Between January 2011 and January 2018 altogether 1188 patients visited an andrology clinic in Budapest, Hungary, for infertility reasons, providing altogether 1345 semen samples. Our center is a certified training center of the European Academy of Andrology and continuously takes part in various international quality control efforts. QuaDeGa and Gamete Expert Andrology Scheme are the running quality controls at our department.
This analysis used chart review of all these patients, which included the following variables: date of visit, date of birth, body measurements (height, weight, waist circumference, and hip circumference), and the results of semen analysis (sperm concentration, total sperm count, progressive sperm motility, and normal sperm morphology) according to current WHO criteria. Age was calculated by subtracting the date of birth from the date of visit.
Body measurements
Height was measured in centimeters with a standardized cloth tape measure. Patients were asked to remove shoes and stand erect with their shoulders relaxed and looking straight ahead. Weight was measured in kilograms using standardized digital scales. BMI was then calculated by dividing the weight in kilograms with the square of the height in meters (kg/m2). We measured the hip and waist circumference in centimeters according to the WHO 2011 guidelines by means of a constant 100 g tension providing tape [20]. Waist circumference was measured at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest. Hip circumference was measured around the widest portion of the buttocks, with the tape parallel to the floor. WHR was then calculated by dividing the waist circumference measurement with the hip measurement (W ÷ H).
Semen analysis
Semen analysis was performed according to the WHO Laboratory Manual for the Examination and Processing of Human Semen 5th edition (2010) [1]. The “Who Laboratory manual for the Examination and processing of human semen” 2010 edition declares that to achieve best results for a standard semen analysis, the sample should be collected after a minimum of 2 days and a maximum of 7 days of sexual abstinence [1]. All our patients adhered our prescribed 3–5 days of abstinence.
From the standardized assessment sperm concentration, progressive sperm motility and normal sperm morphology were selected for further analysis. Sperm concentration was measured in million/milliliters (M/mL) by hemocytometer technique with Neubauer improved cell counting chamber. Diff-Quik® stains were used to evaluate sperm progressive motility and normal morphology. The samples were assessed with 400x magnification on an Olympus CX21 microscope, and progressive motility and normal morphology are expressed as percentage of total cells. Computer-assisted sperm analysis (CASA Sperm Class Analyzer - Microptic Automatic Diagnostic System - Spain) was used on a Nikon Eclipse E200 microscope for the quality control of the data.
Data management and statistical analysis
The data were quality controlled for repeat visits, data entry errors, and influential outliers. Of the 1345 observations, 157 were removed because they were repeat visits, leaving a total of 1188 observations (equaling the first visits of all patients). Of these, one observation was removed because of missing values, and 18 were removed because of data entry errors. Then, loess local regression smooth curve fit plots were created with the proc. loess procedure in SAS V9.4. to visualize influential outliers. Based on this analysis, two observations were removed because they were influential outliers (for both patients, BMI = 54 and WHR = 1.0 with respective large waist and hip circumferences). This analysis also confirmed that the relationships between semen parameters and BMI/WHR are linear. We further assessed our data in order to remove patients with clinical varicocele, orchiditis, epididymitis, and vesiculitis, but the final cleaned dataset did not contain any patients with these conditions. Therefore, the final study sample comprised of 1169 patient observations (98.4% of all patients).
Patient semen parameters were compared by degree of obesity in the following groups: normal weight with BMI less than 25 (438 patients), overweight with BMI between 25 and 29.9 (510 patients) and obese with a BMI above 30 (221 patients), and WHR < = 0.9 (361 patients) and WHR > 0.9. Differences were evaluated with the Kruskal-Wallis non-parametric test.
To compare the regression slopes of the BMI vs. WHR models for each semen parameter, we standardized the BMI and WHR values to range from 0 to 1 using the proc. stdize procedure in SAS V9.4 with the method = range option. We chose this standardization method, because this strictly follows the original distribution of the original variable and makes the two different variables comparable. That means, if we plotted any parameter against either BMI or WHR, the plot with the original values would be identical to the plot with the standardized values, except for the value labels on the axis for BMI/WHR. As such, we included two x axes with our figures: one with the standardized values and one with the original values.
Univariate contingency tables to describe distribution and means procedures were conducted. After the removal of the influential outliers (as described above), the relationship between BMI and WHR, and semen parameters could be fitted as linear. Therefore, scatter plots were created with fitted linear regression lines in order to visualize the relationship between BMI and WHR, and semen parameters. Univariate liner regression models adjusted for age (which is strongly associated both with BMI and WHR, and most likely with indicators for infertility as well) were created to calculate the regression line slope coefficients and their 95% confidence intervals. For reference purposes, we are showing both non-standardized and standardized values for parameter estimates and their confidence intervals, standard errors, t values, p values. For each semen parameter (dependent variable), min/max line plots were created comparing the slope coefficients of BMI vs. WHR (independent variables) for that particular parameter. Statistically significant (p < 0.05) differences between the slopes were considered when the point estimate of the WHR slope coefficient fell outside of the 95% confidence interval of the BMI slope coefficient [21]. SAS V9.4 software (SAS Institute Inc. Cary, NC) was used for data management and analysis, and data visualization.