Maternal risk factors for fetal alcohol spectrum disorders: Distal variables

Abstract

Background:

The objective of this study was to conduct a multivariate analysis of distal influences on maternal risk for fetal alcohol spectrum disorders (FASD).

Methods:

Interviews were conducted with mothers of first grade students whose children were evaluated to assess risk for FASD. Topics included: physical/medical status, childbearing history, demographics, mental health, domestic violence, and trauma.

Results:

Although there is some individual variation in distal maternal risk factors among and within the mothers of children with each of the common diagnoses of FASD, patterns emerged that differentiated risk among mothers of children with FASD from mothers whose children were developing typically. Case control comparisons indicate that mothers of children with FASD were significantly: smaller physically, had higher gravidity and parity, and experienced more miscarriages and stillbirths; were less likely to be married, reported later pregnancy recognition, more depression, and lower formal educational achievement. They were also less engaged with a formal religion, were less happy, suffered more childhood trauma and interpersonal violence, were more likely to drink alone or with her partner, and drank to deal with anxiety, tension, and to be part of a group. Regression modeling utilized usual level of alcohol consumption by trimester and six selected distal variables (maternal head circumference, body mass index, age at pregnancy, gravidity, marital status, and years of formal education) to conclude that these variables explain 57.7% of the variance in fetal alcohol syndrome (FAS) diagnoses, 30.1% of partial FAS (PFAS) diagnoses, and 46.4% of alcohol-related neurodevelopmental disorder (ARND) diagnoses in children with FASD when compared to controls. While the proximal variables explained most of the diagnostic variance, six distal variables explained 16.7% (⅙) of the variance in FAS diagnoses, 13.9% (⅐) of PFAS, and 12.1% (⅛) of ARND.

Conclusions:

Differences of distal FASD risks were identified. Complex models to quantify risk for FASD hold promise for guiding prevention/intervention.

Introduction

Fetal alcohol spectrum disorders (FASD) are a prevalent problem in many populations throughout the world (Lange et al., 2017; Popova et al., 2018). Not only are there four distinct diagnostic phenotypes that vary greatly in their presentation, there is also substantial variation in presentation and severity of traits among individuals within each of the four phenotypes. Much of the phenotypic variation can be associated with, or attributed to, the proximal influences of the major teratogen, ethanol, in alcoholic beverages. The dose and response to alcohol (Stratton et al., 1996), or more specifically, the quantity, frequency, and gestational timing (QFT) of maternal drinking cause a major proportion of the individual variation in the physical, cognitive, and behavioral traits of FASD (May et al., 2013b; Sulik, 2014). Other recreational drugs such as cannabis are also teratogenic, and it is common for people who drink heavily to also consume other drugs before, during, and after pregnancy (Kovács et al., 2023; Lamy et al., 2015). Such co-morbid substance use among pregnant women can cause further developmental delay, more severe physical anomalies, and increase individual variation in the offspring (Fish et al., 2019; Kovács et al., 2023).

Beyond the proximal risk factors of prenatal use of alcohol and other teratogenic drugs, many other variables contribute to the variation and severity of child outcomes. Certainly, genetic and epigenetic traits of both the father and mother (Chung et al., 2021; Thomas et al., 2023) play a role in outcome and trait severity; but multiple other, more distal factors, also contribute to phenotypic variation (McQuire et al., 2020; Montag, 2016; Roozen et al., 2018). A pregnant woman’s nutritional status has been proven influential, for multiple nutritional deficiencies have been associated with more dysmorphia, cognitive and behavioral delays, and growth restriction (Carter et al., 2017; Huebner et al., 2015; May et al., 2014; Wozniak et al., 2015; Zeisel, 2009). Other distal variables and their associations with FASD have been summarized in reviews and meta-analyses. Common distal risk factors for FASD cited in the literature are: socioeconomic status, marital status, religion, gravidity, parity, access to and timing of prenatal care, maternal mental health, intimate partner violence, and age at pregnancy (Esper and Furtado, 2014; McQuire et al., 2020; Montag, 2016; Saxov et al., 2023).

The Major Aim of this Study

In this study, we aimed to define, with data from studies in five communities of essentially one multi-racial and ethnic population, which distal factors might successfully differentiate risk for FASD outcomes. By comparing detailed, case-control descriptions of characteristic traits of mothers who gave birth to children with an FASD diagnosis to mothers of children who were developing in a typical manner, we sought a broad, comprehensive, descriptive understanding of the distal traits most highly associated with mothers of children with the three common diagnoses of FASD. This task was facilitated by several factors that were extant in the study population in the Western Cape Province of South Africa. First, this population has the highest recorded prevalence of FASD ever found in active case ascertainment studies over the past twenty years (16-31%) (May et al., 2022, 2021). Second, this population has been shown to be an accurate alcohol-reporting population by studies comparing data from self-report with data from two biomarkers, ethyl glucuronide (EtG) and phosphatidylethanol (PEth) (Hasken et al., 2023; May et al., 2018). Finally, an extensive examination of an array of multiple proximal and distal risk factors has been implemented in maternal interviews in cross-sectional, epidemiological studies over the past 20 years. These studies have resulted in a rich, large dataset of with an extensive range of variables from multiple domains of potential risk for FASD. We have explored these variables for heuristic purposes to: 1.) share them with other researchers in their most accessible form, and 2.) to lay the groundwork for moving forward to a more quantified, multivariate understanding of risk for FASD.

Materials and Methods

Sampling and Diagnostic Methods

The dataset is comprised of the results of seven population-based, active case ascertainment studies of the prevalence and characteristics of FASD among children in the Western Cape Province of South Africa (May et al., 2022, 2021, 2017, 2016a, 2016b, 2013a, 2007). Seven independent cohorts of 1^st grade children were studied from 2002-2018. Consent was sought from parents and guardians of all enrolled first grade students in public schools in these communities. The consent rate was 78.4%, and the number of consented children was 6,485. There were two modes of entry into the studies. First, enrolled pupils (approximately 35% in each cohort) were selected randomly from entire class rolls, and if they had consent to participate in the study, they were then scheduled to receive full screening for FASD and other known birth defects/anomalies in Tiers II and III of the assessment process. Randomly-selected children who met criteria for one of the four diagnoses within the FASD continuum were ultimately classified as cases and were not part of the control group. If ultimately found to be free of another known anomaly and to be developing in a typical manner (within normal parameters for this age and population), the remaining randomly-selected children formed the control group. The second mode of study entry ensured capture of as many children who might have the hallmark, classical diagnosis of fetal alcohol syndrome (FAS). In this second mode, all consented children were measured for height, weight, and head circumference (occipital frontal circumference - OFC) in Tier I of the study. All children ≤25^th centile on any of these three physical measures, some of whom had already been selected randomly as well, were advanced to Tier II of the study: a dysmorphology examination by pediatric dysmorphologists/medical geneticists skilled in the diagnosis of FASD. After the examinations, all randomly selected children, and all other children who were found to have some of the cardinal dysmorphic features or minor anomalies associated with FASD, were then advanced to complete Tier III of the study: neurobehavioral testing by psychometrists/psychologists and teacher report forms. Neurobehavioral testing utilized the Test of Reception of Grammar (TROG) (Bishop, 1989), Raven (Raven, 1981), WISC-IV Digit Span (Wechsler, 2003), and/or Kaufmann Assessment Battery for Children (KABC-II) (Kaufman and Kaufman, 2004). Achenbach Teacher Report Forms (TRF) (Achenbach and Rescorla, 2001) were completed by the primary school teacher for each child in Tier III. Children who had initially screened small on one or more of the three physical features and had no other indicators of a possible FASD diagnosis, did not continue further. Also in Tier III, all mothers of children participating in this tier were administered structured interviews with experienced staff and queried over an extensive array of proximal and potential distal risk factors. All medical examiners, psychologists and psychometrists interviewers were blinded from knowledge of prior risk or protective factors throughout their interaction with the children and mothers.

Case diagnosis and controls

A final diagnosis for each child was made in confidential, multidisciplinary case conferences in which data from physical exams, neurobehavioral/testing, teacher reports, and information on alcohol, other drug exposure, and other potential risk factors were presented. In a roundtable manner, the data were reviewed, discussed, and assessed for each child by the (no longer blinded) examining physicians, testing personnel, and interviewers of the research team while two-dimensional frontal and profile facial pictures of each child were projected on conference room screens. The physicians assigned the final diagnosis for each child using the Revised U.S. Institute of Medicine guidelines for FASD (Hoyme et al., 2016, 2005; Stratton et al., 1996). Possible diagnoses within the FASD continuum were: fetal alcohol syndrome (FAS), partial fetal alcohol syndrome (PFAS), alcohol-related neurodevelopmental disorder (ARND), alcohol-related birth defects (ARBD), and not FASD (Hoyme et al., 2016, 2005). Because all children in the final control group were randomly-selected, and examinations and testing verified that the control children were developing within the normal ranges for growth, morphology, and neurobehavior, the control group is likely representative of the majority of the children in this population. After the diagnostic criteria were applied to the children, the resultant groups of mothers with completed questionnaire data totaled 2,515.

A summary of the proximal risk factors: measuring alcohol, tobacco, and other drug use

Mothers of study children were interviewed on a variety of proximal and distal risk factors using timeline-follow-back techniques (Sobell et al., 2001; Sobell and Sobell, 1992) to reconstruct maternal drinking behavior, alcohol and other drug use at the time of the interview, and more importantly, before, during, and after the index pregnancies which had occurred six to eight years prior to the interview.

In less than 15% of cases, when the mother of the index child was not available for interview due to mobility, mortality, or abandonment, the current guardians and/or other knowledgeable informants (e.g., relatives of the birth mother) were interviewed about the mother’s alcohol, tobacco, and other drug use during the index pregnancy. Some limited data on the fathers of the index child also were collected from the mothers of the index children which have been summarized in another manuscript on proximal risks for FASD (May et al., epub [ahead of print]). Standard drinks were measured by the South African standard drink criterion: 12 g of absolute alcohol equals one standard unit. South African standard drink measures are: one 340 ml beer with 5% alcohol by volume; 120 ml of wine at 12.4% by volume; and 50 ml of liquor at 40% (80 proof). Current use of alcohol and other drugs were asked first to calibrate the respondent’s memory and then the retrospective questions of substance use during the index pregnancy were asked. Tobacco use was generally practiced via hand-rolled cigarettes in the earliest samples, but in later samples it became more common for some mothers to use machine-rolled, packaged cigarettes. Each cigarette type contained approximately 1 gram of tobacco. In the first section of results, this paper briefly summarizes data collected on the quantity, frequency, and gestational timing (QFT) on multiple proximal variables of alcohol, tobacco, and other drug use. These proximal data have been analyzed and presented in detail in another manuscript (May et al., epub [ahead of print]).

Measurement of distal influences of maternal and paternal risk

The major focus of this study was on describing the large array of potential distal risk factors for FASD. All maternal interviews were conducted as medical history exams: mother’s height, weight, and head circumference were measured, and questions of childbearing history, nutrition, and demographics preceded any questions regarding alcohol and other drug use. More detail on the sampling, diagnostic process, and maternal interviews used in these studies can be found in other publications from each of the individual cohort samples (May et al., 2022, 2021, 2017, 2016a, 2016b, 2013a, 2007).

Many of the individual, distal influences on maternal and environmental risk analyzed in this paper have been identified previously in one or more smaller sample studies in South Africa (May et al., 2009, 2008; May and Gossage, 2011), the United States (Chambers et al., 2019; May et al., 2020a, 2020b, 2020c), and Canada (Popova et al., 2020). Furthermore, literature reviews and meta-analyses have been published recently that have also identified many of these same factors from other studies (Esper and Furtado, 2014; McQuire et al., 2019; Montag, 2016; Popova et al., 2022; Roozen et al., 2018). In this study, the aggregated samples provide larger numbers and a large variety of distal variables that were collected directly in most cases from the (birth) mothers of children with FASD and from randomly-sampled representative controls from the same communities. This manuscript should provide uniquely significant statistical power and veracity from informants with a proven record of accurate reporting on alcohol use and other domains strategically relevant to identifying risk for FASD (Hasken et al., 2023; May et al., 2018).

The distal variables are presented by the following substantive, categorical groupings: maternal demographic and physical traits; childbearing history, marriage, and childcare patterns; index pregnancy health, health practices/behaviors, and general medical history; mental health history and status; religious practices; happiness, trauma history, and experience with domestic violence; and social settings and social influences on drinking behaviors.

Standard scales were used in the interviews including: the Childhood Trauma Questionnaire-Short Form (Bernstein et al., 2003), Davidson Trauma Scale (Davidson et al., 1997), Self-Report Symptoms Checklist (SSCL-51) (Uhlenhuth et al., 1983), Happiness Scale (Meyers and Smith, 1995), and interpersonal/domestic violence (World Health Organization, 2005). The Childhood Trauma Scale is a retrospective assessment of adverse experiences during childhood. Each of the 28 items are ranked from 1 = “never true” to 5 = “very often true” with a total score of 140 (Bernstein et al., 2003). The Davidson Trauma Scale (Davidson et al., 1997) assesses symptoms of post-traumatic stress disorder (PTSD) and queries both the frequency and severity of lived traumatic experiences, which were rated on a 5-point scale ranging from 0 = “not at all” to 4 = “everyday” for frequency and 0 = “not at all distressing” to 4 = “extremely distressing” for severity. The total score ranged from 0 to 136. The SSCl-51 (Uhlenhuth et al., 1983) assesses common psychological distress symptoms in the past month with each of the 51 items ranked from 0 = “not at all” to 2 = “a lot of trouble”. The SSCL-51 total score range is 0 to 102. The Happiness Scale assessed 10 items and had respondents rank their satisfaction from 0 = “completely unhappy” to 10 = “completely happy”. The interpersonal violence scale queried on behaviors of partners in a yes/no manner (World Health Organization, 2005).

Other definitions of many individual variables and measures utilized in this study are conventional and intended to be similar in usage as in clinical medicine and public health practice (Porta, 2014). For example: body mass index (BMI) is defined and measured as “weight in kilograms divided by the square of height in meters”; gravidity is “the number of pregnancies (complete and incomplete) experienced by a woman”; and parity is the number of full-term children or live births previously borne by a woman (Porta, 2014).

Statistical Analysis

There was some variation in the number of respondent data points available for analysis for each of the variables in Tables 1-4. This was due to some changes made in questionnaire items utilized in the seven cross-sectional samples over time, or from missing data from a lack of access to every mother. Data were analyzed with SPSS (IBM, 2021). Case control analysis was employed for a detailed description of mothers of children with an FASD diagnosis for comparison with mothers of randomly-selected children with typical development. In Tables 1-4, statistical significance for nominal and ordinal-level data was determined using chi-square, and for interval level data, one way analysis of variance (ANOVA). To minimize the likelihood of Type I error (a false positive), Bonferroni-adjusted p-values were employed to determine the level of significance. The Bonferroni-adjusted adjusted values were calculated for each specific domain cluster as .05 divided by number of measures in the domain cluster and these values are indicated in each section of Tables 1-4. For post-hoc analysis of the ANOVA findings, Dunnett’s C tests, which also control for Type I error, measured the significance of bi-variate comparisons of each group with an alpha of 0.05.

To provide a simple demonstration of the value of quantifying the relative effect of select distal variables on risk for FASD births, sequential, binary logistic regression analyses were undertaken using child diagnosis vs. ‘not-FASD’ as the outcomes of interest. The initial step was the proximal variable, alcohol consumption, which included three measures of alcohol consumption: mean total drinks per week in the first trimester, in the second trimester, and in the third trimester. The alcohol variables were log-transformed to improve normality and model fit. The subsequent steps (2-7) added one distal factor per step as a predictor variable. Each of these distal factors were demonstrated to be promising predictors of risk in the case control comparisons, but these models are not presented as final, comprehensive, best-fit models. They were calculated and presented for demonstration purposes.

Results

General Drinking Patterns of Maternal Aggregates and Some Proximal Data Prior to Pregnancy

While the primary focus is on distal risk factors, some proximal risk factor data on the pattern of alcohol use prior to pregnancy is summarized in Figures 1 and 2. In Figure 1A, the percentage of women (prevalence) who drank each day of a typical week is presented by the FASD diagnosis of their child. Most drinking occurs on the weekends, with the highest prevalence for all diagnostic groups and controls on Saturday: mothers of children with FAS (93%), PFAS (88%), ARND (89%), and controls (79%). A higher percentage of mothers of children with any FASD diagnosis drank each day of the week compared to mothers of controls. Figure 1B presents data on the percentage of total alcohol intake consumed by mothers of children in each diagnostic group by day of the week. On Friday, mothers of children with FAS consumed 30.7% of their weekly alcohol intake, 53.5%, on Saturday, 10.5% on Sunday, and 5.3% on the remaining days (Monday-Thursday) of the week. Therefore, the total weekend consumption for mothers of the children with FAS was 94.7%. Similarly, mothers of children with PFAS and ARND consumed most of their weekly intake on the three weekend days: PFAS (97.4%) and ARND (98.7%). Mothers of typically-developing controls who did drink prior to pregnancy reported a similar pattern (98.3% on weekends), and abstinence was the modal practice on other weekdays.

Figure 1.

A) Percentage (prevalence) of women who consumed alcohol in the 3 months prior to pregnancy by day of alcohol consumption – among those who reported any alcohol consumption 3 months prior to pregnancy. B) Percent of alcohol consumed per day in the 3 months prior to pregnancy by day of alcohol consumption – among those who reported any alcohol consumption 3 months prior to pregnancy

Figure 2.

Summary Proximal Drinking Measures of Mothers of Children with FASD and Typically-Developing Children (controls) among those who reported drinking prior to pregnancy and each trimester

Finally, as Figure 2 indicates, the mothers of children with any FASD diagnosis drank more drinks per drinking day (DDD), total drinks per week (TDPW), and more drinking days per week (DDPW) than did mothers of typically-developing controls prior to pregnancy and in the first trimester. This same pattern was reported in 2^nd and 3^rd trimesters and is also presented in Figure 2. Mothers of children with FAS were the least likely to reduce their alcohol intake in 3^rd trimester, and mothers of children with PFAS were most likely of the FASD groups to reduce drinking in 3^rd trimester.

Maternal Demography & Physical Traits

Table 1 indicates that 78% of the mothers of randomly-selected, typically-developing controls were “Coloured” (mixed-race), while 93% of the mothers of children with FAS were “Coloured”, as were mothers of children with PFAS (87%) and ARND (90%), all significantly different from controls. Age was significantly different across the groups. The mothers of children with FAS were older at interview and pregnancy (28.5) than mothers of the children with PFAS (25.7) and ARND (25.9), but not mothers of controls (28.5) in post-hoc comparisons.

The physical size measurements analyzed by ANOVA and post-hoc Dunnet C comparisons indicate that mothers of children with FAS, PFAS, and ARND were all significantly shorter, weighed less, had a smaller mean head circumference (OFC), and a lower body mass (measured by BMI and upper arm circumference) than mothers of controls. In bivariate, post-hoc comparisons, all groups were significantly different from one another on these variables except for mothers of PFAS vs. ARND.

Childbearing, Marriage, and Childcare

Gravidity, parity, and the number of living children were all significantly higher among the mothers of children with FAS, PFAS, and ARND than mothers of controls (Table 1). Furthermore, in post-hoc, bi-variate comparisons, the FAS maternal group was significantly more likely than the mothers of children with PFAS, ARND, and controls to be high gravidity (3.7), parity (3.3), and have more living children (3.1) than controls (2.9, 2.6, and 2.6, respectively). The mothers of children with PFAS and ARND were also more likely to exceed controls in these variables, but this difference was not significant in bi-variate comparisons. Miscarriages and stillbirths were also more likely to have occurred among the FASD groups, and infant deaths were more likely to be experienced by mothers of children with FASD than controls.

Table 1.

Maternal Size, Demographic, and Childbearing Variables as Distal Risk Factors for FASD in South Africa

	FAS				PFAS				ARND				Controls
	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)
Physical Size
Race/ethnicity1
Black	31	5.2			50	10.9			26	8.2			196	19.1
Coloured	557	93.1			401	87.4			283	89.6			798	77.7
White/Asian/Other	10	1.7			8	1.7			7	2.2			33	3.2			<.001
Age at pregnancy1	579		28.5	(6.7)	465		25.7	(6.5)	321		25.9	(6.6)	1047		28.5	(6.7)	<.001A,B,C
Age at interview1	569		36.2	(6.7)	464		33.4	(6.6)	317		33.6	(6.7)	1032		33.5	(6.7)	<.001A,B,C
Maternal height (in cm)1	538		155.1	(6.4)	424		157.0	(6.4)	300		157.4	(6.9)	993		159.0	(6.6)	<.001A,B,C,E,F
Maternal weight (in kg)1	537		57.7	(15.5)	423		65.5	(17.7)	298		64.6	(18.0)	992		73.0	(17.9)	<.001A,B,C,E,F
Maternal upper arm circumference (in cm)5	448		26.1	(4.8)	378		28.1	(5.0)	291		28.0	(5.8)	839		30.0	(4.8)	<.001A,B,C,E,F
Maternal head circumference OFC (in cm)1	538		54.6	(2.0)	425		55.2	(2.4)	299		55.1	(2.2)	989		55.7	(2.6)	<.001A,B,C,E,F
Maternal body mass index (BMI)1	537		24.0	(6.3)	422		26.6	(6.9)	298		26.1	(7.1)	990		28.8	(6.7)	<.001A,B,C,E,F
Childbearing & Childcare
Gravidity1	622		3.7	(1.6)	487		3.2	(1.6)	324		3.3	(1.5)	1069		2.9	(1.4)	<.001A,B,C,E,F
Parity1	610		3.3	(1.4)	481		3.0	(1.4)	319		2.9	(1.3)	1057		2.6	(1.3)	<.001A,B,C,E,F
Miscarriages1	475		.3	(.7)	408		.3	(.7)	275		.3	(.6)	912		.2	(.5)	.004C
Stillbirths1	452		.1	(.3)	390		.0	(.2)	265		.1	(.2)	877		.0	(.2)	<.001A,C
# of living children1	620		3.1	(1.3)	487		2.9	(1.3)	324		2.8	(1.2)	1067		2.6	(1.2)	<.001A,B,C,E,F
Had an infant death (within 1st year of life)1
Yes	70	11.3			33	6.9			18	5.6			38	3.6
No	547	88.7			447	93.1			303	94.4			1017	96.4			<.001
# of partners with whom mother has had children1	475		1.8	(1.0)	355		1.8	(1.1)	185		1.8	(1.0)	739		1.5	(0.8)	<.001C,E,F
# of times legally married1
0	401	67.2			317	66.0			226	71.3			504	48.5
1	191	32.0			152	31.7			82	25.9			506	48.7
2+	5	0.8			11	2.3			9	2.8			30	2.9			<.001
Months in relationship with COI biological father1	585		4.1	(1.8)	456		3.8	(1.7)	298		3.9	(1.5)	1137		3.6	(1.8)	<.001A,B,C
Currently sexually active1
Yes	383	75.1			325	76.1			210	68.4			637	75.4
No	127	24.9			102	23.9			97	31.6			220	24.6			.071
Any children placed in foster1 care
Yes	98	15.9			41	8.4			26	8.0			45	4.2
No	520	84.1			445	91.6			298	92.0			1024	95.8			<.001
Relatives ever taken care of your children for long periods of time1
Yes	195	32.0			142	29.8			83	25.8			232	21.8
No	414	68.0			334	70.2			239	74.2			831	78.2			<.001

Significant (p<.05) post-hoc Dunnett C comparisons between

^AFAS & PFAS

^BFAS & ARND

^CFAS & Randomly-Selected Controls

^DPFAS & ARND

^EPFAS & Randomly-Selected Controls

^FARND & Randomly-Selected Controls

Bonferroni-adjusted level of significance for physical size: p≤.007; for childbearing and child care: p≤.005

¹Includes samples: I, II, III, IV, V, VI, VII (Total possible n=2515)

²Includes samples: I, II, III, IV, V (Total possible n=1775)

³Includes samples: I, II, III, IV, V, VI (Total possible n=2074)

⁴Includes samples: VI, VII (Total possible n=746)

Regarding marital relationships, mothers of children with FASD have had children by more male partners, on average, than controls (1.8 vs. 1.5). They were also significantly less likely than controls to have been legally married and had shorter relationships with the biological fathers of the index children, especially the mothers of children with FAS. Foster placement was most common for the offspring of mothers of children with FAS (15.9%), followed by mothers of children with PFAS (8.4%), ARND (8.0%), and controls (4.2%). Finally, mothers of children with FAS (32%) were most likely to have had their relatives take care of their children for long periods of time, and mothers of controls were the least likely (21.8%, p<.001).

Index Pregnancy Health, Practices, and History

Pre-term labor and self-reported psychological conditions were more common among mothers of children with FASD than mothers of controls (Table 2). As reported in the previous table, mothers of children with FAS were significantly older when pregnant with the index child (28.5) than the other groups. Also significant was that mothers of children with FASD were the least likely to have planned the index pregnancy and were significantly later in the pregnancy recognition than controls, especially ARND group (week 11.5 of gestation) vs. controls (9.5 weeks). Mothers of children with FAS were the latest in seeking prenatal care (4.1 months) while controls were the earliest (3.6 months). Timely access to prenatal care is difficult in these towns and rural areas where transportation is limited, and local mobile clinics dates are often weeks apart.

Table 2.

Health, Pregnancy History, Medical History, and Mental Health

	FAS				PFAS				ARND				Controls
	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)
Health during pregnancy
Virus (n, %)1
Yes	0	0.0			1	0.4			2	1.0			0	0.0
No	322	100.0			268	99.6			197	99.0			692	100.0			.032
Preterm labor (n, %)1
Yes	50	15.5			23	8.6			23	11.6			31	4.5
No	272	84.5			245	91.4			175	88.4			662	95.5			<.001
Psychological condition4
Yes	4	3.0			11	8.7			3	2.2			6	1.9
No	130	97.0			115	91.3			134	97.8			316	98.1			.003
Child of Interest (COI) Pregnancy History
Age (years) when pregnant with COI1	579		28.5	(6.7)	465		25.7	(6.5)	321		25.9	(6.6)	1047		25.8	(6.7)	<.001A,B,C
Was COI planned1
Yes	158	27.7			148	32.6			95	30.8			429	41.4
No	402	70.4			302	66.5			211	68.5			596	57.5
Sort of	11	1.9			4	0.9			2	0.6			12	1.2			<.001
Using birth control prior to pregnancy with COI1
Yes	247	43.9			226	50.1			140	45.0			525	50.9
No	316	56.1			225	49.9			171	55.0			506	49.1			.027
Aware of pregnancy with COI (weeks)7	412		10.9	(7.3)	361		10.8	(6.8)	263		11.5	(6.6)	791		9.5	(6.6)	<.001C,E,F
Month prenatal care sought1	528		4.1	(1.8)	427		3.8	(1.7)	290		3.9	(1.5)	998		3.6	(1.8)	<.001A,B,C,F
Weight gained in pregnancy (kg)1	79		7.6	(8.7)	65		7.3	(7.3)	31		9.3	(7.1)	310		9.7	(9.3)	.091
Take prenatal vitamins with COI1
Yes	402	72.3			343	77.1			238	77.3			733	72.1
No	154	27.7			102	22.9			70	22.7			283	27.9			<.001
Take prescribed medications with COI1
Yes	445	80.2			378	84.8			272	88.3			810	79.3
No	110	19.8			68	15.2			36	11.7			211	20.7			<.001
Compared to current eating habits, during pregnancy1
Ate about the same	100	18.4			76	17.4			62	20.4			193	19.3
Usually ate less	170	31.3			102	23.3			79	26.0			238	23.8
Usually ate more	273	50.3			259	59.3			163	53.6			568	56.9			.027
Hungry in pregnancy because of lack of food in home1
Yes	78	13.7			44	9.7			36	11.4			75	7.2
No	491	86.3			410	90.3			279	88.6			961	92.8			<.001
Medical History
Ever had tuberculosis (TB)1
Yes	178	31.0			92	20.0			54	17.1			120	11.5
No	397	39.0			368	80.0			261	315			920	88.5			<.001
Ever been tested for HIV5
Yes	467	90.5			403	93.7			293	93.9			866	95.6
No	49	9.5			27	6.3			19	6.1			40	4.4			.002
Ever been diagnosed with HIV5
Yes	37	6.6			18	4.1			15	4.9			40	3.9
No	523	93.4			425	95.9			294	95.1			979	96.1			.098
Ever had a sexually transmitted infection5
Yes	117	23.1			108	25.7			66	21.2			146	16.3
No	390	76.9			312	74.3			245	78.8			749	83.7			<.001
Ever had sex under the influence of alcohol5
Yes	171	34.0			121	29.2			114	37.1			173	19.4
No	332	66.0			294	70.8			193	62.9			720	720			<.001
Mental Health
Ever experienced depression in lifetime6
Yes	42	12.7			38	15.0			35	16.3			97	16.0
No	288	87.3			216	85.0			180	83.7			508	84.0			.553
Duration of depression (in months) among those with depression in lifetime6	35		38.0	(89.5)	35		33.7	(66.3)	33		23.6	(47.5)	95		26.8	(51.7)	.730
Experienced post-partum depression5
Yes	7	2.1			9	3.6			9	4.2			28	4.7
No	321	97.9			243	96.4			204	95.8			574	95.3			.281
How stressful is life2
Not at all stressful	140	33.7			91	28.8			56	33.1			258	37.1
Somewhat stressful	70	16.8			56	17.7			35	20.7			136	19.5
Medium stressful	54	13.0			46	14.6			24	14.2			119	17.1
Very stressful	137	32.9			109	34.5			50	29.6			161	23.1
Extremely stressful	15	3.6			14	4.4			4	2.4			22	3.2			.016
Cause of stress: Alcohol or other drugs2
Yes	65	17.2			54	18.6			11	7.2			53	8.1			<.001
No	314	82.8			236	81.4			142	92.8			598	91.9
Cause of stress: marital problems/relationships2
Yes	107	28.2			86	29.7			41	26.8			118	18.1
No	272	71.8			204	70.3			112	73.2			534	81.9			<.001
Cause of stress: children/family2
Yes	146	38.5			117	40.3			61	39.9			178	27.3
No	233	61.5			173	59.7			92	60.1			474	72.7			<.001
Cause of stress: unemployment/financial2
Yes	61	16.1			64	22.1			30	19.6			140	21.5
No	318	83.9			226	77.9			123	80.4			512	78.5			.152
Cause of stress: neighborhood2
Yes	9	2.4			9	3.1			9	5.9			19	2.9
No	370	97.6			281	96.9			144	94.1			633	97.1			.199
Cause of stress: Anger-associated problems2
Yes	44	11.6			36	12.4			10	6.5			23	3.5
No	335	88.4			254	87.6			143	93.5			629	96.5			<.001

Significant (p<.05) post-hoc Dunnett C comparisons between

^AFAS & PFAS

^BFAS & ARND

^CFAS & Randomly-Selected Controls

^DPFAS & ARND

^EPFAS & Randomly-Selected Controls

^FARND & Randomly-Selected Controls

Bonferroni-adjusted level of significance for health: p≤.0167; for COI pregnancy history: p≤.005; for medical history: p≤.0083; for mental health: p≤.0042

¹Includes samples^: I, II, III, IV, V, VI, VII (Total possible n=2515)

²Includes samples: I, II, III, IV, V (Total possible n=1775)

³Includes samples: I, II, III, IV, V, VI (Total possible n=2074)

⁴Includes samples: VI, VII (Total possible n=746)

⁵Includes samples: II, III, IV, V, VI, VII (Total possible n=2326)

⁶Includes samples: II, III, IV, VII (Total possible n=1643)

⁷Includes samples: III, IV, V, VI, VII (Total possible n=2082)

⁸Includes samples: II, IV (Total possible n=790)

Weight gain during pregnancy was not significantly different across maternal groups, although gain tended to be lowest among mothers of children with FAS and PFAS. Prenatal vitamins were utilized most by mothers of children with PFAS and ARND as were prescribed medications during the index pregnancy. Eating habits did not change significantly during pregnancy for any of the groups; but mothers of children with FAS, ARND, and PFAS were all more likely to report hunger due to a lack of food.

General Medical History

Mothers of children with FASD were significantly more likely to report having had tuberculosis and being tested for HIV which may be indication of more frequent seeking of overall health care (Table 2). Both diseases were common in the Western Cape Province during these studies. A diagnosis of HIV was reported by more mothers of children with FAS, but this difference was not statistically significant. More mothers of children with FASD reported having sexually transmitted infections (21-26%) than controls (19%) and were more likely to have had sex while under the influence of alcohol.

Mental Health Status

Lifetime prevalence of depression was reported by 13-16% and duration was as long at 24-38 months for each of the maternal groups, but not significantly different across groups. Post-partum depression (2-5%) was relatively low and not statistically different among groups. Stress was reported by all groups (63-71%) but was not statistically different across groups. Causes of stress were reported as significantly greater among the mothers of children with FASD for: alcohol and other drugs, marital relationships, children/family, and anger problems (Table 2).

Social Variables: Socioeconomic Status, Marital Status, and Religion

The educational achievement of mothers of children with any FASD diagnosis were significantly lower (p<.001 and <.05 on all bi-variate comparisons, except for PFAS vs. ARND) than controls: the mean for mothers of children with FAS was 6.9 years, 8.3 for PFAS, and 8.4 for ARND (see Table 3). Mothers of controls were the most likely to be married (33%), and mothers of children with FASD were significantly less likely to be married: ARND:13.2%, FAS: 17.3%, PFAS: 20.1%. Single and/or living with a partner or parent were the most common residence patterns in these communities. Mothers of children with FASD were significantly more likely than controls to: work for money, be employed seasonally, earn less money individually and in their household, and live on farms. The number of people living in the mother’s household and household composition were similar across all groups.

Table 3.

Social Variables: SES, Religion, and Happiness

	FAS				PFAS				ARND				RSC
	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)	N	%	Mean	(SD)
SES
# of years of school completed1	560		6.9	(3.0)	447		8.3	(2.9)	316		8.4	(2.8)	1034		9.6	(2.7)	<.001A,B,C,E,F
Marital status during pregnancy (n, %)1
Married	104	17.3			94	20.1			42	13.2			345	33.0
Widowed/divorced/separated/single	40	6.7			33	7.1			17	5.3			84	8.0
Living with partner	305	50.8			155	33.2			114	35.7			225	21.5
Living with parent(s)	151	25.2			185	39.6			146	45.8			393	37.5			<.001
Work for money (n, %)
Yes	452	75.8			299	63.2			224	70.2			362	34.5
No	144	24.2			174	36.8			95	29.8			687	65.5			<.001
Employment status during pregnancy (n, %)1
Full time	200	34.8			195	43.2			116	36.9			481	47.2
Part time	30	5.2			27	6.0			28	8.9			76	7.5
Seasonal	168	29.3			109	24.2			82	26.1			185	18.2
Unemployed	176	30.7			120	26.6			88	28.0			276	27.1			<.001
Usual Occupation (n, %)1
Factory worker	73	12.7			89	19.3			58	18.5			204	19.8
Farm worker	273	47.4			161	34.9			128	40.9			218	21.2
Housewife	59	10.2			37	8.0			21	6.7			72	7.0
Domestic worker	60	10.4			39	8.5			28	8.9			114	11.1
Office worker / other	37	6.4			86	18.7			32	10.2			295	28.7
Unemployed for any reason	74	12.8			49	10.6			46	14.7			126	12.2			<.001
How much women earns per week (in Rands)1	554		299	(316)	443		469	(690)	306		384	(374)	994		582	(1038)	<.001A,B,C,F
How much household earns per week (in Rands)1	537		816.8	(691.0)	420		1180.5	(1514.0)	292		958.7	(926.2)	974		1446.6	(1891.3)	<.001A,C,E,F
Total monthly income (all sources – wages, grants) (in Rands)1	554		5105	(10549)	434		6110	(10549)	297		5600	(9253)	988		6807	(9566)	<.001A,C,E,F
Number of people with an income in household5	530		2.2	(1.5)	437		2.2	(1.4)	316		2.1	(1.5)	914		2.0	(1.2)	.022
Number of adults in household5	453		2.2	(2.0)	338		2.2	(1.7)	183		2.1	(1.8)	725		2.0	(1.5)	.071
Number of minors in the home5	455		2.9	(1.9)	338		3.1	(1.7)	183		2.9	(1.8)	727		2.7	(1.5)	.046E
Living location
Rural	294	50.9			179	39.3			125	39.3			275	26.4
Urban (conventional)	252	43.6			270	58.8			187	58.8			737	70.7
Urban (informal settlement)	32	5.5			12	1.9			6	1.9			30	2.9			<.001
Months lived in current location1	536		142.9	(130.0)	433		148.2	(126.4)	308		158.1	(138.4)	1013		139.3	(117.8)	.122
Religion
Practice a religion1
Yes	505	89.9			403	89.4			282	90.7			974	95.5
No	57	10.1			48	10.6			29	9.3			46	4.5			<.001
How often attend church – among those who practice a religion1
Never	36	7.2			19	4.8			16	5.8			20	2.1
Not very often	110	22.1			70	17.8			69	24.8			148	15.4
Often (1-2 per month)	189	38.0			155	39.4			109	39.2			347	36.1
Very often (weekly)	162	32.6			149	37.9			84	30.2			446	46.6			<.001
How many times visit church in past month1	562		2.6	(3.1)	433		2.7	(2.9)	302		2.3	(3.0)	1000		3.2	(3.6)	<.001C,E,F
How often pray1
Never	19	3.6			14	3.3			4	1.3			13	1.3
Not very often	32	6.0			23	5.4			22	7.3			44	4.4
Often (1-2 per month)	56	10.6			30	7.0			25	8.3			66	6.7
Very often (weekly)	423	79.8			359	84.3			250	83.1			868	87.6			.003
Trauma
Childhood trauma scale8	185		68.3	(9.5)	122		68.6	(10.2)	76		69.0	(8.0)	272		68.5	(8.7)	.959
SSCL (self-reported syndrome checklist)8	185		19.9	(19.0)	124		28.0	(24.7)	73		20.6	(21.3)	278		18.3	(21.4)	<.001A,E
Davidson trauma scale: frequency8	216		2.9	(8.6)	143		5.9	(13.7)	87		3.9	(11.0)	301		3.6	(10.4)	.072
Davidson trauma scale: severity8	216		3.2	(9.2)	143		5.9	(13.7)	87		3.9	(10.2)	300		3.7	(11.0)	.128
Happiness Scale [0-10, [unhappy to happy]
Job or educational progress1	336		7.5	(2.9)	261		7.6	(2.8)	218		7.6	(2.9)	601		7.6	(2.7)	.898
Money management1	337		7.7	(2.7)	261		7.2	(2.9)	218		7.6	(2.8)	602		7.6	(2.7)	.220
Social life1	337		8.1	(2.4)	261		7.9	(2.6)	217		8.3	(2.3)	602		8.4	(2.2)	.020E
Cultural life1	337		8.1	(2.4)	261		7.8	(2.7)	217		8.3	(2.4)	602		8.4	(2.1)	.014E
Religion or spiritual life1	337		8.7	(2.2)	261		8.5	(2.5)	217		8.9	(2.0)	602		9.0	(1.8)	.007E
Personal habits/health1	337		8.1	(2.6)	260		8.0	(2.6)	217		8.4	(2.4)	601		8.5	(2.0)	.006C,E
Marriage/family relationships1	337		8.1	(2.6)	260		7.9	(2.7)	217		8.1	(2.5)	602		8.5	(2.3)	.005E
Legal issues1	337		8.9	(2.1)	261		8.8	(2.2)	217		8.7	(2.4)	600		9.1	(1.9)	.118
Emotional life1	336		8.4	(2.2)	260		7.9	(2.5)	217		8.1	(2.6)	602		8.3	(2.3)	.022A
Communication1	337		8.5	(2.1)	261		8.4	(2.2)	217		8.4	(2.2)	601		8.7	(2.0)	.172
General Happiness1	337		8.6	(2.1)	260		8.4	(2.3)	216		8.5	(2.1)	601		8.7	(2.0)	.319
Alcohol Sobriety1	337		7.4	(3.2)	261		7.7	(3.0)	217		7.7	(3.0)	602		9.2	(2.0)	<.001C,E,F
Drug Use1	336		9.4	(1.5)	261		9.5	(1.3)	217		9.5	(1.5)	602		9.7	(1.2)	<.001C

Significant (p<.05) post-hoc Dunnett C comparisons between

^AFAS & PFAS

^BFAS & ARND

^CFAS & Randomly-Selected Controls

^DPFAS & ARND

^EPFAS & Randomly-Selected Controls

^FARND & Randomly-Selected Controls

Bonferroni-adjusted level of significance for SES: p≤.0038; for religion: p≤.0125; for trauma p≤.0125; for happiness scale: p≤.0038

¹Includes samples^: I, II, III, IV, V, VI, VII (Total possible n=2515)

²Includes samples: I, II, III, IV, V (Total possible n=1775)

³Includes samples: I, II, III, IV, V, VI (Total possible n=2074)

⁴Includes samples: VI, VII (Total possible n=746)

⁵Includes samples: II, III, IV, V, VI, VII (Total possible n=2326)

⁶Includes samples: II, III, IV, VII (Total possible n=1643)

⁷Includes samples: III, IV, V, VI, VII (Total possible n=2082)

⁸Includes samples: II, IV (Total possible n=790)

Also in Table 3, 89-96% of all women in these communities reported practicing a religion (mostly Christian), yet mothers of children with FASD were significantly less likely than controls to practice a religion, attend church often, pray frequently, or attend church in the past month (FAS - 2.6 times, PFAS - 2.7, ARND - 2.3, and controls - 3.6).

Trauma Scales

On the Childhood Trauma Scale-Short Form (Bernstein et al., 2003), women in all groups scored equally high (8.3 to 69.0 out of 140). On the symptoms and syndromes checklist (SSCL) for adult experiences, the mothers of children with PFAS scored significantly higher (28 out of 102), followed by the ARND group (20.6), the FAS group (19.9), and controls (18.3), which is a significant difference (Table 3). On the Davidson Trauma Scale, there was no statistical difference between the groups regarding reports of frequency or severity of trauma.

Happiness Scale

The concluding section in Table 3 presents data on the Happiness Scale (Meyers et al., 2011; Meyers & Smith, 1995) where a score of 0 is very unhappy and a score of 10 is very happy. The most significant difference in the maternal groups was that mothers of children with FASD, especially mothers of children with FAS, were significantly less happy than controls with their alcohol/sobriety status (7.4-7.7) and drug use (9.4-9.5) than controls (9.2 and 9.7, respectively). Approaching significance with Bonferroni-adjustments, mothers of children with FASD were less happy with their social life, cultural life, religion/spiritual life, personal habits and health, marriage and family relationships, and emotional life.

Domestic and Interpersonal Violence

In Table 4, lifetime experience with domestic violence is presented via the Intimate Partner Violence Scale (WHO, 1999). On nine of the eleven items, mothers of children with any FASD diagnosis experienced more violence than mothers of controls. Partners had insulted, prohibited their partner from leaving home, screamed, threatened hurt, slapped, threw a potentially hurtful object, pushed or shoved, hit and/or kicked, dragged or beaten up the mothers of the index child. All of these items were significant at the Bonferroni-adjusted value of p<.0045, and all violence items were experienced most by mothers of children with FAS. Approaching significance were: experienced choking, burning, and threatened or used a weapon, which were also more common among mothers of children with FASD.

Social Setting, Drinking Companions, and Reasons for Drinking

Mothers of children with FASD, especially those with FAS were significantly more likely to drink alone or with their partners than controls. Drinking at home by mothers of children with FASD approached significance (p=.019).

Reasons for drinking cited significantly more by mothers of children with FASD than controls were: boredom, because others drank, to feel less anxious, to reduce tension/calm nerves, to forget worries, and to be part of a group. Finally, significantly more of the friends of mothers of children with FAS (85%), PFAS (79%), and ARND (86.9%) drank than do friends of the controls (66%).

Regression Modeling: Selected Predictors of Maternal Risk for Specific FASD Diagnoses

Stepwise, binary logistic regression was used to model the effect of a major proximal measure (alcohol by quantity, frequency, and timing by trimester) and select distal maternal risk variables for a diagnosis of FAS. Table 5, section A, considers the effect of alcohol alone (model 1) on a diagnosis of FAS with the effect of six distal variables added in models 2-7. The block and model p-values indicate that each variable selected was statistically significant at p=0.05. The alcohol block (model 1 - measured by total number of drinks in 1^st trimester (log), total drinks in 2^nd trimester (log), and total drinks in 3^rd trimester (log)) contributed the highest estimated explanatory value as indicated by the Nagelkerke R² of .479, or 47.9% of the variance. In models 2-4, maternal head circumference (OFC) adds approximately 1% to explained variance, maternal BMI adds 1.7%, and age at pregnancy adds 2.2%. Gravidity (model 5) adds the least to estimated risk at 0.5%, marital status (model 6) adds 1.5%, and formal maternal education (model 7) adds the most of the distal variables at 2.8%. Overall, this relatively simple, seven-step model explains approximately 57.5% of the variance in predicting maternal risk for FAS.

Table 5.

Seven Step Binary Logistic Regression Models Predicting an FAS Diagnosis¹ (n=1026)

A. Summary model fit statistics for an FAS diagnosis
Model #	Step/block variables	Model Nagelkerke R square	Block Chi-square	Block p-value	Model Chi-square	Model p-value
1	Alcohol	.479	441.851	<.001	441.851	<.001
2	Maternal OFC	.488	10.268	.001	452.119	<.001
3	Maternal BMI	.505	20.00	<.001	472.119	<.001
4	Age at pregnancy	.527	499.298	<.001	499.298	<.001
5	Gravidity	.532	6.024	.014	505.332	<.001
6	Marital status	.547	18.371	<.001	523.693	<.001
7	Years of formal education	.575	35.155	<.001	558.848	<.001

B. Coefficients and Odds Ratios predicting an FAS diagnosis1
					95% C.I.for EXP(B)
	B	S.E.	Sig.	Odds Ratio	Lower	Upper
Step 1:
Total Drinks per Week – 1st trimester (log)	1.624	.245	<.001	5.074	3.142	8.195
Total Drinks per Week – 2nd trimester (log)	.773	.339	.022	2.167	1.116	4.207
Total Drinks per Week – 3rd trimester (log)	−.035	.325	.914	.966	.511	1.825
Step 2: Maternal OFC	−.063	.039	.107	.939	.870	1.014
Step 3: Maternal BMI	−.051	.015	<.001	.951	.922	.980
Step 4: Maternal age at Pregnancy	.054	.017	.002	1.056	1.021	1.092
Step 5: Gravidity	.082	.075	.273	1.086	.937	1.257
Step 6: Marital status2			.001
Divorced/single/widow	.371	.387	.337	1.449	.679	3.092
Unmarried, living with partner	.964	.255	<.001	2.623	1.591	4.323
Unmarried, living with parents	.877	.294	.003	2.404	1.350	4.281
Step 7: Years of formal education	−.195	.034	<.001	.823	.770	.879
Constant	2.154	2.144	.315	8.616

¹Reference randomly-selected, typically-developing control children

²Reference: married

Table 5B, presents the beta coefficients and odds ratios (OR) for the final model. First (p<.001, OR=5.1) and second (p=.022, OR=2.2) trimester drinking each add significantly to the risk for FAS, but not drinking in third trimester. Marital status as a risk factor for FAS was significant in two categories, unmarried living with partner (p<.001, OR=2.6) and unmarried, living with parents (p=.003, OR=2.4) when compared to the risk for a married mother. Significant risk was also identified with lower maternal BMI (p<.001, OR=0.951), higher maternal age (p=.002, OR=1.1), and lower formal education (p<.001, OR=.823).

Therefore, these maternal variables were statistically significant for predicting an FAS diagnosis. The proximal variable of alcohol QFT for first and second trimester added the most predictive value. Distal variables: low maternal education, higher maternal age, and lower BMI also contributed significantly. Overall, these six distal variables contributed substantially less (16.7% or ⅙) to the risk for a diagnosis with FAS than did the dosage of alcohol exposure to the fetus by QFT (the proximal variable).

In Tables 6 and 7, these same regression models are replicated with dependent variables of PFAS and ARND diagnoses. In Table 6, section A, alcohol dosage explained approximately 26% of the variance of maternal risk for an PFAS diagnosis while maternal head circumference (OFC), gravidity, marital status, and years of formal education were also significant variables of risk. The combined effect of these four significant distal variables contribute only 16% (or ⅐) of the overall explained maternal risk, which is 30.1%.

Table 6.

Seven Step Binary Logistic Regression Models Predicting an PFAS Diagnosis¹ (n=929)

A. Summary model fit statistics for an PFAS diagnosis
Model #	Step/block variables	Model Nagelkerke R square	Block Chi-square	Block p-value	Model Chi-square	Model p-value
1	Alcohol	.259	186.892	<.001	186.892	<.001
2	Maternal OFC	.266	5.867	.015	192.729	<.001
3	Maternal BMI	.267	0.898	.343	193.657	<.001
4	Age at pregnancy	.269	1.491	.222	195.148	<.001
5	Gravidity	.276	5.530	.019	200.678	<.001
6	Marital status	.288	9.657	.022	210.336	<.001
7	Years of formal education	.301	11.251	<.001	221.587	<.001

B. Coefficients and Odds Ratios predicting an PFAS diagnosis1
					95% C.I.for EXP(B)
	B	S.E.	Sig.	Odds Ratio	Lower	Upper
Step 1:	1.371	.220	<.001	3.941	2.561	6.064
Total Drinks per Week – 1st trimester (log)
Total Drinks per Week – 2nd trimester (log)	.506	.316	.109	1.659	.894	3.079
Total Drinks per Week – 3rd trimester (log)	−.106	.306	.729	.899	.493	1.639
Step 2: Maternal OFC	−.066	.034	.055	.936	.876	1.001
Step 3: Maternal BMI	−.008	.013	.543	.992	.966	1.018
Step 4: Maternal age at Pregnancy	.011	.016	.481	1.011	.980	1.043
Step 5: Gravidity	.131	.072	.068	1.140	.990	1.311
Step 6: Marital status1			.017
Divorced/single/widow	−.054	.363	.882	.948	.466	1.929
Unmarried, living with partner	.398	.242	.100	1.489	.927	2.391
Unmarried, living with parents	.722	.255	.005	2.059	1.249	3.394
Step 7: Years of formal education	−.107	.032	<.001	.899	.844	.957
Constant	2.054	1.870	.272	7.797

¹Reference: randomly-selected, typically-developing control children

²Reference: married

Table 7.

Seven Step Binary Logistic Regression Models Predicting an ARND Diagnosis¹ (n=814)

A. Summary model fit statistics for an ARND diagnosis
Model #	Step/block variables	Model Nagelkerke R square	Block Chi-square	Block p-value	Model Chi-square	Model p-value
1	Alcohol	.408	242.406	<.001	242.406	<.001
2	Maternal OFC	.408	0.068	.794	242.474	<.001
3	Maternal BMI	.417	6.153	.013	248.627	<.001
4	Age at pregnancy	.420	2.259	.133	250.886	<.001
5	Gravidity	.423	2.103	.147	252.990	<.001
6	Marital status	.448	17.695	<.001	270.685	<.001
7	Years of formal education	.464	11.714	<.001	282.399	<.001

B. Coefficients and Odds Ratios predicting an PFAS diagnosis1
					95% C.I.for EXP(B)
	B	S.E.	Sig.	Odds Ratio	Lower	Upper
Step 1:
Total Drinks per Week – 1st trimester (log)	2.044	.303	<.001	7.725	4.267	13.983
Total Drinks per Week – 2nd trimester (log)	1.022	.367	.005	2.778	1.354	5.698
Total Drinks per Week – 3rd trimester (log)	−.507	.357	.155	.602	.299	1.212
Step 2: Maternal OFC	.016	.051	.753	1.016	.920	1.122
Step 3: Maternal BMI	−.032	.018	.079	.968	.934	1.004
Step 4: Maternal age at Pregnancy	.033	.021	.120	1.033	.992	1.076
Step 5: Gravidity	.121	.098	.220	1.128	.931	1.368
Step 6: Marital status1			<.001
Divorced/single/widow	−.036	.553	.949	.965	.327	2.852
Unmarried, living with partner	.733	.351	.037	2.081	1.046	4.138
Unmarried, living with parents	1.387	.364	<.001	4.002	1.963	8.162
Step 7: Years of formal education	−.144	.042	<.001	.866	.797	.940
Constant	−3.648	2.769	.188	.026

¹Reference: randomly-selected, typically-developing control children

²Reference: married

In section B of Table 6, in the final model, the following variables contributed significantly to the estimated maternal risk for PFAS: one proximal variable, usual drinks per week in first trimester (p<.001, OR =3.94) and distal variables of marital status (unmarried, living with parents, p=.005, OR=2.06), and fewer years of formal education (p<.001, OR= 0.89). The odds ratio of less than one indicates an inverse relationship which means that the higher the mother’s education, the lower the likelihood for a diagnosis of PFAS. Two other variables approached significance as potential risk factors: higher gravidity (p=.068, OR=1.14) and lower maternal OFC (p=0.55, OR=0.936). The six distal variables contributed about 13.9% (or ⅐) of the total variance explained (R²=46.4%) for the diagnosis of PFAS.

Table 7, parts A and B, address risk for the diagnosis of ARND in the offspring. Modeling in section A indicates that the proximal variable, usual alcohol dosage per week during pregnancy was a significant predictor which explained approximately 41% of the likelihood of an ARND diagnosis. Only three distal variables add significantly to risk: higher maternal BMI, marital status, and fewer years of formal education. The selected distal variables contributed significantly to the total variance (46%) explained for ARND risk. However, the distal variables contributed only about 12.1% (⅛) of the total variance explained for an ARND diagnosis.

In model 7 (Table 7B), both total drinks per week in 1^st and 2^nd trimester (p<.001, OR=7.73 and p=.005, OR=2.7, respectively) contribute significantly to a diagnosis of ARND. Among the distal variables, only unmarried status, both living with a partner (p=.037, OR=2.08) or with parents (p<.001, OR=0.866), and fewer years of formal education were significantly correlated with ARND. Approaching significance (p=.079, OR=0.968) was the correlation between lower maternal BMI and ARND.

Discussion

A Summary of Risk Modeling: A Heuristic Demonstration of the Promise of Quantifying Risk

In the models of risk for each diagnosis, some variables were found to significantly elevate risk for FASD. For FAS, all alcohol use variables were significant as were: lower maternal OFC, BMI, fewer years of formal education, higher age at pregnancy, higher gravidity, and being unmarried when pregnant. Variables associated with increased maternal risk for PFAS were: alcohol use; lower OFC; fewer years of formal education; and higher gravidity. Significantly increasing maternal risk for ARND were: alcohol use; lower BMI and years of formal education; and unmarried status. Therefore, this rather simple model confirms: alcohol use during pregnancy is highly teratogenic (known since 1973) and distal variables also play a significant role in the birth of a child with FASD, especially FAS.

Table 8 presents a summary of the significant maternal risk factors delineated by the regression analysis (model 7) for each of the three most common FASD diagnoses. Drinking quantity per week during pregnancy, the major proximal risk factor, was the most influential risk factor for each diagnosis, explaining approximately 48%, 26%, and 41% of the association with FAS, PFAS, and ARND diagnoses, respectively. The six distal variables in the model contributed less to the associated risks for an FASD explaining 16.7% (⅙) of the total variance for an FAS diagnosis, 13.9% (⅐) for a PFAS diagnosis, and 12.1% (⅛) for ARND.

Table 8.

Significant Maternal Risk and Protective Factors for the Three Most Common Diagnoses of Fetal Alcohol Spectrum Disorders: FAS, PFAS, and ARND

FAS	PFAS	ARND
Estimated variance explained: Total Variance = R2 = 57.5% Contribution of six distal variables = 16.7% (⅙) of the 57.5%	Estimated variance explained: Total Variance = R2 = 30.1% Contribution of six distal variables = 13.9% (⅐) of the 30.1%	Estimated variance explained: Total Variance = R2 = 46.4% Contribution of six distal variables = 12.1% (⅛) of the 46.4%
Significant Risk: Usual total drinks per week: 1st and 2nd trimester Lower maternal body mass index Marital Status Unmarried, living with partner Unmarried, living with parents Fewer years of formal education	Significant Risk: Usual total drinks per week: 1st trimester Marital Status Unmarried, living with parents Fewer years of formal education	Significant Risk: Usual total drinks per week: 1st and 2nd trimester Marital Status Unmarried, living with partner Unmarried, living with parents Fewer years of formal education
	Potential Risk: Gravidity (approached significance, p=.068) Smaller maternal head circumference (OFC) (approached significance, p=.055)	Potential Risk: Lower maternal body mass index (approach significance, p = .079)

A high quantity of usual total drinks per week for 1^st and 2^nd trimesters were the most influential risks for both FAS and ARND, and 1^st trimester for PFAS. This presents confirmatory evidence that alcohol consumed in quantities that produce a high blood alcohol concentration (binge drinking) are highly teratogenic, particularly when consumed in the first and second trimesters. Many previous active case ascertainment studies over the past two decades also identified this heavy, binge-like drinking pattern to be most risky for FASD in both South Africa and the United States (May et al., 2022, 2013b, 2008; Chambers, et al., 2021). Furthermore, clinicians have long suspected that small maternal head circumference might be an indicator of intergenerational prenatal alcohol use, for the mother may have experienced prenatal alcohol exposure herself, and some individual mothers might have qualified for a diagnosis within the FASD spectrum at some time in her life. Furthermore, the regression analysis provided quantified evidence that lower maternal weight (BMI) was a significant risk for FAS and possibly ARND births. Marital status, that is, being unmarried when pregnant, placed women in this population at higher risk for bearing a child with an FASD diagnosis. Lower levels of formal education was also associated with a higher risk for offspring with FASD. The inverse interpretation is that higher levels of formal education were associated with lower levels of risk. Finally, gravidity qualified as a significant, associated risk factor for PFAS.

Additional Modeling is Needed

While the modeling presented in this study is clearly useful for a new and partially quantified understanding of some distal contributions to the risk for FASD, attempts at a truly holistic/comprehensive understanding of risk for FASD must be pursued further in the future. Our modeling here was meant to be heuristic, for the six variables utilized to estimate distal risk contributions for three FASD diagnoses were ones that were well established in previous case control studies, and are relatively easy to measure. Maternal head size, BMI, age at pregnancy, gravidity, marital status, and years of formal education are all quantifiable and/or easily categorized. Similarly, the proximal variables of alcohol and other drug use are quantified with basic measures of dosage, frequency, and timing of consumption. However, many other distal influences, documented in case control comparisons, also have promise to be potential and significant predictors of maternal risk.

Maternal mental health status, trauma experiences, happiness, and interpersonal violence before, during, and after pregnancy were significantly different in case control comparisons in this study. Similarly, maternal experience with domestic violence, interpersonal violence, social relationships, reasons for drinking and drinking patterns and venues were quite different between mothers of children with FASD and controls. We believe that these topics/concepts await more qualitative analysis and new theoretical and quantitative measurements with different/better measurement techniques than demonstrated here. Such research will shed more light on important social, psychological, and contextual domains of life that may play a highly significant role in risk for FASD. Several systematic literature reviews and meta-analyses have demonstrated that other researchers have explored the importance of alcohol use to cope with adverse life experiences, social pressures, and feelings of despair (Esper and Furtado, 2014; McQuire et al., 2020; Popova et al., 2022; Roozen et al., 2018). Furthermore, literature analyses rarely are able to quantify distal contributions to overall risk. Even the more grounded studies that identify these concepts of risk in particular cases, populations, or clinic settings generally concentrate on the descriptive and qualitative aspects of maternal risk and are, therefore, limited in their ability for translating their contributions to a broader, more comprehensive understanding of severity of risk to child outcomes (Ceccanti et al., 2014; Fletcher et al., 2018; Watt et al., 2017).

In the future it will be important to measure a broader assessment of the effects of negative emotions such as depression, mental health status, despair, the effect of exposure to violence, poverty, and chaos which create misery, a lack of happiness, and other “negative emotional states” as they occur in a broader environmental context. Concepts such as “hyperkatifeia” (Koob et al., 2020) may be at play in prenatal drinking risk and therefore, FASD. It may be that individual concepts of depression, or other standard psychiatric variables, experience with interpersonal violence, trauma, low socioeconomic status, and other triggers of despair and misery can be summarized by a single measurement of prenatal hopelessness and lack of happiness that can be quantified in a manner that will explain this large area of distal influence on maternal drinking, and therefore, on the likelihood of a diagnosis of FASD. Therefore, one or more composite scales that summarize the risk or protective influences for FASD births is needed.

Paternal Risk

The father’s role, both biological and social-psychological, in the birth of a child with FASD has not been studied sufficiently in either human or in animal models. In a previous paper (May et al., epub [ahead of print) we summarized our limited data on paternal drinking prior to and during pregnancy. Broadly speaking, more comprehensive research on paternal behavior and the biological contributions specific to the birth of a child with FASD are warranted (Abel, 2004; Bakhireva et al., 2011; Meng and Groth, 2018; Thomas et al., 2023).

Strengths and Limitations

Strengths of this study are found in the large number of informants covering a substantial array of potential risk factors. Second, this study was conducted in a population with very high prevalence of for FASD that has also demonstrated an ability to provide a frank and accurate disclosure of the sensitive issues affecting prenatal drinking, health, and social behavior. Third, the multidisciplinary, bilingual team collecting the data has been working together on FASD research with both children and their mothers for most of the past two decades in these populations.

Weaknesses in this study can be found in the fact that the number of data points for each domain of inquiry was variable because the questionnaire was changed and adapted throughout the fifteen-years of data collection. A second weakness is that the standard mental health, trauma, and psychological distress scales may not be as meaningful or cogent when utilized in this highly challenged population. In other words, the concepts queried were likely meaningful to a middle-class population living in more predictable, stable, and more advantaged socioeconomic conditions, but were not as meaningful in this population. Third, there is now evidence that new and broader concepts of life’s social and psychological challenges need to be developed that will more fully resonate with the informants in this population to better describe the full impact of distal influences on prenatal drinking and developmental delay in the offspring. Fourth, the regression risk models that we explored with this study were intended to provide a preliminary test of the efficacy of developing a more comprehensive, multi-factorial quantified risk model for FASD in children that include an array of distal factors. New multivariate models must be explored for a more complete understanding of risk for FASD.

Table 4.

Domestic Violence, Social Setting, Reasons for Drinking, and Day of Week when Drink, and Media Exposure

	FAS		PFAS		ARND		RSC
	N	%	N	%	N	%	N	%
Domestic violence/intimate partner violence – in life
Partner ever insulted you5
Yes	214	43.1	167	40.6	123	40.3	287	32.2
No	282	56.9	244	59.4	182	59.7	603	67.8	<.001
Prohibited you from leaving home5
Yes	142	28.7	81	19.9	67	22.0	178	20.0
No	352	71.3	326	80.1	238	78.0	712	80.0	.001
Screamed at you5
Yes	304	61.3	237	57.5	174	57.2	393	44.2
No	192	38.7	175	42.5	130	42.8	496	55.8	<.001
Threatened to hurt you5
Yes	183	37.1	135	32.8	106	34.9	222	25.0
No	310	62.9	276	67.2	198	65.1	666	75.0	<.001
Slapped you5
Yes	289	58.4	200	48.7	167	54.8	307	34.5
No	206	41.6	211	51.3	138	45.2	582	65.5	<.001
Threw something at you that could5 hurt you
Yes	136	27.5	82	20.1	66	21.6	128	14.4
No	358	72.5	326	79.9	239	78.4	760	85.6	<.001
Pushed or shoved you5
Yes	224	45.2	155	37.9	113	37.2	242	27.2
No	272	54.8	254	62.1	191	62.8	647	72.8	<.001
Hit you5
Yes	187	37.8	118	28.8	92	30.2	182	20.4
No	308	62.2	292	71.2	213	69.8	708	79.6	<.001
Kicked, dragged or beat you up5
Yes	170	34.3	109	26.5	88	28.9	148	16.6
No	325	65.7	302	73.5	217	71.1	743	83.4	<.001
Choked or burnt you5
Yes	73	14.8	59	14.4	42	13.8	89	10.0
No	421	85.2	352	85.6	262	86.2	801	90.0	.027
Threatened to use or did use a weapon against you5
Yes	94	19.0	62	15.1	47	15.5	112	12.6
No	401	81.0	349	84.9	257	84.5	779	87.4	.016
Woman usually drinks – among those who reported drinking in past year
Alone (n, %)2
Yes	63	20.9	37	19.3	21	17.8	28	9.3
No	238	79.1	155	80.7	97	82.2	274	90.7	<.001
With family (n, %)2
Yes	97	32.2	74	38.7	42	35.6	113	37.4
No	204	67.8	117	61.3	76	64.4	189	62.6	.434
With friends (n, %)2
Yes	257	85.4	158	82.7	104	88.1	241	79.8
No	44	14.6	33	17.3	14	11.9	61	20.2	.131
With partner (n, %)2
Yes	146	48.5	71	37.2	59	50.0	99	32.8
No	155	51.5	120	62.8	59	50.0	203	67.2	<.001
At Home2
Yes	237	78.7	138	72.6	86	72.9	204	67.3
No	64	21.3	52	27.4	32	27.1	99	32.7	.019
At family’s home2
Yes	70	23.3	47	24.7	25	21.2	76	25.1
No	231	76.7	143	75.3	93	78.8	227	74.9	.837
At friend’s home2
Yes	177	58.8	101	53.2	64	54.2	152	50.2
No	124	41.2	89	46.8	54	45.8	151	49.8	.200
At a shabeen2
Yes	25	8.3	17	8.9	9	7.6	17	5.6
No	276	91.7	173	91.1	109	92.4	286	94.4	.488
Why the woman chooses to drink – among those who reported drinking in past year
To celebrate (n, %)2
Yes	171	58.4	116	61.7	83	72.2	209	69.7
No	122	41.6	72	38.3	32	27.8	91	30.3	.008
To be sociable (n, %)2
Yes	179	61.1	111	59.4	78	67.8	179	59.7
No	114	38.9	76	40.6	37	32.2	121	40.3	.442
To be polite (n, %)2
Yes	31	10.6	22	11.7	10	8.7	34	11.3
No	262	89.4	166	88.3	105	91.3	266	88.7	.852
To help relax (n, %)2
Yes	176	60.1	121	64.4	76	66.1	175	58.3
No	117	39.9	67	35.6	39	33.9	125	41.7	.370
Boredom2
Yes	110	37.5	54	28.7	30	26.1	46	15.3
No	183	62.5	137	71.3	85	73.9	254	84.7	<.001
Overcome shyness2
Yes	38	13.0	21	11.2	13	11.3	17	5.7
No	255	87.0	167	88.8	102	88.7	283	94.3	.021
Because others drink2
Yes	117	39.9	62	33.0	38	33.0	60	20.1
No	176	60.1	126	67.0	77	67.0	239	79.9	<.001
To feel less anxious2
Yes	56	19.1	36	19.1	13	11.3	25	9.3
No	237	80.9	152	80.9	102	88.7	275	91.7	<.001
To feel less tense/nervous2
Yes	76	25.9	52	27.7	31	27.0	41	13.7
No	217	74.1	136	72.3	84	73.0	259	86.3	<.001
To forget worries2
Yes	119	40.6	56	29.9	31	27.0	41	15.7
No	174	59.4	131	70.1	84	73.0	253	84.3	<.001
To be part of group2
Yes	101	34.5	49	26.1	43	37.4	52	17.3
No	192	65.5	139	73.9	72	62.6	248	82.7	<.001
Social Group Drinking
How many friends drink1
None	73	15.2	78	21.3	36	13.1	300	34.4
Some	132	27.6	124	33.9	102	37.2	281	32.2
Half	41	8.6	27	7.4	16	5.8	43	4.9
Most	95	19.8	59	16.1	46	16.8	118	13.5
All	138	28.8	78	21.3	74	27.0	131	15.0	<.001

Bonferroni-adjusted level of significance for domestic violence: p≤.0045; with whom women drinks: p≤.0063; why woman chooses to drink: p≤.0045; social drinking groups: p≤.05

¹Includes samples: I, II, III, IV, V, VI, VII (Total possible n=2515)

²Includes samples: I, II, III, IV, V (Total possible n=1775)

³Includes samples: I, II, III, IV, V, VI (Total possible n=2074)

⁴Includes samples: VI, VII (Total possible n=746)

⁵Includes samples: II, III, IV, V, VI, VII (Total possible n=2326)

⁶Includes samples: II, III, IV, VII (Total possible n=1643)

⁷Includes samples: III, IV, V, VI, VII (Total possible n=2082)

⁸Includes samples: II, IV (Total possible n=790)

Abbreviations:

ARND

alcohol-related neurodevelopmental disorder

BMI

body mass index

DDD

drinks per drinking day

DDPW

drinking days per week

FASD

fetal alcohol spectrum disorders

FAS

fetal alcohol syndrome

PFAS

partial fetal alcohol syndrome

TDPW

total drinks per week