August 2008
Feature Story

How to improve the reporting of critical results—that’s what Anand S. Dighe, MD, PhD, spoke about last spring at the Roche SmartLab meeting in Chicago. Dr. Dighe is an assistant professor at Harvard Medical School and director of the core laboratory in the Department of Pathology, Massachusetts General Hospital, Boston. An edited version of his talk follows.

What is our responsibility around critical values, and what is our liability? The duty of a laboratory is to provide the standard of care. The standard of care, however, is a murky concept. The general definition of the standard of care is how similarly qualified staff would manage this patient’s care under the same or similar circumstances. A more concrete definition encompasses what your procedure should be, what other people are doing, and what the regulatory agencies expect.

There are two recent lawsuits that I’m aware of in which violation of the standard of care was alleged. In one, a hospital laboratory defined a glucose of greater than 250 mg/dL as a critical value. Of course, a glucose value of greater than 250 is not that uncommon, such that there will be numerous glucoses reported in a large laboratory if 250 is used as the critical value limit. The laboratory staff realized this as well, and since they were busy, they decided to stop calling glucoses in the range of 250 to 500. They didn’t go to their medical policy committee and change the critical values list; they just decided, ‘We really can’t make all these phone calls.’ The patient in the lawsuit had a glucose of 302. It wasn’t called to the care team. There were many delays in care, and the patient subsequently died of diabetic ketoacidosis. The lawyers for the plaintiff’s family went online and found the lab’s critical values policy, and it clearly stated that a glucose of 250 was a life-threatening critical value and should be called within 30 minutes. The hospital and the lab were sued, and there was a large out-of-court settlement. Whether or not the critical value callback would have averted this tragic outcome is difficult to determine, but clearly each lab’s critical value practice must match its written policy.

A second issue that may give rise to a lawsuit is the lack of an effective escalation policy. That’s a lawsuit waiting to happen in probably all of our laboratories due to the complexity of care, especially outpatient care. In this case, it was a reference lab doing outpatient laboratory testing. A critically elevated INR of 9.0 was resulted at 2:00 AM. The laboratory tried to reach the outpatient clinician. They didn’t have a pager number. The answering service wasn’t helpful. So customer service ended up leaving several urgent messages on the clinician’s answering machine early Friday morning. What they didn’t know was that the hematologist was off on Friday, and so no one was informed of the critical value. The patient took Coumadin the next day and ended up having an intracranial bleed and permanent disability. Bottom line, the laboratory didn’t have an effective escalation policy. The lawsuit was settled out of court for a large sum of money.

George Lundberg, MD, brought the concept of critical values to the attention of the national lab community in the 1970s, and his definition of a critical value is still a fine one: A laboratory result that suggests the patient is in imminent danger unless appropriate therapy is initiated promptly. CLIA ’88 furthered this concept and now critical values are incorporated into the Joint Commission national patient safety goals and the CAP checklists. From the Joint Commission: “Measure, assess, and if appropriate, take action to improve the timeliness of reporting and the timeliness of receipt by the responsible licensed caregiver of critical tests and values.” The CAP checklist item is very similar: “Does the laboratory have procedures for immediate notification of a physician (or other clinical personnel responsible for patient care) when results of certain tests fall within established ‘alert’ or ‘critical’ ranges?”

Step one is formulating a critical values list. Of course we all have had a list for many years, but it’s important to take a hard look at what you’re doing now and compare it to other similar health care facilities. Most have their critical values list online, which facilitates comparison. There has to be a medical staff approval process. Anytime you change a critical value, you need to take that critical value list to some kind of board or committee that represents the medical community at your institution so it can review and approve it.

Non-critical “critical values”—it’s important to get those off your list. We want to have a list in the spirit of life-threatening emergent values. Maybe there is a small gastrointestinal bleed that’s getting worse over the course of several months, and we want to inform the clinician of the slowly changing but not yet critically low hematocrit value, but we don’t need to do it within a 30- or 40-minute time frame. It’s important to replace these with something else, like auto generated courtesy e-mails or alerts in your electronic medical record.

And then there’s the issue of “false-positive“ critical values. No. 1 for everybody is probably potassium. We have 37,000 critical value phone calls a year, and about 8,000 of those are for potassium. Of the 439,000 potassiums we do, 1.8 percent of them are critical. A small percentage, but it ends up being a large number of critical value phone calls. It’s important to pay attention to things like the percentage of critical value potassiums. We were noticing that our outpatients had a high percentage of elevated potassiums. At the same time, I was getting occasional calls and e-mails saying, ‘Your lab’s giving me the wrong result. Every time I redraw, it’s always normal.’ What we found is that a lot of these high potassiums had hemoglobin A1cs done on them. At that time, we were recommending that hemoglobin A1cs be sent on ice. The practices would ice the hemoglobin A1c and then put the green-top tube right next to the ice during transport. Thus, effectively they were icing a green-top, giving us the effect of pseudohyperkalemia. We performed a study and determined that our policy of icing the A1cs was unnecessary. We got rid of that, and immediately we dropped our percentage of potassium critical callbacks back down to baseline.

One key aspect of critical values is who’s going to make the phone call. Traditionally, virtually all the phone calls for critical values were made by the people performing the tests. In some sense, that’s the natural person to make the phone call. It’s the person who is most invested in it, and they may know the critical value is coming if they’re verifying the result or doing a repeat. However, this is probably not the best use of testing personnel. Critical callback is one of these intermittent tasks that has to be done really well. The multitasked person who’s the lynchpin of your lab, generating results and dealing with a machine that’s down or out of range QC—that person is not in the best position to take on the responsibility for communicating life-threatening critical values. So the real question is how can we communicate critical values effectively in a centralized call-center model?

Laboratory call centers were quite rare five or six years ago, but last year we did a survey regarding critical values and almost 20 percent of medical centers now have a centralized call center where laboratory critical values are communicated. Centralized callback takes the burden of these essential yet distracting phone calls away from the performing technologists. It’s important, though, if you’re going to set up a centralized call center, to give the call center staff access to good information. They’ve got to have a callback module, they’ve got to understand exactly how to use it, and they’ve got to have directories and contacts for all your outpatient physicians. And, in my opinion, at least one person in your call center needs to be a technologist. You need to have someone with real clinical judgment to know what to do when you can’t reach someone, to apply your escalation policy in the way you intend it to be applied. This was the problem in that lawsuit with the INR—the call center staff didn’t understand how important an INR of 9.0 is when the patient’s just about to take another dose of Coumadin. A technologist would understand that this is a true medical emergency.

Who receives the phone calls? It’s the lab’s responsibility to ensure that the critical value is received by the responsible licensed caregiver, not to just get it to the person who picked up the phone on the floor or practice. The Joint Commission has clarified this by saying, “If critical values are not communicated directly, the lab must demonstrate that there are no significant additional delays.” In reaching a solution it is important to broaden this discussion to your nursing staff and outpatient clinicians, and really have them invested in the process. Because if you’re not communicating directly to the clinicians, they’re the ones who are going to have to keep this documentation to help your organization to be in compliance. The communication handoff documentation can be accomplished via logbooks on the care unit. We really don’t like logbooks, though. An inspector will focus in on the one row in the logbook that doesn’t have any notation, and then you’re cooked. But we use logbooks because in some situations we are not able to communicate directly to the responsible licensed caregiver. This whole issue is challenging, so I would say avoid logbooks and communicate directly to the caregiver if you can.

A further angle to this is that in recent surveys, the Joint Commission has been insistent about the documentation of communication to the physician. Even though the standard clearly says “the responsible licensed caregiver,” the JC really wants to see evidence of the ultimate communication to the physician. The argument is that the physician is the person who can really change the management of the patient. Both the JC and the CAP are going to be looking for that. So we have to get our clinicians involved in the process of critical value communication to make this work

“Vital values” is generally the term applied to test results that are not life-threatening but can strongly impact the patient, such as a positive culture for tuberculosis. These correspond to the results that may be recommended to be flagged as “orange” or “yellow” critical values. No one would argue that these values shouldn’t be brought to the attention of the clinician, but there are much better ways than a phone call. The real task here is to work with your electronic medical record group. Many of the systems providers use to view laboratory results have features that can bring new results to the provider’s attention. That’s the appropriate place for this, not a phone call. In many cases the vital value can be communicated with your electronic medical record. And that’s why it’s important to know how your clinicians are looking at the laboratory tests. You want to identify where the holes are. If there are still tests that are hard for them to find or you don’t have an EMR, then perhaps a courte­sy callback would be helpful, but these should be the exception, not the rule.

The question becomes, though, how many times do we annoy our providers? If you get an outpatient critical value in our system, you’re potentially going to be alerted in four different ways. We’re going to alert you by calling you about that critically high INR. You’re going to get an outpatient alert in ResultsManager, the electronic medical record that outpatient providers use to look at results. The provider is going to get a ResultsManager-­driven e-mail, and lastly, the result is going to be flagged in the main electronic medical record. It’s important to figure out which of these various communications options are appropriate for critical values and which are appropriate for courtesy callbacks.

The ultimate solution is some way to alert the clinician automatically. The result comes off your analyzer and goes to your LIS, and your system then puts on the clinician’s pager a message such as, ‘You have a critical value, please call the laboratory.’ This process would be relatively straightforward if you knew whom to page. It may not be the person who ordered the test. For inpatients, the person who ordered the test may be the tired intern who went off to bed or who’s on a different service now.

So at our institution we’ve created policies and software to define three key roles: the attending physician, the responsible physician (the staff physician in charge of the overall management of the patient), and the responding clinician (the person who’s available now). The responding clinician is the person we’re going to call or page for a critical value. Our next step is to do this for outpatients. It’s obviously much more complicated in the outpatient arena, because patients flow between different outpatient providers, specialists, and primary care physicians, and the responding clinician, in general, is who ordered the test.

Setting goals and measuring yourself is an essential part of any critical values program. We all have thousands of critical values per year that we have to reliably communicate. The goal is that 100 percent of critical values will be called within X minutes. An informal survey shows that X seems to be anywhere from 30 minutes to an hour. Most people seem to be on the lower end of that. Our goal is 40 minutes in our hospital: The laboratory has 30 minutes, and we permit 10 minutes for communication on the floor. We use this second time period only if we’re not communicating directly to the responsible licensed clinician. For us, when we first looked at where we were several years ago, and we were not happy with this result at all, only 82 percent of the values were called within 30 minutes. Most of them got called within the first few minutes, so the median was really low, making you think you are doing a reasonable job. However, there was this long tail of delayed critical values that was the problem. So what we do now is to measure every day the percentage less than 30 minutes and use both daily and monthly reports as the basis for our process improvements. Our remaining issue is our outpatient critical values. For inpatient and ED patients we’re now near 100 percent, but for outpatients we still have issues. You have to plan for failure. You can’t implement a system that assumes your outpatient physicians are going to be available. That would be a very bad system.

The typical scenario is a critically elevated potassium that gets reported in the evening on an outpatient. The practice is closed, so we start with the ordering physician and work with his or her answering service to find out who the covering physician is if the ordering physician is not available. If that process is not successful, we get the pathology resident on call involved. They’re going to take a look at the medical record and make a judgment. Is this something we need to call the patient about? Can they find another clinician responsible for the patient’s care? Can they say, ‘Okay, we can just send an e-mail to this clinician. This is something they have already acknowledged in today’s note’?

In the escalation policy, we have to have someone in the loop who can make a medical decision. If you’re putting this in the hands of only your lab techs, you’re putting them in a risky situation. They’re having to make a medical judgment. We have to take it to someone who can make a medical decision, and even potentially call the patient at home. We’ve certainly called the patient directly on rare occasions. But that’s something that a laboratory director and/or a resident should do. And if you don’t have a pathology resident, then the lab director—anyone who really has an understanding of critical values—should be available to make the decision of what to do. In some centers these patients are referred to a doctor on call in the emergency room. Every time we have to use our escalation policy and get the resident or lab director involved we file a safety report. These are bad outcomes. Even if we got ahold of someone eventually, the fact that we had to use our escalation policy shows there is a break in the system that needs to be fixed.

The last topics are critical results and critical tests. Two years ago, the Joint Commission surprised many of us by clarifying that “critical results“ and “critical tests“ are two different things. Critical results—also known as critical values—are results that always require rapid communication. So, for example, a high potassium, low platelets, high PTTs. Critical tests (also known as stat tests), on the other hand, are a separate list of tests that always require rapid reporting of results even if normal. Tests such as troponin T for emergency room patients. The requirements for monitoring critical results versus critical tests are different. For critical tests, the JC requests that the monitoring be from test ordering to test resulting. This is challenging for most centers. Test ordering is a difficult thing to monitor. The time of order may not be the same as the time you want the test drawn. It is in a theoretical sense the appropriate thing to monitor, but practically it’s very difficult to get an accurate order time and then to use that order time as the basis for process improvement. The best thing to do is measure all the times you have available—the lab time, the receipt time, the result time, and, if available, the order time.

The critical tests are typically applied primarily to inpatient and ED tests. The good thing about the critical test standard is that they’re leaving it up to the institution to make up its list. There’s a lab I know of that has a single test on its critical tests list. We have 11 on ours. If you don’t have a good system in place to monitor all these times, you had better keep your list pretty simple. We all do a lot of this monitoring already. We all have some sense of our turnaround time. But there’s this extra regulatory requirement, so we really have to think about what we designate as a critical test.

There are always going to be occasional samples that take longer, whether it be repeats, clotted samples, or when an analyzer was down. We want to focus on our outlier percentage. Whenever the outlier percentage does not meet the goal, the lab staff have to file a report, work it up with the supervisor, and make sure we understand why the delay in testing occurred, and we then use these data to plan our process improvements.

At the institutional level it is important to realize as well that we are talking about critical results, not just critical lab results. The institution has to identify other areas where critical values may occur. We have encouraged our colleagues in radiology, echo, EKG and anatomic pathology to develop guidelines. The difference here, of course, is that you have a physician in the loop, so the physician can decide if it’s a critical result or not. But the documentation and monitoring requirements are the same. They’re going to have to document that they identified a particular result as a critical value and that it was communicated to the licensed caregiver.

The last approach to critical values is the utilization approach. Are all these tests really necessary? In many cases, they’re really not. Studies have shown that anywhere from 15 to 50 percent of tests are likely to be redundant. Recently, over-ordering of lab tests was rampant in our ICU. We were observing complete metabolic panels and CBCs with differentials performed daily on patients who were incredibly stable. We added guidelines that restricted our routine labs to a short list and rapidly saw a 35 percent drop in laboratory testing with no adverse impact on patient care.

Critical values are not something that you can afford to do an average job on. You have to ask yourself several crucial questions: Do our policies improve outcomes, and are they medically sound? Are we in agreement with the standard of care? Is all our communication documented? Do we have a written goal for our critical value reporting time? Do we review critical values reported outside the goal? Do we have a written escalation policy? Critical values have the potential for being life-threatening emergencies, and for the sake of our patients, we need to have policies and procedures that are reliable and provide real safety for our patients.