Multiple Imputation of Missing Categorical and Continuous Values via Bayesian Mixture Models With Local Dependence

Imputation in nonparametric quantile regression with complex data

Research on economic need in the United States dates back at least to Jacob Riis's classic 1890 book, How the Other Half Lives. In the century since that publication, government efforts to reduce poverty have expanded enormously. Along with the expansion of anti-poverty programs, the demand for research on the nature of economic need in this country has mushroomed. A primary motivation for beginning the Survey of Income and Program Participation (SIPP) was to provide more and better information on families' level of eco­ nomic need and their use of government programs. This paper analyzes the ex­ tent to which SIPP has fulfilled that promise and presents recommendations for further changes and improvements in SIPP that might make it more useful to those who want to investigate poverty and anti-poverty programs. Data on income and program participation serve four basic functions in un­ derstanding poverty. First, income statistics are used to calculate measures of economic need, such as the official poverty rate, which are used as a measure of economic and social progress. Second, comparative measures of economic need across different subgroups of the population provide a measure of the relative progress and well-being of different groups in society, and help focus attention on those whose needs remain unmet. Third, data on program partici­ pation and family income provide information on who receives government transfer income and on how effective these programs are in reducing economic need. Fourth, repeated observations over time provide evidence on how all of these issues evolve. In particular, as anti-poverty programs and the macro

BackgroundMissing data is a pervasive problem in clinical research. Generative adversarial imputation nets (GAIN), a novel machine learning data imputation approach, has the potential to substitute missing data accurately and efficiently but has not yet been evaluated in empirical big clinical datasets.ObjectivesThis study aimed to evaluate the accuracy of GAIN in imputing missing values in large real-world clinical datasets with mixed-type variables. The computation efficiency of GAIN was also evaluated. The performance of GAIN was compared with other commonly used methods, MICE and missForest.MethodsTwo real world clinical datasets were used. The first was that of a cohort study on the long-term outcomes of patients with diabetes (50,000 complete cases), and the second was of a cohort study on the effectiveness of a risk assessment and management programme for patients with hypertension (10,000 complete cases). Missing data (missing at random) to independent variables were simulated at different missingness rates (20, 50%). The normalized root mean square error (NRMSE) between imputed values and real values for continuous variables and the proportion of falsely classified (PFC) for categorical variables were used to measure imputation accuracy. Computation time per imputation for each method was recorded. The differences in accuracy of different imputation methods were compared using ANOVA or non-parametric test.ResultsBoth missForest and GAIN were more accurate than MICE. GAIN showed similar accuracy as missForest when the simulated missingness rate was 20%, but was more accurate when the simulated missingness rate was 50%. GAIN was the most accurate for the imputation of skewed continuous and imbalanced categorical variables at both missingness rates. GAIN had a much higher computation speed (32 min on PC) comparing to that of missForest (1300 min) when the sample size is 50,000.ConclusionGAIN showed better accuracy as an imputation method for missing data in large real-world clinical datasets compared to MICE and missForest, and was more resistant to high missingness rate (50%). The high computation speed is an added advantage of GAIN in big clinical data research. It holds potential as an accurate and efficient method for missing data imputation in future big data clinical research.Trial registrationClinicalTrials.gov ID: NCT03299010; Unique Protocol ID: HKUCTR-2232

INTRODUCTION: Despite an excellent prognosis and survival rate, quality of life (QOL) in thyroid cancer is lower than expected. QOL in thyroid cancer survivors is similar to persons with colon and gynecologic cancers, but worse than breast cancer.1 The SF-36 QOL questionnaire has shown to be a valid tool for assessing QOL in thyroid cancer subjects.2 We evaluated QOL in both thyroid nodule and thyroid cancer patients in an institutional cancer registry. METHODS: The Short Form-36 data version 2 (SF-36) QOL data was obtained from single institution prospective bioinformatics thyroid cancer and benign nodule registry. Physical and mental health scores from the SF-36 questionnaire were obtained from both thyroid cancer and nodule patients. Physical and mental QOL responses were scored on a scale from 20-80 and categorized as either “the same or better than the average population,”; “below average,”; or “well below average.” A two sample Wilcoxon rank sum test and a chi-squared test were used to compare QOL between thyroid cancer and nodule subjects, using QOL as a continuous or categorical variable. Univariate descriptive statistics and bivariate analyses were performed using a Wilcoxon Rank Sum and Chi-squared test for categorical QOL data and Kruskal-Wallis for continuous QOL data. RESULTS: We analyzed 321 thyroid cancer and 32 nodule subjects who completed the SF-36 at a single point in time after diagnosis. There was no difference between the groups with regard to sex, age or QOL scores overall. Average age was 43 and 48 years in cancer and nodule groups respectively. Average Physical QOL score was 50.8 (standard deviation SD + 8.8) and 29.6 (SD+ 12.1), respectively (p=0.42). Average Mental QOL score was 48.9 (SD + 9.9) and 48.3 (SD + 8.1) respectively, p=0.16. Physical QOL score was significantly decreased as cancer stage increased when evaluating results as a continuous variable: Stage 1: 51.8 + 8.7; Stage 2: 50.6 +9.5; Stage 3: 48.7 + 7.2; Stage 4: 47.4 + 9.2. Results for mental QOL as a continuous variable by cancer stage showed improvement in reported QOL with increasing cancer stage: Stage 1: 48 (SD+ 9.9); Stage 2: 50.6 (SD + 11.1); Stage 3: 51.1 (SD + 9.13); Stage 4: 53 (SD+ 6.4), p=0.03. CONCLUSIONS: We found no overall difference in physical nor mental QOL between patients with thyroid cancer or benign thyroid nodules. Overall, QOL was in the “same or better than average” for all respondents, but low cancer stage was associated with higher physical and lower mental QOL scores.REFERENCES:1. Barbus, Elena & Peştean, Claudiu & Iulia Larg, Maria & Piciu, Doina. (2016). Quality of life in thyroid cancer patients: A literature review. Clujul Medical. 90. 10.15386/cjmed-7032. Lubitz CC, De Gregorio L, Fingeret AL, et al. Measurement and Variation in Estimation of Quality of Life Effects of Patients Undergoing Treatment for Papillary Thyroid Carcinoma. Thyroid. 2017;27(2):197-206. doi:10.1089/thy.2016.0260

One of the methods of increased privacy protection is the creation of synthetic data sets. In this article we consider the differences that emerge between synthetic and nonsynthetic data in one of the very few attempts to create a synthetic data file for a major household survey data set, the Survey of Income and Program Participation (SIPP). The data we use are nonsynthetic and synthetic versions of the SIPP linked to administrative earnings histories, known as the SIPP Gold Standard File (GSF) and the SIPP Synthetic Beta (SSB), respectively. We present a set of results on short-run earnings dynamics estimated on both the SSB and the GSF, focusing on earnings volatility—the variance of short-run earnings growth rates—as well as an error components decomposition of inequality into permanent and transitory components. We find that short-run instability—both volatility and the transitory component of earnings inequality—is higher in the SSB than the GSF although the differences are somewhat dependent on modeling choices and the treatment of low earnings. Differences between the two data sets emerge both because cross-sectional inequality is higher in the SSB than in the GSF and because the dynamics of earnings over both shorter and longer periods appear to be different.

The dynamics of participation in government programs can be studied using longitudinal data from the Survey of Income and Program Participation (SIPP). This paper examines some of the issues involved in such analysis, focusing first on the kinds of analysis that SIPP data allow. The second section discusses the major problems with the SIPP that complicate longitudinal analyses. A concluding section suggests changes that could simplify and improve the data.

Missing Data and Imputation: A Practical Illustration in a Prognostic Study on Low Back Pain

Neutrophil-to-lymphocyte ratio on admission is an independent risk factor for the severity and mortality in patients with coronavirus disease 2019

It is a common challenge for the geosciences to jointly model the uncertainty in continuous and categorical regionalised variables and to reproduce observed spatial correlation and complex relationships in realisations. The demand for computational efficiency in the case of high-dimensional data and large simulation domains has led practitioners to utilise approaches based on decorrelation/recorrelation and independent simulation. Among such approaches the method of min/max autocorrelation factors (MAF) has proven to be a practical technique for industrial purposes. This study presents a hybrid model for joint simulation of high-dimensional continuous and categorical variables. Continuous variables are transformed to Gaussian random functions (GRFs) via anamorphosis functions and categorical variables are obtained by truncating one or more GRFs based on the plurigaussian model. MAF factors are then derived from all GRFs. After independent simulation of MAF factors, different realisations of continuous and categorical variables are obtained via back-transformation of MAF factors followed by back-transformation for continuous and truncation for categorical variables, respectively. The proposed algorithm is illustrated through a case study.

Latent class based multiple imputation approach for missing categorical data

We sought to identify factors that are associated with hypotension after the induction of spinal anesthesia (SpA) by using an anesthesia information management system. Hypotension was defined as a decrease of mean arterial blood pressure of more than 30% within a 10-min interval, and relevance was defined as a therapeutic intervention with fluids or pressors within 20 min. From January 1, 1997, to August 5, 2000, data sets from 3315 patients receiving SpA were recorded on-line by using the automatic anesthesia record keeping system NarkoData. Hypotension meeting the predefined criteria occurred in 166 (5.4%) patients. Twenty-nine patient-, surgery-, and anesthesia-related variables were studied by using univariate analysis for a possible association with the occurrence of hypotension after SpA. Logistic regression with a forward stepwise algorithm was performed to identify independent variables (P < 0.05). The discriminative power of the logistic regression model was checked with a receiver operating characteristic curve. Calibration was tested with the Hosmer-Lemeshow goodness-of-fit test. The univariate analysis identified the following variables to be associated with hypotension after SpA: age, weight, height, body mass index, amount of plain bupivacaine 0.5% used for SpA, amount of colloid infusion before puncture, chronic alcohol consumption, ASA physical status, history of hypertension, urgency of surgery, surgical department, sensory block height of anesthesia, and frequency of puncture. In the multivariate analysis, independent factors for relevant hypotension after SpA consisted of three patient-related variables ("chronic alcohol consumption," odds ratio [OR] = 3.05; "history of hypertension," OR = 2.21; and the metric variable "body mass index," OR = 1.08) and two anesthesia-related variables ("sensory block height," OR = 2.32; and "urgency of surgery," OR = 2.84). The area of 0.68 (95% confidence interval, 0.63-0.72) below the receiver operating characteristic curve was significantly greater than 0.5 (P < 0.01). The goodness-of-fit test showed a good calibration of the model (H = 4.3, df = 7, P = 0.7; C = 7.3, df = 8, P = 0.51). This study contributes to the identification of patients with a high risk for hypotension after SpA induction, with the risk increasing two- or threefold with each additional risk factor. By using automated data collection, 5 (chronic alcohol consumption, history of hypertension, body mass index, sensory block height, and urgency of surgery) of 29 variables could be detected as having an association with hypotension after spinal anesthesia induction. The knowledge of these risk factors should be useful in increasing vigilance in those patients most at risk for hypotension, in allowing a more timely therapeutic intervention, or even in suggesting the use of alternative methods of spinal anesthesia, such as titrated continuous or small-dose spinal anesthesia.

The authors use data from the 1984 Survey of Income and Program Participation (SIPP) to investigate whether employer-provided health insurance reduced worker mobility (a phenomenon termed job-lock). The SIPP provides information on variables--particularly pension receipt, job tenure, and spouse job change--that were omitted from previous surveys and are, the authors argue, key to the estimation of well-defined mobility models. For dual-earner married men and women, the authors estimate a model that accounts for the interaction between husbands' and wives' job change decisions. For both married and single individuals, the results provide fairly strong evidence of job-lock among women, but only weak indications of job-lock among men. The authors speculate that this finding reflects higher health care use by women than by men.

The average age of grandparents in the U.S. in 2022 was 68 years. Grandparenthood is more common among older adults (aged 65 and older) than those in midlife (aged 40-64). In 2022, the percentage of older adults who were grandparents (67%) was more than 2.5 times greater than the percentage among those in midlife (24%). Data from the 2023 panel of the Survey of Income and Program Participation (SIPP) are used to update Wu, 2018 &amp; Westrick-Payne, 2023 comparing midlife grandparents to those at older ages across demographic characteristics in 2022. The SIPP is a nationally representative household-based survey that directly asks respondents if they are a grandparent, making it one of the few national surveys that identify non-resident grandparents. This profile is number three in our 2024 series on grandparents in the U.S.

In this paper, we consider the role of all work limiting health conditions on divorce behavior but we distinguish between less (exogenous) predictable versus more (endogenous) predictable shocks. We divide the reported health conditions in the 2004 Survey of Income and Program Participation (SIPP) into those that appear to arrive randomly conditional on age for married men and women versus those that appear to be more predictable. The six work limiting health conditions that appear to arrive largely exogenously are cancer, carpal tunnel syndrome, deafness or serious trouble hearing, paralysis, thyroid trouble and tumor cyst or growth. How the onset of work limiting conditions then affects divorce is studied across different demographic groups. The primary result of the study is that the onset of health problems that are unpredictable generally does not increase the likelihood of divorce while the divorce behavior is closely related to the onset or evolution of a broader, arguably endogenous set of health conditions. However, the patterns of response are shown to vary by race and ethnicity. For example, the onset of health conditions that do appear to be statistically related to demographic factors known to correlate with poor health behaviors appear to be associated with a statistically significant increase in the probability of divorce particularly among minority (Black and Hispanic) groups. The data used in the study are drawn from the SIPP, particularly from the topical module on “work disability history”, and the retrospective histories are used to construct a panel over the individuals’ lives. The primary estimation method is a panel linear probability model with fixed-effects.

We propose a new methodology that does not assume a prior specification of the statistical properties of the measurement errors and treats all sources as noisy measures of some underlying true value. The unobservable true value can be represented as a weighted average of all available measures, using weights that must be specified a priori unless there has been a truth audit. The Census Bureau's Survey of Income and Program Participation (SIPP) survey jobs are linked to Social Security Administration earnings data, creating two potential annual earnings observations. The reliability statistics for both sources are quite similar except for cases where the SIPP used imputations for some missing monthly earnings reports.