In 1999, as SETMA moved from the pursuit of electronic medical records to the pursuit of electronic patient records (for more on this see Your Life Your Health at www.setma.com the section on "Medical Records"), we began leveraging the power of electronics to improve pateint care.  Over a ten-year period, we saw our performance on diabetes care improve every year.

As SETMA has analyzed our treatment of patients with diabetes, we have generated the following data from our electronic medical record. This data represents treatment of diabetes over a period of eight years and shows a consistent improvement in the treatment of diabetes.

As SETMA has analyzed our treatment of patients with diabetes, we have generated the following data from our electronic medical record. This data represents treatment of diabetes over a period of eight years and shows a consistent improvement in the treatment of diabetes.

• Between 2000 and 2001, there was a significant improvement in the Hgb A1Cs.
• Another improvement is seen between 2004 and 2005.
• And, another improvement is seen between 2006 and 2007.

If we are going to learn from the past, we need to know if something intentional happened which would account for this improvement in Hgb A1C and which would account for that improvement being consistently sustained, as there has been an uninterrupted improvement in our diabetes care over the past 8 years. To discover what could have caused this improvement in diabetes care requires a review of SETMA's history. As we examined these results, we realized that in 2000, we developed our disease management tool. In 2003-2004, we introduced our ADA approved DSME program and in 2006, we were successful in recruiting an endocrinologist.  Going forward, our strategy is to increase the use of the disease management took, increase the use of diabetes education and to increase the use of endocrinology for difficult and recalcitrant disease.

As we developed more disease management tools, we realized that there were three life-style changes we wanted everyone of our patients to practice.  We wanted them to lose weight, exercise and stop smoking, so we developed the LESS Initiative.  In this program (for more see Your Life Your Health on our website), for eight years, we have assessed every patients’ weight with a BMI, BMR, body fat and an explanation of who much weight they need to lose in order too improve their health. We give every patient a written personalized exercise program and we address active smoking, secondary smoking and now tertiary smoking with every patient we see. 

These and other screening and prevention programs for diabetes and hypertension, enable us to provide primary and secondary disease prevention for our patients.

The following is a further example of our analysis of our diabetes treatment plan:

Mean, Standard Deviation, Median, Mode

As we looked at the data and tried to draw conclusions about it, we realized that we needed more statistical analysis than just the average (the mean). We need to know the median, the mode and the standard deviation and we needed to know them by provider.

• For a data set, such as the HgbA1C values, the mean (average) is the sum of the observations divided by the number of observations. The mean is often quoted along with the standard deviation. In that case, the mean describes the central location of the data (often called the average) and the standard deviation describes the spread. The mean may be 6.5% in the case of Hgb A1C which is excellent, but if the standard deviation is 1.6, the range would be from 4.0 to 8.1. The 8.1% is not good.
• A median is described as the number separating the higher half of a sample from the lower half. At most half the population has values less than the median and at most half have values greater than the median.
• The mode is the value that occurs the most frequently in a data set.

The analysis by provider in SETMA's treatment of diabetes showed the following. (The provider names have been removed.) As is often the case the worst numbers were found in the case of the best physicians because they see the sickest patients. As you analyze data, you begin to be able to devise a plan for future efforts at improvement of care.

• The age of the patients - younger patients tend to have better control
• The activity of the patients - older patients tend to be more sedentary
• The nutrition of the patients - nursing home patients and elderly often are under-nourished and will thusly skew the HgbA1Cs downward.
• Socio-economic status - patients with lower incomes have more difficulty eating right.
• Educational status of patients - often people with higher education are more motivated and better able to understand the complexities of DSME.
• How long the patient has been a diabetic can influence the HgbA1C.
• How long the patient has been cared for by our clinic. It would appear - and we shall examine this - that the longer a patient sees us, the better their HgbA1Cs will be.
• There are other factors in the care of a patient with diabetes which have equal and possibly superior important to the HgbA1C, i.e, blood pressure, etc. We will be looking at those factors.

Analysis and the future - plans for a 2008 breakpoint

As you analyze the data above, remember that the higher the median (the higher the value which represents the midpoint of your data set, i.e., 50% of our values are above the median and 50% are below), the higher your mean (average) will be and in general the higher your standard deviation will be.

For analysis purposes remember that if your standard deviation is ZERO, the mean (average), the median and the mode will be the same. The problem with the mean (average) as a standard of excellence is that many patients will still be experiencing end-organ damage, even though your mean (average) may be below the ADA target of 7.0. The goal is to lower the standard deviation, the mode and the median which will be reflected in an improved mean (average). With the treatment of diabetes, as with any other biological-system-based data-set, it will be impossible to have a zero standard deviation but the result of improving the care for each individual will be the decreasing and improvement of your standard deviation.

In planning for the creation of a new breakpoint in 2009, we believe that our improvement in the care of diabetes will come as:

• Our nurses initiate the utilization of the Diabetes Disease Management Tool - remember the first break point in the improvement of diabetes care in SETMA was the development and use of the disease management tool in NextGen. Our next breakpoint will be a combination of things including the revitalization of our use of the disease management tool for diabetes.
• Our healthcare providers use the disease management tool and measure their performance with the Consortium for Physician Performance Improvement data set which is built into our diabetes disease management tool.
• Also, the first thing every provider should review for ANY patient with diabetes is the date and value of their last HgbA1C. That is easily done as both of these data points are displayed on the front page of the Diabetes Disease Management tool along with all other critical indicators for quality improvement in the care of diabetes.
• We continue auditing the above again and publishing that data to all providers so that everyone can compare their performance with their colleagues.
• We query our system and involved all patient not to goal in diabetes education (Diabetes Self Management Education) and in specialty care.

SETMA's goal for 2009 is going to be for all providers to improve their median HgbA1Cs by .30 at a minimum. This new breakpoint will result from the understanding that we have gained from our data analysis. The first three breakpoints happened almost by accident but they will each be a significant part of our next breakpoint.

Remember, the better your data, the better your planning can be and the better your results can be. In 2010, we will report how we did in 2009.