Abstract
Drinking alcohol during pregnancy is a risk factor in a range of adverse birth outcomes, including fetal alcohol spectrum disorders, and is a major health concern. For this behaviour to change one of the necessary conditions is for women to have an accurate perception of the risks drinking during pregnancy poses. A major obstacle to this is the presence of unrealistic optimism which leads to women believing they are less at risk than others. This study examined a sample of women (N = 129) from a community in the Northern Cape Province in South Africa with a high prevalence of fetal alcohol spectrum disorder for signs of unrealistic optimism. A questionnaire about the perception of personal and general risk was administered during a one-on-one interview and responses compared. Neither a Student’s t-test (t(115) = −1.720, p = .088, 95% confidence interval [−0.180, 0.013]) nor a Wilcoxon matched-pairs signed-rank test (z = −1.72, p = .285) showed a significant difference. The perception of risk posed by drinking during pregnancy to others, knowledge of fetal alcohol spectrum disorder, and the perception of how easy it would be for the participant to quit drinking were significant predictors of the perception of personal risk. Only the perception of personal risk predicted the perception of general risk. There was no evidence that participants believed themselves to be less at risk than their peers when it came to the risks of drinking during pregnancy. Future directions for research into unrealistic optimism and drinking during pregnancy are discussed.
Alcohol consumption during pregnancy is a major health concern and has been identified as a possible contributing factor to a number of adverse birth outcomes including sudden infant death syndrome (SIDS), stillbirths, preterm labour, and spontaneous abortions (Grinfeld, 2009; Iyasu, Hsia, Mandell, & Willinger, 2002; Odendaal, Steyn, Elliott, & Burd, 2009). The teratogenic effects of alcohol in utero are the causal factor in fetal alcohol spectrum disorders (FASDs), an umbrella term for a range of disabilities associated with alcohol-exposed pregnancies (AEPs; Hoyme et al., 2016; Institute of Medicine, 1996). Individuals with FASD frequently suffer from growth deficits, birth defects, and cognitive deficits (Clarke & Gibbard, 2003; Hoyme et al., 2016). There is as yet no definitive cognitive behavioural profile associated with FASD (Davis, Gagnier, Moore, & Todorow, 2013); there is, however, strong evidence that the domains of attention, executive function, and memory are key areas of deficit (Coles, 2001; Crocker, Vaurio, Riley, & Mattson, 2009; Davis, Desrocher, & Moore, 2010; Davis et al., 2013; Rasmussen, Horne, & Witol, 2006).
The above primary disabilities associated with AEP have serious and lifelong consequences. Individuals with FASD are also at risk of secondary disabilities which include mental health problems, inappropriate sexual behaviour, disrupted school experience, trouble with law enforcement, and alcohol or drug problems (Clark, Lutke, Minnes, & Ouellette-Kuntz, 2004; Streissguth, Barr, Kogan, & Bookstein, 1996; Streissguth et al., 2004).
South Africa has the highest reported prevalence of FASD worldwide (May et al., 2013; Roozen et al., 2016). In South Africa, it is estimated that 26.3% of women drink alcohol, and of these 13.9% reported heavy episodic drinking (World Health Organization [WHO], 2014), which is a key risk factor in FASD (May et al., 2004). FASD in South Africa is not confined to the wine-growing regions or to particular population groups (Urban et al., 2008, 2016, 2015).
Perception of how risky a behaviour is plays a role in shaping behaviour. The extent of this influence is unclear and disputed (Brewer et al., 2007; Brewer, Weinstein, Cuite, & Herrington, 2004; Van der Pligt, 1994; Weinstein et al., 2007); however, there is evidence that risk perception has a significant impact on health behaviours (Brewer et al., 2007; Weinstein et al., 2007). Risk perception is amenable to change and therefore a possible vector for affecting behaviour change (Ayers & Myers, 2012).
The perception of the risk of drinking during pregnancy is inversely related to alcohol consumption during pregnancy (Testa & Reifman, 1996). Universal prevention interventions aim to raise awareness of the risks surrounding drinking during pregnancy assuming that increased knowledge of the risks will lead to behaviour change (Barry et al., 2009; Hankin, 2002). The efficacy of these interventions remains unclear (Anderson, Chisholm, & Fuhr, 2009; Ayers & Myers, 2012; Barry et al., 2009; Hammer & Inglin, 2014; Ősterberg, 2004).
The impact of risk information is affected by how accurately the risk to oneself is perceived; however, the perception of personal risk is mediated by a number of biases and heuristics (Hilbert, 2012; Van der Pligt, 1994, 1996). Unrealistic optimism is one of these cognitive biases and has proven to be a robust phenomenon in health psychology where individuals judge themselves less at risk of experiencing negative life events than the average person. Perception of personal risk is strongly associated with behaviour change (Brewer & Hallman, 2006; Sjöberg, 2003).
An important modifier of the perception of personal risk is the perceived control a person has over the risk behaviour. The more a person feels that they can control the risk behaviour, the less at risk of negative consequences they see themselves (Sjöberg, 2000). In this study that would suggest that women who believe it would be easy to stop drinking during pregnancy, would be more prone to unrealistic optimism.
To the best of our knowledge, there have been no studies examining whether there is evidence of unrealistic optimism with regard to drinking during pregnancy in South Africa. To better inform future interventions, this study set out to examine whether there was evidence of unrealistic optimism with regard to drinking during pregnancy in a community at high risk of AEPs.
Method
Participants
The sample consisted of women who had been participants in an antenatal programme called the Healthy Mother Healthy Baby© programme. The programme was implemented by the Foundation for Alcohol Related Research (FARR) in De Aar in the Northern Cape Province of South Africa between 2009 and 2012. De Aar is the main centre of the Emthanjeni municipality in the Northern Cape. The Northern Cape province is one of the areas with the highest reported levels of drinking among women (Parry, 2005; Peltzer & Ramlagan, 2009), and epidemiological studies in the Northern Cape Province have reported the highest FASD prevalence worldwide, with a FAS prevalence of 67.2/1000 and a partial fetal alcohol syndrome (PFAS) prevalence of 20.8/1000 (Urban et al., 2008). The overall unemployment rate in the area is 28%, with the unemployment rate for persons between 15 and 34 at 37.2%. Educational attainment is generally low overall with only 24% of inhabitants completing secondary schooling. Up to 11% of the community have received no formal schooling (Statistics South Africa, 2012).
Inclusion criteria were age of 18 years or older, completion of the HMHB© programme and, as part of that programme, a completed Alcohol Use Disorders Identification Test (AUDIT; Babor, Higgins-biddle, Saunders, & Monteiro, 2001). Exclusion criteria were not being able to speak English or Afrikaans. The HMHB© programme is only offered in Afrikaans or English so this was not a significant concern.
Random selection of participants from FARR’s records was not feasible as the community is highly mobile and finding previous participants would be challenging given the time frame of the study. FARR’s community workers were however recruited from the community and still had contact with some of the previous participants. Convenience sampling was therefore used with the community workers approaching eligible participants as they came into contact with them in the community. All women approached consented to participate (N = 129), but one withdrew during the interview for unspecified reasons.
Instruments
The questions on risk perception were extracted from a questionnaire used in a study investigating the perception of the risk of drinking during pregnancy (Testa & Reifman, 1996). The questions were selected as they dealt with a similar population (women who drank before pregnancy recognition) and asked a similar research question (how do these women perceive the risk of drinking during pregnancy). It was not a standardised measure and was used as a guide for the construction of the questionnaire used in this study. Changes were made based on newer risk perception research which suggested ways to improve the validity of questions asked (Brewer et al., 2007, 2004). Other questions were adapted from the maternal interview used by FARR in FASD prevalence studies (May et al., 2000; Urban et al., 2008). These included demographic variables, history of pregnancy and adverse birth outcomes, and smoking history. A final set of 10 true or false questions assessed participants’ knowledge of FASD. The questions were based on freely available resources on FASD (Best Start Resource Centre, 2005; National Institutes of Health, 2006). Questions on drinking behaviour were not included in the questionnaire, as an AUDIT was administered when the participants initially entered the HMHB© programme, and this was used as a proxy for current drinking.
The questions on risk perception had to be adapted to avoid pitfalls frequently found in risk perception questionnaires, specifically lack of specified risk target (risk to oneself or risk to others) and not basing the possible outcome on the performance or non-performance of health behaviours (Brewer et al., 2007, 2004; Sjöberg, 2000). Risk perception questions were asked twice, once with the respondent as the risk target and once with an average woman in their community as risk target. The two sections of risk perception questions were separated by the questions regarding pregnancy and smoking habits to prevent responses being duplicated.
For the risk perception questions, participants had to rate their agreement with a statement on the risk of drinking during pregnancy for themselves or the average woman in their community on a 5-point Likert scale ranging from definitely not to definitely yes. Cronbach’s alpha was calculated separately for the questions about personal risk (α = .871) and general risk (α = .841). The perception of personal risk and general risk were strongly correlated (r(115) = .762, p < .001).
Procedure
Participants were interviewed at the FARR centre in De Aar between March and June 2013. The interviews were conducted by FARR community workers during their regular working hours.
Ethical considerations
Ethical approval was obtained from Stellenbosch University before the commencement of the study (#S12/11/274). Informed consent was obtained from all participants. As part of their contracts, all FARR community workers sign confidentiality agreements, contravention of which can lead to immediate dismissal. They also undergo community worker training with FARR where the importance of confidentiality and what that entails is thoroughly discussed. The community workers were experienced in conducting interviews and had already built rapport with the participants during the HMHB© programme.
All participant identifying data were removed and replaced with a study number prior to data analysis. Original copies were kept in a locked office with only the primary investigator having access.
Data analysis
Data were entered into SAS Studio version 3.5. As there is some dispute with regard to whether a Student’s t-test or a Wilcoxon’s matched-pairs signed-rank test is more appropriate for a Likert scale (De Winter & Dodou, 2010), both analyses were conducted on the data. As there were data on previous FASD births in this sample, it was examined whether having a previous FASD birth impacted on the perception of control over drinking.
Simple linear regression analyses were performed on variables that are thought to have an impact on risk perception and on variables included in the risk profile of mothers who are at risk of having children with FASD. Two analyses were done on each variable, one with the perception of personal risk as dependent variable and one with the perception of general risk as dependent variable. Significant predictors were then entered into a stepwise multiple regression analysis. Initially, the perception of general risk was excluded from the multiple regression analysis with the perception of personal risk as dependent variable, and the perception of personal risk was excluded from the analysis with the perception of general risk as dependent variable. This was done as the two variables were expected to be strongly correlated, which could have obscured lesser, yet significant predictors (Field, 2009).
Results
The mean age of participants was 29 and the majority had received some years of schooling. The majority of participants were multigravidous and multiparous. Participants had completed the HMHB© programme at least 14 months prior to their interview (Table 1). Participants obtained a mean score of 8.75 (SD = 1.72) out of 10 on the FASD knowledge questions. Based on the AUDIT data obtained during the HMHB© programme, 85 participants were classified as low-risk drinkers, 17 as risky drinkers, and 25 as high-risk drinkers. No AUDIT score was available for one of the participants.
Summary of demographic variables.
SD: standard deviation.
N = 128.
Includes women who are currently married, have been married previously, and who have been married more than once.
There may have been a response set as some participants answered definitely yes to all questions including ones that were reverse scored. To control for this, all these participants (N = 13) risk perception answers were not included in further analysis. The mean score for the perception of personal risk on the 5-point Likert scale was 4.29 (SD = 0.8), with a higher score indicating higher risk perception.
There were only nine participants who had already given birth to a child with FASD. Compared to the other participants, there was a significant negative correlation between a FASD diagnosis and the perception of how easy it is to stop drinking (r(115) = −.259, p = .005). The size of the effect is, however, small.
The Student’s t-test for related samples did not find a significant difference (t(115) = −1.720, p = .088, 95% confidence interval, CI: [−0.180, 0.013]) between the perception of personal risk and the perception of general risk. The results of the Wilcoxon matched-pairs signed-rank test were not significant (z = −1.72, p = .285).
With personal risk as dependent variable, the linear regression analysis found (a) the perception of how easy it would be to personally quit drinking, (b) knowledge of FASD, (c) the perception of general risk, and (d) the perception of how easy it is for others to quit were significant predictors. With general risk as dependent variable, the perception of how easy it would be to personally quit drinking, knowledge of FASD, and the perception of personal risk were significant predictors, as well as the perception of how easy it is for others to quit drinking.
In the multivariate analysis of significant predictors of the perception of personal risk, the perception of general risk accounted for 59.5% of the variance in the perception of personal risk. There was a strong positive correlation between the perception of personal and general risk (r(115) = .762, p < .001). FASD knowledge and perceived control over own drinking accounted for a further 8.3% of the variance. The perception of how easy it would be for others to quit drinking was excluded by the analysis. A positive correlation with perception of personal risk was found for both FASD knowledge (r(112) = .575, p < .001) and perceived control over drinking (r(115)= .437, p < .001). Table 2 gives a summary of the significant predictors of personal risk. Excluding the perception of general risk from the multivariate analysis revealed no other significant predictors.
Results from final step of stepwise regression analysis related to the perception of personal risk.
FASD: fetal alcohol spectrum disorder.
N = 109.
When the significant predictors of the perception of general risk were entered into a stepwise multiple regression analysis, with the perception of personal risk excluded, FASD knowledge and perception of how easy it would be for oneself to stop drinking accounted for 24% of the variance. The perception of how easy it is for others to stop drinking was excluded by the analysis. With the perception of personal risk included these predictors did not remain significant, with the perception of personal risk accounting for 59.5% of variance. The correlation with perception of general risk was positive for the perception of personal risk (r(115) = .762, p < .001), FASD knowledge (r(112) = .460, p < .001), and perceived control over drinking (r(115) = .326, p < .001). Table 3 gives a summary of the predictors of the perception of general risk.
Results from final step of stepwise regression analyses related to the perception of general risk.
FASD: fetal alcohol spectrum disorder.
N = 109.
Discussion
The women in this sample seemed to be aware of the risks associated with drinking during pregnancy. There was also no evidence that they believed themselves to be less at risk than their peers when it came to the risks of drinking during pregnancy. How pregnant women perceive the risks of drinking during pregnancy has not been widely investigated. In previous studies on unrealistic optimism, alcohol use was among the behaviours that showed the most evidence of unrealistic optimism (Sjöberg, 2000, 2003), and it was therefore expected that unrealistic optimism should be present when investigating drinking during pregnancy. That we found no evidence for this in this sample warrants further investigation. It would be necessary to conduct a follow-up study controlling for the limitations of this study to determine if the lack of unrealistic optimism is a robust finding. Thereafter, a study of women in the general population, not only those who took part in the HMHB©, could determine whether the participants in that programme are unique in their lack of unrealistic optimism.
If women in this population are aware of their personal risk in terms of drinking during pregnancy, it should have a significant impact on their drinking behaviour during pregnancy. The suggested lack of unrealistic optimism does not appear to be reflected in the drinking behaviour of the population as the prevalence of FASD in the area (Chersich et al., 2012). It is not clear whether this is partly due to women drinking regardless of accurate risk perception or whether this points to a lack of generalisability of this study’s findings. Due to the robustness of the phenomenon in other studies and settings, it would be valuable to investigate what factors in this sample contribute to the apparent lack of unrealistic optimism with regard to drinking during pregnancy. This could serve to improve interventions in other populations and settings.
It is possible that the assumption of normality was violated in this population. As all participants had taken part in the HMHB© programme, their knowledge of FASD and therefore their perception of the risks of drinking during pregnancy may be negatively skewed. As the community workers were well known to potential participants, it is possible that there could have been sampling bias as a community worker may have felt more comfortable approaching certain women they have worked with than others. As the interviewers were the same community workers who took them through the HMHB© programme, it is also possible that the data may be influenced by response bias, as they may want to please the interviewers. It may have led to participants overstating their perceived risk of drinking during pregnancy. The lack of unrealistic optimism may therefore not be generalisable to individuals in the broader community who had not taken part in the HMHB©. An additional limitation in terms of generalising the findings was that there were no data available to compare the women in the study sample with the women in the general population based on drinking behaviour and demographic information. In this sample, the strongest predictors of risk perception were knowledge about the risk (FASD knowledge) and perceived control over behaviour (how easy it would be to quit drinking). Increased knowledge was positively correlated with increased risk perception as predicted in the literature; there was, however, also a positive correlation between perceived control and perception of risk. Based on risk perception theory, however, a negative correlation is expected. This may be because the majority of women (66.9%) did not drink regularly if at all. These women will likely respond that they would be able to quit easily, not because of their perceived control, but rather because of the nature of their alcohol use. Previous adverse birth outcomes were not significant predictors of the perception of risk.
Having previously had a child diagnosed with FASD, however, did significantly correlate with the perception of how easy it is to stop drinking. Women who have experienced the negative consequences of drinking during pregnancy were more likely to see stopping alcohol use during pregnancy as difficult. This suggests that they are more aware of the realities of trying to chance drinking behaviour. It may be that they tried to reduce alcohol intake during a previous pregnancy, yet their child was still born with FASD. Their perception of how easy it is to stop drinking was likely changed due to experience.
This study provides support for the importance of increasing population wide awareness of the dangers of drinking during pregnancy. With the knowledge of FASD shown to be a significant predictor of perception of both personal and general risks, it suggests that, although it may not directly lead to behaviour change, awareness regarding the dangers of AEP can influence behaviour by means of increasing the perception of risk. When excluding the perception of personal and general risks from the regression analyses, FASD knowledge is the strongest predictor of risk perception for both personal and general risks. Based on the FASD questions in the questionnaire, it also appears that there is a high level of FASD knowledge in the population. FASD knowledge only explained an additional 6% of the variance in the perception of personal risk, emphasising once again that increased knowledge does not guarantee an increase in risk perception.
This study serves as a preliminary investigation of unrealistic optimism in this population. The results are limited due to the small sample size and additional studies with larger samples are still needed. The major strength of this study, however, is that the sample was drawn from a population where research has previously been conducted on FASD. Based on the previous research, it could be inferred that there is a significant amount of drinking during pregnancy in the community (Chersich et al., 2012). The presence of unrealistic optimism could reasonably be expected, and the lack of evidence therefore raises interesting questions and challenges this expectation. As convenience sampling was used, it is possible that the highest risk individuals were not included in the sample skewing the results. The interviews were conducted by the community workers who spread the prevention message in the community which may have biased responses. Audio of the interviews were, however, recorded and there was no indication that interviewers led participants to desirable answers. As there were indications that there may have been a response set with some of the participants, future investigations should include more reverse scored answers to control for this.
This study suggests fruitful avenues for future research. It would be informative to conduct a similar investigation into unrealistic optimism with regard to drinking during pregnancy among a population not yet exposed to the same level of FASD awareness. A further investigation into risk perception in the same population could inform future interventions by indicating what makes women perceive the risk to themselves as the same as to others and therefore possibly more accurately. This would ideally be addressed through qualitative methods. If possible future investigations should include more women who drink at risky levels. The large proportion of women who did not drink at all may have skewed the data.
Conclusion
This article gathered data on how women in a South African community perceive the risk of drinking during pregnancy. We questioned whether they are unrealistically optimistic about their own risk and also examined which variables predict high or low perception of risk. The predictors of risk perception confirmed the importance of spreading the knowledge of FASD, as this is of course an indispensable part of risk perception. It also affirms once again that although the knowledge of the risk of drinking during pregnancy is a necessary condition for accurate risk perception, it is not sufficient. The lack of evidence of unrealistic optimism emphasises the need for further research. Can something be learnt from this population about factors that increase accurate risk perception? Will a different sample of women from this community also show no evidence of unrealistic optimism? There are still many questions to be answered about the perceived risk of drinking during pregnancy, and to guide efforts to combat the prevalence of FASD, it is important that our understanding of this be improved.
Footnotes
Acknowledgements
MT is supported by the National Research Foundation (South Africa), and is a lead investigator with the Centre of Excellence in Human Development (Witwatersrand University). We acknowledge and thank the women who participated in the study.
Declaration of conflicting interests
Since the start of the study, the primary investigator has been employed by FARR.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.