Maintaining quality assurance for sonographic nuchal translucency measurement: lessons from the FASTER Trial.

OBJECTIVE: To evaluate nuchal translucency measurement quality assurance techniques in a large-scale study. METHODS: From 1999 to 2001, unselected patients with singleton gestations between 10 + 3 weeks and 13 + 6 weeks were recruited from 15 centers. Sonographic nuchal translucency measurement was performed by trained technicians. Four levels of quality assurance were employed: (1) a standardized protocol utilized by each sonographer; (2) local-image review by a second sonographer; (3) central-image scoring by a single physician; and (4) epidemiological monitoring of all accepted nuchal translucency measurements cross-sectionally and over time. RESULTS: Detailed quality assessment was available for 37 018 patients. Nuchal translucency measurement was successful in 96.3% of women. Local reviewers rejected 0.8% of images, and the single central physician reviewer rejected a further 2.9%. Multivariate analysis indicated that higher body mass index, earlier gestational age and transvaginal probe use were predictors of failure of nuchal translucency measurement and central image rejection (P = 0.001). Epidemiological monitoring identified a drift in measurements over time. CONCLUSION: Despite initial training and continuous image review, changes in nuchal translucency measurements occur over time. To maintain screening accuracy, ongoing quality assessment is needed.

The effect of the elapsed time between blood draw and processing on the recovery of fetal cells from maternal blood.

OBJECTIVE: To test the hypothesis that a delay in initial fetal cell enrichment processing of maternal blood samples (defined as the time between blood draw and the initial density gradient centrifugation step) compromises the ability to recover fetal cells, we performed a randomized comparison of immediate (within 4 hours of draw) versus delayed (between 18-24 hours of draw) processing. METHODS: Four centers participated: two centers utilized flow cytometry (FLOW), and two centers utilized magnetic-activated cell sorting (MACS) techniques. Each center collected 34 samples. The outcome was the percentage of gamma positive (gamma(+)) cells for FLOW or the number of nucleated red blood cells (NRBCs) for MACS, found in the final enriched cell population. Both outcomes reflect cell properties that are potentially fetal in origin, thus making them representative of the ability to recover fetal cells. RESULTS: Our results did not support our hypothesis that delay in processing compromises fetal cell recovery. Instead, in MACS processing, we observed an increase in recovered NRBCs when blood sample processing was delayed compared with immediate processing. There was no significant difference in gamma(+) cells with FLOW. CONCLUSION: Time-related changes in the density of target cells, perhaps associated with their progress towards apoptosis during the delay period, may result in increased intact fetal cells with the study methods utilized.

Fetal gender and aneuploidy detection using fetal cells in maternal blood: analysis of NIFTY I data. National Institute of Child Health and Development Fetal Cell Isolation Study.

OBJECTIVES: The National Institute of Child Health and Human Development Fetal Cell Isolation Study (NIFTY) is a prospective, multicenter clinical project to develop non-invasive methods of prenatal diagnosis. The initial objective was to assess the utility of fetal cells in the peripheral blood of pregnant women to diagnose or screen for fetal chromosome abnormalities. METHODS: Results of fluorescence in situ hybridization (FISH) analysis on interphase nuclei of fetal cells recovered from maternal blood were compared to metaphase karyotypes of fetal cells obtained by amniocentesis or chorionic villus sampling (CVS). After the first 5 years of the study we performed a planned analysis of the data. We report here the data from 2744 fully processed pre-procedural blood samples; 1292 samples were from women carrying singleton male fetuses. RESULTS: Target cell recovery and fetal cell detection were better using magnetic-based separation systems (MACS) than with flow-sorting (FACS). Blinded FISH assessment of samples from women carrying singleton male fetuses found at least one cell with an X and Y signal in 41.4% of cases (95% CI: 37.4%, 45.5%). The false-positive rate of gender detection was 11.1% (95% CI: 6.1,16.1%). This was higher than expected due to the use of indirectly labeled FISH probes in one center. The detection rate of finding at least one aneuploid cell in cases of fetal aneuploidy was 74.4% (95% CI: 76.0%, 99.0%), with a false-positive rate estimated to be between 0.6% and 4.1%. CONCLUSIONS: The sensitivity of aneuploidy detection using fetal cell analysis from maternal blood is comparable to single marker prenatal serum screening, but technological advances are needed before fetal cell analysis has clinical application as part of a multiple marker method for non-invasive prenatal screening. The limitations of the present study, i.e. multiple processing protocols, are being addressed in the ongoing study.

Favorable attitudes toward testing for chromosomal abnormalities via analysis of fetal cells in maternal blood.

PURPOSE: The NICHD Fetal Cell Isolation Study (NIFTY) was a multicentered project to isolate fetal cells from maternal blood to detect fetal chromosomal abnormalities. The project included a psychosocial component, which is the basis of this article. We examined the attitudes of high-risk pregnant women toward the availability of a maternal blood test to identify fetal chromosomal abnormalities, how women would respond to hypothetical normal and abnormal maternal blood testing results, and the factors associated with a woman's preference to have an invasive procedure in response to a normal maternal blood test. METHODS: High-risk pregnant women (N = 854) planning to have prenatal diagnostic invasive testing (amniocentesis or chorionic villus sampling) completed a survey. RESULTS: The women highly favored maternal blood testing. Almost all women would seek invasive testing after an abnormal blood test. Only half of the women would seek invasive testing after a normal blood test; these women were older, more willing to terminate their pregnancy, and valued the increased accuracy of invasive testing more highly than women who would not have invasive testing after a normal maternal blood test. CONCLUSIONS: Women having invasive diagnostic testing welcome a noninvasive procedure that uses fetal cells in maternal blood, and its availability would decrease invasive testing by approximately 50%. Research needs to examine the attitudes and anticipated responses of other risk groups as well as the effects of information about maternal blood test sensitivity and specificity on attitudes and responses.

Patients' role in the use of radiology testing for common office practice complaints.

BACKGROUND: Radiological studies are an important component of ambulatory medical costs, and guidelines often focus on their appropriate use. However, little is known about the correlates of the use of those services, particularly the influence of patients' preferences on physicians' utilization decisions. OBJECTIVES: To study patients presenting for outpatient treatment of respiratory problems and low back pain, and to examine the magnitude of the effect of the patients' perceived need for radiological studies (radiology preference score) on use of those services. DESIGN: Cross-sectional survey. SETTING: Office practices of generalist physicians in predominantly rural areas of 8 states. PARTICIPANTS: A total of 52 generalist physicians agreed to enroll consecutive Medicare-eligible patients making office visits for respiratory problems or low back pain. Of 1785 eligible patients invited to participate, 132 (7%) refused and 1137 (69%) of 1653 returned questionnaires. MEASUREMENTS: Radiology utilization rates (plain film, computed tomographic scan, or magnetic resonance image scan) were determined by patient self-report. To assess perceived need for radiological studies, we asked patients how necessary they believed an x-ray film was in the evaluation of 4 common complaints (respiratory problems, low back pain, knee pain, and knee swelling). A summary radiology preference score was created from 3 of the 4 items, excluding the item referring to the patients' index complaint. RESULTS: Six hundred fifteen respiratory and 522 low back pain patients were enrolled; mean ages were 69 and 64 years, respectively. Radiology utilization rates were 37% for respiratory and 26% for low back pain patients. In multiple logistic regression models, for respiratory patients radiology utilization was related significantly to the radiology preference score (odds ratio [OR] for fourth quartile compared with first quartile, 1.94; 95% confidence interval [CI], 1.11-3.37; P = .02), to having a physician who owned radiology equipment (OR, 1.81; 95% CI, 1.23-2.66; P = .002), and current smoking (OR, 1.58; 95% CI, 1.04-2.41; P = .03). For low back pain patients, radiology utilization was significantly related to the radiology preference score (OR for fourth compared with first quartile, 2.55; 95% CI, 1.29-5.06; P = .007), bothersomeness of the pain (OR for fourth compared with first quartile, 3.74; 95% CI, 1.74-8.04; P<.001), and a diagnosis of osteoporosis (OR, 1.67; 95% CI, 1.01-2.75; P = .04). CONCLUSIONS: Patients' perceived need for radiological studies was significantly associated with use of those services for outpatients with respiratory problems and low back pain. These findings suggest that patients communicate their wishes to physicians, either directly or indirectly, regarding services they think are necessary. Differences in physicians' adherence to guidelines regarding radiology utilization may in part reflect variations in patients' perceived need for those services. Efforts to educate patients about when radiological studies are medically indicated may be an important complement to practice guidelines or other utilization-related financial incentives.

Quantifying comorbidity in a disease-specific cohort: adaptation of the total illness burden index to prostate cancer.

OBJECTIVES: Numerous studies have demonstrated the importance of comorbid illness when analyzing medical outcomes. The purpose of this study was to adapt a generic comorbidity index, the Total Illness Burden Index (TIBI), for use in men with prostate cancer, and to evaluate the usefulness of the new instrument in adjusting for the impact of comorbidity on functional outcomes in a prostate cancer cohort. METHODS: The TIBI uses patients' self-report of symptoms and diagnoses to determine not only the presence but also the severity of comorbidities in each of 16 body system domains. To create the TIBI-P (prostate cancer modification), some domains were added and others were modified according to clinical criteria. The TIBI-P was completed by 1638 men with prostate cancer followed up longitudinally in 29 urology practices in the United States. TIBI-P scores were calculated for each patient and analyzed with scores on the SF-36 quality-of-life questionnaire and with patient report of days confined to bed. RESULTS: After adjusting for age and income, lower SF-36 scale scores and increases in confinement to bed were associated with a greater burden of comorbid illness as measured by the TIBI-P, independent of the extent of prostate cancer. The TIBI-P explained 24% of the variance in the SF-36 physical functioning domain score. CONCLUSIONS: The TIBI-P is a powerful measure of the impact of comorbid illness on the quality of life and functioning among patients with prostate cancer. This index may prove valuable in research on clinical and economic outcomes of prostate cancer.

Tutorial in biostatistics. An introduction to hierarchical linear modelling.

Hierarchical linear models are useful for understanding relationships in hierarchical data structures, such as patients within hospitals or physicians within hospitals. In this tutorial we provide an introduction to the technique in general terms, and then specify model notation and assumptions in detail. We describe estimation techniques and hypothesis testing procedures for the three types of parameters involved in hierarchical linear models: fixed effects, covariance components, and random effects. We illustrate the application using an example from the Type II Diabetes Patient Outcomes Research Team (PORT) study and use two popular PC-based statistical computing packages, HLM/2L and SAS Proc Mixed, to perform two-level hierarchical analysis. We compare output from the two packages applied to our example data as well as to simulated data. We elaborate on model interpretation and provide guidelines for model checking.

Breast cancer care in older women: sources of information, social support, and emotional health outcomes.

BACKGROUND: The authors studied older women with breast cancer and asked: 1) where do older women get information regarding breast cancer care and how helpful do they perceive each of these sources to be? and 2) what aspects of social support are associated with older women's general and breast cancer specific emotional health outcomes? METHODS: To be eligible, women had to be at least 55 years of age and newly diagnosed with TNM Stage I or II breast cancer. Data were collected from women's surgical records and a 35-minute, computer-assisted telephone interview. RESULTS: Nearly all women rated information that was provided by their breast cancer physicians as very or somewhat helpful. Written materials provided by breast cancer physicians also were frequently rated as very or somewhat helpful. Women's marital status, religious service attendance, ratings of their physicians' technical and interpersonal care, and perceptions of their own abilities to communicate with their physicians were significantly associated with both general and breast cancer specific emotional health outcomes (all P < 0.05). CONCLUSIONS: Although older women obtained information regarding breast cancer from a variety of sources, they relied heavily on their physicians for information. To care most effectively for this group of patients, an increased understanding of the relation between the processes and outcomes of breast cancer care is needed Identifying older women with breast cancer at risk for poor emotional health outcomes and developing methods to enhance physician-patient communication in this setting may improve these outcomes.

Metabolic control and prevalent cardiovascular disease in non-insulin-dependent diabetes mellitus (NIDDM): The NIDDM Patient Outcome Research Team.

PURPOSE: Cardiovascular disease is a major cause of morbidity and death in non-insulin-dependent diabetes mellitus (NIDDM). While hyperglycemia is clearly related to diabetic microvascular complications, it contribution to large-vessel atherosclerosis is controversial. PATIENTS AND METHODS: We performed an analysis of the association between glycemic control and prevalent cardiovascular disease in 1,539 participants in the NIDDM Patient Outcomes Research Team study who were under usual care in a health maintenance organization. Prevalent cardiovascular disease and its risk factors were identified by self-administered questionnaire. Cardiovascular disease was defined by the presence of coronary heart disease, peripheral vascular disease, and/or cerebrovascular disease. Glycohemoglobin and lipid levels were obtained from a computerized laboratory database. RESULTS: The mean age of participants was 63 years (range 31 to 91); 51% were women. The mean duration of NIDDM was 9 years (range < 1 to 50), 35% took insulin, and 48% took sulfonylureas. Mean glycohemoglobin was 10.6%. Sixty percent had hypertension, 16% currently smoked cigarettes, and the mean total high-density lipoprotein (HDL) cholesterol ratio was 5.7. Fifty-one percent had cardiovascular disease. Cardiovascular disease prevalence remained constant across increasing quartiles of glycohemoglobin for both men and women. In contrast, prevalent cardiovascular disease was associated with established cardiovascular disease risk factors including age (67 versus 59 years, P < 0.0001), hypertension (66% versus 54%, P < 0.0001), current cigarette smoking (17% versus 13%, P < 0.005), and total/HDL cholesterol ratio (5.9 versus 5.6, P < 0.005). Cardiovascular disease was also associated with duration of NIDDM (11 versus 8 years, P < 0.0001). In multiple logistic regression analysis controlling for established cardiovascular disease risk factors and diabetes duration and therapy, glycohemoglobin remained unassociated with cardiovascular disease. CONCLUSIONS: Glycemic control is not associated with prevalent cardiovascular disease in this large population of individuals with NIDDM. Conventional cardiovascular disease risk factors are independently associated with cardiovascular disease and be a more promising focus for clinical intervention to reduce atherosclerotic complications in NIDDM.

The care of older persons with diabetes mellitus: families and primary care physicians.

OBJECTIVES: To determine the extent to which family members participate in the day-to-day management of diabetes mellitus in older persons, and in older diabetics' medical encounters, and to identify patient and family member characteristics associated with this participation. DESIGN: A longitudinal observational study, with baseline data being reported herein. SETTING: Three primary care practice settings in Seattle, Washington, Boston, Massachusetts, and Indianapolis, Indiana. PARTICIPANTS: Family members of patients 70 years of age or older participating in the Patient Outcomes Research Team (PORT) Study of type II diabetics. MAIN OUTCOME MEASUREMENTS: The two dependent variables represent, respectively, the extent of family members' assistance with diabetes-related care and participation in older diabetics' medical encounters. RESULTS: The 357 family members enrolled were older (mean age = 66.3 years), were mostly women (76.2%), and were usually the spouses of diabetic patients (71.3%). Between 22% and 50% of family members reported helping with various aspects of diabetes care; 35.6% of family members participated regularly in their diabetic patients' medical encounters. A multiple linear regression model relating family assistance with diabetes-related care to patient and family member characteristics included four variables: patients' physical function, and the family member's relationship to the patient, assistance with basic activities of daily living (ADLs), and understanding of diabetes management issues (all P < .05). A multiple logistic regression model relating family member participation in the medical encounter to patient and family member characteristics also included four variables: patient age and physical function, and family member assistance with instrumental activities of daily living (IADLs) and with diabetes-related care (all P < .05). CONCLUSION: The family members studied frequently assisted older diabetics with diabetes-specific care; more than one-third were regular participants in older diabetics' medical encounters. Family member involvement in the day-to-day management of diabetes and in the medical encounter was more likely when patients were functionally disabled. Health care systems and physicians need to educate their older patients, and involved family members when patients are frail, about diabetes-related care issues and support them in their roles in the management of diabetes as well as other chronic diseases.

Sex differences in academic advancement. Results of a national study of pediatricians.

BACKGROUND: Although the numbers of women in training and in entry-level academic positions in medicine have increased substantially in recent years, the proportion of women in senior faculty positions has not changed. We conducted a study to determine the contributions of background and training, academic productivity, distribution of work time, institutional support, career attitudes, and family responsibilities to sex differences in academic rank and salary among faculty members of academic pediatric departments. METHODS: We conducted a cross-sectional survey of all salaried physicians in 126 academic departments of pediatrics in the United States in January 1992. Of the 6441 questionnaires distributed, 4285 (67 percent) were returned. The sample was representative of U.S. pediatric faculty members. Multivariate models were used to relate academic rank and salary to 16 independent variables. RESULTS: Significantly fewer women than men achieved the rank of associate professor or higher. For both men and women, higher salaries and ranks were related to greater academic productivity (more publications and grants), more hours worked, more institutional support of research, greater overall career satisfaction, and fewer career problems. Less time spent in teaching and patient care was related to greater academic productivity for both sexes. Women in the low ranks were less academically productive and spent significantly more time in teaching and patient care than men in those ranks. Adjustment for all independent variables eliminated sex differences in academic rank but not in salary. CONCLUSIONS: Lower rates of academic productivity, more time spent in teaching and patient care and less time spent in research, less institutional support for research, and lower rates of specialization in highly paid subspecialties contributed to the lower ranks and salaries of female faculty members.

A comparison of various methods of collecting self-reported health outcomes data among low-income and minority patients.

In a randomized trial of different data collection methods, we challenged the untested assumption that reliable data cannot be obtained from lower-income and/or minority patients by self-administered questionnaires. We tested three methods of data collection among a sample of lower-income and minority patients (n = 697) in Indianapolis at a site for the Type II Diabetes Patient Outcomes Research Team. The study included a questionnaire literacy screening instrument to assess patients' functional literacy. Based on their functional literacy, patients were randomized to one of three methods of data collection: mail-out/mail-back, hand-out/assisted, or the in-home interview. We constructed a tiered system for reassigning nonresponders to alternative methods of data collection, using the in-home interview as the fall-back strategy. We compared the response rates, item completion rates, and internal consistency reliabilities of self-reported health status measures between patients with and without literacy limitations and across the three methods of data collection. Patients with and without literacy limitations, across methods of data collection, provided high-quality data, as evidenced by high item completion rates (> 84%) and high reliability assessments (internal consistency reliability coefficients > .80) for each health status measure. As part of the tiered study design, nonresponders randomized to either the mail-out/mail-back or the hand-out/assisted method were interviewed. These patients were significantly older, had significantly lower education and income levels, and had significantly poorer self-reported visual function as compared with those who responded to the originally assigned method. We conclude that expensive, labor-intensive data collection methods, such as in-home interviews, are not necessary for many low-income, minority patients to generate high-quality, reliable health status data. Using appropriate screening instruments, those patient subgroups needing special help can be screening instruments, those patient subgroups needing special help can be identified and targeted for more expensive data collection methods. This tiered approach has policy implications for the cost, feasibility, and quality of data collection in health outcomes research.

Development and testing of a new measure of case mix for use in office practice.

Case mix has been shown to be of critical importance in studies of effectiveness and quality of care using health outcomes. How these variables are defined, combined, and used to adjust or increase precision in tests for differences in health outcomes has been a source of controversy. Because existing measures were developed to adjust mortality and have marginal relevance for the adjustment of functional status outcomes, especially in ambulatory settings, the authors developed a measure of case (or patient) mix that is specifically designed to adjust functional status outcomes measured in office practice or out-of-hospital settings. This measure, developed as part of Type II Diabetes Patient Outcomes Research Team project, uses patients' reports of symptoms and conditions, as well as patients' ratings of symptom intensity to characterize total disease burden. It differs from other measures of case mix in lack of dependence on diagnoses. Separate measures were developed for each of 15 different disease categories (e.g., chronic lung disease) grouped by body system affected. Within each measure, questionnaire items were combined to rate the severity of that disease on a 1 to 4 scale, according to definitions provided by clinicians. A single, global measure was developed by aggregating the 15 measures, weighted according to the expected impact of each disease category on functional outcomes and disability. In a sample of 1,738 patients, significant relationships were observed between the global case mix measure and functional status, disability days, and service utilization.(ABSTRACT TRUNCATED AT 250 WORDS)

Foot infections in diabetic patients. Decision and cost-effectiveness analyses.

OBJECTIVE: To examine the cost-effectiveness of approaches to the diagnosis and treatment of patients with type II (non-insulin-dependent) diabetes mellitus (NIDDM) who have foot infections and suspected osteomyelitis. DESIGN: Decision and cost-effectiveness analyses were performed using a Markov model. We examined the prevalence of osteomyelitis, the major complications and efficacies of long-term antibiotic therapy and surgery, and the performance characteristics of four diagnostic tests (roentgenography, technetium Tc 99m bone scanning, indium in 111-labeled white blood cell scanning, and magnetic resonance imaging). Data were drawn from the English-language literature using MEDLINE searches and bibliographies from selected articles. SETTING: Primary care. PATIENTS: Patients with NIDDM who had foot infections and suspected osteomyelitis but no signs of systemic toxicity. INTERVENTIONS: Following hospitalization for surgical débridement and intravenous antibiotic therapy: (1) treatment for presumed soft-tissue infection, (2) culture-guided empiric treatment for presumed osteomyelitis, (3) 71 combinations of diagnostic tests preceding antibiotic therapy for osteomyelitis, (4) 71 combinations of tests preceding amputation, and (5) immediate amputation. MAIN OUTCOME MEASURES: Quality-adjusted life expectancy, average costs. RESULTS: Culture-guided empiric treatment for osteomyelitis with 10 weeks of oral antibiotic therapy has similar effectiveness to testing followed by a long course of antibiotic therapy if any test result is positive. However, empiric treatment is the least expensive strategy. CONCLUSIONS: Noninvasive testing adds significant expense to the treatment of patients with NIDDM in whom pedal osteomyelitis is suspected, and such testing may result in little improvement in health outcomes. In patients without systemic toxicity, a 10-week course of culture-guided oral antibiotic therapy following surgical débridement may be as effective as and less costly than other approaches.

Principles and practice of case mix adjustment: applications to end-stage renal disease.

For optimal case mix adjustment, it is necessary to build risk or severity models to fit the disease and its outcome. For cardiac disease, for example, several risk models predict the outcome of mortality. These risk models include variables that, by themselves in multivariable models, predict death. If a variable is associated with death independent of treatment or quality of care, it has to be adjusted for and, therefore, included in the model. Similar models have to be developed and tested for end-stage renal disease. A complete risk model would consist of age, sex, the severity of the primary disease of interest (here, end-stage renal disease), not just the presence but also the severity of all co-morbid disease, and, finally, a measurement of functional status or quality of life. This last measurements is associated with outcome beyond and independent of the above noted severity components. These factors (or "severity" or "patient mix" dimensions) then have to be developed in relation to a specific outcome or response variable. Cholesterol and hypertension are risk factors for long-term events and mortality, but not for 2-year symptoms and quality of life. Thus, the model has to be defined by the type of outcome and by the time interval over which the variable acts. For end-stage renal disease, at least one of the major types of outcome will be quality of life. Our case mix measures for hospital co-morbidity and for office practice chronic disease are defined by and intended to predict functional status in the short run, because that is the outcome of interest for dollars spent and for most of the care rendered to patients with chronic disease before the very end stages. Creating a valid case mix or severity measurement aimed at or defined by quality of life thus involves the creation of new models, the testing of those models, the revision based on the empirical testing, and, finally, generalizing to other sites beyond which the data were collected. Useful case mix measurements will be parsimonious, feasible, reliable, validated against the outcome of interest, shown to work in diverse settings, and acceptable to users, especially clinicians.

Osteonecrosis of the humeral head in sickle cell disease.

The prevalence and incidence of osteonecrosis (ON) of the humeral head in sickle cell disease was determined by a study of 2524 patients who were entered into a prospective study and followed for an average of 5.6 years. At entry, 5.6% had roentgenographic evidence of ON in one or both shoulders. There was little difference in age-adjusted prevalence among genotypes, but there were striking differences in age-specific rates. Observed at ages ranging from five to 24 years, 3.25% of sickle cell anemia (S/S) patients, but only 1.1% of sickle cell disease (S/C) patients, had ON. No S/beta+ thalassemia patients younger than 25 years of age had ON on entry. The highest age-adjusted incidence rate was found in S/S patients with concomitant alpha-thalassemia (4.85 per hundred patient-years), followed by S/beta zero-thalassemia (4.84 per hundred patient-years), S/beta+ thalassemia (2.61 per hundred patient-years), S/S without alpha-thalassemia (2.54 per hundred patient-years), and S/C (1.66 per hundred patient-years). Only 20.9% of patients reported pain or had limited range of movement at the time of diagnosis. Sickle cell disease is a frequent cause of ON of the humeral head, especially in children and young adults.

Sickle cell disease as a cause of osteonecrosis of the femoral head.

BACKGROUND AND METHODS: Osteonecrosis of the femoral head is an important complication of sickle cell disease. We studied 2590 patients who were over 5 years of age at entry and followed them for an average of 5.6 years. Patients were examined twice a year, and radiographs of the hips were taken at least twice: at study entry and approximately three years later. RESULTS: At study entry, 9.8 percent of patients were found to have osteonecrosis of one or both femoral heads. On follow-up, patients with the hemoglobin SS genotype and alpha-thalassemia were at the greatest risk for osteonecrosis (age-adjusted incidence rate, 4.5 cases per 100 patient-years, as compared with 2.4 in patients with the hemoglobin SS genotype without alpha-thalassemia and 1.9 in those with the hemoglobin SC genotype). Although the rate of osteonecrosis in patients with the hemoglobin SC genotype did not differ significantly from that in patients with the hemoglobin SS genotype without alpha-thalassemia, osteonecrosis tended to develop in these patients later in life. Intermediate rates of osteonecrosis were observed among patients with the hemoglobin S-beta zero-thalassemia and the hemoglobin S-beta(+)-thalassemia genotypes (3.6 and 3.3 cases per 100 patient-years, respectively). Osteonecrosis was found in patients as young as five years old (1.8 cases per 100 patient-years for all genotypes). The frequency of painful crises and the hematocrit were positively associated with osteonecrosis. The mean corpuscular volume and serum aspartate aminotransferase level were negatively associated. Twenty-seven patients had hip arthroplasty during the study; 10 were under 25 years of age. Five of the 27 required reoperation 11 to 53 months after the initial operation. CONCLUSIONS: Osteonecrosis of the femoral head is common in patients with sickle cell disease, with an incidence ranging from about 2 to 4.5 cases per 100 patient-years. Patients with the hemoglobin SS genotype and alpha-thalassemia and those with frequent painful crises are at highest risk. The overall prevalence is about 10 percent. The results of hip arthroplasty are poor.