Clinical approach to non-responsive pneumonia diagnosed in the primary care office: a retrospective feasibility study


Department of Family Medicine,Center for Urban Population Health, Aurora UW Medical Group

Presentation Notes

Presented at 2013 Aurora Scientific Day, Milwaukee, WI


Background/significance: Community Acquired Pneumonia (CAP) is commonly diagnosed in the primary clinical setting and often successfully treated. Treatment failures occur in 6-7% of patients treated in the ambulatory setting. However, the management of non-responsive pneumonia has been poorly studied and is not clearly understood.

Purpose: The aim of our study was to determine the feasibility of use of the Aurora data warehouse to characterize clinical features and clinician approach to NRP.

Methods: This is a retrospective study of medical records taken from the Aurora Health Care database based on ICD9 codes for pneumonia (codes 480-486). Patient visits from 10/5/2006 to 9/30/2011 were reviewed. Our initial target sample size was 200 cases and 200 controls. We requested records seeking a clinical or radiographic diagnosis of pneumonia, which was initially made in an outpatient setting or urgent care clinic by a primary care provider. We excluded individuals who initially presented to an emergency room or specialist. Cases were those patients diagnosed with pneumonia which failed to improve within 10 days of antibiotic prescription or worsened after the fifth day following the prescription of a course of antibiotics. Controls were those with acute pneumonia, not meeting NRP definition. Variables included: race, gender, age, smoking, co-morbid conditions, duration of cough, fever, antibiotics given, change in antibiotics, and type/ duration of symptoms. Categorical data were analyzed utilizing t tests and the Mann-Whitney test; categorical variables were analyzed by chi-square or Fischer exact tests. Statistical significant was defined as p values less than 0.05.

Results: Review of 339 records revealed 15 cases of NRP and 13 controls. Duplicate cases accounted for 28/339, 153 records were non-pneumonia (frequently Chlamydia [code 1 digit off]) and 12 were unknown diagnosis. Excluded pneumonia cases included 68 incomplete records and 50 specialist/ED/hospital admission. Cases were younger than controls (59 vs. 46 years, p=0.02). There were no other significant differences. Sample size was limiting.

Conclusion: Discrepancies in diagnosis and the incomplete records found in the data warehouse during this time period precluded efficient obtainment of adequate sample size for this study. Use of Smart Chart for data retrieval may significantly improve efficiency. Older age is a risk for NRP in this population.

Document Type


Link to Full Text