Data Collection and participants selection
The NHANES study is a nationally representative study of population in the United States, which is also a cross-sectional survey based on a national sample of non-institutionalized population in the USA. It is conducted by the U.S. National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC). The survey consist of three main parts. Initial screening of qualified participants through the questionnaire. Then, extensive interviews are conducted, including age, gender, race, medical history and health status. What is more, physical examination and clinical evaluation are performed in specially designed mobile examination centers (MECs). In the process of data acquisition, all interviewers have received the training plan and reached the required standards. NHANES started in 1999 and is an ongoing annual survey with data published every 2 years and made publicly available online. This study gained Institutional Review Board (IRB; project identification code protocol #2011-17) approval by the NCHS in line with the revised Helsinki Declaration[16]. Informed consent was provided by all study participants before the data collection and examination procedures. More NHANES data and information are available at https://www.cdc.gov/nchs/nhanes/index.htm.
Participants from the NHANES were included in this population-based cross-sectional research. There were 19,932 participants in the NHANES from two 2-year survey cycles: 2011-2012 and 2013-2014. We screen participants according to the exclusion criteria listed below: (1) subjects without cognitive performance score (n =16,997 ); (2) subjects without UFR data (n =166 ); (3) subjects with missing data for covariate (n=45 ). Eventually, 2,724 eligible individuals of the NHANES survey were included in our study (Figure 1).
The whole informed consents from each eligible participant were obtained after explaining the whole process of the research. All experimental methods were performed in accordance with the relevant guidelines and regulations of the CDC.
Measurement of Urinary Flow Rate
The UFR was measured by uroflowmetry (mL/min). The calculation formula of UFR is UFR=V/t, where V is the volume of the present urine sample and t is the time duration between the former urination and the present urine collection[17]. The participants had to record their last urination time before coming to the MECs. Then, at the centers, they would record the voiding time and volume of the urine sample and calculate the UFR for three times. The specimens were collected in different containers to guaranteeing enough data for various analyses. The composite UFR (mL/min) was measured by dividing the total urine volume collected by the total time covered by all collected voids[18].
Measurement of Cognitive Function
The following 3 cognitive function measurements designed to assess a wide range of neurocognitive function across a variety of demo graphic backgrounds were studied: the Digit Symbol Substitution Test (DSST), the Animal Fluency Test (AFT) and the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) immediate recall test. The three assessment methods of cognitive function score are detailed in the Supplementary File.
Covariates
Multivariate model contains variables that may confound the association between UFR and cognitive function. For covariates, continuous variables included age (year). Categorical variables included: sex (male; female), poverty income ratio (PIR)(<1; ≥1), body mass index (BMI)(<25; ≥25<30; ≥30), alcohol intake per week (0; 1; 2), marital status (married or living with partner; living alone) ,smoking status (never; former; current) and comorbidity index (0, 1, 2, 3 ,4 ,5). BMI values were calculated by dividing participants' weight (in kilograms) by their height (in square meters)[19]. Diabetes mellitus, congestive heart failure, coronary artery disease, chronic obstructive pulmonary disease (chronic bronchitis and/or emphysema) and hypertension, cancer consisted of comorbid conditions. The number of subjects reported conditions were then combined to generate an ordinal comorbidity index[20].
Statistical analysis
The statistical analysis was performed according to the CDC analytical reporting guidelines for complex NHANES data analysis (https://wwwn.cdc.gov/nchs/nhanes/tutorials/default.aspx). A sample weight was assigned to each person participating in NHANES. Therefore, we accounted for masked variance and used the proposed recommended weighting methodology. Continuous variables were expressed as mean ±standard deviation. Categorical variables were expressed in frequency or as a percentage. Weighted linear regression model (for continuous variables) or weighted chi-square test (for categorical variables) were used to calculate the differences among different UFR groups (tertiles). To investigate whether UFR is correlated with cognitive function in selected participants, our statistical analysis consisted of two main steps.
First, weighted multivariate logistic regression model were employed. We estimated three models: crude model, no covariates were adjusted; model I,only adjusted for gender, age and BMI data; in the final model (model II), model I + other covariates presented in Table 1 (i.e. PIR; marital status; comorbidity index; alcohol intake per week and smoking status).
Moreover, the subgroup analyses were then performed using weighted stratified logistic regression models to further determine the correlation between UFR and cognitive function. To ensure the robustness of data analysis, we did the sensitivity analysis.
All analyses were performed using the statistical software packages R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empower stats.com, X&Y Solutions, Inc., Boston, MA). All P values less than 0.05 (two-sided) were considered statistically significant.