Predictors of the Diagnosis and Antibiotic Prescribing to Patients Presenting with Acute Respiratory Infections BY RYAN JOERRES CAPSTONE COMMITTEE MEMBERS: DENNIS J. BAUMGARDNER, MD, AJAY K. SETHI, PH.D., MHS. CHRISTOPHER J. CRNICH, MD, MS.
Outline Introduction Project Methods Results Discussion/Limitations Conclusion
Introduction Emergence of antibiotic resistance is multi-factorial Major component: prescription of antibiotics for non-bacterial acute respiratory infections Common cold is nearly uniformly viral in etiology Acute bronchitis bacterial etiology 5-10% of the time National Ambulatory Medical Care Survey 2001: 58.7% of acute bronchitis diagnoses given antibiotics 2005: 59% Routine antibiotic prescription for uncomplicated acute bronchitis is not recommended and evidence-based studies proving otherwise are lacking
Introduction 2005 study of family physicians by Phillips and Hickner: physician s use of the term chest cold rather than bronchitis improved patient satisfaction when an antibiotic is NOT prescribed. Clinicians preferentially choose to verbally diagnose a chest cold or URI, rather than bronchitis, to obtain patient satisfaction when not prescribing antibiotics Hypothesis: clinicians choose a bronchitis ICD-9 billing code to justify the prescribing of antibiotics What are predictors of both diagnosis and antibiotic prescription?
Project Methods Aurora Health Care database 98 Outpatient clinics, Oct 2007-Sept 2008 Random sample of N=3,513 patients ages 8-49 (8-18,19-49) presenting for first time Diagnosed by a primary care clinician: Nurse practioner, physician assistant, primary care physician (MD vs. DO) (1) Acute Bronchitis (ICD-9 Code 466.0) (2) Common Cold/URI (ICD-9 Code 460 / 465) Predictors assessed by Chi-square test, T test and Mann-Whitney. Multivariate analysis performed using regression models entering significant (p<0.05) or borderline significant (p<0.15) variables identified by univariate analysis Patient age, gender, race/ethnicity Time of season Clinician type and specialty Geographical location of the clinic, Percent rural of clinic
Crude Results 61% Female 65% Caucasian, 4% African American, 3% Hispanic, <1% Asian, 27% unknown Mean age: 28.6, Median age: 30 > 70% of patient visits occurred during flu season (October March) 51.5% (1810/3513) given cold/uri diagnosis Antibiotics prescribed almost 58% of the time When an antibiotic was prescribed, Macrolides were the most commonly prescribed drug (62.5%) by all clinicians
Results Patients aged 8-18 Rate of acute Rate of prescribing Rate of prescribing given acute bronchitis bronchitis code code versus Cold/URI (P-value) nurse practioners 35% 47% 86% vs. 26% () physician assistants Low sample Low sample Low sample pediatricians 21% 34% 89% vs. 19% () family medicine 37% 53% 84% vs. 34% () Patients aged 19-49 Rate of acute bronchitis code Rate of prescribing Rate of prescribing given acute bronchitis code versus Cold/URI (P-value) nurse practioners 53% 60% 84% vs. 33% () physician assistants 48% 58% 88% vs. 31% () Internists 60% 71% 84% vs. 52% () family medicine 57% 63% 83% vs. 61% () Total values per age group Patients aged 8-18 Patients aged 19-49 Patients aged 8-49 86% vs. 27% 84% vs. 38% 84% vs. 33% X 2 = 315.0 P-value X 2 = 541.25 P-value X 2 = 926.22 P-value
Multivariate analysis of predictors of acute bronchitis diagnosis across all three age groups Ages 8-18 Variable P value Odds Ratio Males 0.038 1.38 95% Confidence Interval 1.02-1.86 Family medicine vs. pediatrician 2.16 1.58-2.96 Ages 19-49 Age 1.02 1.01-1.03 Males 1.36 1.15-1.61 MD/DO vs. NP/PAs 0.011 1.28 0.64-0.94 Ages 8-49 Age 1.04 1.04-1.05 males 1.34 1.16-1.54
Multivariate analysis of predictors of antibiotic prescribing across all three age groups Ages 8-18 Variable P value Odds Ratio 95% Confidence Interval Family medicine vs. pediatricians 0.001 1.72 0.41-0.81 Ages 19-49 Acute bronchitis diagnosis Acute bronchitis diagnosis 16.67 0.04-0.09 8.33 0.10-0.15 Ages 8-49 Age 1.01 0.98-0.99 Acute bronchitis diagnosis 10.00 0.09-0.12
Discussion/Limitations Across all age groups, being a male patient was found to be the greatest predictor of being diagnosed with acute bronchitis Across all age groups, diagnosis of acute bronchitis was found to be the greatest predictor of being prescribed an antibiotic Single regional medical database retrospective Generalizability of results limited Preclusion of Hawthorne effect Clinician demographics not provided Clinician practice behaviors and influences on practice behaviors not provided / observed
Conclusion Overprescribing of antibiotics is common in patients presenting with acute respiratory infections by all primary care specialties Efforts should be made to reduce inappropriate antibiotic use for: containing costs, preventing side effects, and limiting the risk of acquiring and spreading drug resistant bacterial strains. Additional studies are needed to: identify the differences in diagnosis and antibiotic prescribing practices amongst primary care specialties investigate whether choice of bronchitis diagnosis is sometimes used to justify the prescribing of antibiotics amongst primary care clinicians.
Acknowledgements Dennis J. Baumgardner MD Ajay K. Sethi PhD, MHS Christopher J. Crnich MD, MS Barbara L. Duerst RN, MS Lisa M. Joerres MD Linda M. Joerres Andrew T. Joerres Nolan J. Ruekert Julianna G. Bosnjak Evan T. Joerres