Neighborhood-level disadvantages increase risk for invasive pneumococcal disease

Abstract

Background:

Streptococcus pneumoniae (Spn) infection remains common worldwide despite recent vaccine efforts. Invasive pneumococcal disease (IPD) is the most severe form of Spn infection. Known individual risk factors for IPD include male gender and African American race. However, area-level socioeconomic factors have not been assessed. We examined the association of neighborhood-level disadvantages and risk of IPD in a tertiary medical center located in a socioeconomic diverse urban area in the Southeastern United States.

Methods:

Patients hospitalized with culture-confirmed Streptococcus pneumoniae (Spn) infection from 01/01/2010 – 12/31/2019 were identified from electronic health record (EHR). The cohort’s demographic and clinical information were obtained from EHR. Patients’ residential address was geocoded and matched to 2015 area deprivation index (ADI). The association of ADI and IPD was evaluated using logistic regression after controlling for the demographic information (age, sex, race) and clinical factors (BMI, smoking status, alcoholism, immunosuppressive status, vaccination status, comorbidities).

Results:

A total of 268 patients were hospitalized with culture-positive Streptococcus pneumoniae infection and 92 (34.3%) of them had IPD. The analysis showed that higher neighborhood deprivation (ADI in 79−100) was associated with increased risk of developing IPD in younger patients with age less than 65 (p = 0.007) after controlling for the individual demographic information and clinical factors.

Conclusions:

ADI is a risk factor for IPD in younger adults. Community-level socioeconomic risk factors should be consid- ered when developing prevention strategies such as increasing vaccine uptake in high risk population to reduce the disease burden of IPD.

INTRODUCTION

Streptococcus pneumoniae (Spn) is a gram-positive bacterium that can cause a range of infections in humans, such as pneumonia, sinusitis, otitis media, meningitis, and other illnesses. Spn remains the leading cause of community-acquired pneumonia (CAP) in developed countries, including the U.S., despite the existence of the 23-valent pneumococcal polysaccharide vaccine (PPSV23) and the 13-valent conjugated polysaccharide vaccine (PCV-13).¹ The most severe form of Spn infection is invasive pneumococcal disease (IPD), defined as the isolation of Spn from a normally sterile site such as blood or cerebrospinal fluid. IPD carries a high mortality rate, with a recent meta-analysis placing overall mortality at 20.8%.²

Studies have demonstrated that individual risk factors for Spn infection or IPD include advanced age, male gender, and African American race.^3–5 However, there is limited data on whether and how neighborhood-level disparities predict IPD risk, and existing studies have varied in breadth and quality. A few studies have explored the relationship between poverty levels and IPD. Still, their analysis has not considered other relevant socioeconomic determinants of health (SDOH), such as employment, home value, and education.^3,6 Another limitation of existing studies is a focus on the era before introducing PCV-13.^3,6,7 Thus, there is a need to reevaluate neighborhood-level risk factors and their ability to predict the risk of IPD in a post-PCV-13 era.

The area deprivation index (ADI) was developed to measure the socioeconomic status of a specific geographic area based on census data (census block group) and is used to identify relatively disadvantaged areas. It is a composite score of 17 SDOH in four major domains: education, income/employment, housing and household characteristics.⁸ The ADI is calculated on a national scale (valued from 1 to 100) and a state specific scale (valued from 1 to 9), with higher numbers indicating more disadvantages in both scales. ADI has been used to study outcomes in pulmonary fibrosis,⁹ national rehospitalization rates,¹⁰ and in the setting of other infectious diseases such as CMV and MRSA.^8,11

In this study, we aimed to evaluate if neighborhood-level socioeconomic disadvantages would affect the risk of developing IPD and we found that higher ADI in younger adults (<65 years old) was associated with a higher incidence of IPD after adjusting for race, gender, smoking status, alcoholism, pneumococcal vaccination status, immunosuppressive status, comorbidities and BMI. Given that younger populations are not universally recommended to receive pneumococcal vaccines, focused efforts on increasing vaccine uptake in at-risk population residing in socioeconomic disadvantaged areas should be considered to reduce IPD disease burden in this vulnerable demographic group. Furthermore, additional prospective larger studies are needed to validate the results from this study and to examine the causal relationship between ADI and IPD risk.

METHODS

Study design and participants

This was a cross-sectional observational study of patients hospitalized with culture-confirmed Streptococcus pneumoniae (Spn) infection at the University of Alabama at Birmingham Hospital from January 1, 2010 to December 31, 2019. Demographic, clinical and census block group data of the cohort were obtained from EHR. The study protocol (IRB-300,009,246) was approved by the Institutional Review Board of the University of Alabama at Birmingham.

Measures

The outcome measure was incidence of IPD. IPD-positive cases were defined as Spn isolated from a sterile site such as blood, pleural fluid, peritoneal fluid or cerebrospinal fluid (CSF), while IPD-negative status was defined as Spn isolated from sputum, tracheal aspiration or bronchoalveolar lavage (BAL).

Area deprivation was assessed using the 2015 national ADI. We matched each patient hospitalized with Spn with their national ADI values based on their census block group. Known individual demographic and clinical risk factors for IPD include male gender, African American race and smoking, and Spn infection overall is more common in the adult population older than 65 years and smokers. Interestingly, while obesity is associated with higher incidence of pneumonia, hospitalized obese patients have better survival.^3,4,12–17 Collectively, we included three demographic covariates (gender, age, race) and six clinical covariates (BMI, alcoholism, pneumococcal vaccination status, immunosuppressive status, comorbidities, and smoking status). Alcoholism is defined as more than 4 drinks a day. Patients who received at least one pneumococcal vaccine (either PSV23 or PCV13) is defined as vaccinated. Immunosuppressive status equaling to yes includes patients who have HIV infection/AIDS, chronic steroid usage which is larger than 20 mg prednisone per day, chronic usage of immunosuppressive medications such as TNF-α inhibitors or status post solid organ or hematopoietic stem cell transplantation, current solid organ or hematological malignancies, and patients who were receiving chemotherapies. To assess individual comorbidities, we use Charlson Comorbidity Index (CCI). The Charlson Comorbidity Index is a single comorbidity score for a patient summarizing 17 categorized and weighted comorbidities including congestive heart failure, chronic obstructive pulmonary disease, and peripheral vascular disease. A score of zero indicates that no comorbidities were found. The higher the score, the more likely the predicted outcome will result in mortality.¹⁸

We used national ADI because national ADI is scaled from 1100, which provides more granular socioeconomic information than the state ADI, which is scaled from 1 to 9. Additionally, although the cohort is based in the Southeast U.S., using the national ADI allowed for greater comparison and broader understanding of our findings. We first checked the linear relationship between the national ADIs and the logit of IPD status, and it showed that the linear association assumption was not appropriate. Hence, we grouped our patients into two disadvantage levels based on the national ADI distribution with similar patient number in each group: areas with ADI values from [2−79] (less disadvantaged) and (79−100] (more disadvantaged). Age was treated as categorical instead of continuous, because Spn infection overall is more common in adult population 65 and older. Moreover, routine pneumococcal vaccination is recommended for all patients 65 and older, regardless of risk factors, but is only recommended for patients 19−64 years of age who are immunocompromised or have other risk factors such as chronic renal disease or alcoholism.⁵ Because of the potential impact of age, vaccination and IPD, we grouped our patients into two age categories: younger (18−64) and older (65 and older). BMI was treated as categorical: underweight ≤ 18.5, normal weight = 18.6 − 24.9, overweight = 25 − 29.9, and obese ≥30. Race was categorized into three groups: non-Hispanic White, African American, and other (Hispanic, Asian, and multiple races). Charlson Comorbidity Index was treated as a continuous covariate.

Statistical analysis

The distribution of individual characteristics was obtained for the overall sample and by IPD status, as shown in Table 1. Bivariate analysis between IPD status and every covariate were conducted using Chi-square test or Fisher’s exact test, and effect sizes (Cramer’sV or Cohen’s d) and p-values are reported in Table 1. Multiple logistic regression was used to examine how ADI associates with IPD status after controlling for the demographic covariates (age, sex, race) and clinical factors (BMI, smoking status, alcoholism, pneumococcal vaccination status, immunosuppressive status and comorbidities). It was assumed that age group (18−64 vs 65 and older) may play a role in how ADI associates with IPD status; therefore, the interaction term of age and ADI group was included in the multiple logistic regression. Statistical tests were 2-sided and were performed using a 5% significance level (α = 0.05). Analyses were performed using R software, version 4.2.0.

Table 1.

Descriptive Statistics of the Study Population: overall and by IPD status (n = 268).

		IPD
	All Spn	Yes 92 (34.3 %)	No 176 (65.7 %)	Effect Size	p value
Age
Younger adult (18–64)	215 (80.2%)	68 (73.9%)	147 (83.5%)	V = 0.105	0.061
Older adult (≥65)	53 (19.8%)	24 (26.1%)	29 (16.5%)
Gender
Male	178 (66.4%)	58 (63.0%)	120 (68.2%)	V = 0.043	0.400
Female	90 (33.6%)	34 (37.0%)	56 (31.8%)
Race
Non-Hispanic White	154 (57.5%)	36 (39.1%)	118 (67.0%)
African American	102 (38.1%)	52 (56.5%)	50 (28.4%)	V = 0.278	<0.001
Others	12 (4.5%)	4 (4.3%)	8 (4.5%)
Smoking
Former	63 (23.5%)	26 (28.3%)	37 (21.0%)
Current	138 (51.5%)	32 (34.8%)	106 (60.2%)	V = 0.252	<0.001
Never	67 (25.0%)	34 (37.0%)	33 (18.8%)
Alcoholism
Yes	33 (12.3%)	11 (12.0%)	22 (12.5%)	V<0.001	0.898
No	235 (87.7%)	81 (88.0%)	154 (87.5%)
Immunosuppression
Yes	88 (32.8%)	48 (52.2%)	40 (22.7%)	V = 0.289	<0.001
No	180 (67.2%)	44 (47.8%)	136 (77.3%)
Vaccination
Yes	91 (34.0%)	47 (51.1%)	44 (25.0%)	V = 0.289	<0.001
No	177 (66.0%)	45 (48.9%)	132 (75.0%)
BMI
Underweight	22 (8.2%)	12 (13.0%)	10 (5.7%)
Normal	103 (38.4%)	39 (42.4%)	64 (36.4%)	V = 0.160	0.08
Overweight	65 (24.3%)	20 (21.7%)	45 (25.6%)
Obese	78 (29.1%)	21 (22.8%)	57 (32.4%)
ADI
[2–79]	136 (50.7%)	37 (27.2%)	99 (72.8%)	V = 0.144	0.013
(79–100]	132 (49.3%)	55 (41/7%)	77 (58.35)
CCI (Median, Range)	5 (0, 20)	7 (0, 20)	4 (0, 18)	d = 0.6	<0.001

CCI: Charlson Comorbidity Index; V: Cramer’s V; d: Cohen’s d.

RESULTS

The study sample included 268 hospitalized patients. The distribution of demographic and clinical characteristics for the overall Spn patients cohort is summarized in Table 1. In our sample, 80.2 % patients were under 65 years old with an average age of 51.7; 66.4 % were males; and 57.4 % were non-Hispanic White and 38.1 % were African Americans. 75.0 % were current or former smokers. 53.4 % were overweight or obese. 34.3 % of the overall Spn patients had IPD. The median Charlson Comorbidity Index is 5 in all Spn patients.

The interaction term of age and ADI group turned out to be significant in the multiple logistic regression. To explore further on how age affects the relationship between ADI and IPD risk, post-hoc analysis was done by pairwise comparisons by age group. The results were summarized in Table 2. In the younger adult group (18–64 years old), patients with higher national ADI (i.e., (79–100]) had higher risk of developing IPD compared to patients with lower national ADI (i.e., (2–79]) (odd ratio (OR)=2.856, 95% CI 1.340 – 6.084, p = 0.007) (Table 2). On the contrary, although there was a trend suggesting older patients (65 and older) with higher national ADI had lower risk of developing IPD, there was no statistical significance between the two ADI groups in terms of odds of IPD (Table 2).

Table 2.

Odds ratio estimates from the multiple logistic regression models in younger adults (18−64) and older adults (≥65).

Age Range	ADI Groups	Odds Ratio	95 % CI	p-value
18–64	(79,100] vs [2–79]	2.856	1.340–6.084	0.007
≥65	(79,100] vs [2–79]	0.344	0.091–1.304	0.116

DISCUSSION

We conducted a cross-sectional analysis of data from a tertiary medical center located in a socioeconomically diverse urban area in the Southeastern United States to examine the relationship between neighborhood-level socioeconomic status and incidence of IPD. Our findings showed that high neighborhood deprivation was a potential risk factor for IPD in younger adults (18–64). Young adults residing in deprived neighborhoods had higher risk of developing IPD, after adjusting for covariates including gender, race, BMI, alcoholism, pneumococcal vaccination status, immunosuppressive status, comorbidities, and smoking status. However, higher ADI is not associated with higher risk in older adults (≥65).

This study has several limitations. The cross-sectional design prevented us from making causal inferences about the observed relationship between area deprivation and IPD prevalence. Although we attempted to reduce the effects of relative geographic limitation of our study population by using the national ADI, results may not be generalizable to areas outside of the urban areas of the U.S. with high socioeconomic deprivation. As our population was exclusively composed of adults 18 and over, we were not able to draw any conclusions about socioeconomic status and IPD in children. Another limitation of this study was that we did not assess whether patients were co-infected with other pathogens such as influenza which is well-known for promoting the development of secondary bacterial infections and may increase the degree of Spn invasiveness.^19,20

We also did not know which serotypes of Spn our patient population was infected with and thus were not able to appreciate pathogen-specific factors relevant to the development of invasive disease. There are at least 100 serotypes of Spn, each of which is defined by the presence of distinct capsular polysaccharides. Capsule is a key Spn virulence determinant that prevents opsonization, phagocytosis, and complement inactivation.²¹ Different serotypes have varying degrees of invasiveness, antibiotic tolerance, and vaccine escape, all factors that would influence development of IPD in our population.^21–24

It is worth noting that the majority of our cohort with either Spn or IPD was male. This was unsurprising as male sex is a known risk factor for pneumococcal disease, even after adjusting for race.⁴ In fact, disparities between males and females in terms of IPD rates were found to be highest for children. The reasons for this disparity are unclear, though differences in smoking habits and the effects of sex hormones have been suggested as possible factors. Given the effects of ADI and male sex on increasing the risk for IPD, it may be wise to consider focusing more closely on improving vaccination uptake in young males living in socioeconomically disadvantaged areas.

Our study showed that ADI could be a clinically useful tool for identifying populations who are at risk of developing IPD and may benefit in developing preventive measurements such as increasing vaccination uptake strategies in high-risk population. While adults aged 18−64 do not have a universal pneumococcal vaccine indication, our study suggests those who live in substantially economically disadvantaged areas may benefit from screening for routine pneumococcal vaccination candidacy. Given the impact of ADI on IPD, we believe further actions to promote vaccination uptake in at-risk adults aged 18–64 and maintain these records in electronic medical records may contribute to advancing multiple national objectives defined by The United States Department of Health and Human Services in “Healthy People 2030.” These objectives include reducing disparities related to social determinants of health, increasing records of vaccination, maintaining existing vaccination practices, and increasing vaccine uptake in individuals over the age of 19.²⁵ In line with CDC recommendations for increasing COVID-19 vaccine uptake, integrating discussions of the pneumococcal vaccine with high-risk younger adults who present to clinic and partnering with community ambassadors to lower vaccine hesitance may help reduce the burden of IPD within high-risk communities such as those with a high area deprivation index.²⁶ Furthermore, additional prospective larger studies are needed to validate the results from this study and to examine the causal relationship between ADI and IPD risk.

CONCLUSIONS

Neighborhood deprivation is a potential risk factor for IPD, particularly in younger adults (18−64). In addition to individual demographic and clinical risk factors for IPD such as male gender, African American race and smoking, living in a socioeconomic-deprived neighborhood significantly increased the rate of IPD in younger adults. Neighborhood-level socioeconomic risk factors should be considered when developing prevention strategies such as increasing vaccine uptake to reduce the disease burden of IPD.