Predictive value of ICD-9-CM codes used in vaccine safety research.

OBJECTIVES: To assess how well selected ICD-9-CM diagnosis codes predict adverse events; to model bias and power loss when vaccine safety analyses rely on unverified codes. METHODS: We extracted chart verification data for ICD-9-CM diagnosis codes from six Vaccine Safety Datalink (VSD) publications and modeled biases and power losses using positive predictive value (PPV) estimates and ranges of code sensitivity. RESULTS: Positive predictive values were high for type 1 diabetes (80%) in children, relative to WHO criteria, and intussusception (81%) in young children, relative to a standard published case definition. PPVs were moderate (65%) for inpatient and emergency department childhood seizures and low (21%) for outpatient childhood seizures, both relative to physician investigator judgment. Codes for incident central nervous system demyelinating disease in adults had high PPV for inpatient codes (80%) and low PPV for outpatient codes (42%) relative to physicians' diagnoses. Modeled biases were modest, but large increases in frequencies of adverse events are required to achieve adequate power if unverified ICD-9-CM codes are used, especially when vaccine associations are weak. CONCLUSIONS: ICD-9-CM codes for type 1 diabetes in children, intussusception in young children, childhood seizures in inpatient and emergency care settings, and inpatient demyelinating disease in adults were sufficiently predictive for vaccine safety analyses to rely on unverified diagnosis codes. Adverse event misclassification should be accounted for in statistical power calculations.

Trivalent inactivated influenza vaccine safety in children: assessing the contribution of telephone encounters.

We assessed the contribution of telephone medical care encounters to surveillance of adverse events (AE) following trivalent influenza vaccination in children age 6 months to 17 years. We used retrospective, self-controlled, case-series analysis to estimate adverse event incidence rate ratios for post-vaccination risk intervals relative to 15-28 days prior to vaccination. We confirmed possible vaccination reactions by medical record abstraction. Detection of 10 of 20 elevated incidence rate ratios required telephone data. We conclude that telephone encounters substantially contribute to the detection of possible influenza vaccination reactions, primarily local injection site and systemic reactions.

Influence of clinical outcome and outcome period definitions on estimates of absolute clinical and economic benefits of influenza vaccination in community dwelling elderly persons.

Studies assessing the clinical and economic benefits of vaccination in the elderly have used different clinical outcomes (e.g. hospitalizations for pneumonia or influenza versus hospitalizations for respiratory and cardiovascular causes) and different outcome periods (e.g. peak versus total influenza season) on which to base estimates of clinical effectiveness and cost effectiveness. We explored the implications of these varying approaches by comparing two health economic analysis models of influenza vaccination of community-dwelling elderly persons. We developed computerized models using clinical data from 3 large US HMOs for the 1998-1999 and 1999-2000 influenza seasons. The primary health economic model used a broad definition of clinical events and outcome period and included hospitalizations for all respiratory and cardiovascular events that occurred during the entire influenza season. The alternative model used more restrictive definitions and included pneumonia or influenza hospitalizations occurring during the peak influenza season. The results of Monte Carlo simulation showed that, with the more inclusive primary model, influenza vaccination resulted in net medical care cost savings due to fewer respiratory or cardiovascular hospitalizations of Dollars 71/person vaccinated (5th-95th percentile Dollars 32-118) and net savings of Dollars 809/year of life saved (5th-95th percentile Dollars 331-1450). In contrast, the alternate model found costs of Dollars 3.50/person vaccinated (5th-95th percentile Dollars -11 to 5) and net costs of Dollars 91/year of life saved (5th-95th percentile Dollars -309 to 126). Our findings confirm that influenza vaccination of the elderly is most likely cost saving and supports policies and programs that advocate routine immunization of all persons 65 and older. They also highlight how different outcome definitions can influence the results of health economic analyses.

Incidence of herpes zoster, 1997-2002.

We estimated age-specific herpes zoster (HZ) incidence rates in the Kaiser Permanente Northwest Health Plan (KPNW) during 1997-2002 and tested for secular trends and differences between residents of two states with different varicella vaccine coverage rates. The cumulative proportions of 2-year-olds vaccinated increased from 35% in 1997 to 85% in 2002 in Oregon, and from 25% in 1997 to 82% in 2002 in Washington. Age-specific HZ incidence rates in KPNW during 1997-2002 were compared with published rates in the Harvard Community Health Plan (HCHP) during 1990-1992. The overall HZ incidence rate in KPNW during 1997-2002 (369/100,000 person-years) was slightly higher than HCHP's 1990-1992 rate when adjusted for age differences. For children 6-14 years old, KPNW's rates (182 for females, 123 for males) were more than three times HCHP's rates (54 for females, 39 for males). This increase appears to be associated with increased exposure of children to oral corticosteroids. The percentage of KPNW children exposed to oral corticosteroids increased from 2.2% in 1991 to 3.6% in 2002. Oregon residents had slightly higher steroid exposure rates during 1997-2002 than Washington residents. There were significant increases in HZ incidence rates in Oregon and Washington during 1997-2002 among children aged 10-17 years, associated with increased exposure to oral steroids.

Quality assessments of HMO diagnosis databases used to monitor childhood vaccine safety.

OBJECTIVE: To assess the quality of automated diagnoses extracted from medical care databases by the Vaccine Safety Datalink (VSD) study. METHODS: Two methods are used to assess quality of VSD diagnosis data. The first method compares common automated and abstracted diagnostic categories ("outcomes") in 1-2% simple random samples of study populations. The second method estimates positive predictive values of automated diagnosis codes used to identify potential cases of rare conditions (e.g., acute ataxia) for inclusion in nested case-control medical record abstraction studies. RESULTS: There was good agreement (64-68%) between automated and abstracted outcomes in the 1-2% simple random samples at 3 of the 4 VSD sites and poor agreement (44%) at 1 site. Overall at 3 sites, 56% of children with automated cerebella ataxia codes (ICD-9 = 334) and 22% with "lack of coordination" codes (ICD-9 = 781.3) met objective clinical criteria for acute ataxia. CONCLUSIONS: The misclassification error rates for automated screening outcomes substantially reduce the power of screening analyses and limit usefulness of screening analyses to moderate to strong vaccine-outcome associations. Medical record verification of outcomes is needed for definitive assessments.

Medically recorded allergies and the risk of childhood acute lymphoblastic leukaemia.

Data on five allergic conditions were abstracted from the medical records of 180 cases of childhood acute lymphoblastic leukaemia (ALL) and 718 matched controls. Odds Ratios (OR) and 95% Confidence Intervals (CI) were estimated for composite variables and for individual allergies using conditional logistic regression modelling. Allergies were divided into late and early diagnoses (those made within the year before the matched case's ALL diagnosis and those made earlier, respectively). Among the early diagnoses, atopy or hives was significantly associated with ALL (OR=2.20; 95% CI: 1.16-4.16). Significant associations were found for late diagnoses of atopy or hives (OR=3.78; 95% CI: 1.00-14.29) and of asthma (OR=3.10; 95% CI: 1.39-6.95). None of the other allergic conditions were associated with ALL. These results are contrary to those of prior studies of childhood ALL and allergy.

Drinking patterns, health care utilization, and costs among HMO primary care patients.

A survey of 8,034 primary care patients in a health maintenance organization examined the relationship between alcohol consumption and health care costs and service use. Costs were estimated from service use data for 1 year before and 2 years after study enrollment. No strong, consistent relationships were identified between multiple indicators of drinking patterns and either health care costs or service use. Compared with total costs among very light drinkers, former drinkers were higher, lifetime abstainers were similar, and persons in the higher drinking levels tended to have lower but not significantly different costs. Drinking patterns did not appear to be an important predictor of short-term health care costs or service use in this setting. Further study of former drinkers is warranted to examine the role of alcohol-related illnesses in the decision to quit drinking.

Childhood vaccinations, vaccination timing, and risk of type 1 diabetes mellitus.

OBJECTIVES: To evaluate suggested associations between childhood vaccinations, particularly against hepatitis B and Haemophilus influenzae type b, and risk of developing type 1 diabetes; and to determine whether timing of vaccination influences risk. METHODS: We conducted a case-control study within 4 health maintenance organizations (HMOs) that participate in the Vaccine Safety Datalink project of the Centers for Disease Control and Prevention. Study eligibility was restricted to children who met the following criteria: 1) born during 1988 through 1997; 2) HMO member since birth; 3) continuously enrolled for first 6 months of life; and 4) at least 12 months of HMO membership before diabetes incidence date (or index date for controls) unless incidence date was before 12 months of age. All 4 HMOs maintain registries of their members who have diabetes, and we used the registries to identify potential cases of diabetes. We conducted chart reviews to verify that potential cases met the World Health Organization epidemiologic case definition for type 1 diabetes mellitus (ie, a physician's diagnosis of diabetes plus treatment with daily insulin injections). We defined the incidence date of diabetes as the first date that the child received a diagnosis of diabetes. We attempted to match 3 controls to each case. Controls had the same eligibility criteria as cases and were matched to individual cases on HMO, sex, date of birth (within 7 days), and length of health plan enrollment (up to the incidence or index date). The index date for controls was defined as the incidence date of the case to which the control was matched. Chart abstraction was performed by trained chart abstractors using standardized forms. In addition to complete vaccination histories, the chart abstraction forms for both cases and controls included information on sociodemographic characteristics, selected medical conditions, history of breastfeeding, and family medical history. We used conditional logistic regression to estimate the odds ratio (OR) of diabetes associated with vaccination, with vaccine exposure defined as before the diabetes incidence date (or index date for controls). RESULTS: Two hundred fifty-two confirmed cases of diabetes and 768 matched controls met the study eligibility criteria. The OR (95% confidence interval) for the association with type 1 diabetes was 0.28 (0.07-1.06) for whole cell pertussis vaccine (predominantly in combination as diphtheria, tetanus toxoids and pertussis vaccine), 1.36 (0.70-2.63) for measles-mumps-rubella, 1.14 (0.51-2.57) for Haemophilus influenzae type b, 0.81 (0.52-1.27) for hepatitis B vaccine, 1.16 (0.72-1.89) for varicella vaccine, and 0.92 (0.53-1.57) for acellular pertussis-containing vaccines. Compared with children who had not received hepatitis B vaccine, the OR of diabetes was 0.51 (0.23-1.15) for children vaccinated at birth and 0.86 (0.54-1.35) for those first vaccinated against hepatitis B at 2 months of age or later. Race and ethnicity and family history of diabetes were independently associated with risk of type 1 diabetes, but adjustment for these factors did not materially alter the ORs for any of the vaccines. CONCLUSIONS: In this large, population-based, case-control study, we did not find an increased risk of type 1 diabetes associated with any of the routinely recommended childhood vaccines. Our study adds to previous research by providing data on newer vaccines, including hepatitis B, acellular pertussis, and varicella vaccines. For the older vaccines, our results are generally in agreement with previous studies in not finding any increased risks. Ours is the first epidemiologic study to evaluate the possibility that timing of vaccination is related to risk of clinical diabetes in children. Our results on hepatitis B vaccine do not support the hypothesis; risk of type 1 diabetes was not different between infants vaccinated at birth and those who received their first vaccination later in life. The results of our study and the preponderance of epidemiologic evidence do not support an association between any of the recommended childhood vaccines and an increased risk of type 1 diabetes. Suggestions that diabetes risk in humans may be altered by changes in the timing of vaccinations also are unfounded.

The risk of seizures after receipt of whole-cell pertussis or measles, mumps, and rubella vaccine.

BACKGROUND: The administration of the diphtheria and tetanus toxoids and whole-cell pertussis (DTP) vaccine and measles, mumps, and rubella (MMR) vaccine has been associated with adverse neurologic events, including seizures. We studied the relation between these vaccinations and the risk of a first seizure, subsequent seizures, and neurodevelopmental disability in children. METHODS: This cohort study was conducted at four large health maintenance organizations and included reviews of the medical records of children with seizures. We calculated the relative risks of febrile and nonfebrile seizures among 679,942 children after 340,386 vaccinations with DTP vaccine, 137,457 vaccinations with MMR vaccine, or no recent vaccination. Children who had febrile seizures after vaccination were followed to identify the risk of subsequent seizures and other neurologic disabilities. RESULTS: Receipt of DTP vaccine was associated with an increased risk of febrile seizures only on the day of vaccination (adjusted relative risk, 5.70; 95 percent confidence interval, 1.98 to 16.42). Receipt of MMR vaccine was associated with an increased risk of febrile seizures 8 to 14 days after vaccination (relative risk, 2.83; 95 percent confidence interval, 1.44 to 5.55). Neither vaccination was associated with an increased risk of nonfebrile seizures. Analyses of automated data alone gave results similar to the analyses of the data from medical-record reviews. The number of febrile seizures attributable to the administration of DTP and MMR vaccines was estimated to be 6 to 9 and 25 to 34 per 100,000 children, respectively. As compared with other children with febrile seizures that were not associated with vaccination, the children who had febrile seizures after vaccination were not found to be at higher risk for subsequent seizures or neurodevelopmental disabilities. CONCLUSIONS: There are significantly elevated risks of febrile seizures on the day of receipt of DTP vaccine and 8 to 14 days after the receipt of MMR vaccine, but these risks do not appear to be associated with any long-term, adverse consequences.

Influenza vaccine effectiveness in preventing hospitalizations and deaths in persons 65 years or older in Minnesota, New York, and Oregon: data from 3 health plans.

This study developed methods and determined the impact of influenza vaccination on elderly persons in 3 large health plans: Kaiser Permanente Northwest, HealthPartners, and Oxford Health Plans. Data for the 1996-1997 and 1997-1998 seasons were extracted from administrative databases. Subjects were health plan members > or = 65 years old. Comorbid conditions collected from the preceding year were used for risk adjustment with logistic regression. The virus-vaccine match was excellent for year 1 and fair for year 2. Both years, during peak and total periods, vaccination reduced all causes of death and hospitalization for pneumonia and influenza: hospitalizations were reduced by 19%-20% and 18%-24% for years 1 and 2, respectively, and deaths were reduced by 60%-61% and 35%-39% for the same periods. These results show that all elderly persons should be immunized annually for influenza. The methods used in this study are an efficient cost-effective way to study vaccine impact and similar questions.

Measles-mumps-rubella and other measles-containing vaccines do not increase the risk for inflammatory bowel disease: a case-control study from the Vaccine Safety Datalink project.

CONTEXT: A link between measles virus-containing vaccines and inflammatory bowel disease (IBD) has been suggested by recent studies. OBJECTIVE: To address whether receipt or timing of measles-containing vaccine (MCV) increases risk for IBD. DESIGN: A case-control study. SETTING: Four large health maintenance organizations (HMOs) that are part of the Centers for Disease Control and Prevention's Vaccine Safety Datalink project. PATIENTS OR OTHER PARTICIPANTS: A total of 155 persons with codes from International Classification of Diseases, Ninth Revision specific for IBD, born between 1958 and 1989 and enrolled from birth to the onset of disease, were identified. Up to 5 controls were matched by sex, HMO, and birth year. INTERVENTION: None. MAIN OUTCOME MEASURES: Risk for IBD, Crohn's disease, and ulcerative colitis. RESULTS: Past vaccination was not associated with an increased risk for Crohn's disease (odds ratio [OR] for measles-mumps-rubella vaccine [MMR], 0.4; 95% confidence interval [CI], 0.08-2.0), ulcerative colitis (OR, 0.8; 95% CI, 0.18-3.56), or IBD (OR, 0.59; 95% CI, 0.21-1.68). Risk for IBD was not increased among children vaccinated who were younger than 12 months (OR for MMR, 0.61; 95% CI, 0.15-2.45) or aged 12 to 18 months (OR, 0.86; 95% CI, 0.28-2.59) relative to unvaccinated children. Children vaccinated with MMR who were older than 18 months were at significantly decreased risk for IBD (OR, 0.16; 95% CI, 0.04-0.68). Neither past vaccination nor age at vaccination with other MCV was associated with increased risk for Crohn's disease, ulcerative colitis, or IBD. Risk for Crohn's disease, ulcerative colitis, or IBD was not elevated in the time immediately following vaccination with either vaccine. CONCLUSIONS: Vaccination with MMR or other MCV, or the timing of vaccination early in life, did not increase the risk for IBD.

Does influenza vaccination prevent asthma exacerbations in children?

OBJECTIVE: Influenza can exacerbate asthma, particularly in children. The effectiveness of influenza vaccine in preventing influenza-related asthma exacerbations, however, is not known. We evaluated influenza vaccine effectiveness in protecting children against influenza-related asthma exacerbations. STUDY DESIGN: We conducted a population-based retrospective cohort study with medical and vaccination records in 4 large health maintenance organizations in the United States during the 1993-1994, 1994-1995, and 1995-1996 influenza seasons. We studied children with asthma who were 1 through 6 years of age and who were identified by search of computerized databases of medical encounters and pharmacy dispensings. Main outcome measures were exacerbations of asthma evaluated in the emergency department or hospital. RESULTS: Unadjusted rates of asthma exacerbations were higher after influenza vaccination than before vaccination. After adjustment was done for asthma severity by means of a self-control method, however, the incidence rate ratios of asthma exacerbations after vaccination were 0.78 (95% CI: 0.55 to 1.10), 0.59 (0.43 to 0.81), and 0.65 (0.52 to 0.80) compared with the period before vaccination during the 3 influenza seasons. CONCLUSIONS: After controlling for asthma severity, we found that influenza vaccination protects against acute asthma exacerbations in children.

Maternal smoking cessation and relapse prevention during health care visits.

BACKGROUND: Although effects of maternal smoking during pregnancy could be alleviated if women quit early in pregnancy, most do not. Relapse rates among quitters are high. OBJECTIVE: To test the effects of a low-intensity, smoking-cessation/relapse-prevention intervention delivered by clinic staff and providers and based on stages-of-change constructs of the transtheoretical model and brief motivational interviewing techniques. METHODS: A quasi-experimental prospective cohort design employed in obstetric, in-patient, and pediatric care delivery settings of a large health maintenance organization in Portland, Oregon. Subjects were pregnant smokers registered for their first prenatal visit. Primary outcome measures were sustained (self-reported) quit rates during pregnancy and smoking abstinence between 6 and 12 months after delivery. RESULTS: Regression analyses found statistically significant improvement for intervention women in sustained pregnancy quit rates (OR=2.7, CI=1. 2-5.7) and on smoking abstinence between 6 and 12 months after delivery (OR=2.4, CI=1.1-5.3). CONCLUSIONS: While these outcomes are based on self-report only, they emerged despite variable delivery of the intervention across clinics and represent clinically meaningful improvements in rates of nonsmoking. The intervention supports women who want to quit smoking during pregnancy and improves the likelihood of their remaining nonsmokers for the long term.

The incidence of Kawasaki syndrome in West Coast health maintenance organizations.

BACKGROUND: Kawasaki syndrome (KS) causes an acute vasculitis of unknown etiology. It is a leading cause of acquired heart disease of children in Japan and the United States. METHODS: We examined the incidence of KS in a well-defined population group of children < or =6 years of age, using data collected through the Vaccine Safety Datalink (VSD) project. The VSD database contains information on >1 million children enrolled in four West Coast health maintenance organizations (HMOs). RESULTS: During 1993 through 1996 a total of 234 physician-diagnosed KS patients were reported in the 4 HMOs; 152 (65.0%) were boys and 195 (83.3%) were <5 years of age. The incidence of KS among children <5 years of age in the HMOs ranged from 9.0 to 19.1 per 100,000 person years. KS incidence was higher among boys in 3 of the sites. In the 2 sites with the highest number of KS patients, a seasonal occurrence of KS in winter and early spring was observed. Overall 226 (96.6%) of the KS patients were reported to have been hospitalized; hospitalization rates for children <5 years of age ranged from 9.0 to 16.8 per 100,000 person years. CONCLUSIONS: The incidence of KS in the HMOs was similar to that reported in other population-based studies in the United States and higher than estimates for Australia and several European countries.

Does influenza vaccination exacerbate asthma? Analysis of a large cohort of children with asthma. Vaccine Safety Datalink Team.

CONTEXT: Although influenza vaccination is recommended for children with asthma, only a minority are vaccinated. One reason for low influenza vaccine coverage among children with asthma may be concern that influenza vaccination may induce an exacerbation of asthma. OBJECTIVE: To evaluate the safety of influenza vaccination in children with asthma, we studied the incidence of hospitalizations and emergency department visits for asthma following influenza vaccination. DESIGN: Retrospective cohort study-analysis of population-based computerized medical and vaccination records. SETTING: : Four large health maintenance organizations on the West Coast of the United States. SUBJECTS: Children with asthma 1 through 6 years of age, identified by search of computerized databases of medical encounters and pharmacy prescriptions. MAIN OUTCOME MEASURES: Exacerbations of asthma. RESULTS: In unadjusted analyses vaccination was associated with high rates of asthma exacerbations. However, after adjusting for asthma severity using a self-control method, the incidence rate ratios of asthma exacerbations after vaccination were 0.58 (95% confidence interval, 0.36-0.95), 0.74 (95% confidence interval, 0.47-1.17), and 0.98 (95% confidence interval, 0.76-1.27) during the 3 influenza seasons. CONCLUSIONS: After controlling for asthma severity, we found that influenza vaccination does not result in acute asthma exacerbations in children. Concern about possible exacerbation of asthma is not a valid reason to not vaccinate children with asthma against influenza.

The Vaccine Safety Datalink: immunization research in health maintenance organizations in the USA.

The Vaccine Safety Datalink is a collaborative project involving the National Immunization Program of the Centers for Disease Control and Prevention and several large health maintenance organizations in the USA. The project began in 1990 with the primary purpose of rigorously evaluating concerns about the safety of vaccines. Computerized data on vaccination, medical outcome (e.g. outpatient visits, emergency room visits, hospitalizations, and deaths) and covariates (e.g. birth certificates, census data) are prospectively collected and linked under joint protocol at multiple health maintenance organizations for analysis. Approximately 6 million persons (2% of the population of the USA) are now members of health maintenance organizations participating in the Vaccine Safety Datalink, which has proved to be a valuable resource providing important information on a number of vaccine safety issues. The databases and infrastructure created for the Vaccine Safety Datalink have also provided opportunities to address vaccination coverage, cost-effectiveness and other matters connected with immunization as well as matters outside this field.

Influenza vaccination in children with asthma in health maintenance organizations. Vaccine Safety Datalink Team.

We assessed vaccination coverage and predictors of influenza vaccination in asthmatic children in four large Health Maintenance Organizations. We studied 68,839 children with asthma at four Health Maintenance Organizations (HMOs) in the 1995-1996 influenza season and 34,032 children at two HMOs in the 1996-1997 influenza season. In both seasons only 9-10% were vaccinated against influenza. Children who were hospitalized, had an emergency department visit for asthma or a prescription for a beta-agonist prior to the influenza season, were more likely to be vaccinated. Overall, 61% of the unvaccinated asthmatic children had made an outpatient clinic visit during months when influenza vaccination would have been appropriate. Vaccination coverage could be increased by taking advantage of all opportunities to vaccinate children with asthma whenever they make clinic visits in the fall and early winter.

The validity of medicaid pharmacy claims for estimating drug use among elderly nursing home residents: The Oregon experience.

This study compared computerized Medicaid pharmacy claims data for nursing home residents with chart data to establish how well the claims data identified those receiving drugs within three different psychoactive drug classes (yes/no for each class) and how well the claims estimated total within-class average daily dose. Percent agreement, positive predictive value (PPV), and negative predictive value (NPV) for drug exposure were over 85% for each class. Kappas were excellent for antipsychotics and antidepressants, and good for anxiolytics. Correspondence was lower for average daily dose. Using an algorithm that credits some but not all doses associated with overlapping claims, correlations ranged from 0.97 to 0.66. Agreement on therapeutic dose was excellent for antipsychotics (kappa = 0.81) and fair to good for antidepressants and anxiolytics (kappa = 0.63, and kappa = 0.52, respectively). The findings suggest that Medicaid pharmacy claims data are reasonably accurate for quality assurance and accreditation purposes.

Controlled trials of CQI and academic detailing to implement a clinical practice guideline for depression.

BACKGROUND: The release of the Agency for Health Care Policy and Research (AHCPR)'s Guideline for the Detection and Treatment of Depression in Primary Care created an opportunity to evaluate under naturalistic conditions the effectiveness of two clinical practice guideline implementation methods: continuous quality improvement (CQI) and academic detailing. A study conducted in 1993-1994 at Kaiser Permanente Northwest Division, a large, not-for-profit prepaid group practice (group-model) HMO, tested the hypotheses that each method would increase the number of members receiving depression treatment and would relieve depressive symptoms. METHODS: Two trials were conducted simultaneously among adult primary care physicians, physician assistants, and nurse practitioners, using the same guideline document, measurement methods, and one-year follow-up period. The academic detailing trial was randomized at the clinician level. CQI was assigned to one of the setting's two geographic areas. To account for intraclinician correlation, both trials were evaluated using generalized equations analysis. RESULTS: Most of the CQI team's recommendations were not implemented. Academic detailing increased treatment rates, but--in a cohort of patients with probable chronic depressive disorder--it failed to improve symptoms and reduced measures of overall functional status. CONCLUSIONS: New organizational structures may be necessary before CQI teams and academic detailing can substantially change complex processes such as the primary care of depression. New research and treatment guidelines are needed to improve the management of persons with chronic or recurring major depressive disorder.