- Email: [email protected]
PHIShing for Clarity in the Management of Apparent Life-Threatening Events
Even more importantly, determining such a test can function properly in the setting of pancreatic dysfunction will require the inclusion of subjects with impaired glucose tolerance and diabetes, both type 1 and type 2. In addition, longitudinal studies are needed to determine whether the index can in fact predict the development of glucose intolerance or diabetes. The fact that C-peptide correlated better than insulin is not surprising. As the authors mention, serum insulin concentrations are affected by changes in insulin clearance. In addition, C-peptide is more reliable when extended to multicenter trials and clinical care, because insulin assays are notoriously variable and multiple insulin assay methods exist, contributing to the high degree of variability between centers. Furthermore, insulin assays may not be accurate in a subject who has received previous insulin injections or infusions, due to anti-insulin antibodies that can interfere with the assay. Insulin assays also cross-react with insulinomimetic compounds and proinsulin, the secretion of which varies as pancreatic function declines.5 It is convenient that results were equivalent whether the challenge was glucola or Boost shakes. Boost may increase the availability and tolerability of such a test in the clinical pediatric setting. Boost does not require refrigeration, is more palatable, and is less expensive. In addition, because Boost is not composed solely of glucose, it may be a challenge to the pancreas that is more representative of an actual meal and may cause less nausea. However, because the American Diabetes Association guidelines for the diagnosis of diabetes are based on the glucose response to glucola, not Boost, a clinician wishing to diagnose impaired glucose tolerance or diabetes will still need to use glucola. Although Bacha et al’s results are quite promising, the 15-minute C-peptide–to-insulin ratio as a measure of pan-
creatic -cell function currently remains applicable only to the research setting. There is still no evidence that this technique is applicable to patients with abnormal pancreatic function, or that the ratio is predictive of declining -cell function over time. In addition, the 15-minute time point provides no information about the existence of diabetes or prediabetes, which still requires measurement of glucose at the fasting and 2-hour time points after ingestion of a standard dose of glucola. Therefore, it is not yet time to add the proposed 15-minute index to a standard clinical glucose tolerance test. The intriguing possibility remains, however, that the inclusion of an additional time point at 15 minutes will one day provide important information regarding -cell function and the future risk of diabetes. Kristen Nadeau, MD Philip S. Zeitler, MD, PhD Division of Endocrinology Department of Pediatrics University of Colorado Denver School of Medicine Aurora, Colorado
REFERENCES 1. Bacha F, Gungor N, Arslanian S. Measures of -cell function during glucose tolerance test and liquid mixed-meal and the hyperglycemic clamp. J Pediatr 2007;152: 618-21 . 2. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979;237:E214-23. 3. Uwaifo GI, Fallon EM, Chin J, Elberg J, Parikh SJ, Yanovski JA. Indices of insulin action, disposal, and secretion derived from fasting samples and clamps in normal glucose-tolerant black and white children. Diabetes Care 2002;25:2081-7. 4. Hansen T, Drivsholm T, Urhammer SA, Palacios RT, Volund A, Borch-Johnsen K, et al. The BIGTT test: a novel test for simultaneous measurement of pancreatic -cell function, insulin sensitivity, and glucose tolerance. Diabetes Care 2007;30:257-62. 5. Mitrakou A, Vuorinen-Markkola H, Raptis G, Toft I, Mokan M, Strumph P, et al. Simultaneous assessment of insulin secretion and insulin sensitivity using a hyperglycemia clamp. J Clin Endocrinol Metab 1992;75:379-82.
PHIShing for Clarity in the Management of Apparent Life-Threatening Events
n apparent life-threatening event (ALTE) is defined as “an episode in the first year of life that appears potentially life threatening to the observer and is characterized by some combination of color change, apnea, alteration in muscle tone, and choking or gagging.”1 A major difficulty inherent in this definition is that it is usually a parent for whom this event is apparent. The frightening
A ALTE GE ICD-9 PHIS
604
Apparent life-threatening event Gastroesophageal International Classification of Diseases, 9th edition Pediatric Health Information System
Editorials
nature of the event affects the observations and memory of the reporting party and limits the value of the history. Management is relatively straightforward for the child who presents with obvious signs and symptoms but is more problematic for the child who is
See related article, p 629 Reprint requests: Jack M. Percelay, MD, MPH, E.L.M.O. Pediatrics, Pediatric Board Member, Society of Hospital Medicine, 1735 York Ave, Apt 23B, New York, NY 10128. E-mail: [email protected]. J Pediatr 2008;152:604-6 0022-3476/$ - see front matter Copyright © 2008 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2008.01.030
The Journal of Pediatrics • May 2008
clinically well on presentation, particularly if this was a single event requiring little or no intervention and no obvious cause is identified after a careful history and physical examination. Recent prospective studies have attempted to identify historical features on presentation to the emergency department that warrant admission2 and tests that are of value in evaluating an ALTE.3 National guidelines for ALTEs are notably absent. No publications have met the rigor to qualify for www.guidelines.gov. Development of a national standard of care for ALTEs is made more complex by the requirement to involve multiple subspecialties given the wide range of possible causes and presentations for ALTEs. Moreover, there is no formal International Classification of Diseases–9th revision (ICD-9) diagnostic code for ALTE that captures the clinical entity described above. Given the potential for a devastating outcome after an ALTE and fears of liability, ALTE management is subject to marked local variation and a tendency toward extensive but frequently unproductive evaluation and management. In this issue of The Journal, Tieder et al4 use the power of the Pediatric Health Information System (PHIS) database to document the wide range of variation of ALTE management across leading freestanding children’s hospitals and to call for the development of national evidencebased standards. The PHIS database is an administrative database that includes hospital stay, demographic, and diagnostic data, as well as information on medication, diagnostic tests, and selected procedures. Mortality and readmission rates can be determined, but individual chart review and more specific indicators of morbidity are not accessible. With this tool, the authors were able to identify retrospectively 12,067 hospital admissions over a nearly 5-year period, with ICD-9 discharge codes potentially identifiable as an ALTE. The lack of a specific ICD-9 code for ALTEs forced the authors to use a collection of proxy codes to identify ALTEs, specifically apnea, respiratory problems of newborns, syncope, altered consciousness, transient loss of consciousness, and cyanosis. Certain comorbid conditions were excluded (prematurity, congenital heart disease, and some genetic conditions), but discharge diagnoses such as seizures, meningitis, trauma, bronchiolitis, and gastroesophageal (GE) reflux were not. In this way, the authors attempted to collect patients who presumably presented primarily with an ALTE of undetermined cause and subsequently had an underlying condition identified, rather than patients with a primary diagnosis such as status epilepticus, who presented with incidental cyanosis. The in-hospital mortality rate for this population was 0.56%. As would be expected, the most common discharge diagnoses were GE reflux and lower respiratory tract infections, with bacterial infections, cardiovascular conditions, and seizures cited as the next most frequently identified diagnoses. Cardiovascular disorders and GE reflux had increased odds risks for readmission. Tieder et al4 then analyzed length of stay and hospital Editorials
costs, resource utilization, and treatments by individual hospitals while controlling for demographic and diagnostic factors to determine the extent to which hospital-level variation contributed to differences in ALTE management. The differences were dramatic. At some institutions, 5% to 10% of admissions had a 1-day length of stay compared with 25% to 30% of admissions at other institutions. Other institutions had a 5-day length of stay for 40% of their admissions compared with 20% of admissions at other institutions. Most patients underwent a complete blood count and chest radiography, with relatively little variation across institutions, compared with a smaller percentage of patients who underwent sleep testing and electroencephalography, but with marked variation across institutions. The authors rightly conclude that the variation of management across institutions reflects lack of a national consensus for the appropriate evaluation and treatment of ALTEs. Local beliefs prevail over national evidence-based standards. Limitations of the study are acknowledged. A retrospective review of administrative data sets is clearly inferior to prospective collection of clinical data at the individual chart level. More importantly, the lack of a uniform ICD-9 code for ALTEs suggests that variable coding across institutions would further contribute to variability in the data. However, given the extensive range of proxy codes used by the authors, and the large sample size of the study, one is hard pressed to assume that a prospective study would not reveal similar marked variations in practice management across institutions. Nonetheless, it would have been helpful if the authors had been able to obtain approval from the Institutional Review Board to review (a sample of) qualifying charts from their home institution to validate the proxy codes they selected to use as surrogates for the clinical diagnosis of ALTE. Similarly, it would have been insightful if approval from the Institutional Review Board were granted for review of the charts of the 68 patients who died. Did these children present with obvious signs and symptoms of serious illness warranting immediate evaluation and treatment or were they well-appearing without an obvious underlying condition? Would any of them have been sent home from the emergency department with the Claudius criteria of age ⬎ 1 month and a single event?2 The strength of PHIS as a tool is the relative ease with which a large sample can be examined and wide variations in practice management identified. The limitations of PHIS as a tool are the inability to link utilization data to outcomes and the failure to collect clinically meaningful data. Readers and researchers are left to wonder which approach is best. Nihilists will assume that patient populations and outcomes across institutions are generally similar, point to the variation in utilization, and argue that less is more. Those predisposed to a more extensive evaluation need only mention the 0.56% mortality figure cited in this article to argue for supersizing the ALTE workup menu. The truth most likely lies in the middle. Without access to outcomes data, the authors are unable to use the PHIS database to support a specific approach to the well-appearing patient with an ALTE. They can raise questions about the evaluation and treatment of GE reflux on the basis of previous studies and 605
suggest additional research, but PHIS data alone cannot jumpstart the process beyond the call to action. Nonetheless, Tieder et al4 have provided all the information we need to know about variation in ALTE management. There is too much. The next steps are to determine which approach provides the best outcomes and to move from single-site studies to multisite studies. The American Academy of Pediatrics Committee on Coding and Nomenclature has submitted a proposal for adding an ICD-9 code specifically for ALTE (unpublished data, 2008). Such a code is crucial for future studies that use either administrative data sets or individual chart review. Individual institutions should develop their own guidelines, with current studies used as a reference. Some institutions will err on the side of doing less, some on the side of doing more. But by reducing variability among practitioners in a single institution, strategies can be compared by measuring outcomes at Institution A versus Institution B. National and international organizations should collaborate across subspecialties on the development of ALTE guidelines that particularly address “the otherwise well-appearing child without an obvious underlying condition.”4 If consensus cannot be reached, areas for investigation should be highlighted. The study by Tieder et al4 has left us with many questions about the optimal management of ALTEs. The variability is a confession of ignorance. It is unfortunate that the PHIS database is not linked to outcomes data that would provide for more direction. These data were initially presented at the 2007 Pediatric Academic Societies meeting, along with several other PHIS studies that highlighted variation in management of complicated pneumonia,5 the duration of parenteral antibiotics for osteomyelitis,6 and the inpatient management of urinary tract infections.7 Similar scattergrams of interinstitutional variability were reported for these clinical entities. In no cases could the PHIS researchers assert that either earlier video-assisted thoracoscopy, earlier
606
Editorials
conversion to oral antibiotics, or earlier voiding cystourethrograms were associated with equal, better, or worse outcomes. There must be some way to overcome the limitations of the PHIS database as it is currently constructed to create linkages to outcomes data. Privacy concerns need to be respected for both individual patients and institutions, but these restrictions are limiting the ability of qualified researchers to identify best practices in areas where prospective, randomized clinical trials have not yet been conducted. The natural variability in practice patterns that exists among hospitals participating in the PHIS database represents an opportunity to advance the care of children internationally. It is unacceptable to be able to report only that wide variation exists. Clarity requires reporting outcomes and suggesting optimal management strategies. Information is power. With great power comes great responsibility. The PHIS community has the responsibility to report outcomes, as well as utilization. Jack M. Percelay, MD, MPH E.L.M.O. Pediatrics New York, New York
REFERENCES 1. National Institutes of Health. Consensus Development Conference on Infantile Apnea and Home Monitoring, Sept 29 to Oct1, 1986. Consensus statement. Pediatrics 1987;79:292-9. 2. Claudius I, Keens T. Do all infants with apparent life-threatening events need to be admitted? Pediatrics 2007;119:679-83. 3. Brand DA, Altman RL, Purtill K, Edwards KS. Yield of diagnostic testing in infants who have had an apparent life-threatening event. Pediatrics 2005;115;885-93. 4. Tieder JS, Cowan CA, Garrison MM, Christakis DA. Variation in inpatient resource utilization and management of apparent life-threatening events. J Pediatr 2008; 152:629-35. 5. Zaoutis T, Keren R, Heydon K. Parenteral vs. oral antimicrobial therapy for the treatment of acute osteomyelitis. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada. 6. Shah SS, Cara MD, Bell LM. Factors associated with prolonged length of stay and repeat procedures in children with complicated pneumonia. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada. 7. Conway PH, Keren R. Inpatient urinary tract infections: variability in care and outcomes. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada.
The Journal of Pediatrics • May 2008
creatic -cell function currently remains applicable only to the research setting. There is still no evidence that this technique is applicable to patients with abnormal pancreatic function, or that the ratio is predictive of declining -cell function over time. In addition, the 15-minute time point provides no information about the existence of diabetes or prediabetes, which still requires measurement of glucose at the fasting and 2-hour time points after ingestion of a standard dose of glucola. Therefore, it is not yet time to add the proposed 15-minute index to a standard clinical glucose tolerance test. The intriguing possibility remains, however, that the inclusion of an additional time point at 15 minutes will one day provide important information regarding -cell function and the future risk of diabetes. Kristen Nadeau, MD Philip S. Zeitler, MD, PhD Division of Endocrinology Department of Pediatrics University of Colorado Denver School of Medicine Aurora, Colorado
REFERENCES 1. Bacha F, Gungor N, Arslanian S. Measures of -cell function during glucose tolerance test and liquid mixed-meal and the hyperglycemic clamp. J Pediatr 2007;152: 618-21 . 2. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979;237:E214-23. 3. Uwaifo GI, Fallon EM, Chin J, Elberg J, Parikh SJ, Yanovski JA. Indices of insulin action, disposal, and secretion derived from fasting samples and clamps in normal glucose-tolerant black and white children. Diabetes Care 2002;25:2081-7. 4. Hansen T, Drivsholm T, Urhammer SA, Palacios RT, Volund A, Borch-Johnsen K, et al. The BIGTT test: a novel test for simultaneous measurement of pancreatic -cell function, insulin sensitivity, and glucose tolerance. Diabetes Care 2007;30:257-62. 5. Mitrakou A, Vuorinen-Markkola H, Raptis G, Toft I, Mokan M, Strumph P, et al. Simultaneous assessment of insulin secretion and insulin sensitivity using a hyperglycemia clamp. J Clin Endocrinol Metab 1992;75:379-82.
PHIShing for Clarity in the Management of Apparent Life-Threatening Events
n apparent life-threatening event (ALTE) is defined as “an episode in the first year of life that appears potentially life threatening to the observer and is characterized by some combination of color change, apnea, alteration in muscle tone, and choking or gagging.”1 A major difficulty inherent in this definition is that it is usually a parent for whom this event is apparent. The frightening
A ALTE GE ICD-9 PHIS
604
Apparent life-threatening event Gastroesophageal International Classification of Diseases, 9th edition Pediatric Health Information System
Editorials
nature of the event affects the observations and memory of the reporting party and limits the value of the history. Management is relatively straightforward for the child who presents with obvious signs and symptoms but is more problematic for the child who is
See related article, p 629 Reprint requests: Jack M. Percelay, MD, MPH, E.L.M.O. Pediatrics, Pediatric Board Member, Society of Hospital Medicine, 1735 York Ave, Apt 23B, New York, NY 10128. E-mail: [email protected]. J Pediatr 2008;152:604-6 0022-3476/$ - see front matter Copyright © 2008 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2008.01.030
The Journal of Pediatrics • May 2008
clinically well on presentation, particularly if this was a single event requiring little or no intervention and no obvious cause is identified after a careful history and physical examination. Recent prospective studies have attempted to identify historical features on presentation to the emergency department that warrant admission2 and tests that are of value in evaluating an ALTE.3 National guidelines for ALTEs are notably absent. No publications have met the rigor to qualify for www.guidelines.gov. Development of a national standard of care for ALTEs is made more complex by the requirement to involve multiple subspecialties given the wide range of possible causes and presentations for ALTEs. Moreover, there is no formal International Classification of Diseases–9th revision (ICD-9) diagnostic code for ALTE that captures the clinical entity described above. Given the potential for a devastating outcome after an ALTE and fears of liability, ALTE management is subject to marked local variation and a tendency toward extensive but frequently unproductive evaluation and management. In this issue of The Journal, Tieder et al4 use the power of the Pediatric Health Information System (PHIS) database to document the wide range of variation of ALTE management across leading freestanding children’s hospitals and to call for the development of national evidencebased standards. The PHIS database is an administrative database that includes hospital stay, demographic, and diagnostic data, as well as information on medication, diagnostic tests, and selected procedures. Mortality and readmission rates can be determined, but individual chart review and more specific indicators of morbidity are not accessible. With this tool, the authors were able to identify retrospectively 12,067 hospital admissions over a nearly 5-year period, with ICD-9 discharge codes potentially identifiable as an ALTE. The lack of a specific ICD-9 code for ALTEs forced the authors to use a collection of proxy codes to identify ALTEs, specifically apnea, respiratory problems of newborns, syncope, altered consciousness, transient loss of consciousness, and cyanosis. Certain comorbid conditions were excluded (prematurity, congenital heart disease, and some genetic conditions), but discharge diagnoses such as seizures, meningitis, trauma, bronchiolitis, and gastroesophageal (GE) reflux were not. In this way, the authors attempted to collect patients who presumably presented primarily with an ALTE of undetermined cause and subsequently had an underlying condition identified, rather than patients with a primary diagnosis such as status epilepticus, who presented with incidental cyanosis. The in-hospital mortality rate for this population was 0.56%. As would be expected, the most common discharge diagnoses were GE reflux and lower respiratory tract infections, with bacterial infections, cardiovascular conditions, and seizures cited as the next most frequently identified diagnoses. Cardiovascular disorders and GE reflux had increased odds risks for readmission. Tieder et al4 then analyzed length of stay and hospital Editorials
costs, resource utilization, and treatments by individual hospitals while controlling for demographic and diagnostic factors to determine the extent to which hospital-level variation contributed to differences in ALTE management. The differences were dramatic. At some institutions, 5% to 10% of admissions had a 1-day length of stay compared with 25% to 30% of admissions at other institutions. Other institutions had a 5-day length of stay for 40% of their admissions compared with 20% of admissions at other institutions. Most patients underwent a complete blood count and chest radiography, with relatively little variation across institutions, compared with a smaller percentage of patients who underwent sleep testing and electroencephalography, but with marked variation across institutions. The authors rightly conclude that the variation of management across institutions reflects lack of a national consensus for the appropriate evaluation and treatment of ALTEs. Local beliefs prevail over national evidence-based standards. Limitations of the study are acknowledged. A retrospective review of administrative data sets is clearly inferior to prospective collection of clinical data at the individual chart level. More importantly, the lack of a uniform ICD-9 code for ALTEs suggests that variable coding across institutions would further contribute to variability in the data. However, given the extensive range of proxy codes used by the authors, and the large sample size of the study, one is hard pressed to assume that a prospective study would not reveal similar marked variations in practice management across institutions. Nonetheless, it would have been helpful if the authors had been able to obtain approval from the Institutional Review Board to review (a sample of) qualifying charts from their home institution to validate the proxy codes they selected to use as surrogates for the clinical diagnosis of ALTE. Similarly, it would have been insightful if approval from the Institutional Review Board were granted for review of the charts of the 68 patients who died. Did these children present with obvious signs and symptoms of serious illness warranting immediate evaluation and treatment or were they well-appearing without an obvious underlying condition? Would any of them have been sent home from the emergency department with the Claudius criteria of age ⬎ 1 month and a single event?2 The strength of PHIS as a tool is the relative ease with which a large sample can be examined and wide variations in practice management identified. The limitations of PHIS as a tool are the inability to link utilization data to outcomes and the failure to collect clinically meaningful data. Readers and researchers are left to wonder which approach is best. Nihilists will assume that patient populations and outcomes across institutions are generally similar, point to the variation in utilization, and argue that less is more. Those predisposed to a more extensive evaluation need only mention the 0.56% mortality figure cited in this article to argue for supersizing the ALTE workup menu. The truth most likely lies in the middle. Without access to outcomes data, the authors are unable to use the PHIS database to support a specific approach to the well-appearing patient with an ALTE. They can raise questions about the evaluation and treatment of GE reflux on the basis of previous studies and 605
suggest additional research, but PHIS data alone cannot jumpstart the process beyond the call to action. Nonetheless, Tieder et al4 have provided all the information we need to know about variation in ALTE management. There is too much. The next steps are to determine which approach provides the best outcomes and to move from single-site studies to multisite studies. The American Academy of Pediatrics Committee on Coding and Nomenclature has submitted a proposal for adding an ICD-9 code specifically for ALTE (unpublished data, 2008). Such a code is crucial for future studies that use either administrative data sets or individual chart review. Individual institutions should develop their own guidelines, with current studies used as a reference. Some institutions will err on the side of doing less, some on the side of doing more. But by reducing variability among practitioners in a single institution, strategies can be compared by measuring outcomes at Institution A versus Institution B. National and international organizations should collaborate across subspecialties on the development of ALTE guidelines that particularly address “the otherwise well-appearing child without an obvious underlying condition.”4 If consensus cannot be reached, areas for investigation should be highlighted. The study by Tieder et al4 has left us with many questions about the optimal management of ALTEs. The variability is a confession of ignorance. It is unfortunate that the PHIS database is not linked to outcomes data that would provide for more direction. These data were initially presented at the 2007 Pediatric Academic Societies meeting, along with several other PHIS studies that highlighted variation in management of complicated pneumonia,5 the duration of parenteral antibiotics for osteomyelitis,6 and the inpatient management of urinary tract infections.7 Similar scattergrams of interinstitutional variability were reported for these clinical entities. In no cases could the PHIS researchers assert that either earlier video-assisted thoracoscopy, earlier
606
Editorials
conversion to oral antibiotics, or earlier voiding cystourethrograms were associated with equal, better, or worse outcomes. There must be some way to overcome the limitations of the PHIS database as it is currently constructed to create linkages to outcomes data. Privacy concerns need to be respected for both individual patients and institutions, but these restrictions are limiting the ability of qualified researchers to identify best practices in areas where prospective, randomized clinical trials have not yet been conducted. The natural variability in practice patterns that exists among hospitals participating in the PHIS database represents an opportunity to advance the care of children internationally. It is unacceptable to be able to report only that wide variation exists. Clarity requires reporting outcomes and suggesting optimal management strategies. Information is power. With great power comes great responsibility. The PHIS community has the responsibility to report outcomes, as well as utilization. Jack M. Percelay, MD, MPH E.L.M.O. Pediatrics New York, New York
REFERENCES 1. National Institutes of Health. Consensus Development Conference on Infantile Apnea and Home Monitoring, Sept 29 to Oct1, 1986. Consensus statement. Pediatrics 1987;79:292-9. 2. Claudius I, Keens T. Do all infants with apparent life-threatening events need to be admitted? Pediatrics 2007;119:679-83. 3. Brand DA, Altman RL, Purtill K, Edwards KS. Yield of diagnostic testing in infants who have had an apparent life-threatening event. Pediatrics 2005;115;885-93. 4. Tieder JS, Cowan CA, Garrison MM, Christakis DA. Variation in inpatient resource utilization and management of apparent life-threatening events. J Pediatr 2008; 152:629-35. 5. Zaoutis T, Keren R, Heydon K. Parenteral vs. oral antimicrobial therapy for the treatment of acute osteomyelitis. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada. 6. Shah SS, Cara MD, Bell LM. Factors associated with prolonged length of stay and repeat procedures in children with complicated pneumonia. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada. 7. Conway PH, Keren R. Inpatient urinary tract infections: variability in care and outcomes. Presented at the 2007 Pediatric Academic Societies’ Annual Meeting; May 5-8, 2007; Toronto, Ontario, Canada.
The Journal of Pediatrics • May 2008