Multicenter Study on Reproducibility of Sperm Morphology Assessments

Transcription

1 Archives of Andrology Journal of Reproductive Systems ISSN: (Print) (Online) Journal homepage: Multicenter Study on Reproducibility of Sperm Morphology Assessments W. Ombelet, E. Bosmans, M. Janssen, A. Cox, M. Maes, U. Punjabi, V. Blaton, J. Gunst, G. Haidl, E. Wouters, C. Spiessens, M. S. Bornman, E. Pienaar, R. Menkveld & C. J. Lombard To cite this article: W. Ombelet, E. Bosmans, M. Janssen, A. Cox, M. Maes, U. Punjabi, V. Blaton, J. Gunst, G. Haidl, E. Wouters, C. Spiessens, M. S. Bornman, E. Pienaar, R. Menkveld & C. J. Lombard (1998) Multicenter Study on Reproducibility of Sperm Morphology Assessments, Archives of Andrology, 41:2, , DOI: / To link to this article: Published online: 09 Jul Submit your article to this journal Article views: 39 View related articles Citing articles: 24 View citing articles Full Terms & Conditions of access and use can be found at

2 MULTICENTER STUDY ON REPRODUCIBILITY OF SPERM MORPHOLOGY ASSESSMENTS W. OMBELET E. BOSMANS M. JANSSEN A. COX M. MAES The Genk Institute for Fertility Technology, ZOL-Ziekenhuizen, Genk, Belgium U. PUNJABI Fertility Clinic, University Clinic Antwerp, Belgium V. BLATON J. GUNST IVF Laboratory, St. Janshospital, Brugge, Belgium G. HAIDL Klinik und Poliklinik fir Haut- und Geschlechtskrankheiten, Universitiits-Hautklinik, Bonn, Germany E. WOUTERS C. SPIESSENS Leuven University Fertility Centre, UZ Gasthuisberg, University of Leuven, Belgium M. S. BORNMAN E. PIENAAR Centre for Fertility Studies, University of Pretoria, Pretoria, South Africa R. MENKVELD Reproductive Biology Unit, Tygerberg University Hospital, Tygerberg, South Africa C. J. LOMBARD South African Medical Research Council, Tygerberg, South Africa Address correspondence to Dr. Willem Ombelet, Genk Institute for Fertility Technology, ZOL-Campus St. Jan, Schiepse Bos 6, B-3600 Genk. ARCHIVES OF ANDROLOGY 41: (1998) Copyright 8 I998 Taylor & Francis /98 $ OO 103

3 104 W. Ombelet et al. Sperm morphology has always been considered an important tool in evaluating a man s fertilizing potential. The objective of this multicentric study was to evaluate intra- and interindividual variability and between-laboratory variation using the same or different criteria of sperm morphology assessment. Semen samples were obtained from 20 males and 32 smears were made of all samples. Eighty coded smears (4 per patient) were sent to 8 laboratories for morphology assessment. The centers applied different classification systems (strict criteria, WHO 1987, DUsseldorf criteria) and participants were asked to analyze the 80 smears twice, with an interval of 1 week between each participant s two analyses. Intraclass correlations between repeats showed that sperm morphology can be assessed with acceptable within observer reproducibility. Expected increases in imprecision were observed up to coefficients of variation of >30% with decreasing morphology scores, regardless of the classification system used. Agreement in correct classification of samples as normal/abnormal was obtained in 80% of cases. Differences in reproducibility between slides may reflect an important source of heterogeneity due to smear preparation. These results emphasize the importance of external quality control systems to improve the value of sperm morphology assessments in the investigation of the male partner in a subfertile couple. Keywords human spermatozoa, laboratory standardization, sperm morphology Most clinicians believe in the importance of sperm morphology as a useful indicator of male fertility potential. Indeed, numerous reports emphasize the clinical relevance and predictive value of this single semen parameter, not only in vitro [5, 23, 28, 3 I], but also in vivo [29, 30, 34, 351. For a test to be useful, comparable and reliable results among and within individual observers have to be demonstrated. A high variability and inaccuracy in sperm morphology assessment was reported before, not only between different technicians, but also within evaluations by the same technician [2, 7, 8, 13, 15, 20, 27, 391. This variation can be explained by several factors, such as (1) different techniques of preparation, fixation, and staining of smears, (2) differences in the interpretation of protocols, (3) inherent problems due to the use of multiple classification systems [32, 331 and methods with a lack of consensus on normal morphology, and (4) the various levels of experience of the observers [2, 13, 25, 26, 27, 391. The source of variation should be kept as small as possible to increase the consistency of sperm morphology assessments. Comparable and reliable results among and within different technicians, working in the same laboratory and using the strict criteria, were reported in two studies [4, 261. Menkveld et al. [26] also claimed that the application of strict criteria for normality should result in a better predictive power and reproducibility compared to other criteria for sperm morphology assessment, i.e., the WHO criteria (1987) [36]. This multicenter study, including 8 centers with a recognized reputation in ART (assisted reproductive technologies), was organized to compare sperm morphology determinations in terms of analytical variability (intra- and interassay repeatability), between-laboratory reproducibility, and between method or criteria variation. MATERIALS AND METHODS Patients Semen samples were obtained from 20 males enrolled in an IVF program. Concentration, total count, and motility parameters were normal according to the WHO guidelines [36] for all specimens. The majority of those samples, 16 out of 20, originated from patients with poor sperm morphology. From each of these 20 samples, 32 smears were prepared, air dried, and

4 Reproducibility of Sperm Morphological Assessments 105 randomly coded by the organizing laboratory. All 8 participating centers received 80 coded smears, 4 slides per patient, in a complete blind fashion without any patient information. Fixation, staining of the smears, and subsequent determination of the morphology score (% normal forms observed within a series of 100 [Labs A, B, D, E, G, HI and 300 sperm cells [Labs C, F]) were performed according to the routine methods of each center. All centers reported to use Papanicolaou staining, as recommended by the WHO (1987) [36]. Prior to shipment all slides had been stored dry at room temperature for weeks to minimize effects due to mailing delays. All observers, with the exception of Lab E, were experienced (>5 years of experience). Concerning the classification system applied for sperm morphology assessment in the different laboratories: one center applied the WHO (1987) criteria [36] (Lab A), one the Dlisseldorf criteria [17, 191 (Lab C), while the remaining 6 centers used the strict Tygerberg criteria [23] (Labs B, D, E, F, G, H). Participants analyzed the 80 samples twice with 1-week interval. The slides were examined in a blind fashion on both occasions. Records of all morphology results, obtained on at least 100 sperm cells, were sent to the organizing center for statistical analysis. Statistical Analysis Morphology data obtained in each participating center were analyzed to construct box and whisker plots summarizing the distribution of all morphology measures per center in order to compare median values, upper and lower quartiles, and ranges for the pooled morphology data of the 80 slides. Box and whisker plots were then constructed per center on data sets obtained on the four separate series of 20 slides, each series composed of one slide per patient to provide a descriptive representation of the data obtained together with different types of variations observed. Each box and whisker plot covering one series of 20 slides (of 20 patients) provides information on biological variation among sperm morphology results as scored by each of the 8 participating centers. This variation is described by its median value, a box delineating the 25th to the 75th percentile and whiskers covering all the values that are not outliers. The first 4 sets may be compared among each other for analytical variation due to observer scoring as well as to smear preparation andor staining effects. The second set of 4 series of diagrams represent the distribution of repeated measurement values (evaluated 1 week later) for the same 4 sets of slides analyzed previously. Comparison of these sets among each other can reveal systematic bias if similar differences between series are observed in the first and second analysis. To assess the repeatability in the scoring of sperm morphology, coefficients of repeatability (CR) can be calculated from the standard deviations of the differences between the two measurements. These coefficients, which can also be analyzed according to Bland and Altman [3] and read from Bland and Altman plots, should certainly be a more precise approach to evaluate repeatability, but such data cannot be shown here due to the high number of figures resulting from the different combinations of repeats. We used another approach to measure the consistency in morphology measurements, namely intraclass correlations. To obtain data on the reproducibility in function of the magnitude of the morphology score (expressed as % normal sperm cells), precision profiles were constructed for results obtained by Lab A using WHO criteria, for Lab C using Dusseldorf criteria, and for Lab E, which determined sperm morphology according to strict criteria. The precision profile expressed as the CV (coefficient of variation) in function of the morphology index was obtained by a

5 106 W. Ombelet et al. logarithmic regression curve fit. Box and whisker plots for the CVs obtained in all the centers allow comparison of the reproducibility between centers and summarizes the distribution of data obtained on the same sample/patient series in different centers. A mixed-effect model, with patients considered as a random effect and slide and week/ occasion considered as fixed effects, was used to estimate the intraclass correlations between repeats within slides, between slidedweeks, and between repeats within patients. Furthermore, this model was used to estimate the following variance components for each center separately: the variance due to patients, the variance due to slides within patients, and the variance due to repeats within the slides. ANOVA was used to compare between slide repeats within patients as well as repeats within slides obtained through reanalyses after 1 week. Results were expressed as F-values and were joined to the box and whisker summaries per center as well. Interrater agreement on morphology estimation between centers was analyzed according to the method outlined by Cohen [6] using Kappa statistics. Kappa statistics were performed on the summary values for each patient and each center. K can vary between 1 and 0 with K = 1 if there is a perfect agreement between centers and K = 0 if there is no agreement better than chance. RESULTS Box and whisker plots shown in Figure 1 demonstrate not only the difference in the distribution of morphology values according to the applied criteria (WHO, DUsseldorf, strict), but also the differences in ranges between centers using the same (strict) criteria, illustrating differences in interpretation of normalcy. Boxes depict the 25th and the 75th percentiles of the morphology scores observed with indication of the median value. Whisker plots indicate the largest and smallest nonoutlier values. In Figure 2 we illustrated in one diagram per center the differences observed between slides and differences obtained between weeks. Every diagram in this figure has its own scale of % normal sperm morphology and those scales are quite different as demonstrated by the median values obtained for the total data set in each center (Figure 1): Lab A, 36%; Lab B, 2%; Lab T 80 I I ~ A WHO C Dusseld. E Strict 0 Strict B Strict D Strict F Strict H Strict... a E - Median Figure 1. Distribution of morphology assessment values obtained in 8 centers using different criteria: WHO (I), Diisseldorf (I), and strict criteria (6).

6 Reproducibility of Sperm Morphological Assessments 107 V h f e on morpbdogy : Dikwldd uhrb (Lab C) Vuhcc on morpbdw : Sirid uitfria (LAB D) 1 I I, P Vulance m morpbdogy : wid uitd (Lab C) Figure 2. Box and whisker plots per center representing distribution of data obtained on 4 series of 20 slides analyzed week 1 and reanalyzed week 2. Corresponding F values for differences observed between slides and between weeks are presented of the bottom of each graph.

7 108 W. Ombelet et al. C, 17%; Lab D, 9%; Lab E, 8%; Lab F, 2%; Lab G, 5%; and Lab H, 4%. Figure 2 summarizes the results obtained per center on the 4 within-patient repeats in the first 4 sets of box plots at the left side of the diagram. The results obtained after reanalyses of the same 4 series of slides have been plotted on the right side of the diagram. Data shown in the box and whisker plots are the median value, the 25th and the 75th percentiles of the morphology scores obtained for the boxes, and largest and smallest nonoutlier values for the whiskers. Reproducibility between series is measured by ANOVA for repeated measures and expressed as F values with their corresponding p values. In this approach, two factors were included, namely repeats within patients (1-4) and weeks (1, 2) for the data of each center. The F value obtained in the ANOVA to estimate consistency in morphology among slides obtained from the same patient (4 repeats) are shown below at the left; the corresponding F values for comparison of repeated measures in time on the same slides are shown at the right side of each diagram. In contrast to Figure 1, in which all observations per center are compiled, Figure 2 shows only data for each series of 20 slides. In five out of seven laboratories with sufficient data available for ANOVA analysis (Lab H was excluded since an important number of slides were broken during transport and not available for analysis), differences between blind repeated measures within patient on the 4 series of slides were significant 0) < 0.05). Surprisingly, in the repeated measures within slide (second analysis, 1-week interval), only one center out of seven (Lab C) showed a significant difference between first and second analysis of the same slides. Precision profiles shown in Figure 3(a-c) provide information on coefficients of variation that may be expected in sperm morphology assessments in function of the criteria used (WHO, Diisseldorf, and strict criteria). The?h CV varied between 17 and 62%, with ranges up to 85% for low counts of normal sperm morphology. A summary of precision data expressed in YD CV is shown per center as box and whisker plots in Figure 3d. These plots clearly demonstrate that even for centers using the same strict criteria, median CV values may vary between 30 and 100%. Intraclass correlation coefficients between repeated assessments on series of slides, calculated from a mixed-effect model (SAS statistical program), are presented in Table 1. Intraclass correlations are defined as the correlation between different measurements made on the same individual or slide. The correlation between slides and between repeated measurements in the same patient is generally lower than the correlation between measurements within the same slide. Even so, the intraclass correlation is dependent on the variance in each center and therefore provides a relative measure of consistency in morphology measurements. Only center E scored significant low using this approach. However, Table 2 shows a break down of the variance components found for the different centers. Variance between patients is given for each center and this variance will largely depend on the method used. The thud column expresses the variance due to repeats within patients or due to slide variation. The last column expresses the random variation, which should be seen as the variance due to repeats within the slides. For the WHO and Dtisseldorf criteria, the variance attributable to repeats within patients is more than twice the variance attributable to the variance due to repeats within the slides or measurement error. For strict criteria users, the results are a mixture, with contributions in variance due to repeats within slides from 2.5% to as high as 24.3% for a center starting with morphology scoring following strict criteria (Lab E). The variance attributable to within-patients repeats varies from 6.2 to 15.1 % over all centers, suggesting important variation between slides.

8 Reproducibility of Sperm Morphological Assessments a 10 2o 30 Y) 50 ea m o 10 a0 a Morphology index Maphology index xcv m W m I,a,... Table 3, showing the Kappa statistics between every pair of laboratories involved in the study, demonstrates that the overall correlation between results or the correlation between morphology interpretation scores obtained for the 20 patients in the different laboratories is rather poor. According to Altman [l], the K value can be interpreted as follows: <0.20, poor; , fair; , moderate; , good; and , very good. Good correlation of results exist between the pairs of laboratories A-D (0.89), C-G (0.86), IIG (0.74), and E-G (0.83). Described in absolute figures, complete agreement (100%) about the classification of samples as normal or abnormal for semen morphology was reached for 11 of the 20 patients analyzed (55%). DISCUSSION Routine semen analysis remains the most important laboratory practice undertaken in the investigation of the male partner in a subfertile couple. The usefulness of sperm morphology assessment as a predictor of a man's fertilizing capacity has often been challenged due to different classification systems, various slide preparation techniques, and inconsistency of analysis

9 ~ 110 W. Ombelet et al. Table 1. Intraclass correlation between repeats of sperm morphology assessments (slides series repeats and intraclass correlation between 4 slides, evaluated twice within I-week interval) Intraclass correlation between repeats Slides series repeats Center s1 s2 s3 54 Lab A Lab B Lab C Lab D Lab E Lab F Lab G Lab H lntraclass correlation between 4 slides lntraclass overall Center Week 1 Week 2 correlation (n = 8) Lab A Lab B Lab C Lab D Lab E Lab F Lab G Lab H Table 2. Variance components for the 8 centers Variance due Variance due to slides Variance due to repeats Center to patients within patients within slides (6.2%) 1.3 (6.4%) 21.4 (10%) 7.7 (15.1Yo) 2.6 (6.7%) 0.8 (10.1%) 4.0 (10.8%) 1.1 (9.2%) 8.2 (2.2%) 2.8 (13.8%) 11.3 (5.3%) 4.8 (9.4%) 9.4 (24.3%) 0.2 (2.5%) 3.4 (9.2%) 1.7 (14.3%) Note. Values in parentheses are the percentages of total variation in the data.

10 Reproducibility of Sperm Morphological Assessments 111 Table 3. Results of Kappa analysis for correlation between sperm morphology results obtained in 8 centers A WHO B Strict C DUsseld. D Strict E Strict F Strict G Strict A WHO B Strict C DUsseld. D Strict E Strict F Strict G Strict H Strict (0.8) (0.1) 0.27 (0.6) I 0.89 (0.005) 0.0 (0.9) 0.63 (0.08) (0.2) 0.0 (0.9) 0.69 (0.08) 0.58 (0.1) (0.6) 0.0 (0.9) 0.17 (0.7) 0.41 (0.3) 0.32 (0.6) (0.07) 0.0 (0.9) 0.86 (0.02) 0.74 (0.04) 0.83 (0.04) 0.23 (0.7) 1 NA NA NA NA NA NA NA Note. NA, not analyzed. within and between laboratories [32, 331. Freund [ 161 was the first to report on a collaborative study in which 47 participants evaluated photomicrographs of 500 spermatozoa. The disappointing result of this study revealed a chaotic situation, with almost as many descriptions of normality as there were observers involved. In 197 1, Eliasson [ 141 stressed the importance of performing accurate and precise evaluation of sperm morphology to obtain better reproducibility scores between and within laboratories. Freund and Eliasson [ 14, 161 strongly believed in the advantage of one classification system, allowing different centers to compare results. Our multicentric study showed that the evaluation of different slides originating from the same semen sample, resulted in dramatically different values for sperm morphology, not only when different morphology classification protocols were used, but also when the same criteria were applied. This between-laboratory variation in morphology assessments, regardless the use of different criteria, was previously reported [7, 24, 381. Our data showed significant differences between series of slides originating from the same semen sample in 5 out of 7 laboratories. On repeated analysis (interval: 1 week), a lack of significant difference was observed when the same slides were compared. These differences observed may be due to heterogeneity in smear preparation as well as reproducibility difficulties in morphology assessment. Furthermore, the nonblind repeated measures reflect the variability due to scoring of morphology by the observer, since the same slides were used for interpretation. Our results suggest that heterogeneity in smear preparation may result in significant differences in morphology scoring. This finding was never reported before and will surely be a major but difficult variable to correct if our results are confirmed by other investigators. Since most centers nowadays use classification systems with low cutoff values for normality (WHO [37], 30%; strict criteria, 14%; Diisseldorf criteria, 30%), it may be logical that more spermatozoa per slide need to be evaluated to overcome this problem of reproducibility, as suggested by Davis and Gravance [ 101 using computerized systems. Irrespective of the criteria used, we observed an increase in imprecision (higher coefficient of variation) with decreasing morphology scores. At the cutoff level for normality, all centers obtained an acceptable CV (<30%), but for morphology scores toward zero, imprecision, as defined in our study, is dependent on the magnitude of the range of values and therefore becomes unacceptably high, close to the natural boundary 0%. This effect was seen independent of criteria use. However, good overall CV profiles for all criteria

11 112 W. Ombelet et al. were seen for high and normal ranges and at the cutoff levels with a ranking in precision of WHO 1987 > Dusseldorf > strict Tygerberg criteria, a confirmation of the results published by Comhaire et al. [7]. The observation of a very good intraclass correlation between repeats of slide (intraobserver reproducibility) was previously reported [4, 261. This finding was confirmed by our data, except for laboratory E, which showed lower reproducibility, probably because this laboratory was only recently involved in morphology determinations (years of experience < 1 year). Interrater agreement as defined by Kappa statistics and interpreted according to the criteria formulated by Altman [ 11 (good if K exceeds the 0.61 value) showed some unexpected results. First of all, a good correlation was found between Lab A using WHO criteria and Lab D using strict criteria. A similar observation could be made for the interrater agreement between Lab C using Dusseldorf criteria and Lab G, a reference laboratory for strict criteria. On the other hand, important discrepancies in interrater agreement were observed among the users of strict criteria, which constitutes an important argument to stimulate efforts to further standardize methodology of sperm morphology assessments. Preparation of smears, fixation of smears, staining procedures, minimal number of sperm cells to be evaluated, and different aspects in interpreting normal morphology may all contribute to the important differences that have been observed in this study. In a recent report on the results of a questionnaire on sperm morphology methodology [33], we described a wide and complex variation of different sperm preparation methods, staining procedures, and classification systems worldwide. The observation that half of the responding centers reported using more than one method of semen preparation and sperm morphology evaluation was an unexpected but striking finding.. To conclude, reproducibility in sperm morphology assessments between centers will need a well-documented reappraisal in the near future. This may not lead to the premature conclusion that this semen parameter is useless in the prediction of a man s fertilizing potential. We believe that the findings reported in this study probably will be followed by similar conclusions for other semen parameters, such as sperm motility and sperm concentration. The actual methods for performing such analyses are poorly standardized between laboratories and should be frequently subjected to internal quality control and external quality assessment procedures. According to Matson [24], laboratories should be able to define their own normal ranges in order to minimize errors resulting in incorrect diagnosis, rather than using published criteria and cutoff levels for normality. Another, and possibly the best, solution would be the introduction of more accurate computerized systems with better reproducibility data. Recent reports about CASA systems on sperm morphometry are hopeful, but need confirmation [9, 12, 19, 21, 221. The validity and usefulness of CASA systems have still to be proven in clinical studies with special attention to availability, cost-benefit, and reliability advantages over visual methods. In the meantime, a substantial need for quality control in the seminology laboratories is highly recommended. REFERENCES 1. Altman DG (1991): Some common problems in medical research. In: Practical Statistics for Medical Research. Altman DG (Ed). London: Chapman & Hall, pp Baker HWG, Clarke GN (1987): Sperm morphology: consistency of assessment of the same sperm by different observers. Clin Reprod Fertil 5:37-43.

12 Reproducibility of Sperm Morphological Assessments Bland JM, Altman DG (1986): Statistical method for assessing agreement between two methods of clinical measurement. Lancet 1 : Bosmans E, Janssen M, Cox A, Vandeput H, Ombelet W (1991): Morphological characterization of human spermatozoa according to strict criteria: a laboratory evaluation. Hum Reprod (Abstracts from the 7th meeting of ESHRE and the 7th World Congress on IVF and Assisted Procreation), pp Chan SYW, Wang C, Chan STH, Ho PC, So WWK, Chan YF, Ma HK (1989): Predictive value of sperm morphology and movement characteristics in the outcome of in vitro fertilization of human oocytes. J Vitro Fertil Embryo Transfer 6: Cohen J (1960): A coefficient of agreement for nominal scales. Ed Psycho1 Meas 20: Comhaire F, Schoonjans F, Vermeulen L, De Clercq N (1994): Methodological aspects of sperm morphology evaluation: comparison between strict and liberal criteria. Fertil Steril 62: Cooper TG, Neuwinger J, Bahrs S, Nieschlag E (1992): Internal quality control of semen analysis. Fertil Steril 58: Davis RO, Bain DE, Siemers RJ, Thal DM, Andrew JB, Gravance CG (1992): Accuracy and precision of the CellForm-Human automated sperm morphometry instrument. Fertil Steril 58: Davis RO, Gravance CG (1993): Standardization of specimen preparation, staining, and sampling methods improves automated sperm-head morphometry analysis. Fertil Steril 59: I. Davis RO, Gravance CG (1994): Consistency of sperm morphology classification methods. J Androl 15 :83-9 I. 12. Davis RO, Gravance CG, Overstreet JW (1995): A standardized test for visual analysis of human sperm morphology. Fertil Steril 63: Dunphy BC, Kay R, Barratt CLR, Cooke ID (1989): Quality control during the conventional analysis of semen, an essential exercise. J Androl 10: Eliasson R (1971): Standards for investigation of human semen? Andrologie 3: Fredricsson B (1978): Morphologic evaluation of spermatozoa in different laboratories. Andrologia I I Freund M (1966): Standards for the rating of human sperm morphology. Internat J Fertil 11: Hofmann N, Freundl G, Florack M (1985): Die Formstorungen der Spermatozoen im Sperma und Zervikalschleim als spiegel testikularer Erkrankungen. Gynakologe 18: Hofmann N, Haider SG (1985): Neue Ergebnisse morphologischer Diagnostik der Spermatogenesestarungen. Gynakologe 18: Hofmann GE, Santilli BA, Kindig SK, Scott RT Jr, Johnson CA (1 996): Intraobserver, interobserver variation of sperm critical morphology: comparison of examiner and computer-assisted analysis. Fertil Steril 65: Jequier AM, Ukombe EB (1983): Errors inherent in the performance of a routine semen analysis. Br J Urol 55: Kruger T (1995): Computer-assisted sperm analysis systems: morphometric aspects.hum Reprod 10 (Suppl 1):4& Kruger TF, Lacquet FA, Sanchez Sarmiento CA, Menkveld R, Ozgur K, Lombard CJ, Franken DR (1996): A prospective study on the predictive value of normal sperm morphology as evaluated by computer (IVOS*). Fertil Steril 66: Kruger TF, Menkveld R, Stander FSH, Lombard CJ, Van der Merwe JP, Van Zyl JA, Smith K (1986): Sperm morphologic features as a prognostic factor in in vitro fertilization. Fertil Steril 46:llI8-1l Matson P (1995): External quality assessment for semen analysis and sperm antibody detection: results of a pilot scheme. Hum Reprod 10: Menkveld R, Kruger TF (1995): Advantages of strict (Tygerberg) criteria for sperm morphology evaluation. Int J Androl 18 (Suppl 2):36-42.

13 114 W. Ombelet et al. 26. Menkveld R, Stander FSH, Kotze TJW, Kruger TF, Van Zyl JA (1990): The evaluation of morphological characteristics of human spermatozoa according to stricter criteria. Hum Reprod 5: Neuwinger J, Behre HM, Nieschlag E (1990): Computerized semen analysis with sperm tail detection. Hum Reprod 5, Oehninger S, Acosta AA, Morshedi M, Veeck L, Swanson RJ, Simmons K, Rosenwaks Z (1988): Corrective measures and pregnancy outcome in in vitro fertilization in patients with severe sperm morphology abnormalities. Fertil Steril 50: Ombelet W, Bosmans E, Janssen M. Cox A, Vlasselaer J, Gyselaers W, Vandeput H, Gielen J, Pollet H, Maes M, Steeno 0, Kruger T (1997): Semen parameters in a fertile versus subfertile population: a need for change in interpretation of semen testing. Hum Reprod 12: Ombelet W, Cox A, Janssen M, Vandeput H, Bosmans E (1996): Artificial insemination 2: using the husband s sperm. In: Diagnosis and Therapy of Male Factor in Assisted Reproduction. Acosta A A, Kruger T F (Eds.). New York: Parthenon, pp Ombelet W, Fourie FLeR, Vandeput H, Bosmans E, Cox A, Janssen M, Kruger T (1994): Teratozoospermia and in-vitro fertilization: a randomized prospective study. Hum Reprod 9: Ombelet W, Menkveld R, Kruger TF, Steeno 0 (1995): Sperm morphology assessment: historical review in relation to fertility. Hum Reprod Update 1: Ombelet W, Pollet H, Bosmans E, Vereecken A (1997): Results of a questionnaire on sperm morphology assessment. Hum Reprod 12: Toner JP, Mossad H, Grow DR, Moshedi M, Swanson RJ, Oehninger S (1994): Value of sperm morphology assessed by strict criteria for prediction of the outcome of artificial (intrauterine) insemination. Andrologia 27: Van Zyl J A, Kotze T J vw, Menkveld R (1990): Predictive value of spermatozoa morphology in natural fertilization. In: Human Spermatozoa in Assisted Reproduction. Acosta AA, Swanson RJ, Ackerman SB, Kruger TF, Van Zyl JA, Menkveld R (Eds). Williams & Wilkons, pp World Health Organization (1987): WHO Laboratory Manual for the Examination of Human Semen and Semen-Cervical Mucus Interaction, 2nd ed. Cambridge University Press, Cambridge, UK. 37. World Health Organization (1992): WHO Laboratory Manual for the Examination of Human Semen and Semen-Cervical Mucus Interaction. 3d ed. Cambridge University Press, Cambridge, UK. 38. Yovich JL, Matson PL (1995): Male subfertility: concepts in Hum Reprod 10 (Suppl 1): Zaini A, Jennings MG, Baker HWG (1985): Are conventional sperm morphology and motility assessments of predictive value in subfertile men? Int J Androl 8: