Robert F. Hart, Ph.D.
Marilyn K. Hart, Ph.D.
This is a case study, using simple tools, that
shows the usefulness of monitoring individual
patient data over time. Here, the blood pressure
of a male patient in his forties was followed
intermittently over a five-and-a-half-year period.
During that time period, two important things
changed: the level of exercise activity and
the stress level of the job.
An excellent initial method of analysis is
the use of an interrupted run chart. The run
chart is much simpler than the control chart
because it requires no assumptions about the
distribution or stability of the data. The run
chart is simply a graphical record of the data
over time. An interrupted run chart is interrupted
at points where there is a known change in the
process or a known interruption in the data
acquisition.
The interrupted run chart for systolic blood
pressure is shown in Figure 1 and for diastolic
in Figure 2. The interruptions to the process
divide the data into four time eras. In the
first three time eras, the patient had only
a moderate exercise level, approximately equivalent
to 0.5 hours of walking per day. In the fourth
era, the patient walked 1.5 hours per day. The
patient had the same job and about the same
stress levels in all except the second era (where
there are only two pairs of measurements). In
the second time era, the patient was temporarily
on a job that he found to be very stressful.
Time eras 1 and 3 are for the "usual"
process with moderate exercise and no particular
job stress. The run charts show the process
to be relatively stable through these two time
eras. One would not be surprised to find higher
than "usual" blood pressures during
the higher stress levels of era 2 and lower
than "usual" blood pressures due to
the increased exercise of era 4.
The systolic run chart identifies the special
nature of eras 2 and 4. The diastolic run chart
shows clearly the special nature of era 2. Although
the diastolic indication of special cause variation
is less clear with era 4, the similarity of
the systolic and diastolic results adds additional
credibility to this interpretation.
The interrupted run charts lead to the conclusion
that the data are stratified by era and the
results cannot reasonably be attributed to chance
-- a "textbook" example of the influences
of stress and exercise on blood pressure.
All too often decisions are made only on the
most recent data point. Instead, the retrospective
analysis shown above would make a good point
of departure for a prospective, or ongoing,
activity to monitor the patient's blood pressure.
Often, ongoing process control efforts use a
run chart on the observations with action limits
beyond which observations would be deemed clinically
important. This approach is simple, straightforward,
and effective.
Figure 1 Interrupted Run Chart on Systolic Blood
Pressure for the Four Eras
Figure 2 Interrupted Run Chart on Diastolic
Blood Pressures for the Four Eras
Acknowledgment: This case study was developed
with the assistance of Richard Stanula of Pleasant
Prairie, Wisconsin, an advocate of statistical
process control and a friend for 20 years.
Reference: Excerpted from Statistical
Process Control for Health Care, by Marilyn
K. Hart, Ph.D. and Robert F. Hart, Ph.D., Duxbury
Press, 2002.
For more information, contact Drs. Robert and
Marilyn Hart at robthart@aol.com.
