Discharge-Based Recovery After Dialysis-Requiring Acute Kidney Injury: A National U.S. Analysis of Sociodemographic and Hospital Factors

Introduction

Acute kidney injury (AKI) is a common and serious complication of hospitalization, affecting millions of patients annually and contributing substantially to morbidity, mortality, and healthcare utilization worldwide.^1–3 Among hospitalized patients, dialysis-requiring AKI (AKI-D) represents the most severe end of the disease spectrum and is associated with particularly high short-term risk.⁴ Advances in critical care and supportive therapies have improved survival among patients with AKI-D, shifting the clinical focus from in-hospital mortality alone toward longer-term outcomes.⁵ As survival improves, understanding what happens to patients after the acute phase of illness has emerged as a central challenge in AKI care.

Recovery following AKI-D is increasingly recognized as a heterogeneous and clinically consequential process rather than a binary outcome. Many survivors experience incomplete renal recovery, prolonged dependence on renal replacement therapy, or significant functional decline after hospitalization.^6,7 These post-acute outcomes are associated with rehospitalization, reduced quality of life, and progression to chronic kidney disease.^8,9 However, recovery trajectories are not routinely captured in administrative or clinical datasets, and standardized definitions of recovery remain variable across studies. As a result, the post-discharge phase of AKI-D has been comparatively understudied despite its importance to patients and health systems.

A growing body of literature has documented substantial sociodemographic disparities in the incidence and mortality of AKI, with differences observed by age, race and ethnicity, insurance status, and socioeconomic context.^10,11 These disparities are often attributed to a combination of baseline health differences, access to care, and structural factors that shape clinical decision-making.¹² Far less is known, however, about whether similar disparities persist during recovery after AKI-D, particularly among patients who survive the initial hospitalization. Understanding recovery through the lens of structural vulnerability is critical, as post-acute outcomes may be especially sensitive to differences in resources, care transitions, and institutional practices that extend beyond the acute illness itself.

Beyond patient-level characteristics, recovery after AKI-D may be shaped by the institutional environments in which care is delivered. Hospital characteristics such as teaching status and bed size may influence recovery through differences in nephrology availability, dialysis discontinuation practices, discharge planning infrastructure, and access to affiliated post-acute care facilities.^13,14 Prior work has demonstrated substantial hospital-level variation in the use of high-intensity interventions and specialty care, even after accounting for patient severity.¹⁵ These differences raise the possibility that hospital context may modify recovery trajectories following AKI-D, either by mitigating vulnerability through coordinated transitions of care or by amplifying disparities when resources are limited. Despite this plausibility, national evidence examining how hospital characteristics influence recovery after AKI-D remains limited.

In this study, we used a nationally representative inpatient dataset to examine recovery trajectories among adults who survived dialysis-requiring acute kidney injury. We focused on discharge-based recovery as a pragmatic indicator of post-acute burden and evaluated differences across sociodemographic groups. In addition, we assessed whether hospital characteristics modified recovery patterns, with particular attention to institutional contexts in which disparities were amplified or attenuated. By integrating patient-level vulnerability with hospital-level context and presenting results using absolute risk metrics, this study aims to advance understanding of recovery after AKI-D and to identify potential system-level leverage points for improving equitable post-acute outcomes.

Methods

Data Source and Study Design

We conducted a retrospective, cross-sectional study using the 2022 Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), the largest all-payer, nationally representative database of U.S. hospitalizations. The NIS approximates a 20% stratified sample of hospital discharges from nonfederal acute care hospitals and includes discharge-level weights that enable generation of national estimates.¹⁶ Sampling strata are defined by hospital characteristics, including region, teaching status, bed size, and ownership. All analyses accounted for the complex survey design in accordance with HCUP analytic guidelines. Because the NIS contains de-identified publicly available data, this study was exempt from institutional review board review by BayCare Health System.

Study Population and Variables

Adult hospitalizations (aged ≥18 years) complicated by acute kidney injury requiring dialysis (AKI-D) were identified using ICD-10-CM diagnosis and procedure codes. Hospitalizations with evidence of pre-existing end-stage kidney disease, defined by chronic dialysis or end-stage kidney disease diagnosis codes, were excluded to ensure that dialysis reflected acute rather than maintenance therapy. We restricted the analytic cohort to hospitalizations in which patients survived to hospital discharge, as discharge disposition is not defined for in-hospital deaths. This restriction was intentional and aligned with the study objective of examining post-acute recovery trajectories among survivors of dialysis-requiring acute kidney injury rather than in-hospital mortality. The final analytic cohort consisted of survivors of dialysis-requiring acute kidney injury.

Patient-level sociodemographic variables included age group (<50, 50–64, 65–74, 75–84, and ≥85 years), sex, race and ethnicity (as categorized in the NIS), primary expected payer (Medicare, Medicaid, private insurance, self-pay, or no charge), and neighborhood income quartile based on the median household income of the patient’s residential ZIP code. These variables were selected a priori given their established associations with access to care, illness severity, and post-acute outcomes.

Hospital-level characteristics were derived from HCUP hospital files and included teaching status (urban teaching, urban non-teaching, or rural), bed size (small, medium, or large), and geographic region (Northeast, Midwest, South, or West). These characteristics were included to capture institutional and structural differences in care delivery environments.

All ICD-10-CM diagnosis codes and ICD-10-PCS procedure codes used to define dialysis-requiring acute kidney injury, exclusion of pre-existing end-stage kidney disease, chronic kidney disease for sensitivity analyses, and related renal replacement therapy procedures are provided in Supplementary Table 1.

Outcomes and Covariates

The primary outcome was dialysis dependence or non-recovery at hospital discharge, operationalized using discharge disposition as a pragmatic, discharge-based surrogate of post-acute recovery burden following dialysis-requiring acute kidney injury. Non-recovery was defined as survival to discharge after receipt of acute dialysis with discharge to a non-home setting, including skilled nursing facilities, long-term acute care hospitals, or inpatient rehabilitation facilities. This outcome reflects the combined consequences of incomplete renal recovery, functional decline, and overall illness severity at discharge rather than isolated physiologic renal function.

Illness severity and physiological risk were accounted for using the All Patient Refined Diagnosis Related Groups (APR-DRG) severity of illness and risk of mortality classifications. These measures are derived from diagnoses, procedures, age, and comorbidities. APR-DRG severity of illness and risk of mortality classifications were used as global inpatient acuity adjustment measures, consistent with prior national AKI studies using administrative data, rather than as renal-specific severity indices. Severity of illness and mortality risk were categorized as minor, moderate, major, or extreme and were included to mitigate confounding by underlying acuity when examining associations between sociodemographic factors, hospital context, and recovery outcomes.

Statistical Analysis

Baseline characteristics of AKI-D survivors were summarized according to recovery status at discharge. Categorical variables were reported as weighted percentages with 95% confidence intervals, and continuous variables were summarized using weighted means or medians as appropriate. All descriptive analyses incorporated NIS discharge weights and survey design variables to generate nationally representative estimates.

Survey-weighted multivariable logistic regression models were used to examine factors associated with dialysis dependence or non-recovery at discharge. Covariates included age group, sex, race and ethnicity, primary payer, neighborhood income quartile, APR-DRG severity of illness, APR-DRG risk of mortality, hospital teaching status, hospital bed size, and geographic region. Adjusted odds ratios with 95% confidence intervals were reported.

To improve interpretability beyond relative measures, adjusted probabilities of dialysis dependence or non-recovery were estimated using marginal standardization. Predicted probabilities were generated by averaging model-based predictions over the observed covariate distribution of the analytic cohort, and absolute risk differences were calculated by comparing adjusted probabilities across exposure groups.

Effect modification by hospital context was evaluated by including interaction terms between primary payer and hospital teaching status, bed size, and geographic region. Adjusted probabilities and absolute risk differences were estimated within hospital strata to assess whether institutional characteristics modified recovery patterns following AKI-D.

Several sensitivity analyses were performed to assess the robustness of findings. First, an alternative outcome definition was examined in which discharge with home health services was classified as non-recovery. Second, analyses were repeated after excluding patients with chronic kidney disease, defined using ICD-10-CM N18 codes excluding end-stage kidney disease. Results were compared qualitatively across models for consistency.

As a secondary analysis, total hospitalization charges were examined among AKI-D survivors stratified by recovery status at discharge. Charges were summarized as weighted medians with interquartile ranges due to right-skewed distributions. Hospitalizations with total charges less than or equal to zero were treated as missing prior to analysis. Reported charges reflect hospital billing amounts and do not represent actual costs or reimbursements.

All analyses were conducted using R software (version 4.3.1).¹⁷ Survey-weighted analyses were performed using the survey and srvyr packages. Two-sided statistical tests were used, and a p value <0.05 was considered statistically significant.

Results

Study Cohort and Discharge Outcomes

Among U.S. hospitalizations complicated by dialysis-requiring acute kidney injury (AKI-D) in 2022, a substantial proportion of patients survived to hospital discharge and were eligible for analysis. Discharge trajectories among AKI-D survivors varied markedly. While many patients were discharged home, a large subset met criteria for dialysis dependence or non-recovery at discharge, reflecting ongoing need for institutional or post-acute care. When weighted to national estimates, approximately 40% of AKI-D survivors were discharged to non-home settings, indicating a considerable post-acute burden of incomplete recovery following AKI-D across U.S. hospitals (Figure 1).

Figure 1 National trajectory of acute kidney injury and dialysis dependence at hospital discharge. Notes: Figure 1 shows a schematic depiction of acute kidney injury (AKI) hospitalizations in the United States using the 2022 HCUP National Inpatient Sample. Among patients with AKI and no pre-existing end-stage kidney disease, a subset required acute dialysis during hospitalization (AKI-D). Among AKI-D survivors, discharge trajectories diverged substantially: approximately 60% were discharged home, whereas nearly 40% met criteria for dialysis dependence or non-recovery at discharge, defined as survival with discharge to a non-home setting following receipt of acute dialysis. Counts and percentages are weighted to national estimates.

Baseline characteristics differed meaningfully by recovery status at discharge (Table 1). Patients who were dialysis dependent or did not recover were older, with a higher proportion in advanced age categories, and were more likely to be female. These patients exhibited substantially greater illness severity, with a predominance of extreme APR-DRG severity of illness and risk of mortality classifications, and experienced longer hospital lengths of stay. Differences were also observed across race and ethnicity, insurance type, and neighborhood income. Hospital characteristics were broadly similar between groups, although modest variation was observed by geographic region and teaching status.

Table 1 Baseline Characteristics of AKI-D Survivors by Dialysis Dependence at Discharge

Factors Associated with Non-Recovery at Discharge

In survey-weighted multivariable logistic regression analyses adjusting for sociodemographic characteristics, illness severity, and hospital factors, several variables were independently associated with dialysis dependence or non-recovery at discharge (Table 2). Older age demonstrated a strong graded association with non-recovery, with patients aged ≥85 years having more than twice the odds of non-recovery compared with those aged 50–64 years (adjusted OR 2.19, 95% CI 1.95–2.46), whereas patients aged <50 years had substantially lower odds (adjusted OR 0.59, 95% CI 0.55–0.63). Female sex was associated with modestly higher odds of non-recovery (adjusted OR 1.11, 95% CI 1.06–1.17).

Table 2 Factors Associated with Dialysis Dependence or Non-Recovery at Discharge Among AKI-D Survivors

Compared with White patients, individuals identifying as Black (adjusted OR 0.87, 95% CI 0.82–0.93), Hispanic (adjusted OR 0.66, 95% CI 0.61–0.72), or Asian/Pacific Islander (adjusted OR 0.57, 95% CI 0.50–0.65) had lower adjusted odds of dialysis dependence or non-recovery.

Primary payer status was also associated with recovery outcomes. Relative to Medicare beneficiaries, patients with Medicaid (adjusted OR 0.80, 95% CI 0.74–0.87), private insurance (adjusted OR 0.82, 95% CI 0.76–0.87), and self-pay status (adjusted OR 0.45, 95% CI 0.38–0.54) had lower adjusted odds of non-recovery at discharge. These payer-related differences in discharge-based non-recovery likely reflect structural differences in access to post-acute care rather than biologic differences in renal recovery.

Extreme APR-DRG severity of illness (adjusted OR 2.31, 95% CI 2.16–2.48) and extreme risk of mortality (adjusted OR 1.58, 95% CI 1.47–1.69) were strongly associated with non-recovery. Hospital-level characteristics, including urban non-teaching status, smaller bed size, and geographic region, were also independently associated with discharge outcomes. Complete results from the fully adjusted multivariable model are presented in Supplementary Table 2.

To contextualize these associations in absolute terms, adjusted probabilities of dialysis dependence or non-recovery were estimated using marginal standardization (Figure 2). Substantial differences in absolute risk were observed across age groups and payer categories after adjustment for clinical severity and hospital characteristics. Medicare beneficiaries and older adults experienced the highest adjusted probabilities of non-recovery, whereas younger patients and those with self-pay or no-charge encounters exhibited substantially lower adjusted probabilities. Differences by neighborhood income quartile were comparatively modest after adjustment.

Figure 2 Adjusted probability of non-recovery after AKI-D by sociodemographic factors. Notes: Figure 2 presents adjusted probabilities of dialysis dependence or non-recovery at discharge across race/ethnicity, insurance status, and neighborhood income to contextualize relative associations in absolute risk terms. Adjusted probabilities were estimated using survey-weighted logistic regression models and marginal standardization over the observed distribution of covariates. Points represent adjusted probabilities and horizontal bars indicate 95% confidence intervals. Models were adjusted for age, sex, race/ethnicity, primary payer, neighborhood income quartile, APR-DRG severity of illness, APR-DRG risk of mortality, and hospital characteristics.

Hospital Context, Sensitivity Analyses, and Secondary Outcomes

The association between insurance status and non-recovery varied across hospital contexts (Figure 3). Absolute risk differences between Medicare and self-pay patients differed by hospital teaching status, bed size, and geographic region, indicating effect modification by institutional characteristics. In some hospital settings, payer-related differences in non-recovery were attenuated, whereas in others they were more pronounced, suggesting that hospital context may amplify or mitigate disparities in recovery following AKI-D.

Figure 3 Hospital context modifies the risk difference in non-recovery after dialysis-requiring acute kidney injury. Notes: Figure 3 illustrates adjusted risk differences in dialysis dependence or non-recovery at discharge (Medicare minus self-pay) among AKI-D survivors across hospital teaching status, bed size, and region. Risk differences were derived from survey-weighted logistic regression models with interaction terms between payer and hospital context, using marginal standardization to estimate adjusted probability. Points represent absolute risk differences and horizontal bars indicate 95% confidence intervals; the dashed vertical line denotes no difference. Models were adjusted for age, sex, race/ethnicity, neighborhood income quartile, APR-DRG severity of illness, APR-DRG risk of mortality, and remaining hospital characteristics.

To integrate both overall recovery burden and payer-related disparities, adjusted probabilities and absolute risk differences were examined jointly by hospital teaching status (Figure 4). Urban non-teaching and rural hospitals exhibited higher adjusted probabilities of non-recovery compared with urban teaching hospitals. These hospital types also demonstrated greater variability in payer-related risk differences, identifying institutional settings where both recovery burden and disparities co-occurred.

Figure 4 Hospital context and post-acute kidney injury recovery burden. Notes: Figure 4 illustrates hospital teaching status–specific adjusted probability of dialysis dependence or non-recovery at hospital discharge following dialysis-requiring acute kidney injury (AKI-D) and the absolute risk difference between Medicare and self-pay patients. The left column shows overall adjusted non-recovery burden, while the right column shows payer-related disparity, expressed as the Medicare minus self-pay risk difference. Color intensity reflects magnitude, with numeric values annotated within each cell. Estimates were derived from survey-weighted regression models using marginal standardization.

Findings were robust across multiple sensitivity analyses (Supplementary Table 3). When discharge with home health services was classified as non-recovery, the direction and magnitude of associations were preserved. Similarly, exclusion of patients with chronic kidney disease did not materially alter the primary results. Across all sensitivity scenarios, adjusted estimates remained consistent with the main analysis.

As a secondary analysis, hospitalization charges differed substantially by recovery status at discharge. Patients who were dialysis dependent or did not recover incurred markedly higher median total charges compared with those discharged home (Supplementary Table 4), reflecting longer hospitalizations and greater intensity of inpatient care.

Discussion

In this nationally representative study of dialysis-requiring acute kidney injury, we found that incomplete recovery at hospital discharge remains common and contributes to a substantial post-acute care burden in the United States. A large proportion of AKI-D survivors did not return directly home, underscoring that survival alone does not equate to recovery. After adjustment for illness severity and hospital characteristics, recovery trajectories varied meaningfully across sociodemographic groups and were further modified by hospital context. Together, these findings highlight that recovery after AKI-D is shaped not only by patient factors but also by the institutional environments in which care is delivered.

Discharge-level non-recovery following AKI-D represents a clinically meaningful marker of vulnerability during the transition from inpatient to post-acute care. Patients discharged to non-home settings may require ongoing dialysis support, intensive rehabilitation, or skilled nursing services, placing them at elevated risk for rehospitalization and long-term functional decline.^18,19 The divergent discharge trajectories shown in Figure 1 emphasize that AKI-D frequently initiates a prolonged recovery phase rather than a discrete inpatient event. While discharge disposition cannot directly capture long-term renal outcomes, it provides a pragmatic indicator of post-AKI burden within administrative data and reflects the immediate consequences of incomplete recovery on patients and health systems.

Marked differences in recovery trajectories were observed across sociodemographic groups, even after accounting for illness severity and hospital characteristics (Figure 2). Older adults and Medicare beneficiaries experienced the highest adjusted probabilities of non-recovery, consistent with greater physiological vulnerability and higher baseline comorbidity burden.²⁰ In contrast, the lower adjusted probabilities of non-recovery observed among self-pay and no-charge patients should not be interpreted as superior physiological recovery. Instead, these findings likely reflect constrained access to post-acute care facilities among uninsured patients, resulting in higher rates of discharge home despite ongoing medical or functional needs. In this context, discharge to home may paradoxically signal heightened vulnerability rather than successful recovery, underscoring the importance of interpreting administrative recovery proxies through an equity lens. Racial and ethnic differences in adjusted probabilities were also evident and should be interpreted within the context of structural inequities in healthcare access, baseline health status, and care transitions, rather than inherent differences in recovery potential.

Beyond patient-level characteristics, hospital context emerged as a key modifier of recovery after AKI-D. As shown in Figure 3, payer-related differences in non-recovery varied substantially across hospital teaching status, bed size, and geographic region, indicating that institutional factors can either amplify or attenuate disparities. The composite patterns displayed in Figure 4 further demonstrate that some hospital settings experience both a higher overall burden of non-recovery and larger payer-related gaps, whereas others mitigate these differences more effectively. These findings suggest that hospitals are not passive settings for AKI recovery but active determinants of post-discharge trajectories, likely through differences in nephrology access, discharge planning resources, and post-acute care coordination; however, these mechanisms cannot be directly measured within the National Inpatient Sample and should be interpreted as hypotheses rather than causal conclusions.

From a policy and practice perspective, the hospital contexts identified in Figures 3 and 4—particularly urban non-teaching hospitals with a high burden of discharge-based non-recovery—represent actionable targets for system-level intervention. These findings suggest that recovery after AKI-D is not solely a patient-level issue but one that may be amenable to improvement through institutional policies governing care transitions and post-acute resource allocation.

Hospitals with a high burden of discharge-based non-recovery—particularly urban non-teaching and rural hospitals—may benefit from standardized AKI-D discharge pathways, early nephrology engagement, and enhanced coordination with post-acute care facilities. While the effectiveness of these strategies cannot be evaluated within the current dataset, they represent actionable targets for quality improvement efforts aimed at improving post-AKI recovery and equity.

The observed heterogeneity in recovery trajectories across hospital contexts has important implications for clinical care and health system planning. Hospitals with a high burden of non-recovery and pronounced disparities may represent priority targets for interventions aimed at improving AKI recovery pathways. Strategies such as early nephrology engagement, standardized dialysis discontinuation protocols, and enhanced discharge planning for AKI-D survivors could help reduce post-acute vulnerability.^21,22 More broadly, these findings support a shift toward viewing AKI-D recovery as a continuum that extends beyond hospitalization, with hospital systems playing a central role in shaping equitable recovery outcomes.

Incomplete recovery following dialysis-requiring AKI was associated with a substantially higher hospitalization burden, reflecting longer lengths of stay and greater intensity of inpatient care.²³ Although cost data were examined as a secondary analysis, these findings complement the recovery patterns illustrated in Figure 1 by emphasizing that non-recovery has implications not only for patients but also for health systems. The convergence of clinical vulnerability and higher resource utilization underscores the downstream impact of incomplete AKI recovery and reinforces the importance of early identification and targeted transitional care for high-risk patients.

This study has several strengths that enhance the interpretability and relevance of its findings. By leveraging a large, nationally representative inpatient dataset, we were able to characterize recovery trajectories after AKI-D across diverse patient populations and hospital settings. The use of survey-weighted methods and marginal standardization allowed us to present absolute risk estimates that are directly interpretable and clinically meaningful. In addition, the explicit examination of hospital context as an effect modifier extends prior work by highlighting institutional factors as potential levers for improving recovery after AKI-D.

Several limitations warrant consideration. First, recovery was assessed using discharge disposition rather than laboratory-based renal function or post-discharge dialysis dependence. Discharge to a non-home setting may reflect non-renal factors such as physical deconditioning or comorbidity-related care needs, and some patients may recover kidney function yet still require institutional care. Accordingly, this outcome should be interpreted as a measure of post-acute recovery burden rather than isolated physiologic renal recovery. Such misclassification would be expected to bias associations toward the null rather than generate spurious disparities, particularly given adjustment for illness severity and the consistency of findings across sensitivity analyses.

Second, dialysis initiation and discontinuation decisions may vary across institutions and clinicians, which could contribute to observed heterogeneity across hospital contexts. Third, residual confounding by unmeasured clinical factors, such as baseline kidney function or hemodynamic instability, cannot be fully excluded. Finally, findings reflect inpatient care patterns in 2022 and may not capture temporal trends in AKI management.

Conclusion

In this nationally representative analysis, a substantial proportion of survivors of dialysis-requiring acute kidney injury experienced incomplete recovery at hospital discharge, underscoring that survival alone does not equate to recovery. Recovery trajectories varied across sociodemographic groups and were modified by hospital context, highlighting recovery after AKI-D as a health system–dependent process shaped by structural vulnerability and institutional environments.

Importantly, lower rates of discharge-based non-recovery among self-pay patients should not be interpreted as favorable health outcomes, as discharge home in this group may reflect constrained access to post-acute care rather than true physiologic recovery. Similarly, racial and ethnic differences in recovery should be understood within the context of structural inequities in access, care transitions, and institutional resources rather than inherent differences in recovery potential.

Together, these findings identify specific hospital settings—particularly those with a high burden of non-recovery—as actionable targets for quality improvement initiatives, including standardized AKI-D discharge pathways, early nephrology engagement, and enhanced coordination with post-acute care. Efforts to improve recovery after AKI-D should extend beyond the acute hospitalization and incorporate equity-focused, system-level strategies to promote more equitable post-acute outcomes.