Agreement Between Automated Pharmacy Data and Self-Reported Medication Use by Drug Benefit Status and Type of Delivery System
Boudreau DM, Doescher MP, Saver BG, Jackson JE, Fishman PA
Denise M. Boudreau, PhD, Group Health Cooperative-Center for Health Studies, 1730 Minor Avenue, Suite 1600, Seattle, WA 98101-1448. E-mail: email@example.com.
Background: Pharmacy utilization data captured in health maintenance organization (HMO) information systems are used for a variety of research purposes. Certain characteristics such as benefit status and type of healthcare delivery system may influence where enrollees fill prescription medications and therefore whether pharmacy utilization data are captured in the database. Yet, few studies have evaluated the accuracy of using automated pharmacy records to categorize exposure of specific medication classes and no studies have evaluated the completeness of these records in relationship to drug benefit status and type of delivery system.
Objective: To describe where seniors enrolled in a mixed-model HMO obtain prescription medications and how source of prescription medications is related to drug benefit status and type of healthcare delivery system. In addition, the study compares the agreement between automated pharmacy records and self-reported medication use for a variety of medication classes among seniors with and without a drug benefit and in a staff versus network model delivery system.
Methods: Subjects were men and women aged 67 and above as of March 2000, continuously enrolled in Group Health Cooperative's (GHC) Medicare + Choice program during the years of 1998-1999, and diagnosed with 1 or more of the following chronic conditions: hypertension, diabetes, congestive heart failure, and coronary artery disease. Sources of prescription medications were obtained through a mailed survey. Current medication use was identified from a self-reported survey and automated pharmacy records. Information on drug benefit status and type of delivery system was obtained from GHC automated information systems.
Results: Where seniors obtain prescription medications is influenced by drug benefit status and type of healthcare delivery system. Overall, moderate to substantial agreement (kappa statistic) was found between pharmacy records and self-reported use for medication categories defined as hormone replacement therapy, cardiac agents, antihypertensives, diabetic agents, statins, and antidepressants. Agreement varied by benefit status and type of delivery system for numerous of the drug categories.
Conclusion: Automated pharmacy records are a valid source for evaluating medication use. However, health benefit status and type of healthcare delivery system should be considered when using pharmacy records.
Epidemiology of Churg-Strauss Syndrome Among Asthma Drug Users
Harrold LR, Andrade SE, Go AS, et al.
Leslie R. Harrold, MD, MPH, Meyers Primary Care Institute/Fallon Healthcare, 630 Plantation Street, Worcester, MA 01605. E-mail: firstname.lastname@example.org.
Background: Churg-Strauss syndrome (CSS) is a systemic vasculitis primarily affecting patients with asthma, but its epidemiology is poorly understood.
Objective: To estimate the incidence of CSS among a large population of asthma drug users.
Methods: A retrospective study was conducted among patients who had been dispensed asthma drugs at 3 health maintenance organizations. Adults who received ≥3 dispensings of an asthma drug during any consecutive 12-month period between January 1, 1995 and June 30, 2000 were identified. Information on patient age, gender, enrollment status, asthma drugs dispensed, and inpatient and outpatient diagnoses and procedures was obtained from the automated databases. Chart reviews were performed on persons identified by combinations of diagnostic and billing codes indicative of CSS. A rheumatologist reviewed abstracted information on all subjects and those who met 2 American College of Rheumatology criteria for CSS were reviewed by 2 clinical experts. Each clinical expert independently rated the cases; disagreements were resolved by consensus. Cases classified as having "probable/definite" CSS were included in these analyses. The incidence of CSS was estimated overall and according to patient gender, age, and calendar year.
Results: From a population of 184 667 asthma drug users contributing 606 184 person-years of exposure, 22 cases of CSS were identified (overall incidence of 3.6 per 100 000 person years; 95% confidence interval, 2.3-5.5). Incidence rates did not differ by gender and age group. The incidence rates for 1995, 1996, 1997, 1998, 1999 and the first 6 months of 2000 were 1.8, 2.2, 5.2, 7.5, 1.4 and 1.4 per 100 000 person years, respectively.
Conclusions: Results from this population-based study suggest a low incidence of CSS in asthma drug users, although the rate is higher than those previously published and most likely represents the lower limit of the true incidence of CSS, as some cases of CSS may not have been identified using this methodology.
Improving Laboratory Monitoring at Initiation of High Risk Drug Therapy: A Randomized Practical Clinical Trial
Raebel MA, Magid DJ, Lyons EE, Miller C, Chester EA, Bodily MA
Marsha A. Raebel, PharmD, Kaiser Permanente Colorado, PO Box 378066, Denver, CO 80237-8066. E-mail: email@example.com.
Background: Monitoring laboratory parameters is indicated before and during therapy for drugs with adverse events related to lack of monitoring, but medical literature suggests that monitoring of these drugs is performed less often than recommended.
Objective: To test whether a computerized tool alerting pharmacists to missing laboratory orders would be effective in increasing the proportion of patients receiving laboratory monitoring at initiation of high-risk drug therapy.
Methods: We conducted a controlled trial in which we randomized the 375 000 members of Kaiser Permanente of Colorado to either have prescribing and laboratory test information sent electronically to the Kaiser Permanente of Colorado Clinical Pharmacy Call Center (CPCC) (intervention) or to receive usual care (control). For 18 high-risk drugs, CPCC pharmacists received reports of drug dispensings plus drug-specific laboratory tests for which either the test was not a) ordered or b) had not been completed by the patient. The CPCC pharmacists contacted patients to remind them about laboratory tests their providers had ordered and/or they ordered the appropriate test(s). Laboratory results ordered by the CPCC pharmacists were returned to them for evaluation. A guideline for managing abnormal laboratory results was developed and approved. Standardized scripts were used to notify patients' providers about recommended actions.
Results: Results from the first 6 months document 1854 (76.7%) drug dispensings with laboratory monitoring in the intervention group compared to 1630 (67.8%) dispensings with laboratory monitoring in the control group ( < .01). Abnormal results were documented in 10% of laboratory tests performed in the intervention group. Allopurinol (15.7%) and amiodarone (12.8%) had the highest percentages of abnormal patient test results.
Conclusions: This study demonstrates that coupling research, information systems, and physicians' and pharmacists' clinical knowledge and skills is effective in improving laboratory monitoring of high-risk drugs.
Preliminary Results of the CARE Study: NSAID Use and Recurrent Colorectal Cancer
Johnson CC, Rolnick S, Hensley Alford S, et al.
Christine Cole Johnson, PhD, Henry Ford Health System, 1 Ford Place, Suite 5C, Detroit, MI 48202. E-mail: firstname.lastname@example.org.
Background: Chronic inflammation has received increasing attention as a causal factor in many diseases, including gastrointestinal cancers. Evidence has emerged to indicate that use of non-steroidal anti-inflammatory drugs (NSAIDs) is chemopreventive for primary colorectal cancer. We investigate the hypothesis that NSAID use may also be protective for recurrent colorectal cancer.
Methods: In a retrospective cohort study of 1025 patients with primary colon cancer diagnosed between 1990-2000 at 2 healthcare systems (Henry Ford Health System and HealthPartners), we examined the effect of NSAID use (current use versus non-use) on colorectal cancer recurrence. Patients were followed from the date of diagnosis until date of colorectal cancer recurrence, death, or last known follow-up. We classified each day of follow-up according to NSAID use recorded in the prescription claims databases as follows: 1) current use, prescription days supply indicated that patient was actively using NSAID and 2) non-use, person-time accumulating beginning 30 days after a prescription days supply (to allow for wash-out or sporadic use). A minimum of thirty-days supply was required before a patient contributed person-time to current use. We compared the incidence rates of recurrence among current NSAID users to that of non-users.
Results: On average, follow-up was 4.7 years. Preliminary results indicate current NSAID use was associated with a 42% reduction in the rate of recurrence (rate ratio = 0.58; 95% confidence interval, 0.19-1.83). Data collection is still ongoing.
Conclusions: Current NSAIDs use appears in our preliminary results to be protective for recurrent colorectal cancer.