Efficient Hospitals

An example of how not to pay hospitals is illustrated by the work of the LDS Hospital in Salt Lake City.(1) The doctors there decided to take steps to reduce their rate of hospital-acquired pneumonia. It turned out to be a big project. First the culture of the intensive care units had to be changed. That is, everyone had to agree to the objective. Then the doctors had to agree to adhere to a standard way of managing patients. Finally, if adherence to the agreed-upon protocols was not monitored, practice quickly shifted back to baseline conditions where each doctor managed patients differently.

The doctors and nurses discovered that to reduce pneumonia it was also necessary to change the way they managed patients. First, they used less sedation in order to keep the patients alert. Second they found that feeding by mouth rather than intravenously reduced pneumonia rate as did early use of antibiotics and measures to prevent stomach ulcers. None of these methods was inherently obvious - they were learned by trial and error.

The management of blood glucose illustrates the rigor of the approach. The normal value for blood glucose is about 100 milligrams/deciliter. In sick patients, even if they do not have diabetes, levels can run up to 200 milligrams/deciliter. In a population of patients, nothing bad seems to happen to any one patient who has high blood glucose. However, experience showed that more of them got pneumonia than otherwise would have (because high blood glucose impairs the functioning of the immune system).

Insulin is used to decrease elevated blood glucose. The reason that all hospitals don’t attempt to keep patients with blood glucoses of 100 is because if the dose of insulin is not carefully followed, it can overshoot resulting in a fall in blood sugar to 40 or 60 mg/dl. This level of blood sugar is called insulin shock and it can permanently injure patients. If a patient is not being very carefully watched, it is safer to tolerate a high glucose level rather than to risk insulin shock.

The nurses in the LDS hospital developed a protocol in which insulin was continuously infused too keep blood glucose in a normal range. The procedure required a lot of nursing attention and frequent measurement.

The result of all this work was to decrease the hospital-acquired pneumonia rate from 12 percent to 3 percent (400 percent decrease). When the hospital examined the finances, its cost was decreased by $5000 per patient. However, despite all the work and money put into the development and maintenance of the treatment protocol, every dollar saved was turned back over to insurers. This story makes the case that in fee-for-service medicine there is no business case for quality.

The work of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) can shed some light on how better to pay hospitals. Figure 1 below shows the expected pneumonia rate after a surgical procedure based upon the patients’ risk profiles (the horizontal red line).(2) The diagonal blue line is composed of over 500 dots, each representing the ratio of the observed hospital-acquired pneumonia rate to the expected rate for an individual hospital.

If, for instance, the expected rate for hospital-acquired pneumonia was 12 percent, the hospitals at the far left of the graph (Good) had a pneumonia rate of 0-1 percent. The hospitals to the far right (Bad) had a rate of 36 percent. There is a four fold (400 percent) difference in the rate of pneumonia among hospitals. The hospital where the blue and red lines cross has exactly the expected rate of pneumonia. Any of us would want to be cared for in a hospital to the left of this one, and not to the right.

Observed/Expected Post-Operative Pneumonia Rates

American College of Surgeons National Surgical Quality Improvement Program

In this instance, a benchmark of quality might be a hospital-acquired pneumonia rate no higher than expected (all the hospitals in the first and second quartiles). One of the deficiencies in our knowledge is that no one is quite sure what the best processes are that allowed the hospitals in the good category to perform so much better than the hospitals in the bad category. That is we don’t know the best processes that resulted in the benchmarks.

An ideal payment system would be to set a payment rate based upon the admitting condition of the patient adjusted for the patient’s risk factors of developing complications. The payment would give hospitals and doctors in the first and second quartiles (those with the lowest pneumonia rates) a comfortable profit margin. The hospitals to the far left with almost no hospital-acquired pneumonia would have a much higher profit margin than the others. Those hospitals in the third and fourth quartiles would either have to improve their performance or they would lose a lot of money. This payment system drives hospitals to meet benchmarks of quality and shares the savings from efficient practice with them.

1. Clemmer TP, Spuhler VJ, Oniki TA, Horn SD. Results of a collaborative quality
   improvement program on outcomes and costs in a tertiary critical care unit. Crit Care Med.
   1999 Sep;27(9):1768-74.
2. National Surgical Quality Improvement Program, American College of Surgeons, Semiannual Report, January 1, 2007-Decmeber 31, 2007, Issued June 22, 2008.