Semen parameter thresholds and time-to-conception in subfertile couples: how high is high enough?

Abstract

STUDY QUESTION

What thresholds for total sperm count, sperm concentration, progressive motility, and total progressive motile sperm count (TPMC) are associated with earlier time-to-conception in couples undergoing fertility evaluation?

SUMMARY ANSWER

Values well above the World Health Organization (WHO) references for total sperm count, concentration, and progressive motility, and values up to 100 million for TPMC were consistently associated with earlier time-to-conception and higher conception rates.

WHAT IS KNOWN ALREADY

Although individual semen parameters are generally not able to distinguish between fertile and infertile men, they can provide clinically useful information on time-to-pregnancy for counseling patients seeking fertility treatment. Compared to the conventional semen parameters, TPMC might be a better index for evaluating the severity of male infertility.

STUDY DESIGN, SIZE, DURATION

We used data from a longitudinal cohort study on subfertile men from 2002 to 2017 and included 6061 men with initial semen analysis (SA) in the study.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Men from subfertile couples who underwent a SA within the study period were included, and 5-year follow-up data were collected to capture conception data. Couples were further categorized into two subgroups: natural conception (n = 5126), after separating those who achieved conception using ART or IUI; natural conception without major female factor (n = 3753), after separating those with severe female factor infertility diagnoses. TPMC was calculated by multiplying the semen volume (ml) by sperm concentration (million/ml) and the percentage of progressively motile sperm (%). Cox proportional hazard models were used to report hazard ratios (HRs) with 95% CIs before and after adjusting for male age, the number of previous children before the first SA, and income. Using the regression tree method, we calculated thresholds for total sperm count, sperm concentration, progressive motility, and TPMC to best differentiate those who were more likely to conceive within 5 years after first SA from those less likely to conceive. We also plotted continuous values of semen parameters in predicting 5-year conception rates and time-to-conception.

MAIN RESULTS AND THE ROLE OF CHANCE

Overall, the median time to conception was 22 months (95% CI: 21–23). A total of 3957 (65%) couples were known to have achieved conception within 5 years of the first SA. These patients were younger and had higher values of sperm concentration, progressive motility, and TPMC. In the overall cohort, a TPMC of 50 million best differentiated men who were more likely to father a child within 5 years. Partners of men with TPMC ≥50 million had a 45% greater chance of conception within 5 years in the adjusted model (HR: 1.45; 95% CI: 1.34–1.58) and achieved pregnancy earlier compared to those men with TPMC < 50 million (median 19 months (95% CI: 18–20) versus 36 months (95% CI: 32–41)). Similar results were observed in the natural conception cohort. For the natural conception cohort without major female factor, the TPMC cut-off was 20 million. In the visual assessment of the graphs for the continuous semen parameter values, 5-year conception rates and time-to-conception consistently plateaued at higher values of sperm concentration, total sperm count, progressive motility, and TPMC compared to the WHO reference levels and our calculated thresholds. For TPMC, values up to 100–150 million were still associated with a better conception rate and time-to-conception in the visual assessment of the curves.

LIMITATIONS, REASONS FOR CAUTION

There was limited information on female partners and potential for inaccuracies in capturing less severe female infertility diagnoses. Also we lacked details on assisted pregnancies achieved outside of our healthcare network (with possible miscoding as ‘natural conception’ in our cohort). We only used the initial SA and sperm morphology, another potentially important parameter, was not included in the analyses. We had no information on continuity of pregnancy attempts/intention, which could affect the time-to-conception data. Finally, most couples had been attempting conception for >12 months prior to initiating fertility treatment, so it is likely that we are underestimating time to conception. Importantly, our data might lack the generalizability to other populations.

WIDER IMPLICATIONS OF THE FINDINGS

Our results suggest that a TPMC threshold of 50 million sperm provided the best predictive power to estimate earlier time-to-conception in couples evaluated for male factor infertility. Higher values of sperm count, concentration and progressive motility beyond the WHO references were still associated with better conception rates and time-to-conception. This provides an opportunity to optimize semen parameters in those with semen values that are low but not abnormal according to the WHO reference values. These data can be used to better inform patients regarding their chances of conception per year when SA results are used for patient counseling.

STUDY FUNDING/COMPETING INTEREST(S)

None.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

Semen analysis (SA) is usually the first and often the only test used to evaluate the reproductive function of a male partner in couples seeking infertility counseling (Practice Committee of the American Society for Reproductive Medicine, 2015). The World Health Organization (WHO) recommendations are commonly used to interpret SA results (World Health Organization, 2010). To define the lower threshold for semen parameters, the most recent WHO-5 criteria use the 5th percentile of semen values derived from a population of men whose partners conceived a child within 12 months of stopping contraceptives (World Health Organization, 2010). These thresholds, however, do not strongly correlate with fertility outcomes and cannot be reliably used to differentiate fertile from infertile men (Nagler, 2011). Moreover, several studies have shown an increased probability of pregnancy with semen parameter values well above the WHO reference levels (Guzick et al., 2001; Slama et al., 2002; Herrera et al., 2020).

While the overall ability of semen parameters to differentiate fertile from infertile men remains limited (Polansky and Lamb, 1988; Guzick et al., 2001; van der Steeg et al., 2011), specific semen parameters, individually or in combination, might provide information on male factor severity and help in predicting time-to-conception (Ayala et al., 1996; Slama et al., 2002; Buck Louis et al., 2014; Barratt et al., 2017). Infertility is commonly defined as the inability to conceive after 1 year of unprotected intercourse (Practice Committee of the American Society for Reproductive Medicine, 2015). However, more than half of couples undergoing an infertility evaluation will achieve a successful pregnancy within 2–3 years either spontaneously or with the help of therapies such as IUI or IVF (Ayala et al., 1996; Jouannet et al., 1988; Buck Louis et al., 2014). Thus, longitudinal studies with adequate follow-up time are needed to assess the relations between semen parameters and time-to-conception, rather than conception success. Time to conception is a more clinically meaningful parameter for counseling patients seeking fertility treatment.

The concept of total motile sperm count (TMC), or more accurately total progressive motile sperm count (TPMC), was first introduced in the 1970s as a better way to assess the severity of male factor infertility by combining information on semen volume, sperm concentration, and progressive motility (Smith et al., 1977; Borges et al., 2016). Smith et al. (1977) were among the first groups to investigate the relation between TPMC and pregnancy rates in a small group of infertile couples and suggested that male partners with lower TPMC contributed to lower couple pregnancy rates. Further studies reinforced the idea that TPMC has a better correlation with natural pregnancy rates than other individual semen parameters (Small et al., 1987; Ayala et al., 1996; Hamilton et al., 2015). TPMC is also correlated with IUI success and might be useful in selecting patients for IUI over IVF (Brasch et al., 1994; Huang et al., 1996; Borges et al., 2016; Madbouly et al., 2017), although the results are still debated (Ombelet et al., 2014; Mankus et al., 2019; Zanetti et al., 2019). In 2015, Hamilton et al. (2015) compared different ranges of TPMC values in predicting natural pregnancy in couples with unexplained or male factor infertility. In the absence of a validated classification, the authors used arbitrary thresholds to define five TPMC groups and considered a TPMC >20 million as normal, according to the Dutch national fertility guidelines at the time. They recommended three prognostic groups, namely TPMC <5 million, 5–20 million, and >20 million (Hamilton et al., 2015). As a result, TPMC is now more widely used to assess male fertility, with a threshold of 20 million sperm commonly referred to as ‘normal’ in both research studies (Borges et al., 2016; Hajder et al., 2016; Blickenstorfer et al., 2019) and clinical practice (Michigan Medicine, Pacific Fertility Center, Utah Center for Reproductive Medicine). However, the methodology and the evidence underpinning these thresholds remain unclear and population-based data are lacking to define the optimal TPMC threshold for fertility prediction.

We hypothesized that higher values of semen parameters are associated with higher conception rates and earlier time-to-conception in men undergoing fertility evaluation. We examined the associations between individual semen parameters and time-to-conception within 5 years of obtaining a semen sample and aimed to define the ranges and thresholds best associated with time-to-conception in a population-based setting.

Materials and methods

Data

We integrated data from the Subfertility Health and Assisted Reproduction and the Environment (SHARE) study with those from the Utah Population Database (UPDB) to define the cohort of men undergoing male fertility evaluation. SHARE combines medical, genealogic, and administrative data with biospecimen data to create a unique resource that currently includes the University of Utah and Intermountain Healthcare's SA results collected from 1996 to 2017. The University of Utah and the Intermountain Healthcare provide healthcare to approximately 85% of all individuals in Utah. Each individual’s information from SHARE was linked to the UPDB to integrate demographic and longitudinal follow-up information, including the number of children before and after SA, and to link the information to their female partners. Median income was estimated based on zip code data. The UPDB is a statewide database that includes approximately 85% of medical records for all Utah residents and links information across several data sources to incorporate medical and vital records, family history data, cancer diagnoses, and comprehensive pedigree data. The full UPDB contains data on over 11 million individuals owing to the longstanding efforts to update records as they become available, including statewide birth and death certificates, hospitalizations, ambulatory surgeries, and driver licenses. Studies using UPDB data have been approved by the University of Utah Resource for Genetic and Epidemiologic Research and its institutional review board.

All men aged ≥ 18 years presenting to the University of Utah fertility clinics in 2002–2017 with semen samples were selected for the study. Patients were excluded if the semen sample was marked as azoospermic; TPMC values were missing; Utah was not recorded as the state of residence at the time of SA; pregnancy was achieved but conception/birth date was missing; and the first SA date was later than the conception date. Pre- and post-vasectomy and sperm preservation samples were not part of this database. Male BMI data were gathered from different sources including clinic visits and driver license and were missing/not linked for 52% of the patients and thus BMI was not included in the analyses. Census data were used to estimate the median household income based on the patients’ zip codes from the address closest or prior to the date of the SA. Spouses were linked to index male partners using the information through different sources including infertility clinic visits and UPDB linkage records. Events such as marriage, divorce, and death are all captured in UPDB allowing a determination of change in partners or termination of a relationship when officially recorded (DuVall et al., 2012).

SA data

SA were performed following the 4th (1999) or 5th (2010) edition of the WHO manual for examination and processing of human semen at the time of sample collection. All SA were performed in a central reference laboratory on samples obtained after 2–7 days of abstinence and were in accordance with the checklist published by Björndahl et al. (2016) (Björndahl et al., 2016). The continuous values for sperm count, concentration and motility were used to define reference values based upon the 5th edition of the WHO manual. Sperm morphology was not included in the analyses owing to the inconsistencies in data, changes in definitions between the WHO manuals, and changes in the methodology in reporting the morphology data in our laboratory over time. TPMC was calculated by multiplying the semen volume (ml) by sperm concentration (million/ml) and the percentage of progressively motile sperm (%).

If a man had more than one SA on record, only the first qualifying semen sample within the study period was used for the analyses. According to the WHO-5 definitions for reference values, a sperm concentration <15 million/ml was defined as oligozoospermia, 15–178 million/ml as normozoospermia, and >178 million/ml as hyperzoospermia (based on the 90th percentile of our data). A total sperm count of 39 million and progressive motility of 32% were considered the reference values based on the thresholds recommended by WHO-5. Because an evidence-based classification is missing for TPMC, we categorized it based on the derived deciles (D1–D10) in our data and also attempted to find the thresholds that provided the highest concordance index in our analyses. We also used the 20 million threshold and the 5-group categories previously used in the literature as <1 million, 1–5 million, 5–10 million, 10–20 million, and ≥20 million (Hamilton et al., 2015).

Study subgroups

The overall study cohort consisted of all men evaluated with a SA for male infertility in the SHARE database. A ‘natural conception’ subgroup was defined after separating those conceptions achieved via IUI or IVF. We also defined a subgroup within the natural conception cohort after excluding those with major female infertility diagnoses based upon the 9th or 10th revision of the International Classification of Diseases codes. Couples were categorized as having a major female factor if they had any of the following diagnoses during the study period: tubal factor including infertility of tubal origin, hydrosalpinx, and history of tubal ligation (codes: 614.1, N70.11, 628.2, N97.1, V26.51, Z31.0); uterine factor including infertility of uterine origin (codes: 628.3, N97.2); ovulation factor including anovulation, amenorrhea and polycystic ovary syndrome (codes: 628, N97.0, 256.4, E28.2, 626, N91.0, N91.1); severe endometriosis (codes: 617.9, N80.9); and age-related as female age >40 years at the time of SA. Couples without any of the above diagnoses who did not use IUI or IVF were categorized as the ‘natural conception without major female factor’ cohort.

Statistical analysis

Data are summarized as percentages for categorical variables and median (25th–75th interquartile range (IQR)) for continuous variables. Conception dates were reverse calculated by deducting a fixed 40 weeks from the birth date according to the birth certificates. Associations between each of the variables and conception outcome were assessed using a univariable Cox proportional hazard regression models predicting time to conception. Within a 5-year period after the first SA, patients were followed until achieving conception or a censoring event (death, divorce, re-marriage, moving outside of Utah, or being lost to follow-up).

Kaplan–Meier curves were used to describe the cumulative incidence of conception within 5 years of the first SA as a function of total sperm count, sperm concentration, progressive motility, and different TPMC categories (including deciles, thresholds from our analysis, and combinations of arbitrary categories including <1 million, 1–5 million, 5–10 million, 10–20 million, and ≥20 million). A log-rank test was used to compare the 5-year conception across the categories.

For each of the SA measures (concentration, progressive motility, total sperm count, and TPMC), a threshold (X) that best distinguishes those more likely from those less likely to conceive within 5 years was selected as the point that splits the data into SA<X and SA>=X that minimizes 10-fold cross-validated deviance assuming proportional hazard (LeBlanc and Crowley, 1992; Therneau et al., 2015). The selected cut-point was then rounded to the nearest integer, which was a multiple of 5. Conception rates at 5 years and median months to conception estimated (using the Kaplan Meier method for those with SA>=X and SA<X) were reported with 95% CIs and log-rank test P-values. Hazard ratios (HR) estimated using Cox proportional hazard models adjusting for age, income, and the number of children prior to SA were also reported with 95% CI and P-values, and the model Concordance (Harrell’s C). These measures of our selected cut-point X were presented alongside the same measures of the cut points suggested by the WHO. We then repeated the same cut point selection procedure in each of the three cohorts (see study subgroups above).

For each semen parameter and in each of the cohorts, we also fit a univariable Cox proportional hazard model with the semen parameter modeled as a restricted cubic spline of 3∼5 knots to allow for non-linear effects. Estimated 5-year conception rates and time-to-conception were plotted against continuous semen parameter values to better visualize the trends and to find approximate semen parameter values that corresponded to plateaus in conception rate and time-to-conception curves.

Finally, to determine which semen parameter is most predictive of 5-year conception, we built a recursive partitioning and regression trees allowing all four parameters (sperm count, concentration, progressive motility, and TPMC) to be included (Breiman et al., 1984). Variable importance for each semen parameter in the selected tree was estimated as the amount of decreased impurity with the inclusion of that parameter. Statistical analyses were conducted in R v.4.0 (RStudio, PBC, Boston, MA, USA) using two-tailed tests and a 0.05 significance level.

Results

Overall cohort

A total of 6061 adult men with SA data were included in the study. Mean male and female age at the time of SA were 32.6 (6.4) and 30.4 (5.5) years, respectively. The majority of men in our cohort (95%) self-identified as White/Caucasian, with only 2.6% Asian, 1% African-American, and <2% other/unknown. Overall, 3957 (65%) of couples had at least one live birth during the study period. Median time to conception was 22 months (95% CI: 21–23). Patient demographics and semen parameters are summarized in Table I. Patients who achieved live birth were younger and had higher values of total sperm count, sperm concentration, progressive motility, and TPMC. Median income and abstinence days were not different between the two groups.

Table I.

Baseline characteristics and semen parameters in the study cohort separated by successful conception within 5 years.

	Total	Conception = No	Conception = Yes	P-value
Overall cohort
Sample size	6061	2104 (35%)	3957 (65%)
Male age at 1st SA, y	32 (28–36)	33 (29–38)	31 (27–35)	<0.001
Female age at 1st SA, y	30 (26–34)	31 (27–36)	29 (26–33)	<0.001
Male BMI, kg/m2	25.7 (23.1–28.3)	25.8 (23.7–29.1)	25.1 (23.1–28.0)	<0.001
Median income, US dollars	49,807 (42 216–57 087)	49 807 (42 832–56 652)	49 807 (41 901–57 679)	0.47
Abstinence, days	4 (3–5)	4 (3–5)	4 (3–5)	0.49
Semen volume, ml	3 .0 (2.1–4.0)	3 .0 (2.0–4.0)	3 .0 (2.2–4.1)	<0.001
Total sperm count, million	223 (96–404)	190 (69–371)	240 (112–417)	<0.001
Sperm concentration, million/ml	76 (37–131)	69 (29–123)	81 (42–136)	<0.001
Progressive motility, %	51 (36–63)	47 (30–61)	53 (39–64)	<0.001
TPMC, million	111 (36–234)	85 (21–202)	123 (47–241)	<0.001
Natural conception cohort
Sample size	5126	2077 (41%)	3049 (59%)
Male age at 1st SA, y	31 (28–36)	33 (29–38)	30 (27–34)	<0.001
Female age at 1st SA, y	29 (26–34)	31 (27–36)	29 (26–32)	<0.001
Male BMI, kg/m2	25.7 (23.1–28.5)	25.8 (23.7–29.2)	25.1 (23.0–28.1)	<0.001
Median income, US dollars	49 807 (41 901–56 937)	49 807 (42 831–56 557)	49 807 (41 901–56 937)	0.91
Abstinence, days	4 (3–5)	4 (3–5)	4 (3–5)	0.43
Semen volume, ml	3 .0 (2.1–4.0)	3 .0 (2.0–4.0)	3 .0 (2.2–4.1)	<0.001
Total sperm count, million	229 (97–407)	190 (69–370)	250 (120–432)	<0.001
Sperm concentration, million/ml	77 (38–132)	68 (29–122)	84 (46–138)	<0.001
Progressive motility, %	51 (36–63)	47 (30–61)	54 (40–64)	<0.001
TPMC, million	113 (37–229)	85 (21–201)	132 (54–250)	<0.001
Natural conception without major female factor
Sample size	3753	1420 (38%)	2333 (62%)
Male age at 1st SA, y	31 (27–35)	32 (28–37)	30 (27–34)	<0.001
Female age at 1st SA, y	29 (26–33)	30 (27–34)	28 (25–32)	<0.001
Male BMI, kg/m2	25.6 (23.1–28.3)	25.8 (23.7–29.0)	25.1 (23.0–28.0)	<0.001
Median income, US dollars	49 792 (41 901–56 557)	49 792 (41 901–56 557)	49 663 (41 901–56 937)	0.34
Abstinence, days	4 (3–5)	4 (3–5)	4 (3–5)	0.38
Semen volume, ml	3 .0 (2.1–4.1)	3 .0 (2.0–4.0)	3 .1 (2.2–4.2)	<0.001
Total sperm count, million	221 (92–403)	178 (62–356)	245 (113–425)	<0.001
Sperm concentration, million/ml	74 (36–130)	64 (25–117)	82 (43–136)	<0.001
Progressive motility, %	51 (35–63)	46 (28–61)	53 (40–64)	<0.001
TPMC, million	110 (34–224)	75 (17–188)	129 (50–247)	<0.001

All values are presented as percentages or median (25th- 75th interquartile ranges). Chi-square test or Wilcoxon rank-sum test were used to compare values between the conception groups.

SA, semen analysis; TPMC, total progressive motile count.

Overall, 89% of the men had a sperm concentration of ≥15 million/ml. Men with sperm concentration ≥15 million/ml had a 49% greater chance of having a child within 5 years of their first SA when compared to oligozoospermic men, after adjusting for male age, median income, and the number of children before SA (HR: 1.49, 95% CI: 1.32–1.68) (Table II). There was no difference in the likelihood of fathering a child between hyperzoospermic and normozoospermic men (data not shown). Eighty percent of men had progressive sperm motility of ≥32%. The chances of fathering a child were 48% greater for men with progressive sperm motility ≥32% (Table II). Eighty-eight percent of men had total sperm count of ≥39 million and the chances of fathering a child were 52% higher compared to those with total sperm count of <39 million.

Table II.

Summary of semen parameters, 5-year conception rate, and hazard ratios for 5-year conception within different categories of semen parameters based upon WHO cut-off values and our calculated thresholds (overall cohort, n = 6061).

Semen parameters	Cut points	N (%)	Conception rate (%, 95% CI)	Months to conception1	Hazard ratio2	Concordance2
				(Median, 95% CI)	(95% CI)
WHO cut-offs
Concentration						0.565
	<15 M/ml	687 (11%)	56.5 (52.7, 60.5)	39.0 (30.4, 46.9)	1.00 (Ref)
	≥15 M/ml	5369 (89%)	68.7 (67.4, 70.0)	20.8 (19.6, 22.1)	1.49 (1.32,1.68)
Progressive motility						0.572
	<32%	1227 (20%)	56.4 (53.6, 59.4)	37.9 (33.2, 46.3)	1.00 (Ref)
	≥32%	4834 (80%)	70.0 (68.7, 71.4)	19.6 (18.3, 20.8)	1.48 (1.35,1.63)
Total sperm count						0.568
	<39 M	741 (12%)	54.6 (51.0, 58.5)	43.4 (35.5, 53.3)	1.00 (Ref)
	≥39 M	5320 (88%)	69.0 (67.7, 70.4)	20.3 (19.3, 21.5)	1.52 (1.35,1.71)
Calculated cut-offs
Concentration						0.567
	<35 M/ml	1388 (23%)	58.3 (55.7, 61.1)	33.8 (29.6, 39.4)	1.00 (Ref)
	≥35 M/ml	4668 (77%)	70.0 (68.6, 71.4)	19.9 (18.6, 21.1)	1.38 (1.26,1.50)
Progressive motility						0.572
	<30%	1112 (18%)	54.9 (51.9, 58.1)	42.9 (34.6, 53.2)	1.00 (Ref)
	≥30%	4949 (82%)	70.0 (68.7, 71.4)	19.6 (18.3, 20.7)	1.54 (1.39,1.70)
Total sperm count						0.574
	<95 M	1499 (25%)	57.8 (55.2, 60.5)	36.4 (31.5, 41.4)	1.00 (Ref)
	≥95 M	4562 (75%)	70.4 (69.0, 71.8)	19.4 (18.0, 20.4)	1.46 (1.34,1.59)
TPMC						0.575
	<50 M	1812 (30%)	57.9 (55.6, 60.4)	36.1 (31.6, 40.7)	1.00 (Ref)
	≥50 M	4249 (70%)	71.3 (69.9, 72.7)	18.9 (17.6, 19.9)	1.45 (1.34,1.58)

All P-values for comparing months to conception as well as hazard ratios between the cut-points are statistically significant at <0.001 for each of the semen parameters.

¹Conception rate at 60 months and time-to-conception using Kaplan–Meier estimates.

²Adjusted for male age, median income, and number of children prior to first SA.

WHO, World Health Organization; TPMC, total progressive motile count.

Calculating best statistical thresholds in predicting 5-year conception, the following values were achieved: 35 million/ml for sperm concentration, 30% for progressive motility, 95 million for total sperm count, and 50 million for TPMC. Time to conception and HRs for each of these categories are presented in Table II. The overall concordance index for these thresholds was similar between different semen parameters and also between the calculated thresholds and the WHO reference values. Also, the concordance indices for dichotomous values within a given range were close. For example, for TPMC, dichotomizing the values in the range of 20–80 million provided close concordance indices, although 50 million provided the highest concordance index (data not shown).

The 5th, 50th (median), and 95th percentiles for TPMC were 1.2 (95% CI: 1.0–1.6), 110.9 (106.2–114.5), and 462.0 (446.8–483.7) million, respectively. Using TPMC alone, a threshold of 50 million sperm resulted in the best discrimination of men who were more likely to father a child within 5 years: those with a TPMC ≥50 million had a 45% greater chance within 5 years in the adjusted model (HR: 1.45; 95% CI: 1.34–1.58) (Table II) and achieved pregnancy earlier with their partners compared to those with TPMC < 50 million (median 19 months (95% CI: 18–20) versus 36 months (95% CI: 32–41)). As the 50 million threshold in our study was higher than the commonly used 20 million threshold, we also compared those with TPMC ≥50 million to TPMC 20–50 million and found 26% higher chances of conception in the former group (HR : 1.26, 95% CI: 1.12–1.41, adjusted model). This 50 million threshold roughly correlated with the lower bound of the 4th decile (i.e. 30% of included men had TPMC < 50 million). The conception rates in different TPMC deciles and the HRs for 5-year conception are provided in Supplementary Table SI. We also further categorized TPMC in 10 groups (<1, 1–5, 5–10, 10–20, 20–50, 50–100, 100–150, 150–200, 200–250, ≥250) to evaluate the possible stepwise increase in 5-year conception rates for thresholds of interest. The Kaplan–Meier curves for these categories are presented in Fig. 1, and the HRs are provided in Supplementary Table SI.

Figure 1.

Kaplan–Meier curves comparing 5-year conception probability within 10 different categories of total progressive motile sperm count. (A) Overall cohort; (B) natural conception cohort; (C) natural conception cohort without major female factor. Categories are total progressive motile sperm count (TMPC) <1, 1–4, 5–9, 10–19, 20–49, 50–99, 100–149, 150–199, 200–249, and ≥250 million from bottom to top. Solid lines mark categories with TPMC < 50 million and dashed lines are categories with TPMC ≥ 50 million. Solid blue line is TPMC 20–49 and the dashed red line is TPMC 50–99.

A graphical representation of estimated 5-year conception rates and time-to-conception for continuous values of different semen parameters and TPMC are provided in Fig. 2A–D. Conception rates and time-to-conception consistently reached a relative plateau at values well beyond the WHO reference values and our selected thresholds. For example, for total sperm count, the plateau was reached closer to 200 million values (Fig. 2B) in comparison to the 95 million threshold from our statistical model and the 39 million reference value from the WHO manual. For TPMC, the plateau was achieved closer to values above 100 million compared to the 50 million threshold from our statistical model (Fig. 2D).

Figure 2.

Estimated conception rates and time-to-conception in the overall cohort from the Cox regression models for individual semen parameters. (A) Sperm concentration; (B) total sperm count; (C) progressive motility; (D) total progressive motile sperm count (TPMC). Solid lines and shaded areas correspond to point estimates and their 95% CIs, respectively (conception rates: green, time-to-conception: purple).

When all the four semen parameters were used to predict time-to-conception in a setting of recursive partitioning and regression tree, TPMC was of the highest variable importance and used to determine the first split of the tree. Adding other semen parameters to the tree did not improve the prediction accuracy much, as measured by the decrease in node impurity, compared to using TPMC alone (data/regression trees not shown).

Natural conception subgroup

A total of 935 couples achieved pregnancy with the help of IUI or ART. In the natural conception subgroup (n = 5126), the conception rate was 59.5%, and the median time to conception was 27 (26–30) months.

In subgroup analyses of the natural conception cohort and based upon the WHO thresholds, the chances of conception were 71% greater for men with sperm concentration ≥15 million/ml (HR: 1.71, 95% CI: 1.48–1.98), 65% greater for men with progressive motility ≥32% (HR: 1.65, 95% CI: 1.47–1.84), and 70% greater for those with total sperm count ≥39 million (HR: 1.70, 95% CI: 1.47–1.95) (Table III). Calculated thresholds for better predicting 5-year conception were 35 million/ml for concentration, 40% for progressive motility, and 95 million for total sperm count. The optimal TPMC threshold for achieving an earlier pregnancy was 55 million in this subgroup. Men with TPMC ≥55 million had a 55% greater chance of conception within 5 years (HR: 1.55; 95% CI: 1.41–1.70) in the adjusted model.

Table III.

Summary of semen parameters, 5-year conception rate, and hazard ratios for 5-year conception within different categories of semen parameters based upon WHO cut-off values and our calculated thresholds (natural conception cohort, n = 5126).

Semen parameters	Cut points	N (%)	Conception rate1 (%, 95% CI)	Months to conception	Hazard ratio2	Concordance2
				(Median, 95% CI)1	(95% CI)
WHO cut-offs
Concentration						0.586
	<15 M/ml	559 (11%)	46.6 (42.4, 51.1)	NA (53.2, NA)	1.00 (Ref)
	≥15 M/ml	4562 (89%)	63.3 (61.8, 64.8)	25.3 (23.3, 27.0)	1.71 (1.48, 1.98)
Progressive motility						0.593
	<32%	1021 (20%)	47.8 (44.7, 51.2)	NA (57.2, NA)	1.00 (Ref)
	≥32%	4105 (80%)	64.8 (63.3, 66.4)	23.2 (21.6, 25.3)	1.65 (1.47, 1.84)
Total sperm count						0.589
	<39 M	620 (12%)	45.8 (41.7, 50.1)	NA (58.8, NA)	1.00 (Ref)
	≥39 M	4506 (88%)	63.6 (62.1, 65.1)	24.7 (22.8, 26.4)	1.70 (1.47, 1.95)
Calculated cut-offs
Concentration						0.590
	<35 M/ml	1144 (22%)	49.6 (46.7, 52.8)	NA (48.1, NA)	1.00 (Ref)
	≥35 M/ml	3977 (78%)	64.8 (63.3, 66.5)	23.3 (21.6, 25.3)	1.54 (1.39, 1.71)
Progressive motility						0.592
	<40%	1509 (29%)	50.0 (47.4, 52.7)	NA (48.4, NA)	1.00 (Ref)
	≥40%	3617 (71%)	66.2 (64.6, 67.9)	22.1 (20.3, 23.8)	1.51 (1.38, 1.66)
Total sperm count						0.595
	<95 M	1254 (24%)	49.7 (46.8, 52.7)	NA (48.2, NA)	1.00 (Ref)
	≥95 M	3872 (76%)	65.2 (63.7, 66.9)	22.6 (21.0, 24.7)	1.59 (1.44, 1.76)
TPMC						0.595
	<55 M	1598 (31%)	50.6 (48.1, 53.3)	58.2 (47.2, NA)	1.00 (Ref)
	≥55 M	3528 (69%)	66.3 (64.7, 68.1)	21.6 (20.1, 23.3)	1.55 (1.41, 1.70)

All P-values for comparing months to conception as well as hazard ratios between the cut-points are statistically significant at <0.001 for each of the semen parameters.

¹Conception rate at 60 months and time-to-conception using Kaplan–Meier estimates. NA= not applicable. The median and 95% CIs can be fully calculated only if enough conception events occurred within 5 years in the specific cohort.

²Adjusted for male age, median income, and number of children prior to first SA.

TPMC, total progressive motile count.

A graphical representation of estimated 5-year conception rates and time-to-conception for continuous values of different semen parameters and TPMC is provided in Fig. 3A–D. Similar to the overall cohort, conception rates and time-to-conception consistently reached a relative plateau at values well beyond the WHO reference values and our selected thresholds.

Figure 3.

Estimated conception rates and time-to-conception in the natural conception cohort from the Cox regression models for individual semen parameters. (A) Sperm concentration; (B) total sperm count; (C) progressive motility; (D) total progressive motile sperm count (TPMC). Solid lines and shaded areas correspond to point estimates and their 95% CIs, respectively (conception rates: green, time-to-conception: purple).

Natural conception cohort without major female factor

A total of 1669 (27%) of female partners had one or more of the female infertility diagnoses, including tubal factor (n = 199, 12%), uterine factor (n = 154, 9%), ovulation factor (n = 1211, 72%), and age-related (n = 341, 20%) in the overall cohort (percentages are among those with a female factor with possible overlaps between the diagnoses). Of the 3753 couples without any of these diagnoses in the natural conception cohort (i.e. who did not use IVF or IUI), 62% achieved conception within 5 years of the first SA. The median time to conception was 23.5 (21.5–25.4) months.

In this subgroup, chances of conception were 78% greater for men with sperm concentration ≥15 million/ml (HR: 1.78, 95% CI: 1.52–2.08), 69% greater for men with progressive motility ≥ 32% (HR: 1.69, 95% CI: 1.50–1.91), and 80% greater for those with total sperm count ≥39 million (HR: 1.80, 95% CI: 1.54–2.11) (Table IV). Calculated thresholds for better predicting 5-year conception were 40 million/ml for concentration, 35% for progressive motility, and 95 million for total sperm count. The optimal TPMC threshold for achieving an earlier pregnancy was 20 million in this subgroup. Men with TPMC ≥20 million had an 80% greater chance of conception within 5 years (HR: 1.80; 95% CI: 1.58–2.06) in the adjusted model.

Table IV.

Summary of semen parameters, 5-year conception rate, and hazard ratios for 5-year conception within different categories of semen parameters based upon WHO cut-off values and our calculated thresholds (natural conception cohort without major female factors, n = 3753).

Semen parameters	Cut points	N (%)	Conception rate (%, 95% CI)1	Months to conception	Hazard ratio2	Concordance2
				(Median, 95% CI)1	(95% CI)
WHO cut-offs
Concentration						0.574
	<15 M/ml	455 (12%)	48.4 (43.7, 53.4)	NA (43.6, NA)	1.00 (Ref)
	≥15 M/ml	3294 (88%)	66.6 (64.9, 68.3)	21.1 (19.3, 22.9)	1.78 (1.52, 2.08)
Progressive motility						0.583
	<32%	794 (21%)	50.7 (47.1, 54.5)	58.7 (43.6, NA)	1.00 (Ref)
	≥32%	2959 (79%)	68.0 (66.3, 69.9)	19.3 (17.4, 21.2)	1.69 (1.50, 1.91)
Total sperm count						0.577
	<39 M	489 (13%)	47.8 (43.2, 52.7)	NA (48.2, NA)	1.00 (Ref)
	≥39 M	3264 (87%)	66.8 (65.1, 68.6)	20.6 (18.7, 22.4)	1.80 (1.54, 2.11)
Calculated cut-offs
Concentration						0.579
	<40 M/ml	1043 (28%)	53.2 (50.1, 56.6)	46.1 (37.8, 58.8)	1.00 (Ref)
	≥40 M/ml	2706 (72%)	68.7 (66.9, 70.6)	19.3 (17.3, 21.0)	1.55 (1.39, 1.72)
Progressive motility						0.582
	<35%	902 (24%)	51.7 (48.3, 55.3)	53.2 (40.4, NA)	1.00 (Ref)
	≥35%	2851 (76%)	68.3 (66.6, 70.2)	18.9 (17.3, 21.0)	1.63 (1.45, 1.83)
Total sperm count						0.586
	<95 M	966 (26%)	52.4 (49.1, 55.8)	48.7 (39.4, NA)	1.00 (Ref)
	≥95 M	2787 (74%)	68.5 (66.7, 70.4)	19.1 (17.3, 20.8)	1.66 (1.48, 1.86)
TPMC						0.584
	<20 M	709 (19%)	48.3 (44.5, 52.3)	NA (50.6, NA)	1.00 (Ref)
	≥20 M	3044 (81%)	68.1 (66.3, 69.9)	19.3 (17.5, 21.1)	1.80 (1.58, 2.06)

All P-values for comparing months to conception as well as hazard ratios between the cut-points are statistically significant at <0.001 for each of the semen parameters.

¹Conception rate at 60 months and time-to-conception using Kaplan-Meier estimates. NA= not applicable. The median and 95% CIs can be fully calculated only if enough conception events occurred within 5 years in the specific cohort.

²Adjusted for male age, median income, and number of children prior to first SA.

TPMC, total progressive motile count; SA, Semen analysis.

A graphical representation of estimated 5-year conception rates and time-to-conception for continuous values of different semen parameters and TPMC is provided in Fig. 4A–D. Conception rates and time-to-conception again reached relative plateaus at values well beyond the WHO reference values and our selected thresholds. For example, for TPMC, the plateau was achieved at values above 100–150 million compared to the 20 million threshold from our statistical model (Fig. 4D).

Figure 4.

Estimated conception rates and time-to-conception in the natural conception cohort without major female factor infertility from the Cox regression models for individual semen parameters. (A) Sperm concentration; (B) total sperm count; (C) progressive motility; (D) total progressive motile sperm count (TPMC). Solid lines and shaded areas correspond to point estimates and their 95% CIs, respectively (conception rates: green, time-to-conception: purple).

Discussion

In our study, 65% of couples undergoing male fertility assessment achieved conception within 5 years. Those with semen parameters above the WHO-5 reference values had greater chances for conception and earlier time-to-conception. Our calculated sperm concentration and total sperm count cut-offs for predicting 5-year conception were higher than the WHO reference values, although the overall concordance indices remained similar when values were dichotomized in a wide range of semen parameter values. For progressive motility, the calculated cut-offs were closer to the WHO reference value. For TPMC, thresholds of 50 and 55 million maximized the prediction of time-to-conception in the overall study cohort and those who achieved natural conception, respectively. In those without major female factor infertility who achieved conception naturally, the TPMC cut-off was 20 million. However, in visual assessment of the graphs for the continuous semen parameter values, 5-year conception rates and time-to-conception consistently plateaued at higher values of sperm concentration, total sperm count, progressive motility, and TPMC compared to the WHO reference levels and our selected thresholds. This is in line with findings from a large European study that showed sperm concentration up to 55 million and total sperm count values up to 145 million, which are well beyond the WHO reference levels, were associated with earlier time-to-pregnancy in fertile couples (Slama et al., 2002). Although our results and threshold values from a subfertile population are not directly comparable to those of Slama et al. (2002) from a fertile population, these findings highlight an opportunity to improve conception rates and time-to-conception for a large number of men with semen values higher than the WHO reference levels. This also applies to men with TPMCs mostly in the range of 20 to about 150 million at the initial screening. These men should be counseled for optimizing the male partner evaluation and treatment to improve sperm parameters if possible.

Individual semen parameters are considered poor predictors of male fertility (Nagler, 2011). A single SA cannot accurately predict future fertility as the correlation between semen parameters and conception probability is minimal (Small et al., 1987; Polansky and Lamb, 1988; Arumugam and Omar, 1992; Guzick et al., 2001; Borges, 2016). However, SA remains the cornerstone of male fertility evaluation and can provide valuable information regarding the severity of male factor infertility. In the present study, 65% of couples in the total study cohort and about 60% of couples in the natural conception cohort achieved pregnancy within 5 years of the initial male factor evaluation with SA. These rates are comparable to other longitudinal studies assessing conception rates in couples undergoing male infertility assessment (Ayala et al., 1996; Jouannet et al., 1988; Polansky and Lamb, 1988). Although not good predictors of overall fertility, multiple studies have shown that individual semen parameters may be helpful in estimating the length of time needed to achieve pregnancy (Ayala et al., 1996; Slama et al., 2002; Buck Louis et al., 2014). Our results align with these reports showing that higher sperm concentration, total sperm count, progressive motility, and TPMC are associated with shorter time-to-conception.

A TPMC threshold of 20 million is commonly used in practice as the lower limit for normal and to decide about further male fertility evaluation. This threshold is mostly adapted from the study by Hamilton et al. (2015) and some earlier reports exploring the utility of different TPMC thresholds for pregnancy success (Small et al., 1987; Brasch et al., 1994). In our study, we defined a TPMC threshold correlating with earlier time-to-conception. Consistent with the results of the study by Hamilton et al. (2015), we found that higher TPMC values are associated with better conception rates and that a 20 million cut-off can be used to predict higher chances of conception and earlier time-to-conception in couples without major female factors attempting natural conception. However, we found that the discrimination power for most thresholds in the range of 20–80 million is very close and the concordance indices will be similar for a wide range of TPMC thresholds. Additionally, when attempting to dichotomize TPMC values, a threshold of 50 million sperm provided optimal predictive ability in the initial screening of subfertile couples and when the presence of female factor infertility is unknown. In the visual assessment of our graphs, men with TPMC values up to 100–150 million (depending on the cohort of interest) still had higher conception rates and earlier time-to-conception before reaching a plateau. This suggests that a large group of men with TPMC >20 million might still benefit from male factor evaluation and optimization of semen quality if possible.

Using a regression tree analysis, we showed that when a single semen parameter is concerned, TPMC is superior to total sperm count, concentration, and progressive motility in distinguishing those more likely to conceive within 5 years as adding the other three did not improve the model fit. However, the concordance indices for different dichotomous categorizations of the included semen parameters were roughly similar and it does not appear that any of the semen parameters outperform the others in a clinically significant way to predict time-to-conception when single cut-offs and dichotomization are concerned. In the visual assessment of the graphs from the regression analyses, there seems to be a linear increase in conception rates and a decrease in time-to-conception as the semen parameter values increase up to a certain level. As expected, these ‘plateau points’ were consistently higher than the WHO reference values and our selected thresholds. These approximate values might better answer the question of ‘beyond what value of sperm concentration or total sperm count or TPMC, is it unlikely to see any further benefits in terms of conception rate or time to conception?’ We believe these curves provide important and more helpful information over the dichotomous thresholds when counseling infertile couples.

A single TPMC threshold cannot be used to differentiate fertile from infertile men, given that even in lower TPMC groups about one-third to one-half of the couples achieved natural conception within 5 years. This finding reinforces the idea that semen parameters should not be used to define male infertility per se but rather to guide treatment options and potentially estimate the time needed to achieve pregnancy. Further, they call into question the standard threshold of 20 million for TPMC being used as the ‘normal’ threshold. Even in couples without major female factor who attempted natural conception, better conception rates and shorter time-to-conception were observed for TPMC values up to 100–150 million before reaching a plateau (Fig. 4D). Thus, it may be reasonable to consider higher threshold values if a single TPMC value is to be used before excluding male factor as a contributing factor. Doing so would likely result in significantly more men being evaluated for male infertility by a reproductive urologist and may result in earlier and more cost-effective treatments for these men.

Several factors should be considered when choosing a TPMC threshold. First, caution should be used in interpreting the thresholds from earlier studies, in which the formula for TPMC calculation is not clearly indicated (Smith et al., 1977; Small et al., 1987; Ayala et al., 1996). It is unclear if these studies used total motility (i.e. progressive plus non-progressive motility) or progressive motility (i.e. motility a + b per WHO-4) in their calculations. Of note, the current WHO-5 criteria define motility as progressive, non-progressive, and immotile, replacing the a, b, c, d grades in the previous version (World Health Organization, 2010). Second, distinction should also be made between pre-wash and post-wash TPMC values and the outcomes of interest in different studies. For example, Brasch et al. (1994) reported that a post-wash TPMC of >20 million was associated with higher conception rates after IUI, while the 20 million threshold suggested by Hamilton et al. (2015) refers to pre-wash samples for predicting spontaneous pregnancy. Notably, the thresholds adopted from IUI and IVF studies cannot be extrapolated to natural conception and vice versa. Lastly, the WHO-5 lower normal limits are based on the 5th percentile of semen parameters in a population of men with proven fertility. As TPMC thresholds are not derived from a normal fertile male population, they should not be used in the same sense as the WHO thresholds. The 5th percentile of the TPMC data is not reported in the WHO-5 manual; thus, direct comparisons between these TPMC thresholds and the WHO thresholds for concentration and progressive motility are not appropriate.

This study has some strengths and limitations that are worth discussing. In our database, details regarding female partner work-up and treatments were not always available, except when they underwent in-patient procedures or received IUI or IVF treatments. Thus, female infertility factors are likely underreported in this study. In theory, the distribution or severity of female factors might not be similar for men across different categories of semen parameters. We used the available information to separate those with more severe female factor infertility. However, we acknowledge that this distinction might not be accurate or capture all the female infertility diagnoses with possible cross-contamination in the subgroups. Among those with female infertility diagnoses, a disproportionally high percentage were coded as ‘ovulation factor’; we believe this could be due to inaccurate or broader use of these codes by the providers in relation to overall fertility evaluation. Nevertheless, conception success is generally high in most couples with these diagnoses (either with or without treatment). Thus, we decided not to exclude these couples from the study but evaluate them in a subgroup analysis.

Conception intent was assumed for couples seeking fertility evaluation, however, we did not have the information on continuity of pregnancy attempts/intention during the study period and after the initial evaluation. Although unlikely, we cannot exclude the possibility of non-random distribution of pregnancy attempts in relation to semen parameters, which could affect the time-to-conception data. We used IUI and IVF records in our database, which do not include all the IVF/IUI procedures and the service provided by private practices in Utah or outside of the state. Although we believe it is safe to assume that most couples seen at our fertility clinics also received their treatments at affiliated facilities, it is likely that some used treatments at outside facilities and were miscoded as ‘natural conception’ in our cohort. We did not adjust for female age in our models owing to some missingness (7%) and also because we back-calculated this variable for some partners based on the birth certificate information. However, we instead used male age for adjustments, which has a strong correlation with female age in our dataset (correlation coefficient: 0.8). We only used the first semen sample in the analyses, as most patients in the database only had one sample available. The exclusion of other semen samples limits the ability to address intra-individual variability in SA results, resulting in potential misclassification of patients; however, this also eliminates post-treatment samples in those who might have received medical or surgical treatments for male infertility and reflect the real-world scenario where most men would expect counseling based on their initial SA results in the absence of multiple samples (Barratt et al., 2017). Additionally, patients with abnormal semen parameters tend to have more than one SA, which would bias studies if only patients with multiple samples are included. We were not able to incorporate morphology data in our analysis; however, some studies have shown that morphology might also be an important predictor for conception success although the results are debated (Guzick et al., 2001; Herrera et al., 2020). As we did not have information on duration of infertility and first attempt to conceive, we used time from the first SA to calculate time-to-conception. Most couples had been attempting conception for >12 months prior to initiating fertility treatment, so it is likely that we are underestimating time to conception. Importantly, our data are limited only to couples living in the state of Utah with a disproportionally high Caucasian population (95%) and our data might lack the generalizability for other populations. We were able to link birth certificates to individuals in our study using the UPDB, facilitating the acquisition of more accurate longitudinal data on conception rates and time-to-conception for each couple as a more robust method to assess the association between semen parameters and fertility outcomes. A large sample size enabled us to analyze subcategories of initial semen parameters, including different TPMC categories.

In conclusion, individual semen parameters are poor predictors of overall conception success. However, semen parameters can be used to estimate time-to-conception, as men with higher sperm concentration, total sperm count, progressive motility, and TPMC have greater chances for conception and shorter time-to-conception. For TPMC and when concerning all couples who underwent infertility assessment with a SA, a threshold of 50 million sperm provided the best predictive power to estimate earlier time-to-conception as a dichotomous threshold. However, improved conception rates and time-to-conception were observed for TPMC values up to 100–150 million. These findings call into question the standard threshold of 20 million commonly used as the lower limit of normal for men undergoing a fertility evaluation. These data can be used to better inform patients regarding their chances of conception when SA results are used for patient counseling and provide an opportunity to optimize the male partner evaluation and treatment to improve sperm parameters for a larger group of subfertile men when possible.

Supplementary data

Supplementary data are available at Human Reproduction online.

Data availability

Data cannot be shared publicly for ethical/privacy reasons. The privacy of individuals represented in UPDB records and confidentiality of the data is strictly protected. Research ethical approval for this study has been granted by the University of Utah's Institutional Review Board and the Utah Resource for Genetic and Epidemiological Research.

Authors’ roles

S.K., J.M.H., L.E.V., E.B.J.: study concept and design; J.M.H., H.A.H., A.P.P.: acquisition of data; C.Z., A.P.P., S.K.: data analysis; S.K., J.M.H., A.W.P.: interpretation of data; S.K., J.M.H.: drafting the article; L.E.V., E.B.J., C.Z., A.P.P., A.W.P., H.A.H.: support with important intellectual content; all authors: final approval.

Funding

None.

Conflict of interest

None declared.