CSS to Increase CMS Acute MI Core Measure Compliance

The computerized clinical surveillance system (CSS), known as TheraDoc, is used to increase Centers for Medicare & Medicaid Services (CMS) core measure compliance among patients with acute myocardial infarction (MI). CSS was recently studied in a retrospective, single-cohort study that found that CSS had a high sensitivity but a specificity too low to warrant its recommendation as a replacement for tools currently used to determine CMS core measure compliance [Hosp Pharm. 2014;49:26-31].

Many healthcare organizations utilize CSS to help identify at-risk patients and improve patient care. CSS may also assist organizations in receiving maximum incentive payments by increasing compliance with CMS core measures. Core measures that must be met for patients experiencing acute MI include medications on arrival, timing of interventions, and medications on discharge. TheraDoc is specifically designed to identify patient compliance with the CMS acute MI core measures regarding discharge medications.

Jerrica E. Shuster, PharmD, BCPS, University of Michigan Hospitals and Health Centers (UMHHC), along with her colleague Michael P. Dorsch, PharmD, MS, BCPS, conducted the study in 2011 at UMHHC to evaluate the accuracy and practicality of the TheraDoc system in 962 patients. The primary end points were the sensitivity and specificity of the alert compared with the institution’s current gold standard, a quality assurance (QA) review by a QA employee who checks all documentation concerning a patient’s hospital stay. Secondary end points included identifying any modifiable causes that may negatively impact the system’s sensitivity or specificity.

TheraDoc integrates data from multiple sources in real time, recognizing any patient with a positive troponin I laboratory result and lacking an order for 1 of the required CMS acute MI discharge medications (aspirin, beta-blocker, angiotensin-converting enzyme [ACE] inhibitor or angiotensin-receptor blocker [ARB] and/or antihyperlipidemic medication).

The patient population consisted of any patient having a CMS acute MI alert or an acute MI QA review. Sensitivity was evaluated by comparing all acute MI patients who were deemed noncompliant according to QA review who had a TheraDoc CMS acute MI alert during their hospital stay. Specificity was evaluated by comparing the number of patients having a false-positive CSS alert to the number of acute MI patients not found to be CMS acute MI noncompliant during QA review. The researchers also studied the timing of CSS alerts from the point of troponin I laboratory collection to determine if a change in timing could have any impact on the alert’s sensitivity and specificity.

The results showed that 226 of the 962 patients (23.5%) who had a CSS alert actually experienced an acute MI. Most of the patients (n=875) had a CSS alert showing the absence of either an ACE inhibitor or an ARB, although only 120 patients required them; however, 99 patients would have incurred the alert anyway due to another missing medication. The median time of alert was 1.03±0.67 hours from the time of troponin I laboratory collection.

Study results also showed TheraDoc had a sensitivity of 100% and a specificity of 4.79%.

The researchers found that all patients who were noncompliant according to QA review were also identified by CSS. QA review found 219 patients deemed to be false-positive who were actually compliant with CMS acute MI core measures; 48 of these patients were not issued a CSS alert. An additional 955 false-positives were identified by CSS alert.

The large number of false-positives likely caused the low specificity, according to the researchers. Explanations for the false-positives included positive troponins in patients who did not actually experience acute MI and patients who experienced acute MI but did not require an ACE inhibitor or ARB.

These factors are not likely able to be modified to increase the specificity. It was also not believed that a change in timing of the alert would affect sensitivity and specificity.

The researchers acknowledged that as a retrospective, single-center trial, some of the results may be reflective of specific institution characteristics—a limitation of the study.

“In the end, I realized 2 things: with the high patient care volume and responsibilities experienced by pharmacists, regardless of the negative study results, CSS are almost always necessary and can be a terrific resource because they provide the required high sensitivity. However, CSS cannot replace clinical judgment, and developing a system with simultaneously high specificity can be extremely difficult, as seen with the acute MI core measure CSS,” Dr. Shuster said in an interview with First Report Managed Care.