Survival rates for regional cohorts of extremely preterm infants have increased through the 1990s;

Outcomes at Age 2 Years of Infants < 28 Weeks Gestational Age Born in Victoria in 2005 Lex W. Doyle, MD, FRACP, Gehan Roberts, PhD, FRACP, Peter J. Anderson, PhD, and the Victorian Infant Collaborative Study Group* Objective To determine the survival rates and neurosensory outcomes of infants born at gestational age 22-27 weeks in the state of Victoria in 2005 and compare theses data with those for similar infants born in the 1990s. Study design This was a population-based study of all extremely preterm (22-27 weeks gestational age) live births in Victoria in 2005 free of lethal anomalies and randomly selected term controls. Survival and quality-adjusted survival rates at age 2 years were determined relative to the controls, and results were compared with regional extremely preterm cohorts born in 1991-92 and 1997. Results Of 270 very preterm live births in 2005, 172 (63.7%) survived to 2 years, not significantly different from the survival rate of 69.6% for those born in 1997. Rates of severe developmental delay and severe disability were lower than in the very preterm survivors born in 1997. Quality-adjusted survival rates in the extremely preterm cohorts rose from 42.1% in 1991-92 to 55.1% in 1997, but did not increase in 2005 (53.4%). Conclusions Survival rates for infants born at 22-27 weeks gestational age have not increased since the late 1990s, but the neurosensory outcome in survivors has improved. (J Pediatr 2010;156:49-53). Survival rates for regional cohorts of extremely preterm infants have increased through the 1990s; 1 however, survival rates cannot increase indefinitely, due to limits on the effect of current medical technology on survival rate. Thus, the emphasis of perinatal intensive care must be on reducing the rates of neurosensory disability rates in survivors, which have remained overly high relative to term controls. Moreover, adverse neurosensory outcomes are more prevalent with diminishing maturity, to the extent that some would not offer intensive care to infants born at < 26 weeks gestation. The rates of neurosensory disability in early childhood have remained stable for very preterm infants born in the 1990s, despite the increased survival of very preterm infants over time. 2 But perinatal practices change constantly, and the increased use of treatments known to improve outcome, such as caffeine therapy, 3 or avoidance of known potentially harmful therapies, such as postnatal corticosteroids, 4 might be expected to improve rates of disability in survivors over time. The objectives of the present study were to determine the survival and neurosensory disability rates of infants born at < 28 weeks gestational age and term controls born in the state of Victoria in 2005 and to compare the data with those from earlier cohorts born in the state in the 1990s, details of which have been published previously. 2 It was hypothesized that survival rates would not have increased significantly since the late 1990s, but neurosensory disability rates would have decreased. Methods All live-born infants at 22-27 completed weeks of gestation born in the state of Victoria in 2005 were included in this study, except for those live births resulting from termination of pregnancy secondary to lethal anomalies. Multiple data sources (the 4 level-iii neonatal intensive care units in the state, the Newborn Emergency Transport Service, and the Victorian Perinatal Data Collection Unit) were cross-checked to verify the number of live births. 5 Gestational age was determined by the best obstetric estimate, based on fetal ultrasound conducted before 20 weeks in most cases. Controls who were both term (> 36 weeks gestation) and normal birth weight (> 2499 g) were randomly selected from each of the 3 level-iii perinatal centers in the state, stratified to balance with preterm survivors for sex, for the mother s health insurance status (as a proxy for social class), and the language spoken primarily in the mother s country of birth (English or other), all variables known to be related to long-term outcome in our cohorts. CI CP GMFCS OR SD Confidence interval Cerebral palsy Gross Motor Function Classification System Odds ratio Standard deviation From the Newborn Services, Royal Women s Hospital (L.W.D.), Melbourne, Australia; Department of Neonatal Medicine, Royal Children s Hospital (G.R.), Melbourne, Australia; Murdoch Children s Research Institute, (L.W.D., G.R., P.J.A.), Melbourne, Australia; and Departments of Obstetrics and Gynaecology (L.W.D.); and Departments of Paediatrics, and Psychology (P.J.A.), University of Melbourne, Melbourne, Australia *A list of additional members of the Victorian Infant Collaborative Study Group is available at www.jpeds.com (Appendix). Supported in part by a grant from Health and Community Services, Victoria, and Project Grant 454413 from the National Health and Medical Research Council, Australia. The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright Ó 2010 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2009.07.013 49

THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 156, No. 1 Survivors were assessed at age 2 years (corrected for prematurity) by pediatricians and psychologists blinded to perinatal details, including prematurity status. Impairments evauated included cerebral palsy (CP), blindness (visual acuity < 20/200 in the better eye), deafness (hearing loss requiring amplification or worse) and developmental delay. Development was assessed using the Bayley Scales of Infant and Toddler Development, 6 and cognitive and language composite scores were obtained relative to the mean and standard deviation (SD) for the normal birth weight controls on the respective scores. Mild developmental delay was characterized by a score on either scale from -2 SD to < -1 SD; moderate developmental delay, by a score on either scale from -3 SD to < -2 SD; and severe developmental delay, by a score on either scale < -3 SD. A child who was unable to complete psychological testing because of presumed severe developmental delay was assigned a score of -4 SD. The criteria for the diagnosis of CP included abnormal muscle tone and delays in motor control and function. 7 Severe disability comprised severe CP (unlikely ever to walk; Gross Motor Function Classification System 8 [GMFCS] level 4 or 5), blindness, or severe developmental delay; moderate disability comprised moderate CP (not walking at age 2 years but expected to walk eventually; GMFCS level 2 or 3), deafness, or moderate developmental delay; and mild disability comprised mild CP (walking at age 2 years; GMFCS level 1), or mild developmental delay. Neurosensory utilities for survivors were assigned as described previously 9-11 according to the severity of the disability imposed by an impairment; 0.4 for severe, 0.6 for moderate, 0.8 for mild, and 1 for no disability. Utilities were multiplied for children with multiple disabilities; thus, the lowest possible utility for a survivor was 0.0384 for a child with severe CP (0.4), severe developmental delay (0.4), blindness (0.4), and deafness (0.6). Infants who died were assigned a utility of 0. The few infants who survived but were not assessed were assigned a utility equal to the mean utility for survivors assessed for their respective cohort. Utilities were summed and divided by the number of live births to calculate the quality-adjusted survival rate. Data were analyzed by SPSS for Windows version 17.0 (SPSS Inc., Chicago, Illinois). Means were contrasted by mean difference and 95% confidence interval (CI) and by linear regression analysis to adjust for confounding variables, calculating adjusted mean differences and 95% CIs. Proportions were compared by c 2 analysis, with relative risks and 95% CIs calculated, and by logistic regression analysis, to adjust for confounding variables, with odds ratios (OR) and 95% CIs calculated from the regression coefficients. Differences between ordered categories were compared using the Mann-Whitney U test or the c 2 test for linear trend. Differences in quality-adjusted survival rates were compared using the Mann-Whitney U test. A P value <.05 was considered statistically significant. Results for the 2005 very preterm cohort were compared with those for the term control group, as well as with those for term and very preterm cohorts born in 1991-1992 and 1997, for which data have been reported previously. 2 Lethal anomalies had been included in the earlier reports but were excluded in this study for comparison of survival rates. The 1991-1992 and 1997 cohorts are the only children selected by gestational age that we assessed at age 2 years, because we lacked the resources necessary to assess all children from all years. The Research and Ethics Committees at the Royal Women s Hospital, Mercy Hospital for Women, and Monash Medical Centre, Melbourne approved the follow-up studies. Written informed consent was obtained from the parents of term controls. Follow-up was considered routine clinical care for the very preterm infants. Results The number of live births at 22-27 weeks gestational age per year was 32% higher in 2005 (n = 288) compared with 1991-1992 (n = 219) (Table I). There were 18 live births from termination of pregnancy for lethal anomalies in 2005, compared with 5 per year in 1991-1992 and 6 in 1997; most terminations were born at 22 weeks gestation. There were more very immature live births free of lethal anomalies per year in 2005 than in the other 2 eras: 23% were born at 22 or 23 weeks gestation in 2005, compared with 19% in 1991-1992 and 16% in 1997. There were 220 term controls for the 2005 cohort. The survival rate to age 2 years was higher at each week of gestation in 1997 compared with 1991-1992, and was significantly higher overall (OR = 2.06; 95% CI = 1.46-2.92; P <.001) (Table I; Figure 1). The overall survival rate was lower in 2005 than in 1997, but the overall difference was not statistically significant (OR = 0.77; 95% CI = 0.52-1.12; P =.17), even after adjustment for gestational age (adjusted OR = 0.75; 95% CI = 0.48-1.17; P =.20). The survival rates between 2005 and 1997 varied with individual weeks of gestation, but the largest discrepancy was at 23 weeks. The follow-up rates were high for both the very preterm and control cohorts born in 2005 (Table II). No term control had CP or blindness, but 1 control was deaf at age 2 years, and only the rate of CP was significantly higher in the very preterm children. The means on the cognitive and language composite scales for the term controls were substantially above the normative mean of 100 for both scales and were significantly higher than for the very preterm group. Significantly more very preterm children than controls had developmental delays and neurosensory disabilities. Rates of CP, blindness, deafness requiring a hearing aid, developmental delay, and neurosensory disability were not significantly different between the preterm cohorts across all eras, although those in 2005 were less likely to have most of these outcomes (Table III). However, the 2005 cohort had significantly lower rates of severe developmental delay and severe neurologic disability than the 1997 cohort, who in turn had significantly higher rates of these problems than the 1991-1992 cohort. Rates of neurosensory disability were similar in each of the term control groups (1991-1992: none, 82%; mild, 15%; moderate, 2%; severe, 2%; 1997: none, 83%; mild, 13%; moderate, 2%; severe, 2%; 2005: none, 79%; mild, 18%; 50 Doyle et al

January 2010 ORIGINAL ARTICLES Table I. Numbers of live births and lethal anomalies, and survival rates to 2 years of age, by gestational age and era Gestational age (completed weeks) 22 23 24 25 26 27 22-27 Outcome 91-92 97 05 91-92 97 05 91-92 97 05 91-92 97 05 91-92 97 05 91-92 97 05 91-92 97 05 Live 37 15 33 52 22 35 63 29 43 88 56 46 98 52 59 100 49 72 438 223 288 birth, n Lethal 3 0 13 3 2 3 0 0 0 2 3 0 1 0 2 1 1 0 10 6 18 anomalies, n Live birth 34 15 20 49 20 32 63 29 43 86 53 46 97 52 57 99 48 72 428 217 270 free of lethal anomalies, n Survival to 2 years, n (% free of lethal anomalies) 0(0) 1(7) 1(5) 5(10) 9(45) 7(22) 21(33) 12(41) 22(51) 51(59) 41(77) 31(67) 71(73) 46(88) 47(82) 77(78) 42(88) 64(89) 225(53) 151(70) 172(64) moderate, 2%; severe, 0%). In all but 2 controls, the neurosensory disability was related to developmental delay. There was 1 child with CP and 1 deaf child in all of the control groups; there were no blind controls. The quality-adjusted survival rate rose with gestational age and was significantly higher in 1997 than in 1991-1992 (P <.001; Mann-Whitney U test) (Figure 1). The difference between 2005 and 1997 was nonsignificant (P =.25; Mann- Whitney U test), although it was narrower than the survival difference for these 2 cohorts. The mean utility per survivor was higher at each week of gestation for the 2005 cohort than for either the 1991-1992 or 1997 cohort, except for the 1991-1992 cohort only at 27 weeks (Figure 2; available at www.jpeds.com). The mean utility per survivor rose significantly with gestational age for both the 1991-1992 cohort (P <.001) and the 1997 cohort (P =.038), but there was no significant change with gestational age in the 2005 cohort (P =.26) (Figure 2). Discussion Survival rates and quality-adjusted survival rates for very preterm infants in Victoria increased through the 1990s, but not in 2005. On the other hand, the rates of severe developmental delay and severe disability were significantly improved in 2005 compared with 1997. The finding of no blind children in the 2005 cohort was probably due to chance, because there have been no substantial changes in management of retinopathy of prematurity over this period. Survival rates for infants of borderline viability from geographically determined cohorts born predominantly in the 1990s have been reviewed recently. 1 Survivors at 22 weeks gestation are unusual; survival rates increase with increasing gestational age, from up to 40% at 23 weeks to 40%-60% at 24 weeks, 60%-80% at 25 weeks, and 70%-80% at 26 weeks. 1 Since that review, Field et al 12 reported survival rates to hospital discharge of infants born in 2000-2005 in the Trent region of England of 9% at 23 weeks, 36% at 24 weeks, and 59% at 25 weeks, at the lower end of the rates reported for births in the 1990s, and lower than in the current study of births in 2005 in Victoria. Of the impairments leading to disability, there is no doubt that CP rates are significantly higher in very preterm survivors, increasing inversely with diminishing gestational age. 13,14 The rate of CP in our 2005 cohort is lower than that in most other regional cohorts, and lower than that in those born in Victoria in the 1990s. One possible explanation for this might be that most survivors would have received caffeine (which is known to reduce CP in very preterm infants 3 ) in 2005 but not in the 1990s. Another possible explanation is that the use of postnatal corticosteroids, also known to cause CP, 4 decreased substantially after the early 2000s compared with the 1990s. The most prevalent impairment was developmental delay. Each cohort was assessed with different versions of the same developmental test (1991-1992 cohort, Bayley Scales of Infant Development; 15 1997 cohort, Bayley Scales of Infant Development, Second Edition; 16 2005 cohort, Bayley Scales of Infant and Toddler Development, Third Edition 6 ). It is unlikely that the varying rate of developmental delay among the cohorts is due to changes in test norms over time, because the results for the very preterm groups were compared with those for contemporaneous, randomly selected term controls, who in turn had a similar rate of developmental delay (approximately 20% overall). Figure 1. Survival and quality-adjusted survival rates compared over 3 eras by gestational age. Outcomes at Age 2 Years of Infants < 28 Weeks Gestational Age Born in Victoria in 2005 51

THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 156, No. 1 Table II. Neurosensory impairments and disabilities in the extremely preterm cohort and term controls from 2005 Extremely preterm Term controls Statistics Survivors, n 172 220 Survivors assessed, n 163 (94.8%) 202 (91.8%) c 2 = 1.3, P =.25 CP 16 (9.8) 0 (0) c 2 = 20.7, P = <.001 Blindness 0 (0) 0 (0) c 2 =0,P = 1.0 Deafness 4 (2.5) 1 (0.5) c 2 = 2.56, P =.11 Cognitive composite scale 97.5 (12.6) 108.9 (14.3) Mean difference, 11.4; 95% CI = 8.6-14.3 Language composite scale 93.9 (14.8) 108.2 (14.8) Mean difference, 14.4; 95% CI = 11.1-17.6 No developmental delay 85 (52.1) 161 (79.7) Mild developmental delay 52 (31.9) 37 (18.3) Moderate developmental delay 20 (12.3) 4 (2.0) Severe developmental delay 6 (3.7) 0 (0) z = 6.0; P <.001* No disability 83 (50.9) 160 (79.2) Mild disability 47 (28.8) 37 (18.3) Moderate disability 27 (16.6) 5 (2.5) Severe disability 6 (3.7) 0 (0) z = 6.2; P <.001* Data are n (% assessed) unless specified otherwise. *Mann Whitney U test for distribution over ordered categories. Our findings once again underscore the need for contemporaneous controls, with the means for the language and cognitive composite scales on the Bayley Scales substantially above the normative mean of 100 for both scales. Had we used only the test means of 100 and SD of 15, we would have seriously underestimated the rates of developmental delay. There are no government-funded developmental support services available in Victoria solely on the basis of extremely low gestational age. Moreover, children do not gain access to services just because their developmental quotient is < -1 SD; generally, they need to have more severe delay (< -2 SD) to qualify. There is considerable variability in the criteria for defining the presence and severity of neurologic disabilities in other geographical studies, including age at which survivors are assessed. 1 Despite these limitations, approximately 25% of very preterm survivors born in the 1990s in regional cohort studies, including those from Victoria, have substantial neurologic disabilities, 1 equivalent to moderate or severe disability as defined in the current study, although the rate was a little lower (20%) in the current cohort born in 2005. The rates of developmental delay in our cohorts may seem low relative to reports from the United States, mainly because US cohorts are almost never geographically determined and come predominantly from high-risk perinatal centers, with an overrepresentation of children from disadvantaged backgrounds, who subsequently score poorly on the Bayley Scales. The strengths of the current study include the avoidance of referral bias by having a geographic cohort, the high followup rate, the contemporaneous term control group, and the blinding of outcome assessors to knowledge of an individual child s group status. A potential weakness is the relatively early age of assessment. These children must be followed into later childhood at a minimum to determine the presence or absence of more important long-term neurologic or other morbidities that can affect functioning at school age. Nonetheless, assessment at age 2 years provides an important snapshot of the child s function that can be used to guide early intervention therapies as needed. Another potential weakness is the relatively small sample size, which Table III. Neurosensory impairments and disabilities in all extremely preterm cohorts 1991-92 1997 2005 Statistics Survivors, n 225 151 172 Survivors assessed, n 219 (97.3%) 149 (98.7%) 163 (94.8%) CP 24 (11.0) 18 (12.1) 16 (9.8) 1997 vs 1991-92: c 2 = 0.11, P =.74, RR = 1.10 (95% CI = 0.62-1.96); 2005 vs 1997: c 2 = 0.41 P =.52, RR = 0.79 (95% CI = 0.39-1.62) Blindness 5 (2.3) 4 (2.7) 0 (0) c 2 trend = 2.61, P =.11 Deafness 2 (0.9) 2 (1.3) 4 (2.5) c 2 trend = 1.45, P =.23 No developmental delay 128* (58.4) 81 (54.4) 85 (52.1) Mild developmental delay 51 (23.3) 32 (22.1) 52 (31.9) Moderate developmental delay 24 (11.0) 14 (9.4) 20 (12.3) Severe developmental delay 16 (7.3) 22 (14.8) 6 (3.7) 1997 vs 1991-92: c 2 = 5.33, P =.021, RR = 2.02 (95% CI = 1.10-3.72); 2005 vs 1997: c 2 = 11.7, P <.001, RR = 0.25 (95% CI = 0.10-0.60) No disability 119 (54.3) 72 (48.3) 83 (50.9) Mild disability 54 (24.7) 35 (23.5) 47 (28.8) Moderate disability 29 (13.2) 19 (12.8) 27 (16.6) Severe disability 17 (7.8) 23 (15.4) 6 (3.7) 1997 vs 1991-92: c 2 = 5.39, P =.02, RR = 1.99 (95% CI = 1.10-3.59); 2005 vs 1997: c 2 = 12.8, P <.001, RR = 0.24 (95% CI = 0.10-0.57) Data are n (% assessed) unless specified otherwise. *Includes 1 child likely with normal development but difficult to assess because he was non-english-speaking, with a normal IQ at age 5 years. 52 Doyle et al

January 2010 ORIGINAL ARTICLES contributed to the lack of statistical significance in the lower rates of some of the adverse neurologic outcomes in the 2005 cohort compared with the 1990s cohorts. In conclusion, survival rates for infants born at 22-27 weeks gestational age in Victoria peaked in the late 1990s, but the neurosensory outcomes in survivors appear to have improved. We anticipate that rates of neurologic impairments and disabilities from other regional cohorts of very preterm infants born in the 2000s will be lower than for those born in the 1990s. n Submitted for publication Mar 20, 2009; last revision received May 26, 2009; accepted Jul 6, 2009. Reprint requests: Professor Lex W. Doyle, Department of Obstetrics & Gynaecology, The Royal Women s Hospital, 20 Flemington Road, Parkville, Victoria, Australia 3052. E-mail: lwd@unimelb.edu.au. References 1. Saigal S, Doyle LW. An overview of mortality and sequelae of preterm birth from infancy to adulthood. Lancet 2008;371:261-9. 2. Doyle LW and the Victorian Infant Collaborative Study Group. Neonatal intensive care at borderline viability: is it worth it? Early Hum Dev 2004;80:103-13. 3. Schmidt B, Roberts RS, Davis P, Doyle LW, Barrington KJ, Ohlsson A, et al. Long-term effects of caffeine therapy for apnea of prematurity. N Engl J Med 2007;357:1893-902. 4. Doyle LW, Halliday HL, Ehrenkranz RA, Davis PG, Sinclair JC. Impact of postnatal systemic corticosteroids on mortality and cerebral palsy in preterm infants: effect modification by risk for chronic lung disease. Pediatrics 2005;115:655-61. 5. Doyle LW, Kitchen WH, Lumley J, McDougall P, Drew J, Yu VY, et al. Accuracy of mortality rates for livebirths 500-999 g birthweight. (letter) Med J Aust 1992;156:72. 6. Bayley N. Bayley Scales of Infant and Toddler Development. Third Edition. San Antonio: Harcourt Assessment; 2006. 7. Kitchen WH, Doyle LW, Ford GW, Murton LJ, Keith CG, Rickards AL, et al. Changing two-year outcome of infants weighing 500 to 999 grams at birth: a hospital study. J Pediatr 1991;118:938-43. 8. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol 1997;39:214-23. 9. Kitchen WH, Bowman E, Callanan C, et al. The cost of improving the outcome for infants of birthweight 500-999 g in Victoria. The Victorian Infant Collaborative Study Group. J Paediatr Child Health 1993;29:56-62. 10. The Victorian Infant Collaborative Study Group. Economic outcome for intensive care of infants of birthweight 500-999 g born in Victoria in the post-surfactant era. J Paediatr Child Health 1997;33:202-8. 11. Doyle LW and the Victorian Infant Collaborative Study Group. Evaluation of neonatal intensive care for extremely low birth weight infants in Victoria over two decades, II: efficiency. Pediatrics 2004; 113:510-4. 12. Field D, Dorling JS, Manktelow B, Draper ES. Survival of extremely premature babies in a geographically defined population: prospective cohort study of 1994-9 compared with 2000-5. BMJ 2008; 336:1221-3. 13. Hagberg B, Hagberg G, Beckung E, Uvebrant P. Changing panorama of cerebral palsy in Sweden, VIII: prevalence and origin in the birth year period 1991-94. Acta Paediatr 2001;90:271-7. 14. Moster D, Lie RT, Markestad T. Long-term medical and social consequences of preterm birth. N Engl J Med 2008;359:262-73. 15. Bayley N. Bayley Scales of Infant Development. New York: The Psychological Corporation; 1969. 16. Bayley N. Bayley Scales of Infant Development. Second Edition. San Antonio, TX: The Psychological Corporation; 1993. Outcomes at Age 2 Years of Infants < 28 Weeks Gestational Age Born in Victoria in 2005 53

THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 156, No. 1 Appendix Additional members of the Victorian Collaborative Study Group include (in alphabetical order): Catherine Callanan, RN (Royal Women s Hospital), Elizabeth Carse, FRACP (Monash Medical Centre), Margaret P. Charlton, MEd Psych (Monash Medical Centre), Mary-Ann Davey, PhD (Victorian Perinatal Data Collection Unit), Noni Davis, FRACP (Royal Women s Hospital), Julieanne Duff, FRACP (Royal Women s Hospital), Rod Hunt, PhD, FRACP (Royal Children s Hospital), Cinzia de Luca, PhD (Royal Women s Hospital), Marie Hayes, RN (Monash Medical Centre), Esther Hutchinson, BSc(Hons) (Royal Women s Hospital), Elaine Kelly, MA (Royal Women s Hospital and Mercy Hospital for Women), Marion McDonald, RN (Royal Women s Hospital), Gillian Opie, FRACP (Mercy Hospital for Women), Michael Stewart, FRACP (Royal Women s Hospital and Royal Children s Hospital), Linh Ung, BSc(Hons) (Royal Women s Hospital), Andrew Watkins, FRACP (Mercy Hospital for Women), Amanda Williamson, BA (Mercy Hospital for Women), and Heather Woods, RN (Mercy Hospital for Women). Figure 2. Mean utility per survivor compared over 3 eras by gestational age. 53.e1 Doyle et al