Biomedical subjects

Jingli Song

Publications and source records attributed to Jingli Song.

5 recordsLinked to original sources

PubMed2006

Risk factors for breast cancer in older women: the relative contribution of bone mineral density and other established risk factors.

AIM: To determine the contribution of bone mineral density (BMD) to breast cancer risk relative to other established breast cancer risk factors in postmenopausal women with osteoporosis. METHODS: Data for this analysis comprised those collected from women randomized to placebo in the MORE and CORE trials (N = 2,576). Risk factors measured at baseline included age, family history of breast cancer, estradiol level, body mass index, prior hormone therapy, BMD and vertebral fracture status. Cox proportional hazards regression models were used to calculate the hazard ratios (HRs) and 95% confidence intervals (CIs). RESULTS: Over a total of 13,698 woman-years of follow-up, 65 incident breast cancers occurred. In univariate analyses, older age and family history of breast cancer were the strongest predictors of breast cancer risk, associated with a 2.4- and 2.6-fold increase in breast cancer incidence. A higher estradiol level was associated with a 1.9-fold increase in breast cancer incidence. The association between femoral neck BMD and breast cancer incidence was only significant after adjustment for age (P = 0.03). The final multivariable model included age, family history, estradiol, BMD, and the BMD-estradiol interaction since the effect of BMD on breast cancer varied by estradiol level (interaction P-value, 0.04); in those with a lower estradiol level, a higher BMD was associated with a 2.6-fold increased in breast cancer. CONCLUSION: Overall, BMD is a relatively weak predictor of breast cancer risk in these postmenopausal women with osteoporosis, after taking into consideration age, family history and endogenous estradiol level.

Aged↗

PubMed2005

Assay validation for left-censored data.

In laboratory validation studies, it is often important to assess agreement between two assays, based on different techniques. Oftentimes, both assays have lower limits of detection and thus measurements are left censored. For example, in studies of Human Immunodeficiency Virus (HIV), the branched DNA (bDNA) assay was developed to quantify HIV-1 RNA concentrations in plasma. Validation of newer assays, such as the RT-PCR (reverse transcriptase polymerase chain reaction) involves assessing agreement of measurements obtained using the two techniques. Both bDNA and RT-PCR assays have lower limits of detection and thus new statistical methods are needed for assessing agreement between two left-censored variables. In this paper, we present maximum likelihood and generalized estimating equations approaches to evaluate agreement between two assays that are subject to lower limits of detection. The concordance correlation coefficient is used as an agreement index. The methodology is illustrated using HIV RNA assay data collected as part of a long-term HIV cohort study.

Computer Simulation↗

PubMed2005

Skeletal effects of raloxifene after 8 years: results from the continuing outcomes relevant to Evista (CORE) study.

UNLABELLED: In the CORE breast cancer trial of 4011 women continuing from MORE, the incidence of nonvertebral fractures at 8 years was similar between placebo and raloxifene 60 mg/day. CORE had limitations for assessing fracture risk. In a subset of 386 women, 7 years of raloxifene treatment significantly increased lumbar spine and femoral neck BMD compared from the baseline of MORE. INTRODUCTION: The multicenter, double-blind Continuing Outcomes Relevant to Evista (CORE) trial assessed the effects of raloxifene on breast cancer for 4 additional years beyond the 4-year Multiple Outcomes of Raloxifene Evaluation (MORE) osteoporosis treatment trial. MATERIALS AND METHODS: In CORE, placebo-treated women from MORE continued with placebo (n = 1286), whereas those previously given raloxifene (60 or 120 mg/day) received raloxifene 60 mg/day (n = 2725). As a secondary endpoint, new nonvertebral fractures were analyzed as time-to-first event in 4011 postmenopausal women at 8 years. A substudy assessed lumbar spine and femoral neck BMD at 7 years, with the primary analysis based on 386 women (127 placebo, 259 raloxifene) who did not take other bone-active agents from the fourth year of MORE and who were > or =80% compliant with study medication in CORE. RESULTS: The risk of at least one new nonvertebral fracture was similar in the placebo (22.9%) and raloxifene (22.8%) groups (hazard ratio [HR], 1.00; Bonferroni-adjusted CI, 0.82, 1.21). The incidence of at least one new nonvertebral fracture at six major sites (clavicle, humerus, wrist, pelvis, hip, lower leg) was 17.5% in both groups. Posthoc Poisson analyses, which account for multiple events, showed no overall effect on nonvertebral fracture risk, and a decreased risk at six major nonvertebral sites in women with prevalent vertebral fractures (HR, 0.78; 95% CI, 0.63, 0.96). At 7 years after MORE randomization, the differences in mean lumbar spine and femoral neck BMD with raloxifene were 1.7% (p = 0.30) and 2.4% (p = 0.045), respectively, from placebo. Compared with MORE baseline, after 7 years, raloxifene treatment significantly increased lumbar spine (4.3% from baseline, 2.2% from placebo) and femoral neck BMD (1.9% from baseline, 3.0% from placebo). BMDs were significantly increased from MORE baseline at all time-points at both sites with raloxifene. CONCLUSION: Raloxifene therapy had no effect on nonvertebral fracture risk after 8 years, although CORE had limitations for fracture risk assessment. BMD increases were maintained after 7 years of raloxifene.

Aged↗

PubMed2005

Assessing intra, inter and total agreement with replicated readings.

In clinical studies, assessing agreement of multiple readings on the same subject plays an important role in the evaluation of continuous measurement scale. The multiple readings within a subject may be replicated readings by using the same method or/and readings by using several methods (e.g. different technologies or several raters). The traditional agreement data for a given subject often consist of either replicated readings from only one method or multiple readings from several methods where only one reading is taken from each of these methods. In the first case, only intra-method agreement can be evaluated. In the second case, traditional agreement indices such as intra-class correlation (ICC) or concordance correlation coefficient (CCC) is often reported as inter-method agreement. We argue that these indices are in fact measures of total agreement that contains both inter and intra agreement. Only if there are replicated readings from several methods for a given subject, then one can assess intra, inter and total agreement simultaneously. In this paper, we present new inter-method agreement index, inter-CCC, and total agreement index, total-CCC, for agreement data with replicated readings from several methods where the ICCs within methods are used to assess intra-method agreement for each of the several methods. The relationship of the total-CCC with the inter-CCC and the ICCs is investigated. We propose a generalized estimating equations approach for estimation and inference. Simulation studies are conducted to assess the performance of the proposed approach and data from a carotid stenosis screening study is used for illustration.

Carotid Stenosis↗

PubMed2002

Overall concordance correlation coefficient for evaluating agreement among multiple observers.

Accurate and precise measurement is an important component of any proper study design. As elaborated by Lin (1989, Biometrics 45, 255-268), the concordance correlation coefficient (CCC) is more appropriate than other indices for measuring agreement when the variable of interest is continuous. However, this agreement index is defined in the context of comparing two fixed observers. In order to use multiple observers in a study involving large numbers of subjects, there is a need to assess agreement among these multiple observers. In this article, we present an overall CCC (OCCC) in terms of the interobserver variability for assessing agreement among multiple fixed observers. The OCCC turns out to be equivalent to the generalized CCC (King and Chinchilli, 2001, Statistics in Medicine 20, 2131-2147; Lin, 1989; Lin, 2000, Biometrics 56, 324-325) when the squared distance function is used. We evaluated the OCCC through generalized estimating equations (Barnhart and Williamson, 2001, Biometrics 57, 931-940) and U-statistics (King and Chinchilli, 2001) for inference. This article offers the following important points. First, it addresses the precision and accuracy indices as components of the OCCC. Second, it clarifies that the OCCC is the weighted average of all pairwise CCCs. Third, it is intuitively defined in terms of interobserver variability. Fourth, the inference approaches of GEE and the U-statistics are compared via simulations for small samples. Fifth, we illustrate the use of the OCCC by two medical examples with the GEE, U-statistics, and bootstrap approaches.

Adult↗