CD26+ Cord Blood Mononuclear Cells Significantly Produce B, T, and NK Cells

BACKGROUND

Umbilical cord blood (UCB) is an alternative source of hematopoietic stem cell transplantation (HSCT), used in Leukemia treatment. CD26+ cells, a fraction of CD34+cells, are a major population of UCB cells which negatively regulate the in vivo homing and engraftment of HSCs. CD26 is highly expressed in various cells such as HSCs, immune cells, fibroblasts, and epithelial cells. It has been shown that the inhibition of the CD26 on CD34+ cells improves the efficiency of Hematopoietic Stem and Progenitor Cell (HPC) transplantation.

OBJECTIVE

To evaluate the relationship between the production of B, T, and NK cells from the CD26+ fraction of cord blood mononuclear cells.

METHODS

Cord blood mononuclear cells were cultured for 21 days using different combinations of stem cell factors (SCF), Flt3 ligand (FL), IL-2, IL-7, and IL-15. The harvested cells were then analyzed by flowcytometry every week for 21 days.

RESULTS

T cell differentiation from CD26 subset of cord blood mononuclear cells increased by using IL-2 and IL-7. Our data showed that IL-2 and IL-7 significantly affected the generation of B cells from CD26 positive cord blood mononuclear cells. On the other hand, NK (NKp46+) derived CD26+ cells increased by IL-15 and IL-2.

CONCLUSION

Taking all into account, we conclude that B, T, and NK cells can differentiate from the CD26+ subset of mononuclear cord blood cells by using key regulatory cytokines.

Tutorial on risk-adjusted P-charts

Several books are available that describe how to construct control charts. Unfortunately none of these books show how to adjust control charts for the severity of the patients' illness. Without such adjustments control charts not only loose face validity but could lead to wrong conclusions. In this article we provide a step-by-step guide regarding how to control for changes in patients' severity of illness across time periods. We illustrate the use of the approach in examining falls in nursing homes. A control chart was constructed to compare observed falls to the residents' risk of fall measured on admission. Analysis led to conclusions radically different from an unadjusted chart.

A survey of 92 quality improvement projects

BACKGROUND

Studies focusing on the impact of improvement efforts on the organization have yielded mixed results, which has increased interest in comparing the processes of improvement used. Data for a convenience sample of 92 quality improvement (QI) projects in 32 organizations were gathered from interviews and self-reported surveys from 1998 to 2000. A self-administered questionnaire was developed to measure 70 characteristics of improvement projects.

RESULTS

Most (80%) of the improvement projects were conducted by hospitals or clinics affiliated with hospitals. The projects took an average of 13 months from the team's first meeting to the end of the pilot study. Project teams met 14 times (approximately once a month) and spent 1.5 hours per meeting. Some projects did not measure the impact, others did not intend to have a specific impact, and still others measured but did not achieve the planned impact.

DISCUSSION

Patients and employees may be benefitting from improvement projects, but organizations may not be leveraging these improvements to reduce cost of delivery or increase market share. Considerable variation in the projects' impact raises the question of the need to improve the improvement methods. Generalization from this study should be made with caution, as data were based on a self-selected convenience sample of organizations. Furthermore, respondents did not complete all items, and missing information may affect the conclusions. The data on current improvement practices that are provided in this study can serve as baseline data against which rapid improvement efforts can be judged.

Tutorial on risk adjusted X-bar charts: applications to measurement of diabetes control

Two data elements are needed for constructing a risk-adjusted X-bar chart. They are a continuous observed outcome collected over time across a sample of patients, and an expected outcome for each patient. The authors selected blood glucose levels to reflect diabetes control. They then reviewed data showing blood glucose levels of 60 Type 2 diabetes patients in a family practice clinic of five providers. The data covered a period of 21 consecutive months. They present the data relating to two of the providers using this set of data to demonstrate how to create a risk-adjusted X-bar control chart.

Continuous improvement evaluation: a framework for multisite evaluation studies

This article proposes the use of the continuous improvement evaluation (CIE), a framework for multisite demonstration or evaluation studies. This framework is designed for studying intervention programs that change during the evaluation. The development of family drug courts is provided as an example. CIE relies on outcome data collected over time and benchmarked across similar cases in comparison sites; thus, this study was designed to collect data on effectiveness of intervention programs at multiple sites and over time. A weight is calculated for similarity of any two cases based on features they share. In statistical process control charts, these weights are used to compare outcomes at the site against the average of similar cases in comparison groups. Once data are benchmarked, program staff meet to discuss process changes that have led to improvements in outcomes. To ensure that intervention programs have access to evaluation reports on demand, information technology is used to collect, clean, and pool data. Computers generate study reports, and evaluators review reports after release to clients. Statistical tools can be used to evaluate changing programs. Traditional evaluators may be concerned about some threats to validity associated with CIE. The article concludes with a discussion of typical threats to validity and how these threats are addressed in the CIE framework.

Management matters: technology succeeds when management innovates

It is widely believed that better technology means better business: now that consumers and patients have direct access to computers and information services, the cost of care should decrease and services improve. Yet even with the advent of computer technology and the phenomenal growth of the Internet, costs have increased and quality problems have persisted. Far more important than the technology, then, is how business is conducted and systems are organized. Despite overwhelming evidence that computer services can significantly reduce the costs of care, healthcare organizations have not adopted the changes, or have tried and failed. This article explores what it will take to succeed. We propose a list of necessary nontechnical changes. Patient expectations will change the nature of care; clinicians' roles and training, the gatekeeper profession, healthcare financing and bundling of services, and capital costs will all change. In the end, management innovations make the difference between the success and failure of new technology. Technology is important, but it is not enough. Without new practices, we can buy the technology but will fail to effectively use it. Unless management modifies the very nature of its business, technology's promise to the healthcare industry will go unfulfilled.

Integrating the Internet into health administration education: a report from AUPHA's Faculty Internet integration task force

HRSA funded a survey to determine what Internet resources would be most useful to AUPHA membership. This manuscript describes the Internet-intensive survey methodology, reports the survey results, and lists the task force recommendations. The task force used sequential questionnaires posted on the Web to gather both potentially useful Internet resource ideas and membership perceptions of the importance of each idea. Resources recommended by survey participants and the Task Force members emphasized potential improvements to the AUPHA and AUPHA-member Web pages.

Continuous self-improvement: systems thinking in a personal context

BACKGROUND

Continuous quality improvement (CQI) thinking and tools have broad applicability to improving people's lives--in continuous self-improvement (CSI). Examples include weight loss, weight gain, increasing exercise time, and improving relationship with spouse. In addition, change agents, who support and facilitate organizational efforts, can use CSI to help employees understand steps in CQI. A STEP-BY-STEP APPROACH: Team members should be involved in both the definition of the problem and the search for the solution. How do everyday processes and routines affect the habit that needs to change? What are the precursors of the event? Clients list possible solutions, prioritize them, and pilot test the items selected. One needs to change the daily routines until the desired behavior is accomplished habitually and with little external decision.

DISCUSSION

CSI is successful because of its emphasis on habits embedded in personal processes. CSI organizes support from process owners, buddies, and coaches, and encourages regular measurement, multiple small improvement cycles, and public reporting.

Internet course on total quality management

The Internet provides new opportunities to information-based industries such as universities. If a vast majority of potential customers of our programs have access to the Internet, then it is conceivable that many universities will begin teaching Internet courses and that such courses could, in many cases, replace or at least augment existing face-to-face classroom activities. This paper is an account of one such course at Cleveland State University and the type of policy issues raised by the course at this institution.

Validity of three measures of severity of AIDS for use in health services research studies

Policy makers and hospital managers often use severity adjustments as a control for patient mix differences when evaluating outcomes of care. Unfortunately, few indices are carefully examined and therefore the evaluations based on these methods of severity adjustment are suspect. This paper examines the accuracy of three indices for measuring the severity of illness of AIDS patients. We examine the Diagnosis-Based Severity Index (DBSI), a modified version of DBSI referred to as MDBSI and the Composite Laboratory Index (CLI) in predicting survival of AIDS patients at one medical centre. We analysed the correlation between indices and months of survival. We also examined the percentage of variance in survival months explained by each index separately and together. Finally, we used survival analysis to examine whether DBSI classifies patients in groups with distinct patterns of survival. Only patients who had died were included in the analysis so that information on the patients' full course of illness was available. Of the 91 patients abstracted, 81 cases had date of AIDS, date of death, and the CLI. These 81 cases were the focus of the analysis. Both CLI and DBSI were predictive of months of survival but were not correlated to each other. Predictions of months of survival were improved if both indices were used together rather than separately. Survival analysis confirmed that patients classified by DBSI had distinctly different survival patterns. Each index measures different aspects of the severity of the patient's condition and when possible both indices should be used together. When laboratory data are not available, e.g. in Medicaid administrative files, the use of DBSI may be reasonable.

Policy issues relevant to evaluation of interactive health communication applications. The Science Panel on Interactive Communication and Health

This article provides an analysis of policy-related issues associated with the evaluation of interactive health communication (IHC) applications. These include an assessment of the current health and technology policy environment pertinent to public (government, education, public health) and private (medical care providers, purchasers, consumers, IHC developers) IHC stakeholders and discussion of issues likely to merit additional consideration by these stakeholders in the future.

Using PDSA (Plan-Do-Study-Act) to establish academic-community partnerships: the Cleveland experience

The Schools of Medicine and Nursing at Case Western Reserve University and the Program in Health Administration at Cleveland State University have created an interdisciplinary course in continuous improvement that emphasizes learning through experience, accommodates a large number of students, and has created new partnerships with Cleveland area health care organizations. An approach that respects these partners as customers and refines the relationships with serial tests of change (e.g., PDSA) has contributed significantly to this program's success.

Rapid improvement teams

BACKGROUND

Suggestions, most of which are supported by empirical studies, are provided on how total quality management (TQM) teams can be used to bring about faster organizationwide improvements.

SUGGESTIONS

Ideas are offered on how to identify the right problem, have rapid meetings, plan rapidly, collect data rapidly, and make rapid whole-system changes. Suggestions for identifying the right problem include (1) postpone benchmarking when problems are obvious, (2) define the problem in terms of customer experience so as not to blame employees nor embed a solution in the problem statement, (3) communicate with the rest of the organization from the start, (4) state the problem from different perspectives, and (5) break large problems into smaller units. Suggestions for having rapid meetings include (1) choose a nonparticipating facilitator to expedite meetings, (2) meet with each team member before the team meeting, (3) postpone evaluation of ideas, and (4) rethink conclusions of a meeting before acting on them. Suggestions for rapid planning include reducing time spent on flowcharting by focusing on the future, not the present. Suggestions for rapid data collection include (1) sample patients for surveys, (2) rely on numerical estimates by process owners, and (3) plan for rapid data collection. Suggestions for rapid organizationwide implementation include (1) change membership on cross-functional teams, (2) get outside perspectives, (3) use unfolding storyboards, and (4) go beyond self-interest to motivate lasting change in the organization.

CONCLUSIONS

Additional empirical investigations of time saved as a consequence of the strategies provided are needed. If organizations solve their problems rapidly, fewer unresolved problems may remain.

Participation in teleconference support groups: application to drug-using pregnant patients

This paper describes our experience using telephone conference calls to conduct support groups for chemically-dependent women. Forty-seven women agreed to participate in regular, weekly support groups that were conducted by two chemical dependency counselors. Counselors attempted to conduct 59 support groups via a telephone conferencing system. Our data indicated that attendance at these sessions was poor. Seventeen sessions had no participants. No group sessions were attended by 4 or more women, and only 3 sessions (7.1%) included 3 participants. Very few clients made regular use of the support groups. Only 4 clients (8.5%) participated in more than 3 group sessions. Our findings suggest that teleconferencing may not be the most effective method for providing support services to chemically-dependent women.

Educating patients at home. Community Health Rap

OBJECTIVES

The authors analyzed the impact of home health education by studying the impact of a computer service called Community Health Rap. When patients call this service, the computer records their questions and alerts an expert who records a response. Subsequently, the computer alerts the patient that the question has been addressed.

METHODS

Subjects included a group of 82 pregnant women who had used cocaine during or 1 month before pregnancy (as reported by the woman) and a group of residents of zip code areas with the lowest income in Cleveland. From the drug-using pregnant women, we collected data regarding satisfaction with Community Health Rap, usage of Community Health Rap per month, self-reported health status (using the General Health Survey), and the extent of drug use (using the Addiction Severity Index). Trained coders also classified the nature of questions posed to the Community Health Rap by either the pregnant women who abuse drugs or the members of target households. Among the pregnant women who abuse drugs, we compared the differences between those who used the service and those who did not. To control for baseline differences between the two groups, analysis of co-variance was used with exit values as the dependent variables, the baseline values as the co-variates, and participation in the Community Health Rap as the independent variable.

RESULTS

Almost half (45%) of poor, undereducated subjects who lived in inner urban areas used the computer service. Content analysis of Community Health Rap messages revealed that subjects had many questions that were of a social nature (regarding sex, relationships, etc), in addition to medical questions. Analysis showed that poor health status, more frequent drug use, lower education, and age did not affect regular use of Community Health Rap service. No health outcomes or utilization of treatment were associated with regular use of Community Health Rap. One exception, however, was that regular users of Community Health Rap reported slower improvement of their pain than those who did not use Community Health Rap.

CONCLUSIONS

These data suggest that expansion of information services to households will not leave the poor and the undereducated population groups "behind." They will use computer services, though such services may not have an impact on their health status or cost of care.

Computer services for patients. Description of systems and summary of findings

This article (1) describes a number of computer services provided to patients' homes, (2) summarizes four evaluation studies reported in this supplement, and (3) suggests direction for future research. The studies reviewed in this article suggest that computer services have little impact on patients' health status but a significant impact on use of services. Some computer services (eg, reminders) increase client visits and other services (eg, voice bulletin board) reduce clinic visits. The article suggests that a paradigm for conducting research on impact of home computer services on care of patients is missing and should be developed.

A randomized trial of a telecommunications network for pregnant women who use cocaine

OBJECTIVES

The authors describe computer services designed to make patients better informed and more motivated to participate in treatment. Patients use these services through their touch-tone telephone; access to a home computer or a modem is not needed.

METHODS

The authors tested the impact of these computer services on the management of 179 pregnant women who had used cocaine during pregnancy or 1 month before pregnancy (as reported by the woman). Patients were randomly assigned to control and experimental groups; only the experimental group had access to the computer services. Patients were enrolled during the prenatal period and followed for 6 months after the birth of a live child. Self-reported data on the subjects' participation in drug treatment programs, health status (using the SF-36), and addiction severity (using the addiction severity index) were collected. The computer collected data on the frequency of the use of the service. The dependent variables were participation in formal treatment during the course of the project, and drug and alcohol use at exit interview. Multiple and logistic regressions were used to identify the effects of the intervention after controlling for demographic and baseline variables.

RESULTS

Data showed that poor, pregnant, undereducated clients who use drugs and lived in multiple residences could use the services; about one-third of clients used the services more than three times a week. Access to the service did not lead to significant change in patients' health status, drug use, or utilization of services. Use of the services did lead to significant changes in participation in drug treatment: subjects who used the service more than three times a week were 1.5 times more likely to participate in formal drug treatment than subjects who used the service less often. Participation in formal drug treatment, however, was not effective in reducing the drug or the alcohol use.

CONCLUSIONS

Almost all patients used the computer services to some extent, but there seems to be a threshold after which the use of the services had a more positive impact.

Electronic self-help and support groups

OBJECTIVES

The authors examined the impact of a voice bulletin board on the following: (1) participation in self-help efforts, (2) expression of emotional support, (3) development of solidarity within the group, (4) utilization of health-care services, and (5) health status of subjects.

METHODS

Subjects were 53 pregnant women who abused drugs. A quasi-experimental design with matched control group and observations before and after intervention were carried out. Clients in the control group were asked to participate in biweekly face-to-face meetings. Clients in the experimental group participated in the voice bulletin board. Experimental subjects had previous experience with computer services. Subjects reported their level of drug use, health status, and utilization of health services. They also reported on their symptoms, attitudes toward use of physician services, loneliness, willingness to disclose information in groups, and sense of solidarity with their group. The content of the communication among the experimental group was recorded and the utterances were classified as to the type of communication. Exit interviews were done 4 months after baseline interviews were conducted. Clients were paid to complete the baseline and the exit questionnaires, and 94% completed the exit questionnaires. The dependent variables were utilization or health status at exit; the co-variate was utilization or health status at baseline; and the independent variable was the group in which the subject participated.

RESULTS

Clients were eight times more likely to participate in the voice bulletin board than in the face-to-face meeting (alpha < 0.01). The majority of the comments left on the bulletin board (54.6%) were for emotional support of each other; no "flaming" or overt disagreements occurred. The more clients participated in the voice bulletin board, the more they felt a sense of solidarity with each other (alpha < 0.001). Members of the experimental group reported significantly lower rates of visiting outpatient clinics than members of the control group (alpha < 0.05). Lower utilization did not lead to poor health status or more drug use: There were no statistically significant differences in the health status and drug use between the experimental and the control groups.

CONCLUSIONS

Voice bulletin boards may be an effective method of providing support to mothers who have a history of drug use. Use of these services may lead to lower cost without worsening patients' health.

Computer reminders improve on-time immunization rates

OBJECTIVES

This study examines the effectiveness of computer-generated telephone reminders in improving infants receiving on-time immunizations. A computer called parents at home, reminded them of their child's visit, and asked if they could keep the appointment. If parents either canceled or failed to honor the appointment, the computer called back a few days later and asked them to reschedule.

METHODS

A medical assistant recruited 124 consecutive mothers to receive automated computer reminders. These mothers' infants were younger than 6 months, were being seen at an outpatient clinic for a first visit, and were patients of three attending physicians and three nurse practitioners. These infants were compared to 89 infants from the same clinic, in the same age range, who were being seen for the first time during the same period by the same providers but not contacted by the medical assistant. Subjects were selected from mothers who brought their infants for their first visit in an outpatient urban clinic that serves predominantly minority clients. A research assistant reviewed patients' medical records and collected the infants' birthday, mothers' age, race, source of payments, and the immunization record of the infants. Immunization was considered to be late if, at the time of the first visit, it was more than 30 days past due for any of the recommended immunizations of the American Academy of Pediatrics, except for Hepatitis B vaccine which was not recommended at the time of the study. The dependent variable was on-time immunization. The independent variables were age of the mother at baseline, age of the child at baseline, and membership in either the comparison or the experimental group. Chi-square tests and logistic regression were used to analyze the data.

RESULTS

The participation rate for appointments for the experimental group was 82%, as compared to a 69% overall participation rate for the clinic providers. The on-time immunization rate for experimental subjects was 67.8%, whereas the comparison group had an on-time immunization rate of 43.4% (differences were significant at alpha levels less than 0.01).

CONCLUSIONS

Computerized reminders sent to the parents led to an increase in participation rate at the clinic and an increase in on-time immunization for their infants.

A comparison of three techniques for rapid model development: an application in patient risk-stratification

Accurately risk-stratifying patients is a key component of health care outcomes assessment. And, many health care organizations increasingly are relying upon automated means for assistance in making patient risk-stratification decisions. Unfortunately, the process of outcome model development, as it is currently practiced, is both time consuming and difficult. We investigated the relative abilities of three modeling techniques (logistic regression, artificial neural network (ANN), and Bayesian) to rapidly develop models for risk-stratifying patients. Our results demonstrated that all three modeling techniques perform equally well in certain situations. However, the Bayesian model with conditional independence had the best overall performance. Unfortunately, none of the models were able to achieve the degree of accuracy which would be required in a medical setting.

A review of factors affecting treatment outcomes: Expected Treatment Outcome Scale

The Expected Treatment Outcome Scales was developed to gather information on clients who abuse drugs or alcohol, to assess their severity of illness, and to evaluate the effectiveness of drug treatment in nonrandomized clinical studies. The scale is based upon a multiattribute value model reflecting the opinions of an expert panel. The experts identified 25 variables, or predictors of relapse, from which 48 questions were constructed. Answers to the questions are individually scored. These scores are summed to produce an overall Expected Treatment Outcome score. This paper focuses on the development and preliminary validation of the Expected Treatment Outcome Scale. Results of our analysis show a correlation of .89 between the experts' average ratings of hypothetical clients and scores based on our scale. This finding suggests that the Expected Treatment Outcome Scale has face validity and accurately simulates the experts' judgments regarding treatment outcome. Further research is necessary to assess the reliability as well as the concurrent and predictive validity of our instrument.

Reaction to "talking" computers assessing health risks

The authors examined reactions to AVIVA-a talking computer that assesses health risk, gives priorities for risk reduction, discusses risk reduction methods, and refers callers to additional information. Subjects' reactions to AVIVA were compared to receiving health information from magazines, television, or health professional. Data were collected from 96 randomly chosen employees of Cleveland State University. Employees were invited to participate based on a stratified sample that encouraged enrollment of men and women and enrollment of faculty, professional and nonprofessional staff. The majority (71%) of subjects with access to AVIVA used it. Those who did not use AVIVA gave various reasons; less than 4% did not participate in AVIVA because they objected to a computer giving advice regarding health risks. Subjects rated AVIVA as more accurate, easier to understand, more convenient, more affordable, easier to use, and more accessible than health education received from television, magazines, or health professionals. None of the at-risk subjects sought additional information from a library of videotapes available to them. Furthermore, there was no statistically significant difference between the control and the experimental groups in the intent to reduce risk factors. Therefore, despite use and satisfaction with AVIVA, the authors concluded that there was no impact on subjects' behavior or intent to change behavior.

Automated monitoring of outcomes: application to treatment of drug abuse

This paper suggests a new approach for lowering follow-up costs, improving the delivery of health care, and monitoring treatment outcomes. An automated telephone follow-up system that calls, identifies, and interviews clients is an alternative method for monitoring patients that may be both reliable and cost-effective. To test the viability of such a system, the authors monitored a patient population that has historically been shown to be difficult to follow: recovering drug users and alcoholics. Forty-two subjects were asked to call the computer and complete interviews on a weekly basis for five months. Clients answered 25 recorded questions by pressing the keys on their telephone pads. The computer automatically analyzed the clients' answers and estimated a probability of relapse for each client. In addition, the computer automatically called subjects who failed to complete interviews at the scheduled times. The study showed that self-reported data collected by a computer are as reliable as data obtained through a written questionnaire and that clients are more willing to respond to computer interviews than to mailed written questionnaires. This study also provides preliminary data on the predictive ability of a questionnaire for predicting relapse.

An evaluation of factors influencing Bayesian learning systems

OBJECTIVES

To examine the influences of situational and model factors on the accuracy of Bayesian learning systems.

DESIGN

This study examines the impacts of variations in two situational factors, training sample size and number of attributes, and in two model factors, choice of Bayesian model and criteria for excluding model attributes, on the overall accuracy of Bayesian learning systems.

MEASUREMENTS

The test data were derived from myocardial infarction patients who were admitted to eight hospitals in New Orleans during 1985. The test sample consisted of 339 cases; the training samples included 100, 400, and 800 cases. APACHE II variables were used for the model attributes and patient discharge status as the outcome predicted. Attribute sets were selected in sizes of 4, 8, and 12. The authors varied the Bayesian models (proper and simple) and the attribute exclusion criteria (optimism and pessimism).

RESULTS

The simple Bayes model, which assumes conditional independence, consistently equalled or outperformed the proper (maximally dependent) Bayes model, which assumes conditional dependence, across all training sample and attribute set sizes. Not excluding model attributes was found to be preferable to using sample theory as an attribute exclusion criterion in both the simple and the proper models.

CONCLUSION

In the domain tested, the simple Bayes model with optimistic exclusion is more robust than previously assumed and increasing the number of attributes in a model had a greater relative impact on model accuracy than did increasing the number of training sample cases. Assessment of applicability of these findings to other domains will require further study. In addition, other models that are between these two extremes must be investigated. These include models that approximate proper Bayes' conditional dependence computations while requiring fewer training sample cases, attribute exclusion criteria between optimism and pessimism that improve accuracy, and ordering techniques for introducing attributes into Bayes models that optimize the information value associated with the attributes in test-sample cases.

An evaluation of factors influencing Bayesian learning systems

This paper examines the influences of situational and model factors upon the accuracy of Bayesian learning systems. In particular, it is concerned with the impact of variations in training sample size, number of attributes, choice of Bayesian model, and criteria for excluding model attributes upon the overall accuracy of the simple and proper Bayes models.

Improving the accuracy of severity indexes: exceptions to the rules

This study explores the difference between additive and non-additive indexes in measuring the severity of myocardial infarction. It shows, as an example, the fallacy of adding severity scores in a straightforward manner. An additive severity index was constructed from the judgments of seven experts. The experts also identified several exceptions to the additive index. The study used the exceptions to modify the additive index and produce a non-additive severity index. The non-additive severity index explained 36% more of the variance in the severity judgments made by five physicians and two nurses on 50 hypothetical cases than the additive index did. In addition, the non-additive index was 3% more accurate in predicting in-hospital mortality of 7,500 patients with myocardial infarction. When the study reduced the noise in the data by ignoring 1,200 rare cases in which stable estimates of mortality rate were unavailable, the prediction of the non-additive index was 13% more accurate than that of the additive index. Statistical tests showed that the differences between the additive and the non-additive indexes were significant at an alpha level below 1%. The practical implications of non-additive severity indexes are discussed. Researchers and physicians who assess the severity of myocardial infarction should systematically explore exceptions that may improve the accuracy of prediction of an additive index.

A computer aid for choosing HMOs: design of a new interface

This paper reports on a new interface design. The presentation is similar to a newspaper style, allowing a familiar format and advertisement. In addition, the design includes steps to ease data collection, features to help users influence others in the organization, and a dynamic allocation of menu lists that reflect user's knowledge and previous interest.

Rehearsing decisions may help teenagers: an evaluation of a simulation game

This paper presents a new approach to preventing adolescent pregnancy. Information alone is not sufficient to prevent teenage pregnancy. The teenagers ability to choose and remain committed to a decision also needs to be developed. Because decision making skills are best learned through practice in an environment with frequent feedback, we have developed a computer game which simulates the consequences of different sexual roles. In addition, the game is intended to increase communication about sex between teenagers and their role models (peers, teachers and/or parents). Increased communication is expected to reduce the feeling of guilt and lead to either consistent abstention from sex or consistent contraceptive use. The paper reports on the development of the computer game and the preliminary evaluation of its impact.

Appraisal of modifiable hospitalization risks

Nine prevention experts rated hospitalization risks of 64 hypothetical healthy adults between 20 and 65 years old. There was substantial agreement among seven out of the nine experts. Pairwise correlations between any two of the experts ranged between 0.66 and 0.86. Decision analytic tools were used to model the average of the experts' ratings. The panel of experts provided us with the factors used, the relationship between the factors, and the relative importance of each factor. An index based on this information was highly correlated with the judgments of seven experts. Thus, we concluded that the scoring procedure can simulate the experts' judgments. Next, the index was used in an interactive computer program to assess modifiable health risks of individuals. This program is provided along with the paper to facilitate further research on validity and impact of the program.

How to construct a subjective index

We present a method of constructing quantitative indices, which is based on the subjective opinions of a panel of experts, and discuss how a Bayesian probability model and panel opinions can be used together to produce an index. Among the advantages of the method are its face validity and ease of construction. Research shows that when expert opinions are solicited according to certain guidelines, subjective methods may be as accurate as the more objective ones. Guidelines along with a brief report of a recent application are also discussed.

An evaluation of multiple trauma severity indices created by different index development strategies

Evaluation of the effectiveness of emergency trauma care systems is complicated by the need to adjust for the widely variable case mix found in trauma patient populations. Several strategies have been advanced to construct the severity indices that can control for these population differences. This article describes a validity and reliability comparison of trauma severity indices developed under three different approaches: 1) use of a multi-attribute utility (MAU) model; 2) an actuarial approach relying on empirical data bases; and 3) an "ad hoc" approach. Seven criteria were identified to serve as standards of comparison for four different indices. The study's findings indicate that the index developed using the MAU theory approach associates most closely with physician judgments of trauma severity. When correlated with a morbidity outcome measure, the MAU-based index shows higher levels of agreement than the other indices. The index development approach based on the principles of MAU theory has several advantages and it appears to be a powerful tool in the creation of effective severity indices.