Bronchopulmonary Dysplasia with apologies to Leo Tolstoy Stephen Messier MD Staff Neonatologist

Disclosures I have no financial disclosures Some slides from this talk have been borrowed from Dr. (Lt. Col.) William Lefkowitz

Overview What has happened with the diagnosis of BPD Current thoughts on therapies Difference between histopathologic disease and clinical markers What implications that has for the development of new therapies

Is BPD Important? National Trends are followed Individual reports to hospitals are trended Prognosticate to parents It’s a 3 – letter diagnosis (PDA, NEC, IVH…) Bread and Butter of Neonatal Practice

Pubmed Results Since Sept 2006 2970 articles 568 review articles 520 on BPD in animals (research?) 208 Clinical Studies 150 RCT (160 any clinical trial) Multiple open studies using BPD as an outcome

Copyright © 2016 American Medical Association. All rights reserved. From: Trends in Care Practices, Morbidity, and Mortality of Extremely Preterm Neonates, 1993-2012 JAMA. 2015;314(10):1039-1051. doi:10.1001/jama.2015.10244 Date of download: 8/29/2016 Copyright © 2016 American Medical Association. All rights reserved.

Current Consensus Definition In Premature infants: 28 days of supplemental oxygen use Oxygen use at 36 weeks gestation No need for supplemental oxygen: Mild BPD Need for < 30% supplemental oxygen: Mod BPD Need for ≥ 30% supplemental oxygen: Severe BPD No mention as to cause of the need for oxygen

The “New BPD” model Any questions about this model? Pathology of Alveolar Growth Inhibition Relationship “C” Wheezing Medications Hospitalization Relationship “A” Relationship “B” Oxygen use at 36 weeks PMA surrogate marker of pathology Any questions about this model? We know relationship “C” indirectly because we know relationships “A” and “B”

How did this happen? Vitamin A and Steroids Both decrease BPD by ~10% No measurable change in long-term pulmonary outcome

if ? ? How did this happen? Pathology of Wheezing Alveolar Growth Inhibition ? Wheezing Medications Hospitalization if ? Oxygen use at 36 weeks PMA If you assume relationship A, does that mean outcomes are not related to disease or surrogate marker?

How did this happen? CuZn superoxide dismutase No change in BPD Dramatic drop in respiratory morbidity (~50% reduction in hospital visits, respiratory meds and ED visits)

if ? ? How did this happen? Pathology of Wheezing Alveolar Growth Inhibition Wheezing Medications Hospitalization ? ? Oxygen use at 36 weeks PMA If you assume relationship C, does that mean oxygen use at 36 weeks is not related to lung pathology?

What prevents BPD? Prevent preterm birth Antenatal steroids Post natal steroids Fluid Restriction Caffeine Vit A?

Honestly, there is not a lot new that seems to work? P > 0.05? Vit E iNO Cromolyn Sodium Surfactant Diuretics Inhaled Steroids Honestly, there is not a lot new that seems to work? Why?

Rates Stagnant New Therapies Elusive More ELBW infants are surviving? Perhaps our understanding of the disease is limited? Perhaps BPD is not really the disease we think it is?

History of BPD

Origins – Northway RDS Prior to 1960s – White out Chest XRay Death Limits of viability ~ 28 wks , 1200g Outcome known at about 5 DOL RDS White out Chest XRay Death Chest XRay without white out Survival with little Resp. Morbidity

New devices, New survival 1967 – Northway publishes article on neonatal lung disease 32 patients (1962-65) > 24 hrs on vent, > 6 days of Oxyhood FiO2 80-100% Mean Age of survivors 34 weeks Mean weight: 2.3 kg

First Neonatal Ventilators Late 1950s – Early 60s Adult ventilators for neonates Engstrom respirator (Volume controlled) Dr. Ian Mc Donald’s “The Puffer” Dr. Forrest Bird’s “Baby Bird” Advent of PEEP, Pressure controlled ventilation More survivors

Prototype neonatal and pediatric ventilator built by Jimmy Schulz, RRT. Three men chatting over the Engstrom Respirator, which helped to treat cases of respiratory paralysis at Beth Israel, now Beth Israel Deaconess Medical Center, in 1963 http://www.cobth.org/ThrowBackThursday.html Donald M. Null et al. Pediatrics 2012;129:S20-S26

Original Baby Bird neonatal ventilator built by Dr Forrest Bird. Donald M. Null et al. Pediatrics 2012;129:S20-S26 ©2012 by American Academy of Pediatrics

From Northway’s Paper Care of the intubated newborn in the ‘60s Supine ETT changed every 48hrs Suctioned every hour PIP 20-40 cmH2O (Bennett resp. and Bird Mark 7) High oxygen concentrations No PEEP No Synchronization

Bronchopulmonary Dysplasia – Autopsy Studies Stage 1: Respiratory Distress Syndrome 2-3 days Airway Injury Stage 2: Tissue regeneration 4-10 days Metaplasia Stage 3: Transition to Chronic Disease 10d – 3 weeks Fibrosis, Smooth Muscle Hypertrophy Stage 4: Chronic Lung Disease Fibrosis and Emphysema

On Autopsy On Autopsy On Chest XRay Necrotizing Bronchitis Vascular changes Inflamation Over inflation and atelectasis Hyperexpansion Air leak Cystic Changes Highlight the altered development of the lung from ventilator induced lung injury and oxygen toxicity

BPD as Defined by Northway Oxygen need at 28 days Changes on Chest XRay consistent with chronic lung disease “We chose to term the syndrome of chronic lung disease described above as bronchopulmonary dysplasia (BPD) to emphasize the involvement of all the tissues of the lung in the pathologic process” - Northway, 1967 -

The 1967 model Northway Northway’s “BPD” outcomes Oxygen use 28d plus Northway’s attributes

Tooley 1970 1972-74 1976 1978 CPAP MV w/ PEEP PaO2<50 FiO2 1.0 PaO2<50 on 0.6 or BW>1750g MV w/ PEEP PaCO2>80 apnea PaCO2>70 PaCO2>60 PaCO2>45 Fio2>0.3 for bwt <1750

Tooley’s Survival data

1979 NIH BPD conference Not everyone is going through the stages 50% jumped 124 Not everyone was on the vent “The worst case of BPD I’ve seen was in a child that was never on the ventilator” Not everyone got the same disease many commented on a growing number of <1000g survivors with a much more mild disease that seemed different than the usual “BPD” Epidemiologic nightmare Everyone brought a different, practical definition to the table

1979 NIH Conference Range of proposed diagnostic criteria Bancalari’s BPD [strict, VILI] IPPV during first week, for 3+ days Clinical signs and O2 (PaO2 < 50) persisting for 28 days CXR “persistent strands of densities in both lungs, alternating with areas of normal or increased lucency (some with bullae)” Tooley’s BPD [open, includes “mild”] Any radiographic abnormality 1 mo, O2 (PaO2 < 60) (or CO2 >45)

? The model Northway’s “BPD” Northway outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria.

Recap Started as oxygen/VILI Practice changed to make it less injurious A new population of survivors (<1000g) Still prolonged oxygen need, but less severe Diagnostic criteria is modified to include these new kids Simplest definition prevails…

The start of the paradigm shift The “disease” was the histopathology Predominantly interstitial and dysplasia Associated with: “Chronic” clinical signs Markedly abnormal CXR Diagnostic criteria attempted to identify those with the histopathologic disease But, eventually even the need for CXR changes were dropped

Shennan and the 36 week rule 1988 Shennan echoes common concern Oxygen use at 28 days is abnormal But if the child has a “normal” long term outcome then who cares? Attempted to separate those with apparently inconsequential lung dysfunction from those that would have bad long-term outcomes Shennan et al, Pediatr 1988;82:527

Shennan and the 36 week rule At least, if we label a kid with BPD/CLD “it should reflect an increased likelihood of abnormal pulmonary signs in infancy.” Sounds intuitive and innocuous, but it represents a dramatic shift in focus Anyone notice what happened? Shennan et al, Pediatr 1988;82:527

The birth of the 36 week rule The “disease” is now the long term outcome (Long Term Outcome Disease = LTOD) 36 weeks on oxygen Is not a disease It’s a screening test for the disease Now, a brief aside about screening tests… Shennan et al, Pediatr 1988;82:527

“The Square” + - + - Disease State Sensitivity = A/(A+C) [TP/all those with disease] Specificity = D/(B+D) [TN/all those without disease] PPV = A/(A+B) [TP/all positives] NPV = D/(C+D) [TN/all negatives] - A True Positive B False Positive + Test Result C False Negative D True Negative -

PPV NPV Patient - centric Test Result Diagnosis True positive 100%-PPV False positive True positive PPV 100%-PPV NPV 100%-NPV True negative False negative Patient - centric

Sensitivity Specificity Disease - centric Test Result Diagnosis True positive False negative Sensitivity 100%-Sensitivity Specificity 100%-Specificity True negative False positive Disease - centric

Accuracy A measure of just being RIGHT (TP+TN)/(TP+TN+FP+FN) Assumes equally bad to have false positive as false negative (overly simplistic) (TP+TN)/(TP+TN+FP+FN) “end accuracies” If 35% of kids have a bad pulmonary outcome Assuming all kids will be healthy is 65% accurate Assuming all kids will be sick is 35% accurate If your “accuracy” is less than one of the “end accuracies” then this may not be the best way to describe the goodness of your test, or, it may be a stinky test

So? What’s most important to you? Because you really want them ALL to be good! SENSITIVITY SPECIFICITY POSITIVE PREDICTIVE VALUE NEGATIVE PREDICTIVE VALUE ACCURACY But real life means you’ll have to sacrifice something Maximize sens/spec when studying a disease Maximize ppv/npv when using as diagnostic

Back to: The birth of the 36 wk rule Shennan’s study 1981 to 1985 480g - 1500g No surfactant No statement about ANS/PNS 94% got MV (43% dx with RDS) [prophylactic] O2 “requirement” PaO2 50-60, Sats 87-93% Women’s College Hospital Toronto, Canada. 758 infants, 644 survived to 28d, all enrolled in 2y f/u clinic. 39 lost to follow-up = 605/644 or 94% … wow. Shennan et al, Pediatr 1988;82:527

The birth of the 36 wk rule Abnormal outcome (up to 2 years old): Death due to non-anomalous cause 16 Requirement for O2 at 40wks CGA 12 Respiratory tract surgery 2 2 or more hosp admiss for respiratory disease 28 Wheezing requiring drug therapy 55 Xray or incr WOB with FTT or CP/MR 24 (composite bad pulm outcome = 20%) “may identify some infants with non-pulmonary pathology” – very broad clinical outcomes Shennan et al, Pediatr 1988;82:527

The birth of the 36 wk rule Shennan et al, Pediatr 1988;82:527

The specifics 28d 35wk 36wk 37wk Accuracy 71% 81% 85% 86% Sens Spec Is there really a BEST accuracy? Are 81, 85 and 86% different enough to pick a single cutoff? 28d 35wk 36wk 37wk Accuracy (they selected 36 weeks based on accuracy) 71% 81% 85% 86% Sens Spec 79% 69% 66% 63% 91% 51% 95% PPV NPV 38% 93% 55% 90% 89%

No correlation with underlying histopathology Sheenan refigured Based on the methods and data the definition should be 37 weeks Real conclusion: the closer you were to discharge on oxygen the more likely you were to have a bad outcome “36” Implied a precision that really was not there (but it’s wicked-easy to use!) There really does not appear to be ONE best number Replaced one “meaningless number” (28) with another No correlation with underlying histopathology (the cord was cut… “disease” was no longer tied to a histopathology)

The model Northway’s “BPD” Northway outcomes Oxygen use 28d Shennan plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (Oxygen use at 36 weeks) Shennan outcomes

Ability to Predict Pulm. Outcomes Oxygen use at 36wks predicts… Sens Spec PPV NPV Wheeze and cough at 2 years (Greenough, 1996) 0.32 0.89 0.47 0.80 Need for Respiratory Meds at 2 years (Palta, 1998) 0.33 0.95 0.48 0.82 Hospital admission for Respiratory problems (Greggorie, 1998) 0.53 0.62 0.39 0.74 Composite pulmonary outome (Davis, 2002) 0.46 0.75 0.57 Need for Pulm Meds (Ebrenkranz, 2005)( n=3848) Rehospitalization 0.54 0.44 0.71 For those with a birth weight less than 1,000g and off of oxygen before 28 days of life (No BPD!), the chances of further respiratory problems (readmissions, need for medications, symptoms, etc.) during the first 2 years of life is…. 27-41% (Seems pretty high for those with no apparent disease) Lefkowitz. J Perinatology 2006

Suddenly… After surfactant, pathology “CHANGES” to alveolar growth inhibition (AGI) How many examples of a disease can you think of where 1) the pathology changes 2) no one bats an eye How can no one question a disease changing it’s pathology? “BPD”/”CLD” become shorthand for “lung dysfunction” So it’s OK to have more than one pathology for this…

What are the Outcomes? Define outcomes of interest Decide on these as a standard so that EVERYONE can use them Asthma: small airways disease, not a component of the histopathology of AGI SHENNAN et al ‘88 (<1500g) Death due to non-anomalous cause Requirement for O2 at 40wks CGA Respiratory tract surgery 2 or more hosp admiss for respiratory disease Wheezing requiring drug therapy Xray or incr WOB with FTT or CP/MR (composite bad pulm outcome = 20%) DAVIS et al ’02 (<1000g) Death after discharge Oxygen dependency after discharge Hospital readmission after respiratory illness Respiratory medication for > 2mo (composite bad pulm outcome = 54%) PALTA et al ’98 (<1500g) use of resp meds > 2wks after d/c up to 2yo Dx of asthma after 2yo Hosp for above resp causes until lost to follow-up (Separate bad pulm outcomes = max 34%) GREGOIRE et al ’98 (<28wk) Pulmonary problems (bronchiolitis, asthma, ALTE, apnea, trach complication) PICU admission Surgery (inguinal hernia, ENT) Persistent respiratory problems (asthma, meds, tracheostomy, home O2) (Separate bad pulm outcome, max=31% resp readmission) EHRENKRANZ et al ’05 (<1000g) Incidence of treatment with pulmonary meds Rehospitalization for pulmonary causes (Separate bad pulm outcomes = 24-47%)

The 1990s model Northway’s “BPD” Northway outcomes Gregoire outcomes Palta outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (Oxygen use at 36 weeks) Shennan outcomes Davis outcomes Ehrenkranz outcomes

The 1990s model Northway’s “BPD” Northway outcomes Gregoire outcomes Husain’s AGI Palta outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (Oxygen use at 36 weeks) Shennan outcomes Davis outcomes Ehrenkranz outcomes

NICHD BPD conference, 2000 Proposed a new definition Describes “magnitude” of supplemental oxygen need during the NICU stay Assume it reflects disease “severity” Severity classifications (<32 wk version) None (not on O2 for 28 days) Mild (on O2 for the 28 days, but off at 36 weeks CGA or discharge) Moderate (on <30% O2, without PPV/CPAP at 36 weeks CGA or discharge) Severe (on ≥30% O2 or PPV/CPAP at 36 weeks CGA or discharge)

Good outcomes Bad outcomes Screening cutoff: Increasing “magnitude” of supplemental oxygen need during first NICU admission Good outcomes Bad outcomes Sensitivity: percent of bad outcome kids to the right of your cutoff line Specificity: percent of good outcome kids to the left of your cutoff line Ideal: all sick kids to the right, all healthy kids to the left

Bad outcomes (actual) Where would you draw a cutoff to maximize sick on one side and healthy on the other? Ehrenkranz et al, Pediatr 2005;116:1353-60

Bad outcomes (actual) Where would you draw a cutoff to maximize sick on one side and healthy on the other? Ehrenkranz et al, Pediatr 2005;116:1353-60

Outcome between d/c and 18-22mo f/u (NICHD database) Ehrenkranz, NICHD <32wks, ≤1000g Pulmonary Meds 28d BPD 38% 36w BPD 44% Consensus mild 30% Consensus moderate 41% Consensus severe 47% Outcome between d/c and 18-22mo f/u (NICHD database) Ehrenkranz et al, Pediatr 2005;116:1353-60

Pulmonary Meds 82 57 22 26 59 86 38 43 47 73 71 67 46 58 64 Sens Spec mild, mod, sev v. none mod, sev v. none mild sev v. none, mild, mod Sens 82 57 22 Spec 26 59 86 PPV 38 43 47 NPV 73 71 67 Accu 46 58 64 Pulmonary Meds Recall that the “end accuracy” (given a 35% bad outcome rate) is 65% Ehrenkranz et al, Pediatr 2005;116:1353-60

Outcome between d/c and 18-22mo f/u (NICHD database) Ehrenkranz, NICHD <32wks, ≤1000g Pulmonary Meds 28d BPD 38% 36w BPD 44% Consensus NO BPD Consensus mild 30% Consensus moderate 41% Consensus severe 47% 27% Outcome between d/c and 18-22mo f/u (NICHD database) Ehrenkranz et al, Pediatr 2005;116:1353-60

Characteristics by severity None Mild Mod Sev BWT 867 791 766 737 EGA 27.8 25.7 25.6 %male 39 43 51 56 %IVH 3-4 9 17 19 25 %PVL 3 5 6 10 %NEC 2+ 8 14 %Postnatal Steroids 46 67 78 If and when you “correct” for all these other predictive factors, “BPD” no longer adds anything to your ability to predict outcome. Prolonged oxygen requirement is a characteristic of those born physiologically immature. Ehrenkranz et al, Pediatr 2005;116:1353-60

QUESTIONS So, what are the signs of Alveolar Growth Inhibition (AGI) in babies? NHLBI: “… there are no tests with which to evaluate the abnormalities resulting from the arrested alveolar and vascular development in BPD, or for monitoring how these abnormalities change with time.” So, how do we know if someone has AGI? We don’t

BPD The current model But rather than talk about At least 2 histopathologies, Multiple outcomes Multiple clinical diseases simplify simplify BPD Northway’s “BPD” Northway outcomes Gregoire outcomes Husain’s AGI Palta outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (Oxygen use at 36 weeks) Shennan outcomes Davis outcomes Ehrenkranz outcomes

The model BPD Northway’s “BPD” Northway outcomes Gregoire outcomes Husain’s AGI Palta outcomes Shennan outcomes Davis outcomes Ehrenkranz outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (UNLINK .. Oxygen use at 36 weeks)

The model BPD Northway’s “BPD” Northway outcomes Gregoire Husain’s AGI Palta outcomes Shennan outcomes Davis outcomes Ehrenkranz outcomes Oxygen use 28d plus Northway’s attributes A lot of other clinical functional disease that don’t fit original Criteria. (UNLINK .. Oxygen use at 36 weeks)

The model BPD Husain’s AGI “BPD” outcomes Oxygen at 36 weeks

The model Husain’s AGI “BPD” outcomes BPD Oxygen at 36 weeks

The model BPD Husain’s AGI “BPD” outcomes Oxygen at 36 weeks Do we mean the pathology of Northway or Husain? Are the outcomes related to: Lung damage Prematurity / physiologic immaturity? Family history Oxygen at 36 weeks

“BPD” means too many things Being on oxygen at 36 weeks Does not mean we know what the lung histopathology looks like Does not mean we know what the outcome will be Changing the incidence of oxygen at 36 weeks Does not mean changes in the underlying histopathology Does not mean changes in long-term outcome

“BPD” means too many things Characteristic of being on oxygen at 36 weeks Histopathology of AGI Long term outcomes of prematurity (neurodevelopmental) Assumption that a change in one means that you’ve changed the other three No supporting evidence for this But there is evidence against it…

? ? ? The model ≠ Reality Husain’s AGI “BPD” outcomes Oxygen at 36 weeks “clinical BPD”

P > 0.05? Vit E iNO Cromolyn Sodium Surfactant Diuretics Inhaled Steroids Honestly, there is not a lot new that seems to work? Why? Because we are not sure what we are targeting!

From Ana Karenina “All happy families are alike; each unhappy family is unhappy in its own way.” Leo Tolstoy – 1875 Yasnaya Polyana, Russia “All normally developed preterm lungs are alike; each BPD lung is abnormal in its own way.” Dr.. Namasivayam Ambalavanan – 2016 UAB - Birmingham

Statements and Questions Lungs can be devastated in the NICU. Being a micropremie is devastating condition. What biomarkers should we be using CT? MRI? PFTs? Inflammatory markers? What outcomes should we be using? Hospitalizations? Medications? PFTs? Is AGI related to long term outcomes? Does AGI heal? How do we identify kids with it? How do we know when it’s gone?

Time to start over To gain mastery over a disease Learn it’s histopathology / pathophysiology Is it AGI? Is it BPD? Is it something else? Learn how to identify it Clinical signs not related to treatment Map out its natural history Map out its consequences On the infant (outcomes) On other diseases (asthma) THEN: we can design trials to alter disease to change the consequences If we change AGI (like in animal models) so what?

Questions

Questions

Definition changes 2000 – NICHD: Oxygen at 28 days and O2 requirement of 21% at 36 wks O2 requirement of 22-30% at 36 weeks O2 requirement of >30% or pos. pressure at 36 weeks

New BPD Epidemiology of BPD changed as care changed CPAP SIMV Volume targeted ventilation HFV Low O2 strategies Steroids Less chronic fibrosis, more arrested development

But what are we treating Old BPD New BPD Supplemental O2 need at 28 days Autopsy or Chest XRay findings Supplemental O2 at 28 days Suppl. O2 need at 36 weeks Histopathologic Disease Surrogate Marker

Histopathologic Disease Surrogate Marker Outcome The race to 36 weeks

Why do we care about BPD Treatment? (Can be managed, but I can’t treat it) Prognosis? What will the first winter be like? How about the second winter? Re admission? Asthma, COPD, Development How good is my NICU? Need something to talk about at conferences

Ability to Predict Pulm. Outcomes Oxygen use at 36wks predicts… Sens Spec PPV NPV Wheeze and cough at 2 years (Greenough, 1996) 0.32 0.89 0.47 0.80 Need for Respiratory Meds at 2 years (Palta, 1998) 0.33 0.95 0.48 0.82 Hospital admission for Respiratory problems (Greggorie, 1998) 0.53 0.62 0.39 0.74 Composite pulmonary outome (Davis, 2002) 0.46 0.75 0.57 Need for Pulm Meds (Ebrenkranz, 2005)( n=3848) Rehospitalization 0.54 0.44 0.71 For those with a birth weight less than 1,000g and off of oxygen before 28 days of life (No BPD!), the chances of further respiratory problems (readmissions, need for medications, symptoms, etc.) during the first 2 years of life is…. 27-41% (Seems pretty high for those with no apparent disease Lefkowitz. J Perinatology 2006

Copyright © 2016 American Medical Association. All rights reserved. From: Trends in Care Practices, Morbidity, and Mortality of Extremely Preterm Neonates, 1993-2012 JAMA. 2015;314(10):1039-1051. doi:10.1001/jama.2015.10244 Date of download: 8/29/2016 Copyright © 2016 American Medical Association. All rights reserved.

Copyright © 2016 American Medical Association. All rights reserved. From: Trends in Care Practices, Morbidity, and Mortality of Extremely Preterm Neonates, 1993-2012 JAMA. 2015;314(10):1039-1051. doi:10.1001/jama.2015.10244 Date of download: 8/29/2016 Copyright © 2016 American Medical Association. All rights reserved.

Copyright © 2016 American Medical Association. All rights reserved. From: Trends in Care Practices, Morbidity, and Mortality of Extremely Preterm Neonates, 1993-2012 JAMA. 2015;314(10):1039-1051. doi:10.1001/jama.2015.10244 Date of download: 8/29/2016 Copyright © 2016 American Medical Association. All rights reserved.

The Physiologic Challenge Maybe the problem is that oxygen “need” is too subjective Walsh et al: challenge at 36 weeks 10% reduction in diagnosis of “BPD” “BPD” already misses too many kids with a bad outcome. Diagnosing fewer kids with “BPD” isn’t going to improve that

The Physiologic Challenge Ehrenkranz data (~1680 cases of 36 week BPD) Move 168 kids from “BPD” to “no BPD” Measures: sen spe ppv npv Baseline 54 62 44 71 Best case 54 69 49 73 Worst case 42 62 38 66

ROC – find the best cutoff when you have a tiered definition Nirvana 100% sensitivity 1 - specificity 100% AUC 1.0 = perfect test, AUC 0.80 = good test, AUC 0.50 = coin flip

Pulm Outcome: ROC severe moderate mild AUC 0.59

Hosp Outcome: ROC AUC 0.57

Neurodevelopment and CP Neurodevelopmental outcome Cerebral Palsy outcome AUC 0.63 AUC 0.63

Home O2 prescribed ROC Why so good? Moderate = on oxygen at 36 wks. In other words, “Mild” = off oxygen at 36 weeks Means being on oxygen at 36 weeks reasonably predicts being prescribed oxygen at discharge. AUC 0.83

36 weeks since Shennan Report 32 89 47 80 33 95 48 82 53 62 39 74 46 Sens Spec PPV NPV Greenough ‘96 32 89 47 80 Palta ‘98 33 95 48 82 Gregoire ’98 53 62 39 74 Davis ’02 46 75 57 Ehrenkranz ’05 54 44 71 NONE of these authors felt the 36 week definition did a good enough job of predicting long term outcome to be of any use