In this study, we examined the nature, timing, and correlates of divorce in families of children with Down syndrome (647), other birth defects (10,283) and no identified disability (361,154). Divorce rates among families of children with Down syndrome were lower than in the other two groups. When divorce did occur in the Down syndrome group, however, a higher proportion occurred within the first 2 years after the child's birth. Mothers and fathers of children with Down syndrome were much more likely to divorce if they were younger, had not graduated from high school, and if fathers were less educated and lived in a rural area. Few effects on divorce were noted for a variety of family structure variables.
Although Down syndrome has been studied for over 100 years, we continue to have large gaps in our knowledge. One continuing issue concerns the effects of children with Down syndrome on their families. Several researchers have spoken of a “Down syndrome advantage,” the idea that due to certain child characteristics, the parents and families of children with Down syndrome cope better than do parents and families of children with other disabilities (Hodapp, Ly, Fidler, & Ricci, 2001; Seltzer & Ryff, 1994). The presence of such an advantage, however, has long been debated, and parental age, education, and socioeconomic status (SES) may all constitute other-than-child variables potentially responsible for the well-being of these families (Cahill & Glidden, 1996).
Unfortunately, much of this debate takes place in the absence of objective data about many aspects of family functioning, particularly data from large-scale studies that focus on Down syndrome. Consider the seemingly straightforward question of parental divorce. Compared to parents of same-aged typically developing children, is divorce more, less, or equally common among parents of children with Down syndrome? Currently, researchers disagree. Gath (1977) compared families of 30 children with Down syndrome to 30 families of same-aged typically developing children, interviewing each group multiple times over the first 2 years of their child's life. By the time the child was 24 months, 6 couples in the Down syndrome group had divorced, separated, or were showing particularly poor marital relationships, whereas none of the 30 control couples showed these signs of marital distress. This early evidence suggests that marriages in the Down syndrome group may be troubled in the child's first 2 years of life.
In contrast, other researchers have not found such high rates of marital problems among parents of children with Down syndrome. Comparing families of 6- to 17-year-old children with Down syndrome to those with heterogeneous mental retardation of similar levels of impairment, Gath and Gumley (1986) noted that 23 of 147 couples (15.6%) were divorced among families of children with heterogeneous mental retardation compared to a divorce rate of only 5.2% (10 of 192) among parents of children with Down syndrome. Thus, for families with school-age children with Down syndrome, divorce was less likely. Similarly, comparable or lower divorce rates have been found in longitudinal and cohort studies. In her 30-year longitudinal study, Carr (1988) noted that parents in both the Down syndrome and control groups were about evenly likely to divorce, with slightly fewer divorces in the Down syndrome group at each age. In a study in which 181 children from Manchester (England) were examined, Cunningham (1996) even concluded that “The divorce rate in the cohort has always been lower than the national average” (p. 89).
A further issue concerns the correlates of parental divorce in the Down syndrome group. In addition to potentially different divorce rates based on the age of the child, various researchers have speculated on parent and family factors. In their study of 6- to 17-year-olds, Gath and Gumley (1986) noted that although good marriages were distributed throughout the social classes, “very poor marriages” (p. 164) tended to occur more frequently in the lowest social classes. Using the Manchester Down syndrome cohort, Sloper, Knussen, Turner, and Cunningham (1991) also found SES risk factors for both maternal and paternal satisfaction and malaise scores, although the exact variables and the strength of such predictors varied across reporters and measures.
Summarizing the existing literature, many questions remain about the amount and correlates of marital difficulties among families of children with Down syndrome. Such unanswered questions persist partly because of the many difficulties involved in performing these studies. Most researchers have access to only small numbers of children and their families, making it difficult to compare divorce rates in families of children with versus those without Down syndrome. Even in the larger studies (e.g., the Manchester cohort), researchers examined a low prevalence event—divorce during certain childhood periods—in moderate-sized (100 to 200) samples. The combination of low prevalence of divorce and small samples yields studies with minimal power to detect even moderate-sized effects. As a result, the amount, timing, and correlates of parental divorce all remain unclear in this group.
In contrast, many aspects of divorce are well known for families of children who do not have disabilities. Within these families, parental divorce occurs less often during the preschool years, with rising rates beginning when children are about age 6 (Cherlin, 1977; Waite & Lillard, 1991). The risk of divorce may be lower if the child is a boy (Morgan, Lye, & Condran, 1988; but see also Diekmann & Schmidheiny, 2004) and is higher among parents who marry younger, have less income, and are less educated (Bramlett & Mosher, 2002). Other factors, such as whether the family resides in a rural versus nonrural area, also affect divorce rates, with nonrural rates exceeding rural rates in most studies (Breault & Kposowa, 1987). Each of these findings has been reported in several large-scale epidemiological studies.
To examine divorce in families of children with Down syndrome, we employed a large-scale, population-based framework. To ascertain divorces of couples with children who have Down syndrome versus those who do not, we utilized the linked databases of all births, divorces, and hospitalizations in the state of Tennessee between 1990–2002. Because we examined records of divorce rather than subjective measures of marital satisfaction, we avoided the problems inherent in self-report measures, such as parental malaise or satisfaction. In addition, by identifying over 600 families of children with Down syndrome, over 10,000 families of children with other birth defects, and several hundred thousand families of children who did not have disabilities, we were able to perform analyses involving family-structure variables that are impossible in studies with small samples.
We focused on three issues. First, we compared the amount and timing of divorce among families of children with Down syndrome to families in the two comparison groups. Second, we examined whether the correlates of divorce, noted in the literature for families of typically developing children, are significant predictors of divorce in families with children who have Down syndrome. Third, we further examined the effect of family structure, such as family size and birth order, on divorce among parents of children with Down syndrome.
The final linked Birth–Divorce–Hospital Discharge dataset for this study included the families of 647 children with Down syndrome and 361,154 families of comparison children who did not have any record of congenital birth defects (the comparison group). From the congenital anomalies codes in the Birth Records, we identified an additional 10,283 families of children with birth defects other than Down syndrome. Such children included those whose birth records indicated diagnoses of one or more of the remaining 21 birth defects. The most prevalent specific categories included other, musculoskeletal/integumental conditions (16.6% of all recorded birth defects); other, urogenital anomalies (11.2%); heart malformations (10.8%); and polydactyly/syndactyly/adactyly (7.7%). Counts by congenital anomalies were not unduplicated in that any one child could be diagnosed with more than a single birth defect. Families of these children constituted our other birth defects group.
To obtain the study dataset, we examined official records for the entire population of children and parents from the state of Tennessee between 1990 and 2002, namely, 5.8 million people. The state features a diverse population, with 79.2% White nonHispanic, 16.4% African American, 2.2% Hispanic, and 2.2% other. Much of Tennessee's land area—including 67 of the state's 95 counties—is designated as rural by the federal government, although the state also has five large cities: Memphis, Nashville, Knoxville, Chattanooga, and Clarksville (all with populations over 100,000). Tennessee has a low rate of in- or out-migration, with an estimate for 2002 of 0.45% (areaConnect, 2004) and an excellent set of state-level administrative records over periods of up to 20 years. Taken together, it is highly likely that individuals born or marrying in Tennessee remain in Tennessee and that their divorces and hospitalizations will be recorded in the vital records and hospital discharge data systems.
To compare children in the two disability groups (Down syndrome and other birth defects) with families in the comparison group, we randomly chose one child from each family as the index child. The child with Down syndrome and the child with any of the remaining 21 other birth defects constituted the index child in families of children in the remaining two groups.
As shown in Table 1, simple comparisons show several differences in parents, families, and children across the three groups. On average, families of children with Down syndrome had more children, and parents were more highly educated and older at the time of the index child's birth. Children with Down syndrome were more likely to be male, latter-born, and the family's last child; were less likely to be the family's first or only child; and equally likely to be the first of multiple children. Although families of children with Down syndrome differ from families of both children without disabilities and those with other birth defects, the significant differences had small or tiny effect sizes. With the exception of mother's age, parent's education, and child's age at parental divorce, there are no significant interactions with divorce status.
From 1990 to 2002, on average, 81,253 children were born each year in Tennessee. Official birth records contained approximately 140 variables that described the infant (birth weight, APGAR scores at 1 and 5 minutes), mother (age, number of live births, race, education, marital status), and father (age, race, education). The birth file has specific fields for Down syndrome and for 21 other specific birth defects. Because only about half of all children with Down syndrome were identified from the birth records (D. Law, personal communication, June 15, 2004), we relied on hospital discharge records to identify additional children with Down syndrome.
Available from 1996, the hospital discharge data system records any hospitalization (in- or out-patient), for all children and adults within the state. In addition to demographic information, the hospital discharge file provides a principal International Classification of Diseases–ICD-9 (World Health Organization, 1977) diagnostic code and up to 8 secondary diagnostic codes, procedures performed, days and units of service required, and insurance/payer information. Using the primary and all secondary diagnosis data, we identified additional children with Down syndrome who did not have this syndrome recorded in their birth record.
From 1980, Tennessee divorce records provide information about divorcing spouses and the family (426,567 cases). In addition to spouses' names, ages, races, and county of residence, these records give the dates of marriage and of divorce, duration of marriage, number of marriages, number of children less than 18 years old in the family at the time of the divorce, and custody type.
With approval from Tennessee's Department of Health, we obtained birth and divorce vital statistics datasets and hospital discharge data system dataset. First, we identified children with Down syndrome and children with other birth defects who were identified at birth. Examining all Tennessee birth records over the years 1990 through 2002, we selected children with Down syndrome who either had a congenital anomaly code of Down syndrome in the birth records or one or more hospital discharge records with the ICD-9 code for Down syndrome (758.00) in any of the diagnosis fields. In approximately 80% of cases, hospital discharge records were linked to birth records using the variables of the child's and mother's social security number. In remaining cases, we linked using father's social security number and the child's birth month, day, and year (child race and gender also had to match).
Having identified children with (a) Down syndrome, (b) other congenital birth defects in the birth records, and (c) without birth defects, we then linked birth records with divorce records (see Sharma et al., 2000, for similar linkage procedures using Missouri birth and hospitalization records). Because both birth and divorce records have multiple potentially good linking variables, we used a stepwise, deterministic linkage procedure to link these records. The variables used were wife and husband identifiers (social security number, name, and date of birth) from the divorce records and mother and father identifiers in the birth records (see Urbano, 2007). The first link variables used were social security numbers. Using social security numbers alone produced successful matches in over 63% of divorces over all years. During the years 1996–2002, a period during which social security numbers were consistently recorded in the divorce records, 85% of divorces of couples with children were matched to birth records. Combinations of the remaining identifiers were used for records that were missing social security numbers. At this point, the revised individual birth records had information recorded pertaining to group classification and divorce status. If parents divorced after the index child's birth, the variable divorce was coded 1 for these children, the date of divorce and the record number in the divorce data set were also recorded. The percentage of unlinkable individual birth records were lower for children in the comparison group compared to the two disability groups, χ2(2, N = 1,056,284) = 6618.2, p < .0001.
Using second-order linkage (Tu & Mason, 2004) to create family groupings, we then collected individual children into family groups using mother's identifiers (social security number, name, date of birth) and birth order from the birth records. All family units consisted of children 13 years of age or younger born in Tennessee between 1990 and 2002 to the same mothers. These family-level data allowed us to examine the influences on parental divorce of family and child characteristics such as gender, birth order, family size, and family structure (see Figure 1).
To compare rates of divorce in the three study groups (Down syndrome, other birth defects, comparison), we performed an initial omnibus logistic regression analysis (SPSS, 2006). The outcome variable was divorced (0 = not divorced, 1 = divorced). The modeled independent variables were group, index child's age at divorce, index child's gender, parents' education, mother's age at birth of index child, rural status, plurality at birth of index child, mother's race, index child number, family size, and index child is only child. All main effects, 2- and 3-way interactions of the independent variables included in the logistic regression model were tested. A summary of the logistic regression with all significant main and interaction effects is presented in Table 2. Because interactions with group are of primary concern, they are reported first. To make it easier to follow the progression from main effects to interactions, we formatted Table 2 grouping main effects and two-way interactions together with the tests of their interaction contrasts. Three-way interactions and the tests of their interaction effects were presented next. Except where noted, only significant main effects and interactions from the omnibus logistic regression model are reported in the Results section.
The first question concerned our ability to identify the large majority of Tennessee's children with Down syndrome who were born between 1990 and 2002. Given that prevalence rates for Down syndrome are generally considered to range from 1 per 1,000 to 1 per 800 live births (Krivchenia, Huether, Edmonds, May, & Guckenberger, 1993; Stoll, Alembik, Dott, & Roth, 1990), we expected between approximately 80 and 100 Down syndrome births each year. Over 13 years of data and including all 1,056,284 Tennessee births, we expected between 1,056 and 1,320 children with Down syndrome. Across all years, we identified 918 children with Down syndrome, or 86.9% of all children given a yearly incidence rate of 1 per 1,000 births, 69.5% given a 1 per 800 incidence rate.
As Figure 2 shows, our identification rates rose markedly from about 1994 onward, becoming stable from 1996 to 2002. We attribute the inability to identify many children with Down syndrome prior to 1996 as mainly due to the lack of hospital discharge records during these years. Indeed, in most every year from 1996–2002, our rates were between the predicted values of 0.12% (or 1 per 800 births) and 0.10% (1 per 1,000), equaling 0.103% overall (or 1 per 970.8 births). Of the 869 families with a child who had Down syndrome, 7 had 2 children with Down syndrome, with 5 sets of twins. Of the 869 families of children with Down syndrome, 74.5% (647) of the mothers were married at the birth of the index child; among families with children who did not have disabilities and with other birth defects, 66.4% (361,154) and 59.3% (10,283), respectively, of the mothers were married, χ2(2, N = 561,945) = 410.27, p < .0001.
Because our rates of children identified as having Down syndrome increased in the post-1996 period, our findings concerning divorce might have been due to this differential detection rate from 1990–2002 compared to 1996–2002. To investigate this possibility, we re-ran all analyses using families of index children born from 1996 through 2002. Due to fewer years of available data (from 1996 vs. 1990), decreases occurred in both the number of participants examined and in the exact rates of divorce (due to fewer years after the child's birth in which the parents could potentially experience divorce). In no case, however, did differences occur in the patterns of group differences or in the correlates of divorce. We, therefore, report below the findings from our entire sample of families of index children born throughout the 1990 to 2002 period.
The families in this study were predominately White non-Hispanic (90%), with the remaining 10% distributed between African American (8%) and other (2%). Race did not significantly alter the Group × Divorce effects.
Prevalence and Timing of Divorce
Families of children with Down syndrome divorced less often than did families in the other two groups. The divorce rate for the families of children with Down syndrome was 7.6% (49/647) compared to a divorce rate of 10.8% (38,950/ 361,154) in the comparison group and 11.2% (1,155/10,283) in the other birth defects group, χ2(2, N = 372,084) = 9.05, p < .05. In addition to the overall effect of group, there were also significant interactions between group and other factors.
Most notably, differences occurred in the timing of divorce in the three groups. For families of children with Down syndrome who did divorce, 32.7% (16 of 49) of all divorces occurred before the child reached 2 years of age compared to only 17.4% (6,796/38,950) of all divorces in the comparison group of children with no disabilities, and 14.9% (172/1,155) in the other birth defects group, χ2(2, N = 40,154) = 13.05, p < .001. Of families who divorce after the birth of the index child, families of children with Down syndrome were almost twice as likely to divorce during the first 2 years of the child's life.
To better understand this finding that when divorce did occur in the Down syndrome group, it more often occurred when the child was below 2 years of age, we performed two additional analyses. First, we examined at what point within the first 2 years these divorces occurred. If, for example, early divorces predominated, one might be tempted to conclude that divorce in families of children with Down syndrome occurred in response to the pre-existing knowledge that the couple was about to give birth to a child with Down syndrome (as opposed to the actual birth and early experience of the child per se). Although based on small numbers, 6 divorces (or 12.2% of all divorces in the Down syndrome group) occurred before the child turned 1 year of age, with 10 (20.4%) occurring when the child was between 1 and 2. We compared these percentages to the 5.2% and 12.2% of divorces in the comparison group in the birth and first years and the 4.2% and 10.7% of divorces in the birth and first years among families of children with other birth defects.
Given that most pregnant women are advised to have amniocentesis if they are 35 years of age or older, we examined the ages of the mothers within the group experiencing divorces before age 1 and from 1 to 2 years. Of the 0-year divorces, the average age of the mothers was 32.33 years, with 3 of the 6 mothers age 35 or above. Within the first year divorces (i.e., when the child was between 1 and <2 years of age), 4 of the 10 mothers were 35 years of age or older, with the average age equaling 31.10 years. In roughly half of all Down syndrome divorces that occurred within either of the first 2 years of the child's life, then, mothers may have known that their newborn would have Down syndrome.
Correlates of Divorce Previously Found Among the General Population
Gender of the index child
In contrast to prior research, the gender of the child did not influence the chances of parental divorce in any group. For families in the group without disabilities, the divorce rates were virtually identical when the index child was a boy (10.7%) or a girl (10.9%), χ2(1, N = 361,154) = 3.07, ns. Similarly, among families in the two groups of children with disabilities, the divorce rates were not significantly different when the child was a boy as opposed to a girl (Down syndrome = 8.1% vs. 7.0%, respectively; other birth defects = 11.4% vs. 11.1%).
Age of parents
In each of the three groups, fathers and mothers in divorced families were younger than those in nondivorced families. Among families in the comparison group, mothers' mean age if divorced was 24.6 years, and if not divorced, 27.3; for fathers, the mean ages were 27.3 and 29.8 years, respectively, ps < .0001. Similar findings—with similar age differentials—were seen among divorced versus married parents of children with other birth defects. As expected from the demographics of Down syndrome, mothers of children with Down syndrome were 4.0 years older—and fathers over 3.5 years older— than parents of comparison and other birth defects children. In the Down syndrome group as well, both mothers and fathers who divorced were younger than were those who remained married (mothers: 29.00 vs. 31.41 years of age; fathers: 29.91 vs. 33.10).
There was also an interaction between maternal age and group. To examine this effect, we first divided ages into quartiles of below 23 years, 23– 27 years, 27–31 years, and above 31 years. In the combined sample, divorce rates decreased linearly across the four quartiles (19.9%, 10.4%, 7.3%, and 5.5%, respectively), χ2(3, N = 371,478) = 8992, p < .001. Within the Down syndrome group, however, the youngest mothers (< 23 years of age) had significantly increased levels of divorce compared to all three groups of older mothers (18.2% vs. 4.5%, 4.9%, and 5.9%, respectively), Group × Age Group, χ2(3, N = 371,478) = 8748, p < .001. Living in a rural area further amplified this age effect, with the oldest mothers (age > 31) of children with Down syndrome living in rural areas having no divorces whatsoever (0/32 compared to a 6.8% divorce rate for older mothers living in nonrural areas).
Education levels of parents
Across all three groups, parents with more years of formal education were less likely to divorce. In the two non-Down syndrome groups, 17.6% (comparison) and 16.9% (other birth defects) of couples were divorced when mothers had not completed high school compared to 10.0% and 10.4% when mothers were high school graduates or beyond, ps < .0001. Similar findings occurred among fathers who had not graduated high school: 17.4% and 15.5% of couples had divorced in the comparison and other birth defects groups, respectively, compared to rates of 9.9% and 10.5% among fathers in these groups who had at least a high school education, ps < .0001. Overall, divorce was almost twice as likely for parents without a high school education in the two non-Down syndrome groups, OR = 1.92, 95% CI = 1.85, 2.00.
This pattern was even more pronounced among families of children with Down syndrome. Among mothers without a high school education, 19.6% (11/56) divorced, compared to a 6.4% (38/ 552) rate among mothers with a high school degree or more formal schooling, Fisher's Exact Test, p < .002. Similarly, whereas 23.6% of fathers who had not graduated from high school were divorced, divorces occurred among only 5.4% of those with at least a high school degree, χ2(1, N = 645) = 30.76, p < .0001. Compared to parents who had high school or greater levels of education, then, the probability of divorce in mothers without high school educations is three times more likely, OR = 3.57, 95% CI = 1.69, 7.14, p < .002; and for fathers without high school educations, the probability was five and a half times more likely, OR = 5.56, 95% CI = 2.78, 10.0, p < .002. Such extremely high relative probabilities of divorce were more pronounced in the Down syndrome group compared to the other two groups, χ2 = 2081.38 (mothers) and 2108.13 (fathers), respectively, ps < .0001.
Rural versus nonrural comparison
Within Tennessee, 67 of the state's 95 counties are designated as rural by the Office of Rural Health Policy. Defined by a low population density and a lack of many services common to more densely populated areas, the rural designation allows qualifying counties to apply for targeted federal programs (Office of Rural Health Policy, 2004).
Although across all groups combined families showed slightly higher, albeit not statistically significant, rates of divorce in rural versus nonrural areas (14.1% vs. 10.5%), the effects of rural living status occurred mostly in the interaction between less educated and rural parents of children with Down syndrome. As Figure 3 shows, in all three groups proportionally more divorces occurred among both rural and less educated fathers. Within the Down syndrome group, however, this interaction effect was much more pronounced. Indeed, 40.9% (9 of 22 families) of rural fathers with less than a high school education experienced divorce. Within the Down syndrome group, such percentages differed significantly from more highly educated rural fathers (2.2%, or 2/92), χ2(1, N = 114) = 30.56, p < .0001, and from less educated fathers living in nonrural areas (17.0%), χ2(1, N = 69) = 4.61, p < .05. Although based on relatively small numbers, exceptionally high rates of divorce characterized less educated, rural fathers of children with Down syndrome.
Effects of Family Structure
Although often discussed in the literature, family structure had little effect when comparing families of children with Down syndrome versus families of the other two groups. No Group × Family Structure interaction was statistically significant. No differential effect on divorce rates occurred when the child was the only child in the family; the first-, last-, or latter-born child; when the family was small (i.e., ≤ 2 children) versus large (3+ children); when there were (or were not) other brothers or sisters in addition to the index child; or when there were older or younger brothers or sisters. Three family structure variables did, however, show significant main effects, ps < .001. In all three groups, divorce rates were lower as the index child increased in birth position (from first, to second, to third child, etc.), when the index child was not the only child in the family, and when the family had more than 2 children.
This study is among the first with an epidemiological approach to studying divorce in families of children with Down syndrome. Using statewide administrative databases over a multiyear period, we were able to bring to family research in Down syndrome a large-scale approach and a set of procedures common in studies of marriage and divorce among families of non-Down syndrome children. As a result of “second-order” linkage procedures (Tu & Mason, 2004), we examined the amount, timing, and correlates of divorce in this population, a topic that heretofore has only been examined in smaller, less representative samples.
This study featured three main findings. The first concerns the prevalence of divorce among parents of children with Down syndrome. In earlier studies of disabilities in general, including the few large-scale, epidemiological studies of this topic, researchers found that divorce occurs proportionally more often among parents of children with (versus without) disabilities (Fujiura, 1988; Hodapp & Krasner, 1995; Mauldon, 1992; Witt, Riley, & Coiro, 2003). Granted, the magnitude of these differences has never been large; in a recent meta-analysis, Risdal and Singer (2004) reported an average of 5.97% more divorces among couples parenting a child with a disability (the median effect size equaled .21). In contrast, in this study we found the opposite pattern: Parents of children with Down syndrome were slightly less likely to divorce than were parents of children without disabilities or with other birth defects. Although the lower prevalence of divorce in families of children with Down syndrome conforms with the earlier hypothesized “Down syndrome advantage,” the reasons for such an advantage may relate to child factors, factors associated with Down syndrome (e.g., older, more educated, and more often married parents), or both child and associated factors.
A related finding concerns the timing of divorce in families of children with Down syndrome. When divorce did occur in these families, it more often occurred during the child's first years of life. Over 30% of all divorces in the Down syndrome group occurred before the child reached 2 years, compared to only 17.4% of early divorces in families of children without disabilities. As the percentage of early divorces in the other birth defects group (14.9%) was similar to the pattern found in the comparison group, this overabundance of early divorce (i.e., when divorce occurs) may indeed be specific to families of young children with Down syndrome.
Although reasons for such early divorce remain unclear, several possibilities seem likely. First, consistent with earlier theories (e.g., Solnit & Stark, 1961), for most parents the birth of a child with Down syndrome involves the violation of parental expectations that may elicit from parents' strong depressive reactions at intermittent times over the child's first few years (Emde & Brown, 1978). Indeed, in Glidden and Schoolcraft's (2003) studies of adoptive versus birth parents of children with disabilities (including Down syndrome), the birth mothers seem to experience fairly high levels of depression during the child's earliest years. Second, in a recent study Most, Fidler, Booth-LaForce, and Kelley (2006) reported that stress among parents of children with Down syndrome increased over the first 3 years of life and seemed to be tied to the child's greater cognitive–linguistic deficits and maladaptive behaviors. Finally, early divorce might link to the high rates and severity of medical issues in children with Down syndrome, particularly during the earliest years (So, Urbano, & Hodapp, 2007). Any or all of these factors might explain the prevalence of early divorce among parents of children with Down syndrome.
A second major finding involved the parent characteristics of low levels of education and rural status. Although more divorce occurred among less educated parents in all groups, parents of children with Down syndrome were especially likely to divorce when either the mother or the father did not have a high school education. Similarly, in the families of children with Down syndrome, those rural families in which the fathers had not graduated high school seemed particularly at risk for parental divorce.
At present, we do not know why Down syndrome families are at most risk for divorce when headed by less educated parents or when the family is both rural and had less educated fathers. There are, however, several likely explanations associated with the ways in which rural areas are poor in many different types of resources. In Tennessee, for example, the per capita income in the state's rural areas is 74% that of metropolitan areas, and 15.8% of the rural population live in poverty versus 12.8% of the metropolitan population (United States Department of Agriculture, 2005). In terms of disability services, 6 of the 9 Tennessee Department of Education 0–3 Early Intervention programs are located in metropolitan areas, as are all three of the state-supported centers that provide services for children with metabolic and/ or genetic conditions. Families must both find and access services, a process that may be more difficult in rural areas and even more difficult for less educated families. It may also be the case that more educated rural fathers were less likely to divorce (2.2% vs. 7% base rate), a finding that, if replicated, would also have policy implications. Issues of parental education and rural living deserve more attention, and future studies are needed that replicate our Down syndrome finding of exceptionally high divorce rates among less educated rural fathers.
A third issue concerns what we did not find. From earlier writings, we expected aspects of the target child or of the family's structure to influence the amount of parental divorce. From the demographics literature, the presence of boys has been shown to lessen the likelihood of parental divorce (Morgan et al., 1988). From the disability literature, we expected that the presence of an older daughter with no disabilities might be helpful to the parents, while, possibly, being less helpful to the daughter herself (Lobato, 1983). Neither finding emerged as a strong predictor of parental divorce in this study. As Table 2 shows, whether the child was first, only, last, or middle born; whether the family was large or small; had sons or daughters in addition to the index child; or featured the presence of an older or younger daughter or son with no disabilities had no differential effect on whether the parents divorced. Indeed, the only three family variables affecting divorce involved latter birth order, not being an only child, and coming from a larger family; and each of these operated similarly in all three groups.
Again, we are uncertain as to why family structure variables played so little a role in family divorce. It may have been that, in some sense, our measure of family functioning—divorce—is simply too molar. There may indeed be important effects on families of having older sisters with no disabilities, bigger families, or other family structure variables, but such variables may be insufficient, by themselves, to show strong associations with marital dissolution. From the other direction, having a child with Down syndrome, particularly for less educated or rural and less educated parents, may produce a type of diagnostic overshadowing that constitutes the sole salient consideration for some couples already in unsteady marriages.
Taken together, the results of this study have important practical and theoretical implications. Practically, parents of newborns can be counseled both about the risks and timing of possible marital discord. For many families, especially those steeped in the still commonly heard notion that “divorce is rampant” among families of children with disabilities, it may be comforting to know that divorce is neither a necessary nor a common outcome of having a child with Down syndrome. At the same time, however, hospital social workers and early interventionists can educate parents about common stresses that arise during the earliest years. Those parents with less education can be especially targeted, as can those parents—especially fathers—who are both less educated and who live in rural areas. For example, like parent groups representing most syndromes, neither of the United States' two main parent groups in Down syndrome currently feature special programs designed for outreach to rural families. If, indeed, such parents are having more difficulties, both parent groups may want to extend their efforts to this less often included group.
More generally, results of this study help to complete the picture of families of children with Down syndrome. Although this syndrome has been the focus of family research for many years, gaps remain, partly due to the widespread use of small-scale samples of convenience. Use of statewide administrative databases circumvents many of these problems. Such large-scale, epidemiological studies allow for (a) larger samples (of all groups) than is generally achieved in most family studies; (b) analyses that include more complicated, complex interactions (as in our findings about the risks of divorce among rural fathers with less than high-school educations); and (c) inclusion into the study itself of groups that are rarely examined, such as parents who are younger, have less education, or live in rural areas.
At the same time, as with many epidemiological studies, this study has its own limitations. One issue concerns the ability, even with sophisticated multivariate techniques, to truly “equate” families that differ on a variety of demographic variables. Another issue concerns our lack of psychological information on the child, parents, or families. Without such data, it remains unclear why divorce in Down syndrome, when it does occur, is more often found in the earliest years after birth or why less educated and rural–less educated parents are having more trouble. With the development of short, easy-to-administer measures of health, depression, coping, burden of care, and well-being (Mezzich & Ustin, 2006), however, it would seem possible to include measures in future epidemiological work that make it possible for researchers to establish the mechanisms underlying their findings.
This study is also characterized by other, more practical limitations. For example, we used administrative records collected throughout the state. Given that these records are important for legal and economic reasons, we presumed that records were, in most cases, carefully collected and resulted in valid data, but we cannot know that for certain. Similarly, although our identification of children with Down syndrome conforms reasonably closely to presumed incidence rates, no “gold standard” exists to determine how precise our procedures were for specificity and sensitivity. Finally, although the other birth defects group allowed us to determine whether our family findings were specific to families of children with Down syndrome, this group included children with many less serious conditions and excluded children with many types of intellectual disabilities that are only diagnosed at later ages.
Even given these limitations, this study provides the first large-scale examination of the amount, timing, and correlates of divorce among families of children with Down syndrome. In contrast to earlier large-scale studies of children with disabilities more generally, divorce in the Down syndrome group occurred slightly less often than among families of children without disabilities or with other congenital birth defects. When divorce did occur among families of children with Down syndrome, higher proportions occurred early, before the child with Down syndrome reached 2 years of age, and were more related to low levels of parental education, especially for fathers who had low education and also lived in rural areas. Although much more remains to be known, this study begins to delineate the processes by which parents are, or are not, adversely affected by parenting their children with Down syndrome.
Hospital discharge data from the Tennessee Hospital Discharge Data System, 1996–2002, and Vital Statistics datasets, 1990–2002, were provided by the Tennessee Department of Health, Office of Health Statistics, Nashville, Tennessee. The authors gratefully acknowledge support for this work from the Vanderbilt University Kennedy Center's National Institute for Child Health and Human Development Grants P30HD 15052 and R03HD 050468. The authors also thank Jennifer Blackford, Elisabeth Dykens, Stephanie So, and Theresa Urbano for their insightful and detailed editorial comments on drafts of this paper. The contents of this study are solely the responsibility of the authors and do not necessarily represent the official views of any of the funding agencies or the Tennessee Department of Health. Requests for reprints should be sent to Richard C. Urbano, Vanderbilt Kennedy Center, 404A MRL, 230 Appleton Place, Peabody Box 40, Vanderbilt University, Nashville, TN 37203. email@example.com