Area deprivation index predicts annual chronic kidney disease screening and chronic kidney disease development among patients with newly diagnosed hypertension and type 2 diabetes in a large midwestern health system: a retrospective cohort study

Introduction

Clinical practice guidelines recommend regular chronic kidney disease (CKD) screening after a diagnosis of hypertension (HTN) or type 2 diabetes (T2DM).1 2 Annual monitoring of kidney function increases early detection of CKD and improves quality of life.3 However, disparities in neighbourhood characteristics can impact access to routine care and poor health outcomes including uncontrolled blood pressure, haemoglobin A1c and cholesterol.4–7

Area deprivation index (ADI) was originally created by the Health Resources and Services Administration and is composed of education, employment, housing quality and poverty measures originally drawn from long-form Census data and updated periodically to incorporate more recent American Community Survey data.8 9 It allows for rankings of neighbourhood by socioeconomic disadvantage of interest at the state or national level. Understanding the relationship with neighbourhood deprivation with screening and CKD development allows for the development and implementation of interventions that can increase equity in screening and CKD development.

The literature has focused on evaluating the association with CKD diagnosis and mortality. In a study of patients with CKD (n=715) under nephrology care in Italy, ADI was not associated with CKD progression, yet increased deprivation was associated with all-cause mortality.6 However, studies evaluating preventative initiatives and screenings primarily focus on cancer screening.10–12 Unfortunately, studies evaluating area deprivation and CKD screening are limited despite affecting one in seven US adults and costing 100 billion annually.13 Therefore, the aim of this study is to evaluate the association of ADI, a national ranking of neighbourhood sociodemographic disadvantage with annual CKD screening and CKD development among patients with newly diagnosed HTN or T2DM.

Methods

Electronic health records (EHR) of medically homed (≥two primary care visits within 2 years during follow-up) patients in Illinois and Wisconsin (n=235 208) with a new HTN (n=1 54 056), T2DM (n=20 514) or both (n=60 638) diagnosis between 2015 and 2018 were extracted using the International Classification of Diseases, Tenth Revision (ICD-10) (online supplemental table 1) and followed for 3 years. Patients with a diagnosis of albuminuria, CKD or end-stage renal disease (ESRD) at the time of HTN or T2DM diagnosis were excluded. Annual CKD screening was defined as one estimated glomerular filtration rate (eGFR) and urinary albumin-to-creatine ratio within 12 months of HTN or T2DM diagnosis and CKD development included CKD and ESRD diagnoses. Patient demographics, comorbidities and laboratory values were merged with national ADI ranks (1–100). Ranks were divided into quintiles (Q1=least deprived, Q5=most deprived) for analysis.

Patients were stratified by diagnosis (HTN only, T2DM only and HTN and T2DM). Demographics, ADI, laboratory values, comorbidities, CKD screening and CKD diagnoses were calculated using medians, IQRs, frequencies or percentages. Multivariable logistic regression models adjusted for age, sex, smoking (never/ever), body mass index (kg/m²), year, Charlson Comorbidity Index14 and patient location (Illinois/Wisconsin) were used to evaluate the relationship between ADI with CKD screening. For CKD development models, CKD screening within 24 months (yes/no) was added to the adjusted models. All statistical analyses were carried out using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA) and R V.4.1.

Results

Cohort demographics, CKD screening and CKD development for the cohort are described elsewhere (in review). Briefly, of the 235 208 patients, 65% had HTN only (n=154 056), 9% had T2DM only (n=20 514) and 26% had HTN and T2DM (n=60 638). Most patients were white (57%) females (55%) with commercial insurance (82%). 27% of patients were black while 8% identified as Hispanic/Latino. 17% of patients were screened within 12 months of HTN and/or T2DM diagnosis, however, screening was highest for patients who developed HTN and T2DM during the study (44%) compared with only T2DM (38%) or HTN (4%). CKD developed for 9% of patients with HTN and T2DM, but 0% for only HTN or only T2DM. Overall, CKD screening rates for all patients was similar for 12 months (17%) and 24 months (18%). However, when stratified by diagnosis, screening rates for patients with both HTN and T2DM were 44% and 46% within 12 and 24 months, respectively. Screening was slightly lower for T2DM-only patients for 12 and 24 months (38% and 39%, respectively). Only 4% of HTN-only patients were screened for CKD at 12 and 24 months.

Associations of annual CKD screening and ADI quintiles among all patients and stratified by diagnosis (HTN only, T2DM only and HTN and T2DM) are depicted in figure 1. Among all patients, the odds of annual CKD screening increased by 54% (OR 1.54; 95% CI 1.48 to 1.60) for the most deprived patients (Q5) compared with the least deprived (Q1), yet among HTN-only patients, the odds of being screened increased to 84% (OR 1.84, 95% CI 1.62 to 1.99). However, among patients with T2DM or HTN and T2DM, associations were lower across all quintiles, especially with the odds of being screened for the most deprived patients (OR 1.20; 95% CI 1.13 to 1.27 and OR 1.13; 95% CI 1.02 to 1.28, respectively).

Request permission

Similar to CKD screening, the odds of developing CKD increased as deprivation increased (figure 2). In Q2–Q4, the odds of developing CKD roughly doubled as deprivation increased when compared with Q1 (Q2 OR 1.24, 95% CI 1.09 to 1.42; Q3 OR 1.45; 95% CI 1.28 to 1.64; Q4 OR 1.87; 95% CI 1.66 to 2.12, respectively). The odds of developing CKD within 3 years of an HTN and T2DM diagnosis among the most deprived patients were 146% greater compared with the least deprived (OR 2.46; 95% CI 2.19 to 2.76).

Request permission

Discussion

Although increased deprivation has been associated with poorer health outcomes including routine clinical care, it was interesting that among the most deprived patients with newly diagnosed HTN and T2DM in this cohort, the odds of being screened were 54% greater compared with the least deprived. Although our increase in the odds of screening were highest among the most deprived, it is possible the most deprived patients are sicker compared with patients in lower deprivation areas and thus are more likely to seek primary care on a more routine basis. More frequent health checks have been associated with increased offering of routine screening.15 Most patients in our cohort were commercially insured (82%), and annual CKD screening is covered under most healthcare plans. More frequent visits to primary care combined with insurance coverage of testing could account for the high screening rates. A study in the European Union observed higher cancer screening rates among patients at lower poverty and education levels, however these findings are likely due to national health insurance and cancer screening initiative programmes.10 It is also possible the single measure of ADI for patients extracted at the time of HTN and/or T2DM diagnosis did not reflect their ADI at the time of screening. ADI data were merged with patient data from EHR records. If patients fail to update their physical address or if there are errors with patient data, it is possible a patient’s true deprivation measurement was not reflected. However, despite patient mobility, addresses have been shown to reflect deprivation for up to 3 years.16

It was also interesting that once stratified by diagnosis, the odds for screening among patients with only T2DM or HTN and T2DM decrease among all quintiles, yet the decrease is greatest among the most deprived patients, still the odds of screening increased up to 84% for HTN-only patients. The large healthcare system where this cohort was derived underwent a merger during the cohort identification period of the study (2018), thus it is possible patients were diagnosed in one healthcare system and screened in another system, especially for the 65% subgroup of HTN-only patients that made up our cohort.

Higher deprivation was associated with an increased odds of developing CKD between 24% and 146% compared with the least deprived group. Although an increase in screening, as reflected in our results could have led to an increase in CKD diagnosis, we controlled for CKD screening in our final models, and therefore it is likely other factors are contributing to the increased rates to CKD among patients with HTN and T2DM. Patients in high deprivation areas are more likely to be of racial and ethnic minorities, experience poverty and limited built environment that exacerbate lifestyles (ie, obesity, diet and physical activity) that contribute to CKD development.17 18

This is one of the few studies evaluating associations between ADI and CKD screening. Although novel, our study is not without limitations. Development of CKD over the study period was based on ICD-10 codes rather than laboratory thresholds for glomerular filtration rate and urinary albumin-to-creatine ratios. While it is possible some patients may have been missing diagnosis codes and CKD prevalence was underestimated due to the limited follow-up time for CKD development after first screening, underdiagnosis rate is likely low. Automated EHR prompts clinicians to add the official diagnosis code and note scrubbers provide opportunities for appropriate documentation. Additionally, while a single capture of ADI among a transient population could have resulted in measurement error, it is likely this error is systematic across the entire cohort. We also acknowledge the high percentage of commercially insured patients limiting the generalisability of the study, yet we did use national ADI rankings as our primary outcome. Finally, of note, is the generalisability with more recent cohorts as clinical guidance shifted to an eGFR calculation excluding race during study follow-up.19 20 However, our results are still generalisable to cohorts of similar or prior time periods.

Conclusion

High deprivation in our cohort was associated with increased CKD screening and CKD development. While CKD development is likely due to upstream factors that predicate area deprivation and lifestyle factors associated with CKD, our finding of increased screening was surprising. Although it is plausible the most deprived patients in this cohort were more likely to be screened, our study highlights the need for systematic routine collection of social factors among patients within healthcare systems to allow for more accurate depictions of their associations with health outcomes.