Food security status and breast cancer screening among women in the United States: Evidence from Health and Retirement Study and Health Care and Nutrition Study

doi:10.21203/rs.3.rs-176608/v1

Download PDF

Research Article

Food security status and breast cancer screening among women in the United States: Evidence from Health and Retirement Study and Health Care and Nutrition Study

https://doi.org/10.21203/rs.3.rs-176608/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 25 Jan, 2023

Read the published version in Cancer Causes & Control →

You are reading this latest preprint version

Purpose: To assesses the impact of food insecurity on biennial breast cancer screenings (i.e., mammography or breast x-ray) among older women in the United States (US).

Methods: Data from the 2014 and 2016 waves of the Health and Retirement Study and the 2013 Health Care and Nutrition Study were used. The analyses were limited to a nationally representative sample of 2,861 women between 50 to 74 years of age, residing in the US. We employed a propensity score weighting method to balance observed confounders between food-secure and food-insecure women and fitted a binary logistic regression to investigate population-level estimates for the association between food security and breast cancer screening.

Results: Food insecurity was significantly associated with failure to obtain a mammogram or breast x-ray within the past two years. Food-insecure women had 54% lower odds of reporting breast cancer screening in the past two years (OR=0.46; 95% CI: 0.30-0.70, P-value <.001) as compared to food-secure women. Additional factors associated with a higher likelihood of receiving breast cancer screenings included greater educational attainment, higher household income, regular access to health care/advice, not smoking, and not being physically disabled or experiencing depressive symptoms.

Conclusion: Results demonstrate a socioeconomic gradient existing in regard to the utilization of regular breast cancer screenings among women. Those who tend to have lower education, lower-income and lack of reliable healthcare access are more likely to be food insecure. Thus, more likely to face the financial, logistical, or environmental barriers in obtaining screening services that accompany food insecurity.

Cancer Biology

Food insecurity

Breast cancer screening

Mammography

Cancer prevention

Older women

Propensity score weighting

With an average lifetime risk of over 13%, breast cancer is the most common cancer diagnosis among women in the US [1]. This trend is expected to continue, as an estimated 276,480 invasive and 48,530 non-invasive (in situ) new breast cancer diagnoses are projected to occur during 2020 [2]. Furthermore, although noteworthy advancements in early detection and treatment have led to gradual declines in breast cancer mortality [3], an estimated 42,170 women will die of breast cancer in 2020 [2]. Early detection reduces breast cancer mortality rates and increases the likelihood of better treatment options [2, 4, 5]. With evidence indicating that mammography is the most effective method for early diagnosis [2, 4], the U.S. Preventive Services Task Force (USPSTF) recommends biennial screening mammography for women aged 50 to 74 [6]. Despite this recommendation, many women are unaware of the significance of regular mammograms as well as how or when to schedule them. One of the main objectives of Healthy People 2020 in the US was to increase the proportion of women aged 50 to 74 years who receive a breast cancer screening to 81.1%. However, these rates stood at 72.8% in 2018, similar to screening rates in 2008 [7]. Social determinants of health play a crucial role in not only how many women access breast cancer screening services but also in their awareness about these preventative services. The social determinants can delay or prevent disadvantaged women from receiving timely mammograms or breast x-rays and reduce the likelihood of diagnosing early enough for intervention to be effective.

Barriers to breast cancer screening have been extensively reported in the literature [8–14]. Education, household income, and a lack of health insurance are frequently identified as some of the barriers women face in obtaining these discretionary health services. However, less emphasis has been placed on individual and family food insecurity. Food insecurity is related to economic instability, which is considered one of the key social determinants of health [15]. Generally, food security is construed as someone with access to enough quality food at all times to be able to live an active and healthy life [16]. Food insecurity, on the other hand, occurs when people experience uncertain or limited access to adequate or nutritious food due to a lack of financial or other resources [17, 18]. Food insecurity persists along racial/ethnic lines and disproportionately affects those from non-Hispanic Black and Hispanic backgrounds. For example, a recent study found that in 2015, both Black and Hispanic households experienced food insecurity at substantially higher rates, 21.5%, and 19.1%, respectively, compared to the national rate of 12.7% [19].

Beyond socioeconomic differences, minorities are more likely to live in impoverished areas that are classified as food deserts. A food desert is an area with limited access to affordable and nutritious food compared to an area with higher access to supermarkets or vegetable shops with fresh foods [20]. Residents of low-income and/or predominantly Black or Hispanic neighborhoods often have less access to supermarkets and quality foods, travel greater distances to obtain fresh foods, and have fewer healthy food options than affluent and/or predominantly white neighborhoods [20]. Food security, access to good nutrition, and high-quality diets are essential to prevent and reduce disease incidences, chronic medical conditions, disease burdens, and precursors for a better quality of life [17, 18]. In 2018, an estimated 11.1% of U.S. households (14.3 million) were food-insecure at some point; a higher than average proportion of these households were adults living alone and racial/ethnic minorities [16].

Food insecurity is associated with higher utilization rates of emergency department and hospitalizations and higher healthcare expenditures [17, 21, 22]. Food insecurity results in significant healthcare burdens, as evidenced by findings that a lack of access to adequate, nutritious food was associated with increased odds of being in the top decile, 5%, and 2% healthcare expenditures [21]. With the financial toll of food insecurity on people, expenditures like discretionary screening services may not be as urgent of a concern. Studies focusing on the associations between food insecurity and preventive healthcare services, including cancer screening services, are scarce. Previous studies have attempted to investigate the associations between food insecurity and other services––such as uptake of recommended vaccines among adult populations [23]. However, to our knowledge, no study to date has provided empirical evidence based on a national survey sample for the association between food insecurity and breast cancer screening rates among women. In this study, we examine the effect of food insecurity on the utilization of recommended mammography screening services among US women aged 50 to 74 years. More specifically, we employed data from the Health and Retirement Study (HRS) and the Health Care and Nutrition Study (HCNS) to test the hypothesis that food insecurity is a potential barrier to breast cancer screening among this population.

Data Sources, Study Settings, and the Sample

In this study, we utilized data from the 2014 and 2016 waves of the HRS and the 2013 HCNS. The HRS is an ongoing, nationally representative, longitudinal survey that focuses on health, employment, income, and family structure among US residents aged 50 years and older. The HRS is conducted biennially by the Institute for Social Research (ISR) at the University of Michigan [24]. The HRS data and documents regarding its design and methodology can be accessed online at https://hrs.isr.umich.edu/. The HRS has several off-year studies, such as HCNS, on a variety of topics. The HCNS was fielded between November 2013 and May 2014. It included a random subsample (n = 12,418) of HRS core respondents who were invited to participate in a mail-out questionnaire. The HCNS provides information about respondents’ healthcare access, food purchases, food consumption, and nutrition; its response rate was 65% (n = 8,073) [25].

We merged HCNS and HRS data on unique respondent identifiers. The final sample included 2,861 female respondents aged 50 to 74 years in 2014. Figure 1 is a flowchart displaying the inclusionary and exclusionary criteria for the final sample. Both the HRS and HCNS are supported by the National Institute on Aging (grant number: NIA U01AG009740) and the Social Security Administration [25]. The HRS protocol was approved by the University of Michigan’s Institutional Review Board. Prior to each wave’s new interviews or re-interviews, participants are provided with a written informed consent information document; they are then read a confidentiality statement and provide oral consent. For the HCNS, consent was inferred by completion and return of the questionnaire; respondents were informed that their participation was voluntary in the study invitation letter [26]. The present study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

[FIG 1 GOES HERE]

Main Study Measures

Information on breast cancer screening was obtained from the 2016 wave of HRS. Participants were asked if they had a mammogram or x-ray of the breast to screen for cancer in the last two years; response options were (‘yes’ or ‘no’). The 2016 question on breast cancer screening was used to preserve the temporal sequence between the exposure and the outcome, and for the outcome to follow food security status assessment and baseline covariates in 2014. Data on food security were extracted from a portion of the HCNS, which assessed food insecurity using the six-item short form of the US Household Food Security Survey Module [27]. The assessment starts by asking the respondents whether the following two statements were often true, sometimes true, or never true for their households in the last 12 months: ‘The food that we bought just didn’t last and we didn’t have enough money to get more.’ and ‘We couldn't afford to eat balanced meals.’ The responses for each statement were scored as ‘1 = often true or sometimes true, 0 = never true.’ Then, they were asked if they, or other adults in the household, ever cut meal size or skip meals because there was not enough money for food: ‘yes’ or ‘no.’ Respondents who answered ‘yes’ where further asked about the frequencies of cutting portions or skipping meals ‘1 = almost every month or some months but not every month, 0 = only 1 or 2 months or no.’ The last two questions evaluate if, in the last twelve months, the respondent ever ate less than they felt they should because there was not enough money for food (‘yes’ or ‘no’) and whether the respondent ever went hungry but did not eat because there was not enough money for food (‘yes’ or ‘no’). The affirmative responses to all six questions were summed to develop the food security status scale score. Respondents with a raw score of 0 or 1 were considered ‘food-secure’ while those with raw scores ranging from 2 to 6 were considered ‘food-insecure’ [27].

Covariates.

We followed the extant literature and the Behavioral Model of Health Services Utilization, developed and modified by Andersen and colleagues [28–31], as a conceptual framework to identify covariates and related factors that could contribute to the associations between food security status and breast cancer screening. This behavioral model posits that individuals' use of health and healthcare services is a function of their predispositions to use health services, factors that enable or impede such use, and their need for care [28–31]. The model includes predisposing factors such as demographic factors, social structure factors, and belief factors (e.g., attitudes, values, and knowledge); individual- and family-level enabling factors (e.g., having a regular provider, household income, urban or rural location); and need factors, including perceived (e.g., self-rated health status), and evaluated (e.g., chronic medical conditions), needs.

Guided by the framework and literature, covariates were extracted from the 2014 wave of the HRS and considered baseline characteristics for survey respondents. Predisposing characteristics included participants’ age and race and ethnicity: ‘non-Hispanic White,’ ‘non-Hispanic Black,’ ‘Hispanic,’ or ‘non-Hispanic others.’ Enabling characteristics included education level (‘less than high school or GED,’ ‘high-school graduate,’ ‘some college,’ or ‘college and above’), annual household income (‘<$20,000,’ ‘$20,000–$35,000,’ ‘$35,001–$50,000,’ ‘$50,001–$75,000,’ and ‘>$75,000’), whether working for pay (‘yes’ or ‘no’), and census region (‘Northeast,’ ‘Midwest,’ ‘South,’ and ‘West’). Other enabling factors included whether the respondent was covered by a government health insurance plan (‘yes’ or ‘no’), had at least one private health insurance plan (‘yes’ or ‘no’), and had a usual place for health care/advice (‘yes’ or ‘no’). Need characteristics included respondent’s body mass index (BMI) (kg/m²) (‘Normal or underweight [< 25],’ ‘Overweight [25–29.9],’ and ‘Obese [≥ 30]’), smoking status (‘current,’ ‘former,’ or ‘never’), number of comorbid medical conditions (‘none,’ ‘one,’ ‘two and more’), having activities of daily living (ADL) limitations (i.e., ‘yes’ or ‘no’), having instrumental activities of daily living (IADL) limitations (‘yes’ or ‘no’), and their Center for Epidemiologic Studies Depression evaluation (CES-D) scale score (‘elevated depressive symptoms [CES-D ≥ 3]’ or ‘normal [CES-D < 3]’).

Statistical Analysis

We first calculated the respondents’ general characteristics and then estimated weighted proportions in crosstabulations of food security status and respondent characteristics. The design-adjusted Satterthwaite Rao-Scott chi-square test was used to assess the significance of cross-tabulated associations and t-test was used to evaluate continuous variables (Table 1). To properly adjust for confounding factors and reduce selection bias associated with food security status that could distort the estimated effects on breast cancer screening among women, we employed the propensity score (PS) method using the PROC PSMATCH procedure [32–34]. PS is generally calculated as the probability of treatment assignment conditional on the baseline covariates [35]. It usually balances the observed baseline covariates between treatment (food-insecure) and control (food-secure) groups, which could mimic the design of a randomized controlled trial [35]. Among different PS methods––such as matching, stratification, and weighting [33, 34]––we applied the propensity score weighting method to create an output dataset in which the distributions of the variables are balanced between the food-insecure and food-secure subgroups. The procedure was performed using the baseline covariates as a set of potential confounding variables with potential correlations to food security status and breast cancer screening. These variables were used to fit a logistic regression model, using food security status as the outcome, to compute the propensity scores through inverse probability of treatment weighting. In this step, original survey weights were incorporated as a predictor variable along with other covariates to enhance the estimates’ external validity [36]. Then the output PS weights were multiplied by original survey weights to form a new weight. The computed new weights were later applied in cross-tabulations between food security status and the baseline covariates (see Table 2) to investigate the significance of differences in characteristics between food security status subgroups post-PS weighting. Conventional regression models were used to output crude odds ratios (ORs) of breast cancer screening for baseline characteristics (Table 3), and crude and adjusted ORs for food security status (Table 4). Lastly, by applying the PS weights, a binary logistic regression model was fitted for breast cancer screening as an outcome to obtain the population average treatment effect (PATE) and to generate the population-average ORs of screening among food-insecure women (Table 4-Model III). Other complex survey sampling design elements, such as clustering and stratification, were properly accounted for in analyses [32, 36] to obtain accurate variance and population level estimates. All analyses were conducted using SAS 9.4 statistical software (SAS Institute Inc., Cary, NC, USA), and the significance threshold was set at P-value ≤ .05.

Table 1
Respondent characteristics (HRS & HCNS, n = 2,861 women, ages 50–74 years, the United States, 2014–2016).
	Total sample (n = 2,861)	Food-secure (n = 2,177)	Food-insecure (n = 684)
Respondent Characteristics	Frequency (Weighted %)	Frequency (Weighted %)	Frequency (Weighted %)	P-value ^a
Food security status
Food-secure	2,177 (79.0)
Food-insecure	684 (21.0)
Had a mammogram or breast x-ray in past two years
Yes	2,184 (76.5)	1,706 (79.0)	478 (67.3)	< .001
No	672 (23.5)	467 (21.0)	205 (32.7)
Predisposing characteristics
Age (years), mean (± SE)	61.6 (± .18)	62.2 (± .20)	59.6 (± .26)	< .001
Race and Ethnicity
Non-Hispanic White	1,790 (75.8)	1,536 (82.5)	254 (50.3)	< .001
Non-Hispanic Black	570 (11.2)	357 (8.6)	213 (21.1)
Hispanic	384 (9.1)	208 (6.0)	176 (21.0)
Non-Hispanic Others	110 (3.9)	72 (2.9)	38 (7.6)
Enabling characteristics
Level of education
Less than high school or GED	566 (16.5)	307 (11.1)	259 (36.4)	< .001
High-school graduate	862 (28.6)	657 (27.9)	205 (31.6)
Some college	785 (27.8)	613 (29.0)	172 (23.3)
College and above	647 (27.1)	599 (32.0)	48 (8.7)
Annual household income
<$20,000	666 (18.9)	319 (10.9)	347 (48.8)	< .001
$20,000–$35,000	483 (14.3)	330 (12.5)	153 (21.0)
$35,001–$50,000	409 (14.2)	328 (14.3)	81 (13.9)
$50,001–$75,000	469 (16.9)	395 (18.2)	74 (12.2)
>$75,000	834 (35.7)	805 (44.1)	29 (4.1)
Working for pay
Yes	1,214 (46.2)	966 (48.5)	248 (37.4)	< .001
No	1,645 (53.8)	1,209 (51.5)	436 (62.6)
Census region
Northeast	397 (15.9)	304 (16.1)	93 (15.1)	.07
Midwest	633 (25.6)	513 (26.9)	120 (20.5)
South	1,237 (38.5)	901 (36.8)	336 (45.0)
West	590 (20.0)	456 (20.2)	134 (19.4)
Covered by government health insurance plan
Yes	1,623 (53.3)	1,220 (51.4)	403 (60.8)	< .01
No	1,225 (46.7)	946 (48.6)	279 (39.2)
Have at least one private health insurance plan
Yes	1,648 (63.1)	1,429 (71.2)	219 (32.8)	< .001
No	1,202 (36.9)	737 (28.8)	465 (67.2)
Have a usual place for health care/advice
Yes	2,580 (92.0)	2,002 (93.4)	578 (87.1)	< .001
No	279 (8.0)	173 (6.6)	106 (12.9)
Need characteristics
BMI (kg/m²)
Normal or underweight (< 25)	596 (22.8)	498 (24.9)	98 (14.8)	< .001
Overweight (25–29.9)	822 (29.1)	646 (30.3)	176 (24.7)
Obese (≥ 30)	1,441 (48.1)	1,032 (44.8)	409 (60.5)
Smoking status
Current	347 (12.6)	215 (10.3)	132 (21.4)	< .001
Former	1,046 (36.4)	813 (36.3)	233 (36.4)
Never	1,456 (51.0)	1,142 (53.4)	314 (42.2)
Comorbid medical conditions
None	418 (15.5)	357 (17.2)	61 (9.4)	< .001
One	663 (25.1)	547 (27.2)	116 (17.2)
Two and more	1,780 (59.4)	1,273 (55.6)	507 (73.3)
ADL limitations
Yes	396 (13.1)	203 (8.8)	193 (29.1)	< .001
No	2,465 (86.9)	1,974 (91.2)	491 (70.9)
IADL limitations
Yes	347 (11.2)	178 (7.4)	169 (25.2)	< .001
No	2,514 (88.8)	1,999 (92.6)	515 (74.8)
CES-D scale
Elevated depressive symptoms (CES-D ≥ 3)	442 (14.3)	210 (8.9)	232 (34.6)	< .001
Normal (CES-D < 3)	2,419 (85.7)	1,967 (91.1)	452 (65.4)
Abbreviations: SE; Standard Error; GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale.
^a The design-Adjusted Rio-Scott Chi-square test or the t-test.

Table 2
Post-propensity score-weighted associations between food security status and baseline respondent characteristics.
	Food-secure	Food-insecure
Respondent Characteristics	Weighted %	Weighted %	P-value
Food security status	54.0	46.0	.15
Had a mammogram or breast x-ray in past two years
Yes	77.6	61.3	< .01
No	22.4	38.7
Predisposing characteristics
Age (years), Mean	61.7	61.2	.41
Race and Ethnicity
Non-Hispanic White	76.6	66.4	.13
Non-Hispanic Black	11.2	15.4
Hispanic	8.6	12.6
Non-Hispanic others	3.6	5.6
Enabling characteristics
Level of education
Less than high school or GED	15.9	17.6	.78
High-school graduate	28.3	29.7
Some college	28.3	29.2
College and above	27.5	23.5
Annual household income
<$20,000	18.8	23.1	.53
$20,000–$35,000	14.1	16.6
$35,001–$50,000	14.0	13.9
$50,001–$75,000	16.9	16.0
>$75,000	36.2	30.4
Working for pay
Yes	46.1	49.0	.58
No	53.9	51.0
Census region
Northeast	15.7	13.8	.91
Midwest	26.4	25.0
South	37.7	38.5
West	20.2	22.7
Covered by government health insurance plan
Yes	53.3	49.9	.56
No	46.7	50.1
Have at least one private health insurance plan
Yes	63.7	62.2	.76
No	36.3	37.8
Have a usual place for health care/advice
Yes	92.4	86.1	.22
No	7.6	13.9
Need characteristics
BMI (kg/m2)
Normal or underweight (< 25)	22.7	18.7	.45
Overweight (25–29.9)	29.4	27.0
Obese (≥ 30)	47.9	54.3
Smoking status
Current	13.4	16.3	.63
Former	35.5	33.3
Never	51.1	50.4
Comorbid medical conditions
None	15.4	18.5	.66
One	25.0	21.8
Two and more	59.6	59.7
ADL limitations
Yes	13.7	14.8	.61
No	86.3	85.2
IADL limitations
Yes	12.0	13.4	.47
No	88.0	86.6
CES-D scale
Elevated depressive symptoms (CES-D ≥ 3)	14.1	16.6	.32
Normal (CES-D < 3)	85.9	83.4
Abbreviations: GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale.

Table 3
Crude associations between respondents’ predisposing, enabling, need characteristics and reporting breast cancer screening in past two years among women 50–74 years (HRS & HCNS, n = 2,861 women, the United States, 2014–2016).
Respondent Characteristics	Total Study Sample Crude OR (95% CI)
Predisposing characteristics
Age in years	0.98 (0.97–1.01)
Race and Ethnicity
Non-Hispanic White	Reference
Non-Hispanic Black	1.23 (0.92–1.64)
Hispanic	1.00 (0.71–1.41)
Non-Hispanic others	0.83 (0.45–1.56)
Enabling characteristics
Level of education
Less than high school or GED	Reference
High-school graduate	1.26 (0.93–1.70)
Some college	1.44 (1.04-2.00)*
College and above	2.16 (1.54–3.02)***
Annual household income
<$20,000	Reference
$20,000–$35,000	1.21 (0.82–1.78)
$35,001–$50,000	1.64 (1.07–2.51)*
$50,001–$75,000	2.80 (2.05–3.83)***
>$75,000	2.50 (1.78–3.50)***
Working for pay
Yes	1.33 (1.05–1.68)*
No	Reference
Census region
Northeast	Reference
Midwest	0.65 (0.48–0.89)**
South	0.84 (0.63–1.13)
West	0.82 (0.59–1.15)
Covered by government health insurance plan
Yes	0.80 (0.63–1.02)
No	Reference
Have at least one private health insurance plan
Yes	2.02 (1.60–2.55)***
No	Reference
Have a usual place for health care/advice
Yes	2.14 (1.53–2.99)***
No	Reference
Need characteristics
BMI (kg/m²)
Normal or underweight (< 25)	Reference
Overweight (25–29.9)	1.11 (0.83–1.48)
Obese (≥ 30)	1.12 (0.82–1.53)
Smoking status
Current	Reference
Former	2.43 (1.68–3.52)***
Never	2.41 (1.67–3.48)***
Comorbid medical conditions
None	Reference
One	1.02 (0.71–1.47)
Two and more	0.96 (0.68–1.37)
ADL limitations
Yes	0.46 (0.34–0.64)***
No	Reference
IADL limitations
Yes	0.49 (0.33–0.72)***
No	Reference
CES-D scale
Elevated depressive symptoms (CES-D ≥ 3)	0.55 (0.42–0.71)***
Normal (CES-D < 3)	Reference
Abbreviations: OR, Odds Ratio; CI, Confidence Interval; GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale. Indicate P-value ≤ .05, * indicate P-value < .01, *** indicate P-value < .001.

Table 4
Logistic regression results for the association between food security status and breast cancer screening among women of ages 50–74 in the United States (HRS & HCNS, n = 2,861, 2014–2016).
	Receipt of mammogram or breast X-ray in past two years
	Model I	Model II ^a	Model III ^b
Respondent Characteristic	Crude OR (95% CI)	Adjusted OR (95% CI)	OR (95% CI)
Food security status
Food-insecure	0.55 (0.42–0.71)*	0.79 (0.56–1.12)	0.46 (0.30–0.70)*
Food-secure	Reference	Reference	Reference
Abbreviations: OR, Odds Ratio; CI, Confidence Interval.
^a The regression model was adjusted for all predisposing, enabling, and need characteristics of the respondents.
^b Propensity score weights were applied.
* ORs were statistically significant at P-values < .001.

Our sample represented nearly 32.4 million women aged 50 to 74 residing in the US during 2014. Within this population, 21% were deemed food-insecure, and 76.5% reported receiving breast cancer screening in the past two years. The sample’s mean (± standard error [SE]) age was 61.6 ± .18 years. Non-Hispanic Whites made up the largest proportion (75.8%) of respondents. More than half of the respondents (54.9%) had some college or college and above education. Respondents had an annual household income of >$75,000 (35.7%), were not working for pay (53.8%), and were from the southern regions of the U.S. (38.5%). Over half (53.3%) of respondents were covered by government health insurance plans, had at least one private health insurance plan (63.1%), and had a usual place for health care/advice (92%). The proportion of respondents who reported being obese, never-smokers, and those with two and more comorbid conditions, any ADL limitation, any IADL limitations, and elevated depressive symptoms were (48.1%), (51.0%), (59.4%), (13.1%), (11.2%), and (14.3%), respectively (see Table 1).

Among the food-insecure, about two-thirds (67.3%) reported receiving breast cancer screening in the past two years. Food-insecure women had a mean (±SE) age of 59.6 ± .26 years; they were predominantly non-Hispanic Whites (50.3%), had less than high school or GED education (36.4%), had an annual household income of <$20,000 (48.8%), were not working for pay (62.6%), and were from the southern regions of the U.S. (45.0%). About 60.8% of food-insecure women reported having government health insurance coverage, 67.2% reported having no private health insurance plan, and 87.1% reported having a usual place for health care/advice. The proportions of women who were obese, never-smokers, those with two and more comorbid conditions, any ADL and IADL limitations, and those with elevated depressive symptoms among the food-insecure group were (60.5%), (42.2%), (73.3%), (29.1%), (25.2%), and (34.6%), respectively (see Table 1).

Table 2 shows an improved covariate balance between food-secure and food-insecure subgroups after applying the PS weights. The weighted proportion of food-insecure women now stands at (46.0%) not significantly different (P-value = .15) than food-secure women. The two groups are substantially balanced on all characteristics, except for the outcome, when we compare them to the pre-PS weighting distributions in Table 1. This relatively balanced distribution facilitated an unbiased estimate of the odds of breast cancer screening in our final model, where the PS weights were applied.

We analyzed crude associations between respondents’ reports of a breast cancer screening in the past two years and their predisposing, enabling, and need characteristics (see Table 3). The likelihood of reporting a screening was positively associated with respondents’ education level; women with a college education or higher were more than twice as likely to have received a screening compared to those who had not completed high school (OR=2.16; 95% CI: 1.54-3.02; P-value <.001). The odds of screening increased nearly three-fold for women reporting an income greater than $75,000 (OR=2.50; 95% CI: 1.78-3.50; P-value <.001; vs. <$20,000) and over 33% for women who reported working for pay (OR=1.33; 95% CI: 1.05-1.68; P-value = .02; vs. no). Furthermore, those reported at least one private health insurance plan (OR=2.02; 95% CI: 1.60-2.55; P-value <.001) and those having a regular place for health care/advice (OR=2.14; 95% CI: 1.53-2.99; P-value <.001) were both twice as likely to have received screenings compared to their counterparts. Health-related practices and health status were also associated with breast cancer screening. Both never (OR=2.41; 95% CI: 1.67-3.48; P-value <.001) and former smokers (OR=2.43; 95% CI: 1.68-3.52; P-value <.001) were twice as likely to report screening than current smokers. Health status was negatively associated with breast cancer screening; those with ADL (OR=0.46; 95% CI: 0.34-0.64; P-value <.001; vs. no ADL limitations) or IADL limitations (OR=0.49; 95% CI: 0.33-0.72; P-value <.001; vs. no IADL limitations) were half as likely to receive these screenings. Those evaluated as having elevated depressive symptoms were 45% less likely to receive a screening (OR=0.55; 95% CI: 0.42-0.71; P-value <.001) compared to those with a normal CES-D score.

Without any adjustments for confounding and selection bias, the crude associations show that (Model I; see Table 4) food-insecure women had 45% lower odds of reporting breast cancer screening in the past two years (OR=0.55; 95% CI: 0.42-0.71, P-value <.001) when compared to food-secure women. The conventional multivariable logistic regression model (Model II; see Table 4) indicates that, after adjusting for predisposing, enabling, and need factors of the respondents, the odds of reporting breast cancer screening in the past two years were still 21% lower among food-insecure women (adjusted OR=0.79; 95% CI: 0.56-1.12) compared to those food-secure; however, the differences were nonsignificant. Model III (see Table 4) shows the average estimated effect of food insecurity on breast cancer screening among women when the PS weights are applied. In this model, food-insecure women had 54% lower odds of reporting screening in the past two years (OR=0.46; 95% CI: 0.30-0.70, P-value <.001) when compared to food-secure women. The results from Model III show more precision with less potentiality for confounding and selection bias.

Drawing from a nationally representative sample of women aged 50 to 74, the present study revealed that food security status had a substantial effect on subsequent breast cancer screening. Respondents who experienced food insecurity were significantly less likely to report engaging in formal breast cancer screening activities (i.e., mammogram or breast x-ray) within the past two years than their food-secure counterparts. This conclusion was drawn after proper adjustments for confounding and selection bias through a PS weighting method utilized to create a balance between food security status subgroups based on a set of observed respondent characteristics, and through redesigning the study to resemble a randomized controlled trial in the best practical way.

The present findings align with the existing research suggesting that food insecurity negatively impacts health services access and utilization among US residents. The existing literature suggests that food insecurity is associated with delaying medical care [37], lower treatment adherence [38, 39], and poor chronic disease management [18, 40]. Berkowitz and colleagues in 2014 attributed these trends to the ‘treat or eat’ dilemma experienced by food-insecure people; in such instances, they are forced to choose between accessing medical services or meeting the basic needs of food and shelter [41]. Our results indicate that women who experience food insecurity may encounter a similar dilemma when considering whether to engage in breast cancer screening. Previous findings have led to a recent call for cancer care providers to integrate food insecurity screening and resources into cancer care [42]. The present study results suggest that addressing the relationship between food insecurity and cancer prevalence must also involve efforts to promote screening behaviors among those who are food-insecure, especially older women at risk for breast cancer.

Among food-insecure women, barriers to breast cancer screening are structural, multilayered, and multi-factorial. Any accumulation, or a combination, of factors addressed in our study, in addition to other unmeasured factors, might partially explain the lower screening rates among food-insecure women. For example, racial and ethnic minority women, those with low socioeconomic status, lower levels of education, and women with comorbid health conditions and disabilities are more likely to be food-insecure, and they are more likely to face logistical, psychological, and cultural barriers to breast cancer screening [11, 14]. Furthermore, among these groups of women, other factors such as lack of health literacy and awareness about breast cancer screening, lack of physician recommendation, concerns about screening success and efficacy, lack of public or private health insurance, and mistrust in healthcare services have been frequently reported to influence their decision to perform or to utilize, breast cancer mammography [1, 11, 14]. Other significant barriers of screening for food-insecure women might include the location of residence and travel times to mammogram centers [12]. In manners similar to ‘treat or eat’ dilemma, food-insecure women could be forced to decide whether to use money and other resources on food, medical care, utilities, or housing [43, 44]. This is more evident for minority women and low-income urban residents who do not have their own transportation and are more likely to depend on public transportation [43]. These facts support the notion that understanding lower breast cancer screening rates among food-insecure women cannot merely be analyzed through their socioeconomic status. Among these women, understanding the full scope of breast cancer screening barriers requires analyses that account for potential barriers at the individual, family, community, policy, and health system levels.

The crude associations between reported breast cancer screenings and respondents’ predisposing, enabling, and need characteristics demonstrated that such screenings were more prevalent among the socioeconomically advantaged––i.e., those with more education, higher incomes, and consistent healthcare access. These findings are consistent with past research suggesting that medical screenings, increased health literacy, and greater perceived control over disease prevention are tied to education level [45, 46]. Similarly, those with higher incomes and reliable healthcare services would be less likely to experience financial, institutional, or environmental barriers in paying for and accessing preventative healthcare such as breast cancer screening services [47]. Conversely, those in socioeconomically disadvantaged groups are more likely to experience food insecurity, which can be a barrier to utilizing preventative health services.

Additional factors that were negatively associated with breast cancer screening in the present study––including smoking and health status––may also be tied to respondents’ socioeconomic status. For example, smoking behaviors, which are most prevalent among those with lower incomes and less education [48], are generally less likely to engage in preventative health behaviors [49]. The mediating role of life stressors, which may be more likely to manifest amongst those of lower socioeconomic standing, on cigarette usage and insufficient usage of preventative health services also merits further investigation [50]. Cancer screenings were also less likely among those with physical disabilities and poor mental health status. Existing scholarship suggests that women with disabilities encounter significant barriers to obtain screenings and may be deterred by factors that include unreliable transportation, lack of preparedness of healthcare providers, undermining or silencing of their questions, scarcity of barrier-free clinics, and the additional effort required to obtain such screenings [51]. Likewise, those experiencing depression often experience limited energy and motivation [52], making it difficult to engage in cancer screening behaviors.

Efforts to address food insecurity among vulnerable populations in the US have shown some promise. For instance, the Supplemental Nutrition Assistance Program (SNAP) is one of the primary mechanisms for reducing food insecurity in these vulnerable subgroups. Recent research has linked SNAP with improved health outcomes both in regard to positive self-assessments and by addressing the ‘treat or eat’ compromise through providing beneficiaries the ability to spend money and resources on their health that would have been spent on food. Moreover, among older adults, such as the women in this study, participation in SNAP report a reduced likelihood of admission to nursing homes or hospitals compared to non-participating counterparts [53]. Although an association between SNAP participation and breast cancer screening rates has yet to be examined, reducing food insecurity among older women could increase their likelihood of utilizing preventative health services. Targeting social determinants of health, such as food insecurity, could trigger a protective mechanism that would promote health among the most disadvantaged populations. Regardless of the mechanism through which food insecurity is negatively associated with breast cancer screenings, the presence of this disparity in a vulnerable population (i.e., food-insecure women) warrants special consideration from policymakers and researchers. Currently, elderly adults contribute to a significant amount of the US disease burden. According to the latest data from the Centers for Medicare and Medicaid Services (2014), women accounted for 56% of total personal healthcare spending. Additionally, women aged 65 years and over accounted for more than 58% of healthcare spending for older adults [54]. Hence, failure to address disparities in preventative health service utilization, including breast cancer screenings, for this demographic subgroup could further worsen their health and increase healthcare expenditures.

The results of our study must be interpreted, bearing several limitations in mind. First, a significant portion of data from HRS, and all data from HCNS, specifically the questions on food security and breast cancer screening, are respondent self-reports, which are prone to bias. For example, there might be underreporting or overreporting of some measures due to social desirability. Second, we followed the guidelines provided by the USPTSF to define the age limits of the studied population; thus, the results are only generalizable to women of ages 50 and 74 in the US. Other agencies, such as the American Cancer Society, have slightly different screening guidelines that include younger age groups below age 50 [3]. Their recommendations encourage women with an average risk of breast cancer to undergo regular screening mammography starting at age 45, and those of ages 45 to 54 to conduct screening annually. Based on those guidelines, there is no cap on age to perform mammography for women as long as their overall health is good, and they have a life expectancy of 10 years or longer [3].

Lastly, while PS weighting is a powerful method for causal associations, it does not control for unobserved covariates. The PS weighing results are interpreted under the assumption of no unobserved confounding. Despite this assumption, unmeasured confounding might still be present, and the assumption cannot truly be tested outside of an actual randomized study. Hence, we were primarily limited to the information provided in the HRS and HCNS surveys. Those unmeasured covariates could have contributed to breast cancer screening differences between the food-secure and food-insecure women.

Among a nationally representative sample of women aged 50 and 74 years residing in the U.S., we found food insecurity to be an additional barrier to breast cancer screening. Since the reasons for lower rates of breast cancer screening among food-insecure women, compared to those food-secure, are multi-faceted, any effort to increase screening rates among food-insecure women should address the complex needs of the targeted population. Community-level programs or campaigns such as breast cancer awareness and screening or other health education programs for women should be designed to incorporate or reflect the sociodemographic profile of food-insecure women.

Further, in efforts to increase screening rates among food-insecure women, healthcare providers have substantial roles and responsibilities. Healthcare providers and primary care physicians can screen their patients for food insecurity and follow up with women to confirm that they are up to date regarding cancer screening recommendations. Our study found that about 87.1% of food-insecure women had a usual place for health care and health advice. Thus, healthcare providers could communicate the health benefits of screening behavior and share the best possible screening options for insured and uninsured food-insecure women. In the US, private health insurance plans and the State Medicaid expansion programs must cover mammography at least biennially and as frequently as once a year for women ages 40 to 74 who are at average risk of breast cancer, without consumer cost-sharing [55]. Other government-sponsored health insurance programs like Medicare Part B cover annual screening mammograms for women over 40 at no cost to consumers [55].

Moreover, programs such as the National Breast and Cervical Cancer Early Detection Program (NBCCEDP) provide an opportunity for uninsured and underinsured women of ages 40 to 64, who are at or below 250% of the Federal Poverty Level, to gain access to breast cancer screening services [55]. Healthcare providers can also help refer food-insecure women to social workers or case managers to facilitate their enrollment into SNAP and other food assistance programs in local communities [56]. Further studies are needed on this topic, specifically to investigate breast cancer screening barriers within food-insecure women.

Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of Interest: The authors do not have any conflicts of interest to declare.

Data availability: The data analyzed for the current study were obtained from the Health and Retirement Study, publicly available to access and download at: About | Health and Retirement Study (umich.edu)

Code availability Code is available upon request from the corresponding author.

American Cancer Society (2020) About Breast Cancer. https://www.cancer.org/content/dam/CRC/PDF/Public/8577.00.pdf. Accessed 25 Aug 2020
American Cancer Society (2020) Cancer Facts & Figures 2020. https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-and-statistics/annual-cancer-facts-and-figures/2020/cancer-facts-and-figures-2020.pdf. Accessed 25 Aug 2020
Oeffinger KC, Fontham ETH, Etzioni R, et al (2015) Breast Cancer Screening for Women at Average Risk: 2015 Guideline Update From the American Cancer Society. JAMA 314:1599–1614. https://doi.org/10.1001/jama.2015.12783
Coleman C (2017) Early Detection and Screening for Breast Cancer. Seminars in Oncology Nursing 33:141–155. https://doi.org/10.1016/j.soncn.2017.02.009
Kim H, Chang CF (2020) Effectiveness of Using Personal Health Records to Improve Recommended Breast Cancer Screening and Reduce Racial and Geographic Disparities Among Women. J Canc Educ. https://doi.org/10.1007/s13187-020-01821-2
United States Preventive Services Taskforce (2016) Recommendation: Breast Cancer: Screening. https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/breast-cancer-screening. Accessed 25 Aug 2020
Healthy People (2020) Cancer: Objectives. https://www.healthypeople.gov/2020/topics-objectives/topic/cancer/objectives. Accessed 25 Aug 2020
Ahmed AT, Welch BT, Brinjikji W, et al (2017) Racial Disparities in Screening Mammography in the United States: A Systematic Review and Meta-analysis. Journal of the American College of Radiology 14:157-165.e9. https://doi.org/10.1016/j.jacr.2016.07.034
Alexandraki I, Mooradian AD (2010) Barriers Related to Mammography Use for Breast Cancer Screening Among Minority Women. Journal of the National Medical Association 102:206–218. https://doi.org/10.1016/S0027-9684(15)30527-7
Henderson LM, O’Meara ES, Haas JS, et al (2020) The Role of Social Determinants of Health in Self-Reported Access to Health Care Among Women Undergoing Screening Mammography. Journal of Women’s Health. https://doi.org/10.1089/jwh.2019.8267
Miller BC, Bowers JM, Payne JB, Moyer A (2019) Barriers to mammography screening among racial and ethnic minority women. Social Science & Medicine 239:112494. https://doi.org/10.1016/j.socscimed.2019.112494
Onitilo A, Liang H, Stankowski R, et al (2014) Geographical and seasonal barriers to mammography services and breast cancer stage at diagnosis. 13
Sarma EA (2015) Barriers to screening mammography. Health Psychology Review 9:42–62. https://doi.org/10.1080/17437199.2013.766831
Watson-Johnson LC, DeGroff A, Steele CB, et al (2011) Mammography Adherence: A Qualitative Study. Journal of Women’s Health 20:1887–1894. https://doi.org/10.1089/jwh.2010.2724
Seligman HK, Laraia BA, Kushel MB (2010) Food Insecurity Is Associated with Chronic Disease among Low-Income NHANES Participants. The Journal of Nutrition 140:304–310. https://doi.org/10.3945/jn.109.112573
Coleman-Jensen A, Rabbitt MP, Gregory CA, Singh A (2019) Household Food Security in the United States in 2018. http://www.ers.usda.gov/publications/pub-details/?pubid=94848. Accessed 25 Aug 2020
Berkowitz SA, Seligman HK, Meigs JB, Basu S (2018) Food Insecurity, Healthcare Utilization, and High Cost: A Longitudinal Cohort Study. Am J Manag Care 24:399–404
Gundersen C, Ziliak JP (2015) Food Insecurity And Health Outcomes. Health Affairs 34:1830–1839. https://doi.org/10.1377/hlthaff.2015.0645
Coleman-Jensen A, Rabbit MP, Gregory CA, et al. (2015). Household Food Security in the United States in 2015, ERR-215, U.S. Department of Agriculture, Economic Research Service, September 2016.
Walker RE, Keane CR, Burke JG (2010) Disparities and access to healthy food in the United States: A review of food deserts literature. Health & Place 16:876–884. https://doi.org/10.1016/j.healthplace.2010.04.013
Berkowitz SA, Basu S, Meigs JB, Seligman HK (2018) Food Insecurity and Health Care Expenditures in the United States, 2011-2013. Health Serv Res 53:1600–1620. https://doi.org/10.1111/1475-6773.12730
Melo A, Matias MA, Dias SS, et al (2019) Is food insecurity related to health-care use, access and absenteeism? Public Health Nutr 22:3211–3219. https://doi.org/10.1017/S1368980019001885
Mahmood A, Kim H, Kabir U, et al (2020) Food Insecurity and Influenza and Pneumonia Vaccines Uptake Among Community-Dwelling Older Adults in the United States. J Community Health 45:943–953. https://doi.org/10.1007/s10900-020-00812-0
Sonnega A, Faul JD, Ofstedal MB, et al (2014) Cohort Profile: the Health and Retirement Study (HRS). Int J Epidemiol 43:576–585. https://doi.org/10.1093/ije/dyu067
Health and Retirement Study (2013) Data Description 2013 Health Care and Nutrition Study (HCNS) (Version 4.0). https://hrsonline.isr.umich.edu/modules/meta/2013/hcns/desc/2013HCNS_data_description.pdf. Accessed 10 July 2020.
University of Michigan Institute for Social Research (2018) Institutional Review Board Information. https://hrs.isr.umich.edu/sites/default/files/biblio/HRS_IRB_Information%28web%29_08_2018.pdf.Accessed 10 July 2020.
United States Department of Agriculture Economic Research Service (2012) U.S. Household Food Security Survey Module: Six-Item Short Form Economic Research Service. https://www.ers.usda.gov/media/8282/short2012.pdf. Accessed 25 Aug 2020
Aday LA, Andersen R (1974) A Framework for the Study of Access to Medical Care. Health Serv Res 9:208–220
Aday LA, Andersen RM (1981) Equity of access to medical care: a conceptual and empirical overview. Med Care 19:4–27
Andersen RM (1995) Revisiting the Behavioral Model and Access to Medical Care: Does it Matter? Journal of Health and Social Behavior 36:1–10. https://doi.org/10.2307/2137284
Andersen RM, Davidson PL, Baumeister SE (2013) Improving access to care. Changing the US health care system: Key issues in health services policy and management. Chapter Two 37
Karabon P (2019) Applying Propensity Score Methods to Complex Survey Data Using PROC PSMATCH. 10
SAS Institute Inc (2018) SAS/STAT® 15.1 User’s Guide. 128
Yuan Y, Yung Y-F, Stokes M (2017) Propensity Score Methods for Causal Inference with the PSMATCH Procedure. 17
Jing B, Diaz-Ramirez LG, Boscardin WJ, Steinman MA (2020) Propensity Score Matching with Survey Data. 14
Dugoff EH, Schuler M, Stuart EA (2014) Generalizing observational study results: applying propensity score methods to complex surveys. Health Serv Res 49:284–303. https://doi.org/10.1111/1475-6773.12090
Kushel MB, Gupta R, Gee L, Haas JS (2006) Housing instability and food insecurity as barriers to health care among low-income americans. J GEN INTERN MED 21:71–77. https://doi.org/10.1111/j.1525-1497.2005.00278.x
Afulani P, Herman D, Coleman-Jensen A, Harrison GG (2015) Food Insecurity and Health Outcomes Among Older Adults: The Role of Cost-Related Medication Underuse. Journal of Nutrition in Gerontology and Geriatrics 34:319–342. https://doi.org/10.1080/21551197.2015.1054575
McDougall JA, Anderson J, Adler Jaffe S, et al (2020) Food Insecurity and Forgone Medical Care Among Cancer Survivors. JCO Oncol Pract JOP1900736. https://doi.org/10.1200/JOP.19.00736
H. K. Seligman, E. A. Jacobs, A. Lopez, et al (2012) Food Insecurity and Glycemic Control Among Low-Income Patients With Type 2 Diabetes. Diabetes Care 35:233–238. https://doi.org/10.2337/dc11-1627
Berkowitz SA, Seligman HK, Choudhry NK (2014) Treat or Eat: Food Insecurity, Cost-related Medication Underuse, and Unmet Needs. The American Journal of Medicine 127:303-310.e3. https://doi.org/10.1016/j.amjmed.2014.01.002
Patel KG, Borno HT, Seligman HK (2019) Food insecurity screening: A missing piece in cancer management. Cancer 125:3494–3501. https://doi.org/10.1002/cncr.32291
Peipins LA, Graham S, Young R, et al (2011) Time and Distance Barriers to Mammography Facilities in the Atlanta Metropolitan Area. J Community Health 36:675–683. https://doi.org/10.1007/s10900-011-9359-5
Weinfield NS, Mills G, Borger C, et al (2014) Hunger in America 2014: National Report Prepared for Feeding America. 177
Brunello G, Fort M, Schneeweis N, Winter-Ebmer R (2016) The Causal Effect of Education on Health: What is the Role of Health Behaviors? Health Econ 25:314–336. https://doi.org/10.1002/hec.3141
Park CL, Cho D, Moore PJ (2018) How does education lead to healthier behaviours? Testing the mediational roles of perceived control, health literacy and social support. Psychology & Health 33:1416–1429. https://doi.org/10.1080/08870446.2018.1510932
Fleary SA, Ettienne-Gittens R, Heffer RW (2013) Perceptions of Preventive Health Care and Healthy Lifestyle Choices for Low Income Families: A Qualitative Study. ISRN Prev Med 2013:. https://doi.org/10.5402/2013/189180
Drope J, Liber AC, Cahn Z, et al (2018) Who’s still smoking? Disparities in adult cigarette smoking prevalence in the United States. CA: A Cancer Journal for Clinicians 68:106–115. https://doi.org/10.3322/caac.21444
Lim SS, Vos T, Flaxman AD, et al (2012) A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380:2224–2260. https://doi.org/10.1016/S0140-6736(12)61766-8
Zhang C, Brook JS, Leukefeld CG, et al (2017) Lack of Preventive Health Behaviors in the Early Forties: The Role of Earlier Trajectories of Cigarette Smoking from Adolescence to Adulthood. Subst Use Misuse 52:1527–1537. https://doi.org/10.1080/10826084.2017.1281310
Angus J, Seto L, Barry N, et al (2012) Access to Cancer Screening for Women with Mobility Disabilities. J Canc Educ 27:75–82. https://doi.org/10.1007/s13187-011-0273-4
Ludman EJ, Ichikawa LE, Simon GE, et al (2010) Breast and Cervical Cancer Screening: Specific Effects of Depression and Obesity. American Journal of Preventive Medicine 38:303–310. https://doi.org/10.1016/j.amepre.2009.10.039
Keith-Jennings B, Llobrera J, Dean S (2019) Links of the Supplemental Nutrition Assistance Program With Food Insecurity, Poverty, and Health: Evidence and Potential. Am J Public Health 109:1636–1640. https://doi.org/10.2105/AJPH.2019.305325
Centers for Medicare & Medicaid Services (2014) U.S. Personal Health Care Spending by Age and Gender. U.S. Personal Health Care Spending by Age and Gender (cms.gov). Accessed 2 Dec 2020
Kaiser Family Foundation (2019) Coverage of Breast Cancer Screening and Prevention Services. In: KFF. https://www.kff.org/womens-health-policy/fact-sheet/coverage-of-breast-cancer-screening-and-prevention-services/. Accessed 31 Aug 2020
Torres J, De Marchis E, Fichtenberg C, Gottlieb L (2017) Identifying Food Insecurity in Health Care Settings: A Review of the Evidence. https://sirenetwork.ucsf.edu/sites/sirenetwork.ucsf.edu/files/SIREN_FoodInsecurity_Brief.pdf. Accessed 31 Aug 2020

Table 1. Respondent characteristics (HRS & HCNS, n=2,861 women, ages 50-74 years, the United States, 2014-2016).


	Total sample (n =2,861)	Food-secure (n =2,177)	Food-insecure (n = 684)
Respondent Characteristics	Frequency (Weighted %)	Frequency (Weighted %)	Frequency (Weighted %)	P-value^a
Food security status
Food-secure	2,177 (79.0)
Food-insecure	684 (21.0)
Had a mammogram or breast x-ray in past two years
Yes	2,184 (76.5)	1,706 (79.0)	478 (67.3)	<.001
No	672 (23.5)	467 (21.0)	205 (32.7)
Predisposing characteristics
Age (years), mean (±SE)	61.6 (±.18)	62.2 (±.20)	59.6 (±.26)	<.001
Race and Ethnicity
Non-Hispanic White	1,790 (75.8)	1,536 (82.5)	254 (50.3)	<.001
Non-Hispanic Black	570 (11.2)	357 (8.6)	213 (21.1)
Hispanic	384 (9.1)	208 (6.0)	176 (21.0)
Non-Hispanic Others	110 (3.9)	72 (2.9)	38 (7.6)
Enabling characteristics
Level of education
Less than high school or GED	566 (16.5)	307 (11.1)	259 (36.4)	<.001
High-school graduate	862 (28.6)	657 (27.9)	205 (31.6)
Some college	785 (27.8)	613 (29.0)	172 (23.3)
College and above	647 (27.1)	599 (32.0)	48 (8.7)
Annual household income
<$20,000	666 (18.9)	319 (10.9)	347 (48.8)	<.001
$20,000–$35,000	483 (14.3)	330 (12.5)	153 (21.0)
$35,001–$50,000	409 (14.2)	328 (14.3)	81 (13.9)
$50,001–$75,000	469 (16.9)	395 (18.2)	74 (12.2)
>$75,000	834 (35.7)	805 (44.1)	29 (4.1)
Working for pay
Yes	1,214 (46.2)	966 (48.5)	248 (37.4)	<.001
No	1,645 (53.8)	1,209 (51.5)	436 (62.6)
Census region
Northeast	397 (15.9)	304 (16.1)	93 (15.1)	.07
Midwest	633 (25.6)	513 (26.9)	120 (20.5)
South	1,237 (38.5)	901 (36.8)	336 (45.0)
West	590 (20.0)	456 (20.2)	134 (19.4)
Covered by government health insurance plan
Yes	1,623 (53.3)	1,220 (51.4)	403 (60.8)	<.01
No	1,225 (46.7)	946 (48.6)	279 (39.2)
Have at least one private health insurance plan
Yes	1,648 (63.1)	1,429 (71.2)	219 (32.8)	<.001
No	1,202 (36.9)	737 (28.8)	465 (67.2)
Have a usual place for health care/advice
Yes	2,580 (92.0)	2,002 (93.4)	578 (87.1)	<.001
No	279 (8.0)	173 (6.6)	106 (12.9)
Need characteristics
BMI (kg/m²)
Normal or underweight (<25)	596 (22.8)	498 (24.9)	98 (14.8)	<.001
Overweight (25–29.9)	822 (29.1)	646 (30.3)	176 (24.7)
Obese (≥30)	1,441 (48.1)	1,032 (44.8)	409 (60.5)
Smoking status
Current	347 (12.6)	215 (10.3)	132 (21.4)	<.001
Former	1,046 (36.4)	813 (36.3)	233 (36.4)
Never	1,456 (51.0)	1,142 (53.4)	314 (42.2)
Comorbid medical conditions
None	418 (15.5)	357 (17.2)	61 (9.4)	<.001
One	663 (25.1)	547 (27.2)	116 (17.2)
Two and more	1,780 (59.4)	1,273 (55.6)	507 (73.3)
ADL limitations
Yes	396 (13.1)	203 (8.8)	193 (29.1)	<.001
No	2,465 (86.9)	1,974 (91.2)	491 (70.9)
IADL limitations
Yes	347 (11.2)	178 (7.4)	169 (25.2)	<.001
No	2,514 (88.8)	1,999 (92.6)	515 (74.8)
CES-D scale
Elevated depressive symptoms (CES-D ≥3)	442 (14.3)	210 (8.9)	232 (34.6)	<.001
Normal (CES-D <3)	2,419 (85.7)	1,967 (91.1)	452 (65.4)
Abbreviations: SE; Standard Error; GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale.
^aThe design-Adjusted Rio-Scott Chi-square test or the t-test.

Table 2. Post-propensity score-weighted associations between food security status and baseline respondent characteristics.

	Food-secure	Food-insecure
Respondent Characteristics	Weighted %	Weighted %	P-value
Food security status	54.0	46.0	.15
Had a mammogram or breast x-ray in past two years
Yes	77.6	61.3	<.01
No	22.4	38.7
Predisposing characteristics
Age (years), Mean	61.7	61.2	.41
Race and Ethnicity
Non-Hispanic White	76.6	66.4	.13
Non-Hispanic Black	11.2	15.4
Hispanic	8.6	12.6
Non-Hispanic others	3.6	5.6
Enabling characteristics
Level of education
Less than high school or GED	15.9	17.6	.78
High-school graduate	28.3	29.7
Some college	28.3	29.2
College and above	27.5	23.5
Annual household income
<$20,000	18.8	23.1	.53
$20,000–$35,000	14.1	16.6
$35,001–$50,000	14.0	13.9
$50,001–$75,000	16.9	16.0
>$75,000	36.2	30.4
Working for pay
Yes	46.1	49.0	.58
No	53.9	51.0
Census region
Northeast	15.7	13.8	.91
Midwest	26.4	25.0
South	37.7	38.5
West	20.2	22.7
Covered by government health insurance plan
Yes	53.3	49.9	.56
No	46.7	50.1
Have at least one private health insurance plan
Yes	63.7	62.2	.76
No	36.3	37.8
Have a usual place for health care/advice
Yes	92.4	86.1	.22
No	7.6	13.9
Need characteristics
BMI (kg/m2)
Normal or underweight (<25)	22.7	18.7	.45
Overweight (25–29.9)	29.4	27.0
Obese (≥30)	47.9	54.3
Smoking status
Current	13.4	16.3	.63
Former	35.5	33.3
Never	51.1	50.4
Comorbid medical conditions
None	15.4	18.5	.66
One	25.0	21.8
Two and more	59.6	59.7
ADL limitations
Yes	13.7	14.8	.61
No	86.3	85.2
IADL limitations
Yes	12.0	13.4	.47
No	88.0	86.6
CES-D scale
Elevated depressive symptoms (CES-D ≥3)	14.1	16.6	.32
Normal (CES-D <3)	85.9	83.4
Abbreviations: GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale.

Table 3. Crude associations between respondents’ predisposing, enabling, need characteristics and reporting breast cancer screening in past two years among women 50-74 years (HRS & HCNS, n =2,861 women, the United States, 2014-2016).

Respondent Characteristics	Total Study Sample Crude OR (95% CI)
Predisposing characteristics
Age in years	0.98 (0.97-1.01)
Race and Ethnicity
Non-Hispanic White	Reference
Non-Hispanic Black	1.23 (0.92-1.64)
Hispanic	1.00 (0.71-1.41)
Non-Hispanic others	0.83 (0.45-1.56)
Enabling characteristics
Level of education
Less than high school or GED	Reference
High-school graduate	1.26 (0.93-1.70)
Some college	1.44 (1.04-2.00)*
College and above	2.16 (1.54-3.02)***
Annual household income
<$20,000	Reference
$20,000–$35,000	1.21 (0.82-1.78)
$35,001–$50,000	1.64 (1.07-2.51)*
$50,001–$75,000	2.80 (2.05-3.83)***
>$75,000	2.50 (1.78-3.50)***
Working for pay
Yes	1.33 (1.05-1.68)*
No	Reference
Census region
Northeast	Reference
Midwest	0.65 (0.48-0.89)**
South	0.84 (0.63-1.13)
West	0.82 (0.59-1.15)
Covered by government health insurance plan
Yes	0.80 (0.63-1.02)
No	Reference
Have at least one private health insurance plan
Yes	2.02 (1.60-2.55)***
No	Reference
Have a usual place for health care/advice
Yes	2.14 (1.53-2.99)***
No	Reference
Need characteristics
BMI (kg/m²)
Normal or underweight (<25)	Reference
Overweight (25–29.9)	1.11 (0.83-1.48)
Obese (≥30)	1.12 (0.82-1.53)
Smoking status
Current	Reference
Former	2.43 (1.68-3.52)***
Never	2.41 (1.67-3.48)***
Comorbid medical conditions
None	Reference
One	1.02 (0.71-1.47)
Two and more	0.96 (0.68-1.37)
ADL limitations
Yes	0.46 (0.34-0.64)***
No	Reference
IADL limitations
Yes	0.49 (0.33-0.72)***
No	Reference
CES-D scale
Elevated depressive symptoms (CES-D ≥3)	0.55 (0.42-0.71)***
Normal (CES-D <3)	Reference
Abbreviations: OR, Odds Ratio; CI, Confidence Interval; GED, General Educational Development; BMI, Body Mass Index; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CES-D, Center for Epidemiologic Studies Depression Scale. Indicate P-value ≤.05, * indicate P-value <.01, *** indicate P-value <.001.

Table 4. Logistic regression results for the association between food security status and breast cancer screening among women of ages 50-74 in the United States (HRS & HCNS, n =2,861, 2014-2016).

	Receipt of mammogram or breast X-ray in past two years
	Model I	Model II^a	Model III^b
Respondent Characteristic	Crude OR (95% CI)	Adjusted OR (95% CI)	OR (95% CI)
Food security status
Food-insecure	0.55 (0.42-0.71)*	0.79 (0.56-1.12)	0.46 (0.30-0.70)*
Food-secure	Reference	Reference	Reference
Abbreviations: OR, Odds Ratio; CI, Confidence Interval.
^a The regression model was adjusted for all predisposing, enabling, and need characteristics of the respondents.
^b Propensity score weights were applied.
* ORs were statistically significant at P-values <.001.

Download PDF

Journal Publication

published 25 Jan, 2023

Read the published version in Cancer Causes & Control →

Reviews received at journal
29 Apr, 2021
Reviewers invited by journal
16 Feb, 2021
Editor invited by journal
15 Feb, 2021
First submitted to journal
25 Jan, 2021

You are reading this latest preprint version

Food security status and breast cancer screening among women in the United States: Evidence from Health and Retirement Study and Health Care and Nutrition Study

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Methods

Data Sources, Study Settings, and the Sample

[FIG 1 GOES HERE]

Main Study Measures

Statistical Analysis

Results

Discussion

Conclusion And Public Health Implications

Declarations

References

Tables

Status:

Journal Publication

Version 1