The art and science of chart review.

BACKGROUND: Explicit chart review was an integral part of an ongoing national cooperative project, "Using Achievable Benchmarks of Care to Improve Quality of Care for Outpatients with Depression," conducted by a large managed care organization (MCO) and an academic medical center. Many investigators overlook the complexities involved in obtaining high-quality data. Given a scarcity of advice in the quality improvement (QI) literature on how to conduct chart review, the process of chart review was examined and specific techniques for improving data quality were proposed. METHODS: The abstraction tool was developed and tested in a prepilot phase; perhaps the greatest problem detected was abstractor assumption and interpretation. The need for a clear distinction between symptoms of depression or anxiety and physician diagnosis of major depression or anxiety disorder also became apparent. In designing the variables for the chart review module, four key aspects were considered: classification, format, definition, and presentation. For example, issues in format include use of free-text versus numeric variables, categoric variables, and medication variables (which can be especially challenging for abstraction projects). Quantitative measures of reliability and validity were used to improve and maintain the quality of chart review data. Measuring reliability and validity offers assistance with development of the chart review tool, continuous maintenance of data quality throughout the production phase of chart review, and final documentation of data quality. For projects that require ongoing abstraction of large numbers of clinical records, data quality may be monitored with control charts and the principles of statistical process control. RESULTS: The chart review module, which contained 140 variables, was built using MedQuest software, a suite of tools designed for customized data collection. The overall interrater reliability increased from 80% in the prepilot phase to greater than 96% in the final phase (which included three abstractors and 465 unique charts). The mean time per chart was calculated for each abstractor, and the maximum value was 13.7 +/- 13 minutes. CONCLUSIONS: In general, chart review is more difficult than it appears on the surface. It is also project specific, making a "cookbook" approach difficult. Many factors, such as imprecisely worded research questions, vague specification of variables, poorly designed abstraction tools, inappropriate interpretation by abstractors, and poor or missing recording of data in the chart, may compromise data quality.

Optimal reporting of health care process measures: inferential statistics as help or hindrance?

In this paper we discuss the appropriate application of inferential statistics to practice profiles and other measures of care. To accomplish our objectives, we first describe the relative merits of measuring three well-recognized domains of medical quality: structure, process, and outcome. Next, we discuss inferential statistics as used in quality improvement. We then describe several common circumstances that arise in the measurement of medical care, giving attention to the application of inferential statistics to each situation. We end with a brief discussion of statistical techniques commonly used in the measurement of quality and challenges that arise with their use.

The pediatric costs of strategies for minimizing the risk of early-onset group B streptococcal disease.

OBJECTIVE: To estimate the pediatric costs associated with the Centers for Disease Control and Prevention (CDC) guidelines for the management of infants born to mothers receiving intrapartum antimicrobial prophylaxis for prevention of early-onset group B streptococcal disease. METHODS: For an annual United States birth cohort of 3.95 million infants, we estimated the cost of pediatric care provided to full-term asymptomatic infants when pediatricians followed the CDC algorithm for the management of infants exposed to intrapartum antimicrobial prophylaxis under culture-based and risk factor-based maternal care approaches. We calculated the relative contribution of pediatric costs to the total costs of preventing a case of early-onset group B streptococcal sepsis. RESULTS: Total pediatric costs were $41 million for a culture-based approach and $33 million for a risk factor-based approach. Hospital and physician costs accounted for more than 78% of this total. The majority (over 95%) of the pediatric costs were associated with vaginal deliveries. Incorporating pediatric costs into previous cost-effectiveness analyses increased the cost per sepsis case averted by as much as 51% for culture-based strategies and by as much as 112% for risk factor-based strategies. Pediatric costs varied with the average length of stay for full-term infants and with the average cost of a hospital day. CONCLUSION: Substantial pediatric costs are associated with the implementation of an obstetric strategy for minimizing the risk of early-onset group B streptococcal disease. Such costs should be included in future cost-effectiveness analyses of different strategies for minimizing the risk of group B streptococcal disease in newborns.

The importance of preparing medical students to manage different types of uncertainty.

Price competition and other aspects of the changing health care environment are threatening many academic health centers (AHCs) and causing them to reassess their education and research missions. In order to design effective AHCs for the next century, medical leaders must define the unique competencies needed by tomorrow's physicians and describe the educational enterprises required to produce physicians with these competencies. Two of the most important of these competencies are the ability to manage the uncertainty associated with creating clinical paradigms and the ability to manage the uncertainty associated with managing care delivery. Creating clinical paradigms involves (1) developing knowledge about disease categories and (2) developing knowledge about the most appropriate therapy for a disease in a particular category. Both these tasks involve uncertainty. The second type of uncertainty is associated with managing care delivery and is largely a matter of optimizing current clinical paradigms. The challenges are (1) to correctly assign patients' diseases to existing disease categories, and (2) to correctly choose and manage the delivery of the most appropriate therapies to these patients. Currently, AHCs are more competent in managing--and educating students to manage--the uncertainty involved in creating clinical paradigms. But there is an increasing demand for physicians who manage the second type of uncertainty associated with care delivery. The authors conclude that in order to remain viable, AHCs, and particularly their medical schools, must broaden their educational goals so that students can learn to manage both forms of uncertainty.

Preferences of pregnant women and physicians for 2 strategies for prevention of early-onset group B streptococcal sepsis in neonates.

OBJECTIVES: To assess preferences of pregnant women, pediatricians, and obstetricians for the policies of the American College of Obstetrics and Gynecology (ACOG) and American Academy of Pediatrics (AAP) for reducing the incidence of neonatal group B streptococcal (GBS) sepsis. DESIGN: An interactive interview using a computer-based decision-making model (analytic hierarchy process) and a self-administered survey assessing the interview process. SETTING: An obstetric clinic at a university center and offices of practicing physicians from the Birmingham, Ala, area. PARTICIPANTS: Ninety-two pregnant women selected by a systematic sampling technique and 40 pediatricians and 40 obstetricians selected randomly. MAIN OUTCOME MEASURE: Ranking of the ACOG and AAP policies and the 5 criteria on which the decision was based: risk of infection to an infant, knowledge of maternal GBS status, risk of anaphylaxis to mother, diagnostic tests received by healthy infants, and cost. Satisfaction with the interview process also was measured. RESULTS: Eighty-three women (90%), 40 pediatricians, and 40 obstetricians (100%) provided responses suitable for analysis. Sixty-seven pregnant women (81%), 26 pediatricians (65%), and 6 obstetricians (15%) preferred the AAP strategy. The ACOG policy was the preferred strategy by 34 (85%) obstetricians. The 3 groups ranked risk of infection in an infant as the most important criterion in their decisions. Ranks for the other criteria differed among the 3 groups. Women ranked knowledge of maternal GBS status more important than did pediatricians and obstetricians. Thirty obstetricians (75%), 35 pediatricians (87.5%), and 72 pregnant women (86.7%) liked the interview. Seventy-three women (88%), 29 pediatricians (72.5%), and 17 obstetricians (42.5%) thought physicians should use this type of interview to assist in managing patients. CONCLUSIONS: Pregnant women, pediatricians, and obstetricians had different priorities when making a decision about GBS policies. These differences led obstetricians to prefer a different policy than that of pediatricians and pregnant women. Obstetricians were less likely to endorse the use of this decision-making technique in their practice than were patients and pediatricians.

Impact of care setting on cost and quality under Medicaid.

Medicaid claims data were used to compare the costs and care quality of ambulatory visits for two childhood illnesses, urinary tract infection (UTI) and suppurative otitis media (OM), in the fee-for-service Medicaid program in Alabama across three care settings: offices where patients had been seen before, offices where patients had not been seen before, and outpatient hospital departments. Forty percent of UTI visits and 46 percent of OM visits occurred in return office settings. Visits to outpatient hospital and first-time office settings were more expensive than those to return office settings, due to the billing of facility fees and the provision of additional services. Adherence to common measures of quality of care for both types of visits was low; 52 percent of UTI visits included urine cultures and 40 percent of OM visits included recheck visits. Adherence to these quality measures was significantly lower in visits occurring in hospital settings.

How rural physicians compare on cost and quality measures for Medicaid ambulatory care episodes.

This study compares the costs and quality of episodes of care for two common childhood illnesses, urinary tract infections (UTI) and otitis media (OM), across providers practicing in rural, small town, and urban counties in Alabama in 1992. The data source is Medicaid claims data for children under age 8 who were treated for these conditions. The study found that episodes cared for by rural providers were less expensive than episodes cared for in other locations, both because fewer rural episodes included outpatient facility charges and because fewer ancillary services were provided in rural settings. Researchers also found that, even controlling for physician characteristics and patient demographic and utilization factors, rural episodes were significantly less likely to include two process measures of quality of care: fewer rural UTI episodes included urine cultures, and fewer rural OM episodes included follow-up visits. This study suggest that, as a group, rural physicians may have a favorable cost profile but a potentially unfavorable care content profile, compared with other physicians. Both practice profile data and explicit care recommendations need to be available to physicians so thy can monitor, defend, or alter their clinical practices.

Comparison of inpatient charges between academic and nonacademic services in a children's hospital.

OBJECTIVE: To compare inpatient hospital charges generated within a children's hospital by academic and nonacademic pediatric services for common medical diagnoses. METHODS: Hospital admissions to a free-standing children's hospital between 9/1/90 and 8/30/94 were selected for patients who were hospitalized 1 to 14 days, with one of six selected diagnoses, and with discharge attending of record either a private pediatrician or an academic subspecialist. Discharge diagnoses, based on ICD-9 codes, included asthma (n = 1983), bronchiolitis (n = 692), gastroenteritis (n = 733), rule out sepsis (n = 1065), urinary tract infection (n = 516), and viral meningitis (n = 288). Charges associated with patient records were dichotomized as above or below the median charge for each diagnostic category. Each category was analyzed separately using a logistic regression model where the dichotomous-dependent variable was charges above the median charge for each diagnosis. Independent variables included physician type, payor status, patient residence, ICD-9 code as primary or secondary diagnosis, patient age, and presence of complicating conditions. RESULTS: By univariate comparison, academic physicians cared for a higher percentage of underinsured patients, and their care was more expensive. Complicated claims were associated with higher charges than uncomplicated claims for all diagnostic categories. Academic and nonacademic physicians were equally likely to generate above-median charges for five of the six diagnostic categories when controlling for confounding factors. A linear regression model in which charge was the dependent variable generated similar results. CONCLUSIONS: Within the same pediatric health care facility, no consistent difference was found between charges incurred on academic vs private inpatient services.

Attitudes of academic pediatricians with a specific interest in child abuse toward the spanking of children.

OBJECTIVE: To evaluate the attitudes of academic child abuse professionals toward spanking, the effect of context and mode of administration on their attitudes toward spanking appropriateness, and what they teach residents about spanking. DESIGN: A survey. PARTICIPANTS: Convenience sample of 114 members of the Ambulatory Pediatric Association's Special Interest Group on Child Abuse and Neglect. MAIN OUTCOME MEASURES: Respondents were asked if spanking was an appropriate disciplinary option for children 2, 5, and 8 years of age who refused to go to bed, ran into the streets without looking, or hit a playmate. Respondents also rated the appropriateness of spanking in 6 additional scenarios where the setting in which spanking occurred was varied. Respondent's teaching practices relative to spanking observed during a clinic visit were also elicited. RESULTS: The response rate was 70%; 39% thought spanking was appropriate sometimes. The context and mode of spanking affected the acceptance of spanking. All respondents thought that some response was appropriate when spanking was observed during a continuity clinic visit. However, only 29% of respondents taught residents how to handle such situations. CONCLUSIONS: Most academic child abuse professionals believe that spanking is inappropriate and their beliefs are influenced by the context in which spanking occurs. Little is taught about how to manage spanking observed in a clinical setting.

Impact of maternal group B streptococcal screening on pediatric management in full-term newborns.

BACKGROUND: The American Academy of Pediatrics strategy to prevent early-onset neonatal sepsis with group B streptococcus (GBS) relies on maternal antepartum GBS cultures, while the American College of Obstetrics and Gynecology strategy does not. OBJECTIVE: To evaluate the impact of the 2 strategies on the care of asymptomatic full-term newborns. DESIGN/SETTING: Self-administered survey mailed to a national random sample of US pediatricians who were members of the American Academy of Pediatrics. PARTICIPANTS: A total of 461 members of the American Academy of Pediatrics who routinely care for newborns. MAIN OUTCOME MEASURE: Self-report of diagnostic and treatment strategies for asymptomatic full-term newborns who were born under different clinical scenarios. Maternal risk factors, antepartum maternal GBS screening status, and maternal treatment with intrapartum antibiotics were varied across the scenarios. RESULTS: Pediatricians treating asymptomatic full-term newborns born to risk factor-negative mothers reported ordering tests (63.3% in GBS-positive cases vs 6.7% with GBS unknown; P = .001) and antibiotics (21.5% in GBS-positive cases vs 0.9% with GBS unknown; P = .001) more frequently when presented with a positive maternal GBS screening result. Maternal intrapartum treatment had little impact on pediatric practice when risk factors were absent. In risk factor-positive mothers, pediatricians reported an increase in their antibiotic usage in response to a positive maternal GBS screen (61.8% in GBS-positive cases vs 36.9% with GBS unknown; P = .001). In risk factor-positive mothers with unknown results of GBS screening, use of intrapartum antibiotics increased the number of pediatricians who reported that they would prescribe antibiotic therapy. CONCLUSIONS: Obstetrical strategies to decrease the risk of neonatal GBS sepsis increase pediatric services provided to full-term healthy newborns. This increase in services by pediatric practices is likely to be greater with the screening-based strategy recommended by the American Academy of Pediatrics.

The changing academic health center. The death of the traditional academic physician.

Organizational change is required if academic health centers (AHCs) are to survive the decreased societal commitment to them. The changes will generate significant emotional responses in the physicians employed by such institutions. This article presents an analogy between the reactions of academic physicians to the changes they are experiencing, and the stages of grief that Dr. Kübler Ross described in terminally ill patients. By placing physician responses in this context, emotional responses to organizational changes can be more easily understood and managed, allowing academic physicians to devote more energy to facing the threats to AHCs in an innovative and constructive manner.

A way to approach the strategic decisions facing academic health centers.

Rapid changes taking place in the various markets served by academic health centers (AHCs) are forcing these institutions to make difficult strategic decisions that may change AHCs' historic priorities. The authors present an approach that can help AHCs visualize possible new configurations of their traditional services of research, education, and clinical care. This approach is based on successful strategic management methods from the private sector and involves a three-dimensional "topography of services" encompassing all possible configurations of AHCs' services. From among the many possible configurations, the authors discuss three in detail. The historic one, which they call the traditional model, is characteristic of AHCs that give high priority to biomedical and clinical research, have broad medical education activities, and deliver comprehensive, high-quality clinical care. In the future, this configuration will be rare, and two others are likely to predominate. First is the "revised" traditional model, which would offer "boutique" clinical services, biomedical research, and medical education for MD-PhD students, residents, and fellows seeking tertiary care or academic careers. The patient care required for undergraduate medical education and clinical research would be provided by partnerships with community-based providers. Second is the academic services model, which would focus on competitive primary and secondary clinical services, health services and operations research, and primary care medical education. The authors discuss the implications of these models for AHCs' organizational structures and faculty incentives. They conclude that the clarity with which AHCs' strategic decisions are made and communicated to faculties and the incentive systems that are selected to motivate faculty and to provide the selected services may ultimately determine which institutions survive.

Patterns of care received by Medicaid recipients with urinary tract infections.

BACKGROUND: Urinary tract infections (UTIs) occur commonly in children and may lead to substantial morbidity. Most experts recommend urine cultures for diagnosing UTIs in children. In addition, most experts recommend imaging studies in a portion of children diagnosed with UTIs. PURPOSE: The purpose of this study was to assess how rates of performance of urine cultures and imaging studies for children in the Alabama Medicaid program diagnosed with a UTI vary by patient demographics, provider characteristics, and service locations. METHODS: The study design was a retrospective review of Alabama Medicaid claims data. Children were included as UTI cases if they had a Medicaid claim for urinary tract infections during 1991, were continuously enrolled in Medicaid for that year, and were younger than 8 years of age. Claims were grouped into episodes of care, and episodes were assigned to a diagnosing physician. Physician locations were classified as rural, suburban, or urban using demographic data. Specific laboratory and imaging procedures were identified using CPT codes (Physician's Current Procedural Technology Codes, 4th Edition). RESULTS: We identified 404 episodes of UTI occurring in 380 children. Only 47% of episodes were associated with claims for urine cultures. Claims for urine cultures were more frequently filed by pediatricians in urban locations. In the subset of 114 patients with multiple UTI episodes, only 68% had imaging studies specific for the urinary tract. Only 44% received both a voiding cystourethrogram and renal ultrasound. CONCLUSIONS: Claims data suggest that physicians underuse urine cultures in diagnosing UTIs in Alabama pediatric Medicaid recipients. Urban-based pediatricians perform better than other types of physicians. Imaging studies are also used less frequently than is commonly recommended.

Impact of patient history on residents' evaluation on child sexual abuse.

OBJECTIVE: To determine if historical information influences residents' interpretation of physical findings in sexually abused children. METHODOLOGY: In a pediatric residency training program, all residents viewed 15 slides of children's genitalia (8 normal, 7 abnormal) with either a history specific for sexual abuse or one which was nonspecific. Three weeks later the same slides were viewed but with the alternate history scenario. The residents were asked if the physical findings were specific for sexual abuse. RESULTS: Sixty-four percent of residents completed both surveys. Correct response rate did not vary by gender or year of training. Responses were most often correct when the slide and history were normal (87%). Responses were least accurate when normal historical information was presented with abnormal slides (49%). A logistic regression model demonstrated that residents were less accurate when history and physical did not agree (95% CI = .54- .78). Reexamination of the data using areas under the Receiver Operating Characteristic (ROC) curve confirmed that residents performed on a less accurate ROC curve when the slide and history were incongruent (p < .01). CONCLUSION: Incongruency between patient history and physical exam findings negatively affected this group of residents' ability to discriminate between abuse and nonabuse findings.

Gun ownership and counseling of Alabama pediatricians.

OBJECTIVE: To examine the personal behavior of primary care pediatricians in Alabama with respect to a gun control policy. The American Academy of Pediatrics supports removal of handguns from homes, safe gun storage in homes with guns, and counseling patients about gun safety practices. DESIGN: Survey. SETTING: Primary care pediatricians. PATIENTS OR OTHER PARTICIPANTS: Population-based sample. INTERVENTIONS: None. MAIN OUTCOME MEASURES: The proportion of pediatricians who were gun and handgun owners was examined. In addition, the safety behaviors and counseling practices reported by pediatricians were examined. RESULTS: The response rate for pediatricians was 67%. Fifty percent of pediatricians reported owning a gun. Thirty-four percent of pediatricians had a handgun in their household. Eleven percent of pediatricians had unsafe gun practices, where unsafe was defined as having a loaded gun in the home or car at least some of the time. The most common reason for owning a handgun or having a loaded gun in the home or car was personal protection. Only a third (33%) of pediatricians routinely counseled their patients about gun safety. Long gun owners were less likely to counsel patients about bike helmet safety and were less likely to use car seats and bike helmets for their own children. CONCLUSIONS: A large proportion of Alabama pediatricians are gun owners, but most store weapons safely. Although long gun owners were less likely to use car seats or bike helmets and less likely to counsel patients regarding bike helmet use, no effect of gun ownership on counseling about gun safety was identified.

The impact of CQI on human resources management.

If CQI is to become a mind-set and not simply a management fad, adjustments need to be made in all aspects of management, especially human resources management. This article will consider the impact of CQI on human resources philosophy and practice in health services organizations. The effects will be illustrated by the experiences of a group of human resources managers and the organizations in which they work.

Quality in the interorganizational setting.

For many patients, important health services are delivered by a process of care that spans several service organizations. Issues influencing the quality of care delivered across organizations are infrequently discussed in the health care quality literature. In this paper, interorganizational quality problems that detract from the care received by children with spina bifida, lead poisoning, and children who have been sexually abused are identified. Using concepts from the organizational behavior literature, both structural and group process approaches to addressing these problems are discussed to enhance the care received by children. A broader conceptualization of care delivery is required if patients whose care spans institutions are to benefit from quality improvement efforts.