525 Variations in the incidence of intraven- tricular hemorrhage in very low birth weight infants have been noted in the literature. 1-4 Lee et al 1 previously re- ported large variations in the crude in- cidence of IVH among Canadian neonatal intensive care units. In their analyses of the outcome of infants with a birth weight ≤1500 g from 36 centers in 1990, the Vermont-Oxford Trials Network reported an overall crude IVH incidence of 26% but with 25th and 75th percentiles of 18% and 38%, respectively. 2 The National Institute of Child Health and Human Develop- ment Neonatal Network found equally large variations in IVH incidence among the 7 participating sites. 3 Oth- ers 4-5 demonstrated a change in inci- dence of IVH over time. However, most previous reports of variation in IVH rates among NICUs did not ad- just for severity of illness or examine reasons for the variation. The aims of this study were to examine the risk- and illness severity–adjusted variation of IVH incidence among NICUs. We used data from the Canadian NICU Network, 1 which collected demo- graphic, severity of illness, outcome, and treatment data on all admissions during a 22-month period in 1996- 1997, from 17 NICUs across Canada. We hypothesized that there were vari- ations in the incidence of IVH among Canadian NICUs, that these variations Variations in intraventricular hemorrhage incidence rates among Canadian neonatal intensive care units Anne R. Synnes, MDCM, FRCPC, MHSc, Li-Yin Chien, MPH, ScD , Abraham Peliowski, MD, FRCPC, Ranjit Baboolal, MBBCh, FRCPC, Shoo K. Lee, MBBS, FRCPC, PhD, and the Canadian NICU Network From the Department of Pediatrics, University of British Columbia, Vancouver, British Columbia, Canada; Centre for Community Health and Health Evaluation Research, Vancouver, British Columbia, Canada; Department of Pediatrics, University of Alberta, Edmonton, Alberta, Canada; Department of Pediatrics, University of Toronto, Toronto, Ontario, Canada. A complete list of the members of the Canadian NICU Network appears at the end of this article Supported by Grant 40503 and Grant 00152 from the Medical Research Council of Canada. Additional funding was provided by the B.C.’s Children’s Hospital Foundation; Calgary Region- al Health Authority; Dalhousie University Neonatal-Perinatal Research Fund; Division of Neonatology, Children’s Hospital of Eastern Ontario; Child Health Program, Health Care Cor- poration of St John’s; The Neonatology Program, Hospital for Sick Children; Lawson Research Institute; Midland Walwyn Capital Inc; Division of Neonatology, Hamilton Health Sciences Corporation; Mount Sinai Hospital; North York General Hospital Foundation; Saint Joseph’s Health Centre; University of Saskatchewan Neonatal Research Fund; and University of West- ern Ontario; Women’s College Hospital. Presented in part at the Annual Meeting of the Pediatric Academic societies, Boston, Mass, May 16, 2000. Submitted for publication June 5, 2000; revision received Sept 7, 2000; accepted Sept 20, 2000. Reprint requests: Shoo K. Lee, MBBS, FRCPC, PhD, Centre for Community Health and Health Evaluation Research, 4480 Oak St, Room E-414, Vancouver, BC, V6H 3V4 Canada. Copyright © 2001 by Mosby, Inc. 0022-3476/2001/$35.00 + 0 9/21/111822 doi:10.1067/mpd.2001.111822 GA Gestational age IVH Intraventricular hemorrhage NICU Neonatal intensive care unit OR Odds ratio SNAP-II Score for Neonatal Acute Physiology, Version II Objectives: To examine the variation in intraventricular hemorrhage (IVH) incidence among neonatal intensive care units and identify potential- ly modifiable risk factors. Study design: Multiple logistic regression analysis was used to examine variations in ≥grade 3 IVH, adjusting for baseline population risk factors, admission illness severity, and therapeutic risk factors. Subjects were born at <33 weeks’ gestational age, admitted within 4 days of life to 1 of 17 par- ticipating Canadian NICU network sites in 1996-97, and had neuroimaging in the first 2 weeks of life. Results: Of 5126 subjects <33 weeks’ gestational age, 3806 had neuroimag- ing reports. Five of 17 sites had significantly (P < .05) different crude inci- dence rates of grade 3-4 IVH (odds ratios [OR] 0.2, 3.2, 2.6, 2.1, 1.9) than the hospital with median incidence. With adjustment for baseline population risk factors, perinatal risks, and admission illness severity, IVH incidence rates remained significantly (P < .05) higher at 3 sites (OR 2.9, 2.3 and 2.1). Inclusion of therapy-related variables (treatment of acidosis and vasopressor use on the day of admission) in the model eliminated all site differences. Conclusions: IVH incidence rates vary significantly. Patient characteris- tics explain some of the variance. Early treatment of hypotension and aci- dosis and mode of delivery are potentially modifiable factors and warrant further study in IVH prevention. (J Pediatr 2001;138:525-31) SYNNES ET AL THE JOURNAL OF PEDIATRICS APRIL 2001 persisted even after adjusting for known risk factors and severity of ill- ness, and that we could identify risk factors that are potentially modifiable to reduce the incidence of IVH. METHODS Population The study population comprised all 19,507 infants admitted to 17 NICUs in the Canadian NICU Network from January 8, 1996, to October 31, 1997. The 17 hospitals, except one, are re- gional tertiary level referral centers and include 75% of level 3 NICU beds in Canada. The NICUs ranged in size from 9 to 70 beds and had an average of 133 to 1129 admissions annually. The data were collected as part of a larger study of practices and outcomes of NICUs 1 across Canada, which had a population of nearly 30 million peo- ple and over 357,000 births in 1996. 6,7 There was no interference in NICU practices and management. Only in- fants <33 weeks’ gestational age who were <4 days old at the time of admis- sion and who had a cranial sonogram, computerized tomography, or magnet- ic resonance imaging (subsequently re- ferred to collectively as neuroimaging) were included in the analysis. Data Collection Data were collected prospectively by trained research assistants and entered directly from patient charts into laptop computers by using a customized data entry program with built-in error checking and a standard manual of protocols and definitions. Data were electronically transmitted to the Cen- tre for Community Health and Health Evaluation Research at the British Co- lumbia Research Institute for Chil- dren’s and Women’s Health. Data management was conducted by the Centre for Community Health and Health Evaluation Research in concert with a steering committee comprising experienced researchers and neonatol- ogists representing each of the 5 geo- graphic regions (British Columbia, prairie provinces, Ontario, Quebec, and the Atlantic provinces) and site in- vestigators representing each of the participating hospitals. Patient infor- mation was collected until death or dis- charge from the NICU. Patients trans- ferred to another hospital were tracked until death or discharge. Definition of IVH Abstractors recorded the neuroimag- ing reports performed within the first 2 weeks after admission using Papile’s classification system. 8 When there was more than one report or if there was bi- lateral IVH, the highest grade was used. Intraparenchymal hemorrhages with or without IVH were classified as grade 4 IVH. All other intracranial bleeds such as subarachnoid, subdural, and tentorial bleeds were excluded. Possible or questionable diagnoses were excluded. Variables Analyzed Variables chosen because they were known risk factors for IVH were grouped into patient characteristics, ob- stetric variables, severity of illness, and therapy variables. Patient characteris- tics included (1) sex; (2) GA based on sonographic and obstetrical data except when this was unavailable, in which case the neonatologist’s best estimate was used; and (3) Apgar score (5- minute score categorized as low if <4 or medium if 4-6 or reference if >6). Ob- stetric variables included (1) use of an- tenatal steroids, (2) vaginal versus ce- sarean section delivery, and (3) inborn versus outborn status. The Score for Neonatal Acute Physiology, Version II, 9 recorded within the first 12 hours of admission, was included as a measure of the severity of illness. Admission day therapy-related variables included (1) treatment of acidosis (sodium bicar- bonate or tromethamine), (2) vasopres- sor (any vasoactive infusion such as dopamine, dobutamine, isoproterenol, nitroprusside, or epinephrine) use not associated with a resuscitation, and (3) surfactant administration. Analyses Analyses were performed by using data for each infant rather than each admission. For this purpose, data from re-admissions and inter-hospital trans- fers for each patient were combined. Risk factors were analyzed by using SPSS (version 7.5) software. 10 Crude incidence rates were calculated for all sites. When the site with the median incidence rate was used as a reference, statistically significant variation was 526 IVH grade None I II III IV No. 2685 545 259 152 165 Median birth weight (g) 1244 1162 995 900 830 Median GA (wk) 29 29 27 26 26 Median SNAP-II 9 14 16 21 29 Median Apgar score at 5 min 88877 Inborn (%) 85.9 84.0 71.8 59.9 70.9 Cesarean section (%) 54.7 38.2 30.9 37.5 38.8 Male (%) 54.2 58.4 62.2 60.5 65.9 Antenatal steroids (%) 71.6 71.8 65.0 61.7 53.9 Vasopressors—day 1 (%) 17.1 18.0 33.2 48.0 60.0 Acidosis treatment–day 1 (%) 6.7 6.8 17.2 25.8 36.0 IVH, Intraventricular hemorrhage; GA, gestational age; SNAP-II, Score for Neonatal Acute Physiology–Version II. Table. Patient characteristics THE JOURNAL OF PEDIATRICS SYNNES ET AL VOLUME 138, NUMBER 4 identified. A P value of <.05 was deemed to be significant. By means of multiple logistic regression, the risk (odds ratio) for IVH at each site was determined, adjusting stepwise for pa- tient characteristics, obstetric vari- ables, and admission illness severity as defined previously. Admission day therapy-related risks were then added to assess their contribution to site vari- ations in incidence of IVH. RESULTS Of the 19,507 infants admitted to the Canadian NICU Network during the study period, 5126 were <33 complet- ed weeks’ GA at birth and <4 days old at admission. Of these infants, 3806 had neuroimaging performed in the first 2 weeks of life. Neuroimaging re- ports were available for 89% of all in- fants <31 weeks’ GA and 58% of all in- fants between 31 and 32 weeks’ GA. Neuroimaging practices differed sig- nificantly between NICUs (range, 63%-100% among infants <33 weeks’ GA). Infants with neuroimaging (in- cluded in study) compared with those infants without neuroimaging (exclud- ed from study) had lower mean birth weight (1190 vs 1730 g), lower mean GA (29 vs 31 weeks), and higher SNAP-II (14 vs 5). The overall inci- dence of IVH (any grade) was 29.4%. Five hundred forty-five (14.3%) in- fants had a grade 1 IVH, 259 (6.8%) had a grade 2 IVH, 152 (4.0%) had a grade 3 IVH, and 165 (4.3%) had a grade 4 IVH. Characteristics of infants with different grades of IVH are shown in the Table. The variation in the incidence rates of diagnosed IVH among the 17 NICUs is shown in Fig 1. The incidence rate ranged from 14.2% to 57.7% for all grades of IVH. The ranges of grade- specific incidence rates of IVH were 6.3% to 29.8% for grade 1, 0% to 25% for grade 2, 0% to 14.1% for grade 3, and 0% to 8% for grade 4 IVH. Inci- dence rates decreased with increasing GA. The incidence of any grade IVH (range in parentheses) by GA cate- gories was 48% (range, 27%–89%), 33% (range, 0%-60%), 23% (range, 8%-50%), and 17% (range, 0% to 30%) for infants <27 weeks’ GA, 27-28 weeks’ GA, 29-30 weeks’ GA, and 31- 32 weeks’ GA, respectively. Fig 2 shows the crude ORs and 95% CIs for each site for severe (grade 3 or 4) IVH in comparison with site H with the median incidence rate (6.2%). Five sites had significantly (P < .05) differ- ent incidence rates; 4 of these sites (C, F, I, and N) had higher rates, and one site (M) had a lower rate. After adjust- ment for patient characteristics (sex, GA, and Apgar score), obstetric risks (antenatal steroids, mode of delivery, inborn vs outborn), and admission ill- ness severity, only 3 sites (C, F, and N) had significantly higher incidence rates than the site with median incidence (H). The adjusted ORs for IVH by site, derived from this regression 527 Fig 1. IVH rate by site. SYNNES ET AL THE JOURNAL OF PEDIATRICS APRIL 2001 model, are shown in Fig 3. When therapy-related variables were added to the model, all the site variations disappeared. Significant (P < .05) IVH risk factors that were included in the final regres- sion model of site variations were: male sex (OR 1.4), GA <27 weeks’ (OR 4.2), GA 27-28 weeks’ (OR 2.1), 5-minute Apgar score <4 (OR 2.1), 5-minute Apgar score 4-6 (OR 1.5), SNAP-II ≥30 (OR 1.6), outborn status 528 Fig 2. IVH crude odds ratios. Fig 3. IVH adjusted odds ratios. THE JOURNAL OF PEDIATRICS SYNNES ET AL VOLUME 138, NUMBER 4 (OR 1.9), vaginal delivery (OR 1.5), partial antenatal steroid treatment (OR 0.6), treatment for acidosis (OR 2.1), and treatment with vasopressors (OR 1.7). Complete antenatal steroid treatment (OR 0.8) was not significant (P = .16), but there was significant collinearity between outborn status and antenatal steroid use. The latter variable became highly significant if outborn status was removed from the analysis. Surfactant use was not signif- icantly associated with IVH and was excluded from the final model. DISCUSSION This study demonstrates a statistical- ly and clinically significant variation in the incidence of diagnosed IVH in pre- mature infants (<33 weeks’ GA) admit- ted to NICUs. This variation existed across all grades of IVH and all GA categories. Previous reports of varia- tions in the incidence of diagnosed IVH among preterm infants admitted to NICUs did not adjust for variations in patient population. In our study, we found that risk adjustment for baseline population risk factors and SNAP-II resulted in a 40% reduction in the number of NICUs with incidence of IVH significantly (P < .05) different from the median incidence. Omission of adjustment for these factors could therefore result in inaccurate compar- isons for a significant number of NICUs, with important implications if the results were used for hospital audit and accreditation purposes. We identified prematurity, male sex, 5-minute Apgar score, severity of ill- ness, outborn status, vaginal delivery, lack of antenatal steroids, and treatment of acidosis or use of vasopressors on the day of admission as significant risk fac- tors for severe (grade 3 or 4) IVH. Some of these factors are impossible (eg, male sex) or difficult (eg, prematu- rity, severity of illness, outborn status) to modify. Exceptions are choice of ce- sarean section versus vaginal delivery, use of antenatal steroids, and vasopres- sor and sodium bicarbonate use. Ment et al 11 previously reported an association between lower IVH rates and cesarean section delivery and an- tenatal steroid use. Other studies re- ported conflicting results as to the ben- efit of cesarean delivery for very low birth weight infants, 12-14 but they did not control for selection bias and ill- ness severity. Large variations in antenatal steroid use persist to this day. Lee et al 1 re- ported that the incidence of antenatal steroid use in infants <35 weeks’ gesta- tion in the study cohort varied from 23% to 76% among Canadian NICUs. It is possible that increased use of an- tenatal steroids among Canadian hos- pitals may further decrease the inci- dence of IVH. However, antenatal steroid use may also be a marker for other risks (eg, outborn delivery). Fur- ther research is needed to determine whether cesarean section or antenatal steroid use can reduce the incidence of IVH among preterm infants. Therapy variables (vasopressor use, treatment for acidosis) were critically important in explaining the site varia- tion, and their inclusion in the final re- gression model eliminated all signifi- cant site variation. Although our results do not permit inference of causal relationships, the association between IVH and admission day use of vasopressors and treatment for acido- sis merits further study. The relation- ship between vasopressor use, acidosis treatment, and IVH is consistent with present concepts of the pathogenesis of IVH in the preterm infant. 15-17 The germinal matrix of the preterm infant, with its friable capillary network and poor supportive stroma, is especially susceptible to hemorrhage. Alterations in arterial blood pressure and cerebral blood flow appear to be important trig- gers. 18 In the animal model, IVH is precipitated either by inducing hy- potension and then rapidly increasing the blood pressure by volume infu- sion 19 or by vasopressors. 20 The rela- tionship between pathogenesis of IVH and sodium bicarbonate administra- tion is not well studied but is likely re- lated to its hyperosmolarity. Although further study is needed, our results suggest that variations in the incidence of IVH may be reduced by improved strategies for prevention and treatment of hypotension and acidosis in preterm infants. It is also possible that vaso- pressor use and acidosis treatment are proxies for other unidentified risk fac- tors (eg, perinatal infection). The incidence of IVH may also be af- fected by other risk factors that were not examined in our study. Thrombo- cytopenia and coagulopathy are recog- nized risk factors but probably do not play a major role 21 and were not exam- ined in our study because they were in- frequent in number. Indomethacin has been identified as a potential treatment for prevention of IVH. 22 Because many of our patients were participants in a randomized control trial of pro- phylactic indomethacin treatment, we were unable to consider indomethacin prophylaxis in our analysis. System variables may also have af- fected IVH rates. Therefore we as- sessed the effect of the size of the NICU on IVH risk. The number of ad- missions at each site was not a signifi- cant predictor of severe IVH (P = .42), nor did it alter the size of the therapy variables effect. The number of admis- sions of preterm infants <33 weeks’ gestation was a significant predictor of severe IVH (P = .01), independent of the use of vasopressors or treatment for acidosis. These exploratory find- ings suggest that expertise in caring for preterm infants may confer benefits in preventing IVH. Limitations We noted a variation between NICUs in the percentage of patients who had cranial sonograms or other neuroimaging. Sites with protocols or a general consensus generally used a birth weight of <1500 g and/or a GA of <32 weeks or <33 weeks as a cutoff for 529 SYNNES ET AL THE JOURNAL OF PEDIATRICS APRIL 2001 screening. Because neuroimaging was not performed in all patients, selection bias may have affected our results. Inter-rater variability in cranial sono- graphic diagnosis is well described, 23 and a diagnostic bias may have been present because we were unable to have radiologists who were blinded to risk factors review all subjects’ neu- roimaging. A residual significant site variation should have been, but was not, found if this had been the case. Only selected admission day therapy- related risk factors were included in the analysis. It is possible that other risks and therapy-related risk factors beyond the first day of admission could also be significant predictors of IVH. However, the elimination of all site variations by addition of 2 admission day therapy-related variables suggests that relevant practice-related risks occur early in life. Risk factors identi- fied in our analysis do not infer a causal relationship, and further study is needed to determine whether inter- ventions can be designed to decrease the risk for IVH. The incidence of IVH may be re- duced by improved prevention and/or treatment for hypotension and acidosis in preterm infants, and this should be a priority for research. The role of ante- natal steroids and cesarean section to reduce the incidence of IVH in preterm deliveries also warrants fur- ther study. Members of the Canadian NICU Net- work: Shoo K. Lee, MBBS, FRCPC, PhD (Coordinator, Canadian NICU Network; Centre for Community Health and Health Evaluation Research, Vancouver, BC); Wayne Andrews, MD, FRCPC (Charles A. Janeway Child Health Centre, St John’s, NF); Ranjit Baboolal, MBChB, FRCPC (North York Hospital, North York, ON); Jill Boulton, MD, FRCPC (St Joseph’s Health Centre, London, ON; previously Mt Sinai Hospital, Toronto, ON); David Brabyn, MBChB, FRACP, FRCPC (Royal Columbian Hospital, New Westminster, BC); David S. C. Lee, MBBS, FRCPC (St Joseph’s Health Centre; London, ON); Derek Matthew, MRCS, FRCPC, SM (Victoria General Hospital, Victoria, BC); Douglas D. McMillan, MD, FRCPC (Foothill’s Hospital, Calgary, AB); Christine Newman, MD, FRCPC (Hospital for Sick Children; Toronto, ON); Arne Ohlsson, MD, FRCPC, MSc (Mt Sinai Hospital, Toronto, ON; formerly Women’s College Hospital, Toronto, ON); Abraham Peliowski, MD, FRCPC (Royal Alexandra Hospital, Edmon- ton, AB); Margaret Pendray, MBBS, FRCPC (Children’s and Women’s Health Centre of British Columbia, Vancouver, BC); Koravan- gattu Sankaran, MBBS, FRCPC (Royal University Hospital, Saskatoon, SK); Bar- bara Schmidt, MD, FRCPC, MSc (Hamilton Health Sciences Corporation, Hamilton, ON); Mary Seshia, MBChB, FRCPC (Health Sci- ences Centre, Winnipeg, MB); Anne Synnes, MDCM, FRCPC, MHSc (Children’s and Women’s Health Centre of British Columbia, Vancouver, BC; formerly Montreal Children’s Hospital, Montreal, PQ); Paul Thiessen, MD, FRCPC (Children’s and Women’s Health Centre of British Columbia, Vancouver, BC); Robin Walker, MD, FRCPC (Children’s Hos- pital of Eastern Ontario and Ottawa General Hospital, Ottawa, ON); Robin Whyte, MBBS, FRCPC (IWK-Grace Health Centre for Women, Children and Families, Halifax, NS). Coordinating Centre: Centre for Com- munity Health and Health Evaluation Re- search (Vancouver, BC): Li-Yin Chien, MPH, ScD; Joanna Sale, MSc; Herbert Chan, MSc; Shawn Stewart, BA. REFERENCES 1. Lee SK, McMillan D, Ohlsson A, Pen- dray M, Synnes A, Whyte R, et al. Variations in practice and outcomes of the Canadian NICU Network: 1996-7. Pediatrics 2000;106:1070-9. 2. Vermont-Oxford Trials network. The Vermont-Oxford Trials network: very low birth weight outcomes for 1990. Pediatrics 1993;91:540-5. 3. Hack M, Horbar JD, Malloy MH, Tyson JE, Wright E, Wright L. Very low birth weight outcomes of the Na- tional Institute of Child Health and Human Development Neonatal Net- work. Pediatrics 1991;87:587-97. 4. Cooke RWI. Trends in preterm sur- vival and incidence of cerebral haem- orrhage 1980-9. Arch Dis Child 1991; 66:403-7. 5. Philip AGS, Allan WC, Titi AM, Wheeler LR. Intraventricular hemor- rhage in preterm infants: declining in- cidence in the 1980’s. Pediatrics 1989; 84:797-80. 6. Statistics Canada. http://www.stat- can.ca/english/Pgdb/People/popula- tion/demo02.htm 7. Statistics Canada. http://www.stat- can.ca/english/Pgdb/People/popula- tion/demo04a.htm 8. Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of subependymal and intraventricular he- morrhage: a study of infants with birth weights less than 1500 grams. J Pedi- atr 1978;92:529-34. 9. Richardson DK, Corcoran JD, Esco- bar GJ, Lee SK, for the Canadian NICU Network, the Kaiser Perma- nente Neonatal Minimum Data Set Wide Area Network, and the SNAP-II Study Group. SNAP-II and SNAPPE-II: simplified newborn ill- ness severity and mortality risk scores. J Pediatr 2001;138:92-100. 10. SPSS Advanced Statistics 7.5. Chica- go: SPSS Inc; 1997. 11. Ment LR, Oh W, Ehrenkranz RA, Philip AGS, Duncan CC, Makuch RW. Antenatal steroids, delivery mode and intraventricular hemorrhage in preterm infants. Am J Obstet Gynecol 1995:172:795-800. 12. Yu VYH, Bajuk B, Cutting D, Orgill AA, Astbury J. Effect of mode of de- livery on outcome of very-low-birth- weight infants. Br J Obstet Gynaecol 1984;91:633-9. 13. Lee KS, Khoshnood B, Sriram S, Hsieh HL, Singh J, Mittendorf R. Re- lationship of caesarean delivery to lower birth weight-specific neonatal mortality in singleton breech infants in the United States. Obstet Gynecol 1998;92:769-74. 14. Olshan AF, Shy KK, Luthy DA, Hickok D, Weiss N, Daling JR. Cae- sarean birth and neonatal mortality in very low birth weight infants. Obstet Gynecol 1984;64:267-70. 15. Volpe JJ. Intraventricular hemor- rhage in the premature infant—cur- rent concepts. Part I. Ann Neurol 1989;25:3-11. 16. Volpe JJ. Neurology of the newborn. Philadelphia: WB Saunders; 1995. p. 390. 17. Papile LA. Intracranial hemorrhage. In: Fanaroff AA, Martin RG, editors. Neonatal-perinatal medicine. St Louis: Mosby–Year Book, Inc; 1997. p. 891-5. 18. Bada HS, Korones SB, Perry EH, Ar- heart KL, Ray JD, Pourcyrous M, et al. Mean arterial blood pressure changes in premature infants and those at risk for intraventricular hemor- rhage. J Pediatr 1990;117:607-14. 19. Goddard-Finegold J, Armstrong D, Zeller RS. Intraventricular hemor- rhage following volume expansion after 530 THE JOURNAL OF PEDIATRICS SYNNES ET AL VOLUME 138, NUMBER 4 hypovolemic hypotension in the new- born beagle. J Pediatr 1982;100:796-9. 20. Goddard J, Lewis RM, Armstrong DL, Zeller RS. Moderate, rapidly induced hypertension as a cause of intraventric- ular hemorrhage in the newborn beagle model. J Pediatr 1980;96:1057-60. 21. Lupton BA, Hill A, Whitfield MF, Carter CJ, Wadsworth LD, Roland EH. Reduced platelet count as a risk factor for intraventricular hemorrhage. Am J Dis Child 1988;142:1222-4. 22. Ment LR, Oh W, Ehrenkranz RA, Philip AG, Vohr B, Allan W, et al. Low dose indomethacin and prevention of intraventricular hemorrhage: a multi- center randomized trial. Pediatrics 1994;93:543-50. 23. Taylor GA, Seibert JJ, DiPietro MA, Wright LL, Verter J, Melee L, et al. Effect of local versus central reading and image quality on sonographic di- agnosis of intracranial hemorrhage [abstract]. Pediatr Res 1996;39:249A. 531 50 Years Ago in The Journal of Pediatrics ACUTE GLOMERULONEPHRITIS:I MPETIGO AS AN ETIOLOGICAL FACTOR McCullough GC, Coffee JY, Trice PA Jr, Stone JJ, Crandall HL. J Pediatr 1951:38:346-8 Acute glomerulonephritis, a relatively common pediatric disorder, was recognized 50 years ago to follow an antecedent, usually streptococcal, infection. At that time, the vast majority of cases of post-streptococcal glomerulonephritis followed pharyngeal infections, and the etiologic mechanism was already presumed to be the product of an antigen-antibody reaction invoked by a “nephrotoxic substance.” In the mid-1940s, a revo- lution was occurring in the practice of pediatrics: antibiotics became available to treat streptococcal infections. McCullough and associates had the opportunity to care for 124 children with acute glomerulonephritis in a community hospital in Fairfield, Alabama, during the new antibiotic era. These authors carefully reviewed their 5-year experience, from 1944 to 1949, to reappraise the nature of the preceding infections. In this group of children, the most common type of infection to precede acute glomerulonephritis was im- petigo (34% of the patients), with acute tonsillitis being a close second (31%). The average length of hospital- ization was 19.5 days. There were 2 deaths. All but 4 of the children were admitted to the hospital because of edema, and 4 presented with convulsions. The authors were surprised that bloody urine was not the original symptom in most of their patients. This publication was one of the first to emphasize the importance of streptococcal skin infections as an- tecedents for acute glomerulonephritis. The authors speculated that their recent access to penicillin for strep- tococcal infections provided preventative therapy for glomerulonephritis, as it did for acute rheumatic fever. The authors described skin infections as being considered “innocuous” and not customarily triggering aggres- sive antibiotic therapy. They urged rapid and more vigorous treatment of impetigo as a means of preventing acute glomerulonephritis. In 2001, acute post-streptococcal glomerulonephritis is seen much less frequently in the United States but remains a significant global problem. Because of the decreased experience with post-infectious glomeru- lonephritis, the diagnosis is now occasionally missed or delayed. Impetigo continues to be an important cause of acute glomerulonephritis, particularly in warm climates. Today, the immunologic mechanisms leading to acute glomerulonephritis are better understood; however, it is still unclear whether early antibiotic therapy for impetigo can prevent glomerulonephritis. Of greater concern is that with the threat of drug-resistant streptococci looming, acute post-infectious glomerulonephritis may become familiar to pediatricians once again, unless effective vaccines are developed. F. Bruder Stapleton, MD Ford/Morgan Professor and Chair Department of Pediatrics University of Washington School of Medicine Seattle, WA 98105 9/37/114024 doi:10.1067/mpd.2001.114024