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, University of South Alabama, Mobile, Alabama Steven Zimmerman, Ph.D., VP Research, Biomedical Quality Control of America, Inc., Mobile, Alabama

ABSTRACT

Historically vital sign data were collected and recorded once per hour while the clinical monitor displayed the data from heart-beat to heart-beat. The manner in which the data were presented reduced the value of the information in the data so that caregivers came to believe the data had little value. Using SPC methods to display, the story of the patient's vital signs becomes more obvious and thus enhances the caregiver's understanding of the story as it is told.

INTRODUCTION

The current clinical monitoring standard of practice is to display raw vital sign data and to expect the caregiver to watch the continuous display of numbers to draw conclusions about what the numbers mean. The current clinical system produces some useful results, only because of the hard work and good judgment of highly trained caregivers.

Since raw data display systems do not work well in industry, engineers and statisticians have developed methods for data collection and analysis, known as statistical process control (SPC). The hypothesis is that a statistical process control (SPC) approach will provide the clinical decision-maker with clinical information which provides a foundation for decision making. The main concern is that many caregivers have only seen clinical data as raw numbers and they do not recognize the story that can be revealed by learning to read the SPC control charts.

Engineers, statisticians, and other math experts can prove that vital sign data reacts when selective patient treatments are changed, but cannot tell what a given SPC reaction means in a selected ICU for a patient with a given illness. The engineers know when there is a statistical significant increase in a vital sign, but only a caregiver can determine if the change in the vital sign is good or bad. Good/bad depends on the patient's condition and the cycle of his or her illness.

Currently, because of a lack of experience with SPC caregivers are not prepared to identify the meaning of small, but significant changes in a patient's vital sign. The problem is cyclical; until clinical monitors contain SPC capabilities, physicians cannot learn to use SPC, but monitor companies will not put SPC in monitors until physicians ask for SPC capability; but physicians will not ask if they do not see the value in using SPC.

TELL ME A STORY

All vital signs seem to contain information that should be useful for clinical decision making. Depending on the patient's condition some vital signs provide better information than others. Our experience tells us that caregivers often specify samples of convenience, that is they order the collection of vital sign data that happen to be easy, economical, and safe to collect. Figure 1 illustrates the vital signs: SaO2 and heart rate collected from a patient prior to and after an operation for nasal polyps. The 1| on the bottom of the figure indicates when the operation occurred. All data to the left of the 1| were collected prior to the operation, all data to the right of the 1| were collected after the operation. The data (prior and post the operation) were collected at approximately the same time of day and under approximately the same conditions using a Nellcor 200 pulseoximeter.

Figure 1 Vital signs before and after nasal surgery

The oxygen saturation (SaO2) behavior (story told) in Figure 1 was as expected. Prior to the polyp operation the patient's average oxygen saturation was lower than after the operation (top most graph in Figure 1) and there was more variation prior to the operation than after the operation (after the vertical bold line in the top graph in Figure 1).

The behavior of heart rate in Figure 1 was not as expected. Prior to the operation the patient's heart rate average and variation (measure using standard deviation) were higher than after the operation. In retrospect, the change in heart rate behavior was logical because the removal of the nasal polyps made it easier for the patient to breath and thus the heart did not have to work as hard as it did prior to the operation.

Why were the results unexpected? Simply none of the caregivers had ever thought about the effect of nasal polyps on heart rate. In retrospect, the SPC charts demonstrated that heart rate was a better measure of the effect of the nasal polyp operation than oxygen saturation. In addition, the analysis tells us that the caregiver should take into account the existence of nasal polyps in a patient that has either high blood pressure and/or a heart condition. The vital sign analysis does not tell the caregiver what to do about nasal polyps, only that heart rate can be effected by the presence of nasal polyps. These results suggest that when treating the entire patient for a heart condition, nasal polyps have the potential of making breathing more difficult.

SELECTED VITAL SIGNS

Standard practice is that each caregiver determines the vital sign that should be monitored for any given patient. Often the decision is based on the available monitors as well as the condition of the patient. Among the commonly monitored vital signs are:

Current practice is to judge the significance of vital signs behavior against an absolute standard for all individuals. When using SPC a patient's vital signs current behavior is compared to a base period in the past. The objective of current practice is to determine if vital sign behavior is in a good or bad range while the objective of SPC is to determine if the behavior of vital sign data has changed. Using the current system is a point is beyond the specification limits code blue (emergency) corrective action is needed. Using SPC when a change occurs, the caregiver determines if the change is good or bad and then if any action is required.

MEASURING THE WRONG PARAMETER

The statistician is quick to point out that the data collected may or may not reflect actual changes in the patient and that all statistical methods can only analyze data behavior and not any particular parameter (vital sign). The act of selecting a parameter is always the job of a caregiver and not the statistician. As noted above, vital signs are selected as measures as a function of how easy they are to measure and what devices are available, as well as the physician's preferences.

Most patient vital sign measurements have some degree of discomfort or in some cases a danger of infection. Some data are collected, because it is the standard of practice to do so. Some vital sign data are of no value for clinical decision making. SPC may be used to determine if a particular vital sign reacts to any treatments for a given patient's condition. If a given vital sign does not react to a treatment used for a particular patient condition, then the caregiver should evaluate the significance of measuring the vital sign and whether the costs and discomfort of using the measure can be justified.

POWER OF CHANGE

The primary thing a control chart does is to identify change, to determine when a change in vital sign data behavior occurs. The primary thing the current clinical monitoring system does is to identify when the data are outside some specification limit.

The ability to identify a change at first seems simple and not too significant. Change is simple, but can be extremely significant, in particular, knowing that a statistically significant change has occurred (in real-time) means that corrective action may be taken, if the change is bad. If a change is good, action may be taken to reinforce results. Also if a treatment has been changed and no statistically significant change occurs in a significant data stream, then the treatment may have had no effect.

Often small changes lead to big changes. Small changes can be early warnings of bigger changes to come in the future. In particular when specification limits are wide, small changes in vital sign behavior as identified by a control chart allow the caregiver to view vital signs as they are changing, before the major event happens.

WHY STATISTICS?

Statistics are numbers, which represent some aspect of the entity or process being monitored. Vital sign measurements are numbers that measure the behavior of a patient's bodily functions. Vital signs are statistics.

Using the manual recording method of the current system it is impossible to collect and record all vital sign measurements for a given patient. Current practice is to collect and record one vital sign per hour or per half an hour maximum. Using computerized SPC methods most vital signs for a patient for a given time period can be collected and recorded. Some monitor-computer combinations collect data every heart beat while others may collect an observation once every five seconds. Some long-term data collecting procedures collect one observation every 15 seconds. The data collection rate is critical relative to data behavior, but we do not yet know how important the data collection rate is relative to caregiver decision-making and monitoring.

Data may be collected and recorded during the time when the patient is connected to a monitor that is connected to a net work or computer. This time period is only a sample of the vital signs of a given patient. To obtain 100% coverage, an individual must be connected to a monitor and computer for life. A sample of once per hour or of an hour of heart-beat to heart-beat data is a sample. Samples are best analyzed using statistical procedures for maximum information.

CURRENT SYSTEM

The current system sounds an alarm whenever an observation is beyond specified specification limits. Initially, caregivers may think that they can identify the significance of an alarm when it occurs. This is an invalid assumption because it is often impossible to know the reason for an alarm, until after it occurs several times. There is no way to know if an alarm is:

The vital sign alarm is just one of many factors considered by the caregiver. Usually, to identify the true meaning of an alarm takes research. Sometimes the reason for the alarm can be easily identified. For example, when a monitor lead falls off the patient. The usual case is that an alarm sounds and the caregiver examines the patients, looks at other factors and determines what clinical action should be taken. If the reason for an alarm is unknown, then a common reaction to an alarm is to instruct the local caregivers to keep an extra eye on the situation until the meaning of the alarm can be determine.

VITAL SIGNS TELL A STORY

Vital signs are trying to tell us a story. SPC methods present vital sign data so that we can read the patient's specific story. Vital signs tell us something about how the body is behaving and how it is reacting to treatment. Monitoring is a critical clinical activity.

The current practice of observing and recording vital sign data one or twice an hour results in data that were of marginal value to caregivers. Because of the manner vital signs were presented, many caregivers have come to assume that vital sign data have little value. To learn to read SPC vital sign charts the caregiver must change their attitudes toward vital sign data and learn to read the story that is being told. SPC gives the physician quality real-time information that can result in better clinical decisions.