Second Stage Of Analysis Research Articles

Risk Alleles for Multiple Sclerosis Identified by a Genomewide Study

Multiple sclerosis has a clinically significant heritable component. We conducted a genomewide association study to identify alleles associated with the risk of multiple sclerosis. We used DNA microarray technology to identify common DNA sequence variants in 931 family trios (consisting of an affected child and both parents) and tested them for association. For replication, we genotyped another 609 family trios, 2322 case subjects, and 789 control subjects and used genotyping data from two external control data sets. A joint analysis of data from 12,360 subjects was performed to estimate the overall significance and effect size of associations between alleles and the risk of multiple sclerosis. A transmission disequilibrium test of 334,923 single-nucleotide polymorphisms (SNPs) in 931 family trios revealed 49 SNPs having an association with multiple sclerosis (P<1x10(-4)); of these SNPs, 38 were selected for the second-stage analysis. A comparison between the 931 case subjects from the family trios and 2431 control subjects identified an additional nonoverlapping 32 SNPs (P<0.001). An additional 40 SNPs with less stringent P values (<0.01) were also selected, for a total of 110 SNPs for the second-stage analysis. Of these SNPs, two within the interleukin-2 receptor alpha gene (IL2RA) were strongly associated with multiple sclerosis (P=2.96x10(-8)), as were a nonsynonymous SNP in the interleukin-7 receptor alpha gene (IL7RA) (P=2.94x10(-7)) and multiple SNPs in the HLA-DRA locus (P=8.94x10(-81)). Alleles of IL2RA and IL7RA and those in the HLA locus are identified as heritable risk factors for multiple sclerosis.

Pretoria medical students' perspectives on the assessable attributes of professionalism

Background Professionalism forms an important aspect of medicine's contract with society, and it is therefore important that it should be assessed and developed in medical schools. For the effective assessment of medical students' professionalism, clear objectives, or outcomes based on a clear definition of professionalism, have to be accepted by society, the faculty and the students. A Physician's Charter, ‘Medical Professionalism in the New Millennium‿, was published by the Annals of Internal Medicine in February 2002. Fifth-year medical students of the University of Pretoria were challenged to comment on the applicability of this Charter's principles and responsibilities in the South African context. The majority of the students did not fully agree with the principles and responsibilities. A following cohort of fifth-year students was requested to define professionalism and describe attributes that could be included in an assessment tool. Methods A qualitative design was employed to explore the students' perceptions. An analysis of the scripts was performed using qualitative content analysis. All the scripts were read twice and the emergent themes (attributes of professionalism) were identified by two researchers. The scripts were reviewed and coded independently to enhance reliability. Investigator triangulation, involving researchers with diverse research backgrounds, was done to validate the identified attributes. The themes were organised into categories (domains of professionalism). The authors discussed differing concepts with the other members of the Professional Attitude Development and Assessment Committee (PADAC) to arrive at a consensus. In the second stage of analysis, the identified themes were compared to the Charter. This was supported by quotations taken from the students' perceptions, as well as from the relevant literature. Tabulations were used to determine the frequencies of the different themes. Results The attributes used to describe professionalism were grouped under four main domains: attitude and personal conduct, teamwork, patient care and professional competence. These correspond, to some extent, with the Charter. Most of the students, however, stressed the humanistic attributes of professional behaviour, which are not mentioned in the Charter. These attributes include empathy, interpersonal relationships, integrity, respect, maturity and teamwork. Conclusion The students' perceptions of professionalism differed from that described in the Charter. The fact that the students value the humanistic part of professionalism could possibly be ascribed to the following factors: • Teaching of professionalism at the University of Pretoria focuses on the humanistic attributes, which form an important part of patient consultations and teamwork. • Medical students are not actively involved in the social contract that is prominent in the Charter. They have very little influence regarding social justice, access to care, distribution of resources and conflict of interest. • The Charter is considered vague and open to interpretation It is recommended that a universally acceptable charter be developed for the Medical School of Pretoria by involving faculty members to validate and accept the students' perspectives. It is also recommended that these mutually accepted attributes then be used as the assessment criteria of a professionalism assessment tool. This should also encourage faculty members to act as good role models. South African Family Practice Vol. 49 (4) 2007: pp.17

Combining GIS with fuzzy multicriteria decision-making for landfill siting in a fast-growing urban region

Landfill siting is a difficult, complex, tedious, and protracted process requiring evaluation of many different criteria. This paper presents a fuzzy multicriteria decision analysis alongside with a geospatial analysis for the selection of landfill sites. It employs a two-stage analysis synergistically to form a spatial decision support system (SDSS) for waste management in a fast-growing urban region, south Texas. The first-stage analysis makes use of the thematic maps in Geographical information system (GIS) in conjunction with environmental, biophysical, ecological, and socioeconomic variables leading to support the second-stage analysis using the fuzzy multicriteria decision-making (FMCDM) as a tool. It differs from the conventional methods of integrating GIS with MCDM for landfill selection because the approach follows two sequential steps rather than a full-integrated scheme. The case study was made for the city of Harlingen in south Texas, which is rapidly evolving into a large urban area due to its vantage position near the US–Mexico borderlands. The purpose of GIS was to perform an initial screening process to eliminate unsuitable land followed by utilization of FMCDM method to identify the most suitable site using the information provided by the regional experts with reference to five chosen criteria. Research findings show that the proposed SDSS may aid in recognizing the pros and cons of potential areas for the localization of landfill sites in any study region. Based on initial GIS screening and final FMCDM assessment, “site 1” was selected as the most suitable site for the new landfill in the suburban area of the City of Harlingen. Sensitivity analysis was performed using Monte Carlo simulation where the decision weights associated with all criteria were varied to investigate their relative impacts on the rank ordering of the potential sites in the second stage. Despite variations of the decision weights within a range of 20%, it shows that “site 1” remains its comparative advantage in the final site selection process.

Multicity Study of Air Pollution and Health Effects in Latin America

SS1-08 Abstract: The overall goal of this project is to use a common analytic framework to examine the association between exposure to outdoor air pollution and health effects in several Latin American and Caribbean (LAC) cities to obtain comparable and updated information on effects of air pollution on several age/cause groups. This will be achieved through developing a common protocol for the design and analysis of time-series data to be followed by all centers involved in the project. An important focus will be the detailed examination of the impact of such exposures in specific subgroups of the population considered most vulnerable such as infants and young children and the analysis by socioeconomic position. In the first steps, we will collect data and perform quality assurance for air pollution and mortality data from the participating cities from Brazil, Mexico, and Chile. Candidate cities so far are: Mexico City, Guadalajara, Toluca, Monterey, and Juarez City (Mexico); Sao Paulo, Rio de Janeiro, Porto Alegre, and Vitoria (Brazil); and Santiago, Antofogasta, Calama, Copiapo, and Temuco (Chile). Assessment of the availability and completeness of data in other large Latin American cities will also be pursued and those who fit the inclusion criteria will be included. Descriptive data of air pollutants and frequency distribution of mortality by cause and age groups will be carried out. First, single-city analysis will be conducted using a standardized analytical protocol (first-stage analysis). Heterogeneity will be explored and data combined using metaregression models (second-stage analysis). Overall and subgroup analyses will be conducted using generalized additive models (GAMs) in Poisson regression. We also explore the potential effect modification of socioeconomic position (SEP) and other variables on the association of air pollution and mortality using multilevel modeling (second-stage analysis). A common SEP indicator is identified that can equally represent the SEP of populations in the different cities. This project aims to contribute to the international scientific discussion on the conduct and interpretation of time-series studies of the effects of short-term exposure and to provide relevant information to decision-makers in the LAC region to support prevention and control strategies of air pollution.

Do Community-Specific Characteristics Modify Ozone Effect Estimates for Mortality?

TAA3-O-05 Introduction: Earlier work found an association between short-term exposure to ozone and increased risk of premature mortality for U.S. urban communities; however, there was heterogeneity among the community-specific estimates. Several studies have shown that factors such as socioeconomic status and air conditioning may affect the relationship between air pollution and health. This research investigates whether community-level characteristics can explain heterogeneity among community-specific effect estimates of ozone on mortality. Methods: Effect estimates for ozone's impact on mortality were generated for 98 U.S. urban communities using daily data from 1987 to 2000 for the average of the same and previous days' ozone (lag 01) and a distributed lag model accounting for ozone exposure over the past week. Adjustments were made for day of the week, temperature and dew point, and long-term trends. Bayesian hierarchical modeling was used to generate regional estimates. A second-stage analysis was used to determine whether effect estimates vary by community-specific characteristics, including socioeconomic indicators (eg, education level, household income), air conditioning use, and race. Results: Preliminary results indicate that effect estimates exhibited some regional patterns with the highest effect in the Northeast, with a 0.64% increase (95% posterior interval, 0.20%–1.07%) in mortality for a 10 ppb increase in the lag 01 ozone based on 16 communities, with the lowest effect in the Southwest at a −0.06% (−0.92% to 0.81%) increase in mortality per 10 ppb increase in lag 01 ozone based on 9 communities. Interquartile increases in the percentage of blacks, unemployment, and poverty were associated with increases in ozone/mortality effect estimates of 62.9%, 79.9%, and 62.0%, respectively (P < 0.05). Discussion and Conclusions: Findings indicate that socioeconomic community-level characteristics modify the relationship between ozone and mortality, and partially explain heterogeneity among estimates. Results show that minorities and low-income populations may face disproportionate health burdens of ozone pollution.

State Versus Private Sector Provision of Water Services in Africa: a Statistical, DEA and Stochastic Cost Frontier Analysis

Under pressure from donor agencies such as the World Bank, a number of developing countries have experimented with the privatisation of water services. This study reviews the existing econometric evidence on the effects of water privatisation in developing economies before presenting new results using statistical, data envelopment analysis (DEA) and stochastic cost frontier techniques and data from Africa. The study finds evidence of better performance in private utilities compared to state-owned utilities according to the statistical and DEA performance measures. But no statistically significant cost differences were discovered between private and public suppliers in the stochastic cost frontier analysis. The paper then considers reasons why water privatisation may prove problematic in lower-income economies, identifying the technology of water provision and nature of the product, transaction costs and possible regulatory weaknesses. However, a second-stage cost function analysis introducing a regulatory variable produced statistically insignificant results.

Two-Stage Designs in Case–Control Association Analysis

DNA pooling is a cost-effective approach for collecting information on marker allele frequency in genetic studies. It is often suggested as a screening tool to identify a subset of candidate markers from a very large number of markers to be followed up by more accurate and informative individual genotyping. In this article, we investigate several statistical properties and design issues related to this two-stage design, including the selection of the candidate markers for second-stage analysis, statistical power of this design, and the probability that truly disease-associated markers are ranked among the top after second-stage analysis. We have derived analytical results on the proportion of markers to be selected for second-stage analysis. For example, to detect disease-associated markers with an allele frequency difference of 0.05 between the cases and controls through an initial sample of 1000 cases and 1000 controls, our results suggest that when the measurement errors are small (0.005), approximately 3% of the markers should be selected. For the statistical power to identify disease-associated markers, we find that the measurement errors associated with DNA pooling have little effect on its power. This is in contrast to the one-stage pooling scheme where measurement errors may have large effect on statistical power. As for the probability that the disease-associated markers are ranked among the top in the second stage, we show that there is a high probability that at least one disease-associated marker is ranked among the top when the allele frequency differences between the cases and controls are not <0.05 for reasonably large sample sizes, even though the errors associated with DNA pooling in the first stage are not small. Therefore, the two-stage design with DNA pooling as a screening tool offers an efficient strategy in genomewide association studies, even when the measurement errors associated with DNA pooling are nonnegligible. For any disease model, we find that all the statistical results essentially depend on the population allele frequency and the allele frequency differences between the cases and controls at the disease-associated markers. The general conclusions hold whether the second stage uses an entirely independent sample or includes both the samples used in the first stage and an independent set of samples.

On the determinants of local government performance: A two-stage nonparametric approach

This article analyzes the efficiency of local governments in the Comunitat Valenciana (Spain) and their main explanatory variables. The analysis is performed in two stages. Firstly, efficiency is measured via (nonparametric) activity analysis techniques. Specifically, we consider both Data Envelopment Analysis (DEA) and Free Disposable Hull (FDH) techniques. The second stage identifies some critical determinants of efficiency, focusing on both political and fiscal policy variables. In contrast to previous two-stage research studies, our approach performs the latter attempt via nonparametric smoothing techniques, rather than econometric methods such as OLS or Tobit related techniques. Results show that efficiency scores, especially under the nonconvexity assumption (FDH), are higher for large municipalities. Thus, there is empirical evidence to suggest that resources may be better allocated by large municipalities. However, the inefficiency found is not entirely attributable to poor management, as second-stage analysis reveals both fiscal and political variables to be explicably related to municipality performance. Moreover, the explanatory variables’ impact on efficiency is robust to the chosen technique—either convex DEA or nonconvex FDH.

Technical Efficiency in the Rural Financial Sector: Evidence from Mexico

The void in the rural financial market in Mexico is more pronounced than ever given the retreat of many state programs. In response, small, heterogeneous semi-formal institutions have begun to fill the gap with financial services to rural Mexicans. This study sheds light on this largely undocumented sector by examining the technical efficiency of various institutional types. Data Envelopment Analysis is used to create a nonparametric frontier of technically efficient firms based on a sample of 350 semi-formal financial institutions. Using this framework, the heterogeneity of the sector is highlighted and most institutions are found to be highly inefficient. In a second stage of analysis, the primary determinants of efficiency are found to be institutional factors rather than client profiles.

Ethnic Origin and Language: Applying International and EU Quality Standards on the Estonian Labour Market

The purpose of the present research is to analyze compatibility of practical outcomes of Estonian legislation and policies resulting in ethnic differences on the labour market with international and European standards on equality and nondiscrimination on the basis of ethnic origin and language.The issue of discrimination in employment based on ethnic origin and language in the Estonian context has a practical meaning. The existence of discrimination as such and the understanding of what constitutes prohibited discrimination are highly controversial topics in Estonian society, politics, and law. National academic legal writings on the issue of discrimination on the basis of ethnic origin and language in employment are lacking and socio-economic inquiries are very limited.Following the Introduction, Chapter II provides an analysis of International (United Nations) and European (Council of Europe) standards in relation to protection from discrimination on the basis of language and ethnic origin in employment. The analysis is built on country-specific recommendations by monitoring bodies on Estonia, general comments, and jurisprudence of relevant international organs as well as commentaries of legal scholars.The second stage of analysis (Chapter III) then addresses the scope of protection from discrimination on the basis of ethnic origin and language in employment at the level of the EU. More specifically, Chapter III addresses in detail the compatibility of recent amendments to the Employment Contracts Act with minimum requirements of protection from discrimination as provided in the European Community Race Directive. Following analysis of the Race Directive, Chapter III discusses the role of the European Employment Strategy and the European Social Fund in promoting change on the Estonian labour market. The research concludes with a clear statement on compatibility of the situation existing on the Estonian labour market with international and EU standards on equality and non-discrimination on the basis of ethnic origin and language.

Association between single‐nucleotide polymorphisms in the SEC8L1 gene, which encodes a subunit of the exocyst complex, and rheumatoid arthritis in a Japanese population

To identify rheumatoid arthritis (RA) susceptibility genes in a Japanese population by conducting a large-scale case-control association analysis and linkage disequilibrium (LD) mapping on chromosome 7q31-34, a candidate susceptibility locus identified in a preliminary genome-wide scan in 53 Japanese families, using single-nucleotide polymorphisms (SNPs). We prepared 728 dense, evenly spaced SNPs with a minor allele frequency >0.15 in each gene locus on chromosome 7q31-34. Using these SNPs, a 2-stage case-control analysis was performed on 760 RA patients (157 men and 603 women) and 806 non-RA controls (189 men and 617 women). Haplotypes and LD mapping results were assessed based on SNP genotypes in 380 controls. Forty-eight SNPs showed allele associations (P < 0.05) in the first set of DNA samples (380 RA cases and 380 non-RA controls; first-stage analysis). For 4 of the SNPs in the SEC8L1 gene, the association was replicated (P < 0.05) in the second, independent set of DNA samples (an additional 380 RA cases and 380 non-RA controls; second-stage analysis). When data from the 2 groups were combined, the most significant allele association was observed with SNP 441, an intronic SNP of the SEC8L1 gene (P = 0.000059). The SEC8L1 SNPs with significant allele associations were all located in a single conserved LD block (block 4). Haplotype analysis revealed the disease-risk (P = 0.0015) and disease-protective (P = 0.0000062) haplotypes. Resequencing of coding exons within block 4 did not identify any nonsynonymous SNPs. Real-time quantitative polymerase chain reaction revealed that SEC8L1 was expressed ubiquitously in human tissues, including fibroblast-like synoviocytes from RA patients. Our locus-wide association and LD analyses identified intronic SNPs and haplotypes in the SEC8L1 gene that are strongly associated with RA. We propose that SEC8L1, which encodes a component of the exocyst complex, is a candidate susceptibility gene for RA in the Japanese population.

Measuring Efficiency of Macro Systems: An Application to Millennium Development Goal Attainment

At least part of the effort toward attainment of health and education Millennium Development Goals (MDGs) could come from improvements in efficiency of delivery mechanisms. Hence, it is important to know which countries—or which regions within countries—are able to attain higher MDG outcomes even after controlling for resource inputs. This information can be useful for policymakers and can enable a second-stage analysis of why is it that some regions or countries are doing better than others. This paper reviews the methodology for measuring efficiency of macro systems using the health and education sectors as examples. A simpler characterization of efficiency that is less dependent on econometric specifications is introduced. As an example, this method of measuring efficiency is applied to assess the efficiency of health system outcomes at the district (kabupaten) level in Indonesia.

Effects of childhood foster care and adoption on adulthood childbearing

Using data from the National Longitudinal Survey of Youth (NLSY), latent growth curve modeling was used to examine the trajectories of adulthood childbearing. Respondents consisted of 195 subjects who were raised by an adoptive family, 78 who raised by a foster family for 4 months or more, 200 who were raised by stepparents for 4 months or more, and 200 who were raised by both biological parents until age 18. In the first-stage analyses, foster-parented respondents showed significantly greater linear increases in the number of childbirths per year and significant deceleration in childbirth, as compared with biological- and adoptive-parented respondents. In the second-stage analyses, there were significant differences between the groups in the effects of education on the initial level of childbirth, with step- and adoptive-parented respondents having significantly more negative relationships for education than did biological-parented respondents. Likewise, adoptive-parented respondents showed a significantly more positive relationship between persistent poverty status and the initial level of childbirth, as compared with biological- and foster-parented respondents.

Cumulative environmental impacts and integrated coastal management: the case of Xiamen, China

This paper examines the assessment of cumulative environmental impacts and the implementation of integrated coastal management within the harbour of Xiamen, China, an urban region in which the coastal zone is under increasing pressure as a result of very rapid economic growth. The first stage of analysis incorporates components of a cumulative effects assessment, including (a) identification of sources of environmental impacts, notably industrial expansion, port development, shipping, waste disposal, aquaculture and coastal construction, (b) selection of a set of valued ecosystem components, focusing on circulation and siltation, water quality, sediment, the benthic community, and mangrove forests, and (c) use of a set of key indicators to examine cumulative impacts arising from the aggregate of human activities. In the second stage of analysis, the paper describes and assesses the development of an institutional framework for integrated coastal management in Xiamen, one that combines policy and planning (including legislative and enforcement mechanisms) with scientific and monitoring mechanisms (including an innovative ‘marine functional zoning’ system). The paper concludes that the integrated coastal management framework in Xiamen has met all relevant requirements for ‘integration’ as laid out in the literature, and has explicitly incorporated consideration of cumulative impacts within its management and monitoring processes.

Cumulative environmental impacts and integrated coastal management: the case of Xiamen, China

This paper examines the assessment of cumulative environmental impacts and the implementation of integrated coastal management within the harbour of Xiamen, China, an urban region in which the coastal zone is under increasing pressure as a result of very rapid economic growth. The first stage of analysis incorporates components of a cumulative effects assessment, including (a) identification of sources of environmental impacts, notably industrial expansion, port development, shipping, waste disposal, aquaculture and coastal construction, (b) selection of a set of valued ecosystem components, focusing on circulation and siltation, water quality, sediment, the benthic community, and mangrove forests, and (c) use of a set of key indicators to examine cumulative impacts arising from the aggregate of human activities. In the second stage of analysis, the paper describes and assesses the development of an institutional framework for integrated coastal management in Xiamen, one that combines policy and planning (including legislative and enforcement mechanisms) with scientific and monitoring mechanisms (including an innovative ‘marine functional zoning’ system). The paper concludes that the integrated coastal management framework in Xiamen has met all relevant requirements for ‘integration’ as laid out in the literature, and has explicitly incorporated consideration of cumulative impacts within its management and monitoring processes.

Low Frequency Streamflow Regimes over the Central United States: 1939–1998

Intra- to multi-decadal (6-30 year) streamflow regimes over the central United States during the late 20th century are identified and investigated here through a two-stage analysis. The first stage ranks mean annual flow rates during 1939-1998, calculates Mann-Whitney U statistics from samples of those rankings over running time windows, and then tests those U statistics for significance. This analysis of the records of 42 Hydro-Climatic Data Network stations over the Great Plains and Midwest reveals consistent patterns of highly ranked annual streamflow during time windows at the end of the century, with most beginning during the late 1960s and early 1970s and ending in either 1997 or 1998. Many of these stations are located in a critical agricultural region known as the Corn Belt. The second stage of analysis compares both the duration of abnormal flow periods and the frequency of hydrological surplus and drought conditions during the high flow years indicated by the first stage, relative to the remaining years of 1939-1998. Among gage stations in the climatologically drier western Corn Belt during the 1980s and 1990s there is a clear tendency toward extended periods of above normal flow, which results in more than a doubling in the average annual frequency of hydrological surplus days relative to previous years. These stations also show more than a 50% reduction in the average annual frequency of hydrological drought days relative to previous years. Similar but less pronounced shifts in hydrological regime are evident in the central and eastern Corn Belt, and in the Mississippi River at Vicksburg during 1973-1998. These results indicate that many areas of the central United States have shifted toward a climate regime of relative hydrological surplus during the closing decades of the 20th century.

Fitting growth models to length frequency data

Abstract A novel two-stage procedure for fitting growth curves to length frequency data collected from commercial fisheries is described. The method is suitable for species in which cohorts are spawned over a limited time period, and samples of length frequency data are collected regularly (e.g. in weekly, fortnightly, or monthly time intervals) over an extended time period. In the first stage of analysis, Gaussian mixtures are fitted separately to the data for each time interval, and summary statistics (component means and standard errors) are extracted. In the second stage, parametric growth models, such as the von Bertalanffy seasonal growth curve, are fitted to the summary data. The error structure in this second stage of analysis incorporates random between-year effects, random within-year age-group effects, random within-year time-interval effects, random within-year age-group and time-interval interactions, and sampling errors. This complex error structure incorporating unbalanced crossed and nested random effects acknowledges that commercial fishing is not an exercise in random sampling, and allows for the inevitable additional sources of random variation in such an enterprise. The method is applied to South Australian southern bluefin tuna length frequency data collected from 1964 to 1989, and leads to the conclusion that juvenile tuna grew faster in the 1980s than in the 1960s, with the 1970s being a decade of highly variable growth.

A Consensus on Second-Stage Analyses in Ecological Inference Models

Since Herron and Shotts (2003a; hereinafter HS), Adolph and King (2003; hereinafter AK), and Herron and Shotts (2003b; hereinafter HS2), the four of us have iterated many more times, learned a great deal, and arrived at a consensus on this issue. This paper describes our joint recommendations for how to run second-stage ecological regressions, and provides detailed analyses to back up our claims.

Relationship of clinical factors to the use of Cox-2 selective NSAIDs within an arthritis population in a large HMO.

To investigate the degree to which physicians use clinical factors to focus use of Cox-2 selective NSAIDs within an arthritis population. Diagnostic codes in the medical records of a large group-model HMO in northern California with approximately 3 million members were examined to identify patients with either rheumatoid arthritis (RA) or non-RA (osteoarthritis or degenerative joint disease). RA and non-RA patients were stratified in deciles of relative risk for gastrointestinal (GI) complications according to patient characteristics identified on the Standardized Calculator of Risk for Events (SCORE) that were associated with use of Cox-2 selective NSAIDs. (The SCORE tool stratifies patients by risk of serious GI complications using patient characteristics that are assigned points during an office visit, including age, health status, diagnosis of rheumatoid arthritis, corticosteroid use, and history of GI ulcer or bleed.) The second stage of analysis examined the percentage of arthritis patients in each SCORE-risk decile who received a Cox-2 selective NSAID, lower-risk NSAID, or traditional NSAID during calendar year 1999. The study population consisted of 144,360 members with an arthritis diagnosis, approximately 4.8% of members in this HMO. The mean age was 62.8 years (SD = 14.1), 61% were female, 10,449 (7%) had rheumatoid arthritis (RA), and 133,911 (93%) had non-rheumatoid arthritis. A diagnosis of RA was the most significant predictor of Cox-2 NSAID use (OR=2.4; 95% CI=1.6-3.5), followed by a history of GI problems (OR=1.5; 95% CI=1.4- 1.6). Female gender, chronic steroid use, and age each increased the odds of receiving a Cox-2 selective NSAID by about 35% (P<0.001 for all). Approximately 8.3% of patients in the highest decile of risk and 1.5% of patients in the lowest decile of risk received a Cox-2 selective NSAID. Clinical characteristics of patients identified on the SCORE (GI-risk) tool were strongly associated with use of Cox-2-selective NSAIDs in this HMO. A 5.5-fold difference in utilization of Cox-2 selective NSAIDs was found among patients determined to be in the highest-risk decile versus patients in the lowest-risk decile. Future research should investigate how nonclinical factors play a role in the treatment decisions made by physicians.

COST EFFICIENCY IN AUSTRALIAN GENERAL INSURERS: A NON-PARAMETRIC APPROACH

Data envelopment analysis is used to calculate pure technical, scale, allocative and cost efficiency indices for a sample of forty-six Australian general insurers. The inputs used are labour, physical capital (in the form of both information technology and plant and equipment), and financial capital. The outputs are net premium revenues for housing-related, transport-related, indemnity-related, mortgage-related, and other insurance, along with invested assets. The results indicate that the major source of overall cost inefficiency would appear to be allocative inefficiency rather than technical inefficiency and that the largest 20% of insurers are significantly more efficient than the remaining firms. A second-stage analysis uses limited dependent variable regression techniques to relate efficiency scores to firm-specific information. Cost efficiency appears to be closely related to asset size but not to stock exchange listing or product diversification or specialization.