Efficient Identification of Patients from Healthcare Facilities (HFs) for Infection Control Surveillance

Background: Prior exposure to HFs is an important risk factor for carriage of multidrug-resistant organisms (MDROs). Rapid identification of patients transferred from other HFs supports infection control activities such as active surveillance and preemptive Contact Precautions but frequently requires manual chart review.

Objective: To develop and validate a novel Point of Origin Flag (POF) based on standard hospital administrative codes to efficiently identify patients admitted from long-term and acute HFs.

Methods: We identified a universal “point of origin” hospital code that categorizes a patient’s immediate location prior to admission as HF versus non-HF and that is required for billing claims in all hospitals (form UB-04). This code is assigned within 24 hours of a patient’s admission using bed reservation and ambulance report form records. We used the code to generate our POF, which can be readily accessed by medical providers and infection control staff in the electronic patient chart. We defined the following institutions as HFs: acute care hospitals, emergency departments, long-term acute care hospitals (LTACHs), and skilled nursing facilities (SNFs); all others were considered non-HFs.

In a cross-sectional prospective study of randomly selected adult inpatients at a 676-bed urban academic medical center from July - August 2010, we validated the POF against a gold standard of patient-reported point of origin. Test performance measures (Kappa statistic, sensitivity, specificity, positive predictive value [PPV] and negative predictive value [NPV]) were calculated for the POF. The accuracy of the POF was compared to another commonly used source of admissions information, the physician history and physical (H&P).

Results: Among 523 patients included in the study, 85 (16%) patients self-reported as being admitted from another HF. Compared to the gold standard, the POF had a sensitivity of 86% (95% confidence interval [CI], 77 to 92%), specificity of 98% (CI 97 to 99), PPV of 91%, and NPV of 97%. The overall kappa was 0.86. The H&P had a sensitivity of 75% (CI 65 to 84), a specificity of 98% (CI 96 to 99), a PPV of 88%, and a NPV of 95%. For identifying patients from SNFs alone, the sensitivity of the POF and H&P were 50% (CI 23 to 77) and 71% (CI 42 to 92), respectively. For identifying patients from acute hospitals and LTACHs, the sensitivity of the POF and H&P were 93% (CI 84 to 98) and 76% (CI 64 to 85), respectively.

Conclusions: The POF is an accurate, efficient, and easily implemented method of identifying adult inpatients admitted from a healthcare facility; its performance is comparable to H&P review but requires less time and effort. The lower sensitivity of the POF for patients from SNFs reflects their more diverse points of entry at our institution compared to those from LTACHs and acute hospitals. This tool can increase infection control situational awareness by identifying patients at high risk for MDROs.