Archives of Gynecology and Obstetrics

, Volume 286, Issue 6, pp 1349–1355

Twins born over weekends: are they at risk for elevated infant mortality?

Authors

    • Department of Public Health, College of Health and Human ServicesWestern Kentucky University
    • Department of Epidemiology and Biostatistics, College of Public HealthUniversity of South Florida
  • Hamisu M. Salihu
    • Department of Epidemiology and Biostatistics, College of Public HealthUniversity of South Florida
    • Department of Obstetrics and Gynecology, College of MedicineUniversity of South Florida
  • Gary English
    • Department of Public Health, College of Health and Human ServicesWestern Kentucky University
  • Cynthia Anozie
    • Department of Public Health, College of Health and Human ServicesWestern Kentucky University
  • Grace Lartey
    • Department of Public Health, College of Health and Human ServicesWestern Kentucky University
  • Getachew Dagne
    • Department of Epidemiology and Biostatistics, College of Public HealthUniversity of South Florida
Maternal-Fetal Medicine

DOI: 10.1007/s00404-012-2463-7

Cite this article as:
Ibrahimou, B., Salihu, H.M., English, G. et al. Arch Gynecol Obstet (2012) 286: 1349. doi:10.1007/s00404-012-2463-7

Abstract

Objective

To assess the impact of the day of birth on twin mortality in a population sample.

Methods

We analyzed weekend versus weekday twin births from the United States national twin birth data for the periods 1989–2002. We computed adjusted hazard ratios (HR) and 95 % confidence intervals (CI) to assess the association between infant mortality and weekday of birth using the Cox proportional hazards model.

Results

The crude rates for all types of mortality were found to be significantly higher for twins born on weekends than on weekdays. After adjustment, only post-neonatal mortality risk was higher on weekends as compared to weekdays [Hazards ratio (HR) = 1.19, CI: 1.04, 1.36]. Twins of white mothers were at greater risk for neonatal death (HR = 1.16, CI: 1.08, 1.24) but were less likely to experience post-neonatal death (HR = 0.68, CI: 0.64, 0.76) as compared to twins of black mothers. We found an interaction between maternal age and weekday of birth. Twins born on weekends to teenage mothers (age <18) had a 35 % greater risk for neonatal death (HR = 1.35, CI: 1.06, 1.71) while those born on weekends to older mothers did not show elevated risk for any of the mortality indices.

Conclusion

Increased risks for post-neonatal death are significantly higher amongst twins born on weekends as compared to weekdays. Further research is required to identify the detailed differences in structure and procedures that result in the disadvantage associated with weekend birth.

Keywords

TwinInfant mortalityNeonatal mortalityPost-neonatal mortalityAdolescent mothers

Introduction

Twin pregnancies are high risk gestations with elevated perinatal mortality rates [1]. Twins, when compared with singletons, have fivefold risk of fetal death, sevenfold elevated risk of neonatal death, and fivefold increased risk of infant death [24]. Singletons also respond differently from multiples or twins to interventions that are designed to lengthen the gestational age at birth [5, 6]. Higher rates of early neonatal mortality have been linked to preterm birth [7]. Assumptions made have suggested that a detailed examination of the time of birth could provide a clearer understanding of neonatal mortality and preterm birth [8].

Studies examining the risk of neonatal, post-neonatal and infant mortality among infants born during a weekend or weekday are, however, scant.

Although infants born on weekends are considerably fewer as compared to weekday births [911], the relative risk of stillbirth or neonatal mortality increases for weekend births [1113]. The provision of care, especially the availability of facilities that assist rapid delivery and prompt resuscitation, would likely be a factor for differential patterns of mortality for weekend versus weekday births [14]. Low staffing levels and reduced hospital efficiency during weekends have been suggested to be the cause of increased weekend mortality as compared to weekdays [11, 15].

Findings from studies suggest variations in the frequency, management and the rate of survival of both twins and singletons but there are inconsistencies with regards to mortality of twins and singletons at different gestational ages [16, 17]. It is also noteworthy that we are unaware of any study among twin births that has examined the contribution of weekend births to early mortality among twins. Accordingly, we undertook this study to estimate the risk for twin infant mortality encompassing neonatal, post-neonatal, and infant death among twins born on weekends as compared to those delivered on weekdays within a large population sample of twins.

Methods

This analysis was performed using the United States linked Birth/Infant death birth cohort dataset for the period of 1989 through 2002. The data were obtained from the National Center for Health Statistics (NCHS) at the Centers for Disease Control (CDC). Birth cohort files are favored for in-depth analysis because they follow defined birth cohorts for a whole year in order to determine the occurrence of mortality. The national matched twin files were not used because they lacked information on the exact date of death unlike the United States unmatched birth/infant death cohort dataset used for this study.

Study variables included in this analysis from the data files include day of birth and death, mode of delivery (cesarean or vaginal), pregnancy and labor complications, method of delivery, maternal socio-demographics and maternal lifestyle factors like smoking. The main outcome of interest was infant mortality (defined as death before 365 days). Infant mortality was further stratified into neonatal death (<28 days) and post-neonatal death (28–364 days). Weekday births refer to births that occurred from Monday through Friday inclusive, while weekend births were those that occurred on Saturdays and Sundays. Because inclusion of holidays as weekend births did not change the result, we provide in this paper results that include only weekends and exclude the few holidays of the year. Covariates considered in our analysis were maternal race, infant sex, maternal age, marital status, maternal education level and maternal smoking. Maternal race was defined as blacks or whites; maternal age was grouped as less than 18 years, 18–35 years and greater than 35 years. Maternal education level was categorized into two groups of less than 12 years and 12 years or more.

The rate of infant mortality was computed by dividing the total number of deaths by the total number of live births and multiplying the outcome by 1,000. Gestational age was established as the interval between the last menstrual period and the time of twin delivery. Also, because the association between neonatal and infant mortality and method of delivery could be confounded by other factors, we compared cesarean and vaginal delivery between weekend and weekday births across these factors to elucidate potential confounders. Chi-square test was used to assess differences in proportions. The Cox proportional hazard model was employed to perform the survival analysis. We used the Cox proportional hazards regression model to derive adjusted hazard ratios (HRs) after testing for non-violation of the proportionality assumption in each case. We confirmed this by plotting the log-negative-log of the Kaplan–Meier estimates of the survival function versus the log of time. The resulting curves were parallel [18]. Adjusted HRs were derived by loading all the variables that were considered to be potential confounders into the model. The computed hazard was determined by:
$$ h(t) = h_{0} (t)x\,{ \exp }\{ b_{ 1} x_{ 1} + b_{2} x_{2} + \cdots + {\text{b}}_{p} x_{p} \} $$
where, the hazard function is represented by h(t) and is determined by a group of covariates (x1, x2, …, xp) whose effect is measured by the size of the individual coefficients (b1, b2, …, bp) and t is the survival time of infancy. All hypothesis tests were carried out with a type 1 error rate set at 5 %. Analysis was conducted using the R statistical software, version 2.13.2. This study was approved by the institutional review board at the University of South Florida.

Results

A total of 879,966 twins were recorded in the database. Of this number, 170,269 (19.3 %) were born on the weekend while 709,697 (80.7 %) were born during the week. In Table 1, the frequency of the socio-demographic characteristics of the mothers is presented. In both the weekend and weekday delivery categories, over 80 % of the mothers were whites, non-smokers and fell between the ages of 18 and 35. In general, when comparison was made between mothers by day of infant birth (i.e. weekend versus weekday birth), significant differences were observed in maternal marital status, maternal age, prenatal smoking status and maternal race. The overall percentage of weekend births as compared to weekday births were lower among whites, mothers with less than 12 years of education, mothers that were married and mothers without pre-natal smoking. We also observed a difference in the sex of the baby as infants born on weekends were more likely to be males.
Table 1

Socio-demographic characteristics of mothers delivering in the United States by day of the week of birth, 1989–2002

 

Weekenda

Weekdaya

p valueb

170,269 (%)

709,697 (%)

Maternal age

  

<0.0001

 Less than 18 years

7,512 (4.4)

26,069 (3.7)

 

 18–35 years

137,990 (81.0)

574,744 (81.0)

 

 Above 35 years

24,767 (14.5)

108,884 (15.3)

 

Maternal race

  

<0.0001

 White

131,957 (80.9)

562,442 (82.8)

 

 Black

31,116 (19.1)

117,223 (17.2)

 

Infant sex

  

<0.0001

 Male

86,419 (50.8)

355,624 (50.1)

 

 Female

83,850 (49.2)

354,073 (49.9)

 

Maternal educational level

  

0.082

 Less than 12 years

76,104 (45.4)

318,854 (45.6)

 

 12 years or more

91,636 (54.6)

380,301 (54.4)

 

Marital status

  

<0.0001

 Married

119,989 (70.5)

519,738 (73.2)

 

 Not married

50,280 (29.5)

189,959 (26.8)

 

Mother’s smoking

  

<0.0001

 No

125,564 (89.6)

533,456 (90.3)

 

 Yes

14,619 (10.4)

57,413 (9.7)

 

aColumns do not total due to missing data

bSignificant p values are in bold font. p values of 0.05 or less were considered significant

The crude frequencies for pregnancy and labor complications in the study population are shown in Table 2. The rates were significantly higher in weekday birth for all pregnancy and labor complications except for placenta previa and anemia. The most remarkable difference was observed for cesarean section for weekday as compared to weekend delivery (43.18/100 for weekday births vs. 10.90/100 for weekend births, p < 0.0001).
Table 2

Rates of pregnancy and labor complications (in percentage) among twin births (weekend versus weekday birth), United States, 1989–2002

Type of complication

Weekend

Weekday

p valueb

(%)

(%)

Anemia

0.64

2.59

0.074

Eclampsia

0.19

0.76

0.007

Insulin-dependent diabetes

0.69

2.99

0.007

Chronic hypertension

0.18

0.86

<0.0001

Placental abruption

0.28

0.79

<0.0001

Placenta previa

0.09

0.38

0.092

Caesarean section

10.90

43.18

<0.0001

Only yes responses are reported

bSignificant p values are in bold font. p values of 0.05 or less were considered significant

Infant mortality rates by weekend versus weekday birth in the study population are presented in Table 3. Significant differences were observed in neonatal, post-neonatal and infant mortality between the two groups. Upon stratification by neonatal, post-neonatal and infant mortality, we observed that the neonatal mortality rate was higher on weekends (14.73 per 1,000) than on weekdays (10.30 per 1,000) (p < 0.0001). Post-neonatal death (6.13/1,000 for weekend vs. 4.55/1,000 for weekday, p < 0.0001) and infant death (16.32/1,000 for weekend vs. 11.98/1,000 for weekday, p < 0.0001) were also greater on weekends than on weekdays with infant death having the greatest occurrence and post-neonatal death having the smallest occurrence.
Table 3

Rates of neonatal, post-neonatal and infant mortality among twins (weekend versus weekday birth), United States, 1989–2002

Rates per 1,000 live births

Weekend

Weekday

p valuea

N (%)

N (%)

Neonatal death (<28 days)

1,747 (14.73)

5,208 (10.30)

<0.0001

Post-neonatal death (28–364 days)

1,026 (6.13)

3,191 (4.55)

<0.0001

Infant death (1–364 days)

2,773 (16.32)

8,399 (11.98)

<0.0001

aSignificant p values are in bold font. p values of 0.05 or less were considered significant

Table 4 shows the summary results for the adjusted HRs for neonatal, post-neonatal and infant deaths in relation to day of delivery and selected risk factors. The risk for post-neonatal death was increased for twins born during weekend as compared to those born during the week (HR = 1.19, CI: 1.04, 1.36). There was no significant difference between weekday and weekend deliveries for both neonatal death (HR = 0.90, CI: 0.80, 1.00) and infant death (HR = 0.99, CI: 0.92, 1.08). Single motherhood was noted as the most significant risk factor associated with post-neonatal death (HR = 1.54, CI: 1.43, 1.65). The risk for infant death was also increased among unmarried as compared to married women (HR = 1.19, CI: 1.13, 1.24).
Table 4

Adjusted hazards ratios (HR) for the association between day of the week of birth, pregnancy risk factors and early mortality among twins, United States, 1989–2002

Predictors

Neonatal death

Post-neonatal death

Infant death

HR (CI)

HR (CI)

HR (CI)

Day of the week

 Weekend

0.90 (0.80,1.00)

1.19 (1.04, 1.36)

0.99 (0.92,1.08)

 Weekday

1.00

1.00

1.00

Marital status

 Married

1.00

1.00

1.00

 Non married

0.99 (0.94,1.05)

1.54 (1.43, 1.65)

1.19 (1.13, 1.24)

Maternal age

 Less than 18 years

1.00 (0.86, 1.15)

0.91 (0.76, 1.08)

0.92 (0.83, 1.03)

 18–35 years

1.00

1.00

1.00

 More than 35 years

0.90 (0.83, 0.98)

0.87 (0.78, 0.98)

0.95 (0.86, 0.97)

Maternal race

 White

1.16 (1.08, 1.24)

0.68 (0.64, 0.76)

0.95 (0.91, 1.01)

 Black

1.00

1.00

1.00

Infant sex

 Female

0.72 (0.68, 0.75)

0.72 (0.68, 0.77)

0.72 (0.70, 0.75)

 Male

1.00

1.00

1.00

Maternal educational level

 Less than 12 years

1.00

1.00

1.00

 12 years or more

0.90 (0.85, 0.96)

0.73 (0.68, 0.79)

0.83 (0.79, 0.87)

Maternal smoking

 Yes

0.93 (0.88, 0.99)

1.15 (1.07, 1.23)

0.97 (0.93, 1.02)

 No

1.00

1.00

1.00

Type of delivery

 Caesarean section

0.83 (0.79, 0.87)

1.05 (0.98, 1.13)

0.97 (0.94, 1.01)

 Vaginal birth

1.00

1.00

1.00

 Gestational age

0.99 (0.98, 0.994)

1.00 (0.99,1.004)

0.99 (0.99, 0.997)

 Birth weight

0.99 (0.98, 0.998)

0.99 (0.99, 0.999)

0.99 (0.99, 0.998)

Significant p values are in bold font. p values of 0.05 or less were considered significant

HR Hazard ratios, CI 95 % confidence intervals

Female twins were at lower risk for all forms of infant death as compared to male twins. Lower risk of neonatal (HR = 0.90, CI: 0.83, 0.98), post-neonatal (HR = 0.87, CI: 0.78, 0.98) and infant death (0.95 (0.86, 0.97)) was observed in mothers that were more than 35 years old when compared with those 35 years old or less. Furthermore, lower risk for all forms of infant death was observed in mothers with 12 or more years of education (neonatal HR = 0.90, CI: 0.85, 0.96; post-neonatal HR = 0.73, CI: 0.68, 0.79; Infant HR = 0.83, CI: 0.79, 0.87).

Maternal smoking increased the risk for post-neonatal death by 1.15 times. Twins born through cesarean section were at lower risk of neonatal death when compared with twins born through vaginal delivery (HR = 0.83, CI: 0.79, 0.87). The HR shows that neonatal death was 1.16 times likely to occur among white mothers when compared with black mothers. However, the risk of post-neonatal death was lower among white mothers as compared to black mothers (HR = 0.68, CI: 0.64, 0.76). Some of these findings remained the same regardless of whether gestational age was accounted for or not in the adjusted model. Lower risk of neonatal and infant death was associated with increased gestational age while infants with increased birth weight were at low risk for all three mortality indices. It is to be noted that both gestational age and birth weight were loaded onto the adjusted models as continuous variables.

Table 5 presents the summary for infant mortality showing the effects of weekend birth interaction with some risk factors. Weekend twin birth by teenage mothers was found to increase the likelihood of neonatal death (HR = 1.35, CI: 1.06, 1.71). Compared to white mothers, weekend twin birth to black mothers increased the risk of post-neonatal death by about 20 %.
Table 5

Adjusted hazards ratios (HR) showing the effect of interaction of day of the week of birth with maternal race and age on neonatal, post-neonatal and infant mortality, United States, 1989–2002

Predictors

Neonatal death

Post-neonatal death

Infant death

HR (CI)

HR (CI)

HR (CI)

Weekend: maternal race White

1.09 (0.96, 1.24)

0.87 (0.75, 1.00)

0.99 (0.90, 1.09)

Weekend: maternal age <18 years

1.35 (1.06, 1.71)

1.00 (0.73, 1.38)

1.11 (0.92, 1.35)

Weekend: maternal age >35 years

0.96 (0.80, 1.14)

1.23 (0.98, 1.53)

1.04 (0.90, 1.19)

Significant p values are in bold font. p values of 0.05 or less were considered significant

HR Hazard ratios, CI 95 % confidence intervals

Discussion

The increased incidence of multiple births within the previous decades was due partially to delayed child-bearing amongst women, but mainly fueled by the enhanced use of assisted reproductive techniques [2, 19, 20]. The US national data used for our study is a large population database which produces statistically stable and precise results and allows for an assessment of the effects of day of the week of birth in relation to twin mortality as a result of the ample sample size. Based on crude estimates, the study findings showed an increased risk of neonatal, post-neonatal and infant death among women who delivered on weekends than on weekdays. However, after adjustment, this finding was confirmed for post-neonatal death only. Our study findings to some extent reaffirm the evidence from previous studies that there is an increased risk associated with weekend delivery than weekday delivery [14, 21]. We cannot eliminate the likelihood that patients admitted on weekends were in a more critical condition than those admitted on weekdays. Nevertheless, a higher severity of illness among patients admitted to intensive care units in hospitals on weekend questions the adequacy of medical care and staffing patterns on weekends [13]. The study findings also showed higher rate of cesarean section among women who delivered on weekdays than on weekends which could be due to planned cesarean deliveries. Unfortunately, the dataset lack the information of whether the cesarean was elective versus emergency.

Another significant finding in this study is the evidence that emphasizes the importance of maternal education. Maternal education appeared to be inversely associated with infant mortality because a lower risk was found among births to mothers that had 12 or more years of education as compared to those with less than 12 years of education. Maternal age was also influential in our findings. The results showed that an increased risk of neonatal mortality was associated with teenage mothers who delivered during the weekend. Previous studies have shown that teenage pregnancies, in general, are associated with an increased risk of neonatal mortality [22, 23]. The increased risk may be linked to biological immaturity [24] although other factors such as unplanned or emergency cesarean sections might be relevant as likely explanations for the heightened risks for neonatal mortality in general and among twins born to teenagers during the weekend in particular. It is reasonable to expect care providers to plan cesarean deliveries during the weekdays when staff is adequate and readily at hand as compared to the weekend period. This would imply the preponderance of planned cesarean sections during the weekdays as compared to the weekends. This proportional difference could have confounded our findings since the weekend cesarean sections would disproportionately consist of high-risk (emergency or unplanned) cesarean sections. This would make twins delivered through cesarean sections on weekends to have higher neonatal mortality. Unfortunately, the population database used for this study lacks information on the relevant variables that would permit a more detailed quantitative analysis of the extent of this possible source of confounding. Interpretation of our findings should therefore, take this shortcoming into consideration.

The study finding also showed that female twins were at lower risk for all forms of infant death as compared to male twins. The understanding of why males have poorer survival compared to females is sparse and only a fraction of the difference can be explained by the often more hazardous lifestyle of males [25]. It seems likely that also biological factors contribute to the different survival pattern. This is supported by the fact that the males’ disadvantage begins in uterus [26]. The differential in survival could be linked to the genetic difference between males and females: females have two X chromosomes, males have one X chromosome and a small Y chromosome. This means that females are a mosaic of two different cell lines: one cell line with the X chromosome from the father being active and another cell line with the X chromosome from the mother being active. Males have only one cell line because they receive only one X chromosome. Female scenario with two cell lines in all organs provides a health advantage, which is clearly the case for X-linked diseases, such as color blindness, hemophilia and Duchenne’s disease [26]. It seems plausible, not only in the case of such X-linked diseases but also for mutations with more subtle effects, that having two cell lines offers an advantage to females compared to males. This advantage will tend to be greatest in tissues with many cell divisions such as blood cells or mucosa cells. Furthermore, the predominance could be tissue specific with one cell line being predominant in one organ due to better survival while the other cell line could be more frequent in another organ. Overall, this may contribute to the longer lifespan of females which may begin in utero [27].

Previous researches have suggested that twin infants born to black parents have a higher risk of fetal and infant mortality when compared with their white counterparts [4, 28]. Counterintuitively, this study showed a slightly higher risk of neonatal mortality among whites as compared to blacks. However, the interaction between weekend births and maternal race demonstrated that white mothers were at a lower risk of post-neonatal mortality as compared to black mothers. According to a study by Mackey and colleagues, it was suggested that social and economic conditions, ecological stressors and some distinct lifestyle practices among black mothers could make the weekend a high risk time for early labor and hence, an important cause of infant mortality [29].

The results in this study bear some limitations. We could not assess weekend mortality and its association with level of staff on duty and workload allocated to these staff unlike previous researches that ascribed an elevated mortality risk with weekend admissions to excess staff workload [30] and sub-optimal staffing levels [31]. Additionally, we could not dispute the fact that most patients admitted during the weekends were more seriously ill than those admitted on weekdays. Although we did adjust for birth weight and gestational age (both representing indicators of infant well-being at birth), residual confounding due to illness severity could not be ruled out as influential in our findings.

Our study as well as some other investigators that evaluated the risk of infant death during weekends as compared to weekdays did not consider the time of birth [11, 12]. This disregards the fact that the majority of weekday births occur outside the normal working hours when the quality of care could be said to be similar to what obtains on weekends [14]. On the other hand, studies that measured the time of birth provided an in-depth analysis of the 24-h period and were able to identify the high-risk care period [8, 32, 33].

A key strength of our study is the substantial sample size of the data used in our analysis. The data were extracted from over 800,000 twin birth records making it the largest population-based study on twin delivery during the weekend. The large sample size also confers several advantages associated with study validity: there were no issues of sample error hence, parameters estimated are a reflection of true population value. The use of a national population database also makes our outcome less likely to be influenced by selection bias and provides valuable and reliable information for future studies in perinatal twin research.

Conflict of interest

None.

Copyright information

© Springer-Verlag 2012