increased risk of death and poor neurological outcome when compared with a normal heart rate. No study reporting on predictive value of heart rate in the prognosis of a non-traumatic coma in children.
Hypotension was a poor prognostic sign; two-third of 18 children who were hypotensive at admission died. Further, all the survivors of hypotension except one, had moderate or severe neurological deficit. This is similar to the study by Johnston and Seshia,[11] wherein 14 of 15 hypotensive children died. Interestingly, the risk of mortality and neurological deficits was similar in hypertensive and normotensive patients. This data supports the view that acute hypertension in comatose patients may be neuro protective and should not be brought down.
Various abnormalities in respiratory pattern may be seen in coma depending on the region of brain involved. It was therefore expected that an abnormal breathing pattern would predict a poorer outcome. The prognosis was best with a normal respiratory pattern while apnea at admission had highest risk of mortality. This is similar to a study from Canada[11] in which about 60% of apneic children died. Abnormal breathing pattern including apneustic, ataxic, Cheyne-Stokes respiration More Details or neurogenic hyperventilation, also had significant association with increased mortality and moderate to severe neurological disability.
Pupillary signs were very good predictors of survival and neurologic outcome. Non-reactive pupils at admission as well as at 48 hours were strong predictors of a fatal outcome. In the study by Seshia et al 68% of children with fixed dilated pupils for more than 2 hr died[1]. Ogunmekan made similar observations in a large retrospective study from Nigeria.[8] However, it should be appreciated that about one-fourth of children with non-reactive pupils may survive.
Neurological outcome and disability also correlated with abnormality of pupillary size. Severe deficits were seen in 28% of those with unequal reacting pupils, and in 18% of those with equal reacting but dilated pupils.
Presence of the corneal reflex indicates functional interconnections in the pons. Absent corneal reflexes in children with deep coma prognosticated a poor outcome.
Presence of doll's eye movements suggests intact interconnections between cranial nerve nuclei III, IV and VI via the medial longitudinal fasciculus and intact vestibular input to this system. Asymmetrical or partial absence of eye movements therefore, generally indicates asymmetrical brain stem lesion in mid brain or pons while complete absence of doll's eye movements suggests bilateral structural brainstem abnormality or severe metabolic-toxic encephalopathies.
In the present study, 80% of children, with preserved EOM at presentation, survived whereas, 54% of children with absent EOM died. None of the children with impaired eye movements had normal neurological outcomes; all of them had moderate to severe disability and were not capable of self-care. This goes with the earlier studies that suggest absent or impaired EOM as a sensitive index for prognosticating the outcome. Preserved EOM have been associated with favourable outcome in previous studies.[8], [10] Seshia et al[11] observed that 67% of children who had normal EOM recovered and 16% died; in contrast, all those with absent EOM died and needed assisted ventilation.
In the present study, abnormality of motor pattern was predictive of higher risk of death. Interestingly, it was flaccidity that was associated with higher risk as compared to hypertonia. This was similar to findings of Seshia et al[1] who observed that 84% of those with a normal motor pattern had a normal outcome whereas 82% of those who were flaccid died.
The present study shows clinical signs such as poor pulse volume, absent EOM or non-reactive pupils were strong predictors of death or severe handicap following severe non-traumatic coma. The study also reaffirms that clinical variables and GCS score remain the most readily available tools for assessment of non-traumatic coma, to identify those who are likely to die and those having the greatest potential for recovery. This is particularly helpful in resource-limited countries for directing the limited resources for maximal benefit.
References
1. Seshia SS, Seshia MMK, Sachdeva RK. Coma in childhood. Dev Med Child Neurol 1977; 19: 614-628.
2. Sofiah A, Hussain HM. Childhood non-traumatic coma in Kuala Lumpur, Malaysia. Ann Trop Pediatr 1997; 17: 327-331.
3. Nayana Prabha PC, Nalini P, Serane VT. Role of Glasgow Coma Scale in pediatric nontraumatic coma. Indian Pediatr 2003; 40: 620-625.
4. Ahuja GK, Mohan KK, Prasad K, Behari M. Diagnostic criteria for tuberculous meningitis and their validation. Tuber Lung Dis 1994; 75: 149-152.
5. Kirkham FJ. Non-traumatic coma in children. Arch Dis Child 2001; 85: 303-312.
6. Plum F, Posner JB. The Diagnosis of Stupor and Coma. 3rd edn. Philadelphia; FA Davis CC; 1980.
7. Vijayakumar K, Knight R, Prabhakar P, Murphy PJ, Sharples PM. Neurological outcome in children with non-traumatic coma admitted to a regional paediatric intensive care unit. Arch Dis Child 2003; 88: A30-32.
8. Ogunmekan AO. Non-traumatic coma in childhood: etiology, clinical findings, morbidity, prognosis and mortality. J Trop Pediatr 1983; 29: 230 - 232.
9. Matuja WB, Matekere NJ. Causes and early prognosis of non-traumatic coma in Tanzania. Muhimbili Medical Centre experience. Trop Geogr Med 1987; 39: 330-335.
10. Bates D, Caronna JJ, Cartlidge NE, Knill -Jones RP, Levy DE, Shaw DA, Plum F. A prospective study of nontraumatic coma: methods and results in 310 patients. Ann Neurol 1977; 2: 211- 220.
11. Johnston B, Seshia SS. Prediction of outcome in non-traumatic coma in childhood. Acta Neurol Scand 1984; 69: 417-427.
12. Tasker RC, Matthew JD, Helms P, Dinwiddie R, Boyd S. Monitoring in non-traumatic coma. Part I: invasive intracranial measurements. Arch Dis Child 1988; 63: 888-894.
13. Ishikawa T, Asano Y, Morishima T, Nagashima M, Sobue G, Watanabe K, Yamaguchi H. Epidemiology of acute childhood encephalitis. Aichi Prefecture, Japan, 1984-90. Brain Dev 1993; 15: 192-197.
14. Holbrook PR, Fields IA. Drowning and near drowning. In Holbrook PR, ed. Textbook of Pediatric Critical Care . WB Saunders Co., Philadelphia, PA, 1993; 91-93.(Bansal Arun, Singhi Sunit)