1 Prevalence/Incidence of Infertility

Infertility is defined by WHO-ICMART as a disease of the reproductive system defined as the failure to achieve a clinical pregnancy after 12 of more months of regular unprotected sexual intercourse. As all women lose their ability to conceive with age, this definition only holds within the womens’ reproductive age-span between the menarche and menopause. On average the woman’s fertility is highest in the early and mid-twenties, then slowly declines and drops even faster at 35 years and older.

According to a Scottish study, nearly 20% of couples within their reproductive lifespan failed to conceive within 12 months and 12% failed to conceive within 24 months [1]. A Spanish cross-sectional study found that pregnancy was not achieved within 6, 12 and 24 months of starting to attempt conception in 20%, 11% and 4.4% of women, respectively [2].

Couples that did conceive in the end were apparently not really infertile. In view of the reduced fertility with prolonged time of unwanted non-conception, these couples can better be referred to as subfertile [3].

Couples can most likely be considered infertile when failing to conceive within five years. A worldwide systematic review and meta-analysis concluded that within a period of five years, 1.9% of women aged 20–44 years who wanted to have children were unable to have their first live birth (primary infertility), and 10.5% of women with a previous live birth were unable to have an additional live birth (secondary infertility) [4]. This study found that levels of infertility did not change significantly between 1990 and 2010.

2 You Are Planning to Do a Fertility Study

If you are planning a study in the field of fertility or any other research field, you first need a clear-cut research question. The FINER criteria of Hull and co-authors can help here [5]. Is your study question or aim Feasible, Interesting, Novel, Ethical and Relevant? Does it include a clear-cut outcome? If yes, proceed with your proposal and think with what design your question can most likely be answered. According to the hierarchy of evidence, the most reliable evidence comes from randomised clinical trials, followed by cohorts, case-control studies, cross-sectional studies and case studies.

3 Study Design

Cohorts, cross-sectional and case-control studies have observational designs. Case-control and cross-sectional studies are retrospective in nature, meaning that the study looks backward, that is, data are based on events that have already happened. Cohort studies can also be prospective, meaning that the data was collected from the time point at the start of the cohort on until a certain period of time. A prospective study watches for outcomes, such as the development of a disease, during the study period and relates this to other factors such as suspected risk or protection factors.

For all these observational studies, all efforts should be made to avoid sources of bias such as the loss of individuals to follow up during the study. For retrospective studies, bias and confounding can never be completely controlled for as the database was not specifically made for your question.

Both cohorts and case-control studies can be used to evaluate risk factors and diagnostic tests. Though a cohort is generally a better design than a case-control, for cohorts the outcome of interest should be common; otherwise, the number of outcomes observed will be too small to be statistically meaningful. When the events under study are rare, a case-control study is the better alternative. Cross-sectional studies take a sample of the population (the cross-section) at one time point. This design is well set to determine prevalence and can be used to study associations when development over time is less important. Cross-sectional studies are relatively quick and easy.

A recently published example of a retrospective cohort evaluated the association between vanishing twin syndrome (VTS) and adverse perinatal outcome in 253,000 singleton deliveries [6]. Though statistical analysis included multiple logistic regression models to control for possible confounders, the study was retrospective, that is, data were not collected for this specific question. We will not be sure whether VTS was more often diagnosed in the more complicated pregnancies and more often missed in uncomplicated pregnancies. This is a typical association study. The observed association between VTS and adverse perinatal outcome does not imply that there is a causal relationship.

An example of a case-control study is a recently published study that evaluated whether in a group of women with heart disease (cases) and in healthy women (controls) the second pregnancy was less complicated and resulted in a larger baby [7]. As such cases will be less common, the case-control design is likely to be the most feasible option here.

Observational studies do not try to influence anything but just measure what happened; there is no intervention by the researcher. As a result observational studies never provide information on causation but only on a possible association. In experimental studies, like a controlled clinical trial, the researcher intervenes to change something (e.g., gives some patients a drug) and then observes what happens. As the variables under study are controlled for, we can make conclusions on causation. To evaluate the effectiveness of a new intervention the experimental design needs to be used. A cohort study will not be able to detect a direct relationship between the intervention and treatment success.

For scientific evaluation of treatment, the randomised clinical trial (RCT) is widely accepted as the gold standard. Data from clinical trials are considered to represent the highest level of evidence that can be used to inform about the effectiveness of treatment strategies. The parallel design is the commonest trial design, involving a comparison between two groups, an experimental versus a control group. As this type of trial is the most easily understood by researchers as well as patients, this is the simplest trial design and a lot of examples in fertility, obstetrics and gynaecological research can be given. A trial normally has two arms, one intervention arm and a control arm. Occasionally a trial may have three arms, for example, aspirin + low molecular weight heparin (LMWH) versus aspirin versus placebo for the treatment of unexplained recurrent pregnancy loss (RPL) [8] or in vitro fertilisation (IVF)–single embryo transfer (SET) versus IVF in a modified cycle versus intrauterine insemination (IUI) in women with unexplained subfertility [9].

There is a great need for well-performed trials in all areas of clinical medicine. With the results of these trials clinicians will be able to make better and more evidence-based decisions about diagnostics and treatment of their patients. ‘Evidenced-based medicine’ (EBM), defined in 1992 by Gyatt as ‘the integration of individual clinical expertise with the best available external evidence from systematic research’, is considered the cornerstone of decision-making in current clinical medicine [5].

To make a randomised trial feasible and to increase the quality, issues like recruitment, loss-to-follow-up, power/sample size, randomisation methods, data-collection, statistics and ethics need to have been adequately addressed before starting the trial. Doing a randomised study has costs. It is advisable to work together with other clinical centres as multicentre trials are generally more successfully completed.