Higher income was an independent predictor for correctly answering all of the four maternal nutrition questions. Higher income, having a greater number of children and not being a WIC participant were associated with a greater number of correct responses to all four nutritional knowledge questions in a multivariate linear regression analysis. We also found higher income and more education were the only independent predictors in multivariate analysis for increased odds of mothers reading nutritional labels. In spite of the increased educational resources that the vast majority of our low-income participants had access to through the prenatal WIC Program, having a lower income was still independently associated with both a lower maternal nutritional knowledge and a reduced likelihood of reading nutritional labels.
Other studies that have evaluated the relationship between being a WIC participant and nutritional knowledge and nutritional attitudes have controlled income level by using WIC eligible participants as the control group (Ponza et al. 2004
), and as such, have not compared the nutritional knowledge of WIC participants with non-participants from other income brackets. In this study, most of the WIC-eligible participants were already participating in WIC. Only 18 out of 125 (13%) of those 1.85 times below the poverty line, or meeting the income criteria for WIC, were not participating in WIC. Additionally, we had a sizeable percentage of the study population that had an annual income greater than $50 000 per year (43.0%). Approximately 28% of the sample also owned their own homes in the San Francisco Bay area, suggesting that a good percentage of the sample falls within a much higher income bracket, providing a strong contrast to the WIC participants. In spite of additional education on nutrition for WIC participants, lower income WIC participants still performed poorly on our questions on nutritional knowledge (2.7% responded to all knowledge questions correctly) and reported that they never read nutritional labels (29.5%). However, non-WIC participants also did not perform exceptionally well on the questions on nutritional knowledge (only 16.4% answered all the knowledge questions correctly) and 11.4% reported that they never read nutritional labels. There was a wide range in correct responses to the nutrition knowledge questions with 68.4% of WIC participants correctly identifying the appropriate age to introduce cow’s milk but only 13.0% knowing the appropriate age to introduce low-fat products, suggesting that certain areas of infant and child nutrition need increased focus.
Our results contrast with the report by Hendricks et al. (2006)
, which found that the most important factor in determining compliance with the AAP feeding guidelines was higher education using a large sample population from around the United States. In our study, maternal education was significantly associated with nutrition knowledge (answering all questions correctly) in univariate analysis but was not significant after controlling for household income and other variables. In most populations, maternal education and income levels are correlated, and the relationship between the two and child health is complex and not always easily discernable. In our study, maternal education and income were correlated at an r
= 0.72 suggesting that education levels serve as a somewhat good proxy for income levels although, ultimately, certain factors associated with income levels were more predictive of nutritional knowledge. Educational levels, however, were predictive of reading nutrition labels, which corresponds with the results for other nutrition label studies.
Other studies, which have evaluated the frequency of nutrition label reading, have found a similar percentage of participants usually reading nutritional labels (35.2%) from a survey of 1450 adult residents of Washington State (Neuhouser et al. 1999
), and 25.3% usually reading nutrition labels in a population of African-Americans in North Carolina (Satia et al. 2005
). Both of these studies found a higher percentage of nutritional label reading among women, which could account for the high levels of participants from our study who reported always reading nutrition labels (32.9%). Similarly, 21.8% of the population of African-Americans in the study by Satia et al. (2005)
stated that they never read labels, which is comparable to the 20.7% reported by our population. These studies found that increased years of education were associated with usual or often reading nutrition labels with college graduates, those with some college and those having an advanced degree reading nutrition labels more frequently. However, neither of these two studies evaluated the relationship between reading nutrition labels and income levels.
Studies of nutrition label reading have found that individuals who read more nutritional labels are more likely to eat a low-fat diet (Neuhouser et al. 1999
) and a diet high in fruits and vegetables (Kreuter et al. 1997
). The fact that lower income and less well-educated participants are less likely to read labels indicate that these participants may also be eating a diet that is higher in fats and with fewer fruits and vegetables. A study by Lin et al. (2004)
found that use of nutritional labels was related to a combination of knowledge and attitudes of nutrition and label use including: capability of label use, knowledge of fats, belief in the efficacy of diets to reduce illness and an awareness of the relationship between consumption of certain nutrients and health problems. In trying to intervene to get WIC participants to read more nutritional labels, it may not be sufficient to simply educate WIC participants about the benefits of label reading, but rather, it may be essential to also address nutritional attitudes.
Nutrition studies have demonstrated that while nutritional knowledge is necessary for positive health outcomes, it is also important for interventions to focus on attitudes that impact food intake, behavioural intention or label reading. Nutritional attitudes may be most predictive of consumption of specific foods including fat intake (Shepherd & Towler 1992
). A study with college students that evaluated the relationship between knowledge, attitudes and nutrition label use found that both nutrition education and a positive attitude were most strongly associated with label reading. In this study, attitudes mediated the pathway between knowledge and label reading (Misra 2007
). In the study by Satia et al. (2005)
of African-Americans in North Carolina, the strongest predictors of nutrition label use were all associated with attitudes including the belief of a link between diet and cancer, confidence in being able to eat more fruits and vegetables and confidence in being able to eat less fat and lose weight. In a study evaluating reasons for differences in consumption of fruits and vegetables between boys and girls, preference was more predictive than nutritional knowledge or intention (Bere et al. 2008
It is not clear what are the appropriate measures and interventions that are necessary to close the health gap between lower and higher SES children. In our study, we found that lower income mothers and mothers with less education performed more poorly in our evaluation of nutritional knowledge and used labels less frequently. It is possible that in spite of the additional nutritional interventions that WIC participants receive, it is not sufficient to place them par with mothers with more access to financial and educational resources. In addition to potentially increasing nutritional education interventions, possibly focused on areas that women performed particularly poorly on such as appropriate first complementary foods and age to introduce low-fat products, future WIC programming efforts also need to address nutritional attitudes.
Additionally, it is possible that psychological factors that impact attitudes and behaviours including parental depression are important variables that can mediate the effect between income, poverty and child health, suggesting that WIC programmes and other intervention programmes may need to address other measures outside of material hardship and nutritional education (Pickett & Wilkinson 2007
). These studies suggest that comprehensive approaches may be most effective to ensure better childhood health outcomes. It is likely that SES and educational background interact with behaviour and attitudes so it is important for intervention programmes to address attitudes, including factors such as maternal mental health, that may impact behavioural intention for any intervention that addresses the advantages that higher SES and more education confer.
Some factors that impact maternal nutritional attitudes and behavioural intentions may be associated with structural factors that will be difficult to change. For example, the complexities of the relationship between income and child health is illustrated by Pickett & Wilkinson (2007)
who found that all indicators of child health were worse in US states with higher levels of income inequality. Most WIC evaluation studies compare the nutritional knowledge or attitudes of WIC participants with WIC-eligible participants, thereby controlling for structural factors such as societal income inequality levels. However, in spite of the difficulties of altering structural factors, we argue that it is important to evaluate the nutritional knowledge and behaviours of WIC participants with those of higher income and better-educated individuals who routinely have better health outcomes. The nutritional knowledge and attitudes of higher income and better-educated individuals should be the endpoint goal for intervention studies that hope to reduce nutritional deficiencies.
Limitations of the study
Our study is limited by the relatively high percentage of missing responses for the income variable – 15.2% – which corresponds with the non-response for income and economic data found in other surveys including the 15% reported by Hendricks et al. (2006)
. We also did not collect income as a continuous variable but only had categorical data on income. However, our data do provide useful information on the nutritional information and practices of mothers and should help guide further studies and policies relating to nutritional guidance and labelling.
Our study was also limited by questioning mothers within 1–4 days after giving birth, when many women were still recuperating from labour and delivery process and may have been still been feeling the effects of drugs. However, as some studies suggest that participants may introduce complementary foods as early as 7–10 days post-partum, we wanted to evaluate maternal knowledge in this area, and the possible impact of WIC prenatal education on knowledge in this and other areas of early infant feeding. We also used a convenience method of sampling so our population group was not representative of WIC participants in San Francisco or San Francisco pregnant women, in general. However, in spite of using this type of methodology to recruit participants, our population was surprisingly representative of San Francisco populations as demonstrated by Census 2000
. As 27.8% of our population was homeowners, 43.1% was a college graduate or higher with 43.0% reporting an income >$50 000 per year, this actually compared favourably with US Census 2000
results for San Francisco county, which has a home ownership rate of 35%, 45% having a bachelor’s degree or higher and a median household income of $51 815. However, our sample had a higher percentage of Latinas (35.1% vs.14.1% in the Census) and a lower percentage of Asians (20.7% vs. 37.1%) (US Census Bureau 2000
Another limitation of our study was that we did not evaluate women’s ability to read nutritional labels through the use of validated health literacy measures and validated measures of mathematical skills. Future interventions will only be successful if women have basic health literacy and math skills that they can build on to develop the nutritional knowledge necessary to successfully read nutrition labels (Rothman et al. 2006