Multicollinearity is an important issue affecting the results of regression analysis. LASSO developed in recent years has great advantages in selecting explanatory variables, processing high-dimensional data, and solving multicollinearity problems. This method adds a penalty term to the model estimation, which can compress the regression coefficients of some unnecessary variables to zero and then remove them from the model to achieve the purpose of variable screening. This paper focuses on the LASSO method and compares it with optimal subsets, ridge regression, adaptive LASSO, and elastic net results. It is found that both LASSO and adaptive LASSO have good performance in solving independent variable multicollinearity problems and enhancing model interpretation and prediction accuracy.
Humans ; Regression Analysis

Objective: To investigate the current situation of job involvement of nurses in military hospitals in Henan Province and analyze the influencing factors, so as to provide reference for improving the level of job involvement of military nurses. Methods: In February 2022, the employed nurses of 4 military hospitals in Henan Province were investigated by convenient sampling method. A total of 663 questionnaires were collected, including 632 valid questionnaires, with an effective recovery rate of 95.32%. The self-designed questionnaire was used to investigate the basic information of nurses, the Job Involvement Scale was used to investigate the job involvement of nurses, the Emotional Labor Scale for Nurses was used to investigate nurses' emotions, and the Work-Family Conflict Scale was used to investigate the work-family conflict of nurses. Independent sample t-test and univariate analysis of variance were used to compare the job involvement of military employed nurses with different demographic characteristics, Pearson correlation analysis was used to explore the correlation between emotional labor, work-family conflict and job involvement, and hierarchical regression analysis was used to explore the impact of relevant variables on the job involvement of military employed nurses. Results: The total average score of job involvement of military employed nurses was (3.68±1.13), and the scores of vitality, dedication and focus were (3.64±1.15), (3.74±1.25) and (3.67±1.21) respectively. The total score of emotional labor of nurses was 33-80 (62.95±8.12), with an average score of (3.93±0.51). The total score of work-family conflict was 18-94 (55.16±13.53), with an average score of (3.06±0.75). Professional emotional regulation, patient-centered emotional inhibition and standardized emotional play were positively related to the job involvement (r=0.46, 0.41, 0.22, P<0.01). Time-based conflict, stress-based conflict and behavior-based conflict had negative correlation with the job involvement (r=-0.12, -0.23, -0.20, P<0.01). In hierarchical regression analysis, after controlling demographic variables, emotional labor and work-family conflict accounted for 17.2% and 4.2% of the variation of job involvement. Conclusion: The job involvement of military employed nurses tends to be at a moderate level. Emotional labor and work-family conflict can significantly affect their job involvement.
United States ; Humans ; Hospitals, Military ; Family Conflict ; Surveys and Questionnaires ; Regression Analysis ; Nurses ; Job Satisfaction

Body mass index (BMI) has been increasing globally in recent decades. Previous studies reported that BMI was associated with sex hormone levels, but the results were generated via linear regression or logistic regression, which would lose part of information. Quantile regression analysis can maximize the use of variable information. Our study compared the associations among different regression models. The participants were recruited from the Center of Reproductive Medicine, The First Hospital of Jilin University (Changchun, China) between June 2018 and June 2019. We used linear, logistic, and quantile regression models to calculate the associations between sex hormone levels and BMI. In total, 448 men were included in this study. The average BMI was 25.7 (standard deviation [s.d.]: 3.7) kg m^-2; 29.7% (n = 133) of the participants were normal weight, 45.3% (n = 203) of the participants were overweight, and 23.4% (n = 105) of the participants were obese. The levels of testosterone and estradiol significantly differed among BMI groups (all P < 0.05). In linear regression and logistic regression, BMI was associated with testosterone and estradiol levels (both P < 0.05). In quantile regression, BMI was negatively associated with testosterone levels in all quantiles after adjustment for age (all P < 0.05). BMI was positively associated with estradiol levels in most quantiles (≤80^th) after adjustment for age (all P < 0.05). Our study suggested that BMI was one of the influencing factors of testosterone and estradiol. Of note, the quantile regression showed that BMI was associated with estradiol only up to the 80^th percentile of estradiol.
Male ; Humans ; Body Mass Index ; Cross-Sectional Studies ; Gonadal Steroid Hormones ; Regression Analysis ; Estradiol ; Testosterone

OBJECTIVE: To describe the secular trends of age at menarche and age at natural menopause of women from a county of Shandong Province. METHODS: Based on the data of the Premarital Medical Examination and the Cervical Cancer and Breast Cancer Screening of the county, the secular trends of age at menarche in women born in 1951 to 1998 and age at menopause in women born in 1951 to 1975 were studied. Joinpoint regression was used to identify potential inflection points regarding the trend of age at menarche. Average hazard ratios (AHR) of early menopause among women born in different generations were estimated by performing multivariate weighted Cox regression. RESULTS: The average age at menarche was (16.43±1.89) years for women born in 1951 and (13.99±1.22) years for women born in 1998. The average age at menarche was lower for urban women than that for rural women, and the higher the education level, the lower the average age at menarche. Joinpoint regression analysis identified three inflection points: 1959, 1973 and 1993. The average age at menarche decreased annually by 0.03 (P < 0.001), 0.08 (P < 0.001), and 0.03 (P < 0.001) years respectively for women born during 1951-1959, 1960-1973, and 1974-1993, while it remained stable for those born during 1994-1998 (P=0.968). As for age at menopause, compared with women born during 1951-1960, those born during 1961-1965, 1966-1970 and 1971-1975 showed a gradual decrease in the risk of early menopause and a tendency to delay the age at menopause. The stratified analysis presented that the risk of early menopause gradually decreased and the age of menopause showed a significant delay among those with education level of junior high school and below, but this trend was not obvious among those with education level of senior high school and above, where the risk of early menopause decreased and then increased among those with education level of college and above, and the corresponding AHRs were 0.90 (0.66-1.22), 1.07 (0.79-1.44) and 1.14 (0.79-1.66). CONCLUSION The age at menarche for women born since 1951 gradually declined until 1994 and leveled off, with a decrease of nearly 2.5 years in these years. The age at menopause for women born between 1951 and 1975 was generally delayed over time, but the trend of first increase and then decrease was observed among those with relatively higher education levels. In the context of the increasing delay in age at marriage and childbearing and the decline of fertility, this study highlights the necessity of the assessment and monitoring of women' s basic reproductive health status, especially the risk of early menopause.
Female ; Humans ; Aged ; Menarche ; Menopause ; Regression Analysis ; Fertility ; China/epidemiology* ; Age Factors

Objective: To understand the characteristics and trend of the premature death rate of 4 major chronic diseases in Ji'nan from 2015 to 2020. Methods: The death cause surveillance data and population data during 2015-2020 in Ji'nan were collected, and abbreviated life table, Joinpoint regression analysis and other methods were used to analyze the characteristics and change trends of the premature death rates of 4 major chronic diseases. Results: The crude mortality rate and age standardized mortality rate changes for the 4 major chronic diseases from 2015 to 2020 range from 568.65/100 000 to 604.06/100 000 and 366.77/100 000 to 432.48/100 000, respectively. The annual premature death rate of 4 major chronic diseases declined by 3.33% averagely from 2015 to 2020 (95%CI: -6.25%--0.32%), which might be explained by the declines of the premature death rates of cardiovascular and cerebrovascular diseases [average annual percentage change (AAPC)=-3.23%, 95%CI: -6.32%--0.05%] and cancer (AAPC=-3.58%,95%CI:-6.83%--0.21%). The average decline rate in women (AAPC=-4.19%,95%CI:-7.56%- -0.70%) was higher than that in men (AAPC=-2.92%,95%CI: -5.65%--0.11%). Conclusions: The premature death rate of 4 major chronic diseases showed a downward trend in Ji'nan from 2015 to 2020. Men should be considered as a key population in the prevention and control of 4 major chronic diseases, and attention should also be paid to the non-significant declines in the premature death rates of chronic respiratory diseases and diabetes.
Cerebrovascular Disorders ; Chronic Disease ; Diabetes Mellitus ; Female ; Humans ; Male ; Mortality, Premature ; Regression Analysis

Reduced rank regression is an extended multivariate linear regression model with the function of dimension reduction. It has been more and more widely used in nutritional epidemiology research to understand people's dietary patterns in recent years. However, there has been no existing Stata package or command to implement reduced rank regression independently. Therefore, we developed a new user-written package named "rrr" for its implementation in Stata. This paper summarizes the methodology of reduced rank regression, the development and functions of the Stata rrr package and its application in the China Kadoorie Biobank dataset, with the aim of facilitating the future wide use of this statistical method in epidemiology and public health research.
China ; Humans ; Models, Statistical ; Public Health ; Regression Analysis

Objective: To introduce and compare four analysis methods of multiple parallel mediation model, including pure regression method, method based on inverse probability weighting, extended natural effect model method and weight-based imputation strategies. Methods: For the multiple parallel mediation model, the simulation experiments of three scenarios were carried out to compare the performance of different methods in estimating direct and indirect effects in different situations. Dataset from UK Biobank was then analyzed by using the four methods. Results: The estimation biases of the regression method and the inverse probability weighting method were relatively small, followed by the extended natural effect model method, and the estimation results of the weight-based imputation strategies were quite different from the other three methods. Conclusions: Different multiple parallel mediation analysis methods have different application situations and their own advantages and disadvantages. The regression method is more suitable for continuous mediator, and the inverse probability weighting method is more suitable for binary mediator. The extended natural effect model method has better performances when the residuals of two parallel mediators are positively correlated and the correlation degree is small. The weight-based imputation strategies might not be appropriate for parallel mediation analysis. Therefore, appropriate methods should be selected according to the specific situation in practice.
Bias ; Computer Simulation ; Humans ; Mediation Analysis ; Models, Statistical ; Probability ; Regression Analysis ; Research Design

OBJECTIVE: This study aims to investigate the association of metabolic phenotypes that are jointly determined by body mass index (BMI) or fat mass percentage and metabolic health status with the ten-year risk of cardiovascular disease (CVD) among Chinese adults. METHODS: Data were obtained from a cross-sectional study. BMI and body fat mass percentage (FMP) combined with the metabolic status were used to define metabolic phenotypes. Multiple linear regression and logistic regression were used to examine the effects of metabolic phenotypes on CVD risk. RESULTS: A total of 13,239 adults aged 34-75 years were included in this study. Compared with the metabolically healthy non-obese (MHNO) phenotype, the metabolically unhealthy non-obese (MUNO) and metabolically unhealthy obese (MUO) phenotypes defined by BMI showed a higher CVD risk [odds ratio, OR (95% confidence interval, CI): 2.34 (1.89-2.89), 3.45 (2.50-4.75), respectively], after adjusting for the covariates. The MUNO and MUO phenotypes defined by FMP showed a higher CVD risk [ OR (95% CI): 2.31 (1.85-2.88), 2.63 (1.98-3.48), respectively] than the MHNO phenotype. The metabolically healthy obese phenotype, regardless of being defined by BMI or FMP, showed no CVD risk compared with the MHNO phenotype. CONCLUSION General obesity without central obesity does not increase CVD risk in metabolically healthy individuals. FMP might be a more meaningful factor for the evaluation of the association of obesity with CVD risk. Obesity and metabolic status have a synergistic effect on CVD risk.
Adipose Tissue/anatomy & histology* ; Adult ; Aged ; Body Mass Index ; Cardiovascular Diseases/etiology* ; China/epidemiology* ; Cross-Sectional Studies ; Female ; Humans ; Male ; Metabolic Diseases/etiology* ; Middle Aged ; Obesity/complications* ; Phenotype ; Regression Analysis ; Risk Factors

Joinpoint regression was used to analyze the trend of syphilis and gonorrhea incidence rate among 10-19 year old adolescents in Liaoning Province from 2006 to 2020. The syphilis and gonorrhea data in Liaoning Province were reported in the infectious disease monitoring system of China's disease prevention and control information system. From 2006 to 2020, a total of 7 721 cases of syphilis in 10-19 year old adolescents were reported in Liaoning Province, with an incidence rate about 0.90/100 000-22.13/100 000. The incidence rate of syphilis in women was higher than that in men. Adolescents infected with stage Ⅰ and stageⅡ syphilis accounted for 72.6%. There were 2 726 patients with gonorrhea, with an incidence rate about 1.29/100 000-10.74/100 000. The incidence rate of gonorrhea in men was higher than that in women. Joinpoint regression model analysis showed that the incidence of syphilis generally took 2012 as the inflection point. From 2006 to 2012, the average annual growth rate of syphilis incidence rate among adolescents was 67.30% (P<0.001). The average annual growth rate of syphilis incidence rate in adolescents from 2012 to 2020 was -0.02% (P=0.994).The overall incidence of gonorrhea incidence rate took 2015 as the inflection point. From 2006 to 2015, the average annual growth rate of juvenile gonorrhea incidence rate was 23.95% (P<0.001). The average annual growth rate of gonorrhea incidence rate in adolescents from 2015 to 2020 was 4.06% (P=0.492). Overall, from 2006 to 2020, the incidence rate of syphilis and gonorrhea among 10-19 year old adolescents in Liaoning Province increased slowly. The primary and secondary prevention strategies were significantly effective in reducing the risk of sexually transmitted diseases.
Adolescent ; Adult ; Child ; Female ; Gonorrhea/prevention & control* ; Humans ; Incidence ; Male ; Regression Analysis ; Sexually Transmitted Diseases/epidemiology* ; Syphilis/epidemiology* ; Young Adult

BACKGROUND: Ambient temperature may contribute to seasonality of mortality; in particular, a warming climate is likely to influence the seasonality of mortality. However, few studies have investigated seasonality of mortality under a warming climate. METHODS: Daily mean temperature, daily counts for all-cause, circulatory, and respiratory mortality, and annual data on prefecture-specific characteristics were collected for 47 prefectures in Japan between 1972 and 2015. A quasi-Poisson regression model was used to assess the seasonal variation of mortality with a focus on its amplitude, which was quantified as the ratio of mortality estimates between the peak and trough days (peak-to-trough ratio (PTR)). We quantified the contribution of temperature to seasonality by comparing PTR before and after temperature adjustment. Associations between annual mean temperature and annual estimates of the temperature-unadjusted PTR were examined using multilevel multivariate meta-regression models controlling for prefecture-specific characteristics. RESULTS: The temperature-unadjusted PTRs for all-cause, circulatory, and respiratory mortality were 1.28 (95% confidence interval (CI): 1.27-1.30), 1.53 (95% CI: 1.50-1.55), and 1.46 (95% CI: 1.44-1.48), respectively; adjusting for temperature reduced these PTRs to 1.08 (95% CI: 1.08-1.10), 1.10 (95% CI: 1.08-1.11), and 1.35 (95% CI: 1.32-1.39), respectively. During the period of rising temperature (1.3 °C on average), decreases in the temperature-unadjusted PTRs were observed for all mortality causes except circulatory mortality. For each 1 °C increase in annual mean temperature, the temperature-unadjusted PTR for all-cause, circulatory, and respiratory mortality decreased by 0.98% (95% CI: 0.54-1.42), 1.39% (95% CI: 0.82-1.97), and 0.13% (95% CI: - 1.24 to 1.48), respectively. CONCLUSION Seasonality of mortality is driven partly by temperature, and its amplitude may be decreasing under a warming climate.
Cardiovascular Diseases/mortality* ; Cause of Death ; Climate Change/mortality* ; Cold Temperature/adverse effects* ; Hot Temperature/adverse effects* ; Humans ; Japan/epidemiology* ; Mortality/trends* ; Regression Analysis ; Respiratory Tract Diseases/mortality* ; Seasons ; Time