MEDICAL CARE UTILIZATION IN TIME AMONG ELDERLY MEDICARE ENROLLEES IN AN HMO

Investigation into the medical care utilization of elderly Medicare enrollees in an HMO (Kaiser - Portland, Oregon): The specific research topics are: (1) The utilization of medical care by selected determinants such as: place of service, type of service, type of appointment, physician status, physician specialty and number of associated morbidities. (2) The attended prevalence of 3 chronic diseases: hypertension, diabetes and arthritis in addition to pneumonias as an example of acute diseases. The selection of these examples was based on their importance in morbidity/or mortality results among the elderly. The share of these diseases in outpatient and inpatient contacts was examined as an example of the relation between morbidity and medical care utilization. (3) The tendency of individual utilization patterns to persist in subsequent time periods. The concept of contagion or proneness was studied in a period of 2 years. Fitting the negative binomial and the Poisson distributions was applied to the utilization in the 2nd year conditional on that in the 1st year as regards outpatient and inpatient contacts.^ The present research is based on a longitudinal study of 20% random sample of elderly Medicare enrollees. The sample size is 1683 individuals during the period from August 1980-December 1982.^ The results of the research were: (1) The distribution of contacts by selected determinants did not reveal a consistent pattern between sexes and age groups. (2) The attended prevalence of hypertension and arthritis showed excess prevalence among females. For diabetes and pneumonias no female excess was noticed. Consistent increased prevalence with increasing age was not detected.^ There were important findings pertaining to the relatively big share of the combined 3 chronic diseases in utilization. They accounted for 20% of male outpatient contacts vs. 25% of female outpatients. For inpatient contacts, they consumed 20% in case of males vs. 24% in case of females. (3) Finding that the negative binomial distribution fit the utilization experience supported the research hypothesis concerning the concept of contagion in utilization. This important finding can be helpful in estimating liability functions needed for forecasting future utilization according to previous experience. Such information has its relevance to organization, administration and planning for medical care in general. (Abstract shortened with permission of author.) ^

TIME SERIES ANALYSIS AS INPUT FOR PREDICTIVE MODELING: PREDICTING CARDIAC ARREST IN A PEDIATRIC INTENSIVE CARE UNIT

The first manuscript, entitled "Time-Series Analysis as Input for Clinical Predictive Modeling: Modeling Cardiac Arrest in a Pediatric ICU" lays out the theoretical background for the project. There are several core concepts presented in this paper. First, traditional multivariate models (where each variable is represented by only one value) provide single point-in-time snapshots of patient status: they are incapable of characterizing deterioration. Since deterioration is consistently identified as a precursor to cardiac arrests, we maintain that the traditional multivariate paradigm is insufficient for predicting arrests. We identify time series analysis as a method capable of characterizing deterioration in an objective, mathematical fashion, and describe how to build a general foundation for predictive modeling using time series analysis results as latent variables. Building a solid foundation for any given modeling task involves addressing a number of issues during the design phase. These include selecting the proper candidate features on which to base the model, and selecting the most appropriate tool to measure them. We also identified several unique design issues that are introduced when time series data elements are added to the set of candidate features. One such issue is in defining the duration and resolution of time series elements required to sufficiently characterize the time series phenomena being considered as candidate features for the predictive model. Once the duration and resolution are established, there must also be explicit mathematical or statistical operations that produce the time series analysis result to be used as a latent candidate feature. In synthesizing the comprehensive framework for building a predictive model based on time series data elements, we identified at least four classes of data that can be used in the model design. The first two classes are shared with traditional multivariate models: multivariate data and clinical latent features. Multivariate data is represented by the standard one value per variable paradigm and is widely employed in a host of clinical models and tools. These are often represented by a number present in a given cell of a table. Clinical latent features derived, rather than directly measured, data elements that more accurately represent a particular clinical phenomenon than any of the directly measured data elements in isolation. The second two classes are unique to the time series data elements. The first of these is the raw data elements. These are represented by multiple values per variable, and constitute the measured observations that are typically available to end users when they review time series data. These are often represented as dots on a graph. The final class of data results from performing time series analysis. This class of data represents the fundamental concept on which our hypothesis is based. The specific statistical or mathematical operations are up to the modeler to determine, but we generally recommend that a variety of analyses be performed in order to maximize the likelihood that a representation of the time series data elements is produced that is able to distinguish between two or more classes of outcomes. The second manuscript, entitled "Building Clinical Prediction Models Using Time Series Data: Modeling Cardiac Arrest in a Pediatric ICU" provides a detailed description, start to finish, of the methods required to prepare the data, build, and validate a predictive model that uses the time series data elements determined in the first paper. One of the fundamental tenets of the second paper is that manual implementations of time series based models are unfeasible due to the relatively large number of data elements and the complexity of preprocessing that must occur before data can be presented to the model. Each of the seventeen steps is analyzed from the perspective of how it may be automated, when necessary. We identify the general objectives and available strategies of each of the steps, and we present our rationale for choosing a specific strategy for each step in the case of predicting cardiac arrest in a pediatric intensive care unit. Another issue brought to light by the second paper is that the individual steps required to use time series data for predictive modeling are more numerous and more complex than those used for modeling with traditional multivariate data. Even after complexities attributable to the design phase (addressed in our first paper) have been accounted for, the management and manipulation of the time series elements (the preprocessing steps in particular) are issues that are not present in a traditional multivariate modeling paradigm. In our methods, we present the issues that arise from the time series data elements: defining a reference time; imputing and reducing time series data in order to conform to a predefined structure that was specified during the design phase; and normalizing variable families rather than individual variable instances. The final manuscript, entitled: "Using Time-Series Analysis to Predict Cardiac Arrest in a Pediatric Intensive Care Unit" presents the results that were obtained by applying the theoretical construct and its associated methods (detailed in the first two papers) to the case of cardiac arrest prediction in a pediatric intensive care unit. Our results showed that utilizing the trend analysis from the time series data elements reduced the number of classification errors by 73%. The area under the Receiver Operating Characteristic curve increased from a baseline of 87% to 98% by including the trend analysis. In addition to the performance measures, we were also able to demonstrate that adding raw time series data elements without their associated trend analyses improved classification accuracy as compared to the baseline multivariate model, but diminished classification accuracy as compared to when just the trend analysis features were added (ie, without adding the raw time series data elements). We believe this phenomenon was largely attributable to overfitting, which is known to increase as the ratio of candidate features to class examples rises. Furthermore, although we employed several feature reduction strategies to counteract the overfitting problem, they failed to improve the performance beyond that which was achieved by exclusion of the raw time series elements. Finally, our data demonstrated that pulse oximetry and systolic blood pressure readings tend to start diminishing about 10-20 minutes before an arrest, whereas heart rates tend to diminish rapidly less than 5 minutes before an arrest.

The effect of triage team on time-in-department in a pediatric emergency department

Objective. To determine whether the use of a triage team would reduce the average time-in-department in a pediatric emergency department by 25%.^ Methods. A triage team consisting of a physician, a nurse, and a nurse's assistant initiated work-ups and saw patients who required minimal lab work-up and were likely to be discharged. Study days were randomized. Our inclusion criteria were all children seen in the emergency center between 6p and 2a Monday-Friday. Our exclusion criteria included resuscitations, inpatient-inpatient transfers, left without being seen, leaving against medical advice, any child seen outside of 6p-2am Monday-Friday and on the weekends. A Pearson-Chi square was used for comparison of the two groups for heterogeneity. For the time-in-department analysis, we performed a 2 sided t-test with a set alpha of 0.05 using Mann Whitney U looking for differences in time-in-department based on acuity level, disposition, and acuity level stratified by disposition. ^ Results. Among urgent and non-urgent patients, we found a statistically significant decrease in time-in-department in a pediatric emergency department. Urgent patients had a time-in-department that was 51 minutes shorter than patients seen on non-triage team days (p=0.007), which represents a 14% decrease in time-in-department. Non-urgent patients seen on triage team days had a time-in-department that was 24 minutes shorter than non-urgent patients seen on non-triage team days (p=0.009). From the disposition perspective, discharged patients seen on triage team days had a shorter time-in-department of 28 minutes as compared to those seen on non-triage team days (p=0.012). ^ Conclusion. Overall, there was a trend towards decreased time-in-department of 19 minutes (5.9% decrease) during triage team times. There was a statistically significant decrease in the time-in-department among urgent patients of 51 minutes (13.9% decrease) and among discharged patients of 28 minutes (8.4% decrease). Urgent care patients make up nearly a quarter of the emergency patient population and decreasing their time-in-department would likely make a significant impact on overall emergency flow.^

The economic cost of type 2 diabetes attributable to physical inactivity in the United States and Texas in 2007

Type 2 diabetes has grown to epidemic proportions in the U.S., and its prevalence has been steadily increasing in Texas. The physical activity levels in the population have remained low despite it being one of the primary preventive strategies for type 2 diabetes. The objectives of this study were to estimate the direct medical costs of type 2 diabetes attributable to not meeting physical activity Guidelines and to physical inactivity in the U.S. and Texas in 2007. This was a cross sectional study that used physical activity prevalence data from the 2007 Behavioral Risk Factor Surveillance System (BRFSS) to estimate the population attributable risk percentage (PAR%) for type 2 diabetes. These data were combined with the prevalence and cost data of type 2 diabetes to estimate the cost of type 2 diabetes attributable to not meeting Guidelines and to inactivity in the U.S. and Texas in 2007.^ The cost of type 2 diabetes in the U.S. in 2007, attributable to not meeting physical activity Guidelines was estimated to be $13.29 billion, and that attributable to physical inactivity (no leisure time physical activity) was estimated to be $3.32 billion. Depending on various assumptions, these estimates ranged from $7.61 billion to $41.48 billion for not meeting Guidelines, and $1.90 billion to $13.20 billion for physical inactivity in the U.S. in 2007. The cost of type 2 diabetes in Texas in 2007 attributable to not meeting physical activity Guidelines was estimated to be $1.15 billion, and that attributable to physical inactivity (no leisure time physical activity) was estimated to be $325 million. Depending on various assumptions, these estimates ranged from $800 million to $3.47 billion for not meeting Guidelines, and $186 million to $1.28 billion for physical inactivity in Texas in 2007. These results illustrate how much money could be saved annually just in terms of type 2 diabetes cost in the U.S. and Texas, if the entire adult population was active enough to meet physical activity Guidelines. Physical activity promotion, particularly at the environmental and policy level should be a priority in the population. ^

Idiographic Self-Monitoring Instruments to Empower Client Participation and Evaluate Outcome in Intensive Family Preservation Services

Intensive Family Preservation Services seek to reflect the values of focusing on client strengths and viewing clients as colleagues. To promote those values, Intensive Family Preservation Programs should include a systematic form of client self monitoring in their packages of outcome measures. This paper presents a model of idiographic self-monitoring used in time series, single system research design developed for Family Partners, a family preservation program of the School for Contemporary Education in Annandale, Virginia. The evaluation model provides a means of empowering client families to utilize their strengths and promote their status as colleague in determining their own goals, participating in the change process, and measuring their own progress.

EMPLOYMENT AND EXPOSURES IN THE PETROLEUM REFINING AND PETROCHEMICAL INDUSTRIES AND THE RISK OF LUNG CANCER (EPIDEMIOLOGY, ASSESSMENT)

This dissertation addresses the risk of lung cancer associated with occupational exposures in the petroleum refining and petrochemical industries. Earlier epidemiologic studies of this association did not adjust for cigarette smoking or have specific exposure classifications. The Texas EXposure Assessment System (TEXAS) was developed with data from a population-based, case-comparison study conducted in five southeast Texas counties between 1976 and 1980. The Texas Exposure Assessment System uses job and process categories developed by the American Petroleum Institute, as well as time-oriented variables to identify high risk groups.^ An industry-wide, increased risk for lung cancer was associated with jobs having low-level hydrocarbon exposure that also include other occupational inhalation exposures (OR = 2.0--adjusted for smoking and latency effects). The prohibition of cigarette smoking for jobs with high-level hydrocarbon exposure might explain part of the increased risk for jobs with low-level hydrocarbon exposures. Asbestos exposure comprises a large part of the risk associated with jobs having other inhalation exposures besides hydrocarbons. Workers in petroleum refineries were not shown to have an increased, occupational risk for lung cancer. The increased risk for lung cancer among petrochemical workers (OR = 3.1--smoking and latency adjusted) is associated with all jobs that involve other inhalation exposure characteristics (not only low-level hydrocarbon exposures). Findings for contract workers and workers exposed to specific chemicals were inconclusive although some hypotheses for future research were identified.^ The study results demonstrate that the predominant risk for lung cancer is due to cigarette smoking (OR = 9.8). Cigarette smoking accounts for 86.5% of the incident lung cancer cases within the study area. Workers in the petroleum industry smoke significantly less than persons employed in other industries (p << 0.001). Only 2.2% of the incident lung cancer cases may be attributed to petroleum industry jobs; lifestyle factors (e.g., nutrition) may be associated with the balance of the cases. The results from this study also suggest possible high risk time periods (OR = 3.9--smoking and occupation adjusted). Artifacts in time-oriented findings may result because of the latency interval for lung cancer, secular peaks in age-, sex-specific incidence rates, or periods of hazardous exposures in the petroleum industry. ^

Cost analysis of opt-in consent procedures for the Texas statewide immunization information system, ImmTrac

Background. Childhood immunization programs have dramatically reduced the morbidity and mortality associated with vaccine-preventable diseases. Proper documentation of immunizations that have been administered is essential to prevent duplicate immunization of children. To help improve documentation, immunization information systems (IISs) have been developed. IISs are comprehensive repositories of immunization information for children residing within a geographic region. The two models for participation in an IIS are voluntary inclusion, or "opt-in," and voluntary exclusion, or "opt-out." In an opt-in system, consent must be obtained for each participant, conversely, in an opt-out IIS, all children are included unless procedures to exclude the child are completed. Consent requirements for participation vary by state; the Texas IIS, ImmTrac, is an opt-in system.^ Objectives. The specific objectives are to: (1) Evaluate the variance among the time and costs associated with collecting ImmTrac consent at public and private birthing hospitals in the Greater Houston area; (2) Estimate the total costs associated with collecting ImmTrac consent at selected public and private birthing hospitals in the Greater Houston area; (3) Describe the alternative opt-out process for collecting ImmTrac consent at birth and discuss the associated cost savings relative to an opt-in system.^ Methods. Existing time-motion studies (n=281) conducted between October, 2006 and August, 2007 at 8 birthing hospitals in the Greater Houston area were used to assess the time and costs associated with obtaining ImmTrac consent at birth. All data analyzed are deidentified and contain no personal information. Variations in time and costs at each location were assessed and total costs per child and costs per year were estimated. The cost of an alternative opt-out system was also calculated.^ Results. The median time required by birth registrars to complete consent procedures varied from 72-285 seconds per child. The annual costs associated with obtaining consent for 388,285 newborns in ImmTrac's opt-in consent process were estimated at $702,000. The corresponding costs of the proposed opt-out system were estimated to total $194,000 per year. ^ Conclusions. Substantial variation in the time and costs associated with completion of ImmTrac consent procedures were observed. Changing to an opt-out system for participation could represent significant cost savings. ^

Factors associated with the retention of HIV-infected children in care in Botswana

Lost to follow up (LTFU) in the care and treatment of HIV/AIDS represents a particularly problematic aspect when evaluating the success of treatment programs. Identifying modifiable factors that lead to LTFU would be important if we are to design effective retention interventions. The purpose of this study was to identify the challenges faced by children seeking care and treatment at a large HIV Clinic in Botswana. In order to identify those factors, we used mixed methods from different sources of information available at the Baylor Clinic. The first method involved a case-control study through which we interviewed a select representation of children 1-18 years who, at some point in time, have attended clinic at Baylor Clinic in Gaborone, Botswana. We document this in detail using the first journal article. We defined LTFU as patients who had not attended clinic for more than 6 months at the onset of the study; the comparison group was recruited from among those who have attended clinic at any point in the 6 months leading to the start of study. Factors were compared between the cases and controls. The second methodology involved conducting in-depth interviews with health providers to elicit their opinions and experiences dealing with patients at the at the Baylor clinic in general and the LTFU patients in particular. We document this methodology and its findings in the second journal article. ^ We found that most patients that are LTFU failed to engage with the clinic. Most of the LTFU made only one visit to the clinic (47.66%) as compared to less than 1% in the control group (P<0.01, 2-tailed Fisher's exact test). Among the interviewed patients, psychosocial factors such as stigma, religious beliefs, child rebellion and disclosure of HIV status concerns were characteristic of the LTFU population, but psychosocial issues were not cited among the comparison group. We also found that these psychosocial aspects of the patients point towards a bigger problem of mental health that needs to be addressed. Socioeconomic factors such as lack of transport, school-related activities and forgetting check-up dates were cited predominantly by the controls than cases. ^ From these findings, there is need to target interventions towards engaging pediatric patients at their initial clinic visit. Such interventions would focus on psychosocial support, as well as involving faith-based organizations in planning joint responses.^

Investigation of the mechanism of increased melanoma incidence in UV -irradiated mouse skin

The availability of transplantable, syngeneic murine melanomas made it possible to study the potential effects of UV radiation on the growth and progression of melanomas in an animal model. The purpose of my study was to determine how UV-irradiation increases the incidence of melanoma out-growth, when syngeneic melanoma cells are transplanted into a UV-irradiated site. Short term intermittent UVB exposure produces a transitory change in the mice which allows the increased outgrowth of melanoma cells injected into the UV-irradiated site. One possible mechanism is an immunomodulatory effect of UVR on the host. An alternative mechanism to account for the increased tumor incidence in the UV-irradiated site, is the release of inflammatory mediators from UV-irradiated epidermal cells. A third possibility is that UVR could induce the production and/or release of melanoma-specific growth factors resulting in increased melanoma outgrowth.^ My first step in distinguishing among these different possible mechanisms was to characterize further the conditions leading to increased development of melanoma cells in UV-irradiated mouse skin. Next, I attempted to determine which of the 3 proposed mechanisms was most likely. To do this, I defined the specificity of the effect by examining the growth of additional C3H tumorigenic cell lines in UV-irradiated skin. Second, I determined the immunogenicity of these tumor cell lines. The tumor cell lines exhibiting increased tumor incidence are restricted to those tumor cell lines which are immunogenic in normal C3H mice. Third, I determined the effect of UVR on melanoma development did not occur in immunosuppressed mice.^ Because of results from these three lines of investigation suggested that the effect was immunologically mediated, I then investigated whether specific immune reactions were affected by local UV irradiation. To accomplish this, I investigated the effect of UVR on cutaneous immune cells and on induction of contact hypersensitivity (CHS), and I also determined the effect of UVR on the development and the expression of systemic immunity against the melanoma cells. There is no clear cut relationship between the number of Langerhans or Thy1+ cells and the UV effect on tumor incidence. Furthermore, there was no suppression of CHS in the UV-irradiated mice. While the development of systemic immunity is significantly reduced, it appears to be sufficient to provide in vivo immunity to tumor challenge. However the elicitation of tumor immunity in immunized mice can be abrogated if tumor challenge occurs in the site of UV irradiation. This investigation provides new information on an effect of UVR on the elicitation of tumor immunity. Furthermore, it indicates that UV radiation can play a role in the development of melanoma other than just in the transformation of melanocytes. ^