Twins vs singletons—Long‐term health outcomes

Abstract

Introduction

Multiple gestations are a risk factor for most pregnancy complications. The current study aimed to study whether offspring born after twin pregnancies are at increased risk for long‐term health complications.

Material and methods

A retrospective cohort study was conducted in a large medical center, including all offspring born between the years 1991–2021, which were followed‐up until 18 years of age. Hospital‐based diagnoses of the offspring were categorized into main groups of morbidities: cardiac, respiratory, infectious, neurological, malignancy, and metabolic. Incidence of hospitalization with diagnoses from each main group was compared between twins and singletons, as well as time to first hospitalization. Cox proportional hazard models were used to study the association between twins vs singletons and hospitalizations by grouped morbidities, while adjusting for maternal age, ethnicity and gender, besides maternal recurrence in the cohort.

Results

A total of 369 478 offspring were included in the analysis; of these 11 986 (3.2%) were twins and 357 492 (96.8%) were singletons. Twins were more likely to be delivered preterm (odds ratio = 17.65, 95% CI: 16.74–18.60), by cesarean delivery and following infertility treatments. Incidence of hospitalizations with all morbidity groups was slightly, some significantly, higher among twins, including cardiac: 1.9% vs 1.5%, respiratory; 8.4% vs 7.1%, neurological: 7.7% vs 7.4%, infectious: 26.0% vs 24.1%, and malignancies: 0.7% vs 0.4%. The risk remained higher in the multivariable analyses (adjusted hazard ratios ranging between 1.09–1.75). When stratifying by gestational age at delivery, the risk for most morbidities was lower among twins vs singletons born in similar gestational ages.

Conclusions

Twins as compared to singletons are at increased risk for most morbidities due to their risk of being born earlier.

The rates of multiple gestations, including mainly twin gestations, have increased in recent years. Twins face higher risk of health complications that continue throughout childhood, due to their risks of being born earlier.

Abbreviations

GDM

gestational diabetes mellitus

PTD

preterm delivery

Key message.

The rates of multiple gestations, including mainly twin gestations, have increased in recent years. Twins face higher risk of health complications that continue throughout childhood, due to their risks of being born earlier.

1. INTRODUCTION

The rates of multiple gestations, including mainly twin gestations, have increased in recent years, primarily due to older maternal age and the use of assisted reproductive technologies. ¹ , ² Twin gestations entail a greater load on most maternal systems, including respiratory, cardiovascular and endocrine functions, ³ , ⁴ , ⁵ , ⁶ and they are a risk factor for most pregnancy and delivery complications, including preterm delivery (PTD), low birthweight, gestational diabetes mellitus (GDM), pregnancy‐related hypertensive disorders and cesarean deliveries. ⁷ , ⁸

Offspring born following complicated pregnancies are exposed to suboptimal prenatal environments and have been shown to be at an increased risk for morbidities throughout life. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ For instance, prematurity is associated, among others, with an increased risk for respiratory, neurological and infectious morbidities, ⁹ , ¹⁰ , ¹¹ GDM is associated with an increased risk for asthma, obesity and respiratory morbidities, ¹² , ¹³ , ¹⁴ and pregnancy‐related hypertensive disorders are associated with an increased risk for offspring respiratory, endometabolic and infectious morbidities. ¹⁵ , ¹⁶ These findings, in most cases, are regardless of gestational age and are based on long follow‐up studies, suggesting that the risk continues throughout life and not only immediately following the delivery.

Offspring born preterm are considered to be at increased risk for health complications throughout life, both due to the pathology leading to the PTD, and due to the immaturity. Among twins, however, the length of gestation is physiologically shorter, and PTDs are more likely to be of a benign nature. ¹⁷

With increasing rates of twin gestations, it is important to study whether twins, as compared to singletons, are at an increased risk for health complications throughout life. The current study aimed to clarify whether twins are at an increased risk for long‐term health complications, and whether this risk is related to gestational age at delivery.

2. MATERIAL AND METHODS

A retrospective cohort study was conducted at the Soroka University Medical Center which is a single large regional tertiary medical center located in southern Israel.

The study included all births between the years 1991 and 2021, excluding offspring with congenital malformations or chromosomal abnormalities. The study population included two ethnic groups: Jewish and Arab‐Bedouin, which are distinct in many demographic and cultural characteristics. While Jewish women's lifestyle, cultural and demographic characteristics are comparable to western countries, Arab‐Bedouin women are more traditional. They live in small towns or settlements, some in shacks, with higher rates of illiteracy, and are among the lowest socioeconomic level in the country. ¹⁸ Consanguineous marriages are common among Arab‐Bedouin, and due to their traditional lifestyle, pregnancy termination is uncommon, therefore the incidence of congenital malformations and chromosomal abnormalities are high among this ethnic group. ¹⁹

Both ethnic groups are Israeli citizens, and as such, are eligible for free health care coverage by the Israeli National Health Insurance law.

The independent variable was defined as being born following twin vs singleton gestation (i.e., the study groups).

The dependent variables were defined as incidence of hospitalizations with a diagnosis from a list of diagnoses by any of the following categories: cardiac, respiratory, infectious, neurologic, malignancies and metabolic morbidities.

Background characteristics included maternal demographic and pregnancy characteristics, including maternal age, parity, ethnicity (Bedouin Arabs vs. Jewish), mode of conception (spontaneous vs. following infertility treatments), insufficient prenatal care (defined as having ≤4 visits, which started after first trimester), ²⁰ pregnancy complications (including gestational diabetes mellitus or hypertension) and mode of delivery (cesarean vs. vaginal).

All pediatric hospitalization records during the study period were abstracted and over 6000 diagnoses and procedures were categorized into groups according to the systems involved, including, but not limited to: cardiac, respiratory, infectious, neurological, malignancies and metabolic morbidities. The list of all included diagnoses by categories is presented as Table S1. For each offspring, follow‐up was initiated at delivery and terminated at either of the following: the offspring reached the age of 18 years, the end of the study period, or at the onset of an event. An event was defined as the first offspring hospitalization with any diagnosis from the list, per category. For instance, follow‐up was terminated for respiratory morbidities at age 2 if the offspring was hospitalized with asthma for the first time at that age; however, follow‐up continued for cardiac, infectious, neurological, malignancies and metabolic morbidities.

The obstetrics medical records contained data regarding the pregnancy follow‐up and delivery, maternal socioeconomic characteristics and comorbidities, and were based on both hospital and outpatient records.

A link was made between delivery information and offspring hospitalization records based on both the maternal and offspring personal identification numbers. Being the only medical center in the area, it was assumed that any health problem requiring emergency care or hospitalization would be referred to this center (and identified as an “Event” in this cohort), regardless of the studied exposure (i.e., twins or singletons), allowing long‐term follow‐up of the offspring delivered in this medical center. In case no hospitalizations with the studied diagnoses have been identified, the offspring was defined as a censored case. While loss to follow‐up may have occurred, there is no reason to believe it would be associated with the twinning status, so this may have led to an underestimation of the actual effect.

2.1. Statistical analyses

Background characteristics and outcomes were compared between the study groups using chi‐square or t‐tests. For the comparison of maternal and pregnancy characteristics, data are presented per pregnancy (i.e., for the twin group, one value is presented). Variables associated with the study group were suspected as confounding variables and were considered in the multivariable analysis.

Kaplan Meier survival analysis was used to compare cumulative survival between singletons and twins. The assumptions for the proportional hazard models were verified before inclusion in the multivariable model.

Cox proportional hazard models were used to study the association between twins vs singletons and the risk for hospitalization with each category of morbidities, while adjusting for the background characteristics. Additionally, the models were adjusted for the dependence between siblings in the cohort, among both twins and siblings from different pregnancies. The models were not adjusted for gestational age to minimize collider bias. ²¹ The cause of PTDs are diverse and often unknown; however, among singletons as compared to gestation of twins, the cause is more likely to be pathological. Both pathology and prematurity have adverse effects on offspring health, and since singletons are more likely to be exposed to in utero pathology, adjusting for gestational age may bias the observed measure of effect.

The incidence of the hospitalizations by groups of morbidities, and for all‐morbidities combined (i.e., a composite outcome of all morbidities and indicates the time until first hospitalization from any morbidity), was compared between twins and singletons by gestational ages at delivery.

All analyses were two‐sided with α < 0.05 and β = 0.2.

2.2. Ethics statement

The study protocol was approved by the Soroka University Medical Center IRB committee (# 0357‐19‐SOR) on December 5, 2022, and was exempt from the requirement for informed consent.

3. RESULTS

During the study period, 395 499 offspring were delivered at Soroka University Medical Center. After excluding cases of congenital malformations or chromosomal abnormalities (n = 25 567, 6.5%, leaving 369 932 offspring) multiple gestations with more than twins were excluded (n = 454, 0.1%), leaving a total of 369 478 offspring which were included in the analysis, of which 11 986 (3.2%) were twins and 357 492 (96.8%) were singletons.

Table 1 presents maternal, pregnancy and offspring characteristics by study group. Compared to mothers of singletons, mothers of twins were older (29.74 ± 5.6 vs. 28.26 ± 5.8 years), more likely to be Jewish (51.2% vs. 45.6%), conceived following infertility treatments (24.1% vs. 1.7%), with lower parity (29.8% vs. 24.2% were primipara) and have pregnancy complications, including gestational diabetes mellitus (8.4% vs. 4.8%) and pregnancy‐related hypertensive disorders (10.6% vs. 4.7%). Twins were more likely to be female (50.5% vs. 49.0%), delivered preterm (57.2% vs. 7.0%), with low birthweight (<2500 g) (61.8% vs. 7.1%) and by cesarean delivery (53.8% vs. 14.1%).

TABLE 1.

Maternal, pregnancy and offspring characteristics by study group.

	Twins	Singleton		p‐value
Maternal and pregnancy characteristics	n = 6002 (3.2%)	n = 357 492 (96.8%)	OR (95% CI) for twins vs singletons
Maternal age (mean ± SD)	29.74 ± 5.6	28.26 ± 5.8		<0.001 (df = 6208)
Parity				<0.001
1	1786 (29.8)	86 454 (24.2)
2–4	2887 (48.1)	185 353 (51.9)
≥5	1329 (22.1)	85 584 (23.9)
Ethnicity			1.25 (1.19–1.32)	<0.001
Bedouin	2930 (48.8)	194 570 (54.4)
Jewish	3072 (51.2)	162 922 (45.6)
Smoking	41 (0.7)	2564 (0.7)	0.95 (0.70–1.29)	0.81
Obesity	115 (1.9)	4100 (1.1)	1.68 (1.40–2.03)	<0.001
Insufficient prenatal care	330 (5.5)	36 450 (10.2)	0.51 (0.46–0.57)	<0.001
Infertility treatments	1448 (24.1)	6237 (1.7)	17.91 (16.79–19.09)	<0.001
Gestational diabetes mellitus	502 (8.4)	17 037 (4.8)	1.82 (1.66–2.00)	<0.001
Hypertension disorders of pregnancy	638 (10.6)	16 789 (4.7)	2.41 (2.22–2.62)	<0.001
Gestational age (mean ± SD)	35.58 ± 2.9	39.02 ± 1.9		<0.001 (df = 6083)
Preterm delivery	3428 (57.2)	25 123 (7.0)	17.65 (16.74–18.60)	<0.001

Delivery and offspring characteristics	n = 11 986 (3.2%)	n = 357 492 (96.8%)
Birthweight (mean ± SD)	2293 ± 566	3197 ± 521		<0.001 (df = 12 676)
Low birthweight	7407 (61.8)	25 313 (7.1)	21.23 (20.41–22.07)	<0.001
Small for gestational age	522 (4.4)	16 265 (4.5)	0.96 (0.87–1.04)	0.34
Cesarean delivery	6452 (53.8)	50 504 (14.1)	7.09 (6.83–7.36)	<0.001
Gender			0.94 (0.91–0.98)	<0.001
Male	5930 (49.5)	182 209 (51.0)
Female	6056 (50.5)	175 283 (49.0)
Low Apgar 5 (<7)	257 (2.2)	2058 (0.6)	3.81 (3.34–4.34)	<0.001
Perinatal mortality	374 (3.1)	2854 (0.8)	4.00 (3.59–4.46)	<0.001

The follow‐up time, which ranged from 0–18 years, was shorter in twins compared to singletons in all categories of morbidities, due to their increasing rates over the years of the cohort (from 1.3% in the year 1991 to 3.7% and 3.5% in the years 2019 and 2020, respectively). Among twins, the mean years of follow‐up were 10.85 ± 6.6, 10.15 ± 6.6, 10.33 ± 6.4, 10.99 ± 6.3, 8.37 ± 6.9 and 11.02 ± 6.3, while among singletons the years of follow‐up were 11.37 ± 6.5, 10.76 ± 6.6, 10.84 ± 6.5, 11.45 ± 6.3, 9.09 ± 7.0 and 11.5 ± 6.3 for cardiac, respiratory, infectious, neurological, malignancy and endocrine categories of morbidities, respectively (p < 0.001 for all).

Table 2 presents the incidence of hospitalizations by morbidity category in each study group. The incidence of hospitalization in all but the metabolic morbidities, was higher among twins compared to singletons. For instance, incidence rates of cardiac‐related hospitalizations were 1.9% vs 1.5% (hazard ratio [HR] = 1.33; 95% CI: 1.17–1.52) and malignancy‐related hospitalizations were 0.7% vs 0.4% (HR = 1.77; 95% CI: 1.41–2.22) among twins and singletons, respectively. The differences between singleton and twins in time to first hospitalization per morbidity category were also observed in the Kaplan–Meier survival curves (Figure 1), in which survival was lower among twins in all but the metabolic morbidity categories.

TABLE 2.

Incidence, hazard ratios and adjusted hazard ratios for main groups of morbidities by study group.

	Twins n = 11 986 (3.2%)	Singleton n = 363 750 (96.8%)	Hazard ratio (95% CI)	Adjusted a hazard ratio (95% CI)
Cardiac	229 (1.9)	5408 (1.5)	1.33 (1.17–1.52)	1.38 (1.21–1.58)
Respiratory	1007 (8.4)	25 535 (7.1)	1.23 (1.16–1.31)	1.26 (1.18–1.34)
Infectious	3115 (26.0)	86 056 (24.1)	1.15 (1.11–1.19)	1.17 (1.13–1.21)
Neurological	917 (7.7)	26 463 (7.4)	1.08 (1.02–1.16)	1.09 (1.03–1.17)
Neoplasm	80 (0.7)	1387 (0.4)	1.77 (1.41–2.22)	1.75 (1.38–2.24)
Endocrine	140 (1.2)	3755 (1.1)	1.09 (1.00–1.20)	1.17 (0.99–1.39)
Any morbidity	3833 (32.0)	107 384 (30.3)	1.15 (1.12–1.19)	1.17 (1.13–1.21)

^a Adjusted for offspring gender, maternal age and ethnicity.

FIGURE 1.

Kaplan–Meier log of cumulative survival curves for the differences between singleton and twins in time (in days) to first hospitalization, per morbidity category.

The unadjusted and adjusted hazard ratio results for the Cox hazard analyses are presented in Table 2. The multivariable models were adjusted for maternal age, ethnicity and offspring gender. Results of the models show that twins as compared to singletons were at increased risk for all but endocrine morbidities.

Figure 2 presents a comparison between singletons and twins regarding the incidence of each morbidity group by gestational age at delivery. Regarding several morbidities, such as cardiac and malignancies, the difference between twins and singleton was not clear across all gestational ages; however, the incidence of respiratory and neurological morbidities was lower in twins vs singletons across most gestational ages.

FIGURE 2.

Incidence of the studied morbidities categories by gestational age at delivery among singletons and twins.

4. DISCUSSION

In this large population‐based retrospective cohort, with a follow‐up of up to 18 years, immediate pregnancy outcomes were worse among twins, and they were more likely to be hospitalized due to most of the studied groups of morbidities. After stratification by gestational age at delivery, however, twins vs singletons were less likely to be hospitalized with respiratory or neurological morbidities, suggesting the adverse effect on health is due to the shorter gestation in twins.

Comparison of hospitalization rates between different studies and setting is challenging, and it is unclear whether the rates observed in the current study are standard. However, based on Yorita et al., ²² the incidence of hospitalization with infectious related morbidities is expected to be ~7% during the first year of life. This rate is similar to the rate observed in our population.

The increased risk for hospitalizations with most studied morbidities among twins was most likely due to their increased risk for shorter gestation and PTDs.

Prematurity is associated with a greater risk for many morbidities, including neurological, infectious, cardiac and endocrine morbidities, both immediately after delivery, as well as later in life. ⁹ , ¹⁰ , ¹¹ , ²³ Findings in the current study are in agreement with this as the risk for hospitalizations with any of the studied groups of morbidities was higher among offspring born at earlier gestational ages.

Offspring born prematurely were likely to be exposed in utero to the pathology causing the early delivery, such as infections or placental pathologies. In the case of multiple gestation, the length of gestation is physiologically shorter and the fetuses less likely to be exposed to pathological factors. This explains the better survival of twins vs singletons delivered at an early gestational age. ²⁴

Twins are exposed to suboptimal in utero environment due to micronutrient deficiencies, excess placenta or implantation pathologies. ²⁵ , ²⁶ In the current study, twin pregnancies were more likely to be diagnosed with pregnancy complications and adverse pregnancy outcomes. Offspring born following pregnancies complicated with GDM, pregnancy‐related hypertensive disorders and other complications, are at an increased risk of health complications. ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ It can therefore be expected that along with the shorter gestation twins would be at greater risk for health complications.

The stratification by gestational age suggests twins were at lower risk for several hospitalization groups, and the complications diagnosed during twin gestation may pose a lower risk to the fetus. This is in alignment with several studies which found that the risk for a low Apgar score, neonatal respiratory morbidities and intensive care unit admission was lower in twins compared to singletons among pregnancies complicated by GDM or pregnancy‐related hypertensive disorders. ²⁷ , ²⁸ , ²⁹ The fetus in a multiple gestation may have adjusted and possibly compensate for some early life insufficiencies, via epigenetic or other mechanisms, yielding a more resilient offspring. ²⁹

The main strength of this study is the large population‐based cohort in a setting which allowed a long follow‐up period to study long‐term health effects of early life exposures. This also enabled the research of rare, less‐studied morbidities, such as childhood malignancies.

One of the limitations of this study is the possible survival bias. Our cohort represents only offspring who survived the pregnancy and delivery. This bias may be selective as twin gestations are more complicated and more likely to end preterm and with perinatal mortality. Therefore, “weaker” offspring are not represented in the cohort, and the included offspring may be stronger and more resilient and possibly at lower risk for long‐term morbidities, especially among the twins.

Another limitation arises from the retrospective nature of the study, which was based on existing computerized medical records, with unavailable data on important, possibly confounding variables, such as corticosteroid treatment which may have been in use, more likely among twins vs singleton pregnancies, ³⁰ offspring exposures after delivery including breast‐feeding and environmental factors.

The accuracy of the collected data may be another limitation, and in particular potential differences in accuracy between the personnel contributing to the hospital records.

5. CONCLUSION

With increasing rates of multiple gestations, it is critical to study the long‐term health effects of twinning. Twins may face higher risk of a range of morbidities due to their risks of being born earlier. These risks involve many health complications that continue throughout childhood.

AUTHOR CONTRIBUTIONS

TW conceptualized and designed the study, carried out the analyses, drafted the initial manuscript, and revised the manuscript. IY conceptualized and designed the study, critically reviewed the manuscript for important intellectual content, assisted in drafting the initial manuscript. RS designed the data collection instruments, collected data, and assisted with the analyses. ES conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

CONFLICT OF INTEREST STATEMENT

The authors have no conflict of interest to declare.

Supporting information

Table S1.

Wainstock T, Yoles I, Sergienko R, Sheiner E. Twins vs singletons—Long‐term health outcomes. Acta Obstet Gynecol Scand. 2023;102:1000‐1006. doi: 10.1111/aogs.14579