Site hosted by Angelfire.com: Build your free website today!
Using SPC Searching for the natural patient limits (NPL) for short-term newborns using SaO2 support

Steven Ringer, MD, Ph.D., Director of Newborn Services, Brigham and Women's Hospital, Harvard Medical School, Boston MA
Richard Slavin, Technical Director of Neonatal Respiratory Therapy, Brigham and Women's Hospital, Harvard Medical School, Boston MA
Marjorie L. Icenogle, Ph.D., Assistant Professor of Management, University of South Alabama
Naomi Seiler, Research assistant, Brigham and Women's Hospital, Harvard Medical School, Boston, MA
Steven Zimmerman, Ph.D., Professor of Quality and Systems Management, University of South Alabama

Abstract

Caregivers use a variety of methods to provide oxygen support for short-term newborns including nose-plugs, hoods, vents, and CPAP. Many, caregivers assume that all oxygen delivery systems have the same characteristics. The objective of this study is to examine the different delivery systems relative to their effect on the natural patient limits (NPL), that is its own average, standard deviation (sd), and distribution. Researchers used a standard process control chart approach to identify the underlying stable system of chance causes for each oxygen delivery system. Data were collected for patients using each system along with detail annotations of events. All data associated with assignable causes were eliminated. The final data were examined for stability and a process capability study was performed on the data. After eliminating all causes, each delivery system was found to have its own average, standard deviation, and distribution. None of the systems seemed to satisfy the assumption of normality. A conflict was found between the current practice, the upper specification limit selected, and the output data.

Introduction

Short-term newborns usually have under-developed lungs and require breathing support in terms of SaO2. If the oxygen saturation in a patient's blood falls below 84 percent the patient could die by choking and some physicians feel that if the oxygen saturation in a newborn's blood stays above 97 percent for a period of time in short-term newborns, the patient could experience blindness later in life.

Sample size

A sample of 405 heart beat and 455 oxygen saturation measures was taken for 26 short-term newborns at the Brigham and Women's Hospital during the summers of 1996 and 1997. All caregiver actions were annotated and recorded in computer data files by an observer sitting at bedside during patient treatment. The data recording and observation process was under the direct supervision of the primary physician.

The control chart analysis was performed using software developed by Biomedical Quality Control of America, Inc. Through this step the study was a blind study, because no changes could be made in the software during the data collection process.

The eliminating of data associated with an external causal action as recorded by the observer required judgement. The study was still blind in that the analyst had no idea of how his judgement would effect the final analysis.

Treatment

The primary treatment for under-developed lungs is to provide the patient with oxygen support. Among the oxygen delivery methods are nose-plugs, hoods, vents, and continuous positive air pressure (CPAP). Our hypothesis was that the natural patient limits, that is, the standard deviation for patients supported by each support method vary according to support method.

Steps

The steps in our analysis were:

1) To collect vital sign data along with annotations on clinical actions and patient condition for each treatment group plus patients with respiratory problems on room-air.
2) To eliminate all data associated with external causes in an attempt to find the natural variation of each patient group.
3) To check the researchers' stability assumptions using process control charts.
4) To calculate the standard deviation for each group and the upper natural patient limit (UNPL) and the lower natural patient limit (LNPL) on the assumption that the oxygen saturation data are normally distributed.
5) To test the data to see if it followed the classic bell shape normal distribution pattern.

Steps 1 and 2 were completed with ease. In step 3 we found that there was still a degree of possible casual instability for patients within each treatment group (Figure 1). Since we could not justify the removal of the unstable data we proceeded with our study under this limitation. We calculated the standard deviation and the UNPL and LNPL using 3-sigma limits and the assumption of normality for each treatment group.

Figure 1 Stable Control Chart

Results

The results of this analysis was:

A measure of the SaO2 treatment is the standard deviation. Continuous positive air pressure (CPAP) has the lowest standard deviation, while room-air, that is no treatment had the largest. In our last step we found that the assumption of normality was not valid

Specification Limits

The amount of oxygen saturation in a newborn's blood has both an upper specification (USL) and lower specification limit (LSL). The LSL is in the range of 83 to 86 with 84 being the most popular that we have encountered. If the oxygen saturation falls below the LSL, then there is a danger of choking, brain damage from lack of oxygen, and even death.

Some physicians feel that during development, if the oxygen saturation in a baby's blood around the eyes is "near" 100% then there is a danger of blindness when the child reaches approximately 20 years of age. There is another group of physicians who feel the blindness is due to excessive light. Pulse oximetry devices are noninvasive devices used to indirectly measure oxygen saturation in the blood using a light transmitted by a finger or other type of probe. The device truncates it output to the nearest percent. It outputs a 97, 98, 99, or 100, it cannot output a 97.5! Pulse oximetry is used because it is inexpensive, easy to use, and noninvasive. Our data were collected using pulse oximetry.

Figure 2 illustrates a comparison of the observed data versus a normal distribution with the overall average and standard deviation. Figure 3 illustrates a similar comparison for just the CPAP data. The visual test tells us that the data indicates that neither distribution is normal. For completeness we ran a Kolmogorov-Smirnov (KS) test and obtained the expected results.

Figure 2 All data (total number of observations 405,455)

Figure 3 CPAP data (number of observations 242,620)

The results for non-CPAP patients as a group are illustrated in Figure 4. The distribution of observation has shifted towards a lower oxygen saturation level and the shape of the distribution is approaching a more normal looking distribution. The non-CPAP cases include room-air, vent, and hood patients. The study indicates that more data are needed with additional stratification of the patients according to the oxygen support being given.

Figure 4 Non-CPAP (number of observations 162,835)

The USL used for pulse oximetry varies from 96 to 98 percent with the most common set at 97 percent, which is the normal value for adults. The percent of output at 97 or above was 55 percent for all the data, 31 percent for non-CPAP data, and 71 percent for the CPAP data. There is a problem with the USL of 97. The problem is more critical for the patients using CPAP then for the patients not on CPAP.

Actions

Our analysis tells us that the patients on CPAP have high levels of oxygen saturation (relative to the danger of blindness). The analysis indicated that in order to keep oxygen saturation levels under 97 percent, patients using CPAP must be controlled such that their average oxygen saturation is around 94 percent. Question: can this be done with an increase in the standard deviation? Observation, if the oxygen saturation can be controlled, a process control chart most likely will be needed.

Conclusions

The USL for newborn must consider the type of oxygen support provided the patients. Based on this limited study, the USL for CPAP supported patients is more critical than for the other patients. More data are needed so that the study can be organized to handle each type of oxygen support as an individual issues. Our study is a good first step. The study is limited by:

1. The number of patients selected
2. The amount of data available
3. The lack of 100% stable data or an explanation of the results obtain, the failure of the data to test normal (for the CPAP patients)
4. The lack of precision in our SaO2 measurement (to a whole number only).

We feel the study is a good first step that points the way to additional research.

References

1. Klein, Sanford L, A Glossary of Anesthesia and Related Terminology, Medical Examination Publishing CO., Inc. 1984
2. Laffel, Glenn, Robert Luttman, and Steven M. Zimmerman "Using Control Charts to Analyze Serial Patient-Related Data", Quality Management in Health Care, 1994 2(1), p.70-77 Volume 3 Number 1 Fall 1994.
3. Plsek (1992), "Introduction to control charts," Quality Manage Health Case. p.65-73.
4. Zimmerman, Steven M, and Steven Ringer, "Issues in Clinical Monitoring," Computers in Industrial Engineering Vol. 31 No ½, pp 451-454, 1996.
5. Zimmerman, Steven M., Robert N. Zimmerman, Lonnie D. Brown, and Shannon S. Brown, (1992) "Using Moving Average Process Control Charts in Biomedical Applications," Proceedings- Ninth International Conference of the Israel Society of Quality Assurance, 1992, November 1992, p.761-764.
6. Zimmerman, Steven M., Lonnie D. Brown, Shannon S. Brown, and Richard L Goldhamer, M.D. (1990), "Quality Control Charts for Patient Data." The 8th International Conference of Israel Society for Quality Assurance Transactions November 26-29, 1990 Jerusalem.
7. Zimmerman, Steven M., Lonnie Brown, Shannon Brown, and Robert N. Zimmerman (1992), "Using the Theory of Runs in a Biomedical Application," 46th Annual Quality Control Congress Transactions May 18-20, p.903-908.

Home
Articles about BioMedQC material
Dept-Z@BioMedQC.com