• Data Mining Helps Prevent HAIs—and Saves Hospital $2 Million

    Sep 20, 2011

    By Jim Maher and David Stricklin

    After implementing an automated surveillance system, Robert Wood Johnson University Hospital Hamilton reduced its healthcare-associated infection marker rate by more than 20 percent and achieved a 3.90 ROI.

    Robert Wood Johnson University Hospital Hamilton (RWJ Hamilton)-a 284-bed acute care hospital in Hamilton, N.J.-treats more than 200,000 patients annually. It is among a small number of U.S. hospitals to receive the Malcolm Baldrige National Quality Award.

    In 2006, the hospital's CFO-along with clinical and administrative leadership teams-recognized that reducing healthcare-associated infections (HAIs), or nosocomial infections, was an important clinical-and financial-opportunity for the hospital. The financial cost associated with HAIs (i.e., in additional hospital days, tests, and interventions) is astonishing: from $9,969 for a case of ventilator-associated pneumonia to as high as $36,441 for a bloodstream infection.a

    In addition to being costly, HAI rates are now tied to payment rates. Beginning in FY13, when the Centers for Medicare & Medicaid Services enacts a new Value-Based Purchasing program, Medicare payments to acute care hospitals could be reduced-or increased-based on their performance or improvement on certain quality measures, including HAIs.b In addition, many commercial payers are beginning to link HAI rates to payments via pay-for-performance and other value-based purchasing approaches.

    To identify, prevent, and manage HAIs, hospitals should ideally monitor every patient in every location on a continuous basis. But few hospitals have enough people, staff time, or resources to achieve this ideal state-at least with the use of traditional manual processes.

    The development of automated surveillance and data mining systems in recent years has made the prospect of hospitalwide surveillance of HAIs both realistic and highly effective for hospitals of all sizes. Five years after implementing such a system, RWJ Hamilton has protected hundreds of lives, saved millions of dollars, and achieved an almost fourfold ROI (see table and exhibit below).

    Impact Analysis: RWJ Hamilton Automated Surveillance Data Mining Service

    Baseline period (10/1/05-9/30/06) versus active period (10/1/06-9/30/10)

    • Total NIM rate: 20.59% decrease*
    • Unique NIM rate: 21.90% decrease*
    • Lives protected: 722
    • Length of stay days avoided: 5,313
    • Direct cost savings: $3.1 million
    • Bottom-line savings: $2.1 million
    • ROI: 3.90

    *NIM rate = Nosocomial infection-or hospital-acquired infection (HAI)-marker rate; NIM is a tool that automatically screens for likely HAIs through a clinically validated algorithm that reviews, among other data sources, laboratory results, bacterial susceptibility to antibiotics, and patient history.

    Exhibit: NIM Rate Change by Source  

    How Automated Surveillance Works

    Traditional infection surveillance methods rely on the infection preventionist reviewing clinical data for indicators that point to a patient having acquired an HAI. The efficiency of this process is influenced by a number of factors-both internal and external. The effort required to gather multiple data sets from different hospital IT systems prior to review, combined with the ongoing effort to keep up with rapidly changing state and federal infection reporting requirements, can be daunting.

    Due to resource limitations and reporting mandates, infection prevention teams typically focus on patients who have a high risk of acquiring HAIs, such as patients in intensive care units with certain devices (i.e., central lines and ventilators). Though infections discovered through targeted surveillance have a high cost and a high mortality relative to other infection types, surveillance targeted in an intensive care unit may only identify 20 percent to 30 percent of device-related infections and fewer than 50 percent of infections due to multiple drug-resistant organisms, such as Methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococci.c

    Automated surveillance uses technology to identify and report HAIs. By making more efficient use of limited human resources through automated surveillance systems, the scope of infection surveillance may be expanded. Freed from their clerical burdens, infection preventionists may spend more time interacting with nursing and medical staff in patient care areas, facilitating interventions to correct clinical process breakdowns instead of reacting to the outcomes (i.e., HAIs) of broken processes.

    Using a NIM to Screen for HAIs

    The solution chosen by RWJ Hamilton incorporates the use of a nosocomial infection marker (NIM) in the screening process. The peer-reviewed tool automatically screens for likely HAIs through a clinically validated algorithm that reviews-among other data sources-laboratory results, bacterial susceptibility to antibiotics, and patient history for similar infections. The NIM has been specific in the identification of HAIs d, and its objective, reproducible nature lends itself to clinical and financial analytics.

    RWJ Hamilton's infection prevention team can access a scorecard that allows for efficient review of infection trends in each nursing unit and for any infection type. Trending is also available for any infection type at the unit, facility, and enterprise levels.

    In addition, the clinical coordinators receive weekly line listings of HAI-marker patients to investigate and review. The unit educators also receive this information to identify staff training opportunities by type of HAI. One unit may need to focus on reducing blood-related HAIs, while another unit may need to focus on respiratory-related HAIs. The chief nursing officer (CNO) reviews the unit scorecards and works with clinical coordinators and the training and development department to ensure quality improvement initiatives are coordinated and areas of opportunity are addressed.

    Automated trend analyses are sent to the hospital to support the rapid, proactive identification of emerging trends in infection and infection prevention, such as patient bacterial colonization and specimen contamination, potentially before a breakdown is revealed through an outbreak of infection. Additionally, RWJ Hamilton infection preventionists consult with certified infection preventionists through their automated surveillance vendor on best practices and root cause analytics specific to each discovered trend. This patented early detection system acts as a smoke alarm, providing information to caregivers to address quality issues earlier in the surveillance process and enabling targeted, proactive intervention.

    The NIM tool can also be integrated with hospital cost accounting data and used as a viable indicator of the financial impact of infection prevention activities. Monthly and quarterly reports allow management, senior leadership, and hospital board members to see measurable improvements in infection prevention, length of stay, and cost reductions, as well as significant gains to the hospital's bottom line.

    Improvement Examples

    The automated surveillance system enables clinicians to recognize emerging trends more quickly-and more efficiently focus time and resources on staff education and infection prevention.

    For example, in the first quarter of 2011, RWJ Hamilton's infection prevention team noted an increase in respiratory HAI markers. A task force-which comprised the CNO, vice president of quality, and representatives from nursing, laboratory, respiratory, environmental services, and infection control-was charged with investigating the issue and identifying solutions. The data pointed to an area of concern, and the task force refocused its efforts on education and process improvement. Many times this is the case-the policies and procedures are correct, but sometimes the practice does not adhere.

    The result was a change in practice that not only decreased the number of respiratory HAI markers, but also resulted in a decrease in other types of HAI markers (see the exhibit below).

    Exhibit: Decreases in Unique NIM Rate and Total NIM Rate

    "The timeliness of the electronic data allowed us to react quickly, and the objectiveness of the data provided us with the validation of the improvement for the solutions that were put in place," says Anne Dikon, director of infection control, RWJ Hamilton.

    Automated surveillance is also empowering staff across RWJ Hamilton in improving processes that contribute to HAIs.

    For example, the laboratory used surveillance data to streamline its process for reviewing specimen collection techniques and the criteria used to determine when certain specimens required more complex, expensive work-up. Environmental services was also engaged to review the effectiveness of room cleaning supported by unit-specific trending on the prevalence of organisms, such as C. difficile, that are difficult to eradicate in the patient care environment. Even patient transportation was engaged to review its protocol for moving patients with certain devices in place.

    Healthier Patients, Money Saved

    RWJ Hamilton has gone beyond targeted surveillance to make a real difference in improving patient care outcomes and reducing costs.

    Everyone at RWJ Hamilton-from senior leadership and management to frontline staff-has been empowered to recognize the clinical and financial impact of preventing and managing HAIs, as well as the clear benefits of a cross-disciplinary effort to achieve the goal of reducing infections.

    Jim Maher is CFO, Robert Wood Johnson University Hospital Hamilton, Hamilton, N.J., and a member of HFMA's New Jersey Chapter (jmaher@rwjuhh.edu).

    David Stricklin is regional service director, CareFusion MedMined Services, CareFusion Solutions LLC, Birmingham, Ala. (david.stricklin@carefusion.com).


     a. Scott II, R.D., "The Direct Medical Costs of Healthcare-Associated Infections in U.S. Hospitals and the Benefits of Prevention," Centers for Disease Control and Prevention, March 2009.

     b. Clark, C., "CMS Releases Value-Based Purchasing Incentive Program," HealthLeaders Media, Jan. 11, 2011.

     c. Weber, D.J., Sickbert-Bennett, E.E., Brown, V., et al., "Comparison of Hospitalwide Surveillance and Targeted Intensive Care Unit Surveillance of Healthcare-Associated Infections," Infection Control & Hospital Epidemiology, December 2007, pp. 1361-1,366.

    d. Brossette, S., Hacek, D.M., Gavin, P.J., et al., "A Laboratory-Based, Hospital-Wide, Electronic Marker for Nosocomial Infection: The Future of Infection Control Surveillance?" American Journal of Clinical Pathology, January 2006, pp. 34-39.