Relationship between patient complaints and surgical complications.

BACKGROUND: Patient complaints are associated with increased malpractice risk but it is unclear if complaints might be associated with medical complications. The purpose of this study was to determine whether an association exists between patient complaints and surgical complications. METHODS: A retrospective analysis of 16,713 surgical admissions was conducted over a 54 month period at a single academic medical center. Surgical complications were identified using administrative data. The primary outcome measure was unsolicited patient complaints. RESULTS: During the study period 0.9% of surgical admissions were associated with a patient complaint. 19% of admissions associated with a patient complaint included a postoperative complication compared with 12.5% of admissions without a patient complaint (p = 0.01). After adjusting for surgical specialty, co-morbid illnesses and length of stay, admissions with complications had an odds ratio of 1.74 (95% confidence interval 1.01 to 2.98) of being associated with a complaint compared with admissions without complications. CONCLUSIONS: Admissions with surgical complications are more likely to be associated with a complaint than surgical admissions without complications. Further research is necessary to determine if patient complaints might serve as markers for poor clinical outcomes.

Predictors of influenza virus vaccination status in hospitalized children.

OBJECTIVE: To determine predictors of influenza virus vaccination status in children who are hospitalized during the influenza season. METHODS: A cross-sectional study was conducted among children who were hospitalized with fever between 6 months and 3 years of age or with respiratory symptoms between 6 months and 18 years of age. The 1999 to 2000 influenza vaccination status of hospitalized children and potential factors that influence decisions to vaccinate were obtained from a questionnaire administered to parents/guardians. RESULTS: Influenza vaccination rates for hospitalized children with and without high-risk medical conditions were 31% and 14%, respectively. For both groups of children, the vaccination status was strongly influenced by recommendations from physicians. More than 70% of children were vaccinated if a physician had recommended the influenza vaccine, whereas only 3% were vaccinated if a physician had not. Lack of awareness that children can receive the influenza vaccine was a commonly cited reason for nonvaccination. CONCLUSIONS: A minority of hospitalized children with high-risk conditions had received the influenza vaccine. However, parents' recalling that a clinician had recommended the vaccine had a positive impact on the vaccination status of children.

Corticosteroid prescription filling for children covered by Medicaid following an emergency department visit or a hospitalization for asthma.

OBJECTIVE: To identify predictors of corticosteroid prescription filling following an emergency department (ED) visit or a hospitalization for asthma. DESIGN: A retrospective cohort study. PATIENTS: Tennessee children (defined as those aged 2-17 years in this study) covered by Medicaid were included in the cohort if they had an ED visit or a hospitalization for asthma between July 1, 1995, and December 31, 1997. MAIN OUTCOME MEASURES: Prescriptions filled in the child's name for an oral corticosteroid within 7 days of the latest ED visit or hospitalization for asthma. RESULTS: Of 6035 Tennessee children covered by Medicaid with an ED visit for asthma and of 2102 covered by Medicaid with a hospitalization for asthma during the study period, less than half (44.8% following an ED visit and 55.5% following a hospitalization) had prescriptions filled for oral corticosteroids within 7 days. Factors independently predicting a child's not having an oral corticosteroid prescription filled included older age, black race, and residence in rural regions of the state. Conversely, children with oral corticosteroid prescriptions in the previous 6 months were more likely to have oral corticosteroid prescriptions filled following an ED visit for asthma, and children with more than 3 beta-agonist prescriptions in the previous 6 months were more likely to have oral corticosteroid prescriptions filled following a hospitalization for asthma. CONCLUSIONS: Overall, fewer than half of Tennessee children covered by Medicaid had an oral corticosteroid prescription filled following an ED visit or a hospitalization for asthma. Age, race, and county of residence predicted failure to have a prescription filled. Further study is needed to determine whether variations in corticosteroid prescription filling relate to physician practice, family behavior, or both.

Principles of patient safety in pediatrics.

The American Academy of Pediatrics and its members are committed to improving the health care system to provide the best and safest health care for infants, children, adolescents, and young adults. In response to a 1999 Institute of Medicine report on building a safer health system, a set of principles was established to guide the profession in designing a health care system that maximizes quality of care and minimizes medical errors through identification and resolution. This set of principles provides direction on setting up processes to identify and learn from errors, developing performance standards and expectations for safety, and promoting leadership and knowledge.

Changes in continuity of enrollment among high-risk children following implementation of TennCare.

BACKGROUND: TennCare, Tennessee's Medicaid managed care program, was introduced in 1994 with the goals of controlling spending and of improving access to health care. OBJECTIVE: To assess changes in the continuity of enrollment following the implementation of TennCare for 2 groups: infants in the first year of life (defined as persons aged 0-12 months in this study) and children hospitalized with a chronic health condition (defined as persons aged 0-18 years in this study). DESIGN: Retrospective cohort analysis. SETTING AND POPULATION: Infants born during 1992 or 1995 to women enrolled in Medicaid or TennCare and 0- to 18-year-old children enrolled in Medicaid or TennCare who were discharged from a hospital during 1992 or 1995 with a chronic health condition. MAIN OUTCOME MEASURES: For infants, failure to enroll an infant in the first 30 days of life or subsequent gaps in enrollment for 7 days or longer during the first year of life. For children hospitalized with a chronic health condition, any gap in enrollment lasting 7 days or longer by 1 year after discharge from a hospital. RESULTS: There was a reduction in the proportion of infants without continuous enrollment in the first year of life following TennCare (19.4% after vs 25.1% before TennCare; odds ratio, 0.69; 95% confidence interval, 0.67-0.72). Improvements in continuity of enrollment for infants occurred despite an increase in the proportion of infants who were not enrolled in TennCare in the first 30 days of life, even though their mother was enrolled at delivery (14.0% after vs 8.0% before TennCare; odds ratio, 1.86; 95% confidence interval, 1.78-1.96). There was a decrease in the proportion of children hospitalized with a chronic health condition who had subsequent gaps in enrollment by 1 year following discharge from a hospital (14.3% after vs 23.3% before TennCare; odds ratio, 0.52; 95% confidence interval, 0.46-0.59). CONCLUSION: For infants in the first year of life and for children hospitalized with a chronic health condition, implementation of TennCare improved continuity of coverage.

Comparison of perinatal outcomes among TennCare managed care organizations.

OBJECTIVE: To compare perinatal outcomes among the managed care organizations (MCOs) providing care to beneficiaries enrolled in TennCare, Tennessee's capitated Medicaid managed care program. DESIGN: Retrospective cohort analysis. SUBJECTS: Infants born in Tennessee during 1995 to women enrolled in TennCare. PRIMARY OUTCOME MEASURES: Prenatal care use, birth weight (BW), death in the first 60 days of life, and delivery of extremely low BW (<1000 g) infants in hospitals without level 3 neonatal intensive care units. RESULTS: During 1995, 34 402 infants were born to mothers enrolled in TennCare. The MCOs differed widely in the demographic characteristics of their enrollees. In addition, there were small differences in prenatal care utilization, but no differences in BW outcomes among the MCOs. In multivariate analysis, however, infants born to women enrolled in 1 MCO were 2.8 times more likely to die in the first 60 days of life than were infants born to women enrolled in the largest MCO (OR: 2.81; 95% CI: 1.31-6.03). Women enrolled in this same MCO seemed to have a higher proportion of extremely low BW (<1000 g) infants delivering in a hospital lacking a level 3 neonatal intensive care unit (38% vs 20% in the largest MCO). CONCLUSION: The differences among MCOs in early infant death and in the delivery of high-risk infants in hospitals lacking appropriate neonatal facilities suggest that monitoring of care delivery to vulnerable children should include assessment of appropriate use of specialized services.

Early childhood mortality from community-acquired infections.

In this study, the authors identified maternal and child characteristics that were independent predictors of death from infectious diseases acquired in the community and determined if these factors could be used to identify groups of children with excess risk of mortality from infection. A historical cohort study was conducted of children less than 5 years of age between 1985 and 1994 (the study period), who were born in Tennessee, and had complete information on their birth certificates. The primary outcome was death from infection identified from death certificates and confirmed through medical record review. Among the 1,014,976 children less than 5 years of age, who contributed 3,351,568 child-years of follow-up, there were 247 deaths from infections (7.4 deaths from infections per 100,000 child-years). Respiratory infections accounted for approximately one half of the deaths. Children having three or more older siblings or birth weight of less than 1,500 g had a 3-fold and 10-fold increased risk of death from infection, respectively, while children with both characteristics had a nearly 20-fold increased risk that persisted beyond the first year of life. Interventions should be focused on prevention of these infections in vulnerable children. At-risk children should be targeted for careful follow-up and early hospitalization when signs of infection develop.

Sociodemographic factors identify US infants at high risk of injury mortality.

OBJECTIVES: To identify sociodemographic predictors of infant injury mortality and to compare trends in injury mortality rates for high- and low-risk US infants from 1985 to 1991. DESIGN: Historical cohort. SETTING/STUDY PARTICIPANTS: The National Center for Health Statistics linked US infants (<1 year) born from 1985 to 1991 with death certificates. MAIN OUTCOME MEASURES: Multivariate regression was used to identify sociodemographic factors associated with injury mortality. The adjusted relative risks (RRs) of maternal age, education, marital status, number of other children, and infant birth weight were used to categorize infants into risk groups. We compared trends in injury rates for the highest and lowest risk groups. RESULTS: There were 5963 injury deaths and 18.6 million infant years or 32.1 injury deaths per 100 000 infant years. Highest risk infants were born to mothers who were younger than 20 years compared with older than 30 years (RR, 3.25; 95% CI, 2.92-3.63), had less than a high school education compared with a college education (RR, 2.22; 95% CI, 1.95-2.53), had more than 2 other children compared with no other children (RR, 3.15; 95% CI, 2.88-3.45), were unmarried (RR, 1.67; 95% CI, 1.57-1.78), or had birth weights 2500 g (RR, 3.36; 95% CI, 2.94-3.84). Infants in the highest risk group (21.0% of the population) had a >10-fold increased risk of injury mortality compared with the lowest risk group (18.1% of the population) and there was no evidence that this disparity was narrowing. CONCLUSIONS: Sociodemographic predictors of infant injury mortality include maternal age, education, number of other children, marital status, and infant birth weight. Based on these factors, 1 in 5 infants in the United States can be identified at birth as having a >10-fold increased risk of injury mortality compared with infants in lowest risk group. Programs to reduce injuries in these high-risk groups are urgently needed.

Identifying medical center units with disproportionate shares of patient complaints.

BACKGROUND: A pilot study was conducted to learn whether an academic medical center's database of patient complaints would reveal particular service units (or clinics) with disproportionate shares of patient complaints, the types of complaints patients have about those units, and the types of personnel about whom the complaints were made. RESULTS: During the seven-year (December 1991-November 1998) study period, Office of Patient Affairs staff recorded 6,419 reports containing 15,631 individual complaints. More than 40% of the reports contained a single complaint. One-third of the reports contained three or more complaints. Complaints were associated with negative perceptions of care and treatment (29%), communication (22%), billing and payment (20%), humaneness of staff (13%), access to staff (9%), and cleanliness or safety of the environment (7%). Complaints were not evenly distributed across the medical center's various units, even when the data were corrected for numbers of patient visits to clinics or bed days in the hospital. The greatest proportion of complaints were associated with physicians. DISCUSSION: Complaint-based report cards may be used in interventions in which peers share the data with unit managers and seek to learn the nature of the problems, if any, that underlie the complaints. Such interventions should influence behavioral and systems changes in some units. SUMMARY AND CONCLUSIONS: Further experience should indicate how different types of complaints lead to different kinds of interventions and improvements in care. Tests of the system are also currently under way in several nonacademic community medical centers.

The influence of obstetric no-fault compensation on obstetricians' practice patterns.

OBJECTIVES: The objectives were to determine level of satisfaction among obstetricians with the no-fault insurance programs in Florida and Virginia and to study any reported practice patterns attributable to implementation of no-fault compensation. STUDY DESIGN: Structured surveys were conducted with 119 obstetricians in Florida and Virginia. RESULTS: More than 90% of obstetricians were enrolled in no-fault insurance programs, but only 13% reported having had a patient compensated by a no-fault program. Only 14% knew of a colleague with a patient who had been compensated. Despite no-fault compensation, threat of lawsuits was a factor in 39% of cases of physicians who quit practicing obstetrics. The no-fault programs did not cause obstetricians to report increases in their obstetric caseloads or in their fraction of patients at high risk. Overall, obstetricians were far more satisfied with the no-fault system than with the tort system. Still, more than half of the respondents expressed dissatisfaction with premiums assessed by no-fault insurance. CONCLUSION: Obstetricians who knew about the no-fault programs were generally satisfied with their performance. However, the no-fault programs have not built a constituency with physicians, and the programs are relatively small in their scope of coverage. No-fault compensation thus has had minor impact on reported obstetric practice. To be effective in improving patient access, no-fault compensation must be broader in scope.

What health professionals can do to identify and resolve patient dissatisfaction.

BACKGROUND: Patient satisfaction affects consistency of self-care, health outcomes, level of service utilization, choice of health professionals, and decisions to sue in the face of adverse outcomes. Understanding patients' specific dissatisfactions may help health professionals and administrators identify and rectify organizational deficiencies before they become costly. COMMON CAUSES OF COMPLAINTS: As part of a series of research projects, more than 12,000 patient/family complaint narratives were examined in which patients or patients' family members told interviewers or patient advocates about the care they received from their health professionals in both inpatient and outpatient settings. Complaints may be categorized as involving issues of care and treatment, communication, humaneness, access and availability, environment, and billing/payment. STRATEGIES FOR RESOLVING COMPLAINTS: Even though caregivers may not have control over all the factors that lead to dissatisfaction, they can often hear and address complaints. As a result, they may not only contribute to quality of care but improve the systems in which they practice. The challenges are how to prevent dissatisfaction in the first place, and, if it does occur, to identify and if possible rectify patient concerns. Three case studies are provided. CONCLUSION: All health professionals must be involved in efforts to resolve problems that compromise patient care. Some problems could be prevented if administrators and leaders used complaint data to recommend new policies and procedures or to identify and counsel with health care team members who generate disproportionate numbers of complaints. If all are involved in both prevention and problem solving, resources devoted to uncovering, understanding, and resolving patient complaints are likely to prove cost-effective.

Predictors of mortality from fires in young children.

BACKGROUND: In the United States in 1994, fires claimed 3.75 lives per 100 000 child years and accounted for 17.3% of all injury deaths in children <5 years of age. OBJECTIVES: To conduct a historical cohort study that uses maternal demographic characteristics to identify young children at high risk of fire-related deaths, thus defining appropriate targets for prevention programs. METHODS: The cohort consisted of children born to mothers who resided in the state of Tennessee between 1980 and 1995. Information was obtained by linking birth certificates, 1990 census data, and death certificates. Children were eligible for the study if they were <5 years of age at any time within the study period and if key study variables were present (99.2% of births). Birth certificates provided information on maternal characteristics including age, race, education, previous live births, use of prenatal care, and residence (in standard metropolitan statistical area). Child characteristics included gender, gestational age, and birth type (singleton/multiple gestation). Neighborhood income was estimated by linking the mother's address at the time of birth to the 1990 census (block group mean per capita income). The study outcome was a fire resulting in at least one fatality (fatal fire event) during the study period, identified from death certificates (coded E880 through E889 in the International Classification of Diseases, 9th rev). We calculated the fatal fire event rate corresponding to each stratum of maternal/child characteristics. We assessed the independent association between each characteristic and the risk of a fatal fire event from a Poisson regression multivariate analysis. RESULTS: During the study period, 1 428 694 children contributed 5 415 213 child years to the cohort: there were 270 deaths from fire (4.99 deaths per 100 000 child years) and 231 fatal fire events. In the multivariate analysis, factors associated with greater than a threefold increase in fatal fire events included maternal education, age, and number of other children. Compared with children whose mothers had a college education, children whose mothers had less than a high school education had 19.4 times (95% confidence interval [CI], 2.6-142.4) an increased risk of a fatal fire event. Children whose mothers had more than two other children had 6.1 times (95% CI, 3.8-9.8) an increased risk of a fatal fire event compared with children whose mothers had no other children. Children of mothers <20 years of age had 3.9 times (95% CI, 2.2-7.1) increased risk of a fatal fire event compared with children whose mothers were >/=30 years old. Although both maternal neighborhood income and race were associated strongly with increased rates of fatal fire events in the univariate analysis, this association did not persist in the multivariate analysis. Other factors that were associated with increased risk of fatal fire events in the multivariate analysis were male gender and having a mother who was unmarried or who had delayed prenatal care. The three factors associated most strongly with fire mortality were combined to create a risk score based on maternal education (>/=16 years, 0 points; 13 to 15 years, 1 point; 12 years, 2 points; <12 years, 3 points); age (>/=30 years, 0 points; 25 to 29 years, 1 point; 20 to 24 years, 2 points; <20 years, 3 points); and number of other children (none, 0 points; one, 1 point; two, 2 points; three or more, 3 points). The lowest-risk group (score <3) included 19% of the population and had 0.19 fatal fire events per 100 000 child years. In contrast, highest-risk children (score >7) comprised 1.5% of the population and had 28.6 fatal fire events per 100 000 child years, 150 times higher than low-risk children. Children with risk scores >5 contributed 26% of child years but experienced 68% of all fatal fire events. If the fatal fire event rate for all children had been equal to that of the low-risk group (risk score <3), then 95% of deaths from

Effects of pediatrician characteristics on management decisions in simulated cases involving apparent life-threatening events.

OBJECTIVES: To study variations in the way pediatricians would evaluate and manage an infant with an apparent life-threatening event. SUBJECTS AND METHODS: A survey was mailed to the chief residents of all pediatric residency training programs in the United States in which respondents were presented with a simulated case and asked how they would manage an infant who had experienced an apparent life-threatening event that did not require resuscitation. The survey also explored each physician's tolerance of uncertainty, knowledge of apparent life-threatening events, experience, fear of litigation, responsiveness to parental demands, and propensity to order tests. MAIN OUTCOME MEASURES: Presumed decisions to prescribe antibiotics and/or order home apnea monitoring in a simulated case of an infant who had experienced an apparent life-threatening event not requiring resuscitation. RESULTS: Logistic regression analysis revealed 2 characteristics that made significant and independent contributions to respondents' presumed decision to prescribe antibiotics: (1) experience with an adverse outcome, and (2) propensity to order diagnostic tests. Presumed decisions to order a home apnea monitor were notably affected by fear of litigation. CONCLUSIONS: These findings suggest that differences in pediatricians' characteristics contribute to variations in care. Efforts to make management more uniform must consider that decisions are influenced by a host of different characteristics and experiences.

No-fault system of compensation for obstetric injury: winners and losers.

OBJECTIVE: To determine whether Florida's implementation of a no-fault system for birth-related neurologic injuries reduced lawsuits and total spending associated with such injuries, and whether no-fault was more efficient than tort in distributing compensation. METHODS: We compared claims and payments before and after implementation of a no-fault system in 1989. Data came from the Department of Insurance's medical malpractice closed claim files and no-fault records. Descriptive statistics were compiled for tort claims before 1989 and for tort and no-fault claims for 1989-1991. We developed two projection approaches to estimate claims and payments after 1989, with and without no-fault. We assessed the program's performance on the basis of comparisons of actual and projected values for 1989-1991. RESULTS: The number of tort claims for permanent labor-delivery injury and death fell 16-32%. However, when no-fault claims were added to tort claims, total claims frequency rose by 11-38%. Annually, an estimated 479 children suffered birth-related injuries; however, only 13 were compensated under no-fault. Total combined payments to patients and all lawyers did not decrease, but of the total, a much larger portion went to patients. Compensation of patients after plaintiff lawyers' fees rose 4% or 44%, depending on the projection method used. Less than 3% of total payments went to lawyers under no-fault versus 39% under tort. CONCLUSION: Some claimants with birth-related injuries were winners, taking home a larger percentage of their awards than their tort counterparts. Lawyers clearly lost under no-fault. Because of the narrow statutory definition, many children with birth-related neurologic injuries did not qualify for coverage.

Satisfaction with obstetric care: relation to neonatal intensive care.

OBJECTIVE: To examine the relationship between admitting children to a neonatal intensive care unit (NICU) and mothers' satisfaction with obstetric care. METHODS: Mothers of live-born infants who are now normal were interviewed about their perceptions of the interpersonal, organizational, and technical care they and their babies received during pregnancy, delivery, and the neonatal period, and their and their infants' health. Comparisons between the responses of mothers whose babies were admitted to the normal nursery (n = 595) and those whose babies were sent to an NICU (n = 72) were made by chi2 analysis (1 df) and Wilcoxon rank sum tests. RESULTS: Mothers whose babies were admitted to an NICU were more likely to complain that their obstetricians did not explain things in terms they could understand (P < .05); did not give them the right amount of information about what to expect during pregnancy (P < .05); hid something from them before delivery (P < .001); did not explain the reasons for tests performed during delivery (P < .05); misled them about their child's prognosis (P < .001); failed to treat properly a problem during delivery (P < .05); and did not know the latest medical developments (P < .05). CONCLUSION: Even when children do well, admission of newborns to an NICU is associated with greater maternal dissatisfaction with obstetric care.

Perinatal outcomes following implementation of TennCare.

CONTEXT: The abrupt initiation of capitated Medicaid care in Tennessee (TennCare) in 1994 prompted many questions about changes in quality of care. OBJECTIVE: To evaluate the effect on perinatal outcomes of the transition to TennCare in 1994. DESIGN: Before and after retrospective cohort analysis. SETTING AND POPULATION: Births to women residing in Tennessee between 1990 and 1995 with complete demographic information on birth certificates, with a focus on women enrolled in Medicaid giving birth in 1993 (before TennCare) and 1995 (after TennCare). OUTCOME MEASURES: Late prenatal care (after the fourth month of pregnancy) or inadequate prenatal visits, low and very low birth weight, and death in the first 60 days of life. RESULTS: Tennessee residents had 72014 study births in 1993 and 72278 in 1995, of which 37543 (52.1%) and 35707 (49.4%) were to women enrolled in Medicaid at delivery. For these Medicaid births, there were no changes after TennCare in the proportions with late prenatal care (16.2% in 1993 vs 15.8% in 1995), inadequate prenatal visits (5.9% vs 5.6%), low birth weight (9.4% vs 9.0%), very low birth weight (1.6% vs 1.5%), and death in the first 60 days (0.6% both years). These findings were unchanged in multivariate analysis, in analysis of high-risk subgroups, and in analysis of women with demographics characteristic of Medicaid women. CONCLUSION: Study perinatal outcomes did not change among Medicaid births following the transition to TennCare.

Persistently increased injury mortality rates in high-risk young children.

OBJECTIVE: To study trends in injury mortality for low- and high-risk young children. DESIGN AND METHODS: For Tennessee children 0 to 4 years of age, we used birth certificates to obtain data on maternal education, age, and parity; these risk factors were used to classify children into low- and high-risk groups. The outcome was death from injury, as determined from linked death certificates. Between 1978 and 1995, injury mortality rates were calculated for six 3-year periods for low- and high-risk children. RESULTS: There were 1.5 million children 0 to 4 years of age who contributed 4.9 million child-years. The high-risk group contributed 28% of all child-years. There were 673 injury deaths in the high-risk group, 48.9 deaths per 100,000 child-years, and 586 deaths in the low-risk group, 16.8 deaths per 100,000 child-years. The injury mortality rate for low-risk children decreased from 20.7 to 15.7 per 100,000 child-years between the 1978-1980 and 1981-1983 periods; thereafter it remained relatively stable. For high-risk children, the injury mortality rate decreased from 50.9 to 43.5 per 100,000 between the 1978-1980 and 1981-1983 periods, remained mostly unchanged through 1992, and then increased sharply in the 1993-1995 period to 64.1 per 100,000 child-years. The disparity between high- and low-risk children widened from 29.3 (95% confidence interval, 25.1-33.5) excess deaths per 100,000 for 1978 through 1991 to 46.9 (95% confidence interval, 35.9-57.9) in 1993 through 1995. CONCLUSIONS: In Tennessee, maternal education, age, and parity consistently identified a population of children at increased risk of injury mortality. For these high-risk children, there has been no substantial reduction in injury mortality in high-risk young children during the last 18 years.

New use of anticonvulsant medications among children enrolled in the Tennessee Medicaid Program.

OBJECTIVE: To describe the new use of anticonvulsant medications among children enrolled in the Tennessee Medicaid program. DESIGN: A retrospective cohort study. PATIENTS: New users of anticonvulsant medications in 1992 were identified from the 206,098 children (aged 0-18 years) enrolled continuously for 12 months in the Aid to Families With Dependent Children program or foster care program of Tennessee Medicaid. MAIN OUTCOME MEASURES: New users were categorized according to the diagnosis codes of health care encounters occurring 90 days before to 90 days after the first anticonvulsant prescription was filled as having diagnoses consistent with (1) epilepsy or convulsions, (2) neonatal seizures, (3) central nervous system disease, (4) no epilepsy diagnoses but diagnoses for which anticonvulsants might appropriately be used (jaundice, headaches, or psychiatric disorders), or (5) no diagnoses for which an anticonvulsant might appropriately be used. The children in each group were described according to sociodemographic variables, with logistic regression used to analyze variations in the subsequent filling of anticonvulsant prescriptions. RESULTS: Of 647 children continuously enrolled in the Tennessee Medicaid program who were new anticonvulsant users in 1992, 58% had at least 1 health care encounter coded as epilepsy or convulsions, 2% had a diagnosis of neonatal seizures, 8% had central nervous system diagnoses, 16% had specific nonepilepsy diagnoses (jaundice, headache, or psychiatric diagnoses), and 16% had no diagnoses for which anticonvulsants might appropriately be prescribed. For children with epilepsy diagnoses, white race (P = .002) and undergoing tests (P < .001) were independent predictors of a child filling 6 or more prescriptions in the year following the first prescription CONCLUSIONS: A large proportion of new users of anticonvulsants among children enrolled in the Tennessee Medicaid program received these medications for indications other than epilepsy. For children with epilepsy diagnoses, there was considerable variation in the subsequent filling of prescriptions. Further analysis of these variations in practice will allow for the development of policies that will maximize benefit for children who need anticonvulsant therapy, while diminishing unnecessary exposure to potentially toxic drugs for children who do not.