The role of healthcare delivery in the outcome of meningococcal diseas
http://www.100md.com
《英国医生杂志》
1 Infectious Diseases Unit, Department of Paediatrics, Faculty of Medicine, Imperial College of Science, Technology and Medicine, London W2 1PG, 2 Research Unit, Royal College of Paediatrics and Child Health, London W1W 6DE, 3 Faculty of Health and Social Care, University of the West of England, Bristol BS16 1DD, 4 Paediatric Intensive Care Unit and Paediatric Accident and Emergency Department, St Mary's Hospital, London W2 1PG, 5 Department of Paediatrics, Kingston Hospital, Kingston upon Thames KT2 7QB, 6 Centre for Child Health, Queen Mary's School of Medicine and Dentistry, University of London, London E1 1BB
Correspondence to: N Ninis ninisn@gosh.nhs.uk
Objective To determine whether suboptimal management in hospital could contribute to poor outcome in children admitted with meningococcal disease.
Design Case-control study of childhood deaths from meningococcal disease, comparing hospital care in fatal and non-fatal cases.
Setting National statistics and hospital records.
Subjects All children under 17 years who died from meningococcal disease (cases) matched by age with three survivors (controls) from the same region of the country.
Main outcome measures Predefined criteria defined optimal management. A panel of paediatricians blinded to the outcome assessed case records using a standardised form and scored patients for suboptimal management.
Results We identified 143 cases and 355 controls. Departures from optimal (per protocol) management occurred more frequently in the fatal cases than in the survivors. Multivariate analysis identified three factors independently associated with an increased risk of death: failure to be looked after by a paediatrician, failure of sufficient supervision of junior staff, and failure of staff to administer adequate inotropes. Failure to recognise complications of the disease was a significant risk factor for death, although not independently of absence of paediatric care (P = 0.002). The odds ratio for death was 8.7 (95% confidence interval 2.3 to 33) with two failures, increasing with multiple failures.
Conclusions Suboptimal healthcare delivery significantly reduces the likelihood of survival in children with meningococcal disease. Improved training of medical and nursing staff, adherence to published protocols, and increased supervision by consultants may improve the outcome for these children and also those with other life threatening illnesses.
Meningococcal disease remains the most common infectious cause of death in children in many developed countries.1 2 Although treatment on a paediatric intensive care unit improves outcome,3 4 most patients present to their nearest emergency department and many deteriorate so rapidly that death from shock and multiorgan failure often occurs before transfer to a specialist paediatric intensive care unit. The speed with which the diagnosis is made, antibiotics administered, and the complications of shock and multiorgan failure treated is likely to be a major determinant of outcome.5 To test the hypothesis that outcome depends on the quality of health care early in the disease we undertook a national, blinded, case-control study of healthcare delivery in the first 24 hours after admission to hospital in children who died from meningococcal disease compared with those who survived.
Methods
We used the network of regional public health epidemiologists and consultants in communicable diseases in England, Wales, and Northern Ireland and data from the Office for National Statistics to identify cases of meningococcal disease in children aged 0-16 years between 1 December 1997 and 28 February 1999. We used definitions from the Public Health Laboratory Meningococcus Working Group for possible, probable, and confirmed cases of meningococcal disease.6 These definitions are primarily for public health use but we used them to recruit patients through the public health network. We discussed cases of possible or probable meningococcal disease with consultants in disease control and the consultant responsible for the patient. We excluded cases in which it was thought that meningococcal disease was unlikely. For each death (case), we identified three survivors (controls) from the same region of the country matched for age (< 1, 1-4, 5-14, and 15-16 years), corresponding to different risks of mortality.7
A major problem in both the design and analysis of this study was how to control for the expected differences in severity of disease between fatal and non-fatal cases. The children who died were probably more ill than those who survived and would therefore require more medical interventions, which in itself could give rise to greater opportunity for treatment failure. At presentation to hospital, however, children who eventually die are not always sicker than those who survive. Patients presenting with mild disease (for example, with petechial rash and fever only) might progress to severe illness and death if the disease is not recognised and treated early with antibiotics (fig 1). Patients who developed critical illness in hospital or present critically ill might survive or die depending on the speed and quality of care (fig 1). To study failures of healthcare delivery at both stages we identified children who initially presented with mild disease or severe illness and then controlled for the differences in severity of disease in multivariate analysis. To obtain a large enough group of survivors who were severely ill we recruited three controls for each case.
Fig 1 Stages in presentation and progression in children with meningococcal disease. Patients with mild disease on presentation to hospital (A) may progress to severe illness or recover; adequacy of treatment may influence outcome. Patients who are severely ill on presentation to hospital (B) or develop severe illness after presentation at point A may recover or die; adequacy of treatment at point A may influence the outcome
To control for disease severity we used the Glasgow meningococcal septicaemia prognostic score, a well validated severity score that has been shown in numerous studies to predict outcome.8 We also controlled for known factors such as disease presentation (septicaemia or meningitis) and meningococcal serogroup. Furthermore we included the presence of organ failure (see table 1) as a covariate in the multivariate analysis because it is a reliable indicator of disease severity. Finally we assessed failings of fluids and inotrope management in a subgroup of patients who developed cardiovascular failure.
Table 1 Standardised assessment tool for diagnosis of meningococcal disease and complications
Copies of the complete hospital medical and nursing records were received. Some patients were excluded at this stage (because of no microbiological confirmation, absence of any inflammatory markers, atypical clinical presentation, or other confirmed bacterial or viral cause for the illness). All data extracted from clinical material were anonymised and stored with a unique study number.
Standardised evaluation of emergency medical care
Development of a standardised assessment tool
To provide an objective assessment of the promptness and quality of emergency medical care provided, we developed a standardised assessment tool using published and widely accepted criteria for diagnosis and management of meningococcal disease and its complications (table 1).9 Following guidelines in the UK advanced paediatric life support manual,10 we defined the following disease complications (organ failures) namely: cardiovascular failure (shock), respiratory failure, neurological failure, raised intracranial pressure, and haemorrhagic rash. When patients were admitted with tachycardia or tachypnoea but, because of inadequate documentation in the notes, we could not diagnose or rule out a specific organ failure, we categorised the patient as having "abnormal signs only."
Panel
An assessment panel—comprising a consultant in paediatric emergency medicine, a consultant in paediatric infectious diseases, and two consultants in paediatric intensive care—reviewed data on all cases.
Blinded evaluation of patient records using the standardised assessment tool
Vital signs and laboratory results recorded in each patient's notes in the first 24 hours after admission were transcribed on to flow charts in one hour time periods with the time of arrival at hospital taken as time 0 hours. The treatments initiated were also recorded for each hour. The clinical findings and laboratory results were then presented to the panel by revealing the information available at each hour after admission. On the basis of the information available at each hour, the panel members assessed each patient for the presence of diagnostic features of meningococcal disease and its complications. Using the agreed protocol11 they recommended standard management of each complication. CP recorded the panel's decisions and recommended management for each hour. The panel members became aware of the outcome (fatal or not) only after their scoring had been recorded.
By comparing the time after admission at which the panel diagnosed the disease and complications with the time that the hospital team caring for the child reached the diagnosis, and comparing the recommended management of each complication with that which the patient received, we evaluated the actual hospital management, both in terms of timing and the actions undertaken. Delay of more than an hour between the action recommended by the panel and what actually occurred was defined as a failure of care and delay of more than 24 hours in being seen by a consultant as a failure in supervision. The panel assessed whether the failure in care resulted from a failure to recognise the complication or a failure to recognise the severity and to adhere to the protocol. For example, in a hypotensive patient if fluid resuscitation was never instituted at all this was considered a failure to recognise the complication of shock. If fluid resuscitation was started but was inadequate in speed of administration or quantity this was considered to be a failure to appreciate the severity of shock.
The assessment panel scored all patients on admission with the Glasgow meningococcal septicaemia prognostic score,8 and patients were assigned to three groups based on objective clinical features: meningitis (depressed Glasgow coma score, stiff neck, photophobia, and central nervous system failure), septicaemia (shock or multiorgan failure, absence of meningitis), or a mixed picture (some features of meningitis and septicaemia). We also recorded what sort of team (paediatric or adult) primarily cared for the child.
Statistical methods
All statistical analyses were carried out in Stata 8.0 (StataCorp, College Station, TX). We used multivariate conditional logistic regression on matched data with death/survivor status as the outcome variable and failures of care as explanatory variables. Children who died (cases) were matched to survivors (controls) by age group and region of origin. We evaluated a "full" model, which included all the failures of care as well as the effects of potential confounders such as disease severity (Glasgow meningococcal septicaemia prognostic score), disease type, serogroup, organ failure, and whether the patient needed fluid or inotrope therapy. We then used the likelihood ratio test to compare this full model with nested models comprising a subset of failure variables.12 Correlations between explanatory variables were explored by means of univariate logistic regression and Fisher's exact test for contingency tables.
Results
During the study period 190 deaths and 755 survivors were notified (fig 2). We excluded 47 children who died (alternative diagnosis n = 2, death occurred outside hospital n = 28, and incomplete sets of notes n = 17) and 400 survivors (alternative diagnosis n = 106, incomplete sets of notes n = 75, and lack of parental consent n = 219). This left 143 cases and 355 controls to include in the study. Table 2 shows the demographic characteristics of both groups. Organ failure was present in 141 children who died and 169 survivors. For two children who died information in the notes was inadequate for the panel to be able to diagnose any specific organ failure.
Fig 2 Selection of fatal and non-fatal cases for inclusion in the study. Of 945 children with suspected meningococcal disease, we included 143 who died (cases) and 355 who survived (controls)
Table 2 Demographics of children admitted to hospital with meningococcal disease who died (cases) or survived (controls). Figures are number (percentage) of children
Univariate analysis
Table 3 shows the frequency of management failures in cases and controls along with the univariate odds ratios for death. Failures in management were significantly more common in children who died than in survivors. With the exception of serogroup, probability of death was significantly correlated with Glasgow meningococcal septicaemia prognosis score, presence of organ failure, and disease type. Failure to recognise complications, failure to appreciate disease severity, failure in supervision, lack of involvement of a paediatric team in care, and inadequacies of fluid and inotrope administration were all significantly associated with death. Multiple treatment failures significantly increased the risk of death (table 4).
Table 3 Univariate analysis of potential risk factors in management of meningococcal disease
Table 4 Multiple failures in treatment of 480 children with fatal (cases) and non-fatal (controls) meningococcal disease
Multivariate analysis
We excluded sex from the model as it was not significant at the univariate level. We included Glasgow meningococcal septicaemia prognostic score, organ failure, disease type (septicaemia or meningitis), meningococcal serogroup, and the need for fluid or inotrope therapy as potential confounders. The full model indicates that not being under the care of a paediatrician (odds ratio 66.0, 95% confidence interval 3.6 to 1210; P = 0.005), failure of supervision (19.5, 1.8 to 213; P = 0.015), and failure to administer inotropes (23.7, 2.6 to 213; P = 0.005) are independent risk factors for death (table 5). Not being under paediatric care was highly correlated with a failure to recognise complications (P = 0.002; Fisher's exact test). When we removed absence of paediatric care from the model, failure to recognise disease complications became highly significant (6.1, 1.7 to 22; P = 0.006, table 5). This association suggests that failure to recognise complications is one of the consequences of absence of paediatric care. We used the risk factors identified in the multivariate analysis to assess the effect of multiple failures of care on the risk of death. The odds ratio for death with one failure was 8.7 (2.3 to 33) and increased with additional failures (table 6).
Table 5 Multivariate models of treatment of children with meningococcal disease who died or survived. Odds ratios (OR) are for death
Table 6 Multivariate model for multiple failures, with odds ratios for death in children presenting with meningococcal disease
Discussion
Office for National Statistics. Mortality statistics. London: ONS, 2000. (Series DH2 No 26.)
Jones D. Epidemiology of meningococcal disease in Europe and the USA. In: Cartwright K, ed. Meningococcal disease. New York: John Wiley, 1995: 147-57.
Booy R, Habibi P, Nadel S, De Munter C, Britto J, Morrison A, Levin M. Reduction in cases fatality rate from meningococcal disease associated with improved healthcare delivery. Arch Dis Child 2001;85: 386-90.
Thorburn K, Baines P, Thomson A, Hart CA. Mortality in severe meningococcal disease. Arch Dis Child 2001;85: 382-5.
Nadel S, Britto J, Booy R, Maconochie I, Habibi P, Levin M. Avoidable deficiencies in the delivery of healthcare to children with meningococcal disease. J Accid Emerg Med 1998;15: 298-303.
Ramsay M, Kaczmarski E, Rush M, Mallard R, Farrington P, White J. Changing patterns of case ascertainment and trends in meningococcal disease in England and Wales. Commun Dis Rep CDR Rev 1997;7: R49-54.
Stuart JM, Monk PN, Lewis DA, Constantine C, Kaczmarski EB, Cartwright KAV. Management of clusters of meningococcal disease. Commun Dis Rep CDR Rev 1997;7: R3-5.
Thomson APJ, Sills JA, Hart CA. Validation of the Glasgow meningococcal septicaemia prognostic score: a 10 year retrospective survey. Crit Care Med 1991;19: 26-30.
Nadel S, Levin M, Habibi P. Treatment of meningococcal disease in childhood. In: Cartwright K, ed. Meningococcal disease. New York: John Wiley, 1995: 207-43.
Advanced Life Support Group. Advanced paediatric life support manual. London: BMJ Publishing Group, 2001.
Pollard AJ, Britto J, Nadel S, DeMunter C, Habibi P, Levin M. Emergency management of meningococcal disease. Arch Dis Child 1999;80: 290-6.
Hosmer DW, Lemeshow S. Applied logistic regression. Chichester: John Wiley, 1989 (Wiley series in probability and mathematical statistics).
Pathan N, Nadel S, Levin M. Pathophysiology and management of meningococcal septicaemia. J R Coll Physicians London 2000;34: 436-44.(Nelly Ninis, clinical research fellow1, )
Correspondence to: N Ninis ninisn@gosh.nhs.uk
Objective To determine whether suboptimal management in hospital could contribute to poor outcome in children admitted with meningococcal disease.
Design Case-control study of childhood deaths from meningococcal disease, comparing hospital care in fatal and non-fatal cases.
Setting National statistics and hospital records.
Subjects All children under 17 years who died from meningococcal disease (cases) matched by age with three survivors (controls) from the same region of the country.
Main outcome measures Predefined criteria defined optimal management. A panel of paediatricians blinded to the outcome assessed case records using a standardised form and scored patients for suboptimal management.
Results We identified 143 cases and 355 controls. Departures from optimal (per protocol) management occurred more frequently in the fatal cases than in the survivors. Multivariate analysis identified three factors independently associated with an increased risk of death: failure to be looked after by a paediatrician, failure of sufficient supervision of junior staff, and failure of staff to administer adequate inotropes. Failure to recognise complications of the disease was a significant risk factor for death, although not independently of absence of paediatric care (P = 0.002). The odds ratio for death was 8.7 (95% confidence interval 2.3 to 33) with two failures, increasing with multiple failures.
Conclusions Suboptimal healthcare delivery significantly reduces the likelihood of survival in children with meningococcal disease. Improved training of medical and nursing staff, adherence to published protocols, and increased supervision by consultants may improve the outcome for these children and also those with other life threatening illnesses.
Meningococcal disease remains the most common infectious cause of death in children in many developed countries.1 2 Although treatment on a paediatric intensive care unit improves outcome,3 4 most patients present to their nearest emergency department and many deteriorate so rapidly that death from shock and multiorgan failure often occurs before transfer to a specialist paediatric intensive care unit. The speed with which the diagnosis is made, antibiotics administered, and the complications of shock and multiorgan failure treated is likely to be a major determinant of outcome.5 To test the hypothesis that outcome depends on the quality of health care early in the disease we undertook a national, blinded, case-control study of healthcare delivery in the first 24 hours after admission to hospital in children who died from meningococcal disease compared with those who survived.
Methods
We used the network of regional public health epidemiologists and consultants in communicable diseases in England, Wales, and Northern Ireland and data from the Office for National Statistics to identify cases of meningococcal disease in children aged 0-16 years between 1 December 1997 and 28 February 1999. We used definitions from the Public Health Laboratory Meningococcus Working Group for possible, probable, and confirmed cases of meningococcal disease.6 These definitions are primarily for public health use but we used them to recruit patients through the public health network. We discussed cases of possible or probable meningococcal disease with consultants in disease control and the consultant responsible for the patient. We excluded cases in which it was thought that meningococcal disease was unlikely. For each death (case), we identified three survivors (controls) from the same region of the country matched for age (< 1, 1-4, 5-14, and 15-16 years), corresponding to different risks of mortality.7
A major problem in both the design and analysis of this study was how to control for the expected differences in severity of disease between fatal and non-fatal cases. The children who died were probably more ill than those who survived and would therefore require more medical interventions, which in itself could give rise to greater opportunity for treatment failure. At presentation to hospital, however, children who eventually die are not always sicker than those who survive. Patients presenting with mild disease (for example, with petechial rash and fever only) might progress to severe illness and death if the disease is not recognised and treated early with antibiotics (fig 1). Patients who developed critical illness in hospital or present critically ill might survive or die depending on the speed and quality of care (fig 1). To study failures of healthcare delivery at both stages we identified children who initially presented with mild disease or severe illness and then controlled for the differences in severity of disease in multivariate analysis. To obtain a large enough group of survivors who were severely ill we recruited three controls for each case.
Fig 1 Stages in presentation and progression in children with meningococcal disease. Patients with mild disease on presentation to hospital (A) may progress to severe illness or recover; adequacy of treatment may influence outcome. Patients who are severely ill on presentation to hospital (B) or develop severe illness after presentation at point A may recover or die; adequacy of treatment at point A may influence the outcome
To control for disease severity we used the Glasgow meningococcal septicaemia prognostic score, a well validated severity score that has been shown in numerous studies to predict outcome.8 We also controlled for known factors such as disease presentation (septicaemia or meningitis) and meningococcal serogroup. Furthermore we included the presence of organ failure (see table 1) as a covariate in the multivariate analysis because it is a reliable indicator of disease severity. Finally we assessed failings of fluids and inotrope management in a subgroup of patients who developed cardiovascular failure.
Table 1 Standardised assessment tool for diagnosis of meningococcal disease and complications
Copies of the complete hospital medical and nursing records were received. Some patients were excluded at this stage (because of no microbiological confirmation, absence of any inflammatory markers, atypical clinical presentation, or other confirmed bacterial or viral cause for the illness). All data extracted from clinical material were anonymised and stored with a unique study number.
Standardised evaluation of emergency medical care
Development of a standardised assessment tool
To provide an objective assessment of the promptness and quality of emergency medical care provided, we developed a standardised assessment tool using published and widely accepted criteria for diagnosis and management of meningococcal disease and its complications (table 1).9 Following guidelines in the UK advanced paediatric life support manual,10 we defined the following disease complications (organ failures) namely: cardiovascular failure (shock), respiratory failure, neurological failure, raised intracranial pressure, and haemorrhagic rash. When patients were admitted with tachycardia or tachypnoea but, because of inadequate documentation in the notes, we could not diagnose or rule out a specific organ failure, we categorised the patient as having "abnormal signs only."
Panel
An assessment panel—comprising a consultant in paediatric emergency medicine, a consultant in paediatric infectious diseases, and two consultants in paediatric intensive care—reviewed data on all cases.
Blinded evaluation of patient records using the standardised assessment tool
Vital signs and laboratory results recorded in each patient's notes in the first 24 hours after admission were transcribed on to flow charts in one hour time periods with the time of arrival at hospital taken as time 0 hours. The treatments initiated were also recorded for each hour. The clinical findings and laboratory results were then presented to the panel by revealing the information available at each hour after admission. On the basis of the information available at each hour, the panel members assessed each patient for the presence of diagnostic features of meningococcal disease and its complications. Using the agreed protocol11 they recommended standard management of each complication. CP recorded the panel's decisions and recommended management for each hour. The panel members became aware of the outcome (fatal or not) only after their scoring had been recorded.
By comparing the time after admission at which the panel diagnosed the disease and complications with the time that the hospital team caring for the child reached the diagnosis, and comparing the recommended management of each complication with that which the patient received, we evaluated the actual hospital management, both in terms of timing and the actions undertaken. Delay of more than an hour between the action recommended by the panel and what actually occurred was defined as a failure of care and delay of more than 24 hours in being seen by a consultant as a failure in supervision. The panel assessed whether the failure in care resulted from a failure to recognise the complication or a failure to recognise the severity and to adhere to the protocol. For example, in a hypotensive patient if fluid resuscitation was never instituted at all this was considered a failure to recognise the complication of shock. If fluid resuscitation was started but was inadequate in speed of administration or quantity this was considered to be a failure to appreciate the severity of shock.
The assessment panel scored all patients on admission with the Glasgow meningococcal septicaemia prognostic score,8 and patients were assigned to three groups based on objective clinical features: meningitis (depressed Glasgow coma score, stiff neck, photophobia, and central nervous system failure), septicaemia (shock or multiorgan failure, absence of meningitis), or a mixed picture (some features of meningitis and septicaemia). We also recorded what sort of team (paediatric or adult) primarily cared for the child.
Statistical methods
All statistical analyses were carried out in Stata 8.0 (StataCorp, College Station, TX). We used multivariate conditional logistic regression on matched data with death/survivor status as the outcome variable and failures of care as explanatory variables. Children who died (cases) were matched to survivors (controls) by age group and region of origin. We evaluated a "full" model, which included all the failures of care as well as the effects of potential confounders such as disease severity (Glasgow meningococcal septicaemia prognostic score), disease type, serogroup, organ failure, and whether the patient needed fluid or inotrope therapy. We then used the likelihood ratio test to compare this full model with nested models comprising a subset of failure variables.12 Correlations between explanatory variables were explored by means of univariate logistic regression and Fisher's exact test for contingency tables.
Results
During the study period 190 deaths and 755 survivors were notified (fig 2). We excluded 47 children who died (alternative diagnosis n = 2, death occurred outside hospital n = 28, and incomplete sets of notes n = 17) and 400 survivors (alternative diagnosis n = 106, incomplete sets of notes n = 75, and lack of parental consent n = 219). This left 143 cases and 355 controls to include in the study. Table 2 shows the demographic characteristics of both groups. Organ failure was present in 141 children who died and 169 survivors. For two children who died information in the notes was inadequate for the panel to be able to diagnose any specific organ failure.
Fig 2 Selection of fatal and non-fatal cases for inclusion in the study. Of 945 children with suspected meningococcal disease, we included 143 who died (cases) and 355 who survived (controls)
Table 2 Demographics of children admitted to hospital with meningococcal disease who died (cases) or survived (controls). Figures are number (percentage) of children
Univariate analysis
Table 3 shows the frequency of management failures in cases and controls along with the univariate odds ratios for death. Failures in management were significantly more common in children who died than in survivors. With the exception of serogroup, probability of death was significantly correlated with Glasgow meningococcal septicaemia prognosis score, presence of organ failure, and disease type. Failure to recognise complications, failure to appreciate disease severity, failure in supervision, lack of involvement of a paediatric team in care, and inadequacies of fluid and inotrope administration were all significantly associated with death. Multiple treatment failures significantly increased the risk of death (table 4).
Table 3 Univariate analysis of potential risk factors in management of meningococcal disease
Table 4 Multiple failures in treatment of 480 children with fatal (cases) and non-fatal (controls) meningococcal disease
Multivariate analysis
We excluded sex from the model as it was not significant at the univariate level. We included Glasgow meningococcal septicaemia prognostic score, organ failure, disease type (septicaemia or meningitis), meningococcal serogroup, and the need for fluid or inotrope therapy as potential confounders. The full model indicates that not being under the care of a paediatrician (odds ratio 66.0, 95% confidence interval 3.6 to 1210; P = 0.005), failure of supervision (19.5, 1.8 to 213; P = 0.015), and failure to administer inotropes (23.7, 2.6 to 213; P = 0.005) are independent risk factors for death (table 5). Not being under paediatric care was highly correlated with a failure to recognise complications (P = 0.002; Fisher's exact test). When we removed absence of paediatric care from the model, failure to recognise disease complications became highly significant (6.1, 1.7 to 22; P = 0.006, table 5). This association suggests that failure to recognise complications is one of the consequences of absence of paediatric care. We used the risk factors identified in the multivariate analysis to assess the effect of multiple failures of care on the risk of death. The odds ratio for death with one failure was 8.7 (2.3 to 33) and increased with additional failures (table 6).
Table 5 Multivariate models of treatment of children with meningococcal disease who died or survived. Odds ratios (OR) are for death
Table 6 Multivariate model for multiple failures, with odds ratios for death in children presenting with meningococcal disease
Discussion
Office for National Statistics. Mortality statistics. London: ONS, 2000. (Series DH2 No 26.)
Jones D. Epidemiology of meningococcal disease in Europe and the USA. In: Cartwright K, ed. Meningococcal disease. New York: John Wiley, 1995: 147-57.
Booy R, Habibi P, Nadel S, De Munter C, Britto J, Morrison A, Levin M. Reduction in cases fatality rate from meningococcal disease associated with improved healthcare delivery. Arch Dis Child 2001;85: 386-90.
Thorburn K, Baines P, Thomson A, Hart CA. Mortality in severe meningococcal disease. Arch Dis Child 2001;85: 382-5.
Nadel S, Britto J, Booy R, Maconochie I, Habibi P, Levin M. Avoidable deficiencies in the delivery of healthcare to children with meningococcal disease. J Accid Emerg Med 1998;15: 298-303.
Ramsay M, Kaczmarski E, Rush M, Mallard R, Farrington P, White J. Changing patterns of case ascertainment and trends in meningococcal disease in England and Wales. Commun Dis Rep CDR Rev 1997;7: R49-54.
Stuart JM, Monk PN, Lewis DA, Constantine C, Kaczmarski EB, Cartwright KAV. Management of clusters of meningococcal disease. Commun Dis Rep CDR Rev 1997;7: R3-5.
Thomson APJ, Sills JA, Hart CA. Validation of the Glasgow meningococcal septicaemia prognostic score: a 10 year retrospective survey. Crit Care Med 1991;19: 26-30.
Nadel S, Levin M, Habibi P. Treatment of meningococcal disease in childhood. In: Cartwright K, ed. Meningococcal disease. New York: John Wiley, 1995: 207-43.
Advanced Life Support Group. Advanced paediatric life support manual. London: BMJ Publishing Group, 2001.
Pollard AJ, Britto J, Nadel S, DeMunter C, Habibi P, Levin M. Emergency management of meningococcal disease. Arch Dis Child 1999;80: 290-6.
Hosmer DW, Lemeshow S. Applied logistic regression. Chichester: John Wiley, 1989 (Wiley series in probability and mathematical statistics).
Pathan N, Nadel S, Levin M. Pathophysiology and management of meningococcal septicaemia. J R Coll Physicians London 2000;34: 436-44.(Nelly Ninis, clinical research fellow1, )