“What we’ve got here is failure to communicate …”
(Cool Hand Luke)
Every three months, the personnel in my office get a little nervous as we send out the quarterly personal e-mails to physicians. These e-mails, titled “Quality Indicator Deficiency,” are generated from the quality department’s abstraction of quality measures. The e-mails are personally sent from me, the chief medical officer, to the attending physician. For each measure abstracted, an attending is assigned responsibility (some like to call it “blame”). Despite the fact that we are a teaching hospital, we focus on the attendings who oversee the residents, but residents are copied on the e-mail. The deficiency assignment is based on the physician deemed most critical to the delivery of the measure.
For example, for a patient with pneumonia admitted through the ED, the ED physician is assigned the correct antibiotics measure. If the patient was directly admitted to a nursing unit, the admitting attending is assigned the correct antibiotic measure. Similarly, the anesthesia attending, not the surgeon, is given the pre-operative antibiotic measure.
Why the anxiety from my office? Because when the letters go out, there usually is a very rapid response by many physicians, which results in significant scrutiny of the letters. I refer to this as “good conflict.” Not everyone in my office always agrees with the good part, but there is agreement that this leads to significant discussion and awareness. Prior to the letters, we were like many organizations: We reported aggregate data regularly in a variety of forums and formats in order to promote awareness. We had regular conversations with physicians, physician leaders, and hospital staff. There were numerous quality-improvement efforts in measure areas. We continue to do all of these things, but it is the letters that have turned it up a notch.
Responses fall into three categories. First, “The data is wrong.” Second, “This is not my fault. It was so-and-so’s fault.” Third, “A hospital process prevented me from doing the right thing.” The responses have proven so predictable that I have developed standard scripting for my responses.
Occasionally, the data are wrong. All letters are sent after the charts are initially abstracted. The accuracy of a chart abstractor is close to 100%. Addition of a nurse reviewer and a physician reviewer will help us achieve perfection, but the cost of doing this is significant. I find that many physicians, including myself, want perfection when it gets personal, but I just can’t justify the expense. Moreover, when a physician states that the data are wrong, it is more likely that the data are correct, but the physician does not fully understand the specifications of the measure.
Fault is a more interesting response. Deliberately, our quality committee, which determines the assignment rules, has chosen to attach an attending to every single measure, even those driven by hospital staff. This is done because we feel the healthcare team needs to solve these deficiencies, and often this forces a discussion. For deficiencies that are more staff-related, we do send the letter to the physician and unit manager simultaneously. The usual response by the manager is to contact the physician, apologize, then work on a plan that will remedy the situation.
As a recent example, a nurse chose to give a medication prior to an antibiotic for pneumonia and thus missed the six-hour deadline. This resulted in a change in protocol in the ED.
The hospital process response is the most useful to me because for every quarter that we send letters, there are multiple hospital processes that are modified as a result of this response. It is always fascinating to me because almost all of these measures have had good quality-improvement teams, but there are always missed subtle issues or unintended consequences. Because physicians respond quickly and often passionately, the urgency to fix these processes is present, hence facilitating rapid change.
Kill Them with Kindness
When I suggested sending these letters a few years ago, many on the quality staff cringed. Their reluctance was not unfounded: My original letter was a tad abrasive. I have toned down the letter, even to the point that I am now asking for help rather than directly pointing the finger: “I need your help in reviewing the following patient’s chart. In an initial review of quality indicators, a deficiency is noted. This may be because of simple oversight, or the patient might have had a contraindication for therapy for which we cannot find documentation. A physician must document all contraindications. Please review this chart as soon as possible, and if you feel this deficiency is in error, please let me or one of the outcomes managers know immediately. This deficiency has not yet been reported to the federal quality-improvement organization. We may be able to correct the error prior to final submission.”
The last line of the response attaches responsibility: “If we do not hear from you, we will assume the deficiency stands.”
This whole plan is part of a campaign in our hospital to personalize quality data. Recently, several studies have shown that consumers do not review quality data. Other evidence is increasingly backing up the “embarrassment” approach, as some would like to label my methodology. I prefer to call it the “personalizing method.” Physician education these days is largely based on case studies and the evidence-based approach. Personalizing quality data combines both, and might be the ideal approach to taking great leaps in quality.
Some experts have lamented this approach, but I can tell you that when I received two letters regarding the omission of offering a patient with heart failure weight-monitoring instructions, I quickly changed my approach and now make sure that the residents and nurses are giving those instructions. Initially, I was a little defensive about it, but I did the right thing and made modifications. Personalizing does work, and we should embrace it more fully.
It is not easy to personalize. Our present systems—or lack thereof—require abstracting and the review of written notes to achieve high levels of accuracy in personalizing. Without these high levels of accuracy, it is difficult to engage physicians with this data. The future is in electronic capture, but none of this is perfect, either. Hospitals must develop and hone systems to catch physician assignment during multiple interventions of a typical patient stay.
The cost of all this is unknown and most likely significant. But one could easily imagine a tremendous improvement in quality.
My message to hospitalists and hospitalist leaders is “Make the data personal!” Start working on this today, so that within a few years, you, too, will have regular reports. Continue to aggregate quality data based on group performance and review it frequently. If you want more significant impact and greater physician engagement, make it personal. TH
Dr. Cawley is president of SHM.