Assessment of hospital performance with a case-mix standardized mortality model using an existing administrative database in Japan

Background  

Few studies have examined whether risk adjustment is evenly applicable to hospitals with various characteristics and case-mix. In this study, we applied a generic prediction model to nationwide discharge data from hospitals with various characteristics.     

Comparisons of health inequalities in small areas with using the standardized mortality ratios in Korea

OBJECTIVES: This study was performed to compare the standardized mortality ratios among different small areas and to explore the usefulness of standardized mortality ratios in South Korea. METHODS: To calculate the standardized mortality ratio (SMR), we obtained the national deaths certificate data (2004-2006) and national registration population data (2003-2006), and these were provided by the National Statistical Office. The small areas (Eup.Myoun.Dong) were based on the subdivisions of counties. Among the 3,580 small areas classified by the National Statistical Office, 3,571 areas were included in this study. The basic statistics and decile distributions of the SMRs for all the regional levels were calculated, and the small area maps were also produced for some selected regions. To evaluate the precision of SMR, we calculated the 95% confidence intervals of the SMR in selected small areas. RESULTS: The mean and the standard deviation of the SMRs among all small areas were 100.8 and 17.0, respectively. The range was 30.6-211.7 and the inter-quartile range was 20.7. Seoul metropolitan city displayed the lowest mean SMR among 16 regions in South Korea, and 34.6 percent of the small area SMRs belonged to the first decile (the lowest group). On the contrary, the mean SMR of Gyeongsangnam province was highest, and 26.1 percent of the small area SMRs belonged to the tenth decile(the highest group). In some areas, the precision of the SMR, which was calculated by the 95% confidence intervals, remained questionable, yet it was quite stable for almost areas. CONCLUSIONS: The standardized mortality ratios can be useful for allocating health resources at the small area level in Korea.     

Hospital standardized mortality ratios: a tale of two sites

In 2008, Caritas Health Group (CHG), composed of the Edmonton area Catholic acute hospital sites, became part of Covenant Health, Alberta's largest faith-based healthcare provider. In 2005, Capital Health (CH)/CHG began its work with the Canadian Institute of Health Information (CIHI). It is one of eight pioneer health regions in Canada to test the methodology of the hospital standardized mortality ratio (HSMR) developed in the United Kingdom, through Sir Brian Jarman, director of the Dr. Foster Unit (DRF), at the Imperial College in London. This article focuses on understanding the UK's HSMR process and describing its use in two hospitals in the Edmonton area. The hospitals used raw mortality data and chart review processes following the introduction of the HSMR in Canada in order to better understand the intricacy of this measure. The article focuses on the in-house process that became a foundation for an in-depth exploratory use of the CIHI HSMR e-portal for hospital sites within the province of Alberta.     

Modelling RNA-Seq data with a zero-inflated mixture Poisson linear model

RNA sequencing (RNA-Seq) has been frequently used in genomic studies and has generated a vast amount of data. The RNA-Seq data are composed of two parts: (a) a sequence of nucleotides of the genome; and (b) a corresponding sequence of counts, standing for the number of short reads whose mapped positions start at each position of the genome. One common feature of these count data is that they are typically nonuniform; recent studies have revealed that the nonuniformity is partially owing to a systematic bias resulted from the sequencing preference. Existing works in the literature model the nonuniformity with a single component Poisson linear model that incorporates the effects of the sequencing preference. However, we observe consistently that the short reads mapped to a gene may have a mixture structure and can be zero-inflated. A single component model may not suffice to model the complexity of such data. In this paper, we propose a zero-inflated mixture Poisson linear model for the RNA-Seq count data and derive a fast expectation–maximisation-based algorithm for estimating the unknown parameters. Numerical studies are conducted to illustrate the effectiveness of our method.     

A truncated Poisson regression model with applications to occurrence of adenomatous polyps

Adenomatous polyps are considered as the dominant precursor lesion of colorectal cancer. A phase III colorectal cancer prevention trial, conducted by the Arizona Cancer Center, concerns the ability of wheat bran fibre supplement to reduce the recurrence of adenomatous polyps. All participants in the study are to have had colorectal polyps detected and removed during a baseline (qualifying) colonoscopy within three months prior to enrolment. In this paper, our interest focuses on occurrence of adenomatous polyps at the baseline colonoscopy. We use a truncated Poisson model to fit these types of data. We develop a regression model to assess the effects of explanatory factors on the positive counting variable. We fit truncated Poisson parameters by a log-linear regression model and estimate regression parameters by the maximum likelihood procedure. Finally, we apply it to the baseline colonoscopy data from the Wheat Bran Fiber study. © 1997 John Wiley & Sons, Ltd.     

上海市徐汇区1995-2004年20岁以下人群白血病死亡率Poisson回归分析

[目的] 了解1995～2004年上海市徐汇区20岁以下人群白血病死亡率的时间变化趋势及其与性别、年龄的关系。 [方法]拟合Poisson回归模型,分析死亡人数与影响因素之间的关系。 [结果] 徐汇区10年间20岁以下人群白血病死亡率无随时间变化的趋势(z=0．38,P>0．5),女性死亡率低于男性(IRR=0．4706,P<0．1),0-4岁及5 ～9岁的死亡率均低于10～19岁者(IRR1=0．2308,IRR2=0．1538,P<0．05),每年因白血病死亡的20岁以下者平均2．5 人。 [结论] 徐汇区近10年20岁以下人群白血病死亡率稳定,但在性别和年龄上有差异;Poisson回归模型适用于类似资料的统计分析。     

基于空间模型的我国孕产妇死亡率影响因素分析

目的分析我国孕产妇死亡率的空间分布特点,利用空间模型进行拟合优度检验,探讨我国孕产妇死亡率的影响因素。方法采用2005-2014年中国统计年鉴及中国卫生和计划生育年鉴面板数据,运用空间面板数据模型,对影响我国孕产妇死亡率的因素进行定量分析。结果空间滞后模型在0.1的检验水准上,孕产妇死亡率、住院分娩率、每千人口床位数、文盲率差异有统计学意义;其他条件不变时,以上因素每增加1%,孕产妇死亡率会相应降低0.386 2/10万、0.391 1/10万、0.423 2/10万和升高0.316 5/10万。结论孕产妇系统管理率、住院分娩率、妇幼保健专业机构的床位数是孕产妇死亡率的保护因素;文盲率是孕产妇死亡率的危险因素。     

心力衰竭患者治疗效果标准化院内死亡率评价

目的 基于风险调整思想计算标准化院内死亡率, 合理评价医院心力衰竭的治疗质量。方法 收集黑龙江省20家三甲医院2009年1月—2010年10月入院的1 862例心力衰竭患者病历资料, 采用两水平logistic回归模型构建心力衰竭患者院内死亡的风险调整模型, 利用ROC曲线下面积(AUC)评价模型的拟合优度。结果1 862 例心力衰竭患者中, 87例患者在住院期间死亡, 院内死亡率为4.67%。不同特征心力衰竭患者院内死亡率比较, 不同民族、入院时病情、住院天数、心率及是否患呼吸系统疾病、肾脏疾病、心肌病、失盐低钠综合征心力衰竭患者院内死亡率差异均有统计学意义(P<0.05);两水平logistic回归分析结果显示, 年龄较大、有疾病史及患肾脏疾病和失盐低钠综合征的心力衰竭患者院内死亡风险较大, 住院天数较长、入院时病情较轻和患高血压的心力衰竭患者院内死亡风险较小;心力衰竭患者院内死亡风险调整模型的ROC曲线, AUC为0.80, 95%CI=0.75~0.85, P<0.001;风险调整前, 院内死亡率在医院间的变异范围为0~12.82%, 风险调整后, 医院的排序发生变化, 院内死亡率在医院间的变异范围为2.59%~7.62%。结论 风险调整后, 院内死亡率在医院间的变异减小, 粗院内死亡率和标准化院内死亡率对医院的排序不一致, 调整患者风险因素的标准化院内死亡率能合理地评价医院的治疗质量。     

A simple method to calculate the confidence interval of a standardized mortality ratio (SMR)

In analyzing standardized mortality ratios (SMRs), it is of interest to calculate a confidence interval for the true SMR. The exact limits of a specific interval can be obtained by means of the Poisson distribution either within an iterative procedure or by one of the tables. The limits can be approximated in using one of various shortcut methods. In this paper, a method is described for calculating the exact limits in a simple and easy way. The method is based on the link between the chi 2 distribution and the Poisson distribution. Only a table of the chi 2 distribution is necessary.     

The application of a Poisson model to the annual distribution of daily mortality at six Montreal hospitals.

The daily distributions of annual mortality for varying numbers of years between 1965 and 1975 were investigated in three geriatric hospitals and three general hospitals in the Montreal area. Nearly all the observed mortality distributions were found to mimic the classical Poisson distribution, with little departure. In two of the larger hospitals, the matching of the daily mortality distributions with their Poisson models met stringent statistical criteria. In one of them it was even possible to predict the expected mortality frequencies merely from a knowledge of the annual totals. The remaining four hospitals, which included the three geriatric institutions, also exhibited mortalities regarded as highly suggestive of Poisson distributions, although in one of the geriatric hospitals the mortality distribution tended to be somewhat erratic in this respect.     

A comparison of some simple methods to identify geographical areas with excess incidence of a rare disease such as childhood leukaemia.

Six statistics are compared in a simulation study for their ability to identify geographical areas with a known excess incidence of a rare disease. The statistics are the standardized incidence ratio, the empirical Bayes method of Clayton and Kaldor, Poisson probability, a statistic based on the 'Breslow T' test (BT) and two statistics based on the 'Potthoff-Whittinghill' test (PW) for extra-Poisson variance. Two alternative processes of clustering are simulated in which high-risk locations could be caused by environmental sources or could be sites of microepidemics of an infectious agent contributing to a rare disease such as childhood leukaemia. The simulation processes use two parameters (proportion of cases found in clusters and mean cluster size) which are varied to embrace a variety of situations. Real and artificial data sets of small area populations are considered. The most extreme of the artificial sets has all areas of equal population size. The other data sets use the small census areas (municipalities) in Finland since these have extremely heterogeneous population size distribution. Subset selection allows examination of this variability. Receiver operator curve methodology is used to compare the efficacy of the statistics in identifying the cluster areas; statistics are compared for the proportion of true high-risk areas identified in the top 1 per cent and 10 per cent of ranked areas. One of the PW statistics performed consistently well under all circumstances, although the results for the BT statistic were marginally better when only the top 1 per cent of ranked areas was considered. The standardized incidence ratio performed consistently worst.     

Geographical pattern of malignant neoplasm by cluster analysis using standardized mortality ratios (SMRs) in Ibaraki Prefecture,Japan

We investigated the geographical patterns of mortality from eight (males)/ten (females) sites of malignant neoplasm, using cluster analysis with Standardized Mortality Ratios (SMRs), and examined the relationship between the mortality structure and urbanization. To explore the geographical tendencies is important for the prevention of cancers; such as noticing risk factors associated with regional variance. The death rates, by site, gender and age from 1990 to 1994 in Japan, were obtained from Vital Statistics. The deaths and population in municipalities were obtained from "Population of Ibaraki Prefecture". These were represented as averaged values in five-year periods. As an indicator of urbanization and mortality structure, the population density of municipalities and the overall rank scores of SMRs were used, respectively. Cluster analysis formed some distinctive structures. For males, Cluster 1 included four municipalities and three of these were located in the mountainous area in northwest Ibaraki, characterized by high SMRs from bone marrow. Cluster 5 consisted of the mid-south areas, characterized by high SMRs from stomach cancer. For females, the clusters seemed to be characterized by SMRs from esophagus cancer. An association between mortality structure and urbanization was found for females, 0.364 (p<0.01), but not for males, 0.162 (p=0.14).     

Correlation of cerebrovascular disease standardized mortality ratios with dietary sodium and the sodium/potassium ratio among the Japanese population

One thousand seventy-two 24-hr diet duplicate samples were collected from inhabitants of 49 regions in various parts of Japan during the winters of 1977-1981. An additional 238 samples were collected in an adjacent summer. The samples were analyzed for sodium (Na) and potassium (K) by flame atomic absorption spectrometry and for chloride (Cl) with a chloride counter. The winter-summer differences in Na, Cl, and Na/K were essentially negligible. When the regional means of Na, K, Cl, and Na/K were compared with the 1969-1978 standardized mortality ratios of each region, positive and significant correlations were observed between winter Na and the standardized mortality ratios for cerebrovascular disease (P less than 0.01), cerebral infarction (P less than 0.01), and subarachnoid hemorrhage (P less than 0.05) in both males and females. The correlation (P less than 0.01) with the cerebrovascular disease standardized mortality ratio was further confirmed by the values for 1978-1982. In the case of the Na/K ratio, the correlation with the standardized mortality ratio for each of the three diseases was significant for men (P less than 0.01 or 0.05, depending on the disease) but not for women (P greater than 0.05). Both Na and Na/K showed significant associations with the ischemic heart disease standardized mortality ratio in men (P less than 0.05) but not in women (P greater than 0.05). In contrast, no positive association was found between Na, K, Cl, or Na/K and standardized mortality ratios for diabetes mellitus, liver cirrhosis, tuberculosis, or liver cancer (P greater than 0.05). Current blood pressure did not appear to correlate with any of the Na, K, Cl, or Na/K measurements. The validity of the present observation is discussed.     

Testing the equality of two Poisson means using the rate ratio

In this article, we investigate procedures for comparing two independent Poisson variates that are observed over unequal sampling frames (i.e. time intervals, populations, areas or any combination thereof). We consider two statistics (with and without the logarithmic transformation) for testing the equality of two Poisson rates. Two methods for implementing these statistics are reviewed. They are (1) the sample-based method, and (2) the constrained maximum likelihood estimation (CMLE) method. We conduct an empirical study to evaluate the performance of different statistics and methods. Generally, we find that the CMLE method works satisfactorily only for the statistic without the logarithmic transformation (denoted as W(2)) while sample-based method performs better for the statistic using the logarithmic transformation (denoted as W(3)). It is noteworthy that both statistics perform well for moderate to large Poisson rates (e.g. > or =10). For small Poisson rates (e.g. <10), W(2) can be liberal (e.g. actual type I error rate/nominal level > or =1.2) while W(3) can be conservative (e.g. actual type I error rate/nominal level < or =0.8). The corresponding sample size formulae are provided and valid in the sense that the simulated powers associated with the approximate sample size formulae are generally close to the pre-chosen power level. We illustrate our methodologies with a real example from a breast cancer study.     

Childhood leukaemia: a model of pre-obesity.

The prevalence of obesity in children, as in adults, is increasing dramatically. The extent to which this is due to reduced energy expenditure, increased energy intake, or both, is unclear at present. This in part reflects the limitations of existing models of the pre-obese state. In childhood acute lymphoblastic leukaemia (ALL), patients typically gain weight excessively during and after 2 years of therapy, and are at high risk of becoming obese. Previous studies have failed to identify the cause of obesity in these patients. We have tested the hypothesis that excess weight gain in ALL is due to reduced total energy expenditure (TEE), measured using the doubly-labelled water method, and have identified risk factors for excess weight gain in ALL. Pre-obese children with ALL in the dynamic phase of weight gain are less physically active than their peers, with a reduced TEE of approximately 1.2 (95% CI 0.2, 2.2) MJ/d. While other factors might contribute to excess weight gain, lifestyle (i.e. reduced habitual physical activity) plays a central role in ALL. Several considerations suggest that ALL might be a useful model of the pre-obese state: lifestyle is critical to development of obesity in ALL; ALL is relatively common; approximately 70% of patients survive; patients are readily accessible during the 2 years of therapy and beyond.     

Childhood leukaemia in areas with different radon levels: a spatial and temporal analysis using GIS

OBJECTIVES—To evaluate the relation between exposure to ground radon levels and leukaemia among children using existing population and disease registers.
DESIGN—Ecological correlation study.
SETTING—The county of Östergötland in Sweden.
METHODS—Every child born in the county between 1979 and 1992 was mapped to the property centroid coordinates by linking addresses in the population and property registers. Population maps were overlaid with radon maps and exposure at birth and each subsequent year was quantified as high, normal, low or unknown. This was analysed with data from the tumour registry. Standardised mortality ratios (SMRs) were calculated using the age and sex specific rates for Sweden for the year 1995.
RESULTS—90 malignancies occurred among 53 146 children (498 887 person years) who formed the study population. SMRs for acute lymphatic leukaemia (ALL) among children born in high, normal and low risk areas were 1.43, 1.17 and 0.25 respectively. The relative risk for the normal risk group and high risk group as compared with the low risk group was 4.64 (95% CI 1.29, 28.26) and 5.67 (95% CI 1.06, 42.27). The association between ALL and continued residence at normal or high risk areas showed a similar trend. No association between radon risk levels and any other malignancy was seen.
CONCLUSION—Children born in and staying at areas where the risk from ground radon has been classified as low are less likely to develop ALL than those born in areas classified as normal and high risk.