Objectives The authors aim to calculate the number of live births, before and after systemic lupus erythematosus (SLE) diagnosis, in women diagnosed during their reproductive years and to compare this with general population rates.
Methods The authors identified women with SLE using Quebec administrative databases (1 January 1994 to 31 December 2003). The authors determined the number of live births, and calculated the standardised incidence ratio (SIR) of observed to expected live births.
Results 1334 women with SLE were identified. Overall, the number of live births over the interval (559) was below that which would be expected (708) (SIR 0.79; 95% CI 0.73 to 0.86). Compared with the general population, live births were substantially lower after SLE diagnosis (SIR 0.62; 95% CI 0.55 to 0.70) than before diagnosis (SIR 1.01; 95% CI 0.90 to 1.13).
Conclusion After diagnosis, women with SLE have substantially fewer live births than the general population.
Statistics from Altmetric.com
Upon being diagnosed as having systemic lupus erythematosus (SLE), women of reproductive age often want to know if their disease will limit their ability to have children. Up to now, research has focused on a limited but important aspect of this question—estimating the proportion of SLE pregnancies ending in live births. One further step in providing an adequate answer would be to assess live birth rates in women with SLE.
The live birth rate (also known as the fertility rate) is a useful demographic statistic and is defined as the number of live births per 1000 women of reproductive age (usually between 15 and 45 years old) in a given year.1 The live birth rate is thus influenced by two key reproductive outcomes: the number of pregnancies and the number of live births in a group of women. Multiple disease-related factors may limit live birth rates in women with SLE (see online supplementary text).2
The standardised incidence ratio (SIR), which is the ratio of the observed number of events in a sample divided by the expected number of events, allows us to directly compare the live birth rate in women with SLE to the live birth rate in the general population.3 As the live birth rate in the general population is usually recorded according to age group and calendar time, the SIR offers direct comparison of women with SLE, at different reproductive stages and born at different periods, to the general population by providing a summarised measure. Estimating such a measure should definitively answer whether having SLE influences the number of children affected women have.
Previously, we have performed a study determining the SIR of live births in women from an inception cohort of SLE,4 using general population rates as a reference. We observed that women diagnosed as having SLE during their reproductive period had substantially decreased live birth rates compared to the general population (SIR 0.65; 95% CI 0.58 to 0.73). However, this study was limited in that we had information only on date of first birth, not on dates of subsequent births. Thus, we were unable to estimate a SIR for live births specifically before and after SLE diagnosis.
To further investigate if live birth rates are reduced after SLE diagnosis, we performed a population-based study using administrative databases. Our primary objectives were to calculate the number of live births, before and after SLE diagnosis, in a cohort of women diagnosed as having SLE during their reproductive years and to compare this with general population rates using SIR.
Subjects and methods
We identified women with SLE using Quebec administrative databases (Med-Echo and Régie de l’assurance maladie du Québec (RAMQ) physician billing databases, 1 January 1994 to 31 December 2003), which cover all healthcare beneficiaries. Incident SLE cases were women with ≥1 hospitalisation with either a primary or a secondary diagnosis of SLE or ≥2 physicians' claims for SLE within any 2-month-to-2-year period, with no prior diagnosis of SLE in the 5 years preceding the interval. To assess women diagnosed as having SLE during their reproductive period, we only included women aged 15–35 years on 1 January 1994.
We determined the number of live births during the interval as defined by diagnostic and procedure codes for delivery in the Med-Echo and RAMQ physician databases, respectively. As mentioned previously, the SIR is the ratio of the observed number of live births in a sample divided by the expected number of live births. We determined the expected number of live births as follows: we summed the years of follow-up from the subjects' age at the start of the study interval up to the age of 45 years (or the oldest age attained at the end of the study period). For subjects who died during the interval, years of follow-up were summed up to the time of death. We applied age-specific general population birth rates, for the relevant calendar periods, to these years of follow-up to obtain the expected number of births. We then calculated the SIR of observed to expected live births for the overall study interval, as well as before and after SLE diagnosis.
We performed a multivariate Poisson regression to explore potential predictors of live births in women with SLE. Time-dependent predictors, assessed at the time of delivery, included prior hospitalisation with a primary diagnosis of SLE, renal disease (ie, RAMQ billing code for renal biopsy), antiphospholipid syndrome (ie, ICD-9 (International Classification of Diseases, 9th Revision) code for antiphospholipid antibodies and/or any thromboembolic events in either databases), disease duration ≥5 years and residence in rural regions (ie, census <10 000 inhabitants). Because renal disease was defined using renal biopsy, which almost always requires hospitalisation, we included an interaction term between renal biopsy and hospitalisation for SLE in the multivariate model.
We corrected our model for clustering of reproductive outcomes and performed a sensitivity analysis using first births (see online supplementary text).5
The McGill University Research ethics board approved this study.
One thousand three hundred and thirty-four women with SLE were identified (table 1). Overall, the number of live births over the interval (559) was below that which would be expected (708) (SIR 0.79; 95% CI 0.73 to 0.86) (table 2). Compared with the general population, live births were substantially lower after SLE diagnosis (SIR 0.62; 95% CI 0.55 to 0.70) than before diagnosis (SIR 1.01; 95% CI 0.90 to 1.13).
In multivariate analyses of potential predictors of live births in women after SLE diagnosis (table 3), prior hospitalisation for SLE (RR 0.49; 95% CI 0.35 to 0.68) was associated with markedly decreased live births. There were trends for fewer live births in women with disease duration ≥5 years (RR 0.89; 95% CI 0.67 to 1.18) and in those living in rural regions (RR 0.83; 95% CI 0.61 to 1.13).
We did not definitively observe a decrease in live births independently attributable to age at SLE diagnosis ≥30 years (RR 1.10; 95% CI 0.81 to 1.47), antiphospholipid syndrome (RR 0.91; 95% CI 0.65 to 1.29) or renal disease (RR 0.84; 95% CI 0.34 to 2.04). In addition, we did not establish any interaction between renal disease and prior hospitalisation for SLE (RR 1.95; 95% CI 0.72 to 5.31). The multivariate analysis restricted to first births gave similar results, confirming the precision of our model for all births.
We observed that women diagnosed as having SLE during their reproductive period have substantially decreased live birth rates after diagnosis compared to the general population. We did not observe decreased live births in women with SLE prior to diagnosis. The most important predictor of decreased live births after SLE diagnosis was prior hospitalisation for SLE, which likely indicates more severe/active disease.
We were unable to establish an independent association with renal disease, traditionally recognised as a marker of disease activity/severity,6 on live birth rates. The definition we used for renal disease relied on renal biopsy, which often requires hospitalisation. Since the primary discharge diagnosis in patients hospitalised to undergo a renal biopsy may have been SLE, the renal disease effect estimate may have been intermingled with the effect estimate for prior SLE hospitalisation. To account for this possibility, we included in our model an interaction term for renal disease and prior hospitalisation, but it failed to demonstrate an effect of renal disease, with or without prior hospitalisation, on live birth rates. Although our definition of renal disease aimed for high specificity, likely capturing patients with severe nephritis, it may have lacked sensitivity, potentially missing milder forms of nephritis. Only 11% of our SLE cohort was identified as having renal disease; this limited our power to find a small effect.
We were unable to demonstrate a decrease in live births due to antiphospholipid syndrome. No ICD-9 code exists for this syndrome, and no claim-based definition has been validated. Thus, our definition, based on antiphospholipid antibodies and/or thromboembolic events, may have lacked specificity, biasing our effect estimate towards the null value due to non-differential misclassification.
We observed potentially decreased live births in women living in rural regions. This may be explained by limited healthcare accessibility, resulting in adverse pregnancy outcomes, deliberate decision to avoid pregnancy and/or inappropriate counselling on pregnancy. Moreover, this difference may reflect variations in racial distribution between urban and rural regions in Quebec, the latter being predominantly populated by Caucasians.7 Since Caucasians are known to have one of the lowest live birth rates8 and since our multivariate analysis did not account for race, as this variable was not present in the database, this may explain reduced RR for live births in women from rural regions. However, it is unlikely that failure to adjust for race would explain our primary results, the SIR for live births before and after diagnosis. Indeed, in our previous study,4 decreased live birth rates in women with SLE (compared to the general population) persisted after applying race-specific birth rates. Furthermore, non-Caucasian groups such as African–Americans and Hispanics, who have increased live birth rates compared to Caucasians,8 are generally over-represented in North American SLE cohorts (due to their higher rate of SLE).9 10 Thus, if anything, our results are likely conservative (see online supplementary text for further discussion of limitations).
In conclusion, our findings indicate that live birth rates are substantially reduced (compared to the general population) after diagnosis in women with SLE and that disease-related factors, such as prior hospitalisation for SLE, potentially play an important role. These results prompt future research to further characterise disease-related, demographic and psychosocial factors contributing to decreased live birth rates in women with SLE.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Funding This research received no specific funding.
Competing interests None.
Ethics approval Ethics approval was obtained from the McGill University ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.