Because of the sharp rise in medically assisted reproduction and delayed childbearing during the last decades, the twinning rate has increased dramatically in all developed countries (Pison et al., 2015). This rise in twinning rate represents an important public health issue since twin pregnancies are associated with greater health risks for both infants and mothers (Cheong-See et al., 2016). Compared to singletons, twin babies are more often subjected to multiple adverse neonatal outcomes, including preterm birth, small-for-gestational-age, and neonatal death (Cheong-See et al., 2016). In addition, twin pregnancies are further complicated with discordant growth (Miller et al., 2012; Grantz et al., 2016; D'Antonio et al., 2018). Approximately 16% of twin pregnancies have birth weight discordance of at least 20%, which might also contribute to elevated risk of perinatal and neonatal mortality (Blickstein and Kalish, 2003). Consequently, a considerable proportion of twins experience a very early loss of their co-twin.

Previous studies indicate that loss of a co-twin by death in childhood or adulthood is associated with considerable mental morbidities among the surviving twins (Segal and Bouchard, 1993; Woodward, 1988). Our recent findings suggest an increased risk of psychiatric disorders after a loss of co-twin, compared to loss of a full (non-twin) sibling, beyond age 2 (Song et al., 2020). The greater risk increase after loss of a co-twin might be due to the stronger emotional bond between twins (Rosendahl SP and Björklund, 2013; Segal and Ream, 1998) and the greater genetic relatedness, in the case of monozygotic twins (Segal and Ream, 1998; Parkes, 1993; Neyer, 2002). In contrast, with cognitive immaturity and limited (or no) afterbirth interactions, a co-twin loss at birth or during the neonatal period leaves little room for a twin bond to be established and therefore it seems implausible that the mental health of the surviving twin would be affected by such a loss. However, several scientific and media accounts describe unexpected lingering sorrow among twins who lost their co-twin at or shortly after birth, even among the twins that didn’t know they were born as twins (Morgan, 2014; lone twin network, 2020; Woodward, 2010). It is further possible that the parenting of bereaved parents after the perinatal loss of one twin baby may leave the surviving twin vulnerable for mental morbidities (Lamb, 2002). In the total absence of data on the rate of psychiatric disorders among twins who lost a co-twin at birth, we conducted a nationwide population- and sibling-matched cohort study to estimate the extent to which loss of a co-twin at birth is associated with the incidence of psychiatric disorders among surviving twins, after carefully controlling for important confounders such as birth characteristics and familial factors.

In a population-based matched cohort, we included in the study all Swedish-born twins that lost a co-twin at birth, defined as a death of the co-twin within 60 days after birth, between 1973 and 2011- (exposed twins, n = 787), together with two reference groups. The first reference group included 3935 sex-, birth year-, and gestational age (GA)-matched unexposed twins randomly selected from the Swedish twin population that did not experience such a loss, to control for twin pregnancy and twin birth. Because the exposed twins grew up on their own, as singletons, which differs from the social conditions of a twin life, we also included 3935 singletons randomly selected from the singleton population that were individually matched to the exposed twins on sex, birth year, and birth factors (GA, birth weight for GA, and birth order) as the second reference group. In addition to the population-based matched cohort, to address the concern of familial factors such as genetic background and environmental factors during childhood shared within a family (e.g., parenting of the bereaved parents), we compared 569 exposed twins to their full siblings (n = 880) in twin-sibling family cohort. The study entry was the 60th day after birth of the exposed twin (i.e., the index date). We then followed all individuals from the index date until the first diagnosis of any psychiatric disorder, emigration, death, or the end of 2013, whichever occurred first (see details in the later section ‘Materials and methods,’ and Figure 1 and Supplementary file 1-Table 1).

Figure 1

Download asset Open asset

In total, the population-based matched cohort accumulated 169,507 person-years at risk, with a median of 19 years of follow-up (Table 1). The exposed twins had an almost equal sex distribution (54% were male). While there was little difference in birth weight for GA, maternal educational level, and family history of psychiatric disorders, exposed twins tended to have lower Apgar score, older mothers at childbirth, and higher maternal cohabitation rate than matched unexposed twins and singletons. As expected, in the twin-sibling family cohort, maternal characteristics were identical or similar between exposed twins and their full siblings, although birth characteristics (e.g., GA, birth weight for GA, and Apgar score) were different between these two groups (Table 1).

During the follow-up, we identified 1501 individuals with incident psychiatric disorders in the population-based matched cohort, including 178 cases among the exposed twins, 600 among the matched unexposed twins, and 723 among the matched singletons, corresponding to a crude incidence rate (IR) of 12.08, 7.76, and 9.33 per 1000 person-years, respectively (Table 2). Compared to matched unexposed twins, the rate of any psychiatric disorders was increased among the exposed twins. The hazard ratio (HR) was 1.59 (95% confidence intervals [CI] 1.33–1.90) after controlling for birth year, sex, and birth factors (i.e., GA, birth weight for GA, maternal age at birth), and decreased to 1.56 (95% CI 1.30–1.87) when other covariables, including Apgar score, family history of psychiatric disorders, and maternal educational level and cohabitation status were added into the model. The fully adjusted HR was 1.41 (95% CI 1.19–1.69) when the reference group was matched singletons. In the analyses of twin-sibling family cohort, we obtained an HR of 1.43 (95% CI 0.82–2.51) after full adjustment of all abovementioned variables (Table 2).

Although not statistically significant, we observed higher HRs for emotional disorders such as depression and anxiety, compared with neurodevelopmental disorders and other psychiatric disorders (Table 2). Subgroup analyses of the population-based matched cohort indicated that the association between loss of a co-twin at birth and subsequent risk of any psychiatric disorder did not differ by sex, but seemed stronger among individuals without family history of psychiatric disorders. In addition, the relative risk increased further after loss of a same-sex co-twin, than loss of an opposite-sex co-twin, at birth (Table 3). By plotting HRs over attained age, we found the risk elevation to be only evident for psychiatric disorders diagnosed in childhood and early adulthood, that is before age of 25 (Figure 2). With lower precision, analyses of the twin-sibling family cohort revealed similar risk patterns (Table 3). We obtained slightly higher HRs when comparing the exposed twins to their older siblings (1.74, 95% CI 0.54–5.59), than to their younger siblings (1.27, 95% CI 0.48–3.34) (Supplementary file 1-Table 2).

Figure 2

Download asset Open asset

Figure 2—source data 1

Summary data for Figure 2.

https://cdn.elifesciences.org/articles/63514/elife-63514-fig2-data1-v1.xls

Download elife-63514-fig2-data1-v1.xls

In sensitivity analyses, we found no significant effect modification by the presence of psychiatric disorders among parents (Supplementary file 1-Table 2), by the presence of congenital abnormalities, or by the diagnoses of severe somatic diseases during follow-up (Supplementary file 1-Table 3). Moreover, changing the definition of loss of a co-twin at birth to loss of a co-twin within 28 days after birth yielded largely similar estimates, although with lower precision (Supplementary file 1-Table 4).

To our knowledge, this is the first nationwide population-based and sibling-matched cohort study exploring the association between loss of a co-twin at birth or shortly after birth and the subsequent risk of psychiatric disorders. Compared to matched unexposed twins or singletons, as well as their non-twin full siblings, twins exposed to a co-twin loss at birth were at considerably elevated risk of psychiatric disorders, especially emotional disorders, such as depression and anxiety, before age of 25. Notably, this association was independent of multiple important confounders, including birth characteristics and childhood social conditions (by comparing bereaved twins to singletons), and other familial factors (by comparing bereaved twins to their full siblings), indicating that increased clinical alertness of the mental health of surviving twins after very early co-twin loss is warranted. In addition, although the excess risk was not modified by parent’s psychiatric disorder nor the surviving twin’s congenital or other severe diseases diagnosed during follow-up, it seemed more pronounced among twins exposed to early loss of a same-sex co-twin and among twins without family history of psychiatric disorders.

While accumulating evidence supports that both childhood and adult twin loss are associated with increased risk of psychiatric morbidity among the surviving twins (Rosendahl SP and Björklund, 2013; Segal and Ream, 1998), no previous study has addressed whether such emotional reactions can be observed after a very early co-twin loss where limited twin relationship, perception, or memory from the loss could be expected. The absence of evidence is mainly due to the complexity of the research question and lack of high-quality data to address potential confounding by multiple factors, such as twin pregnancy and birth (i.e., suboptimal birth characteristics) but a singleton-like life, familial factors, and genetic susceptibility to diseases. Therefore, with the unique Swedish nationwide data sources, which provide a substantial sample size of exposed twins with detailed data on birth characteristics and familial information, we conducted the present study. By contrasting the rate of psychiatric disorders among surviving twins who were exposed to a co-twin loss at birth with that of several comparison groups, including matched unexposed twins and singletons, as well as the full siblings of the exposed twins, our assessment demonstrates a robust association between early loss of a co-twin and subsequent risk of psychiatric disorders. Given the scarcity of existing data within this area of research, our findings call for further investigation on the possible underlying mechanisms linking the experience of a co-twin loss at birth to mental health decline during adulthood. Particularly, despite the lack of information on zygosity, the higher relative risk observed after early loss of a same-sex co-twin, compared with a loss of opposite-sex co-twin, may indicate the importance of shared genetic background on the formation of a twinship bond. This is similar to the greater grief intensity reported among monozygotic twins who experienced an adult loss of co-twin, compared with dizygotic twins, and consistent with the evolutionary theory suggesting a role of genetic relatedness in the bereavement process (Segal and Blozis, 2002; Segal, 2019).

The major concern in studies of this kind is that, the death of a co-twin at birth may be an indicator of a poor pregnancy or birth conditions or congenital defects that are shared within a twin pair, and hence vulnerability of the surviving twin to various diseases (such as developmental defects and other somatic disorders) during their later lives. In present analysis, we indeed attempted to have a control for birth characteristics, yet we cannot rule out the possibility that both physical vulnerabilities of the surviving twins and their close contacts with health care due to these somatic problems have contributed to a detection of psychiatric disorders in this population. However, similar results were obtained in our sensitivity analyses where the presence of congenital abnormalities and the diagnosis of severe somatic conditions during follow-up were taken into account by subgroup analyses or by additional adjustments, suggesting a minor influence of these factors. Another possible explanation for the observed increased risk of psychiatric disorders among the surviving twins could be altered parenting of the grieving parents. Indeed, given that we also observed a heightened risk of psychiatric disorder in full siblings, especially younger full siblings, of the bereaved twins in our twin-sibling family analysis, altered parenting style among bereaved individuals and its impact on offspring’s mental health need further investigation. Our additional analyses taking into account clinically confirmed psychiatric disorders of the bereaved parents during the follow-up suggest limited mediating role of clinically confirmed parental psychiatric disorders in the association between early co-twin loss and risk of psychiatric disorder. Nevertheless, severe mental illness requiring a clinical diagnosis affects a relatively small proportion of the bereaved parents.

The major merit of our study is the use of population-based cohort design, including 787 exposed twins with a complete follow-up of up to 41 years. In addition to a full consideration of birth factors, the possible influence of twin pregnancy and twin birth was controlled through the twin-twin comparison, while the social conditions during childhood was taken into consideration in the twin-singleton comparison. Through the twin-full sibling comparison, we were further able to control for potential familial confounders, as well as explore possible underlying mechanisms related to parenting. As the largest cohort study on co-twin loss at birth to date, we had sufficient statistical power to perform most of the planned subgroup analyses. Information bias was minimized since the registration and diagnosis of exposure and outcome was compiled prospectively and independently. Furthermore, the availability of rich information on sociodemographic and medical conditions for both the study participants and their parents enabled considerations of a wide range of important confounding factors.

Limitations include the late establishment of outpatient care records in the Swedish National Patient Register (2001-), which may have rendered an underestimated number of psychiatric disorder diagnoses, especially the milder ones. In addition, individuals were relatively young at the end of follow-up (up to 41 years, with median age at follow-up as 19 years). Therefore, the study focused mainly on early-onset psychiatric disorders. The association between loss of a co-twin at birth and psychiatric disorders beyond early adulthood needs to be addressed in further studies. Furthermore, because we identified individuals through the unique personal identification numbers assigned at birth to each Swedish-born person, our study involved merely twins who lost their co-twin at or shortly after delivery. The loss of a co-twin during the early pregnancy period (i.e., vanishing twin) or due to stillbirth, which might also have psychological effects on the surviving twin (A Silent Cry, 2008), is therefore beyond the scope of the present study. Finally, although we made every effort to control for important confounders such as birth characteristics, social and familial conditions, and shared genetic background, we cannot exclude the possibility of residual confounding.

In conclusion, in the Swedish population, exposure to death of a co-twin at birth was associated with a subsequently elevated risk of psychiatric disorders among the surviving twin. These findings call for medical and scientific attention of the mental health of this bereaved population and further exploration of the underlying mechanisms.

One source data file has been provided. Original data are held by Swedish National Board of Health and Welfare, Statistics Sweden and the Swedish Twin Registry. Due to Swedish law on data protection and the ethical approval of the current study, we cannot make the data publicly available. However, any researcher can access the data by obtaining an ethical approval from a regional ethical review board and thereafter request the original data from the Swedish National Board of Health and Welfare, Statistics Sweden, and the Swedish Twin Register. Detailed information on data application can be found at https://www.registerforskning.se/en/ and https://ki.se/en/research/the-swedish-twin-registry.

1. Book

   1. A Silent Cry

   (2008)

   Wombtwin Survivors Tell Their Stories

   Wren Publications.

   • Google Scholar
2. 1. Blickstein I
   2. Kalish RB

   (2003) Birthweight discordance in multiple pregnancy

   Twin Research 6:526–531.

   https://doi.org/10.1375/136905203322686536

   • PubMed
   • Google Scholar
3. 1. Cheong-See F
   2. Schuit E
   3. Arroyo-Manzano D
   4. Khalil A
   5. Barrett J
   6. Joseph KS
   7. Asztalos E
   8. Hack K
   9. Lewi L
   10. Lim A
   11. Liem S
   12. Norman JE
   13. Morrison J
   14. Combs CA
   15. Garite TJ
   16. Maurel K
   17. Serra V
   18. Perales A
   19. Rode L
   20. Worda K
   21. Nassar A
   22. Aboulghar M
   23. Rouse D
   24. Thom E
   25. Breathnach F
   26. Nakayama S
   27. Russo FM
   28. Robinson JN
   29. Dodd JM
   30. Newman RB
   31. Bhattacharya S
   32. Tang S
   33. Mol BW
   34. Zamora J
   35. Thilaganathan B
   36. Thangaratinam S
   37. Global Obstetrics Network (GONet) Collaboration

   (2016) Prospective risk of stillbirth and neonatal complications in twin pregnancies: systematic review and meta-analysis

   BMJ 354:i4353.

   https://doi.org/10.1136/bmj.i4353

   • PubMed
   • Google Scholar
4. 1. D'Antonio F
   2. Odibo AO
   3. Prefumo F
   4. Khalil A
   5. Buca D
   6. Flacco ME
   7. Liberati M
   8. Manzoli L
   9. Acharya G

   (2018) Weight discordance and perinatal mortality in twin pregnancy: systematic review and meta-analysis

   Ultrasound in Obstetrics & Gynecology 52:11–23.

   https://doi.org/10.1002/uog.18966

   • PubMed
   • Google Scholar
5. 1. Grantz KL
   2. Grewal J
   3. Albert PS
   4. Wapner R
   5. D'Alton ME
   6. Sciscione A
   7. Grobman WA
   8. Wing DA
   9. Owen J
   10. Newman RB
   11. Chien EK
   12. Gore-Langton RE
   13. Kim S
   14. Zhang C
   15. Buck Louis GM
   16. Hediger ML

   (2016) Dichorionic twin trajectories: the NICHD fetal growth studies

   American Journal of Obstetrics and Gynecology 215:221.

   https://doi.org/10.1016/j.ajog.2016.04.044

   • PubMed
   • Google Scholar
6. 1. Lamb EH

   (2002) The impact of previous perinatal loss on subsequent pregnancy and parenting

   The Journal of Perinatal Education 11:33–40.

   https://doi.org/10.1891/1058-1243.11.2.33

   • PubMed
   • Google Scholar
7. Software

   1. lone twin network

   (2020) Loss at Birth

   Loss at Birth.

   https://lonetwinnetwork.org.uk/members-stories/loss-at-birth/
8. 1. Miller J
   2. Chauhan SP
   3. Abuhamad AZ

   (2012) Discordant twins: diagnosis, evaluation and management

   American Journal of Obstetrics and Gynecology 206:10–20.

   https://doi.org/10.1016/j.ajog.2011.06.075

   • PubMed
   • Google Scholar
9. Book

   1. Morgan MR

   (2014)

   When Grief Calls Forth the Healing: A Memoir of Losing a Twin

   Open Road Media.

   • Google Scholar
10. 1. Neyer FJ

    (2002) Twin relationships in old age: a developmental perspective

    Journal of Social and Personal Relationships 19:155–177.

    https://doi.org/10.1177/0265407502192001

    • Google Scholar
11. 1. Parkes CM

    (1993) Bereavement and adaptation: a comparative study of the aftermath of death

    Journal of Psychosomatic Research 37:679–680.

    https://doi.org/10.1016/0022-3999(93)90067-P

    • Google Scholar
12. 1. Pison G
    2. Monden C
    3. Smits J

    (2015) Twinning rates in developed countries: trends and explanations

    Population and Development Review 41:629–649.

    https://doi.org/10.1111/j.1728-4457.2015.00088.x

    • Google Scholar
13. 1. Rosendahl SP BP
    2. Björklund A

    (2013) Images of sorrow—Experiences of losing a co-twin in old age

    Health 5:64–73.

    https://doi.org/10.4236/health.2013.512A009

    • Google Scholar
14. Book

    1. Segal NL

    (2019) Evolutionary perspectives on death

    In: Zeigler-Hill V, editors. Evolutionary Perspectives on the Loss of a Twin. Springer. pp. 25–36.

    https://doi.org/10.1007/978-3-030-25466-7

    • Google Scholar
15. 1. Segal NL
    2. Blozis SA

    (2002) Psychobiological and evolutionary perspectives on coping and health characteristics following loss: a twin study

    Twin Research 5:175–187.

    https://doi.org/10.1375/twin.5.3.175

    • PubMed
    • Google Scholar
16. 1. Segal NL
    2. Bouchard TJ

    (1993)

    Grief intensity following the loss of a twin and other relatives: test of kinship genetic hypotheses

    Human Biology 65:87–105.

    • PubMed
    • Google Scholar
17. 1. Segal NL
    2. Ream SL

    (1998) Decrease in grief intensity for deceased twin and non-twin relatives: an evolutionary perspective

    Personality and Individual Differences 25:317–325.

    https://doi.org/10.1016/S0191-8869(98)00050-6

    • Google Scholar
18. 1. Song H
    2. Larsson H
    3. Fang F
    4. Almqvist C
    5. Pedersen NL
    6. Magnusson PK
    7. Valdimarsdóttir UA

    (2020) Risk of psychiatric disorders among the surviving twins after a co-twin loss

    eLife 9:e56860.

    https://doi.org/10.7554/eLife.56860

    • PubMed
    • Google Scholar
19. 1. Woodward J

    (1988) The bereaved twin

    Acta Geneticae Medicae Et Gemellologiae: Twin Research 37:173–180.

    https://doi.org/10.1017/S0001566000004098

    • PubMed
    • Google Scholar
20. Book

    1. Woodward J

    (2010)

    The Lone Twin

    Free Association Books.

    • Google Scholar

Author details

1. Huan Song

   1. West China Biomedical Big Data Center, West China Hospital, Sichuan University, Sichuan, China
   2. Center of Public Health Sciences, Faculty of Medicine, University of Iceland, Reykjavík, Iceland
   3. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden

   Contribution

   Conceptualization, Formal analysis, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   songhuan@wchscu.cn

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3845-8079

2. Fang Fang

   Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden

   Contribution

   Conceptualization, Supervision, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-3310-6456

3. Henrik Larsson

   1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
   2. School of Medical Sciences, Örebro University, Örebro, Sweden

   Contribution

   Data curation, Supervision, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

4. Nancy L Pedersen

   1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
   2. Department of Psychology, University of Southern California, Los Angeles, United States

   Contribution

   Supervision, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

5. Patrik KE Magnusson

   Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden

   Contribution

   Data curation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

6. Catarina Almqvist

   1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
   2. Astrid Lindgren Children’s Hospital, Karolinska University Hospital, Stockholm, Sweden

   Contribution

   Funding acquisition, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

7. Unnur A Valdimarsdóttir

   1. Center of Public Health Sciences, Faculty of Medicine, University of Iceland, Reykjavík, Iceland
   2. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
   3. Department of Epidemiology, Harvard T H Chan School of Public Health, Boston, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5382-946X

• 33,028

  views

• 213

  downloads

• 1

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.