Food insecurity and chronic diseases in low-income older Americans: The role of SNAP receipt in medication underuse

Cardio-metabolic diseases such as diabetes and hypertension are among the leading cause of morbidity and mortality in the United States.1 About 63.1% of adults 60 years or older were living with the hypertension between 2011 and 2014; 25.2% of adults 65% or older are living with diabetes2,3. Less than half (49.4%) of adults who were 60 years or older living with hypertension had their blood pressure controlled (SBP/DBP <140/90 mm Hg) between 2015 and 2016.4 The economic burden of hypertension and diabetes is profound. It has been estimated that between 2003 and 2014, on average, the adjusted annual incremental cost of hypertension among US adults was $131 billion per year compared to their non-hypertensive counterparts5. In 2017, it was estimated that the direct and indirect cost of diabetes care was $327 billion6. These costs are exacerbated by medication non-adherence which results in avoidable hospitalizations7-9. Avoidable costs due to medication non-adherence for diabetes and hypertension was estimated to be $43.2 billion in 201310. Diabetes and hypertension are more prevalent among low-income and food insecure populations of older adults3,4,11. Further, higher prevalence of diabetes and hypertension has been reported among food insecure individuals12. Food insecure individuals living with diabetes and/or hypertension are also more likely to report sub-optimal outcomes such as sub-optimal glycemic or blood pressure control13,14 as well as increased likelihood of cost-related medication non-adherence15,16even after controlling for socio-demographic characteristics 17. Thus, older adults living with type 1 or 2 diabetes who are not adherent to medications are at risk for developing diabetic complications-such as hypoglycemia, hyperglycemic hyperosmolar state or diabetic ketoacidosis and those living with hypertension could develop hypertensive urgency or emergencies with acute target organ disease. These complications could result in emergency department visits or hospitalizations. There is growing recognition of what is being termed the eat or treat trade-off low-income households make between food and prescription drugs to treat chronic illnesses such as diabetes and hypertension. According to Berkowitz et al. (2014), among those who reported a chronic illness in the National Health Interview Survey, 23.4% of adults reported cost-related medication underuse, 18.8% reported food insecurity, and 11% reported both. Interestingly, multivariate results indicated that SNAP participation was associated with a higher risk of each negative outcome, suggesting that the population receiving SNAP is negative selected with respect to material hardship. More recent research using a propensity score matching technique found that SNAP reduced the rate of cost-related medication underuse for older adults (Srinivasan and Pooler 2017). Similar evidence of the higher prevalence of medication non-adherence among food insecure older adults has been documented both in state-based samples (Bengle et al. 2010; Sattler and Lee 2012) as well as in nationally representative data (Alfunai et al. 2015). Further, growing evidence suggests that food insecure older adults have higher healthcare utilization costs (Bhargava and Lee 2016a; Bhargava and Lee 2016b; Bhargava et al. 2012; Berkowitz et al. 2017). SNAP participation is likely to reduce the incidence of diabetic and hypertensive emergencies by improving diet quality. Diet quality among those with diabetes has been shown to be lower among low-income populations and those with food insecurity (Orr et al. 2019). Additionally, food insecurity was associated with inadequate disease control among those with both diabetes and hypertension (Seligman et al. 2010). While it is well documented that both the timing and size of household food expenditures are tied to SNAP benefits (Nord and Prell 2011; Tuttle 2016; Valizadeh and Smith 2019; Kim et al 2019), little research has directly connected the size and timing of SNAP benefits to healthcare utilization using administrative data (Heflin et al. 2016; Arteaga et al. 2018; Ojinnaka and Heflin 2018; Heflin et al. 2019; Heflin et al. 2020) and this research has not focused on the older adults, who are half as likely to participate in SNAP when eligible compared to prime aged adults--42% compared to 83% overall in 2015 (Gray et al. 2017). The current literature on medication non-adherence and food insecurity suffers from several significant limitations. First, most of the research documenting the treat or eat phenomenon uses survey data (Berkowitz et al. 2014; Alfunai et al. 2015; Srinivasan and Pooler 2017), which likely suffers from measurement error in the self-reports of chronic diseases, medication under-use and SNAP participation. Second, within the medical literature the standard approach to study medication non-adherence requires using administrative data containing pharmacy claims but this data lacks information on patient income and SNAP participation (Calip et al. 2017; Kim et al. 2018). 3) Finally, although older adults consume a growing level of medical resources and are comprising a larger share of the total population, most of the research to date has not focused on this age group (Herman et al. 2015). This study allows us to make a significant contribution to the field by combining administrative data on pharmacy claims, SNAP administrative data (containing both benefit level and exact disbursement data) for a census of the population age 60 and older from a Midwestern state from 2007-2014. Expected Results and Impact Results of the proposed study will make a significant research contribution to the field. The use of administrative data containing both prescription drug claims and SNAP benefit data will provide a good empirical test of the survey data reporting high levels of food insecurity and cost-related medication underutilization. Additionally, given the policy variation during our time period due to the AERRA benefit increase in the size of the SNAP benefit, it will be possible establish a causal relationship with the size of the SNAP benefit and the rate of medication adherence as well as the need for subsequent emergency care resulting from the medication non-adherence in either the ED or hospital setting. Additionally, given that Missouri disburses SNAP benefits over 22 days per month in manner that is nearly random, it will also be possible to causally identify if there is a time of the month in which SNAP disbursement reduces the prevalence of either medication non-adherence directly or the emergency complications requiring acute care in the ED or hospital setting for diabetes or hypertension. In addition, the study will yield policy-relevant findings about the connection between medication adherence for the high prevalence diseases of diabetes and hypertension, emergency healthcare utilization patterns for these diseases (both ED and hospitalizations), and SNAP participation. In addition to producing timely causal results linking SNAP benefit size and timing of disbursement to medication non-adherence and health emergencies requiring care in the ED or hospital setting, the study will also include a cost benefit analysis. This CBA will include the costs of medication non-adherence in terms of treatment in the ED and hospital against increases in SNAP benefits. Ultimately, as the Nation considers the rising cost of healthcare, this study will yield policy relevant information about the causal linkages between nutritional support and the health and healthcare utilization of older adults.