The Center for Medicare and Medicaid Services requires hospitals to report central line (CL)–associated bloodstream infection rates to the National Healthcare Safety Network as the number of CL-associated bloodstream infections per 1,000 CL-days. To ensure the validity of these rates, the National Healthcare Safety Network publishes criteria for defining CL-associated bloodstream infections and requires hospitals to validate methods used for determining the number of CL-days occurring. 1 Because daily observation of the number of patients with CLs is time consuming, many hospitals are seeking more efficient methods for reliably establishing the number of CL-days in their facility. Currently, there is a paucity of published data regarding the validity of using electronic medical record (EMR) systems to determine the number of CL-days in hospitals.Reference Wright, Fisher, John, Reynolds, Peterson and Robicsek 2 – Reference Tejedor, Garrett and Jacob 4 As a result, we conducted a study to validate the accuracy of a daily CL-day report generated by an EMR system.
METHODS
A prospective study was conducted at a 1,500-bed hospital that implemented an EMR system (Epic Systems) on its 2 campuses in 2013. Upon implementation of the EMR, infection control personnel visited all intensive care units for 2 weeks to compare ward-based patient-specific nursing documentation on the presence and type of patients’ CLs with data documented by nurses in the lines/drains/airways field in each patient’s electronic record. At the time of CL insertion, nurses enter data into fields regarding type, location, and laterality of CLs placed. On each shift, nurses review data in the lines/drains/airways field and update information as needed. An EMR report writer developed a daily, automated report listing all inpatients at 6 AM who have one or more CLs, the CL type(s) present, and presence of an arteriovenous fistula on the basis of data in the lines/drains/airways field. Daily EMR reports are reviewed by infection control personnel, patients with only an arteriovenous fistula are excluded, and the number of CL-days is calculated for each day.
To determine the validity of the EMR CL-days report, nurse managers performed audits by observation of each patient on their unit to determine whether patients had one or more CLs present. Nurses were given a printed list of patients on their unit and recorded whether or not each patient had one or more CLs in place. Observational audits by nurses were conducted in the morning on all nursing units except psychiatry units. Audits by nurses were conducted on 8 days over a period of 3 months. Paired audits performed on Mondays and Thursdays of the same week were conducted on 3 occasions (6 audits combined) and on Wednesday and Friday of the same week during another week. A paired Monday-Thursday sampling scheme has been found to yield accurate estimates of CL-days.Reference Thompson, Edwards and Bamberg 5 Nurses forwarded data sheets to Hospital Epidemiology for analysis. The number of CL-days observed by nurses on the 8 audit days was compared with the number of CL-days listed in the EMR reports printed on the same days.
RESULTS
Nursing audits were performed on 57 of the hospital’s nursing units located on 2 campuses. Nursing staff did not submit data on 114 patients, who were excluded from the analysis. Also, we excluded 43 patients who were listed on daily EMR reports but for whom observations were not performed because they were not physically present on nursing units as they were undergoing hemodialysis, surgery, or diagnostic examinations at the time of nursing audits. The 8 audits performed by nurses revealed a total of 2,284 patients who had one or more CLs present (range per day, 252–317) and 6,033 patients without a CL. The EMR-generated reports listed a total of 2,288 CL-days (range per day, 255-317) on the 8 days on which nursing audits were conducted. On 6 of the 8 audit days, the number of CL-days listed in the EMR report matched exactly the number of CL-days reported by nurses. On one day, the EMR report listed 255 CL-days while nurse observations revealed 252 patients with one or more CLs (98.8% match). On another day, the EMR report listed 293 CL-days while nurses reported 292 patients with one or more CLs (99.7% match). Overall, of the 2,288 CL-days included in the EMR reports, 2,284 (99.8%) were reported by the nursing audits. All CL-days detected by nursing audits were included in EMR reports. The sensitivity and specificity of the EMR report were 100% and 99.9%, respectively.
DISCUSSION
Our prospective study to validate the EMR report of the number of CL-days was unique in several respects. In 2007-2008, Wright et alReference Wright, Fisher, John, Reynolds, Peterson and Robicsek 2 used an early version of the Epic EMR system to estimate CL-days and compared the results with infection preventionist reviews of the medical records of a random sample of patients for the presence of CLs. The infection preventionist reviewed nursing documentation fields, interdisciplinary team notes, laboratory data, and radiology dictations. Unlike our study, no direct observations of patients were performed.
A medical intensive care unit–based study by Hota et alReference Hota, Harting and Weinstein 3 revealed that electronic models based on information in a proprietary clinical data warehouse correlated well with bedside observations of CL-days. A large study by Tejedor et alReference Tejedor, Garrett and Jacob 4 compared the accuracy of CL reports generated from nursing documentation in a different EMR system (Cerner) with data obtained by nurses who initially conducted bedside observational audits 5 times a week, and then several times per month. Unlike in our study, Tejedor et alReference Tejedor, Garrett and Jacob 4 did not state at which intervals and on which days of the week observational sampling was conducted. Although several CL-day sampling strategies have been described, we conducted most observational surveys on Mondays and Thursdays of the same week because this sampling strategy has yielded the best estimates of CL-days.Reference Thompson, Edwards and Bamberg 5 – Reference Backman, Nobert, Melchreit, Fekieta and Dembry 8 Tejedor et alReference Tejedor, Garrett and Jacob 4 found that initial problems with the accuracy of the EMR reports of CL-days were corrected by a number of interventions, yielding revised reports with excellent accuracy. They emphasized the importance of validating EMR data on the number of CL-days.
We found that the number of CL-days listed in EMR reports, which are based on daily nursing documentation, matched data derived from bedside nursing audits 99.8% of the time. When the EMR report listed a patient as having a CL but nursing audits did not confirm this, patients’ CLs may have been removed between the time the EMR report was generated at 6 AM and when nurses conducted their audits later in the morning. Limitations of the current EMR report include the need for infection preventionists to quickly review the report and exclude patients with only arteriovenous fistulas and that ports are listed whether accessed or not. Limitations of our study include the relatively small number of nursing audits performed and the fact that the study was conducted in only one hospital.
In conclusion, the present study suggests that automated reports of CL-days generated by use of lines/drains/airways data in the Epic EMR can provide accurate estimates of the number of CL-days.
ACKNOWLEDGMENTS
Financial support. None reported.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
