Achieved Versus Intended Pulse Oximeter Saturation in Infants Born Less Than 28 Weeks' Gestation: The AVIOx Study
OBJECTIVE. The objective of this study was to document pulse oximeter saturation levels achieved in the first 4 weeks of life in infants who were born at <28 weeks' gestation, compared with the levels that were targeted by local policy, and examine factors that are associated with compliance with the target range.
METHODS. Infants who were <28 weeks' gestation and ≤96 hours of age were enrolled in a prospective, multicenter cohort study. Oximetry data were collected with masked signal-extraction oximeters for a 72-hour period in each of the first 4 weeks of life. Data were compared with the pulse oximeter saturation target range prescribed by local institutional policy. Factors that were associated with intended range compliance were identified with hierarchical modeling.
RESULTS. Fourteen centers from 3 countries enrolled 84 infants with mean ± SD birth weight of 863 ± 208 g and gestational age of 26 ± 1.4 weeks. Oxygen saturation policy limits ranged between 83% and 92% for lower limits and 92% and 98% for upper limits. For infants who received respiratory support, median pulse oximeter saturation level achieved was 95%. Center-specific medial levels were within the intended range at 12 centers. Centers maintained infants within their intended range 16% to 64% of the time but were above range 20% to 73% of the time. In hierarchical modeling, wider target ranges, higher target range upper limits, presence of a policy of setting oximeter alarms close to the target range limits, and lower gestational age were associated with improved target range compliance.
CONCLUSIONS. Success with maintaining the intended pulse oximeter saturation range varied substantially among centers, among patients within centers, and for individual patients over time. Most noncompliance was above the intended range. Methods for improving compliance and the effect of improved compliance on neonatal outcomes require additional research.
The incidence and severity of retinopathy of prematurity (ROP) and bronchopulmonary dysplasia (BPD), two of the most debilitating complications of premature birth, appear linked to management of supplemental oxygen therapy.1–8 Despite the association between oxygen administration and these adverse outcomes, the optimal oxygen saturation range to maintain in very premature infants to minimize the risk for complications while maximizing benefit is unknown.9
Rigorous control of arterial oxygen saturation as estimated by pulse oximetry (Spo2) may offer a means of reducing the incidence and the severity of ROP and BPD in premature infants. Recent cohort studies reported higher rates of ROP, ROP that required surgery, and BPD in nurseries where policy allowed Spo2 near 100%, compared with nurseries where policy targeted lower Spo2 ranges.7,8,10 Although oxygen-related morbidities may be associated with target Spo2 ranges as established by unit policies, the degree to which actual care practices reflect these policies is not known. Few data are available from extremely premature newborn infants during routine care comparing achieved and intended Spo2 levels.
The primary goal of the AVIOx study was to compare achieved Spo2 levels in infants who were born at <28 weeks' gestation with Spo2 levels that were recommended by local NICU policy. Specifically, the objectives of this exploratory study were (1) to document achieved Spo2 levels in the first 4 weeks of life in infants who were <28 weeks' gestation during routine care, (2) to compare achieved Spo2 levels with the intended Spo2 range, and (3) to examine characteristics that are associated with successful maintenance of the target range.
This was a prospective cohort study in which 14 centers in 3 countries enrolled eligible infants between July 2003 and February 2004. Institutional Review Board or Ethics Committee approval was obtained at each center. Before the study, participating center investigators provided their NICU's written policy or guideline regarding their intended range of Spo2 and oximeter alarm settings. For each center, upper and lower limits of the intended Spo2 range were specified by written policy or physician-ordered standard practice for clinical care during weeks 1 to 4 in infants who were born at <28 weeks' gestation. Oximeter upper and lower alarm settings were the Spo2 values at which the center's routine care oximeter alarms signaled unacceptable saturation levels, as required by written policy. Each center's alarm setting policy was categorized as tight or loose. Tight alarm settings were defined as upper alarm limit not higher than 1 percentage point above the upper limit of the intended Spo2 range and lower alarm limit not >2 percentage points below the lower limit of the intended range.
A convenience sample of consecutive infants who were <28 weeks' gestation by best obstetric estimated gestational age were screened for eligibility. Eligible infants were ≤96 hours of chronological age. Infants were excluded when cyanotic congenital heart disease was present or at the discretion of the center investigator when death was believed to be imminent. Informed parental consent was obtained before enrollment. At each center, enrollment continued until either 5 to 10 infants were enrolled or the end of the study enrollment period was reached.
Participating centers delivered their usual clinical care to enrolled patients in all respects. Oxygen saturation management for each enrolled infant occurred in the usual manner using continuous monitoring with the center's routine daily care oximeters. In addition, each patient was monitored simultaneously with a masked Masimo Signal Extraction Technology (SET) Radical v4 oximeter (Masimo Corp, Irvine, CA). Spo2 sampling was every 2 seconds for a 72-hour monitoring period in each of the first 4 weeks after birth. To maximize capture of typical compliance, the Spo2 display of the SET oximeters was inactivated and their alarms were silenced. Bedside caregivers received no feedback from the SET oximeters other than for lead malfunction. Although bedside caregivers were aware of the general purpose of the study, they received no formal educational or feedback that would lead them to alter oxygen use.
Data from the first 12 hours after birth that were sufficient to calculate the Score for Neonatal Acute Physiology, Version II (SNAP-II score)11 were recorded for all patients, along with baseline data including gestational age, birth weight, race, and gender. At the end of each monitoring period, Spo2 data from the SET oximeters were transferred to a computer using ProFox software (ProFox Associates, San Diego, CA) and then to the study's Data Coordinating Center Internet server. Spo2 data were downloaded from routine care oximeters when feasible, depending on the type of oximeter in use. Electronic collection of actual alarm setting values from routine care oximeters was not possible. Data regarding ventilator or continuous positive airway pressure support, supplemental oxygen therapy, pulmonary illness acuity,12 and Spo2 values recorded hourly by bedside caregivers during each monitoring period were obtained from the medical chart.
Percentage of time spent below, within, and above the intended range was determined by comparing SET oximeter data with the Spo2 range that was prescribed by local institutional policy for all patients. For all analyses, only monitoring periods with “modifiable” Spo2 were studied. In general, Spo2 level was defined as modifiable when it could be adjusted downward by reducing the amount of supplemental oxygen or mechanical respiratory support. For example, for an infant who was breathing a fraction of inspired oxygen (Fio2) of 40%, the Spo2 level could be modified by weaning the Fio2. For this study, a monitoring period was considered to have modifiable Spo2 and was included in analyses when the patient had modifiable Spo2 levels for at least 80% of the monitoring period. A monitoring period was considered “nonmodifiable” when the infant was not on mechanical ventilation or continuous positive airway pressure and was on supplemental oxygen for <80% of the interval. For example, an infant who was breathing room air, was not on mechanical respiratory support, and had Spo2 greater than the upper limit of the center's intended range was considered to have a nonmodifiable Spo2. For each monitoring period with modifiable Spo2, for all such monitoring periods combined for each patient, and for all such monitoring periods combined for each center, percentages of time spent below, within, and above the intended range were calculated. For each center, the median, range, and interquartile range (IQR) were calculated for achieved Spo2 using SET oximeter data, for Spo2 values documented in the medical chart, and also for Spo2 values from the routine care oximeters when available. Pulmonary acuity score was calculated for each monitoring period.12 Percentage of time in the intended range during day versus night shifts was examined using the paired-samples t test. Spo2 data values that were flagged by the study SET oximeters as possible artifact, as a result of suboptimal acquisition of signal, were excluded from analyses. P < .05 was considered statistically significant.
Because monitoring periods were nested within infants and infants were nested within NICUs, factors that were associated with successful maintenance of the intended range were identified with 3-level hierarchical modeling13,14 using HLM 6.0 software (Scientific Software International, Lincolnwood, IL). For all models, proportion of time in the intended range for each monitoring period was the outcome measure, with monitoring period characteristics, infant characteristics, and center characteristics as independent variables. Modeling was accomplished in several steps using iterative generalized least squares. Initially, an unelaborated model was created in which proportion of time in the intended range for each monitoring period was described as a single mean value with random error components attributable to variation between centers, between patients within centers, and between monitoring periods within patients. Independent variables were introduced in subsequent modeling. At the monitoring period level (level 1), the proportion of monitored time in the intended range was modeled as a function of characteristics at the time of monitoring. The relationship between each characteristic and the proportion of monitored time in the intended range was summarized by a regression coefficient. Monitoring period characteristics that were tested included chronological age, postmenstrual age, and the mechanical ventilator status and pulmonary acuity score at the time of monitoring. In the level 2, or patient-level, models, each level 1 regression coefficient including the intercept was treated as an outcome and was modeled as a function of infant characteristics. Patient-level characteristics that were tested included gestational age, birth weight, and SNAP-II score at birth. Finally, level 2 regression coefficients were modeled as a function of level 3, or NICU-level, characteristics. NICU-level characteristics that were tested included the upper limit of the Spo2 target range, the width in percentage points of the target range, and whether the NICU had a policy of tight oximeter alarm limits. The resulting 3-level model included monitoring period-, patient-, and NICU-level estimates of residual variance. Random effects were tested for each significant characteristic identified for assessment of whether its effect on maintenance of the intended range was constant or varied with time, between patients, or between centers. The percentage of variation that was explained by the final model was obtained by comparing the residual variance for the fully elaborated 3-level model with the variance of the unelaborated model described above.14
Characteristics of participating centers and enrolled patients are shown in Table 1. A total of 181 infants who were <28 weeks' estimated gestational age were screened, and 84 were enrolled and monitored. Individual centers enrolled between 2 and 10 infants. Data were collected for 307 monitoring periods of median duration of 67.3 hours, including 263 periods with modifiable Spo2 from 80 patients. Four patients had no monitoring periods with modifiable Spo2. Patients had mean ± SD gestational age of 26 ± 1.4 weeks and birth weight of 863 ± 208 g. Forty-four (52%) patients were white, 21 (25%) were black, 13 (15%) were Hispanic, 1 (1%) was other, and 5 (6%) were of unknown race. Forty-four (52%) patients were male. The overall median Spo2 documented for all patients at all centers during the first 4 weeks of life was 95% (IQR: 91%–97%) for monitoring periods with modifiable Spo2. Overall median Spo2 was 96% (IQR: 93%–98%) for nonmodifiable monitoring periods.
Intended Spo2 ranges and Spo2 values that actually were achieved at each center are shown in Table 2 for all patients during monitoring periods with modifiable Spo2. Twelve centers had a defined Spo2 target range at the time of study survey, and 2 others developed a policy and implemented it before beginning enrollment. None of the participating centers had policies that targeted different Spo2 ranges during different monitoring periods in the first 4 weeks. The lower limits of the intended Spo2 ranges varied from 83% to 92%, and upper limits ranged from 92% to 98%. Twenty different oximeter models from 7 manufacturers were used for routine care by participating centers, all measuring functional hemoglobin saturation. Six centers had a written policy or guideline that required tight oximeter alarm settings as defined by this study.
NICU-specific median achieved Spo2 during monitoring periods with modifiable Spo2 was within the intended range at 12 centers, including 8 centers where median Spo2 was within 1% of the upper limit of their target range. Two centers had median Spo2 above the targeted range. Overall, study patients spent 16%, 48%, and 36% of monitored time below, within, and above their NICU’s intended range, respectively. Successful maintenance of the intended range varied substantially between centers, between patients within centers, and between monitoring periods for individual patients (Table 2). At different centers, overall time within the intended range for all patients combined ranged from 16% to 64% during periods of modifiable Spo2. Individual patients ranged from 0% to 47% time below, 6% to 75% time within, and 5% to 90% time above their center's intended range during their combined monitored periods. Individual monitoring periods with modifiable Spo2 ranged between 0% and 79% time below, 4% and 93% time within, and 1% and 93% time above the intended range.
Spo2 Values From SET Oximeters, Routine Care Oximeters, and Documented Spo2
The distribution of Spo2 values that were obtained with the masked SET oximeters and those that were recorded in the medical chart by bedside staff were similar (Fig 1). To evaluate whether systematic differences between the routine care and SET oximeters could result in underestimation or overestimation of success with maintaining the intended range, we repeated descriptive analyses (Table 3). For 4 centers where data from routine care oximeters were available, SET oximeter Spo2 values were offset to reflect any difference between center-specific median Spo2 from SET oximeters and center-specific median Spo2 from routine care oximeters. For the 10 centers where data from routine care oximeters were not available, the median of Spo2 values that were documented in the medical chart were used. Calculated offset values (median SET Spo2 − median routine care Spo2 or median SET Spo2 − median medical chart Spo2) ranged from −3 to 1 for individual centers. With the use of calculated offset values, the overall percentage of time below, within, and above the intended range for all enrolled patients was 15%, 46%, and 39%, respectively. The percentage of time within the intended range improved for some centers but worsened for others, with individual centers varying between 24% and 59% of time within the intended range. Overall median was 95% (IQR: 92%–97%) for Spo2 values that were obtained from the study SET oximeters and 95% (IQR: 92%–97%) for Spo2 values that were documented in the medical chart for the same monitoring periods.
Date and time that monitoring began as recorded by study personnel were within 1 hour of the date and time recorded by the SET oximeter for 193 monitoring periods with modifiable Spo2 from 70 patients, confirming that the local time that was recorded automatically by the oximeter was accurate. For these patients, total monitored time between 7:00 am and 3:00 pm was not significantly different from monitored time between 11:00 pm and 7:00 am regarding percentage of time within (47.4 ± 16 vs 46.7 ± 15; n = 70; mean difference: 0.7 percentage points; 95% confidence interval [CI]: −1.3 to 2.6; P = .49), above (35.9 ± 19 vs 36.9 ± 20; mean difference: −1; 95% CI: −2.8 to 0.7; P = .59), or below (16.7 ± 12 vs 16.4 ± 13; mean difference: −0.3; 95% CI: −0.1 to 1.7; P = .24) the intended range (paired samples t test). This paired analysis was estimated posthoc to have 80% power to detect a 2.8 percentage point difference in day versus night shift maintenance of the target range. At all centers, center-specific median Spo2 for all patients during the day was within 1 percentage point of the median Spo2 at night.
Centers with a policy of tight oximeter alarm limits maintained patients within the intended range 52.9% (IQR: 46.7%–58.9%) of monitored time, compared with 42.2% (IQR: 24.0%–48.4%) for centers without tight alarm limits (P = .04, Mann-Whitney U test). Overall percentage of time below, within, and above the intended range was 16%, 49%, and 36%, respectively, for 144 monitoring periods that were performed while infants were on mechanical ventilation (56 infants) and 15%, 48%, and 37% for 119 periods during which infants were not on mechanical ventilation (49 infants). Mean percentage of time in the intended range was 50.9 ± 18.2% in week 1 (n = 70 monitoring periods with modifiable Spo2), 49.6 ± 16.8% in week 2 (n = 65), 47.1 ± 17.5% in week 3 (n = 67), and 44.1 ± 16.4% in week 4 (n = 61). This downward trend in compliance with the targeted range was evident for both ventilated and nonventilated infants and was not related to pulmonary acuity scores. Pulmonary acuity scores were not correlated with week of monitoring (data not shown).
Unelaborated hierarchical modeling demonstrated that 45.3% of variation in successful maintenance of the target range was attributable to variation between monitoring periods within patients, 16.8% to variation between patients within centers, and 37.9% to variation between participating centers. In the fully elaborated model, level 3 (center-level) factors that were significantly associated with proportion of time in the intended range were upper limit of the intended range, width of the intended range, and presence of a policy of tight oximeter alarm settings (Table 4). Each increment of 1 percentage point in the target range upper limit was associated with an increase of 4 percentage points in target range compliance. Each increment of 1 percentage point in the target range width was associated with a 1.5-point increase in target range compliance. Gestational age was the only significant level 2 (infant-level) factor that was associated with proportion of monitored time in the intended range. The effect of gestational age was modified significantly by width of the target range. A narrow target range was associated with lower target range compliance at all gestational ages. A wider target range was associated with higher compliance at low gestational ages, but compliance decreased as gestational age advanced. The main effect of gestational age and the modifying effect of width of target range on gestational age both varied significantly between centers. After adjustment for other factors in this model, the presence of a policy of tight oximeter alarm settings was associated with an improvement in compliance with the target range of 7 percentage points. Adjustment of the model for these factors accounted for 42.2% of observed overall variation in successful maintenance of the intended range, including 60.7% of variation between patients within centers and 83% of variation between centers. None of the level 1 (monitoring period) characteristics tested was significant. The hierarchical model accounted for only 1% of variation in compliance with the target range between monitoring periods within patients.
Little information is available to document how oxygen saturation policies translate into clinical practice during neonatal intensive care. This study found that during routine care of infants who were <28 weeks' gestation and on respiratory support in the neonatal period, successful maintenance of the intended saturation range varied widely between participating centers, between patients within centers, and for individual patients over time. Overall, study infants were outside the range that was intended by their caregivers more than half of the time. Successful maintenance of the intended range was associated with wider intended Spo2 ranges, with the presence of a policy that set oximeter alarms close to the limits of the intended Spo2 range, and with higher target range upper limits. Target range compliance decreased with advancing gestational age at birth regardless of chronological age, but this effect was significantly less pronounced when the target range was narrow. After adjustment for these factors, chronological age, mechanical ventilator status, pulmonary acuity score, birth weight, and illness severity at birth were not significantly associated with target range compliance.
Wide target ranges by definition are easier to maintain than narrow ones. Adjustment of this analysis for target range width quantified this effect and allowed more accurate definition of the relationship between target range compliance and other variables. The association of improved compliance with higher target range upper limits may reflect preferences of bedside personnel, unintentional overtreatment of low Spo2 values, or the decreasing slope of the oxygen-hemoglobin saturation curve at its upper plateau. These results do not suggest that wide target ranges or high target range upper limits are preferable as treatment goals for extremely premature infants. It is plausible that setting oximeter alarms close to the target range limits may promote better compliance with the target range, but in the absence of a proven “best” target range, caregivers must weigh the benefits of improved compliance against the potential for staff dissatisfaction with frequent alarms.
Hierarchical modeling was necessary for this analysis because of the nature of the data, in which values that summarized proportion of monitored time in the intended range for each monitoring period were clustered within infants and infants were clustered within centers. Such data do not meet the requirement for independent sampling assumed by nonhierarchical statistical methods. The 3-level model presented here accounted for the majority of variation between centers and between patients within centers. However, the largest source of variation in the proportion of monitored time in the intended range was between monitoring periods within patients. Such variation is caused by factors that change over time. None of the time-dependent monitoring period characteristics tested was significant after adjustment for the factors in Table 4, including mechanical ventilator status, pulmonary acuity score, chronological age, and postmenstrual age. Other time-dependent factors must account for this variation, such as changes in pulmonary status not captured by mechanical ventilator status or pulmonary acuity score or differences in individual target Spo2 ranges among staff who provide bedside care.
Two recent single-center studies documented compliance with the intended range that was comparable to that reported here. A study of 10 extremely low birth weight infants that used tight oximeter alarm settings and a prestudy educational program found 62% compliance with a target Spo2 range of 80% to 92% and noted variation among individual nurses in successful achievement of the intended range.15 Laptook et al16 documented 60% compliance over a 24-hour period in infants with birth weight <1250 g before and after changing the intended target range from 90% to 95% to 88% to 94%. The policy change, accompanied by an educational program and tightening of oximeter alarm upper limits, was associated with an increase in time with Spo2 <80%, with no increase in overall compliance with targeted Spo2 range.
Because masked oximeters were used to estimate time in the target range, systematic differences between the routine care and SET oximeters could result in under- or overestimation of success with maintaining the intended range. Modest differences between oximeters in measured Spo2 are to be expected, with such error occurring in both directions.17 Overall agreement was high between documented Spo2 as recorded in the medical chart by bedside care staff, data that were available from routine care oximeters, and Spo2 values that were obtained from the study SET oximeters, but varied from center to center. These results suggest that this study accurately reflects the general degree of compliance with the intended range among study centers. However, it is possible that individual centers achieved their intended target Spo2 range with a higher or lower rate of success than reported here.
A limitation of this study is the small number of infants enrolled. Therefore, the results may not be representative of compliance with the intended range at all participating centers. Furthermore, the centers that participated in the study may not be representative of all nurseries that care for infants who are <28 weeks' gestation. Intervention bias was reduced by inactivating the Spo2 display on the study oximeters and by not providing an educational program before the study, feedback during monitoring other than routine oximeter care alarms, or feedback regarding performance after monitoring. However, it is not possible to assess the influence, if any, that awareness that a study was in progress may have had on compliance with the intended range compared with routine clinical care conditions before or after the study. Additional investigation is warranted regarding center and clinical characteristics that determine successful maintenance of intended Spo2.
Important questions remain unanswered regarding targeted oxygen saturation and outcome in premature infants. First, what target saturation range is optimal for promoting growth and neurologic development while minimizing risk for ROP, BPD, and other adverse outcomes that are influenced by oxygen therapy? Randomized, controlled trials that enroll thousands of infants will be necessary to answer this fundamentally important question.18 Second, assuming that an optimal target range is identified, what degree of compliance is necessary to achieve maximum benefit? The relationship between outcome and degree of compliance with the target range has received little attention. Finally, what degree of compliance is feasible? This study suggests that a practice intervention that moves median Spo2 away from the target range upper limit might improve compliance at centers where practice currently favors the upper end of the target Spo2 range. However, closed-loop servocontrol systems that link Spo2 feedback to provision of Fio2 may offer the most promising avenue for improvement of target range compliance.19–21 In 12 infants who were <34 weeks' gestational age at birth and receiving nasal continuous positive airway pressure and supplemental oxygen, Urschitz et al19 documented median 90.5% compliance with a target range of 87% to 96% when monitored for 90 minutes with Fio2 administration linked to Spo2 with an automated mechanism. In 14 infants who weighed <1500 g and were monitored for 2 hours, Claure et al21 reported 75% compliance with a target Spo2 range of 88% to 96% using an automated mechanism, compared with 66% compliance that was achieved by a dedicated research nurse who adjusted Fio2 manually. The authors noted that staff who were engaged in bedside care likely would not be able to replicate this success. The results of the current study suggest that additional development of such systems may be necessary if optimal outcomes depend on high compliance with a target range.
This study documented achieved versus intended pulse oximetry saturation levels during the first 4 weeks of life in infants who were born at <28 weeks' gestation and reflects contemporary oxygenation policy and management. These results provide a context in which to interpret recent studies which suggest that oxygen saturation policies are related to the incidence and severity of ROP and BPD. Additional research is needed to assess the benefits and risks of different target oxygen saturation ranges and the effects of varying compliance with the target range on important short-term and long-term outcomes of very premature infants.
This study was funded by the Society for Pediatric Research Student Research Program; Fight for Sight/Prevent Blindness America; The Tufts-New England Medical Center Research Fund; General Clinical Research Center (GCRC)/National Center for Research Resources M01-RR00054, GCRC 5 M01 RR00044, GCRC 5 M01 RR000334; NEI K23 EY/HD00420; and the Arline Silfberg Trust. Oximeters were provided by Masimo Corp.
The AVIOx Study Group included the following: Johns Hopkins School of Medicine (Baltimore, MD): Pamela K. Donohue, Jennifer A. Shepard; University of Alabama (Birmingham, AL): Wally Carlo, Monica Collins; University of Massachusetts Medical Center (Worcester, MA): Jennifer Rylander, Stephen Bean, Francis Bednarek, Tara Loiseau; St Elizabeth's Medical Center (Boston, MA): Gopal Gupta; University of Rochester (Rochester, NY): Dale L. Phelps, Cassandra Horihan, Erica Burnell; Oregon Health & Science University (Portland, OR): De-Ann M. Pillers, David Wheeler, Sue Escoe; University of New Mexico (Albuquerque, NM): Lu-Ann Papile, Conra Backstrom Lacey; Christchurch School of Medicine (Christchurch, New Zealand): Brian Darlow; Royal Maternity Hospital (Belfast, Northern Ireland, United Kingdom): David G. Sweet, Henry L. Halliday; The University of Tennessee Health Science Center (Memphis, TN): Ajay J. Talati, Sheldon B. Korones; University of North Carolina–Chapel Hill (Chapel Hill, NC): Carl Bose, Courtney Winston, Anna Allen; Baystate Medical Center Children's Hospital (Springfield, MA): Kathleen Meyer, Karen Christianson, Bhavesh L. Shah; Children's Hospital (Boston, MA): Anne Hansen; Boston Medical Center (Boston, MA): Alan Fujii; Tufts-New England Medical Center (Boston, MA): James Hagadorn, Cynthia Cole, Tuyet-Hang Nghiem, Anne Furey, Skyler Greene, Ami Vora, Jennifer Cho, Prashant Shrestha, Ekua Abban; Data Coordinating Center: the Division of Clinical Care Research Resources, Tufts-NEMC: Patricia Hibberd, Christopher Schmid, Sarah Brody-Kaplan, Christine Botelho, Marcia Landa, Marina Weisburd, Caitlin Hurley.
- Accepted May 31, 2006.
- Address correspondence to James I. Hagadorn, MD, MS, Division of Neonatology, Connecticut Children's Medical Center, 282 Washington St, Hartford, CT 06106. E-mail:
This study was presented in part at the annual meeting of the Pediatric Academic Societies; May 3, 2004; San Francisco, CA.
The authors have indicated they have no financial relationships relevant to this article to disclose.
- Saugstad O. Is oxygen more toxic than currently believed? Pediatrics.2001;108 :1203– 1205
- The STOP-ROP Multicenter Study Group. Supplemental Therapeutic Oxygen for Prethreshold Retinopathy Of Prematurity (STOP-ROP), a randomized, controlled trial. I: primary outcomes. Pediatrics.2000;105 :295– 310
- Van Marter LJ, Allred EN, Pagano M, et al. Do clinical markers of barotrauma and oxygen toxicity explain interhospital variation in rates of chronic lung disease? The Neonatology Committee for the Developmental Network. Pediatrics.2000;105 :1194– 1201
- ↵Chow L, Wright K, Sola A. Can changes in clinical practice decrease the incidence of severe retinopathy of prematurity in very low birth weight infants? Pediatrics.2003;111 :339– 345
- ↵Tin W, Milligan D, Pennefather P, Hey E. Pulse oximetry, severe retinopathy, and outcome at one year in babies of less than 28 weeks gestation. Arch Dis Child Fetal Neonatal Ed.2001;84 :F106– F110
- ↵Madan A, Brozanski B, Cole C, Oden N, Cohen G, Phelps D. A pulmonary score for assessing the severity of neonatal chronic lung disease. Pediatrics.2005;115(4) . Available at: www.pediatrics.org/cgi/content/full/115/4/e450
- ↵Raudenbush S, Bryk A. Hierarchical Linear Models: Applications and Data Analysis Methods. 2nd ed. Thousand Oaks, CA: Sage Publications, Inc; 2002
- ↵Bohnhorst B, Peter CS, Poets CF. Detection of hyperoxaemia in neonates: data from three new pulse oximeters. Arch Dis Child Fetal Neonatal Ed.2002;87 :F217– F219
- ↵Cole C, Wright K, Tarnow-Mordi W, Phelps D. Resolving our uncertainty about oxygen therapy. Pediatrics.2003;112(pt 1) :1415– 1419
- ↵Claure N, Gerhardt T, Everett R, Musante G, Herrera C, Bancalari E. Closed-loop controlled inspired oxygen concentration for mechanically ventilated very low birth weight infants with frequent episodes of hypoxemia. Pediatrics.2001;107 :1120– 1124
- Copyright © 2006 by the American Academy of Pediatrics