OBJECTIVE. To reduce hyperoxia in very low birth weight infants who receive supplemental oxygen, the Children’s Mercy Hospital neonatal respiratory quality improvement committee introduced the potentially better practice of oxygen saturation targeting and identified strategies to overcome barriers to implementation of this practice.
METHODS. Using rapid-cycle quality improvement projects, this center adapted an oxygen saturation targeting protocol and tracked hourly oxygen saturation as measured by pulse oximetry in very low birth weight infants who received supplemental oxygen.
RESULTS. The percentage of time in the range of 90% to 94% of oxygen saturation as measured by pulse oximetry increased from 20% to an average of 35% after implementation of the protocol. The percentage of time with oxygen saturation as measured by pulse oximetry >98% dropped from 30% to an average of 5% to 10%.
CONCLUSIONS. A well-planned strategy for implementing oxygen saturation targeting can result in a sustained change in clinical practice as well as change in the culture of the NICU regarding the use of oxygen.
- oxygen saturation targeting
- quality improvement
- very low birth weight
- retinopathy of prematurity
- chronic lung disease
Since the advent of pulse oximetry, the measurement of arterial pulse oxygen saturation level (Spo2) has been used as a noninvasive estimation of the adequacy of oxygenation. At best, pulse oximetry can alert the clinician to a potential range of Pao2 that an infant may be experiencing, because the correlation between Spo2 and Pao2 varies considerably between infants.1–3 The sigmoid shape of the oxygen dissociation curve results in large changes in Pao2 in the upper portion of the curve with minimal changes in oxygen saturation of hemoglobin.2,4 The range of Pao2 values that can occur at a given Spo2 value becomes large at higher saturation values.2,3,5
Earlier reviews of pulse oximetry warn of the difficulty of assessing both hypoxemia and hyperoxemia by pulse oximetry alone and suggest that pulse oximetry should not be used exclusive of Pao2 measurements to manage blood oxygenation.2,5 In most NICUs, pulse oximetry gradually replaced transcutaneous Po2 instrumentation, and arterial blood gases were checked less frequently to avoid the complications from indwelling arterial catheters and arterial punctures. The sensitivity and the specificity of pulse oximeters to detect hyperoxemia varies among models. An upper saturation limit of no more than 95% is required to detect a majority of elevated Pao2 values.4,5 Although most clinicians would agree to the merits of avoiding hyperoxia, the practice of many NICUs has evolved to accept Spo2 of 98% to 100% in infants who receive supplemental oxygen regardless of gestational age or the disease process that precipitated the use of oxygen.6 Whereas most NICUs set a lower alarm limit to provide an auditory alert to impending hypoxia, the habit of many units is to set a high alarm limit of 98% to 100%6 or not to set an upper auditory alarm at all. This evolution in clinical practice with the unquestioned use of pulse oximetry for the assessment of oxygenation has resulted in the desensitization of many caregivers to both the potential presence and the hazard of hyperoxia.
Concern regarding the potential benefit or detriment of hyperoxia resulted in 2 multicenter, controlled, randomized, clinical trials: Supplemental Therapeutic Oxygen for Prethreshold Retinopathy of Prematurity (STOP-ROP) and Benefits of Oxygen Saturation Targeting (BOOST).7,8 In STOP-ROP and BOOST, very low birth weight (VLBW) infants who required supplemental oxygen were randomly assigned to a conventional or high Spo2 target range either at the time of diagnosis of prethreshold retinopathy of prematurity (ROP; an average of 35 weeks’ postmenstrual age)7 or at 32 weeks’ postmenstrual age.8 The STOP-ROP trial findings demonstrated that high saturation ranges (96%–99%) did not significantly decrease the proportion of infants who had at least 1 eye progress to threshold ROP and had deleterious effects on chronic lung disease in some infants with no change in growth or neuromotor development.7 Similarly, the infants who were randomly assigned to high oxygen saturation targeting (95%–98%) in the BOOST trial had no beneficial effect on growth or development. These infants had higher rates of dependence on oxygen at 36 weeks’ postmenstrual age and home oxygen therapy.8
Retrospective and observational studies suggest that controlling oxygen saturation ranges from birth in VLBW infants may beneficially affect the rate of severe ROP.7,9,10 After implementing a strict oxygen saturation protocol beginning from birth, Chow et al10 observed a significant decrease in the rate of severe ROP compared with historical rates of ROP. These studies suggest that maintaining oxygen saturations in a range to avoid hyperoxia may result in improved ROP and pulmonary outcomes.
The Children’s Mercy Hospital neonatal respiratory quality improvement (QI) committee joined with multiple other centers through the Vermont Oxford Network to form The Breathsavers Quality Improvement Exploratory Group, which identified reducing chronic lung disease as a primary objective.11 This center adopted the potentially better practice (PBP) of oxygen saturation targeting by developing a protocol and plan. This article describes this single center’s structured QI project of implementing this PBP and expands on the issue of overcoming obstacles to acceptance. The outcome measurement goals were as follows:
100% of caregivers aware of the goal to avoid hyperoxia in VLBW infants who require supplemental oxygen
Saturations in the target range of 90% to 94% >50% of the time
Saturations <98% >90% of the time
This QI project was reviewed by the Children’s Mercy Hospitals and Clinics Pediatric Institutional Review Board before submission for publication. A structured QI method, plan-do-study-act,12 was applied to the introduction and subsequent revision of an oxygen saturation protocol. The following steps were followed: assessment of the problem, literature review, staff assessment, protocol development and tracking tools, education of nursery staff, feedback and frequently asked questions (FAQs), and data collection and use (Table 1). All infants who had birth weights <1500 g and were patients in this center’s NICU after May 1, 2003, and receiving supplemental oxygen were monitored prospectively until 100 days of age to monitor the introduction of this protocol as outlined in Appendix 1. Statistics were performed using a simple linear regression model looking for trends of the percentage of time above, within, and below the target oxygen saturation ranges.
Barriers to implementation of oxygen saturation targeting were identified and targeted (Table 2, Appendix 2). Each week, an average of 8 (range: 5–17) VLBW infants received supplemental oxygen during the first 66 weeks after the protocol was introduced. Before initiation of this project, VLBW infants who received supplemental oxygen were in the Spo2 range of 90% to 94% only 20% of the time. Since implementation of the protocol, the percentage of time in the Spo2 range of 90% to 94% increased to >40% (r2 = 0.22, P < .001). The percentage of time within a wider range of 85% to 94% increased from 22% before intervention to >55% (r2 = 0.22, P < .001; Fig 1A).
The overall percentage of time with Spo2 >94% has dropped from 78% to <40% (r2 = 0.21, P = .001). Tracking the percentage of time spent with Spo2 >94% alone overlooked improvement in decreased time spent at higher saturations. Before intervention, VLBW infants who received supplemental oxygen had Spo2 values of 98% to 100% almost 30% of time. Since implementation of this protocol, the percentage of time with Spo2 of 98% to 100% decreased to ∼5% to 10%, a significant negative trend (r2 = 0.36, P = .001; Fig 1B).
Although the first 6 weeks after implementation of the protocol showed a sustained improvement in decreasing the time spent with Spo2 >94% (Fig 1B), the percentage of time in the goal range and within the alarm limits gradually started to decrease (Fig 1A). Twelve weeks after the start of the protocol, the committee presented an educational program outlining the importance of oxygen saturation targeting, reviewed the data collected to date, and addressed questions. After this second educational effort, the percentage of time within the target range and alarm limit range improved.
Implementation of the protocol raised the concern of a potential increase of hypoxic episodes. Before intervention, VLBW infants who received supplemental oxygen spent 1% of time with Spo2 <85%, and only occasionally did these infants desaturate to Spo2 <70%. Since beginning this protocol, the amount of time with Spo2 <85% increased, varying from 5% to 13% of the time; however, this is not a statistically significant trend (r2 = 0.04, P = .12). No significant trend was noted in the percentage of time with Spo2 <70% (r2 = 0.003, P = .64).
Alternating episodes of hyperoxia and hypoxia adversely affect vascular tone and may play a role in the development of ROP.10 The saturation protocol was revised to limit severe desaturation events, defined as a sustained Spo2 <70% for >1 alarm cycle on the pulse oximeter (3 minutes’ duration), while avoiding rebound hyperoxia. The baseline fraction of inspired oxygen (Fio2) was doubled to bring the Spo2 to 85% and then aggressively weaned to within 3% of the baseline Fio2. These data also were presented to the nursing staff at the second educational conference to allay concerns about more frequent desaturation events.
Data from individual patients were presented weekly to bedside caregivers to facilitate adherence to and discussion of necessary deviation from the protocol goal and alarm saturation ranges. Although implementation of the protocol helped to bring saturations closer to the goal range (Fig 2 A and B), some patients did not tolerate saturations in the low 90s. Bringing these infants into the goal range precipitated frequent and severe desaturations (Fig 2C). These infants’ goal and alarm ranges were modified upward to avoid these events.
In response to uncertainty about which infants should follow the oxygen saturation targeting protocol, cards that identify the correct Spo2 target range and alarm limits were posted on the bedside monitor of each VLBW infant who was receiving supplemental oxygen (concept adapted from Ochsner Clinic Foundation Hospital). The nursery was surveyed regularly to identify the percentage of VLBW infants who were receiving supplemental oxygen and had these cards displayed accurately. These data were used as a measure of awareness of hyperoxia and oxygen saturation targeting. During the first 6 weeks of monitoring card placement, the percentage of correctly placed cards varied from 33% to 78%. Over time, the correct placement of these cards occurred without daily monitoring and reminders and increased to 67% to 100%. There is a statistical correlation with improvement over time (r2 = 0.15, P = .01).
A well-planned strategy for implementing oxygen saturation targeting can result in a sustained change in both clinical practice and culture of the NICU regarding the use of oxygen. Although the optimal oxygen saturation target range to reduce the potential sequelae of hyperoxia for VLBW infants has not been identified through prospective, randomized trials, this center’s neonatal attending staff has accepted the merits of avoiding hyperoxia. The goal of this project was to develop an oxygen saturation targeting protocol that is based on what has been described previously in the literature and to measure the adherence to that protocol, with the emphasis on implementation of evidence rather than its reproduction.
The oxygen saturation targeting protocol as outlined in Appendix 1 was revised periodically over 18 months to overcome barriers that were identified as the process unfolded. Although any NICU that embarks on a change in Spo2 targeting will encounter challenges that are unique to its unit, these guidelines may offer suggestions to improve the likelihood of success.
When oxygen saturation targeting first was identified as a PBP for reducing chronic lung disease, the focus initially was on the Spo2 as the assessment of hyperoxia. In retrospect, a more comprehensive approach could have included strict targeting of Pao2 values in addition to the Spo2 for the duration of ready arterial access. This approach could have the added benefit of closely monitoring correlations between Pao2 and Spo2. The ease of use of pulse oximetry has resulted in a subtle shift in focus from Pao2 to Spo2. A secondary goal has been to educate all caregivers about the limitations of pulse oximetry as the assessment of oxygenation.
The mental model survey13–15 identified specific areas of education to be targeted to implement an oxygen saturation targeting protocol successfully. Conducting this survey as a 1-on-1 interview revealed detailed, useful information for the initial stages of protocol planning. In addition, the nursing review of the protocol by Chow et al10 provided useful feedback that aided in educating nursing staff about how to respond to various scenarios.
The proposed protocol in Appendix 1 was presented to the attending physicians during a regularly scheduled clinical research meeting. The Spo2 target range of 90% to 94% and alarm limits of 85% to 96% were decided at this meeting, although this was not a unanimous decision. A more systematic approach to reaching consensus among attending physicians regarding target saturation range may have resulted in improved support for the protocol. A written survey may have allowed physicians to respond in writing independently to the proposed saturation and alarm ranges. Although consensus-building requires a greater investment of time in the planning stages, a rigorous consensus approach may result in clinical practice guidelines that are better supported, as occurred in this unit’s development of a standard approach to nasal cannula management.16
Detailed tracking of the change in oxygen saturations has been paramount to the success of this project. The Spo2 database for each patient was developed using the nursery’s Quantitative Sentinel charting system. The oxygen saturation value from the beginning of each hour automatically downloaded from the monitor to the Quantitative Sentinel database and was verified by the bedside nurse. Although nursing verification of the number does not completely eliminate the potential of reporting bias, the nursing staff was more likely to verify an automatically downloaded but correct Spo2 that was outside the target range than to enter manually a Spo2 value that was outside the range.
Some caregivers expressed concern that 24 daily data points would not reflect accurately the saturations of any particular infant. Given that an infant is no more likely to be outside the target Spo2 range or alarm limit range at the top of the hour than at any other minute of the hour, hourly data collection captured the overall trend for each infant. The ongoing monitoring of this data allowed targeted education throughout the process and facilitated evaluation of adherence to the protocol.
Approximately 1 month after the initiation of the protocol, the nursing staff were surveyed again to ascertain understanding of the goals of the project. The majority of nursing staff believed that the saturation protocol improved the care of VLBW infants. Resistance to the protocol was encountered when infants had wide fluctuations of Spo2 despite careful adherence to the protocol. Although the perception was that these infants spent the majority of time outside the alarm limits, the data showed otherwise (Figs 1 and 2). The ongoing data collection allowed for presentation of percentage of time in the target range and showed improvements.
The protocol allowed physicians to modify the oxygen saturation target range and alarm limits. The upper alarm limit occasionally was increased to 98% for infants who desaturated quickly on reaching a certain threshold of 88% to 92%. These mostly were extremely low birth weight infants. Physicians were encouraged to return alarm limits to the advised range within the protocol when infant stability allowed. For a brief time, the exception of raising alarm limits for particularly fragile infants became generalized quickly in response to frequent alarms for high saturations. Close monitoring of the data allowed for quick identification and interruption of this trend.
The majority of protocol education targeted nursing and respiratory therapy staff because they provide most of the care for VLBW infants. The committee actively solicited feedback from caregivers in person in addition to that received by e-mail. In response, the committee made a concerted effort to provide timely answers to staff questions regarding implementation of the protocol as illustrated in Appendix 2. In retrospect, more education could have been provided to physicians and nurse practitioners. By not obtaining consensus from all physicians in the development of the protocol, a lack of commitment from some made the implementation of the protocol more difficult for the nursing staff and respiratory therapists. The committee overestimated the commitment to the Spo2 targeting protocol by nurse practitioners and fellows. Because fellows and nurse practitioners are the usual first-line responders to questions regarding the management of VLBW infants, their lack of clear understanding and commitment to the protocol resulted in confusion and conflicting recommendations to bedside nurses.
Review of the data revealed that often infants who received supplemental oxygen via nasal cannula were more difficult to maintain within the target range than those who were on continuous positive airway pressure or endotracheal mechanical ventilation. This observation prompted the development of a nasal cannula protocol to provide consistent directions for use of and weaning from this form of supplemental oxygen.16 The success of any QI project depends on the commitment of both the committee that is developing and overseeing the project and all of caregivers who apply the project. To facilitate success, clinical leaders need to demonstrate support for the project, and caregivers must understand the rationale for the project and be equipped to enact the proposed change.
APPENDIX 1: OXYGEN SATURATION TARGETING PROTOCOL—BREATHSAVERS O2 SATURATION GUIDELINES
Oxygen is a drug. It could be a very dangerous medicine with potentially significant adverse effects in VLBW preterm infants. Avoiding hypoxia is important, but prolonged hyperoxia can lead to oxidative stress and injury. There is no evidence that VLBW infants need to be managed with an Fio2 that leads to Spo2 of 95% to 100%, and these levels are potentially harmful. In addition, repeated episodes of alternating hyperoxia/hypoxia can promote significant alterations in vascular tone in immature infants. By avoiding these episodes, risks to the developing vascular bed in various organ systems could be minimized.
Why implement the Breathsavers O2 saturation guidelines in our NICU?
To lower the incidence of chronic lung disease and possibly lower the incidence of ROP
Which patients are covered by the Breathsavers O2 saturation guidelines?
Patients who are in the NICU and have birth weights <1500 g and are on supplemental O2 unless specifically ordered differently by the doctor/neonatal nurse practitioner (NNP).
What are the Breathsavers O2 saturation guidelines?
Attempt to maintain O2 saturations at 90% to 94% for all patients who have birth weights <1500 g and are on supplemental O2 unless specifically ordered differently by the MD/NNP (eg, cardiac patient, at risk for persistent pulmonary hypertension [PPHN]).
Wean oxygen actively in increments of 1% to 3% to maintain target O2 saturations of 90% to 94% (or ordered target saturation).
Avoid exaggerated decreases in Fio2 that could subsequently lead to hypoxia
Change Fio2 only in small increments
Set saturation monitor alarm limits at
Low 85 and high 96 for all patients who have birth weights <1500 g and are on supplemental O2
Low 85 and high 103 for patients who are on room air
As specifically ordered by the doctor/NNP if other clinical indications (eg, cardiac patients, at risk for PPHN)
Do not change Fio2 frequently up and down to try to maintain target O2 saturations; this can produce dangerous and risky “ups and downs” in infant’s oxygen levels. Change Fio2 only in small increments.
Every increase in Fio2 requires careful assessment of infant and monitor and documentation.
Increase % Fio2 only in increments of 1 to 3 except before procedures or with significant desaturation.
Do not keep increasing Fio2 without notifying MD/NNP; changes in respiratory parameters may be necessary.
Registered nurse (RN)/respiratory therapist (RT) to remain at bedside until return to baseline and saturations have stabilized.
No infant should be left as “stable” if the condition has required an increase of >3% to 5% Fio2.
Inability to maintain O2 saturations of 90% to 94% requires discussion among team members: RN, RT, resident/NNP, and attending/fellow.
How do I implement the Breathsavers O2 saturation guidelines?
Change Fio2 only in small increments.
Routinely increase or decrease Fio2 in increments of 1 to 3
Exception: infant has a history of desaturations with handling and procedures; increase Fio2 5% to 10% before handling/procedures.
Exception: for significant desaturation (persistent Spo2 <70%) increase the Fio2 to double the baseline to bring the Spo2 up quickly to 85%. Once Spo2 ≥85%, wean Fio2 aggressively to within 3% of baseline.
Avoid exaggerated changes in Fio2 (up or down), but avoid prolonged hypoxia
Small incremental changes in % Fio2 probably are safer than large changes (eg, frequent “tweaking” between 40% and 50% is better than periodic changes between 30% and 100%).
Exception: aggressively return Fio2 back to baseline after handling/procedures or significant desaturation spell.
Response to Spo2 alarms
Silence and observe.
At 3 minutes (when alarm sounds again), silence and observe. If saturations are not returning to baseline, then decrease Fio2 by increments of 1% to 3%.
Every 3 minutes, silence and observe. If saturations remain high, then continue to decrease Fio2 by increments of 1% to 3%.
Silence and observe
Evaluate the monitor
(1)Wave form and heart rate: is the saturation reading accurate?
Evaluate the patient
(1)Is the airway patent?
(2)s suctioning needed?
(3)Is repositioning needed?
(4)Is endotracheal tube positioned correctly?
Moderate desaturation spells (Spo2 >70%)
(1)At 3 minutes, if saturation is not increasing, then increase Fio2 by increments of 1 to 3.
(2)Every 3 minutes, if saturation is not increasing, then increase Fio2 by increments of 1% to 3%.
(3)Call doctor/NNP if persistent Fio2 needs of >10% are required to maintain target saturations; changes in respiratory parameters may be needed rather than Fio2.
Significant desaturation spells (Spo2 <70%)
(1)At 3 minutes, if saturation is not increasing, then increase the Fio2 to double the baseline to bring the Spo2 up quickly to 85%.
(2)Once Spo2 ≥85%, aggressively wean Fio2 to within 3% of baseline.
(3)Call doctor/NNP if persistent Fio2 needs of >10% are required to maintain target saturations; Changes in respiratory parameters may be needed rather than Fio2.
Multiple high and low alarms
Silence and observe.
Evaluate the monitor and the patient as above.
Patiently watch wether the saturations will return to target range without changing the Fio2.
Contact the resident/NNP/fellow/attending to discuss
(1)Personalizing the saturation alarm limits
(2)Changes in ventilatory parameters
(3)Nasal continuous positive airway pressure instead of nasal cannula
(4)Hood Fio2 instead of nasal cannula
(5)Changes in nasal cannula flow
(6)Medications (eg, albuterol, caffeine, diuretics, blood)
APPENDIX 2: FAQs FOR THE NICU BREATHSAVERS O2 SATURATION PROTOCOL
FAQ 1: I am taking care of a “twitty” infant. I am chasing his/her saturations up and down because the monitor is alarming all the time either high or low. What should I do?
These infants are very frustrating; evaluate the infant and the monitor for accuracy.
Patiently watch the infant to see whether he or she will recover without changing the Fio2.
Contact the NNP/fellow/resident, and discuss with team (doctor/NNP/RN/RT); do not stay frustrated alone.
May need to consider increasing high saturation alarm limit as needed to keep in target saturation range the majority of the time.
May need other changes: nasal continuous positive airway pressure, nasal cannula flow, ventilator parameters, medications (eg, albuterol, caffeine, blood).
Silence the alarms while watching the infant. Okay to decrease the decibel level of the alarms to minimize the effects to others on the pod while observing the infant.
FAQ 2: I am taking care of a near-term infant (35–38 weeks); he/she may be at risk for PPHN. Should I be following the new O2 saturation guidelines with this infant?
Probably not! The protocol applies to infants with birth weights <1500 g.
Discuss with the doctor/NNP; many near-term and term infants who are on ventilation with respiratory distress/pneumonia/sepsis are at severe risk for PPHN in the first days of life.
In these cases, the risk for PPHN may outweigh the risk for oxygen toxicity and he or she may need to be managed with higher saturations and Po2 levels.
FAQ 3: My infant has a history of desaturations with handling. What should I do with procedures (suctioning, starting an intravenous line, repositioning, or heel sticks)?
It is okay to preoxygenate. Turn up the percentage of Fio2 by 5 to 10 immediately before procedure.
Turn O2 down as soon as possible.
FAQ 4: Should I routinely hand-ventilate my infant (eg, before suctioning or to do an assessment)?
Avoid large changes in Po2 and saturations. Use options other than hand ventilation.
Provide manual breaths from the vent.
Preoxygenate (increase O2 5% to 10%, then turn back down as soon as possible).
Only hand-ventilate when essential.
FAQ 5: Why is the “goal” O2 saturation range different from the O2 saturation monitor alarm settings?
It is impossible to achieve the goal all the time—we recognize this! However, the target O2 saturation range should be the ultimate goal.
The monitor alarm settings are wider than the goal range with the reality that by responding to the alarms, the infant will be in the goal range for an acceptable percentage of the time.
It is okay for the saturations to be lower than the goal but within the monitor alarm range for a period of time. However, if the infant’s saturations consistently are outside the goal saturation range but within the monitor alarm ranges, then changes need to be made.
We thank Marge Ellgen and Jeanette Kinser for hard work in helping us refine the manuscript. We also thank Gay Kurth for ongoing weekly commitment to forwarding data and all of the bedside nurses, RTs, and NNPs for hard work in applying the lessons that we learned. We thank Dr Howard Kilbride for facilitating rigorous quality improvement within the NICU. We also thank the members and facilitators of the Vermont Oxford Network Breathsavers exploratory group for support and willingness to share information.
- Accepted July 18, 2006.
- Address correspondence to Jodi K. Jackson, MD, Children’s Mercy Hospitals and Clinics, 2401 Gillham Rd, Kansas City, MO 64108. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Bucher HU, Fanconi S, Baeckert P, Duc G. Hyperoxemia in newborn infants: detection by pulse oximetry. Pediatrics.1989;84 :226– 230
- ↵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
- ↵Tin W, Milligan DW, 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
- ↵Chow LC, Wright KW, Sola A, CSMC Oxygen Administration Study Group. Can changes in clinical practice decrease the incidence of severe retinopathy of prematurity in very low birth weight infants? Pediatrics.2003;111 :339– 345
- ↵Payne NR, LaCorte M, Karna P, et al. Reduction of bronchopulmonary dysplasia after participation in the Breathsavers Group of the Vermont Oxford Network Neonatal Intensive Care Quality Improvement Collaborative. Pediatrics.2006;118 (5). Available at: www.pediatrics.org/cgi/content/full/118/5/S2/S73
- ↵Horbar JD, Rogowski J, Plsek PE, et al. Collaborative quality improvement for neonatal intensive care. NIC/Q Project Investigators of the Vermont Oxford Network. Pediatrics.2001;107 :14– 22
- ↵Capra F. The Web of Life: The New Scientific Understanding of Living Systems. New York, NY: Anchor Books; 1996
- Zimmerman BJ, Lindberg C, Plsek PE. Edgeware: Insights From Complexity Science for Health Care Leaders. Dallas, TX: VHA Publishing; 1998
- ↵Plsek PE. Redesigning health care with insights from the science of complex adaptive systems. In: IOM Committee on Quality of Health Care in America. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001:309–322
- ↵Jackson JK, Ford SP, Meinert KA, et al. Standardizing nasal cannula oxygen administration in the neonatal intensive care unit. Pediatrics.2006;118 (5). Available at: www.pediatrics.org/cgi/content/full/118/5/S2/S187
- Copyright © 2006 by the American Academy of Pediatrics