Economic evaluation of endometrial scratching before the second IVF/ICSI treatment: a cost-effectiveness analysis of a randomized controlled trial (SCRaTCH trial).

STUDY QUESTION: Is a single endometrial scratch prior to the second fresh IVF/ICSI treatment cost-effective compared to no scratch, when evaluated over a 12-month follow-up period? SUMMARY ANSWER: The incremental cost-effectiveness ratio (ICER) for an endometrial scratch was €6524 per additional live birth, but due to uncertainty regarding the increase in live birth rate this has to be interpreted with caution. WHAT IS KNOWN ALREADY: Endometrial scratching is thought to improve the chances of success in couples with previously failed embryo implantation in IVF/ICSI treatment. It has been widely implemented in daily practice, despite the lack of conclusive evidence of its effectiveness and without investigating whether scratching allows for a cost-effective method to reduce the number of IVF/ICSI cycles needed to achieve a live birth. STUDY DESIGN, SIZE, DURATION: This economic evaluation is based on a multicentre randomized controlled trial carried out in the Netherlands (SCRaTCH trial) that compared a single scratch prior to the second IVF/ICSI treatment with no scratch in couples with a failed full first IVF/ICSI cycle. Follow-up was 12 months after randomization.Economic evaluation was performed from a healthcare and societal perspective by taking both direct medical costs and lost productivity costs into account. It was performed for the primary outcome of biochemical pregnancy leading to live birth after 12 months of follow-up as well as the secondary outcome of live birth after the second fresh IVF/ICSI treatment (i.e. the first after randomization). To allow for worldwide interpretation of the data, cost level scenario analysis and sensitivity analysis was performed. PARTICIPANTS/MATERIALS, SETTING, METHODS: From January 2016 until July 2018, 933 women with a failed first IVF/ICSI cycle were included in the trial. Data on treatment and pregnancy were recorded up until 12 months after randomization, and the resulting live birth outcomes (even if after 12 months) were also recorded.Total costs were calculated for the second fresh IVF/ICSI treatment and for the full 12 month period for each participant. We included costs of all treatments, medication, complications and lost productivity costs. Cost-effectiveness analysis was carried out by calculating ICERs for scratch compared to control. Bootstrap resampling was used to estimate the uncertainty around cost and effect differences and ICERs. In the sensitivity and scenario analyses, various unit costs for a single scratch were introduced, amongst them, unit costs as they apply for the United Kingdom (UK). MAIN RESULTS AND THE ROLE OF CHANCE: More live births occurred in the scratch group, but this also came with increased costs over a 12-month period. The estimated chance of a live birth after 12 months of follow-up was 44.1% in the scratch group compared to 39.3% in the control group (risk difference 4.8%, 95% CI -1.6% to +11.2%). The mean costs were on average €283 (95% CI: -€299 to €810) higher in the scratch group so that the point average ICER was €5846 per additional live birth. The ICER estimate was surrounded with a high level of uncertainty, as indicated by the fact that the cost-effectiveness acceptability curve (CEAC) showed that there is an 80% chance that endometrial scratching is cost-effective if society is willing to pay ∼€17 500 for each additional live birth. LIMITATIONS, REASONS FOR CAUTION: There was a high uncertainty surrounding the effects, mainly in the clinical effect, i.e. the difference in the chance of live birth, which meant that a single straightforward conclusion could not be ascertained as for now. WIDER IMPLICATIONS OF THE FINDINGS: This is the first formal cost-effectiveness analysis of endometrial scratching in women undergoing IVF/ICSI treatment. The results presented in this manuscript cannot provide a clear-cut expenditure for one additional birth, but they do allow for estimating costs per additional live birth in different scenarios once the clinical effectiveness of scratching is known. As the SCRaTCH trial was the only trial with a follow-up of 12 months, it allows for the most complete estimation of costs to date. STUDY FUNDING/COMPETING INTEREST(S): This study was funded by ZonMW, the Dutch organization for funding healthcare research. A.E.P.C., F.J.M.B., E.R.G. and C.B. L. reported having received fees or grants during, but outside of, this trial. TRIAL REGISTRATION NUMBER: Netherlands Trial Register (NL5193/NTR 5342).

Corrigendum: Predicting Colorectal Cancer Recurrence and Patient Survival Using Supervised Machine Learning Approach: A South African Population-Based Study.

[This corrects the article DOI: 10.3389/fpubh.2021.694306.].

Predicting Colorectal Cancer Recurrence and Patient Survival Using Supervised Machine Learning Approach: A South African Population-Based Study.

Background: South Africa (SA) has the highest incidence of colorectal cancer (CRC) in Sub-Saharan Africa (SSA). However, there is limited research on CRC recurrence and survival in SA. CRC recurrence and overall survival are highly variable across studies. Accurate prediction of patients at risk can enhance clinical expectations and decisions within the South African CRC patients population. We explored the feasibility of integrating statistical and machine learning (ML) algorithms to achieve higher predictive performance and interpretability in findings. Methods: We selected and compared six algorithms:- logistic regression (LR), naïve Bayes (NB), C5.0, random forest (RF), support vector machine (SVM) and artificial neural network (ANN). Commonly selected features based on OneR and information gain, within 10-fold cross-validation, were used for model development. The validity and stability of the predictive models were further assessed using simulated datasets. Results: The six algorithms achieved high discriminative accuracies (AUC-ROC). ANN achieved the highest AUC-ROC for recurrence (87.0%) and survival (82.0%), and other models showed comparable performance with ANN. We observed no statistical difference in the performance of the models. Features including radiological stage and patient's age, histology, and race are risk factors of CRC recurrence and patient survival, respectively. Conclusions: Based on other studies and what is known in the field, we have affirmed important predictive factors for recurrence and survival using rigorous procedures. Outcomes of this study can be generalised to CRC patient population elsewhere in SA and other SSA countries with similar patient profiles.

Using Machine Learning Methods Incorporating Individual Reader Annotations to Classify Paediatric Chest Radiographs in Epidemiological Studies.

Introduction: Epidemiological studies that involve interpretation of chest radiographs (CXRs) suffer from inter-reader and intra-reader variability. Inter-reader and intra-reader variability hinder comparison of results from different studies or centres, which negatively affects efforts to track the burden of chest diseases or evaluate the efficacy of interventions such as vaccines. This study explores machine learning models that could standardize interpretation of CXR across studies and the utility of incorporating individual reader annotations when training models using CXR data sets annotated by multiple readers. Methods: Convolutional neural networks were used to classify CXRs from seven low to middle-income countries into five categories according to the World Health Organization's standardized methodology for interpreting paediatric CXRs. We compared models trained to predict the final/aggregate classification with models trained to predict how each reader would classify an image and then aggregate predictions for all readers using unweighted mean. Results: Incorporating individual reader's annotations during model training improved classification accuracy by 3.4% (multi-class accuracy 61% vs 59%). Model accuracy was higher for children above 12 months of age (68% vs 58%). The accuracy of the models in different countries ranged between 45% and 71%. Conclusions: Machine learning models can annotate CXRs in epidemiological studies reducing inter-reader and intra-reader variability. In addition, incorporating individual reader annotations can improve the performance of machine learning models trained using CXRs annotated by multiple readers.

Using self-supervised feature learning to improve the use of pulse oximeter signals to predict paediatric hospitalization.

Background: The success of many machine learning applications depends on knowledge about the relationship between the input data and the task of interest (output), hindering the application of machine learning to novel tasks. End-to-end deep learning, which does not require intermediate feature engineering, has been recommended to overcome this challenge but end-to-end deep learning models require large labelled training data sets often unavailable in many medical applications. In this study, we trained self-supervised learning (SSL) models for automatic feature extraction from raw photoplethysmography (PPG) obtained using a pulse oximeter, with the aim of predicting paediatric hospitalization.  Methods: We compared logistic regression models fitted using features extracted using SSL with models trained using both clinical and SSL features. In addition, we compared end-to-end deep learning models initialized randomly or using weights from the SSL models. We also compared the performance of SSL models trained on labelled data alone (n=1,031) with SSL trained using both labelled and unlabelled signals (n=7,578). Results: Logistic regression models were more predictive of hospitalization when trained on features extracted using labelled PPG signals only compared to SSL models trained on both labelled and unlabelled signals (AUC 0.83 vs 0.80). However, features extracted using SSL model trained on both labelled and unlabelled PPG signals were more predictive of hospitalization when concatenated with clinical features (AUC 0.89 vs 0.87). The end-to-end deep learning model had an AUC of 0.80 when initialized using the SSL model trained on all PPG signals, 0.77 when initialized using SSL trained on labelled data only, and 0.73 when initialized randomly. Conclusions: This study shows that SSL can extract features from PPG signals that are predictive of hospitalization or initialize end-to-end deep learning models. Furthermore, SSL can leverage larger unlabelled data sets to improve performance of models fitted using small labelled data sets.

Endometrial scratching in women with one failed IVF/ICSI cycle-outcomes of a randomised controlled trial (SCRaTCH).

STUDY QUESTION: Does endometrial scratching in women with one failed IVF/ICSI treatment affect the chance of a live birth of the subsequent fresh IVF/ICSI cycle? SUMMARY ANSWER: In this study, 4.6% more live births were observed in the scratch group, with a likely certainty range between -0.7% and +9.9%. WHAT IS KNOWN ALREADY: Since the first suggestion that endometrial scratching might improve embryo implantation during IVF/ICSI, many clinical trials have been conducted. However, due to limitations in sample size and study quality, it remains unclear whether endometrial scratching improves IVF/ICSI outcomes. STUDY DESIGN, SIZE, DURATION: The SCRaTCH trial was a non-blinded randomised controlled trial in women with one unsuccessful IVF/ICSI cycle and assessed whether a single endometrial scratch using an endometrial biopsy catheter would lead to a higher live birth rate after the subsequent IVF/ICSI treatment compared to no scratch. The study took place in 8 academic and 24 general hospitals. Participants were randomised between January 2016 and July 2018 by a web-based randomisation programme. Secondary outcomes included cumulative 12-month ongoing pregnancy leading to live birth rate. PARTICIPANTS/MATERIALS, SETTING, METHODS: Women with one previous failed IVF/ICSI treatment and planning a second fresh IVF/ICSI treatment were eligible. In total, 933 participants out of 1065 eligibles were included (participation rate 88%). MAIN RESULTS AND THE ROLE OF CHANCE: After the fresh transfer, 4.6% more live births were observed in the scratch compared to control group (110/465 versus 88/461, respectively, risk ratio (RR) 1.24 [95% CI 0.96-1.59]). These data are consistent with a true difference of between -0.7% and +9.9% (95% CI), indicating that while the largest proportion of the 95% CI is positive, scratching could have no or even a small negative effect. Biochemical pregnancy loss and miscarriage rate did not differ between the two groups: in the scratch group 27/153 biochemical pregnancy losses and 14/126 miscarriages occurred, while this was 19/130 and 17/111 for the control group (RR 1.21 (95% CI 0.71-2.07) and RR 0.73 (95% CI 0.38-1.40), respectively). After 12 months of follow-up, 5.1% more live births were observed in the scratch group (202/467 versus 178/466), of which the true difference most likely lies between -1.2% and +11.4% (95% CI). LIMITATIONS, REASONS FOR CAUTION: This study was not blinded. Knowledge of allocation may have been an incentive for participants allocated to the scratch group to continue treatment in situations where they may otherwise have cancelled or stopped. In addition, this study was powered to detect a difference in live birth rate of 9%. WIDER IMPLICATIONS OF THE FINDINGS: The results of this study are an incentive for further assessment of the efficacy and clinical implications of endometrial scratching. If a true effect exists, it may be smaller than previously anticipated or may be limited to specific groups of women undergoing IVF/ICSI. Studying this will require larger sample sizes, which will be provided by the ongoing international individual participant data-analysis (PROSPERO CRD42017079120). At present, endometrial scratching should not be performed outside of clinical trials. STUDY FUNDING/COMPETING INTEREST(S): This study was funded by ZonMW, the Dutch organisation for funding healthcare research. J.S.E. Laven reports grants and personal fees from AnshLabs (Webster, Tx, USA), Ferring (Hoofddorp, The Netherlands) and Ministry of Health (CIBG, The Hague, The Netherlands) outside the submitted work. A.E.P. Cantineau reports 'other' from Ferring BV, personal fees from Up to date Hyperthecosis, 'other' from Theramex BV, outside the submitted work. E.R. Groenewoud reports grants from Titus Health Care during the conduct of the study. A.M. van Heusden reports personal fees from Merck Serono, personal fees from Ferring, personal fees from Goodlife, outside the submitted work. F.J.M. Broekmans reports personal fees as Member of the external advisory board for Ferring BV, The Netherlands, personal fees as Member of the external advisory board for Merck Serono, The Netherlands, personal fees as Member of the external advisory for Gedeon Richter, Belgium, personal fees from Educational activities for Ferring BV, The Netherlands, grants from Research support grant Merck Serono, grants from Research support grant Ferring, personal fees from Advisory and consultancy work Roche, outside the submitted work. C.B. Lambalk reports grants from Ferring, grants from Merck, grants from Guerbet, outside the submitted work. TRIAL REGISTRATION NUMBER: Registered in the Netherlands Trial Register (NL5193/NTR 5342). TRIAL REGISTRATION DATE: 31 July 2015. DATE OF FIRST PATIENT'S ENROLMENT: 26 January 2016.

Cost-effectiveness of medically assisted reproduction or expectant management for unexplained subfertility: when to start treatment?

STUDY QUESTION: Over a time period of 3 years, which order of expectant management (EM), IUI with ovarian stimulation (IUI-OS) and IVF is the most cost-effective for couples with unexplained subfertility with the female age below 38 years? SUMMARY ANSWER: If a live birth is considered worth €32 000 or less, 2 years of EM followed by IVF was the most cost-effective, whereas above €32 000 this was 1 year of EM, 1 year of IUI-OS and then 1 year of IVF. WHAT IS KNOWN ALREADY: IUI-OS and IVF are commonly used fertility treatments for unexplained subfertility although many couples can conceive naturally, as no identifiable barrier to conception could be found by definition. Few countries have guidelines on when to proceed with medically assisted reproduction (MAR), mostly based on the expected probability of live birth after treatment, but there is a lack of evidence to support the strategies proposed by these guidelines. The increased uptake of IUI-OS and IVF over the past decades and costs related to reimbursement of these treatments are pressing concerns to health service providers. For MAR to remain affordable, sustainable and a responsible use of public funds, guidance is needed on the cost-effectiveness of treatment strategies for unexplained subfertility, including EM. STUDY DESIGN, SIZE, DURATION: We developed a decision analytic Markov model that follows couples with unexplained subfertility of which the woman is under 38 years of age for a time period of 3 years from completion of the fertility workup onwards. We divided the time axis of 3 years into three separate periods, each comprising 1 year. The model was based on contemporary evidence, most notably the dynamic prediction model for natural conception, which was combined with MAR treatment effects from a network meta-analysis on randomized controlled trials. We changed the order of options for managing unexplained subfertility for the 1 year periods to yield five different treatment policies in total: IVF-EM-EM (immediate IVF), EM-IVF-EM (delayed IVF), EM-EM-IVF (postponed IVF), IUIOS-IVF-EM (immediate IUI-OS) and EM-IUIOS-IVF (delayed IUI-OS). PARTICIPANTS/MATERIALS, SETTING, METHODS: The main outcomes per policy over the 3-year period were the probability of live birth, the average treatment and delivery costs, the probability of multiple pregnancy, the incremental cost-effectiveness ratio (ICER) and finally, which policy yields the highest net benefit in which costs for a policy were deducted from the health effects, i.e. live births gained. We chose the Dutch societal perspective, but the model can be easily modified for other locations or other perspectives. The probability of live birth after EM was taken from the dynamic prediction model for natural conception and updated for Years 2 and 3. The relative effects of IUI-OS and IVF in terms of odds ratios, taken from the network meta-analysis, were applied to the probability of live birth after EM. We applied standard discounting procedures for economic analyses for Years 2 and 3. The uncertainty around effectiveness, costs and other parameters was assessed by probabilistic sensitivity analysis in which we drew values from distributions and repeated this procedure 20 000 times. In addition, we changed model assumptions to assess their influence on our results. MAIN RESULTS AND THE ROLE OF CHANCE: From IVF-EM-EM to EM-IUIOS-IVF, the probability of live birth varied from approximately 54-64% and the average costs from approximately €4000 to €9000. The policies IVF-EM-EM and EM-IVF-EM were dominated by EM-EM-IVF as the latter yielded a higher cumulative probability of live birth at a lower cost. The policy IUIOS-IVF-EM was dominated by EM-IUIOS-IVF as the latter yielded a higher cumulative probability of live birth at a lower cost. After removal of policies that were dominated, the ICER for EM-IUIOS-IVF was approximately €31 000 compared to EM-EM-IVF. The range of ICER values between the lowest 25% and highest 75% of simulation replications was broad. The net benefit curve showed that when we assume a live birth to be worth approximately €20 000 or less, the policy EM-EM-IVF had the highest probability to achieve the highest net benefit. Between €20 000 and €50 000 monetary value per live birth, it was uncertain whether EM-EM-IVF was better than EM-IUIOS-IVF, with the turning point of €32 000. When we assume a monetary value per live birth over €50 000, the policy with the highest probability to achieve the highest net benefit was EM-IUIOS-IVF. Results for subgroups with different baseline prognoses showed the same policies dominated and the same two policies that were the most likely to achieve the highest net benefit but at different threshold values for the assumed monetary value per live birth. LIMITATIONS, REASONS FOR CAUTION: Our model focused on population level and was thus based on average costs for the average number of cycles conducted. We also based the model on a number of key assumptions. We changed model assumptions to assess the influence of these assumptions on our results. The change in relative effectiveness of IVF over time was found to be highly influential on results and their interpretation. WIDER IMPLICATIONS OF THE FINDINGS: EM-EM-IVF and EM-IUIOS-IVF followed by IVF were the most cost-effective policies. The choice depends on the monetary value assigned to a live birth. The results of our study can be used in discussions between clinicians, couples and policy makers to decide on a sustainable treatment protocol based on the probability of live birth, the costs and the limitations of MAR treatment. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by the ZonMw Doelmatigheidsonderzoek (80-85200-98-91072). The funder had no role in the design, conduct or reporting of this work. B.W.M. is supported by a NHMRC Practitioner Fellowship (GNT1082548). B.W.M. reports consultancy for ObsEva, Merck KGaA and Guerbet and travel and research support from ObsEva, Merck and Guerbet. TRIAL REGISTRATION NUMBER: N/A.

Age-related natural fertility outcomes in women over 35 years: a systematic review and individual participant data meta-analysis.

STUDY QUESTION: What is the rate of natural conception leading to ongoing pregnancy or livebirth over 6-12 months for infertile women of age ≥35 years? SUMMARY ANSWER: Natural conception rates were still clinically relevant in women aged 35 years and above and were significantly higher in women with unexplained infertility compared to those with other diagnoses. WHAT IS KNOWN ALREADY: In recent years, increasing numbers of women have attempted to conceive at a later age, resulting in a commensurate increase in the need for ART. However, there is a lack of data on natural fertility outcomes (i.e. no interventions) in women with increasing age. STUDY DESIGN, SIZE, DURATION: A systematic review with individual participant data (IPD) meta-analysis was carried out. PubMed, MEDLINE, EMBASE, the Cochrane Library, clinicaltrials.gov were searched until 1 July 2018 including search terms 'fertility service', 'waiting list', 'treatment-independent' and 'spontaneous conception'. Language restrictions were not imposed. PARTICIPANTS/MATERIALS, SETTING, METHODS: Inclusion criteria were studies (at least partly) reporting on infertile couples with female partner of age ≥35 years who attended fertility services, underwent fertility workup (e.g. history, semen analysis, tubal status and ovulation status) and were exposed to natural conception (e.g. independent of treatment such as IVF, ovulation induction and tubal surgery). Studies that exclusively studied only one infertility diagnosis, without including other women presenting to infertility services for other causes of infertility, were excluded. For studies that met the inclusion criteria, study authors were contacted to provide IPD, after which fertility outcomes for women of age ≥35 years were retrieved. Time to pregnancy or livebirth and the effect of increasing age on fertility outcomes after adjustment for other prognostic factors were analysed. Quality of studies was graded with the Newcastle-Ottawa Scale (non-randomised controlled trials (RCTs)) or the Cochrane Risk of Bias tool (for RCTs). MAIN RESULTS AND THE ROLE OF CHANCE: We included nine studies (seven cohort studies and two RCTs) (n = 4379 women of at least age 35 years), with the observed composite primary outcome of ongoing pregnancy or livebirth occurring in 429 women (9.8%) over a median follow-up of 5 months (25th to 75th percentile: 2.5-8.5 months). Studies were of moderate to high quality. The probability of natural conception significantly decreased with any diagnosis of infertility, when compared with unexplained infertility. We found non-linear effects of female age and duration of infertility on ongoing pregnancy and tabulated the predicted probabilities for unexplained infertile women aged 35-42 years with either primary or secondary infertility and with a duration of infertility from 1 to 6 years. For a 35-year-old woman with 2 years of primary unexplained infertility, the predicted probability of natural conception leading to ongoing pregnancy or livebirth was 0.15 (95% CI 0.11-0.19) after 6 months and 0.24 (95% CI 0.17-0.30) after 12 months. For a 42-year-old woman, this decreased to 0.08 (95% CI 0.04-0.11) after 6 months and 0.13 (95% CI 0.07-0.18) after 12 months. LIMITATIONS, REASONS FOR CAUTION: In the studies selected, there were different study designs, recruitment strategies in different centres, protocols and countries and different methods of assessment of infertility. Data were limited for women above the age of 40 years. WIDER IMPLICATIONS OF THE FINDINGS: Women attending fertility services should be encouraged to pursue natural conception while waiting for treatment to commence and after treatment if it is unsuccessful. Our results may aid in counselling women, and, in particular, for those with unexplained infertility. STUDY FUNDING/COMPETING INTEREST(S): S.J.C. received funding from the University of Adelaide Summer Research Scholarship. B.W.M. is supported by a NHMRC Investigator grant (GNT1176437), B.W.M. reports consultancy for ObsEva, Merck, Merck KGaA, iGenomix and Guerbet. B.W.M. reports research support by Merck and Guerbet. PROSPERO REGISTRATION NUMBER: CRD42018096552.

Does the effectiveness of IUI in couples with unexplained subfertility depend on their prognosis of natural conception? A replication of the H2Oil study.

STUDY QUESTION: Can we replicate the finding that the benefit of IUI-ovarian stimulation (IUI-OS) compared to expectant management for couples with unexplained subfertility depends on the prognosis of natural conception? SUMMARY ANSWER: The estimated benefit of IUI-OS did not depend on the prognosis of natural conception but did depend on when treatment was started after diagnosis, with starting IUI-OS later yielding a larger absolute and relative benefit of treatment. WHAT IS KNOWN ALREADY: IUI-OS is often the first-line treatment for couples with unexplained subfertility. Two randomized controlled trials (RCTs) compared IUI-OS to expectant management using different thresholds for the prognosis of natural conception as inclusion criteria and found different results. In a previous study (a Dutch national cohort), it was found that the benefit of IUI-OS compared to expectant management seemed dependent on the prognosis of natural conception, but this finding warrants replication. STUDY DESIGN SIZE DURATION: We conducted a secondary analysis of the H2Oil study (n = 1119), a multicentre RCT that evaluated the effect of oil-based contrast versus water-based contrast during hysterosalpingography (HSG). Couples were randomized before HSG and followed up for 3-5 years. We selected couples with unexplained subfertility who received HSG and had follow-up or pregnancy data available. Follow-up was censored at the start of IVF, after the last IUI cycle or at last contact and was truncated at a maximum of 18 months after the fertility workup. PARTICIPANTS/MATERIALS SETTING METHODS: The endpoint was time to conception leading to an ongoing pregnancy. We used the sequential Cox approach comparing in each month the ongoing pregnancy rates over the next 6 months of couples who started IUI-OS to couples who did not. We calculated the prognosis of natural conception for individual couples, updated this over consecutive failed cycles and evaluated whether prognosis modified the effect of starting IUI-OS. We corrected for known predictors of conception using inverse probability weighting. MAIN RESULTS AND THE ROLE OF CHANCE: Data from 975 couples were available. There were 587 couples who received at least one IUI-OS cycle within 18 months after HSG of whom 221 conceived leading to an ongoing pregnancy (rate: 0.74 per couple per year over a median follow-up for IUI of 5 months). The median period between HSG and starting IUI-OS was 4 months. Out of 388 untreated couples, 299 conceived naturally (rate: 0.56 per couple per year over a median follow-up of 4 months). After creating our mimicked trial datasets, starting IUI-OS was associated with a higher chance of ongoing pregnancy by a pooled, overall hazard ratio of 1.50 (95% CI: 1.19-1.89) compared to expectant management. We did not find strong evidence that the effect of treatment was modified by a couple's prognosis of achieving natural conception (Akaike's Information Criterion (AIC) decreased by 1 point). The effect of treatment was dependent on when couples started IUI-OS (AIC decreased by more than 2 points). The patterns of estimated absolute chances over time for couples with increasingly better prognoses were different from the previous study but the finding that starting later yields a larger benefit of treatment was similar. We found IUI-OS increased the absolute chance of pregnancy by at least 5% compared to expectant management. The absolute chance of pregnancy after IUI-OS seems less variable between couples and starting times of treatment than the absolute chance after expectant management. LIMITATIONS REASONS FOR CAUTION: This is a secondary analysis, as the H2Oil trial was not designed with this research question in mind. Owing to sample size restrictions, it remained difficult to distinguish between the ranges of prognoses in which true benefit was found. WIDER IMPLICATIONS OF THE FINDINGS: We replicated the finding that starting IUI-OS later after diagnosis yields a larger absolute and relative benefit of treatment. We did not replicate the dependency of the effect of IUI-OS on the prognosis of natural conception and could not identify clear thresholds for the prognosis of natural conception when IUI-OS was and/or was not effective. Because many of these couples still have good chances of natural conception at the time of diagnosis, we suggest clinicians should advise couples to delay the start of IUI-OS for several months to avoid unnecessary treatment. STUDY FUNDING/COMPETING INTERESTS: The H2Oil study (NTR 3270) was an investigator-initiated study that was funded by the two academic institutions (AMC and VUmc) of the Amsterdam UMC. The follow-up study (NTR 6577) was also an investigator-initiated study with funding by Guerbet, France. The funders had no role in study design, collection, analysis and interpretation of the data. B.W.M. is supported by an Investigator grant (GNT1176437) from the Australian National Health and Medical Research Council (NHMRC). K.D. reports receiving travel and speaker fees from Guerbet. B.W.M. reports consultancy for ObsEva, Merck, Merck KGaA, iGenomix and Guerbet. V.M. reports receiving travel- and speaker fees as well as research grants from Guerbet.

Eliminating biasing signals in lung cancer images for prognosis predictions with deep learning.

Deep learning has shown remarkable results for image analysis and is expected to aid individual treatment decisions in health care. Treatment recommendations are predictions with an inherently causal interpretation. To use deep learning for these applications in the setting of observational data, deep learning methods must be made compatible with the required causal assumptions. We present a scenario with real-world medical images (CT-scans of lung cancer) and simulated outcome data. Through the data simulation scheme, the images contain two distinct factors of variation that are associated with survival, but represent a collider (tumor size) and a prognostic factor (tumor heterogeneity), respectively. When a deep network would use all the information available in the image to predict survival, it would condition on the collider and thereby introduce bias in the estimation of the treatment effect. We show that when this collider can be quantified, unbiased individual prognosis predictions are attainable with deep learning. This is achieved by (1) setting a dual task for the network to predict both the outcome and the collider and (2) enforcing a form of linear independence of the activation distributions of the last layer. Our method provides an example of combining deep learning and structural causal models to achieve unbiased individual prognosis predictions. Extensions of machine learning methods for applications to causal questions are required to attain the long-standing goal of personalized medicine supported by artificial intelligence.

What is the prognosis for a live birth after unexplained recurrent implantation failure following IVF/ICSI?

STUDY QUESTION: What is the cumulative incidence of live birth and mean time to pregnancy (by conception after IVF/ICSI or natural conception) in women experiencing unexplained recurrent implantation failure (RIF) following IVF/ICSI treatment? SUMMARY ANSWER: In 118 women who had experienced RIF, the reported cumulative incidence of live birth during a maximum of 5.5 years follow-up period was 49%, with a calculated median time to pregnancy leading to live birth of 9 months after diagnosis of RIF. WHAT IS KNOWN ALREADY: Current definitions of RIF include failure to achieve a pregnancy following IVF/ICSI and undergoing three or more fresh embryo transfer procedures of one or two high quality embryos or more than 10 embryos transferred in fresh or frozen cycles. The causes and optimal management of this distressing condition remain uncertain and a range of empirical and often expensive adjuvant therapies is often advocated. Little information is available regarding the long-term prognosis for achieving a pregnancy. STUDY DESIGN, SIZE, DURATION: Two hundred and twenty-three women under 39 years of age who had experienced RIF without a known cause after IVF/ICSI treatment in two tertiary referral university hospitals between January 2008 and December 2012 were invited to participate in this retrospective cohort follow up study. PARTICIPANTS/MATERIALS, SETTING, METHODS: All eligible women were sent a letter requesting their consent to the anonymous use of their medical file data and were asked to complete a questionnaire enquiring about treatments and pregnancies subsequent to experiencing RIF. Medical files and questionnaires were examined and results were analysed to determine the subsequent cumulative incidence of live birth and time to pregnancy within a maximum 5.5 year follow-up period using Kaplan Meier analysis. Clinical predictors for achieving a live birth were investigated using a Cox hazard model. MAIN RESULTS AND THE ROLE OF CHANCE: One hundred and twenty-seven women responded (57%) and data from 118 women (53%) were available for analysis. During the maximum 5.5 year follow up period the overall cumulative incidence of live birth was 49% (95% CI 39-59%). Among women who gave birth, the calculated median time to pregnancy was 9 months after experiencing RIF, where 18% arose from natural conceptions. LIMITATIONS, REASONS FOR CAUTION: Since only 57% of the eligible study cohort completed the questionnaire, the risk of response bias limits the applicability of the study findings. WIDER IMPLICATIONS OF THE FINDINGS: This study reports a favorable overall prognosis for achieving live birth in women who have previously experienced RIF, especially in those who continue with further IVF/ICSI treatments. However since 51% did not achieve a live birth during the follow-up period, there is a need to distinguish those most likely to benefit from further treatment. In this study, no clinical factors were found to be predictive of those achieving a subsequent live birth. STUDY FUNDING/COMPETING INTEREST(S): This study was funded by the University Medical Center Utrecht, in Utrecht and the Academic Medical Centre, in Amsterdam. NSM has received consultancy and speaking fees and research funding from Ferring, MSD, Merck Serono, Abbott, IBSA, Gedion Richter, and Clearblue. During the most recent 5-year period BCJMF has received fees or grant support from the following organizations (in alphabetic order); Actavis/Watson/Uteron, Controversies in Obstetrics & Gynecology (COGI), Dutch Heart Foundation, Dutch Medical Research Counsel (ZonMW), Euroscreen/Ogeda, Ferring, London Womens Clinic (LWC), Merck Serono, Myovant, Netherland Genomic Initiative (NGI), OvaScience, Pantharei Bioscience, PregLem/Gedeon Richter/Finox, Reproductive Biomedicine Online (RBMO), Roche, Teva, World Health Organisation (WHO).None of the authors have disclosures to make in relation to this manuscript.

Incorporating repeated measurements into prediction models in the critical care setting: a framework, systematic review and meta-analysis.

BACKGROUND: The incorporation of repeated measurements into multivariable prediction research may greatly enhance predictive performance. However, the methodological possibilities vary widely and a structured overview of the possible and utilized approaches lacks. Therefore, we [1] propose a structured framework for these approaches, [2] determine what methods are currently used to incorporate repeated measurements in prediction research in the critical care setting and, where possible, [3] assess the added discriminative value of incorporating repeated measurements. METHODS: The proposed framework consists of three domains: the observation window (static or dynamic), the processing of the raw data (raw data modelling, feature extraction and reduction) and the type of modelling. A systematic review was performed to identify studies which incorporate repeated measurements to predict (e.g. mortality) in the critical care setting. The within-study difference in c-statistics between models with versus without repeated measurements were obtained and pooled in a meta-analysis. RESULTS: From the 2618 studies found, 29 studies incorporated multiple repeated measurements. The annual number of studies with repeated measurements increased from 2.8/year (2000-2005) to 16.0/year (2016-2018). The majority of studies that incorporated repeated measurements for prediction research used a dynamic observation window, and extracted features directly from the data. Differences in c statistics ranged from - 0.048 to 0.217 in favour of models that utilize repeated measurements. CONCLUSIONS: Repeated measurements are increasingly common to predict events in the critical care domain, but their incorporation is lagging. A framework of possible approaches could aid researchers to optimize future prediction models.

IVF for unexplained subfertility; whom should we treat?

STUDY QUESTION: Which couples with unexplained subfertility can expect increased chances of ongoing pregnancy with IVF compared to expectant management? SUMMARY ANSWER: For couples in which the woman is under 40 years of age, IVF is associated with higher chances of conception than expectant management. WHAT IS KNOWN ALREADY: The clinical indications for IVF have expanded over time from bilateral tubal blockage to include unexplained subfertility in which there is no identifiable barrier to conception. Yet, there is little evidence from randomized controlled trials that IVF is effective in these couples. STUDY DESIGN, SIZE, DURATION: We compared outcomes in British couples with unexplained subfertility undergoing IVF (n = 40 921) from registry data to couples with the same type of subfertility on expectant management. Those couples on expectant management (defined as no intervention aside from the advice to have intercourse) comprised a prospective nation-wide Dutch cohort (n = 4875) and a retrospective regional cohort from Aberdeen, Scotland (n = 975). We excluded couples who had tried for <1 year to conceive and also those with anovulation, uni- or bilateral tubal occlusion, mild or severe endometriosis or male subfertility i.e. impaired semen quality according to World Health Organization criteria. PARTICIPANTS/MATERIALS, SETTING, METHODS: We matched couples who received IVF and couples on expectant management based on their characteristics to control for confounding. We fitted a Cox proportional hazards model including patient characteristics, IVF treatment and their interactions to estimate the individualized chance of conception over 1 year-either following IVF or expectant management for all combinations of patient characteristics. The endpoint was conception leading to ongoing pregnancy, defined as a foetus reaching a gestational age of at least 12 weeks. MAIN RESULTS AND THE ROLE OF CHANCE: The adjusted 1-year chance of conception was 47.9% (95% CI: 45.0-50.9) after IVF and 26.1% (95% CI: 24.2-28.0) after expectant management. The absolute difference in the average adjusted 1-year chances of conception was 21.8% (95%CI: 18.3-25.3) in favour of IVF. The effectiveness of IVF was influenced by female age, duration of subfertility and previous pregnancy. IVF was effective in women under 40 years, but the 1-year chance of an IVF conception declined sharply in women over 34 years. In contrast, in woman over 40 years of age, IVF was less effective, with an absolute difference in chance compared to expectant management of 10% or lower. Regardless of female age, IVF was also less effective in couples with a short period of secondary subfertility (1 year) who had chances of natural conception of 30% or above. LIMITATIONS, REASONS FOR CAUTION: The 1-year chances of conception were based on three cohorts with different sampling mechanisms. Despite adjustment for the three most important prognostic patient characteristics, namely female age, duration of subfertility and primary or secondary subfertility, our estimates might not be free from residual confounding. WIDER IMPLICATIONS OF THE FINDINGS: IVF should be used selectively based on judgements on gain compared to continuing expectant management for a given couple. Our results can be used by clinicians to counsel couples with unexplained subfertility, to inform their expectations and facilitate evidence-based, shared decision making. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by Tenovus Scotland [grant G17.04]. Travel for RvE was supported by the Amsterdam Reproduction & Development Research Group [grant V.000296]. SB reports acting as editor-in-chief of HROpen. Other authors have no conflicts.

Predicting the chances of having a baby with or without treatment at different time points in couples with unexplained subfertility.

STUDY QUESTION: Can we develop a prediction model that can estimate the chances of conception leading to live birth with and without treatment at different points in time in couples with unexplained subfertility? SUMMARY ANSWER: Yes, a dynamic model was developed that predicted the probability of conceiving under expectant management and following active treatments (in vitro fertilisation (IVF), intrauterine insemination with ovarian stimulation (IUI + SO), clomiphene) at different points in time since diagnosis. WHAT IS KNOWN ALREADY: Couples with no identified cause for their subfertility continue to have a realistic chance of conceiving naturally, which makes it difficult for clinicians to decide when to intervene. Previous fertility prediction models have attempted to address this by separately estimating either the chances of natural conception or the chances of conception following certain treatments. These models only make predictions at a single point in time and are therefore inadequate for informing continued decision-making at subsequent consultations. STUDY DESIGN, SIZE, DURATION: A population-based study of 1316 couples with unexplained subfertility attending a regional clinic between 1998 and 2011. PARTICIPANTS/MATERIALS, SETTING, METHODS: A dynamic prediction model was developed that estimates the chances of conception within 6 months from the point when a diagnosis of unexplained subfertility was made. These predictions were recomputed each month to provide a dynamic assessment of the individualised chances of conception while taking account of treatment status in each month. Conception must have led to live birth and treatments included clomiphene, IUI + SO, and IVF. Predictions for natural conception were externally validated using a prospective cohort from The Netherlands. MAIN RESULTS AND THE ROLE OF CHANCE: A total of 554 (42%) couples started fertility treatment within 2 years of their first fertility consultation. The natural conception leading to live birth rate was 0.24 natural conceptions per couple per year. Active treatment had a higher chance of conception compared to those who remained under expectant management. This association ranged from weak with clomiphene to strong with IVF [clomiphene, hazard ratio (HR) = 1.42 (95% confidence interval, 1.05 to 1.91); IUI + SO, HR = 2.90 (2.06 to 4.08); IVF, HR = 5.09 (4.04 to 6.40)]. Female age and duration of subfertility were significant predictors, without clear interaction with the relative effect of treatment. LIMITATIONS, REASONS FOR CAUTION: We were unable to adjust for other potentially important predictors, e.g. measures of ovarian reserve, which were not available in the linked Grampian dataset that may have made predictions more specific. This study was conducted using single centre data meaning that it may not be generalizable to other centres. However, the model performed as well as previous models in reproductive medicine when externally validated using the Dutch cohort. WIDER IMPLICATIONS OF THE FINDINGS: For the first time, it is possible to estimate the chances of conception following expectant management and different fertility treatments over time in couples with unexplained subfertility. This information will help inform couples and their clinicians of their likely chances of success, which may help manage expectations, not only at diagnostic workup completion but also throughout their fertility journey. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by a Chief Scientist Office postdoctoral training fellowship in health services research and health of the public research (ref PDF/12/06). B.W.M. is supported by an NHMRC Practitioner Fellowship (GNT1082548). B.W.M. reports consultancy for ObsEva, Merck, and Guerbet. None of the other authors declare any conflicts of interest.

Associations of preconception Body Mass Index in women with PCOS and BMI and blood pressure of their offspring.

Women with polycystic ovary syndrome (PCOS) have unfavorable metabolic profiles. Their offspring may be affected by such risks. The objective of the current study was to disclose associations between preconception health of these women and health of their offspring. 74 women diagnosed with PCOS according to the Rotterdam criteria were screened systematically before conception. Cardiovascular health of their offspring was assessed at 2.5-4 (n = 42) or at 6-8 years of age (n = 32). Multivariate linear regression analysis was performed with adjustments for potential confounders. In the primary analyses the association between preconception Body Mass index (BMI) and offspring BMI was evaluated. Secondly associations between preconception blood pressure, androgens, insulin-resistance (HOMA-IR), and LDL-cholesterol in women with PCOS and BMI and blood pressure of offspring were assessed. Results show that preconception BMI of women with PCOS was positively associated with sex- and age-adjusted BMI of their offspring at 6-8 years of age (ß = 0.55 (95% CI: 0.12 to 0.97), p = .012). No other significant associations were found. In conclusion, our data suggest that preconception BMI in PCOS is significantly associated with offspring BMI at 6-8 year of age. If this suggestion could be confirmed this may provide an opportunity for improving the future health of these children.

Is IUI with ovarian stimulation effective in couples with unexplained subfertility?

STUDY QUESTION: Does starting IUI with ovarian stimulation (IUI-OS) within 1.5 years after completion of the fertility workup increase ongoing pregnancy rates compared to expectant management in couples with unexplained subfertility? SUMMARY ANSWER: IUI-OS is associated with higher chances of ongoing pregnancy compared to expectant management in unexplained subfertile couples, specifically those with poor prognoses of natural conception, i.e. <15% over 6 months or <25% over 1 year. WHAT IS KNOWN ALREADY: IUI-OS is often the first-line treatment for couples with unexplained subfertility. Two randomized controlled trials compared IUI-OS to expectant management using different thresholds for the prognosis of natural conception as inclusion criteria and found conflicting results. A cohort of couples with unexplained subfertility exposed to expectant management and IUI-OS offers an opportunity to determine the chances of conception after both strategies and to evaluate whether the effect of IUI-OS depends on a couple's prognosis of natural conception. STUDY DESIGN, SIZE, DURATION: A prospective cohort study on couples with unexplained or mild male subfertility who could start IUI-OS at any point after completion of the fertility workup, recruited in seven Dutch centres between January 2002 and February 2004. Decisions regarding treatment were subject to local protocols, the judgement of the clinician and the wishes of the couple. Couples with bilateral tubal occlusion, anovulation or a total motile sperm count <1 × 106 were excluded. Follow up was censored at the start of IVF, after the last IUI cycle or at last contact and truncated at a maximum of 1.5 years after the fertility workup. PARTICIPANTS/MATERIALS, SETTING, METHODS: The endpoint was time to conception leading to an ongoing pregnancy. We used the sequential Cox approach comparing in each month ongoing pregnancy rates over the next 6 months of couples who started IUI-OS to couples who did not. We calculated the prognosis of natural conception for individual couples, updated this over consecutive failed cycles and evaluated whether prognosis modified the effect of starting IUI-OS. We corrected for known predictors of conception using inverse probability weighting. MAIN RESULTS AND THE ROLE OF CHANCE: Data from 1896 couples were available. There were 800 couples whom had at least one IUI-OS cycle within 1.5 years post fertility workup of whom 142 couples conceived (rate: 0.50 per couple per year, median follow up 4 months). The median period between fertility workup completion and starting IUI-OS was 6.5 months. Out of 1096 untreated couples, 386 conceived naturally (rate: 0.31 per couple per year, median follow up 7 months). Starting IUI-OS was associated with a higher chance of ongoing pregnancy by a pooled, overall hazard ratio of 1.96 (95% CI: 1.47-2.62) compared to expectant management. The effect of treatment was modified by a couple's prognosis of achieving natural conception (P = 0.01), with poorer prognoses or additional failed natural cycles being associated with a stronger effect of treatment. The predicted 6-month ongoing pregnancy rate for a couple with a prognosis of 25% at completion of the fertility workup over the next six cycles (~40% over 1 year) was 25% (95% CI: 21-28%) for expectant management and 24% (95% CI: 9-36%) when starting IUI-OS directly. For a couple with a prognosis of 15% (25% over 1 year), these predicted rates were 17% (95% CI: 15-19%) for expectant management and 24% (95% CI: 15-32%) for starting IUI-OS. LIMITATIONS, REASONS FOR CAUTION: The effect estimates are based on a prospective cohort followed up for 1.5 years after completion of the fertility workup. Although we balanced the known predictors of conception between treated and untreated couples using inverse probability weighting, observational data may be subject to residual confounding. The results need to be confirmed in external datasets. WIDER IMPLICATIONS OF THE FINDINGS: These results explain the discrepancies between previous trials that compared IUI-OS to expectant management, but further studies are required to establish the threshold at which IUI-OS is (cost-)effective. STUDY FUNDING/COMPETING INTEREST(S): This study was facilitated by (Grant 945/12/002) from ZonMW, The Netherlands Organization for Health Research and Development, The Hague, The Netherlands. B.W.M. is supported by a NHMRC Practitioner Fellowship (GNT1082548). B.W.M. reports consultancy for ObsEva, Merck and Guerbet. S.B. reports acting as Editor-in-Chief of HROpen. The other authors have no conflicts of interest.

Does endometrial scratching increase the rate of spontaneous conception in couples with unexplained infertility and a good prognosis (Hunault > 30%)? Study protocol of the SCRaTCH-OFO trial: a randomized controlled trial.

BACKGROUND: In the Netherlands, couples with unexplained infertility and a good prognosis to conceive spontaneously (i.e. Hunault > 30%) are advised to perform timed intercourse for at least another 6 months. If couples fail to conceive within this period, they will usually start assisted reproductive technology (ART). However, treatment of unexplained infertility by ART is empirical and can involve significant burdens. Intentional endometrial injury, also called 'endometrial scratching', has been proposed to positively affect the chance of embryo implantation in patients undergoing in vitro fertilization (IVF). It might also be beneficial for couples with unexplained infertility as defective endometrial receptivity may play a role in these women. The primary aim of this study is to determine whether endometrial scratching increases live birth rates in women with unexplained infertility. METHOD: A multicentre randomized controlled trial will be conducted in Dutch academic and non-academic hospitals starting from November 2017. A total of 792 women with unexplained infertility and a good prognosis for spontaneous conception < 12 months (Hunault > 30%) will be included, of whom half will undergo endometrial scratching in the luteal phase of the natural cycle. The women in the control group will not undergo endometrial scratching. According to Dutch guidelines, both groups will subsequently perform timed intercourse for at least 6 months. The primary endpoint is cumulative live birth rate. Secondary endpoints are clinical and ongoing pregnancy rate; miscarriage rate; biochemical pregnancy loss; multiple pregnancy rate; time to pregnancy; progression to intrauterine insemination (IUI) or IVF; pregnancy complications; complications of endometrial scratching; costs and endometrial tissue parameters associated with reproductive success or failure. The follow-up duration is 12 months. DISCUSSION: Several small studies show a possible beneficial effect of endometrial scratching in women with unexplained infertility trying to conceive naturally or through IUI. However, the quality of this evidence is very low, making it unclear whether these women will truly benefit from this procedure. The SCRaTCH-OFO trial aims to investigate the effect of endometrial scratching on live birth rate in women with unexplained infertility and a good prognosis for spontaneous conception < 12 months. TRIAL REGISTRATION: NTR6687 , registered August 31st, 2017. PROTOCOL VERSION: Version 2.6, November 14th, 2018.

External validation of a dynamic prediction model for repeated predictions of natural conception over time.

STUDY QUESTION: How well does a previously developed dynamic prediction model perform in an external, geographical validation in terms of predicting the chances of natural conception at various points in time? SUMMARY ANSWER: The dynamic prediction model performs well in an external validation on a Scottish cohort. WHAT IS KNOWN ALREADY: Prediction models provide information that can aid evidence-based management of unexplained subfertile couples. We developed a dynamic prediction model for natural conception (van Eekelen model) that is able to update predictions of natural conception when couples return to their clinician after a period of unsuccessful expectant management. It is not known how well this model performs in an external population. STUDY DESIGN, SIZE, DURATION: A record-linked registry study including the long-term follow-up of all couples who were considered unexplained subfertile following a fertility workup at a Scottish fertility clinic between 1998 and 2011. Couples with anovulation, uni/bilateral tubal occlusion, mild/severe endometriosis or impaired semen quality according to World Health Organization criteria were excluded. PARTICIPANTS/MATERIALS, SETTING, METHODS: The endpoint was time to natural conception, leading to an ongoing pregnancy (defined as reaching a gestational age of at least 12 weeks). Follow-up was censored at the start of treatment, at the change of partner or at the end of study (31 March 2012). The performance of the van Eekelen model was evaluated in terms of calibration and discrimination at various points in time. Additionally, we assessed the clinical utility of the model in terms of the range of the calculated predictions. MAIN RESULTS AND THE ROLE OF CHANCE: Of a total of 1203 couples with a median follow-up of 1 year and 3 months after the fertility workup, 398 (33%) couples conceived naturally leading to an ongoing pregnancy. Using the dynamic prediction model, the mean probability of natural conception over the course of the first year after the fertility workup was estimated at 25% (observed: 23%). After 0.5, 1 and 1.5 years of expectant management after the completion of the fertility workup, the average probability of conceiving naturally over the next year was estimated at 18% (observed: 15%), 14% (observed: 14%) and 12% (observed: 12%). Calibration plots showed good agreement between predicted chances and the observed fraction of ongoing pregnancy within risk groups. Discrimination was moderate with c statistics similar to those in the internal validation, ranging from 0.60 to 0.64. The range of predicted chances was sufficiently wide to distinguish between couples having a good and poor prognosis with a minimum of zero at all times and a maximum of 55% over the first year after the workup, which decreased to maxima of 43% after 0.5 years, 34% after 1 year and 29% after 1.5 years after the fertility workup. LIMITATIONS, REASONS FOR CAUTION: The model slightly overestimated the chances of conception by ~2-3% points on group level in the first-year post-fertility workup and after 0.5 years of expectant management, respectively. This is likely attributable to the fact that the exact dates of completion of the fertility workup for couples were missing and had to be estimated. WIDER IMPLICATIONS OF THE FINDINGS: The van Eekelen model is a valid and robust tool that is ready to use in clinical practice to counsel couples with unexplained subfertility on their individualized chances of natural conception at various points in time, notably when couples return to the clinic after a period of unsuccessful expectant management. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by a Chief Scientist Office postdoctoral training fellowship in health services research and health of the public research (ref PDF/12/06). There are no conflicts of interest.

Predicting the cumulative chance of live birth over multiple complete cycles of in vitro fertilization: an external validation study.

STUDY QUESTION: Are the published pre-treatment and post-treatment McLernon models, predicting cumulative live birth rates (LBR) over multiple complete IVF cycles, valid in a different context? SUMMARY ANSWER: With minor recalibration of the pre-treatment model, both McLernon models accurately predict cumulative LBR in a different geographical context and a more recent time period. WHAT IS KNOWN ALREADY: Previous IVF prediction models have estimated the chance of a live birth after a single fresh embryo transfer, thereby excluding the important contribution of embryo cryopreservation and subsequent IVF cycles to cumulative LBR. In contrast, the recently developed McLernon models predict the cumulative chance of a live birth over multiple complete IVF cycles at two certain time points: (i) before initiating treatment using baseline characteristics (pre-treatment model) and (ii) after the first IVF cycle adding treatment related information to update predictions (post-treatment model). Before implementation of these models in clinical practice, their predictive performance needs to be validated in an independent cohort. STUDY DESIGN, SIZE, DURATION: External validation study in an independent prospective cohort of 1515 Dutch women who participated in the OPTIMIST study (NTR2657) and underwent their first IVF treatment between 2011 and 2014. Participants underwent a total of 2881 complete treatment cycles, with a complete cycle defined as all fresh and frozen thawed embryo transfers resulting from one episode of ovarian stimulation. The follow up duration was 18 months after inclusion, and the primary outcome was ongoing pregnancy leading to live birth. PARTICIPANTS/MATERIALS, SETTING, METHODS: Model performance was externally validated up to three complete treatment cycles, using the linear predictor as described by McLernon et al. to calculate the probability of a live birth. Discrimination was expressed by the c-statistic and calibration was depicted graphically in a calibration plot. In contrast to the original model development cohort, anti-Müllerian hormone (AMH), antral follicle count (AFC) and body weight were available in the OPTIMIST cohort, and evaluated as potential additional predictors for model improvement. MAIN RESULTS AND THE ROLE OF CHANCE: Applying the McLernon models to the OPTIMIST cohort, the c-statistic of the pre-treatment model was 0.62 (95% CI: 0.59-0.64) and of the post-treatment model 0.71 (95% CI: 0.69-0.74). The calibration plot of the pre-treatment model indicated a slight overestimation of the cumulative LBR. To improve calibration, the pre-treatment model was recalibrated by subtracting 0.35 from the intercept. The post-treatment model calibration plot revealed accurate cumulative LBR predictions. After addition of AMH, AFC and body weight to the McLernon models, the c-statistic of the updated pre-treatment model improved slightly to 0.66 (95% CI: 0.64-0.68), and of the updated post-treatment model remained at the previous level of 0.71 (95% CI: 0.69-0.73). Using the recalibrated pre-treatment model, a woman aged 30 years with 2 years of primary infertility who starts ICSI treatment for male factor infertility has a chance of 40% of a live birth from the first complete cycle, increasing to 72% over three complete cycles. If this woman weighs 70 kg, has an AMH of 1.5 ng/mL and an AFC of 10 measured at the beginning of her treatment, the updated pre-treatment model revises the estimated chance of a live birth to 30% in the first complete cycle and 59% over three complete cycles. If this woman then has five retrieved oocytes, no embryos cryopreserved and a single fresh cleavage stage embryo transfer in her first ICSI cycle, the post-treatment model estimates the chances of a live birth at 28 and 58%, respectively. LIMITATIONS, REASONS FOR CAUTION: Two randomized controlled trials (RCT) evaluating the effectiveness of gonadotropin dose individualization on basis of the AFC were nested within the OPTIMIST study. The strict dosing regimens, the RCT in- and exclusion criteria and the limited follow up time of 18 months might have influenced model performance in this independent cohort. Also, consistent with the original model development study, external validation was performed using the optimistic assumption that the cumulative LBR in couples who discontinue treatment without a live birth would have been equal to that of those who continue treatment. WIDER IMPLICATIONS OF THE FINDINGS: After national recalibration to account for geographical differences in IVF treatment, the McLernon prediction models can be introduced as new counselling tools in clinical practice to inform patients and to complement clinical reasoning. These models are the first to offer an objective and personalized estimate of the cumulative probability of a live birth over multiple complete IVF cycles. STUDY FUNDING/COMPETING INTEREST(S): No external funds were obtained for this study. M.J.C.E., D.J.M. and S.B. have nothing to disclose. J.A.L, S.C.O, T.C.v.T. and H.LT. received an unrestricted personal grant from Merck BV. B.W.M. is supported by a NHMRC Practitioner Fellowship (GNT1082548) and reports consultancy for ObsEva, Merck and Guerbet. F.J.M.B. receives monetary compensation as a member of the external advisory board for Merck BV (the Netherlands) and Ferring pharmaceutics BV (the Netherlands), for consultancy work for Gedeon Richter (Belgium) and Roche Diagnostics on automated AMH assay development, and for a research cooperation with Ansh Labs (USA). TRIAL REGISTRATION NUMBER: Not applicable.