The work reported here was funded through a Small Grant from the University of California-Davis supported by the Economic Research Service of the US Department of Agriculture, grant no. K-981834-10. The data were collected through a grant from the National Institutes of Health, grant no. HD 29549.
The Relationship Between Food Insecurity and Obesity in Rural Childbearing Women
Article first published online: 1 FEB 2008
© 2008 National Rural Health Association
The Journal of Rural Health
Volume 24, Issue 1, pages 60–66, Winter 2008
How to Cite
Olson, C. M. and Strawderman, M. S. (2008), The Relationship Between Food Insecurity and Obesity in Rural Childbearing Women. The Journal of Rural Health, 24: 60–66. doi: 10.1111/j.1748-0361.2008.00138.x
For further information, contact: Christine M. Olson, PhD, Division of Nutritional Sciences, 376 MVR Hall, Cornell University, Ithaca, NY 14853; e-mail email@example.com.
- Issue published online: 1 FEB 2008
- Article first published online: 1 FEB 2008
ABSTRACT: Context:While food insecurity and obesity have been shown to be positively associated in women, little is known about the direction of the causal relationship between these 2 constructs. Purpose: To clarify the direction of the causal relationship between food insecurity and obesity. Methods: Chi-square and logistic regression analysis of data from a cohort of 622 healthy childbearing women living in a 10-county rural area of upstate New York and followed from early pregnancy until 2 years postpartum. Findings: Obesity in early pregnancy was associated with increased risk of food insecurity at 2 years postpartum. Initial food insecurity was not associated with increased risk of obesity at 2 years postpartum. Women who were both obese and food insecure in early pregnancy were at greatest risk of major weight gain over the pregnancy and postpartum period. Conclusions: Obesity appears to lead to food insecurity rather than the converse. Obesity combined with food insecurity present the greatest risk for major weight gain in this sample of childbearing women.
Over 32% of American women age 20 years and older were obese with a body mass index (BMI) of 30 or more in 2003-2004.1 The prevalence of obesity among women has nearly doubled since 1976-1980.2 Obesity is more prevalent among lower income3 and rural women.4–6 The socioeconomic gradient in obesity among women nationally is also seen in rural areas,7 as is the recent dramatic increase in its prevalence.8
Food insecurity is a core indicator of nutritional state and is defined as “whenever the availability of nutritionally adequate and safe food or the ability to acquire acceptable foods in socially acceptable ways is limited or uncertain.”9 In 2006, 10.9% of US households were food insecure and 4.0% were food insecure with hunger.10 Food insecurity was more common outside than inside metropolitan areas (12.0% vs 10.8%).
Previous research has shown a higher prevalence of overweight and obesity among women living in food insecure households compared to women living in food secure households,11–17 although 1 recent study by Laraia and colleagues18 found no significant relationship. The results from 1 study conducted in the same geographic area as the research reported here illustrate the relationship.
Olson and colleagues11,14 showed BMI and the prevalence of obesity were significantly higher (P < .05) for women in food insecure households compared to women in food secure households (BMI = 28.2 vs 25.6). As the severity of food insecurity progressed toward hunger, both BMI and the prevalence of obesity decreased. Controlling for the women's height, income level, educational level, single parent status, and employment status in the statistical analysis, household-level food insecurity was still positively related to BMI, with a P value of .06.
All the studies of the positive association between food insecurity and obesity in women are cross-sectional and are constrained in specifying the direction of causality between the 2 constructs. Recently, Wilde and Peterman19 examined the relationship between food insecurity and change in self-reported weight over 12 months. These investigators found that women in households that were marginally food secure were more likely to gain 4.54 kg (10 pounds) or more in a year compared to women in food secure households (OR 1.68; 95% CI 1.21-2.33). Using risk of gaining 2.27 kg (5 pounds) as the outcome, they found that both the marginally food secure women (OR 1.56; 95% CI 1.09-2.23) and the women who were food insecure without hunger (OR 1.43; 95% CI 1.02-2.00) were more likely to gain weight over the year than food secure women.
Wilde and Peterman's findings are an important contribution to understanding the relationship between food insecurity and weight. However, these researchers were not able to include change in food security status over time in their analysis of weight change. The objective of our study was to clarify the direction of the relationship between food insecurity and obesity in women by utilizing data from a study with a longitudinal cohort design and measures of both variables in early pregnancy and 2 years postpartum. This study was conducted with childbearing women, and no studies exist in the literature on the relationship between these 2 variables at this time in women's lives.
Data Source and Population This study used data from the Bassett Mothers Health Project (BMHP), 1 component of a larger NIH-funded project, “Biosocial Influences on Postpartum Weight Retention” (Grant No. HD 29549). The BMHP was an observational cohort study of 622 healthy adult women followed from early pregnancy until 2 years postpartum.
Women were recruited from the population registering for prenatal care at Bassett Healthcare's network of primary care clinics serving a 10-county area of rural upstate New York. Bassett Healthcare was the sole provider of prenatal care in this area at the time of sample recruitment (late 1994 to late 1996). All women entering care were screened for eligibility and eligible women were recruited for participation. All the counties except 1 were rural based on total population (<200,000) and population of the largest place (<20,000).20 (The analysis sample included 2 women from rural areas in the non-rural county.) The participation rate and characteristics of the population-based sample have been described elsewhere.21,22 Only 10% (63) of women dropped out of the cohort by 2 years postpartum and 75% (463 of 622) had a valid weight at this time. The major reason for not having a valid weight was being less than 6 months postpartum from a second birth (54) and thus the weight was not representative of the woman's true weight. The study was approved by the University Committee on Human Subjects of Cornell University and the Institutional Review Board of Bassett Healthcare.
Variables and Measures The major variables were measured using standard, validated instruments. The specific methods and instruments used in the BMHP are described in detail in a previous publication23 and described briefly here. At the time of entry into the study in early pregnancy, food insecurity was measured using 3 items included in the women's medical record from the Institute of Medicine's (IOM) Nutrition Questionnaire published in Nutrition during Pregnancy and Lactation: An Implementation Guide.24 At 2 years postpartum, food insecurity was measured using 3 questions from the US Household Food Security Survey that were included in a mailed questionnaire.25 The US Household Food Security Survey is used by the US government for monitoring the prevalence of food insecurity as part of the Current Population Survey. The instrument for collection of the early pregnancy data was developed in 1993, before the US Household Food Security Survey was developed in 1995. The items used at each time point are shown in Figure 1 and were judged by the research team to be conceptually similar. A positive response to any one of these questions was considered indicative of food insecurity.
BMI was used to determine body weight status categories. In early pregnancy and at 2 years postpartum, women were classified as being low, normal, high, or obese, using the World Health Organization's (WHO)26 BMI categories: low or underweight (<18.5), normal (18.5-25.0), high or overweight (25.1-30.0), and obese (>30.0). The WHO BMI categories were used to facilitate comparisons of results from this study with other studies of obesity. Gestational weight gain was expressed as gaining less than the recommended amount, the recommended amount, or more than the recommended amount according to the IOM BMI categories and gestational weight gain guidelines.24 Major weight gain was defined as weighing 4.55 kg, or 10 pounds more at 2 years postpartum than in early pregnancy (the first 14 weeks of gestation). This amount of weight was selected because it is perceived by most people as a significant weight gain over a pregnancy and 2 years. Household income level and marital status were defined as follows: marital status was either single (never married and not living with the infant's father) or married (ever married including living with the infant's father and currently divorced or separated); household income was categorized as low income (≤185% of the federal poverty line) or not low income (>185% of the federal poverty line.) Additional variables included: mother's age at delivery (>30 years and ≤30 years); parity (nulliparous and parous); years of education (>high school and ≤high school); and clinic where prenatal care was received.
Data Analysis First, associations between food insecurity and obesity within and across the 2 time points of early pregnancy and 2 years postpartum were assessed using chi-square analysis. When frequencies were sparse, a Fisher's exact test was used. Second, multiple logistic regression models were developed for 4 outcomes at 2 years postpartum: food insecurity, obesity, becoming food insecure among those who were initially food secure, and major weight gain. Potential confounding variables included in the models were gestational weight gain, age, parity, household income, years of education, and marital status. Clinic site was included as a control variable. A P value of .05 was considered statistically significant, except when considering potentially confounding variables for inclusion in the regression analyses when a P value of .10 was used. Residual and influence plots were examined to assess model fit. The max-rescaled R-Square statistic is provided for each model.27 This statistic is interpreted as the proportion of variation in the outcome that is accounted for by the model factors.
Subjects with missing data were excluded. To evaluate the impact of the missing data on our models, we used chi-square analysis to compare the proportions of women across categories within the factors listed in Table 1 in the analysis sample (N = 311) to the population-based sample (N = 622) and to the sample with valid weight at 2 years (N = 463). A significant P value indicated that the proportions had shifted and possibly resulted in a biased analysis sample. Analyses were conducted using the SAS statistical package (v8.2 SAS Institute, Cary, NC).
|Population-Based Sample (N = 622) N (%)||Sample With Valid 2-Year Weight (N = 463) N (%)||Analysis Sample With Both Initial & Final Food Insecurity Status (N = 311) N (%)|
|BMI in early pregnancy (time 1)|
|Low||18 (2.9)||12 (2.6)||7 (2.3)|
|Normal||299 (48.1)||225 (48.6)||153 (49.2)|
|High||160 (25.7)||114 (24.6)||83 (26.7)|
|Obese||145 (23.3)||112 (24.2)||68 (21.9)|
|BMI at 2 years postpartum (time 2)|
|Low||15 (2.7)||14 (3.0)||9 (2.9)|
|Normal||256 (45.8)||219 (47.3)||149 (47.9)|
|High||137 (24.5)||111 (24.0)||81 (26.1)|
|Obese||151 (27.0)||119 (25.7)||72 (23.2)|
|Major weight gain (≥4.55 kg) at 2 years postpartum*|
|No||404 (72.3)||353 (76.2)||242 (77.8)|
|Yes||155 (27.7)||110 (23.8)||69 (22.2)|
|Food insecurity status in early pregnancy (time 1)|
|Food secure||376 (80.2)||283 (81.8)||259 (83.3)|
|Food insecure||93 (19.8)||63 (18.2)||52 (16.7)|
|Food insecurity status at 2 years postpartum (time 2)|
|Food secure||397 (78.0)||328 (79.6)||249 (80.1)|
|Food insecure||112 (22.0)||84 (20.4)||62 (19.9)|
|Not low income||338 (56.0)||261 (58.1)||184 (60.7)|
|Low income||266 (44.0)||188 (41.9)||119 (39.3)|
|Married (Ever)||577 (92.8)||434 (93.7)||294 (94.5)|
|Never married (single)||45 (7.2)||29 (6.3)||17 (5.5)|
|Gestational weight gain|
|<Recommended||134 (21.5)||105 (22.7)||71 (22.8)|
|Within IOM recommendations||236 (37.9)||177 (38.2)||112 (36.0)|
|>Recommended||252 (40.5)||181 (39.1)||128 (41.2)|
|Nulliparous||257 (41.4)||179 (38.7)||129 (41.5)|
|Parous||364 (58.6)||284 (61.3)||182 (58.5)|
|Clinic 1||198 (31.8)||148 (32.0)||119 (38.3)|
|Clinic 2||53 (8.5)||34 (7.3)||11 (3.5)|
|Clinic 3||118 (19.0)||97 (21.0)||53 (17.0)|
|Clinic 4||94 (15.1)||80 (17.3)||60 (19.3)|
|Other clinic||159 (25.6)||104 (22.5)||68 (21.9)|
The characteristics of the population-based sample of women who entered prenatal care and the sample with valid weights at 2 years postpartum are shown in Table 1. The sample of women who had data for food insecurity and weight in early pregnancy and at 2 years postpartum is also shown in the far right-hand column of Table 1. This is the primary sample used for the analyses in this paper.
The analysis sample differed from the population-based sample on only 2 variables: The analysis sample had a smaller proportion of women with major weight gain at 2 years postpartum; and there was a shift in the composition of the sample in terms of the clinic that provided primary care in pregnancy. The latter difference was driven by missing data on food insecurity in early pregnancy. The missing data resulted from a change in the prenatal medical record in 1995-1996 and phasing in of this change across clinics. In some clinics, the food insecurity questions were included in an obscure manner in the initial medical record and thus information was not recorded.
The results of the chi-square analyses of the association between food insecurity and obesity within the 2 time periods of early pregnancy and 2 years postpartum and across the 2 time periods are shown in Figure 2. In the within time period analyses, food insecurity was positively and significantly associated with obesity at 2 years postpartum (P= .004) but not during early pregnancy (P= .82). Food insecurity in early pregnancy was not significantly associated with obesity at 2 years postpartum (P= .48), but obesity in early pregnancy was significantly associated with food insecurity at 2 years postpartum (P= .001). For both food insecurity and obesity, status in early pregnancy was significantly associated with status at 2 years postpartum (P < .0001).
Each of the potentially confounding variables was associated with food insecurity or obesity at one or the other time points (P < .10). Thus, all confounding variables and the initial levels of the outcome variables were included in the fully adjusted regression models for both food insecurity and obesity at 2 years postpartum (Table 2). Initial food insecurity was not associated with obesity at 2 years postpartum (OR 1.97; 95% CI 0.44-8.86). Among the confounding variables, parity was a significant predictor of obesity with parous women having a decreased risk of obesity at 2 years postpartum compared to nulliparous women. In contrast, initial obesity status was a significant predictor of food insecurity at 2 years postpartum (OR 2.45; 95% CI 1.21-4.95). Among the confounding variables, low income significantly increased risk of food insecurity at 2 years postpartum.
|Predictors||Outcomes With Odds Ratio and 95% CI [OR (95% CI)]|
|Obese at 2 Years||Food Insecure at 2 Years||Became Food Insecure at 2 Years||Major Weight Gain†|
|Food insecure (early pregnancy)||1.97 (0.44-8.86)||*4.57 (2.16-9.65)||Omitted||1.19 (0.47-2.96)|
|Obese (early pregnancy)||*515.7 (118.8->999)||*2.45 (1.21-4.95)||*2.56 (1.14-5.78)||1.61 (0.77-3.36)|
|Obese + food insecure (early pregnancy)||––||––||––||*7.26 (1.28-41.15)|
|GWG‡ > recommended||1.61 (0.53-4.83)||0.57 (0.29-1.13)||*0.32 (0.13-0.80)||*1.82 (1.01-3.27)|
|Low income||0.93 (0.25-3.44)||*3.06 (1.41-6.64)||*2.70 (1.12-6.48)||0.61 (0.29-1.32)|
|Single||0.33 (0.03-3.62)||1.88 (0.58-6.11)||2.86 (0.61-13.50)||1.43 (0.42-4.79)|
|Age >30 year||1.25 (0.39-3.99)||0.86 (0.41-1.78)||0.81 (0.34-1.96)||0.61 (0.32-1.19)|
|Parous||*0.15 (0.04-0.61)||1.59 (0.75-3.35)||1.87 (0.77-4.58)||0.54 (0.26-1.01)|
|>High school education||0.81 (0.22-2.89)||0.73 (0.35-1.51)||0.78 (0.31-1.94)||0.53 (0.26-1.08)|
|No. of cases/events||72||62||35||68|
Obesity in early pregnancy was a highly significant predictor of obesity at 2 years postpartum (Table 2). While not as dramatic, women who were food insecure in early pregnancy were more than 4.5 times more likely to be food insecure at 2 years postpartum than women who were food secure (OR 4.57; 95% CI 2.16-9.65).
Given these findings on the stability of food insecurity and obesity status across time, we examined 2 more dynamic outcomes (Table 2). First, with becoming food insecure across time as the outcome, initial obesity status increased a woman's risk of becoming food insecure (OR 2.56; 95% CI 1.14-5.78), as did low income. Gaining more than the recommended amount of weight in pregnancy was associated with decreased risk of becoming food insecure. For the weight outcome, we had too few cases of becoming obese over the time period (N = 10) to create statistical models, so we used major weight gain. In this model, the interaction of food insecurity and obese BMI in early pregnancy was significant (OR 7.26; 95% CI 1.28-41.15) and neither main effect was significant. Gaining more weight in pregnancy than is recommended was significant (OR 1.82; 95% CI 1.01-3.27) and increased the risk of major weight gain by over 80%.
Food insecurity in early pregnancy was not associated with increased risk of obesity at 2 years postpartum. Obesity in early pregnancy was significantly associated with increased risk of food insecurity at 2 years postpartum. The latter result was statistically significant when controlling for initial food insecurity, obesity, and confounding variables, whereas, the former was not. This result provides support for the causal direction of the relationship between food insecurity and obesity going from obesity to increased risk of food insecurity rather than food insecurity leading to obesity. Thus, the findings from previous cross-sectional studies of an association between food insecurity and obesity11–17 may be due to obese women being at increased risk of becoming food insecure across time.
Analysis with a more dynamic outcome, becoming food insecure, showed that obesity significantly increased a woman's risk of becoming food insecure. The analyses with weight gain of 10 or more pounds at 2 years postpartum clearly showed that initial obesity status modified the relationship between food insecurity and weight gain over time. It appears that initially obese women are at particularly high risk of weight gain with food insecurity. This finding is consistent with Wilde and Peterman,19 who showed that initial food insecurity was associated with increased risk of weight gain across 1 year in a national sample of nonpregnant women. Our result extends their finding to show that the effect of food insecurity on weight gain is most pronounced in obese women. The significant statistical interaction we found suggests that food insecurity and obesity may be related to some additional common factors that promote weight gain over time.
Research has shown discrimination against overweight and obese women in the labor market.28 Discrimination against obese women in terms of decreased employment opportunities and lower wages may contribute to economic hardship and increased risk of food insecurity, providing support for the plausibility of the causal relationship going from obesity to food insecurity.
The literature also supports the plausibility of the relationship between food insecurity and major weight gain that could over the time lead to the development of obesity. The eating pattern literature supports the idea that food deprivation can result in overeating. Polivy29 found that food restriction or deprivation, whether voluntary or involuntary, results in a variety of changes including the preoccupation with food and eating. Thus, while it seems counter-intuitive that assuring access to adequate food (food security) might be helpful in preventing weight gain and improving weight loss, this may be the case.
While these findings contribute to the understanding of the direction of the causal pathway between food insecurity and obesity, we are still observing an association across 2 points in time and making an inference, based on statistics, about the likely direction of the causal pathway. Taris30 calls this making an inference about “causal priority,” not causality. Our results are influenced by the relative stability of food insecurity and obesity across time. In this sample, 19.3% of women changed food insecurity category from the beginning of pregnancy to 2 years postpartum, whereas only 5.1% changed category for obesity. Thus, there is more change in food insecurity across time to be accounted for by initial obesity than there is variation in obesity across time be accounted for by initial food insecurity.
The results reported here are not likely to be unique to rural populations of women. However, because both obesity and food insecurity are more prevalent in rural communities, our findings may be more relevant to the development of interventions to address the problems in rural communities than in more metropolitan areas. These results from pregnant and postpartum women also are not likely unique to this life stage since they are similar to those of Wilde and Peterman,19 who studied a nationally representative sample of nonpregnant women. It is possible that the results were influenced by the use of different items for measuring food insecurity at the 2 time points. However, the conceptual similarity of the items makes this unlikely.
Additional research is needed. The following questions merit investigation: What are the specific mechanisms through which obesity leads to food insecurity? Why are obese, food insecure women at greater risk for major weight gain than obese, food secure women and non-obese, food insecure women? Are there eating pattern and other dietary intake differences between these groups?
The Healthy People 2010 public health objectives for the nation aim to decrease food insecurity to a prevalence of 6% and obesity in adults to a prevalence of 15%.31 The results presented here suggest that progress toward these objectives may be facilitated by intervention programs and policies focused on ensuring that women and their families have access to sufficient nutritious food.
Among childbearing women, it appears that the causal direction of the relationship between food insecurity and obesity likely goes from obesity to food insecurity. Obesity in early pregnancy was associated with increased risk of food insecurity at 2 years postpartum, while initial food insecurity was not associated with increased risk of obesity at 2 years postpartum. However, women who were both obese and food insecure in early pregnancy were at greatest risk of major weight gain over the time period encompassing pregnancy and the postpartum period. This finding suggests that there may be additional common factors contributing to both nutrition problems.
- 4Obesity in rural women: emerging risk factors and hypotheses. In: CowardRT, DavisLA, GoldCH, Smiciklas-WrightH, ThorndykeLE, VondracekFW, eds. Rural Women's Health. Mental, Behavioral, and Physical Issues. New York , NY : Springer Publishing Company; 2006;63-81., .
- 6Urban and Rural Health Chartbook. Health, United States, 2001. Hyattsville , Md : National Center for Health Statistics; 2001., , , et al.
- 10Household Food Security in the United States, 2006 (Economic Research Report 49). Washington, DC : Economic Research Service, U.S. Department of Agriculture; 2007., , .
- 14Nutritional Consequences of Food Insecurity in a Rural New York State County. Madison , Wis : Institute for Research on Poverty, University of Wisconsin-Madison; 1997. Discussion Paper no. 1120-97., , , .
- 15Food insufficiency and the prevalence of overweight among adult women. Fam Econ Nutr Rev. 2003;15(2):55-57..
- 16Self-reported concern about food security associated with obesityWashington 1995–1999. Morb Mortal Wkly Rep. 2003;52(35):840-842..
- 17How can Californians be overweight and hungry? Calif Agric. 2004;58(1):12-17., , , et al.
- 20Socioeconomic Trends and Well-Being Indicators in New York State, 1950–2000. Albany , NY : The NYS Legislative Commission on Rural Resources; 2004., .
- 24Institute of Medicine. Nutrition during Pregnancy and Lactation. An Implementation Guide. Washington , DC : National Academy Press; 1992.
- 25Guide to Measuring Household Food Security (Rev 2000). Alexandria , Va : Food and Nutrition Service, US Department of Agriculture; 2000., , , , .
- 26WHO Consultation on Obesity. Obesity: Preventing and Managing the Global Epidemic. Geneva , Switzerland : World Health Organization; 2000. WHO Technical Report Series 894.
- 30A Primer in Longitudinal Data Analysis. Thousand Oaks , Calif : Sage Publications Inc.; 2000..
- 31Healthy People 2010. Available at: http://www.healthypeople.gov/Document/HTML/Volume2/19Nutrition.htm_TOC490383122. Accessed September 5, 2006.