The study’s treatment population of patients with diabetes was selected from the six pilot institutions implementing the CCM program, having physical yards of varying security and medication requirement levels. The control population was chosen from the remaining non-pilot prisons also having a complex mix of security and clinical needs. Individuals included in this research were identified as having diabetes and were at least 18 years of age . Sample Selection and Sample Size Eligible inmate-patients were identified via review of the universal health record . Inmate-patients were identified for inclusion into the case group by presence of their record in the diabetes registry, and their residence in one of the CCM pilot sites. The registry was populated with all known diabetic inmates in January 2009 and was continuously updated as new diabetic patients presented for care. The criterion for submission to the registry database was an encounter or admission specifically related to diabetes, as indicated by the presence of a diabetes-specific International Classification of Diseases -9-CM code—codes 250 ; 250.0 ; 250.1 ; 250.3 ; 250.4 ; 250.5 ; 250.6 ; 250.7 ; 250.8 ; 250.9 ; and 362.0 .The registry identified a total of 72 diabetic inmates in the pilot CCM facilities. Each patient case was randomly matched with two controls on the basis of stratifications of age, sex, prison security level, disease condition , and length of incarceration during the time of the study. Control patients were also selected from the registry but were restricted to non-CCM implemented facilities. These 144 control inmate patients represented the highest number of controls that could be matched from non-CCM yards.

As a result, the study design yielded a 1:2 matching. Matching was performed in SQL Server 2003,vertical grow system explicitly grouping identified inmate-patients from the control population to those in the case cohort based on matches on the aforementioned variables. The variables chosen for matching were selected due to their ability to compare individuals within groups based on their prison behavior related to health care utilization. Certain combinations of age, security level, and sex were known to create varying conditions for the seeking or utilization of healthcare beyond the actual need for health care services . To minimize differences in this type of health care utilization behavior, matching was performed on these variables. In order to minimize differences between comparison groups in the type of treatment sought, inmate-patients were matched on the disease state under consideration . Length of incarceration was used as a matching characteristic as another control for behavioral considerations, as inmates may act differently if a known transfer or parole action is pending. The matching criteria for age were based on three levels of stratification: 18–25, 26–45, and 46+. Length of incarceration was employed to match prisoners on the basis of their lengths of stay within the environment. There were four groups for this matching criterion: 0–6 months, 6–12 months, 12–18 months, and 18–24 months. PNo sampling method was required because the full target population was available for review in a manageable and cost-effective manner. As noted above, the total population of inmate-patients under treatment of the intervention method was 72. Logistic regression was the primary method of analysis used on the data that were gathered to address the four research questions.

LeBlanc and Fitzgerald suggest a minimum of 30 participants per predictor variable in the analysis. A rule of thumb states that there should be at least 10 yes and 10 no, and preferably 20, for each predictor variable . With two predictor variables in each analysis, the generally accepted principle suggests 60 to 80 participants to detect significance for the logistic regressions. The data utilized in this portion of the analysis representing 100% of the population available for study met this logistic-analysis standard.Data were analyzed using SPSS v.19, Excel 2010, and SQL Server 2003. Descriptive statistics were initially calculated to describe the sample demographics and the research variables used in the analyses. Logistic regression was then used to assess the association of independent variables of CCM intervention and age to dependent variables of emergency department visits, hospital visits, specialist visits, and length of stay. The regression models were used to explore the four research questions concerning the relationship between the CCM intervention and age of inmate-patient on health care outcome variables. Age—as well as its associated characteristics such as lifestyle patterns, which associate health-related behaviors to age groupings—is commonly utilized in the health literature as having predictive validity in the measurement of interventions . Within the correctional environment, this viewpoint of the literature was generally accepted; however, given the state of egregious neglect that brought the receivership into effect, health interventions would stand on their own merit without regard to age differences. Additionally, because health-related behaviors considered within the literature were not generalizable to daily-life conditions prevalent within the custodial environment, the interventions brought under the receivership would not be subject to the age effect.

Testing of the model in relation to health outcomes then first commenced with the addition of age as a control variable, with CCM as the independent variable, in a multivariate logit model. Multivariate Analysis Logistic regression is appropriate when the dependent variable is dichotomous, meaning there are two possible outcomes for the dependent variable, allowing one to directly estimate the probability of an event’s occurrence . The logistic regression can be used when the predictor variables are continuous, discrete, or a combination of both. This analysis permits the evaluation of the odds of membership in one of the two groups based on the combination of predictor-variable values. The overall model significance for the logistic regression was examined by the effect of the independent variable, presented with a χ2 coefficient. The Nagelkerke R2 was examined to assess the percent of variance accounted for. Predicted probabilities of an event occurring were determined by Exp . Predictors with negative beta weights will be interpreted as the event coded as 1 to not happen. Logistic regressions by design overcome many of the restrictive assumptions of multiple linear regressions. For example, the assumptions of linearity, normality and equal variances are not made. The major assumption is that the outcome variable must be dichotomous. There should be no outliers in the data. A larger sample is recommended fitting with the maximum-likelihood method; using discrete variables requires that there are enough responses in each category. Given the data used in this study, the logistic regression approach is therefore appropriate. To mirror the presentation style found in the subsequent results section, the research questions are broken out into their discrete elements below.The goal of the analysis was to understand the association between exposure to care under the new delivery model of care and subsequent need for expensive modalities of care . With respect to the sampling technique used,drainage planter pot it is believed that no differential selection bias was introduced for selection of the case group because the entire population of diabetics constituted the case group. All known diabetics were a part of the study because they were a part of the registry. Individuals could not opt in or out of the CCM model of care, and when any inmate presented for care for the disease, he or she was entered into the registry. Bias due to follow-up was minimized in the matching design by grouping inmate-patients on the basis of length of incarceration. Bias, however, may have been introduced uncontrollably in the control group for inmate-patients who were misdiagnosed with diabetes as their presentation case issue.Length of stay within the inpatient setting was examined to serve as a proxy for cost, as well as to signify severity of condition at the time of admission.

It was presumed that existing organizational procedures in other areas related to policies on length of stay and discharge activities would control for confounding length-of-stay factors. The average length of stay for an inmate presenting with diabetes as the primary diagnosis was 4.8 days in 2009. An internal target of 6 days or under for lengths of stay of a diabetic admit was commonly accepted within this environment. As such, length of stay was analyzed as a binary outcome . The outcomes of care looked at were: ≥1 hospital visits, ≥1 specialist visits, ≥1 emergency department visits , and length of stay. Due to a lack of integrated data sources, some confounding effects such as treatment under other pilot programs could not be accounted for or analyzed at the time of analysis. Diabetes-related health outcomes found to have an association to the CCM program were those specific to the inpatient setting. After being selected into the study cohort, case patients made fewer visits to the hospital unit and additionally had shorter lengths of stay once admitted than did controls. This suggests that treatment under the new care model was successful in its goals of reducing the need for subsequent high cost care. Table 7 shows the odds ratios derived from the logistical regression output. The OR for hospital visits was rather low at .320 , suggesting that if a hospital visit outcome were to increase by 1, the odds ratio for an inmate-patient to be in CCM would decrease by a factor of .320. Put another way, the likelihood for a person to have an inpatient hospital episode for diabetes is smaller if enrolled in the CCM program than if not. This translates directly to the cost-avoidance feature of the program. As longer inpatient stays relate to higher costs per episode of care, the statistical results concerning length of stay are also quite pertinent to this discussion. The multivariate analysis reveals a statistically significant causal relationship between length of stay and the CCM program, controlling for age of the inmate. The OR for length of stay at an inpatient setting is .191 . No causal relationship between the use of specialists or the emergency department and the CCM program were found . Given that, the only cost avoidance and health outcomes variables shown to have an association to the new program are due to the inpatient care setting. Emergency room admissions could have been problematic to address in terms of conclusions should the findings have indicated a causal relationship between the program and visits. Within the correctional environment, the emergency room can be used as by inmates as means of escaping the standard custodial confines. As a result, it is plausible that inmate-patients would claim the need for emergency diabetic treatment for reasons well beyond actual chronic care indication. From a resource perspective, this is especially relevant, as the inpatient facilities are often over-utilized and understaffed. Should the results be replicable outside of the pilot sites, costs related to over staffing can be reduced, and the quality of care to those in need of a hospital bed possibly can be increased.Surveying the parade as it passed by the reviewing stand and through the West Gate into New Chinatown, Peter SooHoo must have felt a sense of pride. That day, June 25, 1938 was the official opening of the newest Chinatown in Los Angeles, and SooHoo was serving as the Master of Ceremonies. A third-generation Californian, SooHoo was officially the English-language secretary for the Los Angeles Chinatown Project Association, but his role in the construction of New Chinatown had been much more significant than his title implied.1 Four years earlier, city leaders announced that Old Chinatown would be torn down and its residents displaced to build the new Union Station. Since then, Peter SooHoo had acted as a liaison between the railways and the Chinese American business owners and residents set to be displaced by this new train depot. As crews dismantled Old Chinatown block by block, Peter SooHoo worked tirelessly to forestall the evictions. It was as a result of SooHoo’s tireless efforts that so many Chinese Americans were able to stay as long as they had. 2 The desire of city elites to destroy Old Chinatown was not a surprise. In the popular imagination, white artists, authors, and filmmakers had long represented urban China towns as the physical embodiment of the immigrant alien. Since the passage of the 1882 Chinese Exclusion Act when the U.S. government began barring Chinese laborers from entering the country, the figure of the Chinese immigrant became the ethnic Other against which the American citizen was constructed.