ARTICLE IN BRIEF
Based on a decision-tree analysis, investigators speculated that despite increased access to tissue plasminogen activator (tPA) in the hospital, hospitalization is not cost-eff ective compared with the same-day referral to a clinic for evaluation following TIA.
Hypothetically, what would be the most cost-effective option for a patient who has had a transient ischemic attack (TIA) — to be hospitalized for 48 hours or discharged to an urgent-care specialty clinic?
That question was the focus of a novel analysis, which used a decision-tree modeling technique to simulate both clinical decisions in a hypothetical group of 1000 patients — 60- to 74-years-old — and predict what happens next. For each decision and event, from TIA to the possibility of death, the investigators computed and assumed possible outcomes based on data from the most current literature. [See “Assumptions for the Decision Tree” for more on the factors they considered in their analysis.}
The Rush University investigators concluded in a report in the Nov. 16 online edition of Neurology that despite increased access to tissue plasminogen activator (tPA) in the hospital, hospitalization is not cost-effective compared with the same-day referral to a clinic for evaluation following TIA.
Assuming that the patients had a 48-hour stroke risk of 0.6 percent, the researchers found that hospitalization would result in an additional .00026 quality-adjusted life years (QALY) at an additional cost of $5,573, for a price tag of over 21.4 million dollars per QALY.
The 0.6 percent stroke risk was derived from a 2007 study in Lancet Neurology that found that a substantial lowering of risk occurs with urgent evaluation and management of TIA, steps researchers assumed would be taken in both scenarios — at the hospital and the urgent-care clinic.
Higher stroke risks, up to 30 percent, were also run through the model — with similar results. At one year, hospitalization was not found to be cost-effective even when the 48-hour stroke risk was assumed to be 30 percent. That was true even when the acceptability threshold was set at more lenient standard of $100,000 per QALY.
Researchers assumed a much higher usage rate of tPA in the hospital (53 percent) compared with the community clinic (8 percent). They also assumed that each use of tPA saved $967 after one year and $6,672 after 30 years, based on figures in the literature.
The cost of hospitalization was assumed to be $3,052 a day, or $6,104 for the two-day stay. The cost of care at the clinic —including a CT scan of the head and carotid doppler imaging — was assumed to be $528.
The two-day hospital stay assumption was based on a review of 190 consecutive TIA admissions at Rush University from 2006 to 2009.
“In the final analysis, we concluded that most patients could be evaluated as outpatients, assuming that you can do it that day, or within 24 hours of the event,” said the senior study author Shyam Prabhakaran, MD, associate professor of neurology at Rush University Medical Center.
“In the hypothesized scenario where both options are available, a same-day clinic is far more cost-effective than hospitalization. Since early initiation of secondary prevention strategies prevents many strokes, the benefit of admission — the opportunity for thrombolysis if strokes occurred following TIA — would seem to apply to very few patients.”
Standards of care for TIA patients have reduced the risk of stroke, making it much more cost-effective to treat patients in the outpatient setting, Dr. Prabhakaran said.
“SJust by doing some basic interventions —— like starting appropriate antithrombotic therapy, controlling blood pressure, controlling cholesterol, and evaluating for carotid stenosis — the subsequent risk of strokes actually quite low,” he said.
He acknowledged that for some patients with specific risk factors or findings, such as high-grade stenosis, hospitalization may be a cost-effective method of treatment.
And he said that more finely honed ways of assessing risk — including new imaging tools and the ABCD2 score, which uses age, blood pressure, clinical features, duration of TIA symptoms, and diabetes as a simple way to gauge risk — may help lead to more tailored decisions about the best treatment option for patients.
Absent the widespread availability of 24-hour TIA clinics, one option now growing in popularity are observation units within emergency departments, Dr. Prabhakaran said.
“So there may be an intermediate hybrid that is feasible in low-resource areas and still less costly than admitting patients for two days,” he said.
Establishing protocols other than hospitalization for TIA patients is “actually more doable than we think it is,” he said.
In a study published earlier this year in Stroke, patients were assigned to either hospitalization or a TIA triage unit within the emergency room, based on their stroke risk, and their combined stroke rate was just 0.9 percent at 90 days.
Confronting health care costs is only going to grow in importance, he said.
“In the next decade or two, when curbing costs are likely to become a priority, we do have to be better at telling patients that costlier does not always mean better and that appropriate decision-making that avoids unnecessary costs is going to benefit everyone in the long run.
The results notwithstanding, the study authors acknowledged that their analysis requires real-life validation. “Prospective observational studies and ultimately randomized clinical trials comparing urgent outpatient versus inpatient evaluation for TIA patients, are sorely needed,” the study authors wrote. “Until such comparative studies, our model suggests that routine hospitalization of all TIA patients is cost-prohibitive.”
Gregory Albers, MD, director of the Stanford Stroke Center and Coyote Foundation Professor of Neurology and Neurological Sciences, said the value of the paper is that it points out how, in so many TIA cases, it makes little sense to hospitalize.
But he said stratifying the risks of patients is crucial — a point that the paper's overall conclusion might have minimized.
“If somebody were to read this article and say nobody should be hospitalized because it's not [cost- effective is missing the fact that we now can do much more to clarify the prognosis of the patient,” Dr. Albers said.
Treating TIA patients is much more complicated than assessing cost-benefits, he said. Recently published National Stroke Association recommendations, which Dr. Albers helped write, caution that outpatient care can sometimes be suboptimal, and that the patients most appropriate for such care are low-risk patients.
“It doesn't make sense to treat all TIA patients the same,” he said. “When you have a group of patients who have a stroke risk maybe on the order of one in 200 or less over the next week, versus a group where you've got a 15 percent stroke risk, there are issues beyond cost-effectiveness. Imagine if you had a TIA on a Friday and you could be identified as someone who has a 15 percent chance of having a stroke in the next couple of days, and somebody sends you home and you have your stroke on Saturday or Sunday. You would be quite unhappy, whether it was cost-effective to one of these thresholds or not.
“From a quality of care point of view if you have patients and you know their stroke risk is on the order of 10 to 15 percent in the next two days, I think sending that patient home is not an ethical decision. It's not right for the patient. We do all sorts of things that wouldn't stand up to this level of scrutiny in saying that it's going to be cost-effective. But you do it because you're trying to take care of the patient and you don't want somebody to have a bad event at home.”
Philip Gorelick, MD, MPH, director of the Center for Stroke Research at the University of Illinois, said a main challenge is the accessibility of clinics.
“Clinical predictor scores, especially those incorporating neuroimaging findings, may be useful for risk stratification in these patients,”“ he said. “A challenge of providing urgent outpatient TIA diagnosis and treatment is the availability of physicians, clinic staff, and rapid access to necessary diagnostic imaging modalities to rule out etiologies for TIA such as major cardiac sources of embolism or high-grade intra- or extracranial carotid stenosis. Not every facility is equipped to make the financial investment, which may be the case in some general community hospitals, rural settings, and critical-access hospitals. Transfer agreements to hospitals with such vascular-focused resources can overcome the challenge.”
In addition to stroke risk, concerns about adherence, or about socioeconomic factors, might call for hospitalization, he said.
“A relatively high degree of patient adherence may be required for an urgent outpatient TIA clinic program to be successful, and therefore, if adherence or socioeconomic circumstances is an issue, hospitalization may be a better option for rapid diagnosis and treatment of a TIA patient,” Dr. Gorelick said. “The assumptions in the cost-effectiveness analysis here may or may not precisely apply to your practice setting. In any case, rapid diagnosis and treatment of the recent TIA patient is indicated, and exercising good clinical judgment is of utmost importance.”
ASSUMPTIONS FOR THE DECISION TREE
* All patients were accurately diagnosed with TA.
* All patients who had subsequent ischemic strokes. reported to the hospital.
* Strokes were related to the initial TIA.
* No patients had contraindications to tPA, and those who received tPA had it administered intravenously per established standard of care, and within the three-hour time window.
* tPA use would be higher in hospitalized patients if a stroke occurred within 48 hours from initial admission and symptoms were noted within 1 hour of onset.
* All other preventive measures including antithrombotic therapy, carotid revascularization, risk factor modification, and stroke education were similar between hospitalized and clinic patients
* TIA patients who were admitted to the hospital had an average length of stay of 48 hours if no stroke occurred.