Prevalence and Impact of Food Insecurity in Patients Undergoing Pulmonary Function Testing

Abstract

Background

Food insecurity (FI), the lack of consistent access to food for a healthy life, affected 13.5% of US households in 2023 and is associated with several chronic diseases. Limited data exist on the relationship between FI and lung health.

Research Question

In this cross-sectional study of adults presenting for pulmonary function testing (PFT), is FI associated with pulmonary function or respiratory symptom burden?

Study Design and Methods

We conducted a cross-sectional study of adults presenting for PFT between March 2018 and January 2023 in a large academic health network in North Carolina. FI was assessed using a validated 2-item questionnaire. Pulmonary function was measured using percent predicted FEV₁ and percent predicted FVC. Respiratory symptoms were assessed using the Modified Medical Research Council dyspnea scale and COPD Assessment Test. Multivariable models adjusting for sociodemographic factors and other covariates were used to evaluate associations between FI, pulmonary function, and symptoms.

Results

Of 10,805 patients screened, 17.8% reported FI. Patients with FI were more likely to be younger, female, non-Hispanic Black or Hispanic, to actively smoke, and to reside in socioeconomically disadvantaged areas. FI was associated with lower percent predicted FEV₁ (−2.55%) and percent predicted FVC (−1.04%), and increased odds of moderate to severe dyspnea (OR, 1.97). Among patients with COPD, FI was associated with higher symptom burden (mean COPD Assessment Test score, 23 vs 17) and more frequently reported moderate or severe dyspnea (67.5% vs 51.2%), with similar findings observed in patients with asthma.

Interpretation

Our findings suggest that FI is associated with worse pulmonary function and higher respiratory symptom burden in patients undergoing PFT, highlighting the importance of addressing FI as a social determinant of respiratory health. Further research is needed to explore mechanisms and potential interventions to mitigate the impact of FI on lung health.

Take-Home Points.

Study Question: Is food insecurity associated with differences in lung function or respiratory symptom burden in patients presenting for pulmonary function testing?

Results: In this cross-sectional analysis of patients presenting for pulmonary function testing, the prevalence of food insecurity was found to be associated with lower lung function and increased burden of respiratory symptoms while disproportionately affecting patients who were younger, female, non-Hispanic Black or Hispanic, and residing in socioeconomically disadvantaged areas.

Interpretation: These data support our hypothesis that food insecurity is associated with worse pulmonary function and higher respiratory symptom burden in patients undergoing pulmonary function testing.

Food insecurity (FI) is the lack of consistent access to the food needed for an active, healthy life.¹ FI impacted 13.5% of households in the United States, or > 35 million people, in 2023, particularly people with lower incomes and who have been racially and ethnically minoritized.^1,2 Similarly, an average of nearly 11% of North Carolinians were food insecure from 2023. FI is associated with higher rates of numerous chronic medical conditions, including diabetes, dementia, cardiovascular disease, and hypertension, and with worse health outcomes in people living with HIV and surgical outcomes in some cancers.3, 4, 5, 6, 7, 8, 9

FI likely also negatively impacts the health of people living with chronic lung diseases; however, data in this area are limited.¹⁰ Prior studies have shown that household FI is associated with a diagnosis of asthma in adults living in Korea as well numerous studies in children in the United States; however, there were no measurements of pulmonary function in these studies, and whether this can be extrapolated to an adult population in the United States is unclear.11, 12, 13, 14 Finally, FI has previously been shown to correlate with a self-reported diagnosis of both asthma and COPD, with increasing severity of FI strengthening this association; however, there was no measurement of pulmonary function in this study.¹⁵

Given the high prevalence of FI in North Carolina and its association with worse outcomes in chronic diseases, we implemented a FI questionnaire into the introductory screening process for patients presenting for testing in our pulmonary function laboratory. Our objective was to determine the prevalence of FI in patients presenting for pulmonary function testing (PFT) in a large, academic health network, and to determine if FI is associated with pulmonary function and respiratory symptom burden.

Study Design and Methods

We conducted a cross-sectional study of adult patients (≥ 18 years of age) presenting for PFT at 1 of 2 hospitals within the Atrium Health Wake Forest Baptist Network between March 2018 and January 2023. The hospital systems included in the study were Atrium Health Wake Forest Baptist Hospital, an urban, academic medical center and health system located in Winston-Salem, North Carolina, and Wake Forest Baptist Health Lexington Medical Center, a rural, community hospital located in Lexington, North Carolina. In 2018, systematic screening for FI using a validated, 2-item questionnaire (Fig 1) was implemented for all patients presenting for PFT within these hospital systems.¹⁶ This study was approved by the Wake Forest Baptist institutional review board (No. IRB00053117).

Figure 1.

Food insecurity questionnaire.

We included all adult patients who presented for PFT at 1 of these 2 sites and completed the FI questionnaire (Fig 2). For patients with multiple pulmonary function tests during the study period, we used only the most recently available results for analysis. We reviewed the electronic health record (EHR) for all problem lists, encounter diagnoses, and billing diagnoses for any inpatient or outpatient visit during the study period. A diagnosis of COPD was defined using the International Classification of Diseases, 10th Revision (ICD-10) codes J41-44.¹⁷ A diagnosis of asthma was defined using ICD-10 code J45. A prespecified subgroup of patients with obstruction on PFT was also defined using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) definition as having an FEV₁ to FVC ratio < 0.70.¹⁸

Figure 2.

CONSORT Diagram. GOLD = Global Initiative for Chronic Obstructive Lung Disease; PFT = pulmonary function testing; ICD-10 = International Classification of Diseases, 10th Revision.

FI Screening

The validated, 2-item FI questionnaire was included in the standard paper intake form that all patients presenting for PFT completed in the waiting room.¹⁶ Their responses to these questions were then documented by clinic staff in the EHR. Based on the standard scoring, patients who responded sometimes or often to either question were considered food insecure. Patients who answered at least 1 of the 2 screening questions were included in the analysis. Patients who did not complete the questionnaire or who chose do not know to both questions were excluded. Our primary exposure of interest for this study was FI (yes or no).

Study Outcomes

Our primary outcome was pulmonary function as measured by percent predicted FEV₁ (ppFEV1). Secondary outcomes included percent predicted FVC (ppFVC), FEV₁/FVC ratio, and symptom scoring systems obtained when the patient presented for PFT (Modified Medical Research Council [mMRC]; score range, 0 to 4; and COPD Assessment Test [CAT]; score range, 0-40).^19,20 Based on prior studies, we categorized mMRC as no or mild dyspnea (0-1) or moderate to severe dyspnea (≥ 2) because this dichotomy correlates with physical activities of daily living and has been adopted by GOLD for classification and management of COPD.^18,21,22

Covariates

Because sociodemographic characteristics may confound the relationship between FI and pulmonary function, we extracted data on several covariates from the EHR. Covariates were chosen based on current literature¹ and included the following: age, race/ethnicity, sex, smoking status (active, prior, or never), BMI, and insurance status (private/commercial, Medicaid/other public insurance, Medicare, other, or uninsured). The social categories of race and ethnicity were included as covariates because they may indicate exposure to racism, which can worsen FI.¹ Because of small sample sizes in individual categories, race/ethnicity was categorized as non-Hispanic White, non-Hispanic Black, Hispanic, or non-Hispanic other. Non-Hispanic other race/ethnicity included American Indian or Alaska Native, Asian Indian, Chinese, Filipino, Native Hawaiian, other Asian, patient refused, and unknown. We also included the 2023 Area Deprivation Index (ADI) based on the address of residence. The ADI is an established indicator of socioeconomic disadvantage that aggregates estimates from the American Community Survey across domains of income, education, and health care access (range from 0 to 100, with 100 representing the worst socioeconomic disadvantage).^23,24 Individual patient addresses listed in the EHR were geocoded at the Census tract level and linked to their respective ADI. We categorized ADI in quintiles from least socioeconomically disadvantaged to most socioeconomically disadvantaged.

Data Analysis

In bivariate analysis, we used t tests to evaluate associations between continuous variables and FI, and χ² tests to examine the associations between categorical variables and FI. The association between FEV₁ and FI was examined using multivariable generalized linear models. We also used generalized linear models to evaluate the relationship between FI and FEV₁/FVC ratio and CAT score. We used multivariable logistic regression to evaluate the association between mMRC and FI. All multivariable models were adjusted for age, race/ethnicity, sex, smoking status, BMI, insurance status, and ADI. We also conducted secondary analyses evaluating the association between FI and our outcomes stratified by diagnosis (COPD or asthma). All models assessing patients with COPD were adjusted for asthma, and all models assessing patients with a diagnosis of asthma were adjusted for COPD. In all multivariable models, < 10% of patients were missing data; therefore, we used listwise deletion to account for missing data.²⁵

For the primary outcome, COPD was defined using ICD-10 codes as previously defined; this was chosen with the intent to be inclusive of all GOLD stages of COPD, including unclassified or pre-COPD.^18,21 A preplanned secondary analysis of COPD was performed using GOLD defined obstruction, given the well-documented high burden of undiagnosed COPD and the limitations of using International Classification of Diseases codes to define disease.^23,26 Dyspnea was assessed using mMRC scores for all patients, whereas overall symptom burden in patients with COPD was assessed using CAT scores. Moderate to severe dyspnea was defined as an mMRC score ≥ 2. All analyses were conducted using R Software (R Foundation for Statistical Computing, version 4.2.3).

Results

Demographics

A total of 18,011 patients presented for PFT during our study period; of these, 10,805 (60%) completed at least 1 FI screening question and were included. Of the 10,805 who completed FI screening, 1,924 (17.8%) screened positive for FI. Patients with FI were more likely to be younger age (mean age ± SD, 57 ± 12 vs 62 ± 15 years; P < .001), female (OR, 1.30; 95% CI, 1.18-1.44), and non-Hispanic Black (OR, 2.45; 95% CI, 2.19-2.73) or Hispanic (OR, 2.04; 95% CI, 1.53-2.69) (Table 1). Patients with FI had higher odds of actively smoking (OR, 3.09; 95% CI, 2.74-3.48). Patients with FI were more likely to reside in areas with higher mean ADI (mean ± SD, 73 ± 18 vs 63 ± 20; P < .001).

Table 1.

Population Characteristics Across Categories of Food Insecurity

Characteristic	Total (N = 10,805)	Food Secure (n = 8,881)	Food Insecure (n = 1,924)	P Value
Age, y	61 [14]	62 [15]	57 [12]	< .001
Female	6,031 (55.8)	4,856 (54.7)	1,175 (61.1)	< .001
Race and ethnicity				< .001
Non-Hispanic, White	8,161 (75.5)	6,978 (78.6)	1183 (61.5)
Non-Hispanic, Black	2,153 (19.9)	1,522 (17.1)	631 (32.8)
Hispanic	265 (2.5)	197 (2.2)	68 (3.5)
Non-Hispanic, other	226 (2.1)	184 (2.1)	42 (2.2)
BMI, kg/m2				< .001
Mean [SD]	31 [8.2]	30 [8.0]	32 [9.0]
Missing	233 (2.2)	191 (2.2)	42 (2.2)
Smoking history				< .001
Active	2,271 (21.0)	1,471 (16.6)	800 (41.6)
Prior	4,491 (41.6)	3,818 (43.0)	673 (35.0)
Never smoked	3,803 (35.2)	3,396 (38.2)	407 (21.2)
Unknown	240 (2.2)	196 (2.2)	44 (2.3)
Insurance				< .001
Private/commercial	3,476 (32.2)	3,098 (34.9)	378 (19.6)
Medicaid/other public	1,145 (10.6)	700 (7.9)	445 (23.1)
Medicare	5,892 (54.5)	4,908 (55.3)	984 (51.1)
Other	53 (0.5)	46 (0.5)	7 (0.4)
Uninsured	239 (2.2)	129 (1.5)	110 (5.7)
Asthma (ICD-10)	3,509 (32.5)	2,741 (30.9)	768 (39.9)	< .001
COPD (ICD-10)	5,109 (47.3)	3,909 (44.0)	1,200 (62.4)	< .001
COPD (GOLD)a	3,040 (28.1)	2,400 (27.0)	640 (33.3)	< .001
Area Deprivation Index				< .001
Mean [SD]	64 [20]	63 [20]	73 [18]
Missing	1,086 (10.1)	935 (10.5)	151 (7.8)
ppFEV1	79 [22]	80 [22]	75 [22]	< .001
ppFVC	84 [19]	85 [19]	82 [18]	< .001
FEV1/FVC (actual)	73 [13]	74 [13]	72 [14]	< .001
CAT				< .001
Mean [SD]	16 [8.6]	15 [8.2]	21 [8.7]
Missing	596 (5.5)	483 (5.4)	113 (5.9)
mMRC				< .001
0-1	5,259 (48.7)	4,632 (52.2)	627 (32.6)
≥ 2b	4,714 (43.6)	3,565 (40.1)	1,149 (59.7)
Missing	832 (7.7)	684 (7.7)	148 (7.7)

Data are presented as mean [SD], No. (%), or as otherwise indicated. CAT = COPD Assessment Test; GOLD = Global Initiative for Chronic Obstructive Lung Disease; ICD-10 = International Classification of Diseases, 10th Revision; mMRC = Modified Medical Research Council; ppFEV₁ = percent predicted FEV₁; ppFVC = percent predicted FVC.

^aCOPD (GOLD) is defined as those having an FEV₁/FVC < 70.

^bmMRC ≥ 2 is defined as having moderate to severe dyspnea.

Pulmonary Function

In bivariate analysis, we found that patients who reported FI, compared with those that did not, had significantly lower ppFEV₁ (mean ± SD, 75 ± 22 vs 80 ± 22; P < .001) and lower ppFVC (mean ± SD, 82 ± 18 vs 85 ± 19; P < .001). We also found that patients with FI were more likely to have moderate to severe dyspnea than those that were food secure (OR, 2.38; 95% CI, 2.14-2.65). In multivariable models, individuals who reported FI had a lower ppFEV₁ (OR, −2.55; 95% CI, −3.69 to −1.41) and ppFVC (OR, −1.04; 95% CI, −2.02 to −0.07) and increased odds of moderate to severe dyspnea (OR, 1.97; 95% CI, 1.75-2.22) compared with those who did not report FI (Table 2).

Table 2.

Association of Food Insecurity on Pulmonary Function and Symptom Burden^a

Lung Function or Symptom Assessment	All (N = 10,805)	COPD (GOLD) (n = 3,040)	COPD (ICD-10) (n = 2,109)	Asthma (ICD-10) (n = 3,509)
ppFEV1	−2.55a (−3.69 to −1.41)	0.53 (−1.37 to 2.44)	0.33 (−1.21 to 1.88)	1.09 (−0.64 to 2.83)
ppFVC	−1.04a (−2.02 to −0.07)	0.90 (−0.95 to 2.76)	0.80 (−0.51 to 2.12)	1.12 (−0.40 to 2.64)
FEV1/FVC (actual)	−1.58a (−2.21 to −0.96)	0.04 (−0.99 to 1.07)	−0.20 (−1.15 to 0.75)	−0.13 (−1.09 to 0.83)
CAT	4.43a (3.99 to 4.87)	4.03a (3.24 to 4.84)	4.25a (3.67 to 4.84)	NA
mMRC (≥ 2), β (OR)	1.97a (1.75 to 2.22)	1.73a (1.39 to 2.14)	1.91a (1.62 to 2.25)	1.71a (1.40 to 2.09)

Data are presented as β (95% CI) or as otherwise indicated. All models looked at food insecurity adjusting for payor status, smoking status, race/ethnicity, sex, BMI, age, and national Area Deprivation Index quintiles. All COPD models also adjusted for asthma, and the asthma model adjusted for COPD. CAT = COPD Assessment Test; GOLD = Global Initiative for Chronic Obstructive Lung Disease; mMRC = Modified Medical Research Council; ppFEV₁ = percent predicted FEV₁; ppFVC = percent predicted FVC.

^aP < .05.

COPD

Of the 5,109 patients with a clinical diagnosis of COPD as defined by ICD-10 codes (clinical COPD), 1,200 (23.5%) screened positive for FI (Table 3). When assessing patients with GOLD defined obstruction on PFT, 21.1% reported having FI (Table 4). In patients with clinical COPD, we did not find a significant association between FI and ppFEV₁ (68 vs 69, P = .171) or FI and ppFVC (79 vs 80, P = .244). Compared with patients who did not report FI, patients with FI reported a higher respiratory symptom burden (mean CAT, 23 vs 17; P < .001) and more frequently reported moderate or severe dyspnea by mMRC (67.5% vs 51.2%, P < .001). When evaluating only patients with obstruction on PFT, patients with FI had lower ppFEV₁ (58 vs 61, P = .003) but no significant difference in ppFVC (81 vs 82, P = .123), reported higher symptom burden (mean CAT, 22 vs 17; P < .001), and more frequently reported moderate to severe dyspnea (64.4% vs 48.1%, P < .001). When adjusting for covariates, FI was associated with increased symptom burden, including higher CAT scores and mMRC, but not with lung function.

Table 3.

Pulmonary Function and Symptom Burden of Patients With COPD as Defined by ICD-10 codes (Clinical COPD)

	All (n = 5,109)	Food Security (n = 3,909)	Food Insecurity (n = 1,200)	P Value
ppFEV1	69 [22]	69 [23]	68 [21]	.171
ppFVC	80 [19]	80 [19]	79 [19]	.244
FEV1/FVC (actual)	67 [15]	67 [15]	68 [15]	.063
CAT	18 [8.7]	17 [8.3]	23 [8.5]	< .001
MMRC (≥ 2)	2,810 (55.0)	2,000 (51.2)	810 (67.5)	< .001

Data are presented as mean [SD], No. (%), or as otherwise indicated. CAT = COPD Assessment Test; ICD-10 = International Classification of Diseases, 10th Revision; mMRC = Modified Medical Research Council; ppFEV₁ = percent predicted FEV₁; ppFVC = percent predicted FVC.

Table 4.

Pulmonary Function and Symptom Burden of Patients With GOLD Defined Obstruction on PFT (FEV₁/FVC < 0.70)

	All (n = 3,040)	Food Security (n = 2,400)	Food Insecurity (n = 640)	P Value
ppFEV1	60 [21]	61 [21]	58 [20]	.003
ppFVC	82 [20]	82 [20]	81 [19]	.123
FEV1/FVC (actual)	56 [12]	56 [11]	56 [12]	.359
CAT	18 [8.8]	17 [8.4]	22 [8.8]	< .001
MMRC (≥ 2)	1,567 (51.5)	1,155 (48.1)	421 (64.4)	< .001

Data are presented as mean [SD], No. (%), or as otherwise indicated. CAT = COPD Assessment Test; GOLD = Global Initiative for Chronic Obstructive Lung Disease; mMRC = Modified Medical Research Council; PFT = pulmonary function testing; ppFEV₁ = percent predicted FEV₁; ppFVC = percent predicted FVC.

Asthma

Among the 3,509 patients with a clinical diagnosis of asthma, 768 (21.9%) reported FI (Table 5). In patients with a diagnosis of asthma, FI was associated with a lower mean ppFEV₁ (75 vs 79, P < .001), lower mean ppFVC (81 vs 84, P < .001), and increased frequency of moderate to severe dyspnea (66.1% vs 46.8%, P < .001). When adjusting for covariates, FI was associated with increased dyspnea but not with pulmonary function.

Table 5.

Pulmonary Function and Symptom Burden of Patients With Asthma as Defined by ICD-10 Codes

	All (n = 3,509)	Food Security (n = 2,741)	Food Insecurity (n = 768)	P Value
ppFEV1	78 [22]	79 [22]	75 [22]	< .001
ppFVC	83 [19]	84 [19]	81 [18]	< .001
FEV1/FVC (actual)	74 [13]	74 [13]	73 [14]	.210
MMRC (≥ 2)	1,792 (51.1)	1,284 (46.8)	508 (66.1)	< .001

Data are presented as mean [SD], No. (%), or as otherwise indicated. ICD-10 = International Classification of Diseases, 10th Revision; mMRC = Modified Medical Research Council; PFT = pulmonary function testing; ppFEV₁ = percent predicted FEV₁; ppFVC = percent predicted FVC.

Discussion

In this cross-sectional study, we found that FI is prevalent among patients presenting for PFT, occurring in 17.8% of the patients, higher than both the national (13.5%) and statewide (North Carolina, 11%) averages.¹ Patients with FI were more likely to be younger, female, non-Hispanic Black, actively smoking, and residing in socioeconomically disadvantaged areas. Rates of FI were even higher in patients with diagnosed lung disease; nearly 1 in 4 patients with COPD and > 1 in 5 patients with asthma reported being food insecure. A similar prevalence of FI at 13.8% was found in a previous study by de Castro Mendes et al²⁴ using national data, further contrasting the higher prevalence in the present cohort. There are many potential reasons for the high prevalence in our health system. First, our health system may serve a lower socioeconomic population than the US average, as was demonstrated in a study commissioned locally by our county.²⁷ This study also estimated a high prevalence of FI in our city overall at 17% in 2014, and 21 existing food deserts in 2018, leading to increased barriers in access to food in our community. Additionally, this cohort was established by surveying patients presenting for PFT, presumably as part of an evaluation of underlying respiratory symptoms. Because dyspnea and respiratory symptoms appear to be worse in patients with FI, it is possible that we selected for a population at higher risk of FI than the general population. This is a notable finding because it shows that patients presenting for PFT may represent an important population to target for interventions to address FI. Conversely, it is also possible that we may be underestimating the prevalence of FI in patients with chronic respiratory symptoms by specifically screening in the PFT laboratory; individuals from food insecure households are less likely to present for preventative care visits because of cost. Therefore, we may be missing many patients who are not presenting for testing.^28,29

FI and other health-related social needs are associated with many chronic diseases and worse health outcomes.^3,4 To add to this growing literature, we found that FI was also associated with lower pulmonary function and higher respiratory symptom burden. We found that FI was associated with a 2.5% lower ppFEV₁ and a 1% lower ppFVC. These findings are consistent with other studies that have demonstrated a relationship between FI and worse lung function. For example, research conducted by de Castro Mendes et al²⁴ showed that households experiencing high FI exhibit worse lung function, particularly when measured using FVC, and higher odds of spirometric restriction, with a consistent effect observed across various ethnic groups. Notably, although unadjusted analyses showed lower ppFEV₁ and ppFVC in patients with asthma and lower ppFEV₁ in patients with GOLD-defined obstruction, these results did not persist after adjusting for other confounders. Interestingly, however, although there was no statistical difference in patients with COPD or asthma in our adjusted model, worse lung function was noted by both ppFEV₁ and ppFVC in the total population, driven entirely by those without diagnoses of asthma or COPD. The reasons behind this finding are not clear, and more studies are needed to better understand this relationship.

Patients in our study with FI also reported worsened dyspnea, including being nearly twice as likely to report moderate or severe dyspnea by mMRC. In patients with COPD, FI was associated with a higher burden of respiratory symptoms, including both higher CAT and mMRC scores. Belz et al³⁰ also showed that patients with FI demonstrate worse COPD health status by CAT scores, dyspnea, and respiratory-specific quality of life. CAT scores have been shown to correlate with risk of acute exacerbations of COPD and respiratory health-related quality of life; therefore, the 6-point difference in CAT score between patients with FI and food security seen in our study is likely to be clinically significant.^31,32 In patients with asthma, FI was also associated with higher dyspnea scores; this finding is consistent with results from Grande et al,³³ who found that food insecure adults were much more likely to have uncontrolled asthma when compared with those with lower FI, even when controlling for other known confounders.

There are likely multiple potential mechanisms to explain the relationship between FI and respiratory function. FI has been shown to correlate with obesity in several studies, and it is well established that obesity affects both respiratory physiology as measured by PFT, particularly with spirometric restriction, and respiratory symptoms.34, 35, 36, 37 Indeed, the mean BMI in our study was higher in patients with FI. Notably, however, changes in lung function and respiratory symptoms persisted even after adjusting for BMI, indicating that this is likely not the only factor at play. BMI and body composition are not the same; therefore, it is possible that adiposity, which also has been shown to correlate with FI, may contribute to symptoms even when accounting for BMI.^38,39

Stress may also be a potential contributing factor to lower lung function. Several studies have found that patients with FI, particularly those with COPD, report higher perceived stress and depression.^30,40 Separately, a study by Du et al⁴¹ showed that patients with comorbid depression and COPD had increased sputum levels of IL-1 and tumor necrosis factor-alpha, and decreased diurnal variation of salivary cortisol. Further studies are needed, but this may point to potential mechanisms for how chronic stress and depression could promote prolonged activation of proinflammatory cytokines in patients with FI that could lead to increased airway inflammation and decreased lung function. As in our study, FI has been found in several other studies to be associated with smoking.⁴² The relationship between tobacco use and FI is likely bidirectional. FI is associated with increased stress, and many people use tobacco to cope with stress, thus making tobacco cessation more difficult. Individuals living in food-insecure households may use nicotine as an appetite suppressant. The costs of cigarettes, and the subsequent health care expenditures from the development of chronic conditions from tobacco use, may also reduce the overall household budget, making households more likely to be food insecure. Further studies combining tobacco cessation programs with food assistance could be needed.⁴³

Because of financial burdens, patients with FI may also be forced to decide between purchasing food or their prescription medications (eg, inhalers). Becerra et al⁴⁴ demonstrated that patients with FI and asthma were 148% more likely to report a delay in obtaining their inhaler prescription. This highlights the importance of screening to identify patients with FI so that additional support may be provided in accessing both food and their medications, which may in turn improve their overall lung health.

Additionally, FI may correlate with specific dietary deficiencies that lead to worse respiratory health. Studies have implicated specific food sources that could be involved. Specifically, Shaheen et al⁴⁵ showed that a cohort in Hertfordshire, England, with higher intake of fruits, vegetables, fish, and whole grain cereals was associated with a better FEV₁ and lower prevalence of COPD. Furthermore, there have been studies suggesting that patients with a diet rich in dietary fiber had the highest lung function.⁴⁶ Whether these findings represent causation or correlation is unclear; however, it does highlight potential avenues for further investigation.

There are several limitations to our study that should be acknowledged. First, this is a cross-sectional study and cannot determine a causal association between FI and pulmonary outcomes. Second, this sample may not be representative of the general population because it consists of patients from a single health system and only those presenting for PFT. Future studies should consider a broader and more diverse population to generalize findings. Third, it is possible that missing data affected our outcomes; however, the number of patients with missing data in this cohort was small. Notably, however, nearly 45% of those presenting for PFT selected do not know or declined to answer food security questions, limiting the accuracy and interpretation of our results. Notably, patients that declined to answer the food security screening questions were more likely to be male (52.7% vs 44.2%), White (77.9% vs 75.5%), and have Medicaid insurance (15.4% vs 10.6%) and less likely to have never smoked (30.7% vs 35.2%). When comparing PFT data between patients that completed the screening questions and those that did not, patients who completed the screening had a higher mean ppFEV₁ (79% vs 74%) and ppFVC (84% vs 80%), whereas FEV₁/FVC was similar between the 2 groups (73% vs 72%) in unadjusted comparisons. How these differences may have impacted the results of our study and the reasons for these discrepancies are not immediately clear, and future studies should try to further assess these questions. Finally, the accuracy of using International Classification of Diseases codes to identify respiratory diseases, and COPD in particular, is limited, and it is possible that this may have misclassified patients in our study.^47,48 To mitigate this risk, we evaluated patients using both ICD-10 codes and actual obstruction on PFT with similar outcomes.

Interpretation

This study demonstrates a significant association between FI and both lower pulmonary function and higher symptom burden in patients presenting for PFT. Implementing FI screening could help identify patients at risk for worse pulmonary function and symptom burden, and addressing FI could be a vital component in managing and improving respiratory health, particularly among vulnerable populations. Our study highlights an important social determinant for lung health. Further prospective studies are needed to determine potential pathophysiological mechanisms that may be implicated, and longitudinal studies would clarify if improvement in food security can subsequently improve lung function and symptom burden. Health care systems and policymakers should work together to identify and support individuals with FI to enhance their health outcomes and quality of life.

Funding/Support

D. P. is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health [Award K23HL146902]. J. A. P. is supported by the National Institute on Aging [Award K23AG073529]. T. W. L. is supported by the National Center for Advancing Translational Sciences of the National Institutes of Health [Award K12TR004931].

Financial/Nonfinancial Disclosures

D. P. reports personal fees from WellCare of North Carolina, outside of the submitted work. None declared (D. L. N., L. W., C. S. O., S. E. M., S. K. R., R. B., A. M., T. W. L., J. A. P.).