In 2001, the VA Maryland Health Care System (VAMHCS) implemented a re-engineered pharmacy process that translates key scientific evidence into software tools programmed into the Veterans Administrations (VA) Computerized Patient Record System (CPRS). These tools, called MUEs (for Medication Use Evaluation), interpose computerized decision support (CDS) at the time of an electronic prescription (eRx). It is our long-term objective to promote safe and effective drug utilization by integrating CDS into VAs pharmacotherapeutic process, so that nationally developed drug criteria-for-use can be seamlessly applied to individual eRx.
Our immediate objective is to evaluate the long-term effectiveness of MUEs aimed at reducing overutilization of antibiotics in outpatients with uncomplicated acute upper respiratory infections (AURI). Our main hypothesis is that MUE can adjust antibiotic utilization toward consensus guidelines for the treatment of outpatients with AURI. To test this hypothesis, we propose studies aimed at answering the following research questions:
1) Do MUEs targeting azithromycin and gatifloxacin adjust utilization of these drugs toward guideline recommendations for acute bronchitis, acute sinusitis and non-specific upper respiratory tract infections?
2) Did the introduction of MUEs aimed at only two antibiotics improve the overall antibiotic utilization for AURI?
3) Following the MUE introduction, what are the predictors of adherence or non-adherence with the guidelines?
This work uses a retrospective observational case-control study, designed to determine the differences in congruence to national antibiotics use guidelines for AURI, prior to and after MUE implementation. The intervention site is VA Maryland Health Care System; the control site is the VA Salt Lake City Health Care System. The study period begins in 2002, after guideline publication and one year prior to MUE implementation in 2003, and continues for four years post MUE implementation (2003-2006).
The project takes advantage of extensive existing relational databases at both study sites and from already developed AURI case-finding algorithms (CDA) to identify the study population and candidate records for manual chart review. We applied a CDA [(one of 197 "respiratory" ICD-9 code OR Cough Remedies OR Temp >=38 C) AND (positive results of text analysis for AURI symptoms)] to 2.7 million outpatient visits during the study period (1/2002 - 12/2006). We manually reviewed CPRS information on the day of the first visit of all 4776 unique patients identified by the CDA , for guideline-defined symptoms (cough, sinus pressure or pain, sore or erythematous pharynx, nasal discharge etc) that identify specific AURIs (acute bronchitis, sinusitis, pharyngitis, unspecified acute upper respiratory infection). We also abstracted all of the data (vitals signs, orders/reports for chest imaging, and medications of interest, TIU notes, etc) necessary to determine whether antibiotics were indicated or permissible for each of these conditions. We reviewed all reports from chest imaging ordered on the index visit date, and evaluated if these x-rays were compatible with a pneumonia or not. We performed a secondary review of 600 randomly selected CDA-positive CPRS records for cough, cough duration, sputum production, sputum production duration, fever/chills, and shortness of breath: interrater agreement yielded kappa values between 0.81 and 0.87. Similarly, a secondary review of 538 randomly selected chest imaging reports from a total of 2748 yielded a kappa of 0.82 for classification into three probabilities that the findings represented a pneumonia.
Of the 4776 patients identified by the CDA, 2980 had a total of 337 pneumonias and 4014 uncomplicated AURI diagnoses (pharyngitis (1419), bronchitis (2122), sinusitis (371), non-specific URI (102)). For the conditions targeted by the MUE (pneumonia, bronchitis, sinusitis and non-specific URI), overall proportion of unwarranted antibiotics prescriptions decreased from 42% to 32% (p < 0.0001), pre vs. post intervention. This reduction was not distributed equally amongst all antibiotics. For the two antibiotics targeted by the MUE (azithromycin and gatifloxacin), unwarranted prescription proportion for all AURI diagnoses decreased from 22% to 3% (p , 0.0001), with misutilization for patients with bronchitis and normal vital signs, the safest patient subset according to the guidelines, reduced even further (17.7 vs. 1.8% pre vs. post intervention p < 0.0001). In contrast, unwarranted prescriptions for antibiotics not targeted by the MUE remained unchanged (30.1% vs. 30.5%, p = 0.2348), suggesting that the MUE did not simply displace misutilization to alternative antibiotics. The MUE intervention did not prevent utilization of antibiotics when they were absolutely indicated: all cases of pneumonia received antibiotics. Azithromycin or gatifloxacin remained the predominant antibiotics prescribed for pneumonia (27/33 (82%) of total cases pre-MUE vs. 232/304 (76%) post-MUE. Overall, guideline-congruent antibiotic utilization (prescribed when warranted + not prescribed when unwarranted divided by total number of patients with targeted AURI diagnoses) increased from 62.7% to 72.3% (p = 0.0001) pre vs. post intervention. Most of the improvement in congruence was attributable to the MUE-targeted drugs.
The above results indicate that CDS at the time of single prescriptions represent an effective, sustainable approach to modify antibiotics utilization for outpatients with uncomplicated AURI. The results suggest that MUE-mandated guideline exposure is sufficient to markedly improve guideline adherence, neither superimposed surveillance nor audit/feedback is required. The key challenge is thus to expand the intervention to other antibiotics while retaining system tolerability. These results provide key ammunition for further system refinements, extensions to other other drug classes and disease processes, and for grant proposals of systematic process evaluations.
External Links for this Project
- Rattinger GB, Mullins CD, Zuckerman IH, Onukwugha E, Walker LD, Gundlapalli A, Samore M, Delisle S. A sustainable strategy to prevent misuse of antibiotics for acute respiratory infections. PLoS ONE. 2013 Jul 11; 7(12):e51147. [view]
- Gundlapalli, South, Phansalker, Kinney, Shen, Delisle, Perl, Samore. Application of Natural Language Processing to VA Electronic Health Records to Identify Phenotypic Characteristics for Clinical and Research Purposes. Presented at: American Medical Informatics Association on Translational Bioinformatics Research Annual Summit; 2008 Mar 10; San Francisco, CA. [view]
- Delisle S, Fang M, Sun BC, South B, Smith R, Samore MH, Perl. Automated surveillance to detect an influenza epidemic: Which respiratory syndrome should we monitor? Paper presented at: International Society for Disease Surveillance Annual Conference; 2008 Dec 3; Raleigh, NC. [view]
- Rattinger GB, Mullins CD, Onukwugha E, Zuckerman IH, Walker L, South BR, Gundlapalli A, Delisle S. Clinical Decision Support at the Time of An e-Prescription Can Sustainably Decrease Unwarranted Use of Antibiotics for Acute Respiratory Infections. Paper presented at: Infectious Diseases Society of America Annual Meeting; 2010 Oct 21; Vancouver, Canada. [view]
- Rattinger GB, Mullins CD, Onukwugha E, Zuckerman IH, Walker L, Delisle S. Clinical Decision Support at Time of Prescribing To Decrease Inappropriate Antibiotics Use for Treatment of Acute Respiratory Infections. Paper presented at: American Thoracic Society Annual International Conference; 2010 May 14; New Orleans, LA. [view]
- Campbell B, South B, Gundlapalli A, Delisle S, Samore MH, Perl. Concept level evaluation of negotiation processing for an influenza-like illness text classifier. Paper presented at: International Society for Disease Surveillance Annual Conference; 2008 Dec 3; Raleigh, NC. [view]
- Delisle S, South B, Phansalker, Perl, Gundlapalli A, Samore MH. Free-text processing to enhance detection of acute respiratory infections. Paper presented at: International Society for Disease Surveillance Annual Conference; 2008 Dec 3; Raleigh, NC. [view]
- Campbell B, South B, Gundlapalli A, Samore MH, Delisle S. Natural Language Processing: Can it help detect cases and characterize outbreaks? Paper presented at: International Society for Disease Surveillance Annual Conference; 2008 Dec 3; Raleigh, NC. [view]
- South, Chapman, Delisle, Shen, Kalp, Perl, Samore, Gundlapalli. Optimizing a Syndromic Surveillance Text Classifier for Influenza-like Illness: Does Document Source Matter? Presented at: American Medical Informatics Association Annual Symposium; 2008 Jan 6; Washington, DC. [view]
- Delisle S. Re-Purposing the Electronic Medical Record for Public Health. Presented at: McGill University Society and Medicine Conference; 2008 Jan 1; Montréal, Canada. [view]
- Delisle S. Re-Purposing the Electronic Medical Record for Public Health. Paper presented at: AcademyHealth Annual Research Meeting; 2009 Jun 28; Chicago, IL. [view]
- Godish M, Delisle S, Onukwugha E, Lee MT, Shinogle J, Zuckerman M. The Electronic Medical Record As A Surveillance System To Prevent Adverse Outcomes: Finding Patients At-Risk for CHF-Related Hospital Admissions. Paper presented at: American Thoracic Society Annual International Conference; 2010 May 14; New Orleans, LA. [view]
- Delisle S, Fang, Hongzhang, Sun BC, South B, Gaff, Samore MH, Perl. Using biosurveillance-system facsmiles to compare aberrancy-detection methods: Should Biosense use SatScan? Paper presented at: International Society for Disease Surveillance Annual Conference; 2008 Dec 3; Raleigh, NC. [view]