search text   Advanced Search

This Article Abstract Full Text (PDF) P3Rs: Submit a response Alert me when this article is cited Alert me when P3Rs are posted Alert me if a correction is posted Citation Map Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Griffin, M. P. Articles by Moorman, J. R. Search for Related Content PubMed PubMed Citation Articles by Griffin, M. P. Articles by Moorman, J. R. Related Collections Premature & Newborn

Heart Rate Characteristics: Novel Physiomarkers to Predict Neonatal Infection and Death

M. Pamela Griffin, MD^*, Douglas E. Lake, PhD, Eric A. Bissonette, MS, Frank E. Harrell, Jr, PhD^||, T. Michael O'Shea, MD^¶ and J. Randall Moorman, MD

Departments of ^* Pediatrics
Internal Medicine and the Cardiovascular Research Center
Health Evaluation Sciences, University of Virginia Health System, Charlottesville, Virginia
^|| Department of Biostatistics, Vanderbilt University, Nashville, Tennessee
^¶ Department of Pediatrics, Wake Forest School of Medicine, Winston-Salem, North Carolina

	ABSTRACT

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
Objective. Monitoring of regulated physiologic processes using physiomarkers such as heart rate variability may be important in the early diagnosis of subacute, potentially catastrophic illness. Early in the course of neonatal sepsis, there are physiomarkers of reduced heart rate variability and transient decelerations similar to fetal distress. The goal of this study was to determine the degree of increased risk for sepsis, urinary tract infection (UTI), and death when these abnormal heart rate characteristics (HRC) were observed.

Methods. We monitored 1022 infants at 2 tertiary care NICUs, 458 of whom were very low birth weight. We calculated an HRC index from validated regression models relating mathematical features of heart rate time series and histograms to episodes of illness. We calculated the risks for adverse events of sepsis, UTI, and death for infants stratified by HRC measurements.

Results. Compared with infants with low-risk HRC measurements, infants with high-risk HRC measurements had 5- to 6-fold increased risk for an adverse event in the next day and 3-fold increased risk in the next week. Laboratory tests that were relevant to infection added information to HRC measurements. Infants with both high-risk HRC and abnormal laboratory tests had 6- to 7-fold increased risk for an adverse event in the next day compared with infants who had neither.

Conclusion. HRC are noninvasively monitored physiomarkers that identify infants in the NICU who are at high risk for sepsis, UTI, and death.

Key Words: infant • mortality prediction • sepsis • urinary tract infections • physiomarker

Abbreviations: HRC, heart rate characteristics • UTI, urinary tract infection • UVA, University of Virginia • HR, heart rate • WFU, Wake Forest University • CONS, coagulase-negative Staphylococcus • NICHD, National Institute of Child Health and Human Development • VLBW, very low birth weight

The common clinical scenario of abrupt deterioration of a stable infant in the NICU usually prompts the consideration of sepsis. Establishing the diagnosis of neonatal sepsis using current clinical and laboratory methods, though, is difficult.^1,2 Even when tests are performed at the time that the infant is obviously ill clinically, there is no consensus about diagnostic clinical or laboratory findings.^3–5 Blood cultures, which should be the final arbiter in the diagnosis of bacteremia, are not reliable in this clinical setting, where very small amounts are usually obtained for testing.^6–9 Bacteremia, moreover, may be transient and intermittent.

Abnormal heart rate characteristics (HRC) of reduced variability and transient decelerations occur early in the course of neonatal sepsis.^10–12 We previously developed algorithms to detect these HRC and showed significant association of them with sepsis¹⁰ and with death¹¹ at 2 university tertiary care referral NICUs. Here, we evaluate continuous neonatal HRC monitoring as a clinical tool to identify infants who are at increased risk for having sepsis, urinary tract infection (UTI), or death in the NICU.

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
Study Design
We prospectively studied all admissions to the University of Virginia (UVA) NICU from September 1999 to July 2003 and to the Wake Forest University (WFU) NICU from September 1999 to March 2001. Institutional Review Boards at both sites approved the study. We prospectively recorded times and results of blood and urine cultures that were obtained for the clinical suspicion of sepsis and times of death. Laboratory test results were available for the UVA patients from a clinical archive system.

We continuously monitored heart rate (HR) from the bedside electrocardiogram monitors and calculated the HRC index every 6 hours. The HRC index is derived from a multivariable logistic regression model that relates HR statistics to acute neonatal illness in the next 24 hours. The model was initially generated at UVA using data from 316 of the study infants and validated at WFU on 317 of the study infants.¹² The HR parameters of SD, sample asymmetry,¹³ and sample entropy^14,15 report on reduced variability and transient decelerations that occur early in the course of neonatal sepsis.¹⁰ Model coefficients and predictive performance were similar whether the outcome variable was proven sepsis or suspected sepsis or whether the time window was the day before or the day of the event. All models that were generated at 1 site could be validated at the other.

We defined "adverse events" as sepsis, UTI, and death. We defined "sepsis" to be present when a physician suspected the diagnosis, obtained a blood culture that grew bacteria not ordinarily considered to be a contaminant, and initiated antibiotic therapy of 5 days' duration or more. This definition is consistent with the diagnosis of "proven sepsis" offered by the Centers for Disease Control and Prevention.¹⁶ For infections with coagulase-negative Staphylococcus (CONS), our definition is consistent with the National Institute of Child Health and Human Development (NICHD) Neonatal Research Network definition of possible CONS sepsis.¹⁷ We defined UTI to be present when a physician detected signs of illness, obtained a urine sample from a catheter or suprapubic aspiration that grew bacteria, and initiated antibiotic therapy. An abnormal laboratory result was considered to be present when leucocytosis or leucopenia (white blood cell count >20000/µL or <5000/µL), high proportion of circulating immature neutrophils (I:T ratio >0.2), thrombocytopenia (platelet count <100000/µL), or hyperglycemia (blood glucose >180 mg/dL) were recorded within the previous 24 hours.

Statistical Analysis
The HRC index is a continuous measure. To facilitate clinical application of the HRC index, we defined 3 levels of risk on the basis of the model values. The lowest 70% of HRC measurements were less than or equal to the mean value and thus corresponded to a relative risk of 1 or less for acute neonatal illness in the next 24 hours. We categorized these as low risk. Ninety percent of values were less than or equal to twice the mean value, so values between the 70th and 90th percentiles correspond to 1- to 2-fold increase in relative risk, and we categorized these as intermediate risk. HRC values above the 90th percentile correspond to 2-fold and higher risk, and we categorized these as high risk.

The population-based distribution of HRC index is approximately log-normal at the 2 sites but shifted to higher values at WFU.¹² We attribute this to a higher level of illness severity at WFU, which, unlike UVA, has no in-born patients and admits only infants who are referred from other hospitals. The proportion of extremely low birth weight infants (<750 g) is higher at WFU, and there was a trend toward higher mortality rate (7% vs 4%).¹¹ Accordingly, HRC index values were assigned to percentiles of the distribution for each site.

Because of the continuously repeated measures, statistical inference was performed using a cluster bootstrap technique whereby 1000 new samples of the same size were obtained by resampling the infants with replacement.¹⁸ The 2.5th and 97.5th percentiles of the sample of risks are used as lower and upper limits for a 95% confidence interval around the observer risk in the original sample. The P value to measure the statistical significance of increased risk is evaluated by the proportion of resampled observations with risk 1.

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
Study Population
We studied 1022 infants, in whom there were 379 adverse events: 223 episodes of sepsis, 108 of UTI, and 48 deaths. Most events were in very low birth weight (VLBW) infants, as shown in Table 1. In the VLBW infants, 30% (137 of 458) had 1 or more episodes of sepsis, and 34 (7.4%) died. These rates are similar to the NICHD Neonatal Research Network experience.¹⁷ There were 105706 6-hour epochs of HRC, 77.5% from VLBW infants.

View larger version (14K): [in this window] [in a new window]   Fig 1. Calculation of the HRC index. (A) The course for an infant who developed Serratia marcescens sepsis on day 37 of life. Physical examination and laboratory tests (white blood cells: 8000/µL; I:T ratio: 0.02; platelets: 257000/µL) were normal. This panel shows the time course of the HRC index at 6-hourly intervals. The x axis is days, and 0 is the time of the positive blood culture. The y axis is the fold increase in risk for imminent clinical illness calculated from the HRC index. The insets labeled a, b, and c show 10-minute records of HR. (B–D) The time course of HRC parameters over 12 hours before the diagnosis of sepsis. The data points are individual HRC parameter values for 26 4096-beat epochs, and the labeled points were used in the HRC index expression given at the bottom. (B) SD ("X" is the 10th percentile value). (C) Sample asymmetry ("Y" is the median value). (D) Sample entropy ("Z" is the 10th percentile value). The HRC index is calculated by an expression derived from multivariable logistic regression, which yields values of the ß coefficients. The calculated HRC index for this 12-hour epoch predicted more than twice the expected risk for imminent illness and appears as the emphasized data point in the plot of A.

The abnormalities or interest are low SD (reduced variability), high sample asymmetry (transient decelerations), and low sample entropy (both). The goal is to detect intermittent abnormal records without overreporting of very transient abnormalities that might be attributable to innocuous interventions such as feeding. Our approach is to summarize some of the measures by their extremes and others by the median. Using extreme values allows the HRC index to change quickly (2 or 3 consecutive abnormal records have a perceptible effect). Using median values smooths the response.

The labeled data points in Fig 1B to D are the 10th percentile values of SD (point X) and sample entropy (point Z) and the median sample asymmetry (point Y). The expression below the plots shows how these values are weighted and combined to yield the HRC index. The weights, expressed as the coefficients ß, and their statistical significance are determined by regression analysis.¹² The HRC index for this 12-hour period was >2-fold higher than average, indicating >2-fold increase in risk.

HRC Predicts Sepsis and UTI in Infants in the NICU
We examined the relationship of HRC measurements with the incidence of adverse events over the next 7 days. For each HRC measurement, infants were placed into high-, intermediate-, and low-risk groups. Figure 2A shows the cumulative incidence of adverse events as a function of time. The separation of the confidence intervals indicates significant differences among the risk groups at the level of P < .001. Figure 2B shows the risk for events relative to the average risk. Infants with measurements in the low-risk group had less than the average risk, whereas those in the intermediate- and high-risk groups had above-average rates. The relative risk of an adverse event was nearly 3-fold higher than average in the high-risk group for the first 24 hours and remained >2-fold higher for a week. Compared with the low-risk group, the high-risk group had a 5- to 6-fold increase in risk of an event in the next 24 hours and a 3- to 4-fold increase for 1 week.

View larger version (25K): [in this window] [in a new window]   Fig 2. HRC measurements stratify risk for adverse events in the NICU. In A and B, high, intermediate, and low indicate the level of risk as judged by HRC index. In C and D, the labels refer to high-risk HRC and abnormal laboratory tests, either alone or in combination. Nonoverlapping shaded areas, which are 95% confidence intervals determined by bootstrap, signify statistically significant differences. (A) Cumulative incidence of sepsis, UTI, and death in 1022 infants in the UVA and WFU NICUs on the basis of HRC monitoring results of high, intermediate, and low risk. (B) Relative risk for adverse events in 1022 infants in 2 NICUs stratified into HRC groups. (C) Cumulative incidence of sepsis, UTI, and death in 320 VLBW infants in the UVA NICU on the basis of finding of high HRC measurement, abnormal laboratory result, or both. (D) Relative risk for adverse events in 320 VLBW infants in the UVA NICU on the basis of finding of high HRC measurement, abnormal laboratory result, or both.

Combining HRC and Laboratory Tests to Predict Adverse Events in VLBW Infants
Laboratory tests that are relevant to infection are the current tools for diagnosing neonatal sepsis and UTI, and we investigated how HRC monitoring might add information in the VLBW population. In VLBW infants after 1 week of age at UVA, 82% of 6-hourly epochs had neither high HRC nor an abnormal laboratory result, 15% had either high HRC or an abnormal laboratory result, and 3% had both. Figure 2C shows how HRC and laboratory tests influence the incidence rates of an adverse outcome during the next 7 days. For example, the group with high HRC and an abnormal laboratory test had a 11% incidence of adverse outcome at day 3 compared with 2.0% in the group with normal HRC and no abnormal laboratory test.

Figure 2D shows the relative risk of high HRC with an abnormal laboratory test compared with high HRC without an abnormal test. There was a 6- to 7-fold increase in risk in the first day that remained 3-fold at 1 week.

Overall Evaluation of the Predictive Model
Figure 3 compares the predicted and observed increases in risk of an adverse event using a predictive model that was generated on the entire data set of 1022 infants. Intermediate- and high-risk groups are shown as shaded boxes bounded by 95% confidence intervals. There is good agreement between predicted and observed increases in risk and a steep increase in predicted and observed risk of event in the high-risk group.

View larger version (17K): [in this window] [in a new window]   Fig 3. Predicted and observed rates of neonatal sepsis, UTI, and death. The smooth line is the output of the predictive multivariable logistic regression model. The points are the measured event rates, and the shaded region represents the 95% confidence interval. The lighter gray shaded region represents the intermediate-risk zone, and the darker gray shaded region represents the high-risk zone.

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

We evaluated continuous monitoring of HRC in 2 NICUs using a previously validated algorithm to detect reduced variability and transient decelerations. Our new major findings are as follows: (1) the relative risk of an adverse event of sepsis, UTI, or death is severalfold higher in the high-risk group compared with the low-risk group; and (2) HRC monitoring adds information to abnormal laboratory tests.

HRC monitoring in the NICU differs from other forms of HR monitoring such as fetal monitoring. The HRC index is quantitative and based on occurrence of observed episodes of illness in a large population. It is a continuous risk-assessment measure presented as the fold increase in probability of an event. Many infections have subacute onset, and the HRC changes hours to days in advance of severe symptoms.^11,12 The intervention, antibiotics, is generally assumed to be more effective when started early.

HRC monitoring seems to meet methodologic standards defined by Feinstein et al²² for assessing new tests. There should be a satisfactory spectrum of population (all NICU admissions), analysis of subgroups (VLBW), no work-up or review bias (clinicians blinded to results during model development and validation), confidence intervals (for fold increase in risk), indeterminate results handled correctly (no indeterminate results), and test reproducibility (robust performance of predictive models with varying outcomes and time windows, all generated at 1 NICU and validated at another).

HRC monitoring also meets more recent and specific guidelines proposed for evaluation of new screening tests.^23,24 These include an ability to standardize (automated HR analysis from bedside monitor), independence from established risk factors (adds information to birth weight, gestational age, and days of age),^11,12 association with clinical endpoints (sepsis, UTI, and death), population norms (this and previous work^11,12), ability to improve prediction over traditional risk factors (adds information to laboratory tests, birth weight, gestational age, and days of age), generalization to population groups (HRC index generated at 1 NICU and validated at another),¹² and acceptable cost (uses existing electrocardiogram monitor data: no new contact with patient required).

Figure 2A and C shows the predictive accuracy of HRC monitoring. If a single high-risk HRC measurement is considered to be a "positive" test for impending proven sepsis, UTI, or death in the next week, then the positive predictive value is 15%. This increases to 20% if an abnormal laboratory test is also present. If a single low-risk HRC result is considered to be a "negative" test, then the negative predictive value is 95%. The problems with this perspective are well known. First, information is lost when thresholds are used (what if white blood cell and blood glucose results were limited to "positive" or "negative"?). Second, positive predictive values are low for infrequent events such as neonatal sepsis and death.⁵ Laboratory tests for sepsis are usually performed in symptomatic patients, who thus have a higher pretest probability of illness than the infants here, for whom the testing was continuous. Finally, the intent of HRC monitoring is to identify infants with increasing risk so that physicians can evaluate them, not to make a definitive diagnosis.

There are limitations to HRC monitoring. Illnesses other than sepsis and UTI can lead to elevated cytokine levels and thus, in our view,¹⁰ to abnormal HRC. We have noted cases of elevated HRC index in which there was no suspected or documented sepsis or UTI. Instead, there were other important new diagnoses, including acute respiratory failure, exacerbations of chronic lung disease, intraventricular hemorrhage, and necrotizing enterocolitis. HRC are almost always abnormal immediately after surgery. Undoubtedly, there are still other illnesses that may cause abnormal HRC and are not attributable to infection. There inevitably will be epochs of abnormal HRC that go unexplained. Sometimes these recur and escalate, and clinical illness eventually declares itself, as we have previously shown.¹² Moreover, we find that infants with chronic severe respiratory illness, for example, have persistently elevated HRC index in the absence of acute deteriorations. Although abnormal HRC may not be specific for sepsis and UTI, these remain common causes for abnormal monitoring results and ones for which early diagnosis and therapy should be especially helpful.

We conclude that the HRC index holds useful clinical information about risk stratification in the NICU. Moreover, it adds information to abnormal values of common laboratory tests that are relevant to neonatal sepsis. It has the advantage of being noninvasive and continuous, unlike current laboratory tests, which are intermittent and require a blood sample. Continuous HRC monitoring has potential for alerting medical personnel in advance of overt clinical illness in the NICU. More generally, monitoring of regulated physiologic processes using physiomarkers such as HRC may be important in the early diagnosis of many subacute, potentially catastrophic illnesses.

	ACKNOWLEDGMENTS

 
This study was supported by NIGMS-64640; American Heart Association, Mid-Atlantic Affiliate; Children's Medical Center Research Fund, University of Virginia; Virginia's Center for Innovative Technology; and Medical Predictive Science Corporation, Charlottesville, VA.

	FOOTNOTES

 
Accepted Feb 4, 2005.

Reprint requests to (J.R.M.) Cardiovascular Division, Box 801395, University of Virginia, Charlottesville, VA 22908. E-mail: rm3h{at}virginia.edu

Conflict of interest: Medical Predictive Science Corporation of Charlottesville, VA, has a license to market technology related to heart rate characteristics monitoring of newborn infants, and supplied partial funding for this study. Drs Griffin and Moorman have an equity share in this company.

	REFERENCES

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 

1. Escobar GJ. The neonatal "sepsis work-up": personal reflections on the development of an evidence-based approach toward newborn infections in a managed care organization. Pediatrics. 1999;103 :360 –373
2. Polin RA. The "ins and outs" of neonatal sepsis. J Pediatr. 2003;143 :3 –4[CrossRef][ISI][Medline]
3. Malik A, Hui CP, Pennie RA, Kirpalani H. Beyond the complete blood cell count and C-reactive protein: a systematic review of modern diagnostic tests for neonatal sepsis. Arch Pediatr Adolesc Med. 2003;157 :511 –516[Abstract/Free Full Text]
4. Mehr S, Doyle LW. Cytokines as markers of bacterial sepsis in newborn infants: a review. Pediatr Infect Dis J. 2000;19 :879 –887[ISI][Medline]
5. Gerdes JS. Clinicopathologic approach to the diagnosis of neonatal sepsis. Clin Perinatol. 1991;18 :361 –381[ISI][Medline]
6. Kellogg JA, Ferrentino FL, Goodstein MH, et al. Frequency of low level bacteremia in infants from birth to two months of age. Pediatr Infect Dis J. 1997;16 :381 –385[CrossRef][ISI][Medline]
7. Schelonka RL, Chai MK, Yoder BA, et al. Volume of blood required to detect common neonatal pathogens. J Pediatr. 1996;129 :275 –278[CrossRef][ISI][Medline]
8. Neal PR, Kleiman MB, Reynolds JK, et al. Volume of blood submitted for culture from neonates. J Clin Microbiol. 1986;24 :353 –356[Abstract/Free Full Text]
9. Pierce JR, Merenstein GB, Stocker JT. Immediate postmortem cultures in an intensive care nursery. Pediatr Infect Dis. 1984;3 :510 –513[ISI][Medline]
10. Griffin MP, Moorman JR. Toward the early diagnosis of neonatal sepsis and sepsis-like illness using novel heart rate analysis. Pediatrics. 2001;107 :97 –104[Abstract/Free Full Text]
11. Griffin MP, O'Shea TM, Bissonette EA, et al. Abnormal heart rate characteristics are associated with neonatal mortality. Pediatr Res. 2004;55 :782 –788[CrossRef][ISI][Medline]
12. Griffin MP, O'Shea TM, Bissonette EA, et al. Abnormal heart rate characteristics preceding neonatal sepsis and sepsis-like illness. Pediatr Res. 2003;53 :920 –926[CrossRef][ISI][Medline]
13. Kovatchev BP, Farhy LS, Cao H, et al. Sample asymmetry analysis of heart rate characteristics with application to neonatal sepsis and systemic inflammatory response syndrome. Pediatr Res. 2003;54 :892 –898[CrossRef][ISI][Medline]
14. Richman JS, Moorman JR. Physiological time-series analysis using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol. 2000;278 :H2039 –H2049[Abstract/Free Full Text]
15. Lake DE, Richman JS, Griffin MP, Moorman JR. Sample entropy analysis of neonatal heart rate variability. Am J Physiol. 2002;283 :R789 –R797
16. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control. 1988;16 :128 –140[CrossRef][ISI][Medline]
17. Stoll BJ, Hansen N, Fanaroff AA, et al. Late-onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics. 2002;110 :285 –291[Abstract/Free Full Text]
18. Feng Z, McLerran D, Grizzle J. A comparison of statistical methods for clustered data analysis with Gaussian error. Stat Med 1996;15 :1793 –1806[CrossRef][ISI][Medline]
19. Stoll BJ, Hansen N. Infections in VLBW infants: studies from the NICHD Neonatal Research Network. Semin Perinatol. 2003;27 :293 –301[CrossRef][ISI][Medline]
20. Fanaroff AA, Hack M, Walsh MC. The NICHD Neonatal Research Network: changes in practice and outcomes during the first 15 years. Semin Perinatol. 2003;27 :281 –287[CrossRef][ISI][Medline]
21. Stoll BJ, Gordon T, Korones SB, et al. Late-onset sepsis in very low birth weight neonates: a report from the National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr. 1996;129 :63 –71[CrossRef][ISI][Medline]
22. Reid MC, Lachs MS, Feinstein AR. Use of methodological standards in diagnostic test research. Getting better but still not good. JAMA. 1995;274 :645 –651[Abstract]
23. Pearson TA, Mensah GA, Alexander RW, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation. 2003;107 :499 –511[Free Full Text]
24. Mosca L. C-reactive protein—to screen or not to screen? N Engl J Med. 2002;347 :1615 –1617[Free Full Text]

This article has been cited by other articles:

				V. Bhandari, C. Wang, C. Rinder, and H. Rinder Hematologic Profile of Sepsis in Neonates: Neutrophil CD64 as a Diagnostic Marker Pediatrics, January 1, 2008; 121(1): 129 - 134. [Abstract] [Full Text] [PDF]

				M. El-Khatib Physiomarkers of Neonatal Heart Rate Pediatrics, May 1, 2006; 117(5): 1853 - 1854. [Full Text] [PDF]

This Article Abstract Full Text (PDF) P3Rs: Submit a response Alert me when this article is cited Alert me when P3Rs are posted Alert me if a correction is posted Citation Map Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Griffin, M. P. Articles by Moorman, J. R. Search for Related Content PubMed PubMed Citation Articles by Griffin, M. P. Articles by Moorman, J. R. Related Collections Premature & Newborn

	Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2005 American Academy of Pediatrics. All rights reserved.