Strategies are therefore needed to improve both the recognition of delayed test results and the communication of such results to the appropriate health care professional providing care to the patient. Computerized information systems are gaining increasing recognition as a valued tool for reducing medical errors. Computerized tools that can improve communication of laboratory information and assist with monitoring of pending laboratory tests are 2 main strategies that show how information technology can reduce medical errors and prevent adverse patient outcomes.⁴ An automated alerting system for inpatient critical laboratory results has been shown to reduce the time until appropriate treatment is subsequently prescribed.⁵ However, no published studies have evaluated the utility of automated alerting systems for microbiology test results that are pending at the time of hospital discharge.

The Computer-Based Antimicrobial Monitoring (CBAM) system used at Mayo Clinic in Rochester, MN, is an automated, rules-based medical informatics system that monitors antimicrobial and microbiology information among hospitalized patients in near real time. Its objectives are to improve patient care through the reduction of medical errors, enhancement of education for the health care professional, and improvement of hospital cost savings. Each CBAM algorithm or “rule” contains a set of clinical criteria that identifies situations in which the review and/or intervention by an infectious diseases (ID) physician or ID pharmacy specialist may be beneficial to patient care. More than 60 CBAM rules utilize information from a number of integrated electronic data sources, including patient demographic information, calculated metabolic values, laboratory and microbiology data, and medication information. When appropriate, the patient’s primary hospital physician is contacted to discuss the specific clinical issue and optimal antimicrobial and/or other infection management measures as needed.

At our medical center, CBAM has had a significant influence in the optimization of hospital inpatient care, including identification of untreated or suboptimally treated infections, incorrect antimicrobial dosing, adverse drug-to-drug interactions, and suboptimal hepatorenal or serum drug level monitoring. During a 16-month period in 2003-2004, approximately 83% of CBAM interventions resulted in improvements to inpatient care pertaining to antimicrobial and/or infection management.⁶ However, an important limitation of CBAM has been the inability of the system to identify clinical infection and antimicrobial management concerns for outpatients or for those recently dismissed from the hospital.

To assess the significance of overlooked culture results from patients recently dismissed from the hospital and the potential benefits from clinical intervention, we developed a new CBAM rule to identify cultures collected during a patient’s hospitalization that subsequently turned positive after hospital dismissal. We then determined the outcomes of interventions by the ID physicians and ID pharmacy specialist working with CBAM.

Although a relatively short incubation for some slow-growing organisms, a 7-day duration was chosen initially for CBAM of pending cultures after hospital dismissal to best pilot this new CBAM rule. Flagged patients by CBAM had their medical and laboratory records reviewed by an ID physician assigned to CBAM to determine whether (1) the positive culture result was clinically relevant or contributing to disease, (2) the patient was prescribed appropriate antimicrobial therapy on hospital dismissal, and (3) any additional diagnostic (eg, echocardiogram, radiologic imaging, etc) or additional therapeutic measures were needed. If the CBAM-ID physician had any clinical concerns regarding possible suboptimal infection management, the patient’s primary care physician was contacted by the CBAM-ID physician with the corresponding delayed positive culture results and, if needed or requested, given appropriate antimicrobial treatment recommendations and other supporting clinical advice. For the more complex or serious infections and antimicrobial-related issues, a formal ID outpatient consultation was also recommended.

Among the 112 patient interventions performed by the CBAM-ID physician, 32 (29%) were for positive blood cultures (Table 1). Among CBAM-ID physician communications with primary health care professionals that resulted in subsequent alterations in their patients’ post-hospitalization management and that also were clearly documented by those health care professionals in the electronic medical record, 38 (34%) resulted in initiation, alteration, or prolongation of antimicrobial therapy. Among all CBAM interventions, there were no instances in which recommendations by the CBAM-ID physician were not followed or resulted in a subsequent alternative inappropriate antimicrobial change.

Not all CBAM-ID physician interventions necessitated a change in therapy. Twenty-one interventions (19%) resulted in no change in outpatient management given that acceptable therapy had already been provided after discussion by the CBAM-ID physician with the primary health care professional. The outcomes in patient management from the remaining 53 CBAM interventions (47%) could not be determined because of insufficient or absent outpatient documentation in the Mayo Clinic patient medical record. This was a common occurrence in patients subsequently managed outside of the Mayo Clinic system because outpatient documentation by non-Mayo Clinic health care professionals is not typically accessible through our electronic medical record.

Anaerobic blood cultures typically require prolonged incubation periods, and thus results may not be available during short patient hospitalizations. The presence of anaerobic bacteria in the bloodstream (more than one-third of delayed blood culture results in the current study) may also suggest a more complex and often polymicrobial infection depending on the source and bacteria species.

CBAM is not the only mechanism available to notify primary health care professionals of delayed positive culture results. Many hospital microbiology laboratories, including those at Mayo Clinic, will attempt to contact the hospital physician or service about a positive culture from a sterile body site. This system works relatively well for hospitalized patients but poses challenges for those who have been discharged. One problem with delayed positive culture results is that health care professionals taking care of patients in the hospital are commonly not the same health care professionals who follow up in the outpatient setting. Our laboratory technicians do not have access to patient medical records and do not know whom to contact beyond the inpatient health care professional who originally ordered the culture. Even when the inpatient health care professional is contacted by the microbiology laboratory regarding the delayed positive culture result, that information is commonly not communicated to the appropriate primary outside health care professional or to the patients themselves. Additionally, the laboratory technicians may not be qualified to render opinions to the inpatient physician or service regarding the specific clinical relevance or application of positive culture results.

Many community and academic hospitals utilize physician hospitalists and residents to care for inpatients, and the communication between the inpatient health care professionals and outpatient primary health care professionals is often compromised.¹⁰ Resident work hour limitations and frequent patient “sign-offs” to other health care professionals further complicate accurate communication and documentation of pending laboratory results at the time a patient is dismissed from the hospital. During the transition from inpatient to outpatient status, it can be difficult for both the hospital health care professionals following up on pending cultures and the laboratory technicians to determine which person is best suited to address important microbiology test results.

In addition to microbiology, laboratory, and pharmacy data, an important component to the CBAM system is the direct access to inpatient and outpatient Mayo Clinic electronic medical records. The ability to review the actual medical records of patients dismissed from the hospital enables the CBAM-ID physician to both understand the specific clinical application of a patient’s positive culture result and the identity of the patient’s outpatient health care professional. Both the direct and time-efficient notification by CBAM-ID staff to outpatient health care professionals and the opportunities to discuss culture interpretation and management recommendations through expert telephone consultation are why CBAM has a unique and clinically advantageous role at Mayo Clinic. This is not to say that an ID physician is required for this type of medical informatics program to be effective because recognition of delayed positive culture results after hospital dismissal is the primary goal. However, the ID physician can provide additional clinical expertise via a “curbside” or virtual consultation to assist the primary health care professional.

For this rule, CBAM identifies only the post-hospital positive culture results, whereas the CBAM-ID physician is required to interpret the clinical relevancy within the context of each patient’s diagnosis and plan of care. Not all delayed positive culture results or interventions by the CBAM-ID physician require initiation of or change in antimicrobial therapy. Indeed, 421 (79%) of the CBAM-flagged patients reviewed by the CBAM-ID physician did not require intervention or communication with the primary health care professional (Figure); 21 (19%) of the CBAM-ID physician interventions and discussions with primary health care professionals did not involve any change in patient management. These results often reflected an attentive primary health care professional who was aware of the positive culture result and had already implemented an appropriate management plan. Patient-specific information outlined by the primary health care professional, including palliative care, chronic suppression, or specific patient wishes, commonly led to alternative and appropriate antimicrobial treatment plans. In addition, delayed growth of bacteria from some tissue sites may have been expected (eg, repeatedly positive bacterial growth from surgically débrided tissues or fastidious bacteria from a known polymicrobial infection) and did not warrant a change in management.

Because of insufficient subsequent outpatient documentation in the patient medical record, 47% of the CBAM-ID physician intervention outcomes could not be accurately determined. This was often due to the large numbers of nonregional (eg, out-of-state and international) patients at our medical center who are managed by non-Mayo Clinic health care professionals. Although the CBAM-ID physician would routinely contact outside health care professionals regarding delayed positive culture results (as documented in a CBAM note), we did not have access to outside, non-Mayo Clinic patient medical records or the identity of many outside health care professionals to confirm whether CBAM-ID physician interventions and recommendations were carried out. Considering the high number of post-hospitalized patients managed outside of the Mayo system, we think that the overall number of patients who had their treatment plan appropriately influenced by CBAM intervention was significantly higher.

Our study has several limitations. As previously mentioned, our inability to access non-Mayo Clinic outpatient records of patients recently dismissed from our hospitals limited a more accurate and comprehensive determination of CBAM intervention impact. We were also not able to contact most outside non-Mayo health care professionals to retrospectively confirm results of CBAM intervention. These are areas for CBAM system improvement that are currently in development. Additionally, some outpatient modifications in antimicrobial therapy (eg, extension of duration of therapy or outpatient therapy) may have been done without documentation in the medical records. We acknowledge that 7 days can be a relatively short incubation for some slow-growing organisms, including mycobacteria and some fungi. After the current study was completed, we extended the time that the CBAM program monitors pending culture results from 7 days to 1 month after hospital dismissal.