1. CHANGES IN ANTIMICROBIAL PRESCRIBING PATTERNS FOLLOWING A HEALTH CARE PROVIDER EDUCATIONAL INTERVENTION Nadia Shalauta Juzych, ScD, MS*, Mousumi Banerjee, PhD**, Lynnette Essenmacher**, Stephen A Lerner, MD** *Michigan Public Health Institute,**Wayne State University School of Medicine CHANGES IN ANTIMICROBIAL PRESCRIBING PATTERNS FOLLOWING A HEALTH CARE PROVIDER EDUCATIONAL INTERVENTION Nadia Shalauta Juzych, ScD, MS 1, Mousumi Banerjee, PhD 2, Lynnette Essenmacher 2, Stephen A Lerner, MD 2 1 Michigan Public Health Institute, 2 Wayne State University School of Medicine Abstract Problem Statement: Inappropriate use of antimicrobials in the treatment of acute upper respiratory tract infections, which usually have a viral etiology, contributes to emergence and spread of antimicrobial resistance in Streptococcus pneumoniae and other human pathogens. Objective: To reduce antimicrobial use for management of acute upper respiratory tract infections in adult and pediatric patients. Design: Prospective, non-randomized control trial, including baseline (November 15, 1999 to March 31, 2000) and study (November 16, 2000 to March 31, 2001) periods. Setting: Four primary care clinics within a staff model health maintenance organization in Detroit, Michigan. Study Population: Twenty-one primary care physicians at two clinics where the educational intervention was implemented, and nine primary care physicians at two control clinics where no educational programs were presented. Intervention: Interactive case-based educational program on appropriate use of antimicrobials in the treatment of upper respiratory tract infections for pediatric and adult medicine physicians and their staff was presented at two inner city health maintenance organization (HMO) clinics for indigent patients. Outcome Measure: Antibiotic prescribing for acute upper respiratory tract infections during the baseline and study years among the intervention and control groups. Results: Antimicrobial prescribing among the physicians who received the educational intervention decreased 24.6% between the baseline and study years (p<0.0001) for both pediatric and adult medicine physicians. There was no significant decline in rates of antimicrobial prescribing for the control group of physicians between the baseline and study years for pediatric (p=0.35) or adult medicine (p=0.42) physicians. There was a significant difference in the decline in rates of antimicrobial prescribing between the control and intervention groups (p<0.0003) for pediatricians and (p<0.01) for adult medicine physicians. Conclusion: An interactive case-based educational program for physicians and their staff appeared to be an effective means for reducing the prescribing of antibiotics for the treatment of upper respiratory tract infections by primary care physicians in an indigent HMO setting. Funding Source: Blue Cross Blue Shield of Michigan Background and Setting The rise in resistance to penicillin and other classes of antibiotics in Streptococcus pneumoniae, the most frequent bacterial cause of community-acquired pneumonia, otitis media, and meningitis, is due principally to selection from exposure to penicillins and other antibacterial agents. Since S. pneumoniae is often present in the upper respiratory tract of healthy individuals, each course of antibiotic therapy contributes to the selective pressure for emergence and proliferation of antibiotic-resistant strains. An estimated 75% of all antibiotic usage in outpatients in the US is prescribed for respiratory tract infections. The majority of upper respiratory tract infections (URIs), including the common cold, acute bronchitis, and many cases of otitis media, are caused by viruses, for which antimicrobials are not indicated. Previous studies that have documented changes in antimicrobial use for treating a number of uncomplicated viral upper respiratory tract infections through physician education have utilized population-based evaluations to determine the efficacy of their approaches in changing antimicrobial prescribing practices. In this study, we have been able to link diagnosis to antibiotic prescriptions and, therefore, to demonstrate the efficacy of our educational intervention in altering prescribing practices among individual physicians for both pediatric and adult patient populations. Setting The Wellness Plan (TWP) is a staff model health maintenance organization (HMO) that manages four primary care clinics and an outpatient urgent care center in Detroit, Michigan. The Wellness Plan HMO includes approximately 50,000 covered lives, and has a patient population that is more than 90% African-American. Members receive primary care services primarily from TWP physicians, staff, and laboratories. In addition, both clinicians and patients are assigned to one of the four clinics, and rarely do patients see clinicians from one of the other clinic sites. The Wellness Plan clinics contain in-house pharmacies that fill approximately 95% of the prescriptions. Patients with acute illnesses are given same-day appointments at each of the four clinics, and are seen only at the urgent care center when the clinics are closed. Study Aim To evaluate the efficacy of a health care provider educational intervention in reducing unwarranted prescribing of antibiotics for probable viral upper respiratory tract infections. Methods  Prospective, non-randomized control trial of an educational intervention program at two of the health plan’s clinics.  The program involved a half-day educational session that provided an overview of antimicrobial resistance, and  Utilized case study presentations to review appropriate treatment and diagnosis of bronchitis, pharyngitis, sinusitis, and otitis media.  The in-service educational program was conducted with health care staff of two of TWP’s four clinics, designated “intervention” clinics.  Staff of the other two TWP clinics did not participate in the educational intervention and served as a concurrent control population. Data on antimicrobial prescribing for the treatment of select URIs were collected from each of the clinics and evaluated for the periods November 16, 1999 to March 31, 2000 (baseline period) and November 16, 2000 to March 31, 2001 (study period). Intervention Twenty-six pediatricians, internists, and obstetricians participated in the educational programs, along with nurses, physician assistants, and pharmacists of two health plan clinics. The sessions were structured to provide an overview of antimicrobial resistance, followed by a case-based interactive review of appropriate diagnosis and treatment of a number of upper respiratory tract infections: bronchitis, pharyngitis, otitis media, and sinusitis. The participants were divided into “pediatric” and “adult medicine” groups for the case-based review portion of the program. Subjects The evaluation was conducted to determine changes in prescribing of antimicrobials in the treatment of bronchitis, pharyngitis, otitis media, and URIs not otherwise specified (nos) among the physicians of the TWP clinics between the baseline and study years, and between physicians at the intervention and control clinics. For each diagnosis, physicians were included in the analysis if they conducted at least 10 patient visits in each of the baseline and study years for that diagnosis. Measurements Data were gathered from all incident office visits for both pediatric and adult patients during which a diagnosis of one of the following URIs was made: bronchitis, otitis media, pharyngitis, or URI nos. The data were extracted from the administrative database files of TWP as identified by ICD-9 codes recorded for medical billing. Data were extracted for the periods November 16, 1999 to March 31, 2000 (baseline year) and November 16, 2000 to March 31, 2001 (study year). Incident visits were defined as representing the first office visit for any of the diseases of interest. Return visits within 30 days of an incident visit were excluded from the analysis. Patients were excluded for any of the following co-morbidities: asthma, chronic obstructive pulmonary disease (COPD), diabetes, HIV, congestive heart failure, or chronic ischemic heart disease. Calculations were also performed to evaluate the potential for diagnostic shifting.  Prescribing data were extracted from a pharmacy administrative database, linked to each patient visit by date of prescription.  Additional data were collected for each office visit: patient sex, age, and attending physician.  Data were also collected for the year in which each attending physician completed residency training. Statistical Analysis Univariate measures were used to compare the control versus the intervention group with respect to patient demographics. Antibiotic prescribing rate was calculated as the proportion of incident office visits where the patient received an antibiotic prescription. To evaluate the impact of the intervention on antibiotic prescribing rates within and between the intervention and control groups, a generalized linear mixed- effects model was used to control for potential clustering (random effects) of physicians by group. Antibiotic prescribing rates were modeled using binomial error distribution and logit link function. The model included physician-specific random effects. Time (i.e., baseline versus study year), patient gender, and years since residency training had been completed by physicians (categorized as in practice >15 years versus <15 years) were included as fixed effects for the within-group analysis. For the between- group analysis, we added group (intervention versus control) as a fixed effect. The interaction term between time and group was used to test if change in prescription rates from baseline to study year differed between the intervention and control groups. Estimates of variance components were obtained using restricted maximum likelihood method. Model fit was assessed using deviance. Analyses were performed using the GLIMMIX macro in SAS version 8. In addition to the overall (unstratified) analysis, a stratified analysis was performed by physician specialty (pediatrics versus adult medicine), and specific URI diagnosis (pharyngitis, bronchitis, otitis media, and URIs nos). Prescribing Rate Baseline Year Study Year % Change P valueBaseline Year Study Year % ChangeP value All Physicians49.937.6-24.6<0.00014542.8-4.90.25 Internists62.645.1-27.9<0.000175.372.2-4.1%0.42 Pediatricians37.127.5-25.9<0.000131.429.9-4.8%0.35 Table 1. Changes in Antibiotic Prescribing Rates Among Internists and Pediatricians Intervention Group Control Group Findings and Results Total number of physicians included in the evaluation for the study population was 21 (9 pediatricians and 12 internists); control population had 9 physicians (6 pediatricians and 3 internists). A number of physicians were excluded because they had not seen a minimum of 10 patients with a given diagnosis. Patient encounters included in the analysis: baseline year (4429 in study population and 1970 in control population); study year (3338 in study population and 1688 in the control population). Tables 1 and 2 and Figures 1 and 2 summarize the overall findings for the changes in prescribing practices among physicians in both the intervention and control groups as well as the changes in prescribing for specific URIs..  An evaluation of patient gender showed that there was no effect on prescription rates of antibiotics for URIs among the physicians in either the intervention or control groups. Similarly, years since residency training had been completed had no effect on prescription rates by physicians.  Return office visits were tracked within 30 days of the initial diagnosis to evaluate whether the decreases in antibiotic prescription rates resulted in complication of diseases. In the baseline year the mean number of follow-up visits was 1.54 in the intervention group and 1.51 in the control group, with standard deviations of 1.10 and 1.05, respectively. In the study year, the mean was 1.26 follow-up visits in the intervention group and 1.33 in the control group, with standard deviations of 0.66 and 0.67, respectively. Thus, the follow-up visits of the study groups did not differ and declined from the baseline year to the study year. 2. The study included an evaluation of both pediatric and adult patient populations, as well as an evaluation of male and female patients. The study found that there were no differences among antibiotic prescription rates for male and female patients, and that, overall, antibiotic prescription rates were lower among pediatricians prescribing for children (<15 years old). 3. Tracking of follow-up visits showed that the average number of follow-up visits were very similar between the intervention and control populations in both the baseline and study years. This indicates that the decreased prescription of antibiotics for the treatment of URIs within the intervention group did not result in an increase in the number of return visits. This study of the effects of educational intervention has confirmed the results of previous studies of educational interventions to reduce antimicrobial use and has demonstrated the effectiveness of an educational intervention with physicians and their staff in a large, inner- city, staff-model HMO primary care clinic. Although the overall reduction of antimicrobial use by 24.6% among the intervention population is a large step forward in reducing inappropriate use, there is still much room for improvement. In this study, following the educational intervention, nearly 80% of patients who were diagnosed as having bronchitis still received antibiotic prescriptions. Because bronchitis is largely a viral disease, few patients benefit from treatment with antimicrobial therapy and this number should have been close to zero. Similarly, because streptococcal pharyngitis, which does require antimicrobial treatment, is relatively rare in adults, the number of prescriptions of antibiotics for pharyngitis in adults should have been lower. Limitations  Potential for coding errors - data were derived from administrative data files for medical billing  One of the pediatric physicians in the control group inadvertently participated in the educational program. Baseline prescribing rates of antibiotics for treating URI nos for this physician were very low (5%), and changed to 4% in the study year. The overall effect of this physician’s participation in the intervention on the control group is not known, but would have likely served to limit the differences in prescribing rates between the study and control populations.  This study was conducted in partnership with a Medicaid HMO in Detroit, Michigan. Because the patients at these clinics are of lower socioeconomic status, it is unknown whether similar effects would be seen among patients of higher socioeconomic status.  The intervention was conducted during the winter of 2000-2001, when anti-pneumococcal vaccine for children became widely available. We were not able to track immunization rates among the patients in the study and are not able to account for potential decreases in patient encounters or reductions in antimicrobial prescriptions that may have resulted due to the vaccinations. In addition to the public health implications of unnecessary antimicrobial use, cost considerations are significant. In 2001, the cost of prescription drugs rose 17%. From 2001- 2002, the use of prescription drugs accounted for 22% of the total rise in gross healthcare expenditures in the US. In light of the significant impact of prescription medication costs on healthcare expenditures, the development of methods to reduce the use of prescription medications without compromising patient care or outcomes will take on great significance. Conclusions and Implications Recognizing the rise in the rates of antibiotic resistance in Michigan, particularly among S. pneumoniae, we have demonstrated the efficacy of a physician and health care provider educational intervention on reducing antibiotic use in the treatment of uncomplicated viral URIs. This study demonstrated a statistically significant decline of 24.6% in the number of prescriptions for URIs among the intervention clinic physicians. Both the intervention and control groups exhibited a decline in the number of patient encounters for URIs and the number of prescriptions for treating URIs between the baseline and study years. The decline in the control group was likely due to the mild flu and cold season that occurred during the winter of 2000-2001. The decreases in prescriptions for treating URIs may also reflect the impact of both national and local efforts to promote appropriate antibiotic use. 1.14 of the 21 physicians in the intervention group whose antimicrobial prescribing patterns were evaluated in this study participated in the educational program. No significant difference was found between the changes in antimicrobial prescribing patterns of the seven physicians who did not receive the education, compared to the 14 who did attend.  Improvements seen in the antimicrobial prescribing for treating probable viral URIs may have been due to education of other members of the health care staff, as well as peer influence of physicians in the clinic who did participate in the sessions. This finding is particularly interesting, given that the baseline prescribing rates of the physicians who participated in the educational sessions were lower than those of the physicians who did not participate in the program. Prescribing Rate Baseline Year Study Year % Change P valueBaseline Year Study Year % ChangeP value Pharyngitis82.873.9-10.70.00782.674.5-9.80.053 Bronchitis79.678.2-1.80.8474.581.69.50.15 Otitis Media84.374.1-12.1<0.000191.685.2-7.00.007 URI Nos26.116.9-35.2<0.000116.817.96.50.58 Table 2. Changes in Antibiotic Prescribing Rates For Treating Specific URIs Intervention Group Control Group Figure 1. Comparison of Antimicrobial Prescribing at Intervention Clinics: Baseline and Study Years Figure 2. Comparison of Antimicrobial Prescribing at Control Clinics: Baseline and Study Years