Depletion-of-susceptibles bias in influenza vaccine waning studies: how to ensure robust results.

Vaccine effectiveness studies are subject to biases due to depletion-of-persons at risk of infection, or at especially high risk of infection, at different rates from different groups (depletion-of-susceptibles bias), a problem that can also lead to biased estimates of waning effectiveness, including spurious inference of waning when none exists. An alternative study design to identify waning is to study only vaccinated persons, and compare for each day the incidence in persons with earlier or later dates of vaccination to assess waning in vaccine protection as a function of vaccination time (namely whether earlier vaccination would result in lower subsequent protection compared to later vaccination). Prior studies suggested under what conditions this alternative would yield correct estimates of waning. Here we define the depletion-of-susceptibles process formally and show mathematically that for influenza vaccine waning studies, a randomised trial or corresponding observational study that compares incidence at a specific calendar time among individuals vaccinated at different times before the influenza season begins will not be vulnerable to depletion-of-susceptibles bias in its inference of waning as a function of vaccination time under the null hypothesis that none exists, and will - if waning does actually occur - underestimate the extent of waning. Such a design is thus robust in the sense that a finding of waning in that inference framework reflects actual waning of vaccine-induced immunity. We recommend such a design for future studies of waning, whether observational or randomised.

Safety of zoster vaccine in adults from a large managed-care cohort: a Vaccine Safety Datalink study.

OBJECTIVES: The aim of this study was to examine a large cohort of adults who received the zoster vaccine for evidence of an increased risk of prespecified adverse events requiring medical attention. DESIGN: Two self-comparison approaches, including a case-centred approach and a self-controlled case series (SCCS) analysis were used. SETTING: Eight managed-care organizations participating in the Vaccine Safety Datalink project in the United States. SUBJECTS: A total of 193 083 adults aged 50 and older receiving a zoster vaccine from 1 January 2007 to 31 December 2008 were included. MAIN OUTCOME MEASURES: Prespecified adverse events were identified by aggregated International Classification of Diseases, Ninth Revision (ICD-9) codes in automated health plan datasets. RESULTS: The risk of allergic reaction was significantly increased within 1-7 days of vaccination [relative risk = 2.13, 95% confidence interval (CI): 1.87-2.40 by case-centred method and relative rate = 2.32, 95% CI: 1.85-2.91 by SCCS]. No increased risk was found for the following adverse event groupings: cerebrovascular events; cardiovascular events; meningitis; encephalitis; and encephalopathy; and Ramsay-Hunt syndrome and Bell's palsy. CONCLUSIONS: The results of this study support the findings from the prelicensure clinical trials, providing reassurance that the zoster vaccine is generally safe and well-tolerated with a small increased risk of allergic reactions in 1-7 days after vaccination.

Smoking, menopause, and breast cancer.

As part of a multiphasic health examination, breast cancer incidence was studied in a cohort of 84,172 women who reported their smoking habits between 1964 and 1972. After adjustment for race, education, marital status, body mass, parity, age of menarche, and alcohol consumption, women currently smoking at examination had a slight increase in the relative risk of breast cancer; relative risk for light smokers was 0.95 (95% confidence interval--0.80, 1.11); for moderate smokers, 1.22 (1.05, 1.43); and for heavy smokers, 1.19 (0.88, 1.60) compared with non-smokers. Consistent with the hypothesis that estrogen levels mediate a small protective effect of smoking on the risk of breast cancer, we found that moderate and heavy smoking were associated with a 1.29-year and a 1.48-year reduction in the age of menopause. A 1-year reduction in age of menopause was associated with a 0.97 reduction in the risk of breast cancer. However, there was no indication that the overall effect of smoking on breast cancer was protective.

Cost of care for patients in cancer clinical trials.

BACKGROUND: Information on the costs of medical care for patients enrolled in clinical trials is needed by policymakers evaluating ways to facilitate clinical research in a managed care environment. We examined the direct costs of medical care for patients enrolled in cancer clinical trials at a large health maintenance organization (HMO). METHODS: Costs for 135 patients who entered 22 cancer clinical trials (including 12 breast cancer trials) at Kaiser Permanente in Northern California, from 1994 through 1996 were compared with costs for 135 matched control subjects who were not enrolled in such trials. Cancer registry data and medical charts were used in matching the control subjects to the trial enrollees with respect to cancer site, stage, date of diagnosis, age, sex, and trial eligibility. The direct costs of medical care were compared between trial enrollees and the control subjects for a 1-year period, with data on costs and utilization of services obtained from Kaiser Permanente databases and medical charts. RESULTS: Mean 1-year costs for the enrollees in trials were 10% higher than those for the control subjects ($17 003 per enrollee compared with $15 516 per control subject; two-sided P =.011). The primary component of this difference was a $1376 difference in chemotherapy costs ($4815 per trial enrollee versus $3439 per control subject; two-sided P<.001). Costs for the 11 enrollees in trials that had a bone marrow transplant (BMT) arm were approximately double the costs for their matched control subjects (borderline significance: two-sided P=.054). The $15 041 mean cost for the enrollees in trials without BMT was similar to the $15 186 mean cost for their matched control subjects. CONCLUSIONS: Participation in cancer clinical trials at a large HMO did not result in substantial increases in the direct costs of medical care.

Breast cancer survival and treatment in health maintenance organization and fee-for-service settings.

BACKGROUND: Enrollment in health maintenance organizations (HMOs) has increased rapidly during the past 10 years, reflecting a growing emphasis on health care cost containment. To determine whether there is a difference in the treatment and outcome for female patients with breast cancer enrolled in HMOs versus a fee-for-service setting, we compared the 10-year survival and initial treatment of patients with breast cancer enrolled in both types of plans. METHODS: With the use of tumor registries covering the greater San Francisco-Oakland and Seattle-Puget Sound areas, respectively, we obtained information on the treatment and outcome for 13,358 female patients with breast cancer, aged 65 years and older, diagnosed between 1985 and 1992. We linked registry information with Medicare data and data from the two large HMOs included in the study. We compared the survival and treatment differences between HMO and fee-for-service care after adjusting for tumor stage, comorbidity, and sociodemographic characteristics. RESULTS: In San Francisco-Oakland, the 10-year adjusted risk ratio for breast cancer deaths among HMO patients compared with fee-for-service patients was 0.71 (95% confidence interval [CI] = 0.59-0.87) and was comparable for all deaths. In Seattle-Puget Sound, the risk ratio for breast cancer deaths was 1.01 (95% CI = 0.77-1.33) but somewhat lower for all deaths. Women enrolled in HMOs were more likely to receive breast-conserving surgery than women in fee-for-service (odds ratio = 1.55 in San Francisco-Oakland; 3.39 in Seattle). HMO enrollees undergoing breast-conserving surgery were also more likely to receive adjuvant radiotherapy (San Francisco-Oakland odds ratio = 2.49; Seattle odds ratio = 4.62). CONCLUSIONS: Long-term survival outcomes in the two prepaid group practice HMOs in this study were at least equal to, and possibly better than, outcomes in the fee-for-service system. In addition, the use of recommended therapy for early stage breast cancer was more frequent in the two HMOs.

Serum uric acid unrelated to cancer incidence in humans.

We examined the hypothesis that low levels of serum uric acid are associated with elevated cancer risk. A subpopulation (163,830 members) of a large health maintenance organization was followed for a mean of 9.8 years after a multiphasic health checkup at which serum uric acid level was measured. Total cancer incidence as well as site-specific incidence (for lung, colon, and prostate cancer in men and lung, colon, breast, uterine, and cervical cancer in women) was ascertained from hospital discharge records and the Surveillance, Epidemiology and End Results program in the San Francisco-Oakland Bay area. Age-adjusted cancer incidence was not elevated in the lower deciles of serum uric acid level. After adjusting for age, race, education, smoking, alcohol consumption, and body mass, using proportional hazards models, the risk of cancer was not elevated at lower levels of uric acid. Our results suggest that if increased risk of cancer is associated with low serum uric acid, this risk is associated with serum uric levels below those commonly seen in human populations.

Appendectomy, appendicitis, and large bowel cancer.

A cohort of 167,561 persons who received multiphasic health checkups were followed up for cancer development. A history of appendectomy showed slightly negative nonsignificant associations with the development of cancer of the colon, rectum, and all sites combined. By inference, the relation of appendicitis with these cancers was also inverse. Upper 95% confidence limits were compatible only with small positive associations of appendectomy and appendicitis with these cancers. These data do not support the view that removing the appendix increases cancer risk by diminishing immunocompetency. A link between appendicitis and large bowel cancer has been noted in intersociety correlations and has been hypothesized to be due to prevention of both by a high-fiber diet. However, appendicitis does not appear to be a useful predictor of large bowel cancer within a developed society.

Screening prescription drugs for possible carcinogenicity: eleven to fifteen years of follow-up.

Using computerized pharmacy records from 1969 to 1973 for a cohort of 143,574 members of the Kaiser Permanente Medical Care Program, we have been testing associations of 215 drugs or drug groups with subsequent incidence of cancer at 56 sites. This paper presents findings with follow-up through 1984. There were 227 statistically significant (P less than 0.05, two-tailed) associations: 170 positive, 57 negative. Some were undoubtedly chance findings; others were likely due to confounding by unmeasured covariables. However, several associations suggested hypotheses for further studies and/or the need for continued observation. Most notable among findings not previously reported were associations of several antibiotics, both oral and topical, with lung cancer. These associations could not be explained by indications for drug use or by differences in smoking habits between users and nonusers, and suggest a possible link between the occurrence of bacterial infections and risk for cancer. In general, our results continue to suggest that most medications used during that period did not affect cancer incidence substantially. However, for less frequently prescribed medications, our power to detect moderate increases in cancer risk was quite low.

Projected cost-effectiveness of primary angioplasty for acute myocardial infarction.

OBJECTIVES: This study sought to evaluate the cost-effectiveness of primary angioplasty for acute myocardial infarction under varying assumptions about effectiveness, existing facilities and staffing and volume of services. BACKGROUND: Primary angioplasty for acute myocardial infarction has reduced mortality in some studies, but its actual effectiveness may vary, and most U.S. hospitals do not have cardiac catheterization laboratories. Projections of cost-effectiveness in various settings are needed for decisions about adoption. METHODS: We created a decision analytic model to compare three policies: primary angioplasty, intravenous thrombolysis and no intervention. Probabilities of health outcomes were taken from randomized trials (base case efficacy assumptions) and community-based studies (effectiveness assumptions). The base case analysis assumed that a hospital with an existing laboratory with night/weekend staffing coverage admitted 200 patients with a myocardial infarction annually. In alternative scenarios, a new laboratory was built, and its capacity for elective procedures was either 1) needed or 2) redundant with existing laboratories. RESULTS: Under base case efficacy assumptions, primary angioplasty resulted in cost savings compared with thrombolysis and had a cost of $12,000/quality-adjusted life-year (QALY) saved compared with no intervention. In sensitivity analyses, when there was an existing cardiac catheterization laboratory at a hospital with > or = 200 patients with a myocardial infarction annually, primary angioplasty had a cost of < $30,000/QALY saved under a wide range of assumptions. However, the cost/QALY saved increased sharply under effectiveness assumptions when the hospital had < 150 patients with a myocardial infarction annually or when a redundant laboratory was built. CONCLUSIONS: At hospitals with an existing cardiac catheterization laboratory, primary angioplasty for acute myocardial infarction would be cost-effective relative to other medical interventions under a wide range of assumptions. The procedure's relative cost-ineffectiveness at low volumes or redundant laboratories supports regionalization of cardiac services in urban areas. However, approaches to overcoming competitive barriers and close monitoring of outcomes and costs will be needed.

Initial cost of primary angioplasty for acute myocardial infarction.

OBJECTIVES: We sought to evaluate the initial economic cost of primary angioplasty for acute myocardial infarction under varying assumptions about whether a cardiac catheterization laboratory exists, whether services are provided during night and weekend hours and how cardiovascular surgical backup is arranged. BACKGROUND: Primary angioplasty for acute myocardial infarction has resulted in clinical outcomes superior or equal to those obtained with thrombolysis in recent studies, but its future implementation depends greatly on its cost and cost-effectiveness. There is a gap in knowledge about the true economic costs of this procedure, and understanding costs under a variety of hypothetic scenarios is important in planning whether and how the procedure should be offered to broad groups of patients. METHODS: A generalizable spreadsheet model was constructed to calculate the cost of primary angioplasty at a single hospital with assumptions based on data from a large nonprofit health maintenance organization (Kaiser Permanente). The following baseline assumptions were made: 1) A total of 200 patients with myocardial infarction presented to the hospital each year; 2) primary angioplasty was offered for 10 years; 3) the hospital had a cardiac catheterization laboratory; 4) costs of night call for technical personnel and cardiovascular surgical backup were already covered. Other scenarios were modeled to represent different assumptions about existing resources. RESULTS: Under the baseline assumptions, primary angioplasty cost $1,597/procedure. If night call for technical personnel were a new expense, the average cost would be > or = $3,206. If a new cardiac catheterization laboratory needed to be built, costs would range from $3,866 to $14,339/procedure, depending on how cardiovascular surgical backup was provided. Results were sensitive to assumptions about the annual volume of myocardial infarctions, the number of years the procedure was offered and the costs of labor, construction and equipment. CONCLUSIONS: The initial cost of providing primary angioplasty for acute myocardial infarction varies greatly, depending on the setting in which it is provided. To provide information for clinical policy decisions, a cost-effectiveness model is needed that combines these initial costs with data on survival, quality of life and rates and costs of subsequent cardiac procedures.

The cost of acquired immunodeficiency syndrome in northern California. The experience of a large prepaid health plan.

We studied the cost of acquired immunodeficiency syndrome (AIDS) in the Kaiser Permanente Medical Care Program (KPMCP), northern California region. We report the costs of care to the KPMCP and introduce an innovative application of survival methods to cost analysis. From the beginning of the AIDS epidemic in 1981 to the end of June 1987, 866 cases of AIDS were recorded among members of the KPMCP. Estimates of the costs of care of these patients were derived from comprehensive chart reviews of a random sample of 71 patients whose conditions were diagnosed from January 1984 through June 1987. Total mean lifetime costs per patient were $32,816 (median, $28,677), whereas the mean hospital per diem cost was $20,446 per patient. As more care was shifted to outpatient services overtime, overall costs dropped, despite marked increases in the cost of outpatient medications such as zidovudine. The overall estimate of cost compared closely with other estimates of the cost of care in San Francisco, Calif, and it is lower than estimates from elsewhere in the United States, probably because of the low proportion of cases associated with intravenous drug use and the well-developed social support networks available to patients with AIDS in the San Francisco Bay Area.

Evaluating hypertension control in a managed care setting.

BACKGROUND: We conducted a retrospective cohort study on a random sample of adult patients with hypertension in a large health maintenance organization to assess the feasibility of documenting blood pressure (BP) control and to compare different measures for defining BP control. METHODS: Three criteria for BP control were assessed: systolic BP less than 140 mm Hg; diastolic BP less than 90 mm Hg; and combined BP control, with systolic BP less than 140 mm Hg and diastolic BP less than 90 mm Hg. Four methods of assessing hypertension control by the above criteria were examined: proportion of patients with BP under control at 75% and 50% or more of their office visits; the mean of all pressures during the study period; and the BP from the last visit during the study period. RESULTS: The proportion of patients meeting each criterion for control was similar whether we used the mean BP for all visits, the last recorded BP, or control at 50% or more of visits. Control rates were substantially lower when the more stringent assessment, 75% of visits, was used. The proportion of patients with combined BP control at 75% or more of their visits was half that of the other methods. CONCLUSIONS: In this health maintenance organization population, results with the use of the simplest approach, the last BP measurement recorded, were similar to results with the mean BP. Our findings indicate that evaluation of BP control in a large health maintenance organization will find substantial room for improvement, and clinicians should be encouraged to be more aggressive in their management of hypertension, especially with regard to the systolic BP, which until recent years has been underemphasized.

The prevalence of clinically recognized obsessive-compulsive disorder in a large health maintenance organization.

OBJECTIVE: Little is known about the prevalence of obsessive-compulsive disorder (OCD) as recognized in clinical settings. The authors report data on the prevalence of clinically recognized OCD in a large, integrated, group practice health maintenance organization (HMO). METHOD: The authors examined the database of outpatient diagnoses for the 1.7 million people (age >or=6) in the San Francisco Bay Area and Sacramento who were continuously enrolled in Kaiser Permanente from May 1995 through April 1996. OCD diagnoses were confirmed by chart review. RESULTS: The 1-year prevalence of clinically recognized OCD was 84/100,000 (95% confidence interval: 80-89/100,000), or 0.084%. It varied among the 19 clinics within the HMO but was nowhere higher than 150/100,000. Prevalence was higher among women than among men but was higher among boys than among girls. Above age 65, OCD prevalence decreased markedly in both genders. Period prevalence rates increased by 60% as the length of the study period doubled from 1 to 2 years, more than would be expected for a chronic disease requiring regular care. About three-quarters of both children and adults with OCD had comorbid psychiatric diagnoses; major depression was common in both groups. CONCLUSIONS: Although previously reported prevalences of 1%-3% from community studies may have included many transient or misclassified cases of OCD not requiring treatment, the very low prevalence of clinically recognized OCD in this population suggests that many individuals suffering from OCD are not receiving the benefits of effective treatment.

Mental disorders and disability among patients in a primary care group practice.

OBJECTIVE: This article examines social and occupational disability associated with several DSM-IV mental disorders in a group of adult primary care outpatients. METHOD: The subjects were 1,001 primary care patients (aged 18-70 years) in a large health maintenance organization. Data on each patient's sociodemographic characteristics and functional disability, including scores on the Sheehan Disability Scale, were collected at the time of a medical visit. A structured diagnostic interview for current DSM-IV disorders was then completed by a mental health professional over the telephone within 4 days of the visit. RESULTS: The most prevalent disorders were phobias (7.7%), major depressive disorder (7.3%), alcohol use disorders (5.2%), generalized anxiety disorder (3.7%), and panic disorder (3.0%). A total of 8.3% of the patients met the criteria for more than one mental disorder. The proportion of patients with co-occurring mental disorders varied by index disorder from 50.0% (alcohol use disorder) to 89.2% (generalized anxiety disorder). Compared with patients who had a single mental disorder, patients with co-occurring disorders reported significantly more disability in social and occupational functioning. After adjustment for other mental disorders and demographic and general health factors, compared with patients with no mental disorder, only patients with major depressive disorder, bipolar disorder, phobias, and substance use disorders had significantly increased disability, as measured by the Sheehan Disability Scale. CONCLUSIONS: Primary care patients with more than one mental disorder are common and highly disabled. Individual mental disorders have distinct patterns of psychiatric comorbidity and disability.

Stress management and mutual support groups in rheumatoid arthritis.

Stress management and mutual support groups are employed widely in chronic illness, although their efficacy has not been established. To determine the effect of these measures on morbidity and psychologic health in rheumatoid arthritis, 105 patients meeting diagnostic criteria for rheumatoid arthritis were evaluated for depression, life satisfaction, functional disability, and indicators of disease activity. Patients were randomly assigned to one of three groups: (1) stress management; (2) mutual support; (3) no intervention (control). After completion of 10 weekly sessions, identical tests were performed for all patients in the intervention and control groups. Patients in the intervention groups showed greater improvement in joint tenderness than did the control patients but did not differ significantly from the patients in the control group in any of the other outcome measures.

Trends in cardiovascular disease incidence and survival in the elderly.

This study compared the age-specific incidence, postdiagnostic survival, and mortality for cardiovascular disease (CVD) in two cohorts of people aged 65 years and older. All subjects were members of a large prepaid health maintenance organization. The influence of changes in CVD risk factors on these rates also was evaluated. Trends in prevalence, incidence, postdiagnostic survival, and mortality for CVD were examined in both cohorts in 1971 and 1980. Myocardial infarction (MI), angina pectoris, stroke, and congestive heart failure (CHF) were included as CVD outcomes in this analysis. Nine-year prospective data on these diagnoses were abstracted from medical records and computerized hospitalization records for both cohorts. Age-sex-adjusted cardiovascular mortality was lower for both sexes by approximately 20% in the 1980 cohort. Overall survival did not change, whereas cancer mortality increased by 76% in women and 36% in men. With the exception of stroke, there was no increase in age-adjusted or age-specific prevalence. In men, the age-adjusted prevalence of stroke in men was 24% higher in the 1980 cohort. Age-adjusted 9-year incidence of MI, angina pectoris, stroke, and CHF did not change between cohorts in either sex Postdiagnostic, age-adjusted mortality for men with incident stroke was 24% lower in the 1980 cohort, and Postdiagnostic, age-adjusted mortality for men with incident angina was 35% lower in the 1980 cohort. Adjustment for risk factors measured at or before baseline had little influence on cohort differences in CVD incidence or duration of survival after CVD diagnosis. This study confirms other research showing a decline in CVD mortality over the past 20 years. These findings suggest that prevalent angina pectoris is increasing in men, and that survival with stroke and with angina is improving in men. Later diagnosis of incident CHF in men suggests that prevention and early detection may be postponing the development of more serious disease.

Testing for recovery of thyroid function after withdrawal of long-term suppression therapy.

Many persons are given thyroid hormone for long periods without sufficient justification. We selected 45 patients whose chart reviews did not demonstrate a clear need for thyroid replacement. Thyroid hormone was withdrawn and blood samples taken for free thyroxine index and TSH levels at 2, 4, 6, and 8 weeks after thyroid withdrawal. At 8 weeks, 30 of the 45 were normal. A single TSH test 8 weeks after thyroid withdrawal appears to be the most efficient way to separate euthyroid from hypothyroid patients.

Prognosis of gallstones with mild or no symptoms: 25 years of follow-up in a health maintenance organization.

Two-hundred ninety-eight patients with gallstones accompanied by mild or nonspecific symptoms, 123 with asymptomatic gallstones, and 46 with nonfunction on cholecystogram and mild or nonspecific symptoms were followed in the setting of a health maintenance organization for up to 25 years after diagnosis. In a life-table analysis, complications (severe events) developed in about 1% per year of patients with visualized gallstones and in about 2% per year of patients with nonfunction. During each of the first 5 years after diagnosis, all events, both severe and non-severe (including surgery for continuing mild symptoms) occurred in about 6% of the patients with mild symptoms accompanying either gallstones or nonfunction, and in about 4% of patients with asymptomatic gallstones. The annual probabilities for all events tended to decrease as length of follow-up increased. Among patients with stones and mild symptoms, women were more apt to develop any event than men (p = 0.02) and more obese patients were more likely to develop severe events than those who were thinner (p = 0.05). The patients in this study are probably more representative of outpatients with gallstones seen by most primary care physicians than are most groups investigated previously for prognosis. Their event rates are sufficiently similar to those used in published decision analyses, that they should not alter previous conclusions that early elective cholecystectomy will have little positive or negative effect on average life expectancy of patients with gallstones.

Evaluation of ventilating tubes and myringotomy in the treatment of recurrent or persistent otitis media.

In a prospective controlled study of the efficacy and sequelae of ventilating tubes, 44 children with bilateral recurrent acute otitis media (greater than 6 episodes/year) and 13 children with bilateral persistent middle ear effusion (greater than 3 months) received unilateral ventilating tube insertion in a randomly selected ear. The contralateral ears were randomized to receive either myringotomy alone or no surgery. Clinical, otoscopic, tympanometric and audiologic examinations were performed before the study and 2 to 4 weeks later, then at 3-month intervals for up to 2 years and at 36 months after surgical randomization. Medical therapy and antibiotic prophylaxis were used whenever indicated. While the ventilating tubes remained functional (mean duration, 10 months) the ears with a tube had significantly fewer episodes of otitis media than their contralateral ear (P less than 0.001; 95% confidence intervals -0.7, -1.7) and had more hearing improvement (P = 0.005; 95% confidence intervals, -5.9, -1.2). After tube extrusion there was a tendency for surgically treated ears to have more otitis and worse hearing, but not at a significant level. Tympanosclerosis, retraction and atrophy were more common in ears that received tubes. The majority of ears treated medically also improved. There is need for a more cautious and selective use of ventilating tubes.

Efficacy of Haemophilus influenzae type b capsular polysaccharide vaccine.

The efficacy of the Haemophilus influenzae type b (Hib) polyribose phosphate vaccine was evaluated in a population of 120,000 children from 23 through 71 months of age in the Kaiser Permanente Medical Care Program (KPMCP) in Northern California over the 2-year period from June 1, 1985, through May 31, 1987. Approximately 37% of the population were vaccinated by the end of the first year and 60% were vaccinated by the end of the second year. There were 35 cases of Hib disease, 4 of whom were vaccine failures. Cases of Hib disease were identified by multiple modality case ascertainment, consisting of: (1) active surveillance in KPMCP microbiology laboratories; (2) active surveillance on KPMCP pediatric wards by a study physician; (3) retrospective review of computer-stored hospital discharge diagnoses; and (4) a review of all hospitalizations outside the health plan. The medical records of cases, matched controls and a random sample of the population were reviewed to obtain information on vaccination and related variables. Efficacy was evaluated using two complementary methods. In a retrospective surveillance approach, efficacy was estimated to be 68% (95% confidence limits of 4, 89%). In a matched case-control analysis, efficacy was estimated to be 69% (95% confidence limits of -13, 91%). Adjustment for day care attendance and parental occupation slightly reduced the efficacy estimate. Other possible confounders including race, parental education and number of siblings were considered. Four cases of Hib disease were observed within 1 week following receipt of vaccine and before the time when immunity could have developed. There are several plausible explanations for the occurrence of these early cases including the possibility of chance alone.(ABSTRACT TRUNCATED AT 250 WORDS)