Letters to the editor |
Routine use of hysterosalpingography prior to diagnostic laparoscopy in the fertility workup
1 Department of Obstetrics and Gynaecology, Máxima Medical Centre, Veldhoven 2 Department of Clinical Epidemiology & Biostatistics and 3 Centre for Reproductive Medicine, Academic Medical Centre, Amsterdam, The Netherlands
4 To whom correspondence should be addressed at: Department of Obstetrics and Gynaecology, Máxima Medical Centre, P.O. Box 7777, 5500 MB Veldhoven, The Netherlands. E-mail: s.f.coppus{at}amc.uva.nl
Sir,
In the May issue of your journal, Perquin and colleagues report the results of a randomized controlled trial of hysterosalpingography versus laparoscopy early in the fertility workup (Perquin et al., 2006
). Subfertile patients selected for tubal patency testing were randomly allocated to one of two strategies. The first strategy was hysterosalpingography (HSG) followed by diagnostic laparoscopy (DLS) within 1–2 months in case tubal pathology was suspected or 6 months later in case HSG showed no abnormalities and the patient had not become pregnant by that time. The second strategy was to perform DLS immediately. The authors conclude from their trial that routine use of HSG at an early stage in the infertility workup does not influence cumulative pregnancy rates.
We believe that this conclusion is not warranted by the design and the results of the trial. Generally, clinical outcomes are not affected by the test under evaluation but by the intervention that follows a normal or abnormal test result. As such, randomized comparisons of tests can only be validly evaluated in terms of patient outcomes in case an abnormal test result is followed by a standardized and explicit treatment protocol (Bossuyt et al., 2000). In the present study, the authors have allowed a variety of protocols and decisions after both normal and abnormal HSG and laparoscopy. For this reason, it did not come as a surprise that the authors failed to demonstrate a clinically relevant difference between the diagnostic strategies in terms of cumulative pregnancy rate. In the gynaecological literature, several other examples can be found of randomized controlled trials of diagnostic tests or strategies in which the lack of explicit test-treatment decision rules in the study design can be held responsible for a failure to demonstrate health-related benefits from using the test under study (Nienhuis et al., 1997;Oei et al., 1998).
Furthermore, the statistical power of the reported study is probably insufficient to draw this conclusion from the available data. Even if the present study would have used explicit treatment rules, a difference in clinical outcome could only result from patients with discordant test results, as patients with the same test result would obtain the same treatment protocol. Unfortunately, the proportion of discordant test results cannot be inferred from the data provided. Hypothetically, it may be possible that all abnormal HSGs were normal at DLS and vice versa, which would result in a 21% discordance rate. The other extreme would be that all abnormalities detected by HSG were confirmed by DLS, which would result in a 3% concordance rate. Nevertheless, whatever this rate would be, it is clear that between 79 and 97% of the patients who have been included in the study do not contribute to a difference in pregnancy rate and that this large group will dilute any such difference. Larger sample sizes are therefore needed to reliably estimate differences in outcome between testing strategies.
The only conclusion that can be made from this study is that a strategy starting with HSG can prevent DLS in 30% of women. From the data provided, we would therefore prefer a strategy starting with HSG followed by laparoscopy when needed, as this strategy reduces the laparoscopy rate without compromising the pregnancy rate.
In conclusion, when conducting randomized controlled trials of diagnostic tests or strategies, it is not possible to show clinical benefits from the use of the test if normal and abnormal test results are not followed by explicit treatment rules. In fertility care, diagnostic trials without explicit treatment rules will inevitably result in equal pregnancy rates in both groups and should therefore in our opinion be avoided.
References
Bossuyt PM, Lijmer JG, Mol BW. (2000) Randomised comparisons of medical tests: sometimes invalid, not always efficient. Lancet 356:1844–1847.[CrossRef][Web of Science][Medline]
Nienhuis SJ, Vles JS, Gerver WJ, Hoogland HJ. (1997) Doppler ultrasonography in suspected intrauterine growth retardation: a randomized clinical trial. Ultrasound Obstet Gynecol 9:6–13.[Medline]
Oei SG, Helmerhorst FM, Bloemenkamp KW, Hollants FA, Meerpoel DE, Keirse MJ. (1998) Effectiveness of the postcoital test: randomised controlled trial. BMJ 317:502–505.
Perquin DA, Dorr PJ, Craen AJ, Helmerhorst FM. (2006) Routine use of hystero-salpingography prior to laparoscopy in the fertility workup: a multicentre randomized controlled trial. Hum Reprod 21:1227–1231.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||