Pre-eclampsia occurs in 3% to 7% of all deliveries, 1–5 and is a frequent cause of maternal and infant morbidity and mortality. The etiology of pre-eclampsia is unknown, but there is evidence that a maternal immune reaction related to paternal factors may be part of the mechanism. 6 Several studies indicate that the incidence of pre-eclampsia in a second pregnancy with a new partner is almost as high as that of the first pregnancy, 1,5,6,7 where it is most frequent.
Women inseminated with external donor sperm have a higher risk of pre-eclampsia than women inseminated with their partner’s sperm. 4 Men whose partner had had pre-eclampsia are more likely to father a child with pre-eclampsia also with a new woman. 2
Since the risk declines in successive pregnancies in women in stable relationships, there may be an attenuation of the immune response over time. 8 Pregnancies and previous abortions, 6,7,9 especially late ones 10 are reported to be protective. A fetal, intrauterine, or genetic effect on the maternal side is expected to play a role since women who had pre-eclampsia have a high recurrence risk 1,2,10 and daughters of women with pre-eclampsia have themselves an increased risk. 2,11,11a There is, however, evidence that men also genetically contribute to the risk of pre-eclampsia 2,11a
The interval between births is correlated with partner change, time of cohabitation, change in behavioral and environmental factors, and perhaps also with the “responsiveness” of the immune system. If subfecundity is a determinant of pre-eclampsia, 12,13 this interval will confound the effect of paternal change. In general, the interpregnancy interval will have a very different meaning in women who change partner compared with women who do not.
We evaluated the role of the interval between births in the association between pre-eclampsia in second birth and change of partner.
From the Danish National Board of Health we obtained hospital discharge records of all women hospitalized with eclampsia/pre-eclampsia [discharge diagnoses: 637.03, 637.04 637.09, 637.19 (ICD-8), and O14 to O15 (ICD-10)] in Denmark between 1980 and 1994 and who had subsequently given birth. We excluded records where the most recent diagnosis had the additional coding “not found” or “under observation.”
As the study population of women without pre-eclampsia in first birth we combined all cases of pre-eclampsia in second (but not first) birth with pairs of first and second births in the same time period obtained from a random sample of 60,810 mothers.
We excluded multiple births and births where the reported gestational age was 27 completed weeks or less, as no stillbirths were included in the Registry until week 28. We kept births with missing gestational age but checked whether their exclusion had an effect on the results. Non-Danish citizens were also excluded, as information on parity and social status was missing for many of them. We also excluded 208 pairs where one of the children had been given in adoption.
We further excluded 90 mothers from the cohort of women with pre-eclampsia in the first birth and 255 from the cohort with no pre-eclampsia in the first birth, as information on the father was missing in one or both of the births. The cohort of the 8,401 women with a diagnosis of pre-eclampsia in their first birth who had a subsequent birth during follow-up was defined as the “PE+ cohort.” The study population of women with no pre-eclampsia in first birth was defined as the “PE− cohort” (26,596 women), although it was not properly a cohort.
The risk of pre-eclampsia in second birth was estimated within each cohort according to whether the biological father had changed or not from the previous birth. Women who changed partner also changed municipality more often, which may involve both a change in environmental factors and in detection probability. We repeated the main analyses restricting to women who remained in the same municipality between the two births.
Data on paternity and social status were obtained through linkage with Statistics Denmark’s Fertility Database, 14 which covers the Danish population in the fertile age and any child with at least one registered parent with a permanent address in Denmark at the 1st of January 1980.
We categorized social status into three levels from a ten-point classification system used by Statistics Denmark according to the occupation held at the 1st of January of the year in which the child was born. The low social class included unemployed, retired, unskilled manual workers, and unspecified or unknown job levels. In the middle category were office workers, students, skilled manual workers and those assisting the spouse. Finally, in the high category were all high ranking office workers, managers, self-employed office workers and small enterprise or shop owners. The partner who held the highest level determined the social status of the couple.
We defined the interpregnancy interval as the time between the birth of the first child in the pair and the estimated conception date of the second child, omitting pregnancies not leading to birth. Information on smoking during the first trimester was included in the Birth Registry starting from 1991 and we thus repeated some of the analyses on the women where we had information on smoking. We attempted to estimate the paternal effect on the risk of pre-eclampsia and to evaluate its relation with the interval between two births by applying different logistic regression models.
Table 1 reports some estimates of pre-eclampsia from the background sample of the general population. The incidence of pre-eclampsia was relatively stable in time and was overall around 3% among births that reached the 28th gestational week.
Table 2 illustrates some characteristics of the two cohorts. The recurrence of pre-eclampsia in the PE+ cohort was high in both women who did or did not change partner. Since the sampling of the PE+ cohort was of the case-control type, we also present estimates based upon the fraction belonging to the random sample. In this cohort the background risk in the population of pre-eclampsia in second birth was 1.1%. Women who changed partner were younger and of lower social status in both cohorts. They also moved to a different municipality more frequently.
Figure 1 shows that a much longer interpregnancy interval was associated with change of partner in both cohorts. Figure 2 reports the unadjusted and adjusted (for maternal age and social status) odds ratios of pre-eclampsia in second birth according to the length of the interpregnancy interval in women who did not change partner. There was virtually no difference between the unadjusted and adjusted estimates, except for long intervals in the PE− cohort. In this cohort, which had a very low background risk, there was an increasing risk of pre-eclampsia in second birth as the interpregnancy interval became longer. In the PE+ cohort no such trend was present.
In Table 3 we report the estimated effect of paternal change on pre-eclampsia according to different logistic regression models. The top half of the table reports results from the PE+ cohort and the bottom half from the PE− cohort. The right-hand column reports the estimates restricted to women who remained in the same municipality.
Very little, if any, paternal effect can be seen in the PE+ cohort, and the results differ to some extent in women remaining in the same municipality. When stratifying for the interpregnancy interval however, it appears that changing partner may be protective in short and long intervals.
In the PE− cohort the models not including the term for interpregnancy interval suggested an increased risk of pre-eclampsia after change of partner. This effect did not persist when adjusting for the interpregnancy interval. When stratifying by the interval, however, women changing partner and conceiving within three years appeared to have an increased risk of pre-eclampsia, although this effect was mostly seen in women remaining in the same municipality. We saw no adverse effect of paternal change among women waiting longer than three years before having the second child. The interval 3 to 5 years had the lowest risk of pre-eclampsia associated with paternal change, the opposite of what we saw in the PE+ cohort.
There were 13,752 pairs where the second child was born between 1991 and 1994. We disregarded pairs where information on smoking during the first trimester of pregnancy (yes/no) was missing (7.5%). The prevalence of smoking in the first trimester was 23.8% in the PE+ cohort and 30.5% in the PE− cohort. First trimester smoking was strongly associated with having changed partner in both cohorts (approximately 20% of smokers had changed partner as opposed to 8% of non-smokers). Smoking was also associated with low social status, young age, and an interval longer than three years.
After adjustment for maternal age, change of partner, change of municipality, and social status, smoking was associated with a marked decline in the risk of pre-eclampsia in the PE− cohort (OR = 0.61; 95% CI: 0.51–0.72) but not in the PE+ cohort (OR = 0.87; 95% CI: 0.69–1.11). The effect of smoking on pre-eclampsia appeared to be the same regardless of change of partner and was independent on whether the interpregnancy interval was adjusted for or not.
In Table 4 we present the same models as in Table 3 but stratified by smoking. The most remarkable findings were (i) the low odds ratios among women in the PE+ cohort who changed partner and had a second child within 3 years, and (ii) the possible modification of smoking on the effect of partner change in the same interval in women in the PE− cohort.
We had some information on induced and spontaneous abortions occurring between the first and second birth. One abortion of either type did not appear to influence the risk of pre-eclampsia in any way, but two or more spontaneous abortions increased the risk of pre-eclampsia in both cohorts and regardless of change of partner. Too few women reported two or more induced abortions to be able to provide meaningful results. Although the risk of abortions increased with time, adjusting for abortions did not appear to change the estimates of partner change. The odds ratios of pre-eclampsia were 0.89 (95% CI: 0.77–1.03) in the PE+ cohort with no adjustment for the interpregnancy interval and 0.84 (95% CI: 0.67–1.05) after adjustment. Analogously, the estimates were 1.12 (0.98–1.29) and 0.90 (0.73–1.12) in the PE− cohort.
When excluding pairs where gestational age was missing in any of the births (3.4% in the PE+ cohort and 3.8% in the PE− cohort), the odds ratios associated with change of partner changed only marginally. In the PE+ cohort they were 0.87 (95% CI: 0.69–1.08) before adjustment for the interpregnancy interval and 0.80 (0.63–1.00) after adjustment. The corresponding estimates in the PE− cohort were 1.10 (0.95–1.26) and 0.84 (0.72–0.98), respectively.
In this study we observed that the interpregnancy interval had a very different distribution between women who changed partner and women who did not. In the latter group a long interval was associated with an increased risk of pre-eclampsia in second birth among women with no previous history of pre-eclampsia. This feature makes the interpregnancy interval a candidate confounder or effect modifier of the association between primipaternity and pre-eclampsia in parous women. Women who changed partner also moved more frequently and had a different lifestyle and socio-economic status compared with women in stable relationships. All those factors are liable to have an impact on the risk of pre-eclampsia, but time represents an especially difficult aspect to disentangle. Not only does time correlate with the length of sexual cohabitation, subfecundity, and change in diagnostic criteria, but a number of other putative risk factors may have changed over time, interfering with the association between paternal change and pre-eclampsia. In general, the interpregnancy interval is a different determinant in women who change partner compared with women who do not.
Unfortunately, we did not have sufficient data to address effect modification between the interpregnancy interval and change of partner. When limiting the analysis to women conceiving their second child within three years after the birth of the first we saw that women remaining in the same municipality after a change of partner had a reduced risk of pre-eclampsia if they belonged to the PE+ cohort, and a higher risk if they belonged to the PE− cohort, as previously reported. 1,5 The interval between three and five years showed unexpected results, which could be due to chance. It is possible that the effect of a new father may only be found shortly after a previous pregnancy, especially in women who had not had pre-eclampsia in the first birth. We do not know the length of sexual cohabitation but we expect that women who changed partner and had a new child within a short time had a short sexual cohabitation with the new partner and therefore no time to develop an immunity to paternal antigens. 6,8 At least a fraction of the women who waited for a longer time had had the time to develop such immunity, which would explain lower odds ratios in the strata with a longer interval in the PE- cohort, but not the finding for the interval between three and five years in the PE+ cohort.
For the period where we had information on smoking during the first trimester we saw the expected protective effect of smoking on pre-eclampsia in the PE− cohort, 15 and the most surprising finding was the apparent effect modification of smoking on partner change in women with an interval of less than three years. While this finding could easily be due to chance because of the small numbers in the strata, it is also possible that a negative effect of partner change in the PE− cohort may be counterbalanced by a beneficial effect of smoking among women who changed partner, as we saw that smoking and changing partner were strongly correlated.
We had to rely upon the hospital record for the diagnosis of pre-eclampsia, but we tried to exclude all uncertain diagnoses to reduce the number of false positives. We saw that one city with a large university hospital (Aarhus) had a lower incidence of pre-eclampsia compared with the rest of the country. Excluding women living in Aarhus at the time of the second birth did not change the results, and women living in this municipality contributed the same proportion to the PE+ and PE− cohorts despite the difference in incidence and/or threshold of diagnosis.
Most women in Denmark give birth in a hospital and it is very likely that most cases would be reported. Our estimates of the incidence of pre-eclampsia 1–5 and of a decreased risk among smokers are furthermore consistent with the literature. 15 In our study we were able to identify virtually all reported cases of pre-eclampsia in Denmark over a period of 15 years with a small proportion of unidentified fathers, although we expect some misclassification of paternity, 16 which would be a problem for all studies based upon routine registration. We realize that in this paper we provide more questions than answers, but we do show that epidemiologic studies should take into account the interval between two births when evaluating the effect of paternal change on the risk of pre-eclampsia.
1. Li DK, Wi S. Changing paternity and the risk of preeclampsia/eclampsia in the subsequent pregnancy. Am J Epidemiol. 2000; 151 (1): 57–62.
2. Lie RT, Rasmussen S, Brunborg H, Gjessing HK, Lie-Nielsen E, Irgens LM. Fetal and maternal contributions to risk of pre-eclampsia: population based study. BMJ. 1998; 316 (7141): 1343–1347.
3. Van Pampus MG, Dekker GA, Wolf H, Huijgens PC, Koopman MMW, von Blomberg ME, Boüller HR. High prevalence of hemostatic abnormalities in women with a history of severe pre-eclampsia. Am J Obstet Gynecol 1999; 180: 1146–1150.
4. Smith GN, Walker M, Tessier JL, Millar KG. Increased incidence of preeclampsia in women conceiving by intrauterine insemination with donor versus partner sperm for treatment of primary infertility. Am J Obstet Gynecol. 1997; 177 (2): 455–458.
5. Trupin LS, Simon LP, Eskenazi B. Change in paternity
: a risk factor for preeclampsia in multiparas. Epidemiology. 1996; 7 (3): 240–244.
6. Dekker GA, Robillard PY, Hulsey TC. Immune maladaptation in the etiology of preeclampsia: a review of corroborative epidemiologic studies. Obstet Gynecol Surv. 1998; 53 (6): 377–382.
7. Dekker GA. Risk factors for pre-eclampsia. Clin Obstet Gynecol 1999; 42: 422–435.
8. Robillard PY, Hulsey TC, Perianin J, Janky E, Miri EH, Papiernik E. Association of pregnancy-induced hypertension with duration of sexual cohabitation before conception. Lancet. 1994; 344 (8928): 973–975.
9. Eras JL, Saftlas AF, Triche E, Hsu CD, Risch HA, Bracken MB. Abortion and its effect on risk of preeclampsia and transient hypertension. Epidemiology. 2000; 11 (1): 36–43.
10. Campbell DM, MacGillivray I, Carr-Hill R. Pre-eclampsia in second preg-nancy. Br J Obstet Gynaecol. 1985; 92 (2): 131–140.
11. Mogren I, Hogberg U, Winkvist A, Stenlund H. Familial occurrence of preeclampsia. Epidemiology. 1999; 10 (5): 518–522.
11A. Esplin MS, Fausch MB, Fraser A, Kerber R, Mineau G, Carrillo J, Varner MW. Paternal and maternal components of the predisposition to preeclampsia. N Engl J Med. 2001; 344: 867–872.
12. Moore MP, Redman CW. Case-control study of severe pre-eclampsia of early onset. Br Med J (Clin Res Ed). 1983; 287 (6392): 580–583.
13. Glietenberg H, van der Walt A. Viable pregnancy after prolonged periods of secondary infertility. S Afr Med J. 1976; 50 (2): 44–48.
14. Knudsen LB. The Danish Fertility Database. Dan Med Bull 1998; 45: 221–225.
15. Conde-Agudelo A, Althabe F, Belizan JM, Kafury-Goeta AC. Cigarette smoking during pregnancy and risk of preeclampsia: a systematic review. Am J Obstet Gynecol 1999; 181: 1026–1035.
16. Olsen J, Schmidt MM, Christensen K. Evaluation of nature-nurture impact on reproductive health using half-siblings. The computerized square dance study design. Epidemiology 1997; 8: 6–11.