Abstract
Objective.
Thirty-day hospital readmissions in systemic lupus erythematosus (SLE) approach proportions in Medicare-reported conditions including heart failure (HF). We compared adjusted 30-day readmission and mortality among SLE, HF, and general Medicare to assess predictors informing readmission prevention.
Methods.
This database study used a 20% sample of all US Medicare 2014 adult hospitalizations to compare risk of 30-day readmission and mortality among admissions with SLE, HF, and neither per discharge diagnoses (if both SLE and HF, classified as SLE). Inclusion required live discharge and ≥12 months of Medicare A/B before admission to assess baseline covariates including patient, geographic, and hospital factors. Analysis used observed and predicted probabilities, and multivariable GEE models clustered by patient to report adjusted risk ratios (ARRs) of 30-day readmission and mortality.
Results.
SLE admissions (n=10,868) were younger, predominantly female, more likely to be Black, disabled, and have Medicaid or end-stage renal disease (ESRD). Observed 30-day readmissions of 24% were identical for SLE and HF (p=0.6), and higher than other Medicare (16%, p<0.001). Both SLE and HF had elevated readmission risk (ARR 1.08, (95% CI (1.04, 1.13)); 1.11, (1.09, 1.13)). SLE readmissions were higher for Black (30%) versus White (21%) populations, and highest in ages 18–33 (39%) and ESRD (37%). Admissions of Black patients with SLE from least disadvantaged neighborhoods had highest 30-day mortality (9% versus 3% White).
Conclusion.
Thirty-day SLE readmissions rivaled HF at 24%. Readmission prevention programs should engage young, ESRD patients with SLE and examine potential causal gaps in SLE care and transitions.
Keywords: Systemic lupus erythematosus (SLE), Disparities, Comorbidity, Medicare, Quality of Care
INTRODUCTION
Systemic lupus erythematosus (SLE) ranked sixth as a cause of hospital readmissions in an unadjusted 2010 US hospital report, with 27% of patients being readmitted within 30 days (1). This suggests that patients with SLE might have greater risk for hospital readmission than those with ambulatory care-sensitive conditions targeted through the Centers for Medicare & Medicaid Services (CMS) Hospital Readmissions Reduction Program penalties since 2013 (2). Given that CMS uses readmissions as a marker of poor transitional care quality (3), adverse outcomes for patients, and increased costs (4), US hospitals are now penalized with up to a 3% reduction in inpatient Medicare payments for early readmissions related to heart failure (HF) and other targeted conditions (2). Although SLE is not subject to such penalties, given SLE readmissions potentially rivaling CMS-targeted conditions such as HF, we sought to compare 30-day readmissions and mortality between SLE, HF, and the general Medicare population in an adjusted analysis. We hypothesized that adjusted risk of readmission in SLE would be equal to HF. We also examined predictors of readmission, hypothesizing that sociodemographics and social determinants such as neighborhood disadvantage could predict readmission risk. These are first steps toward designing and targeting interventions to improve post-hospitalization SLE care.
In addition to the aforementioned 2010 hospital statistics report (1), a separate single-site US study reported 34% 30-day readmissions in SLE (5). Another five-state US study reported 17% 30-day readmissions for SLE (6). Both reported increased risk in historically disadvantaged populations including Black patients and those receiving Medicaid, a socioeconomically disadvantaged population (5, 6). Likewise, a US national study reported disparities in SLE mortality over a 46-year period, showing higher age-standardized mortality rates among Black patients and residents of the South (7). However, details and mechanisms of how such sociodemographic factors or other social determinants of health impact 30-day post-hospital readmission risk and mortality in SLE remain unclear, and no studies have examined adjusted between-condition comparisons. Examining factors contributing to readmission risk in SLE is critical for developing effective mitigation strategies to improve care quality and reduce readmissions, early mortality, and healthcare costs. Understanding how the causes and patterns of SLE readmissions compare to other common readmission diagnoses, such as HF, will inform if and how strategies for reducing readmissions from HF or other conditions may be applied in SLE. Our investigation focused on a national public insurance sample, Medicare; given that one third of ambulatory patients with SLE (8, 9) and half of hospitalizations for SLE are covered by public insurance, this was considered an appropriate sample.
Using a unique geo-linked national Medicare database, we compared the observed and adjusted 30-day readmission and 30-day mortality among SLE, HF, and the Medicare population with neither condition. We also assessed predictors of 30-day readmissions and mortality in SLE with the goal of identifying targets for readmission prevention strategies.
PATIENTS AND METHODS
Design Overview
We performed a database study using a 20% random sample of all US Medicare adult hospitalizations in 2014 to compare the risk of 30-day readmission and mortality among admissions for SLE, HF, and neither per discharge diagnoses. Admissions with both SLE and HF were classified as SLE.
Sample Definition with Inclusion and Exclusion Criteria
Our admission cohort included qualifying acute care hospital admissions from a nationally representative 20% random sample of Medicare patients with at least one inpatient admission from January through November of 2014; our sample was censored at the end of November so that we could observe subsequent 30-day readmissions. This 20% sample included comprehensive outpatient claims data to define baseline comorbidity and health care utilization from January 1 through December 31, 2013. Inclusion criteria consisted of being at least 18 years old and alive at index hospital discharge, and having at least 12 months of continuous Medicare parts A and B coverage prior to index admission (for comorbidity assessment) and 30 days after discharge (for outcomes assessment) (see Supplementary Figure). We excluded claims for care in long-term acute care facilities, psychiatric hospitals, children’s hospitals, cancer-specific hospitals, and rehabilitation hospitals, given different payment methods for these hospital types, in a similar fashion to other studies. Beneficiaries with other payors beyond Medicare, including capitated health maintenance organizations or railroad benefits, were excluded because they may have incomplete claims data, as were those discharged against medical advice or who were alive at 30 days but with fewer than 30 days of Medicare follow-up coverage. The Minimal Risk Health Sciences Institutional Review Board approved this study with waiver of individual informed consent.
Outcomes
Our primary outcome of interest was all-cause acute care readmission within 30 days of index hospitalization discharge (2). We additionally examined all-cause death within 30 days of each index hospitalization discharge.
Explanatory Variables
To define comparison groups, SLE hospitalization was defined as any inpatient stay with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM; all data were prior to the use of ICD-10 codes) code of 710.0 in any position; this method has a positive predictive value of 99.4% for a confirmed patient with lupus and has been used in prior studies (6, 10). HF hospitalization was identified using ICD-9-CM codes 398.91, 402.11, 402.91, 404.11, 404.13, 404.91, 404.93, and 428.0–428.9 (11). Index admissions with both SLE and HF were classified as SLE admissions (n=2,451). The comparator group consisted of all other hospitalizations within this time period that had neither SLE nor HF as an included diagnosis that were labeled “neither.”
We also used measures to examine geographic and sociodemographic variation in readmissions. Rural/Urban Commuting Area (RUCA) codes, which categorize commuter patterns from isolated rural small town to urban areas, were grouped into four categories: urban, suburban, small towns, and large towns (12). Additionally, we used a detailed assessment of neighborhood (i.e., census block-group) geographical context called the Area Deprivation Index (ADI) (13). The ADI is a validated composite measure of neighborhood-level factors representing socioeconomic disadvantage that predicted readmission risk in a random 5% Medicare sample (14). ADI factors include 17 neighborhood-level measures, such as income, education, employment, and housing quality. The ADI uses granular, nine-digit ZIP codes to map to census-block groups that discretely define neighborhood context disparities impacting health separately from complex relationships with individual socioeconomic status and race. The ADI was constructed as a national relative ranking of quintile or decile of disadvantage. We examined each of these measures to assess geographic variation in our cohort. Visits missing address data for RUCA or ADI (n=30,852) were omitted from final multivariable models (see Supplementary Figure).
Covariates
Informed by well-established theoretical models of readmission risk, covariates of interest drawn from Medicare files included patient and hospital factors. Patient variables included age at date of index hospitalization, sex, race, ethnicity, history of ever receiving Medicaid in the prior year, initial Medicare enrollment reason (i.e., disability, end-stage renal disease [ESRD], or age ≥65), and length of index hospital stay in days. Comorbid conditions were identified (11) and baseline indicator variables were applied to account for specific comorbidities of interest, and to account for any history of HF or SLE. Baseline indicators were created for any diagnosis of diabetes mellitus, chronic obstructive pulmonary disease (COPD), or ESRD (11) occurring in the year before index hospitalization. Likewise, the Hierarchical Classification Conditions (HCC) comorbidity index (15) and Elixhauser (11) comorbidity indicators were also used to adjust for baseline comorbidity. Hospital variables included hospital discharge volume and medical school affiliation. Measures of hospital volume were created empirically from the available Medicare data tertiled to compare readmission at centers with the highest, lowest, and middle discharge volumes.
Other analyses also stratified by primary reason for admission or readmission using Clinical Classifications Software (CCS) category groups (16). Single level CCS codes aggregate primary diagnoses by organ system into 18 potential categories for purposes of description.
Analysis
To compare baseline characteristics of the SLE, HF, and the neither admissions cohorts, p-values were calculated using ANOVA for numeric variables and chi-square for categorical variables to compare among the three cohorts.
We reported observed 30-day readmissions and mortality in SLE, HF, and Medicare admissions with neither condition. We then compared SLE admissions and outcomes by race and across ADI quintiles from least to most disadvantaged to evaluate for differing readmission proportions. Predicted probabilities for SLE, HF, versus neither and for each risk factor were estimated using fitted multivariate general estimating equation (GEE) models for 30-day readmission and mortality adjusted risk ratios (RR). The delta method was used to determine standard errors for predicted probabilities and 95% confidence intervals (CI) are calculated accordingly. (SAS Institute, Cary, NC).
Next, univariable and multivariable GEE models were used with clustering by patient to calculate unadjusted and adjusted risk ratios (RR) and 95% CI of 30-day readmission and mortality to compare the SLE and the HF cohorts with the cohort with neither and to assess risk factors. To reflect overall healthcare utilization, patients were allowed to re-enter with additional index hospital admissions in 2014 for this visit-level dataset, while clustering accounted for repeat hospital observations of the same patient. The GEE models for readmission were applied to index admissions alive at 30 days after discharge. Therefore, the 30-day readmission results are interpreted as conditional on being alive 30 days after discharge and were only calculated among patients alive through day 30 post-discharge. We also conducted sensitivity analyses to evaluate the competing risk of death. For that competing risk model, we omitted the last admission record within 30 days that led to death for that patient to overcome confounding by the presence of death. This allowed considering death as a competing risk when running GEE models for 30-day readmission. Results were consistent, so we report readmission results among the living, consistent with Hospital Readmissions Reduction Program measurement specifications (17). We then separately assessed risk factors for mortality in the full SLE hospital visit-level cohort.
Finally, we used chi-square tests to examine between-group differences in the reasons for index admissions and readmissions by CCS categories, with Bonferroni adjustment for the number of single-level categories. Analyses were performed using STATA version 15 (StataCorp, College Station, TX) and SAS software version 9.4 (SAS Institute, Cary, NC).
RESULTS
Cohort Description
Among 1.39 million hospital admissions in 2014 within a 20% random Medicare sample, we classified three admission groups: those with an inpatient diagnosis of SLE (n=10,868), those with an inpatient diagnosis of HF (n=345,628), and those with neither SLE nor HF, labeled neither (n=1,033,026). Table 1 compares baseline characteristics among these three groups. The hospitalized SLE group was, on average, 17 years younger than the HF group. The SLE group was predominantly female (89% compared to 54% HF and 57% neither, p<0.001), and more than twice as likely to be Black (29% vs. 12–14%, p<0.001). Eighty percent of SLE cases in this cohort were Medicare eligible due to disability or ESRD. Compare to HF or neither, SLE admissions were nearly four times more likely to be Medicare eligible due to ESRD, twice as likely due to disability, and had 50% greater Medicaid eligibility, indicating individual financial need. The SLE group was more often urban and more often resided in the most disadvantaged neighborhood quintile. While several comorbidities were slightly less frequent in SLE, the ESRD rate was doubled (20% vs 11% HF and 5% neither, p<0.001). SLE hospitalizations were more likely to occur at medical school-affiliated hospitals (55% vs 48–49%, p<0.001) and high-volume centers.
Table 1.
Baseline characteristics of Medicare admissions with SLE, HF and Neither
| Overall n = 1,389,522 n (%) | SLE n = 10,868 n (%) | HF n = 345,628 n (%) | Neither n = 1,033,026 n (%) | p value* | ||
|---|---|---|---|---|---|---|
| Patient Variables | ||||||
| Admit age (mean, [SD]) | 73.4, [13.6] | 59.9, [16.7] | 76.9, [12.0] | 72.4, [13.8] | <0.0001 | |
| Age group | 18–33 | 19,594 (1) | 967 (9) | 1,045 (<1) | 17,582 (2) | <0.0001 |
| 34–49 | 72,106 (5) | 2,187 (20) | 8,627 (3) | 61,292 (6) | ||
| 50–64 | 190,008 (14) | 2,834 (26) | 40,411 (12) | 146,763 (14) | ||
| 65–79 | 632,874 (46) | 3,685 (34) | 141,284 (41) | 487,905 (47) | ||
| ≥80 | 474,940 (34) | 1,195 (11) | 154,261 (45) | 319,484 (31) | ||
| Gender | Male | 602,996 (43) | 1,145 (11) | 158,432 (46) | 443,419 (43) | <0.0001 |
| Female | 786,526 (57) | 9,723 (89) | 187,196 (54) | 589,607 (57) | ||
| Race/Ethnicity | White | 1,145,325 (82) | 6,738 (62) | 279,964 (81) | 858,623 (83) | <0.0001 |
| Black | 172,030 (12) | 3,153 (29) | 49,736 (14) | 119,141 (12) | ||
| Asian | 15,033 (1) | 113 (1) | 3,724 (1) | 11,196 (1) | ||
| Native American | 10,197 (1) | 129 (1) | 2,295 (1) | 7,773 (1) | ||
| Other/Unknown | 19,219 (1) | 197 (2) | 4,005 (1) | 15,017 (1) | ||
| Hispanic | 27,718 (2) | 538 (5) | 5,904 (2) | 21,276 (2) | ||
| Medicaid ever | 416,701 (31) | 5,007 (46) | 105,933 (31) | 305,761 (30) | <0.0001 | |
| Medicare reason | ||||||
| Disability | 448,004 (30) | 7,047 (65) | 102,048 (30) | 338,909 (33) | <0.001 | |
| ESRD entitlement | 46,376 (4) | 1,637 (15) | 13,897 (4) | 30,842 (3) | <0.0001 | |
| Index stay days (mean, [SD]) | 4.9, [4.9] | 5.3, [5.3] | 5.9, [5.6] | 4.6, [4.7] | <0.0001 | |
| RUCA | Urban | 1,157,664 (84) | 9,483 (87) | 287,367 (83) | 860,814 (84) | <0.0001 |
| Large City/Town | 113,457 (8) | 771 (7) | 28,512 (8) | 84,174 (8) | ||
| Small Rural | 65,261 (5) | 340 (3) | 16,618 (5) | 48,303 (5) | ||
| Isolated | 49,700 (4) | 252 (2) | 12,507 (4) | 36,941 (4) | ||
| ADI disadvantage | <0.001 | |||||
| Least | 1–20 | 234,523 (17) | 1,519 (14) | 53,126 (16) | 179,878 (18) | |
| 21–40 | 282,819 (21) | 1,901 (18) | 67,276 (20) | 213,642 (21) | ||
| 41–60 | 296,295 (22) | 2,283 (21) | 73,699 (22) | 220,313 (22) | ||
| 61–80 | 300,634 (22) | 2,485 (23) | 77,906 (23) | 220,243 (22) | ||
| Most | 81–100 | 246,820 (18) | 2,454 (23) | 66,317 (20) | 178,049 (18) | |
| HCC Score (mean, [SD]) | 2.84, [2.13] | 3.73, [2.53] | 3.75, [2.15] | 2.54, [2.03] | <0.0001 | |
| COPD | 619,728 (45) | 5,058 (47) | 216,082 (63) | 398,588 (39) | <0.0001 | |
| Diabetes mellitus | 498,378 (36) | 3,263 (30) | 164,453 (48) | 330,662 (32) | <0.0001 | |
| ESRD | 95,331 (7) | 2,157 (20) | 37,528 (11) | 55,646 (5) | <0.0001 | |
| Hospital Variables | ||||||
| Discharge volume | Highest | 472,212 (34) | 4,217 (39) | 118,609 (34) | 349,386 (34) | <0.0001 |
| Middle | 458,409 (33) | 3,742 (34) | 113,834 (33) | 340,833 (34) | ||
| Lowest | 458,901 (33) | 2,909 (27) | 113,185 (33) | 342,807 (33) | ||
| Medical school affiliation | 674,466 (49) | 5,967 (55) | 169,411 (49) | 499,088 (48) | <0.0001 | |
| For-profit status | 207,479 (15) | 1,630 (15) | 48,056 (14) | 152,928 (15) | <0.0001 | |
| Critical access | 45,542 (3) | 159 (1) | 11,430 (3) | 33,953 (3) | <0.0001 | |
| 30-Day Outcome Events | ||||||
| Readmissions | 254,328 (18) | 2,592 (24) | 83,249 (24) | 168,487 (16) | <0.0001 | |
| Deaths | 71,235 (5) | 310 (3) | 27,462 (8) | 1,033,025 (4) | <0.0001 | |
p values calculated using ANOVA for numeric variables & chi-square for categorical comparisons. Critical access designation by CMS Medicare means the facility has 25 or fewer beds, is >35 mi from another hospital, and provides rural access in exchange for cost-based reimbursement. Abbreviations: RUCA = rural urban commuting area, ADI=area deprivation index, HCC=Hierarchical Condition Category, HF=heart failure, COPD=chronic obstructive pulmonary disease, ESRD=end-stage renal disease.
Observed Thirty-Day Readmission and Mortality
The observed 30-day readmission proportion was nearly identical for SLE and HF at 24%, both higher than 16% with neither (Table 1, p<0.001). The 30-day post-discharge mortality in the SLE admissions group was 3%, compared to 8% in the HF group and 4% in the neither group, which were both significantly older than the SLE cohort.
Thirty Day Readmissions Compared
The adjusted 30-day readmission risk ratio in SLE (ARR 1.08 (95% CI 1.04, 1.13)) was higher than the neither cohort but slightly lower than HF (ARR 1.11 (1.09, 1.13)) compared to neither (Table 2; full model in Supplementary Table). In the multivariable analysis, the sociodemographic risk factors that increased readmission included age less than 50, male sex, Medicaid, or disability as the reason for Medicare eligibility. Comorbidity, including higher HCC scores, COPD, diabetes mellitus, and particularly ESRD, were all at higher risk (ESRD ARR 1.17 (1.15, 1.19)), as was length of stay.
Table 2.
Probability and risk ratios for 30-day readmissions in full Medicare admission cohort
| Observed probability, % | Unadjusted risk ratio (95% CI) | Adjusted risk ratio (adjusted 95% CI) | ||
|---|---|---|---|---|
| Neither SLE nor HF | 16.31 | Ref | Ref | |
| SLE | 23.85 | 1.29 (1.23, 1.36) | 1.08 (1.04, 1.13) | |
| HF | 24.09 | 1.37 (1.30, 1.46) | 1.11 (1.09, 1.13) | |
| Age at admission | 18–33 | 26.77 | 1.23 (1.19, 1.27) | 1.19 (1.15, 1.23) |
| 34–49 | 23.62 | 1.29 (1.07, 1.55) | 1.01 (0.98, 1.05) | |
| 50–64 | 21.71 | 1.16 (1.15, 1.17) | 0.99 (0.97, 1.00) | |
| 65–79 | 17.33 | Ref | Ref | |
| ≥80 | 17.08 | 1.04 (1.03, 1.05) | 1.01 (1.00, 1.02) | |
| Gender | Male | 19.46 | Ref | Ref |
| Female | 17.42 | 0.90 (0.89, 0.92) | 0.94 (0.93, 0.94) | |
| Race/Ethnicity | White | 17.47 | Ref | Ref |
| Black | 23.27 | 1.28 (1.19, 1.38) | 1.01 (1.00, 1.03) | |
| Asian | 19.34 | 1.09 (1.05, 1.13) | 1.00 (0.96, 1.04) | |
| Native American | 20.76 | 1.14 (1.09, 1.19) | 1.14 (1.09, 1.19) | |
| Other/Unknown | 17.60 | 0.99 (0.95, 1.02) | 0.95 (0.92, 0.98) | |
| Hispanic | 20.89 | 1.13 (1.10, 1.16) | 0.97 (0.95, 1.00) | |
| Medicaid ever | 22.09 | 1.24 (1.20, 1.29) | 1.04 (1.03, 1.05) | |
| Disability eligibility | 21.53 | 1.16 (1.14, 1.18) | 1.03 (1.01, 1.04) | |
| Length of index stay | Quartile 1 | 14.40 | Ref | Ref |
| Quartile 2 | 16.43 | 1.14 (1.11, 1.17) | 1.09 (1.08, 1.10) | |
| Quartile 3 | 20.29 | 1.37 (1.32, 1.42) | 1.22 (1.21, 1.24) | |
| Quartile 4 | 25.02 | 1.67 (1.59, 1.76) | 1.40 (1.39, 1.42) | |
| RUCA | Urban | 18.56 | 1.09 (1.07, 1.12) | 1.02 (1.00, 1.04) |
| Large City/Town | 17.16 | 1.03 (1.01, 1.06) | 1.00 (0.98, 1.03) | |
| Small Rural | 17.05 | 1.02 (0.99, 1.05) | 1.00 (0.98, 1.03) | |
| Isolated | 16.67 | Ref | Ref | |
| ADI disadvantage | 1–20 (least) | 17.14 | Ref | Ref |
| 21–40 | 17.58 | 1.01 (1.00, 1.03) | 1.01 (0.99, 1.02) | |
| 41–60 | 17.88 | 1.05 (1.00, 1.11) | 1.00 (0.99, 1.02) | |
| 61–80 | 18.70 | 1.06 (1.05, 1.08) | 1.02 (1.01, 1.04) | |
| 81–100 (most) | 20.28 | 1.13 (1.12, 1.15) | 1.03 (1.02, 1.05) | |
| HCC Comorbidity Score | Quartile 1 | 9.37 | Ref | Ref |
| Quartile 2 | 14.16 | 1.47 (1.45, 1.49) | 1.34 (1.32, 1.36) | |
| Quartile 3 | 19.85 | 1.98 (1.95, 2.00) | 1.64 (1.61, 1.66) | |
| Quartile 4 | 29.84 | 2.83 (2.67, 3.01) | 2.00 (1.97, 2.04) | |
| COPD | 21.95 | 1.41 (1.38, 1.44) | 1.06 (1.04, 1.07) | |
| Diabetes mellitus | 21.99 | 1.31 (1.27, 1.34) | 1.02 (1.01, 1.03) | |
| ESRD | 32.46 | 1.86 (1.69, 2.05) | 1.17 (1.15, 1.19) | |
| Discharge Volume | Highest | 19.22 | Ref | Ref |
| Middle | 18.24 | 0.98 (0.94, 1.01) | 0.99 (0.98, 1.00) | |
| Lowest | 17.43 | 0.90 (0.84, 0.96) | 0.98 (0.97, 0.99) | |
| Medical school affiliation | 19.02 | 1.00 (0.90, 1.12) | 1.02 (1.01, 1.03) | |
Univariable model used n=1,389,522; GEE Multivariable model used n=1,358,670 visits with full data. Full model also included 30 Elixhauser Comorbidity Variables (see Supplementary Table); Abbreviations: ADI=Area Deprivation Index, HCC=Hierarchical Condition Category, HF=heart failure, COPD=chronic obstructive pulmonary disease, ESRD=end-stage renal disease, RUCA= rural urban commuting area
In multivariable analysis, greater ADI disadvantage compared to the least disadvantaged quintile predicted higher readmission risk. Notably, Rural/Urban Commuting Areas were not significant predictors of readmissions in univariable or multivariable analyses. Finally, in examining hospital factors, hospitals with the middle and lowest discharge volume demonstrated lower readmission risk.
Thirty-Day Readmission Predictors in SLE
When we stratified by race and ADI deciles, we found that Black patients with SLE disproportionately resided in more disadvantaged neighborhoods. Most admissions and readmissions occurred among such patients who had an up to 37% risk of readmission (Figure 1 A and B). Overall 30-day readmissions in SLE were 30% for Black patients compared to 21% for White/Other, which was statistically significant (p<0.0001). Again, the peak was 37% (20–37%) readmission among admissions of Black patients from the worst ADI decile, versus 19–23% across all ADI disadvantage deciles for White/Other.
Figure 1.
A) Observed 30-day SLE readmission frequency among White/Other Race Ethnicity patients with SLE by ADI decile; B) Frequency of 30-day SLE readmission among Black patients with SLE by ADI decile. C) Frequency of 30-day mortality among White/Other Race Ethnicity patients with SLE by ADI decile; D) Frequency of 30-day mortality among Black patients with SLE by ADI decile.
Next, we examined multivariable factors contributing to 30-day readmissions specifically within the SLE group (Table 3), including sociodemographic, contextual, and hospital indicators. Interestingly, the highest risk of readmission was seen in the 18–33 and 34–49 age groups (39% and 29% respectively), representing 52% and 28% greater adjusted risk, respectively, compared to the typical 65–79 age group for Medicare patients. Longer length of index hospital stay increased readmission risk in multivariable analysis (fourth versus first quartile: ARR 1.37 (1.25, 1.51)). The risk of 30-day readmission increased in the presence of comorbidities, as demonstrated across HCC quartiles (fourth quartile: ARR 2.87 (2.39, 3.45)), and most notably with ESRD with 37% observed readmission (ARR 1.20 (1.07, 1.34)). Rural/Urban Commuting Areas were not predictors of readmissions in univariable or multivariable analyses, and thus are not reported. Hospital factors were not predictive in adjusted models.
Table 3.
Probability and risk ratios for 30-day readmissions in SLE admission cohort
| Observed probability, % | Adjusted predicted probability, % (adjusted 95% CI) | Unadjusted risk ratio (95% CI) | Adjusted risk ratio (adjusted 95% CI) | ||
|---|---|---|---|---|---|
| Age at admission | 18–33 | 38.99 | 33.36 (29.24, 37.48) | 1.94 (1.68, 2.24) | 1.52 (1.29, 1.78) |
| 34–49 | 28.85 | 27.79 (25.05, 30.54) | 1.44 (1.26, 1.64) | 1.28 (1.12, 1.48) | |
| 50–64 | 21.98 | 21.27 (19.41, 23.13) | 1.10 (0.97, 1.24) | 0.99 (0.87, 1.12) | |
| 65–79 | 20.08 | 21.55 (19.75, 23.34) | Ref | Ref | |
| ≥80 | 18.49 | 20.12 (17.26, 22.98) | 0.92 (0.79, 1.07) | 0.93 (0.81, 1.08) | |
| Gender | Male | 24.72 | 24.81 (21.91, 27.70) | Ref | Ref |
| Female | 23.75 | 23.65 (22.65, 24.64) | 0.96 (0.84, 1.11) | 0.95 (0.84, 1.08) | |
| Race/Ethnicity | White | 20.56 | 23.35 (22.07, 24.64) | Ref | Ref |
| Black | 30.29 | 25.04 (23.03, 27.06) | 1.47 (1.33, 1.64) | 1.07 (0.97, 1.19) | |
| Asian | 17.70 | 17.09 (10.59, 23.59) | 0.86 (0.57, 1.31) | 0.75 (0.51, 1.09) | |
| Native American | 31.01 | 27.89 (20.68, 35.10) | 1.51 (1.08, 2.11) | 1.19 (0.92, 1.53) | |
| Other/Unknown | 26.90 | 22.64 (15.90, 29.39) | 1.31 (0.90, 1.90) | 0.98 (0.74, 1.30) | |
| Hispanic | 25.84 | 21.49 (17.74, 25.24) | 1.26 (1.03, 1.53) | 0.94 (0.78, 1.12) | |
| Medicaid ever | 28.02 | 23.78 (22.31, 25.25) | 1.38 (1.26, 1.52) | 1.00 (0.91, 1.10) | |
| Disability eligibility | 25.00 | 23.47 (22.23, 24.72) | 1.15 (1.04, 1.27) | 0.97 (0.88, 1.08) | |
| Length of index stay | |||||
| Quartile 1 | 19.55 | 20.55 (18.96, 22.13) | Ref | Ref | |
| Quartile 2 | 21.27 | 22.28 (20.83, 23.74) | 1.09 (0.99, 1.20) | 1.09 (0.99, 1.19) | |
| Quartile 3 | 25.88 | 25.07 (22.96, 27.19) | 1.32 (1.18, 1.49) | 1.22 (1.10, 1.37) | |
| Quartile 4 | 31.11 | 28.18 (26.44, 29.93) | 1.59 (1.44, 1.76) | 1.37 (1.25, 1.51) | |
| ADI Disadvantage | |||||
| 1–20 (least) | 22.65 | 24.58 (22.18, 27.98) | Ref | Ref | |
| 21–40 | 19.88 | 21.27 (19.30, 23.24) | 0.88 (0.75, 1.03) | 0.86 (0.75, 0.99) | |
| 41–60 | 24.31 | 24.53 (22.64, 26.42) | 1.07 (0.92, 1.25) | 0.99 (0.88, 1.13) | |
| 61–80 | 24.51 | 24.21 (21.98, 0.264) | 1.08 (0.92, 1.28) | 0.98 (0.86, 1.12) | |
| 81–100 (most) | 26.20 | 24.00 (22.01, 25.99) | 1.16 (1.00, 1.35) | 0.98 (0.86, 1.11) | |
| HCC Comorbidity Score | |||||
| Quartile 1 | 10.45 | 12.04 (10.04, 14.04) | Ref | Ref | |
| Quartile 2 | 14.03 | 15.59 (13.96, 17.22) | 1.34 (1.11, 1.63) | 1.32 (1.08, 1.60) | |
| Quartile 3 | 20.15 | 20.65 (19.10, 22.20) | 1.93 (1.61, 2.31) | 1.78 (1.48, 2.13) | |
| Quartile 4 | 36.58 | 33.48 (31.47, 35.49) | 3.5 (2.95, 4.16) | 2.87 (2.39, 3.45) | |
| COPD | 27.64 | 24.89 (23.46, 26.32) | 1.35 (1.22, 1.48) | 1.10 (1.01, 1.19) | |
| Diabetes mellitus | 27.49 | 23.89 (22.21, 25.56) | 1.23 (1.12, 1.36) | 1.01 (0.92, 1.10) | |
| ESRD | 36.76 | 27.41 (24.92, 29.90) | 1.78 (1.62, 1.96) | 1.20 (1.07, 1.34) | |
| Discharge Volume | |||||
| Highest | 24.57 | 23.87 (22.38, 25.36) | Ref | Ref | |
| Middle | 24.69 | 24.36 (22.81, 25.91) | 1.01 (0.91, 1.11) | 0.95 (0.86, 1.06) | |
| Lowest | 21.73 | 22.80 (20.00, 24.60) | 0.88 (0.79, 0.99) | 1.02 (0.93, 1.11) | |
| Medical school affiliation | 24.64 | 23.48 (22.22, 24.74) | 1.08 (0.99, 1.17) | 0.97 (0.90, 1.05) | |
Univariable model used n=10,868; GEE Multivariable model used n=10,640 visits with full data. Abbreviations: ADI=Area Deprivation Index, HCC=Hierarchical Condition Category, COPD=chronic obstructive pulmonary disease, ESRD=end-stage renal disease. Full model shown above.
Thirty-Day Post-Admission Mortality Predictors in SLE
Table 4 shows factors predicting 30-day mortality in SLE. Relative to the 65–79 age group, those younger than 65 had lower 30-day mortality. Female sex also predicted reduced risk of mortality (ARR 0.68 (0.48, 0.97)). Major disease and comorbidity factors predicted increased 30-day SLE mortality including length of index stay, and higher HCC score. Hospital volume and medical school affiliation were not predictive.
Table 4.
Probability and risk ratios for 30-day mortality in SLE admission cohort
| Observed probability, % | Adjusted predicted probability, % (adjusted 95% CI) | Unadjusted risk ratio (95% CI) | Adjusted risk ratio (adjusted 95% CI) | ||
|---|---|---|---|---|---|
| Age at admission | 18–33 | 0.52 | 0.40 (−0.08, 0.88) | 0.11 (0.03, 0.35) | 0.07 (0.02, 0.25) |
| 34–49 | 1.10 | 1.06 (0.55, 1.57) | 0.28 (0.17, 0.46) | 0.28 (0.16, 0.49) | |
| 50–64 | 2.29 | 2.45 (1.75, 3.15) | 0.62 (0.45, 0.86) | 0.63 (0.43, 0.93) | |
| 65–79 | 3.72 | 3.96 (3.07, 4.84) | Ref | Ref | |
| ≥80 | 6.61 | 6.74 (4.79, 8.70) | 2.06 (1.54, 2.75) | 1.76 (1.29, 2.41) | |
| Gender | Male | 4.28 | 4.00 (2.70, 5.30) | Ref | Ref |
| Female | 2.68 | 2.71 (2.35, 3.07) | 0.62 (0.44, 0.86) | 0.68 (0.48, 0.97) | |
| Race/Ethnicity | White | 3.12 | 2.66 (2.25, 3.07) | Ref | Ref |
| Black | 2.54 | 3.49 (2.52, 4.47) | 0.74 (0.56, 0.99) | 1.30 (0.92, 1.82) | |
| Asian | 2.65 | 2.88 (−0.61, 6.37) | 1.09 (0.35, 3.38) | 1.31 (0.39, 4.40) | |
| Native American | 2.33 | 2.77 (0.29, 5.83) | 0.87 (0.29, 2.67) | 1.23 (0.40, 3.82) | |
| Other/Unknown | 3.55 | 5.77 (1.28, 10.26) | 1.09 (0.44, 2.68) | 1.83 (0.79, 4.24) | |
| Hispanic | 1.30 | 2.09 (0.12, 4.05) | 0.39 (0.15, 1.06) | 0.89 (0.37, 2.17) | |
| Medicaid ever | 2.20 | 2.92 (2.25, 3.59) | 0.63 (0.49, 0.82) | 0.98 (0.72, 1.33) | |
| Disability eligibility | 2.02 | 2.59 (2.07, 3.11) | 0.43 (0.34, 0.55) | 0.78 (0.57, 1.08) | |
| Length of index stay | |||||
| Quartile 1 | 1.44 | 1.71 (1.18, 2.24) | Ref | Ref | |
| Quartile 2 | 1.81 | 1.18 (1.30, 2.21) | 1.07 (0.73, 1.58) | 0.90 (0.62, 1.32) | |
| Quartile 3 | 2.68 | 2.25, (1.79, 3.26) | 1.43 (0.92, 2.23) | 1.16 (0.75, 1.79) | |
| Quartile 4 | 6.09 | 5.51 (4.61, 6.41) | 3.2 (2.30, 4.45) | 2.52 (1.81, 3.49) | |
| ADI disadvantage | |||||
| 1–20 (least) | 3.42 | 2.84 (1.94,3.74) | Ref | Ref | |
| 21–40 | 3.52 | 3.07 (2.27, 3.88) | 0.98 (0.66, 1.46) | 0.98 (0.66, 1.46) | |
| 41–60 | 2.72 | 2.91 (2.11, 3.70) | 0.75 (0.50, 1.12) | 0.88 (0.59, 1.34) | |
| 61–80 | 2.86 | 3.07 (2.26, 3.87) | 0.78 (0.52, 1.15) | 1.02 (0.68, 1.53) | |
| 81–100 (most) | 2.16 | 2.39 (1.67, 3.11) | 0.62 (0.41, 0.93) | 0.83 (0.53, 1.31) | |
| HCC Comorbidity Score | |||||
| Quartile 1 | 1.22 | 1.35 (0.67, 2.03) | Ref | Ref | |
| Quartile 2 | 1.40 | 1.40 (0.92, 1.88) | 1.15 (0.62, 2.13) | 1.12 (0.57, 2.20) | |
| Quartile 3 | 2.13 | 2.13 (1.62, 2.64) | 1.90 (1.09, 3.30) | 2.03 (1.11, 3.74) | |
| Quartile 4 | 4.75 | 4.68 (3.88, 5.49) | 4.11 (2.42, 6.97) | 4.17 (2.26, 7.68) | |
| COPD | 3.70 | 2.94 (2.45, 3.43) | 1.78 (1.39, 2.28) | 1.06 (0.81, 1.40) | |
| Diabetes mellitus | 3.03 | 2.42 (1.89, 2.95) | 1.08 (0.83, 1.40) | 0.77 (0.58, 1.01) | |
| ESRD | 2.74 | 3.24 (2.24, 4.23) | 0.95 (0.69, 1.30) | 1.11 (0.77, 1.59) | |
| Discharge Volume | |||||
| Highest | 2.40 | 2.41 (1.88, 2.95) | Ref | Ref | |
| Middle | 3.05 | 3.02 (2.42, 3.63) | 1.34 (1.00, 1.81) | 1.49 (1.09, 2.04) | |
| Lowest | 3.27 | 3.31 (2.58, 4.04) | 1.45 (1.08, 1.95) | 1.36 (1.00, 1.87) | |
| Medical school affiliation | 2.60 | 2.82 (2.33, 3.31) | 0.78 (0.61, 0.98) | 0.87 (0.68, 1.11) | |
Univariable model used n=10,868; GEE Multivariable model used n=10,640 visits with full data. Abbreviations: ADI=Area Deprivation Index, HCC=Hierarchical Condition Category, COPD=chronic obstructive pulmonary disease, ESRD=end-stage renal disease
In multivariable models neither race nor ADI neighborhood disadvantage predicted differential SLE 30-day mortality, but stratification by race and ADI showed differences (Table 4; Figure 1 C and D). Admissions of Black SLE patients from the least disadvantaged neighborhoods had highest 30-day mortality (9% vs. 3% White, unadjusted chi square p=0.005). Sensitivity testing showed this increase occurred despite similar median age across ADI deciles (52.2, range 46.4–55.7 years) and the lowest prevalence of ESRD in this least disadvantaged decile.
Causes of Admission and Readmission
We next examined the primary diagnoses as causes of admission and readmission using 18 CCS diagnosis categories to better understand how clinical conditions associated with 30-day readmission. Comparisons between index admissions with and without SLE and readmissions with and without SLE are shown in Table 5. Overall, the primary diagnoses for index admissions were similar across groups, with all absolute categorical differences less than 5%. The SLE group had more primary readmission diagnoses of musculoskeletal and connective tissue disease, as expected, along with more pregnancy complications, and more injuries and poisoning. Notably, injuries and poisoning ranked second as a cause of readmission in SLE versus fourth in non-SLE readmissions. Infections were higher in index admissions with SLE, but readmissions with infection were no different with or without SLE.
Table 5.
Primary diagnoses by CCS category overall and by readmission and SLE status
| Index Admissions | Readmissions | ||||||
|---|---|---|---|---|---|---|---|
| CCS Category | No SLE Index Admit n=1,378,654 n (%) | SLE Index Admit-No Readmit n=8,276 n (%) | SLE Index Admit + Readmit n=2592 n (%) | p | Non-SLE Patient Readmit n=251,736 n (%) | SLE Patient Readmit n=2,592 n (%) | p |
| Circulatory system | 321,017 (23.3) | 1,546 (18.7) | 508 (19.6) | <0.001 | 57,933 (23.0) | 470 (18.1) | 1 |
| Injury & poisoning | 151,066 (11.0) | 1,020 (12.3) | 312 (12.0) | 0.001 | 28,485 (11.3) | 362 (14.0) | <0.001 |
| Digestive system | 152,825 (11.1) | 962 (11.6) | 335 (12.9) | 0.061 | 29,081 (11.6) | 348 (13.4) | <0.001 |
| Respiratory system | 171,555 (12.4) | 970 (11.7) | 303 (11.7) | 1 | 35,196 (14.0) | 299 (11.5) | <0.001 |
| Infectious & parasitic | 94,019 (6.8) | 626 (7.6) | 231 (8.9) | <0.001 | 24,954 (9.9) | 253 (9.8) | 1 |
| Endocrine, nutritional & metabolic, & immune | 61,412 (4.5) | 374 (4.5) | 156 (6.0) | 0.010 | 13,563 (5.4) | 161 (6.2) | <0.001 |
| Musculoskeletal system & connective tissue | 121,100 (8.8) | 936 (11.3) | 217 (8.4) | <0.001 | 4,698 (1.9) | 135 (5.2) | 0.052 |
| Genitourinary system | 94,202 (6.8) | 525 (6.3) | 137 (5.3) | 0.028 | 17,744 (7.1) | 131 (5.1) | 1 |
| Blood & blood- forming organs | 21,854 (1.6) | 184 (2.2) | 63 (2.4) | <0.001 | 5,992 (2.4) | 85 (3.3) | <0.001 |
| Nervous system & sense organs | 37,359 (2.7) | 312 (3.8) | 80 (3.1) | <0.001 | 6,893 (2.7) | 83 (3.2) | 1 |
| Symptoms, signs, & ill-defined factors | 23,566 (1.7) | 216 (2.6) | 61 (2.4) | <0.001 | 4,589 (1.8) | 82 (3.2) | 0.006 |
| Skin & subcutaneous tissue | 30,269 (2.2) | 214 (2.6) | 65 (2.5) | 0.518 | 4,276 (1.7) | 75 (2.9) | 1 |
| Neoplasms | 57,940 (4.2) | 174 (2.1) | 59 (2.3) | <0.001 | 10,217 (4.1) | 46 (1.8) | 0.001 |
| Mental illness | 30,875 (2.2) | 135 (1.6) | 34 (1.3) | 0.001 | 6509 (2.6) | 40 (1.5) | <0.001 |
| Complications of pregnancy, childbirth & puerperium | 2,629 (0.2) | 41 (0.5) | 16 (0.6) | <0.001 | 240 (0.1) | 12 (0.5) | <0.001 |
| Residual codes, unclassified | 5,376 (0.4) | 30 (0.4) | 12 (0.5) | 1 | 1,227 (0.5) | * | |
| Congenital anomalies | 1,590 (0.1) | 11 (0.1) | * | 138 (0.1) | * | ||
| Perinatal conditions | * | * | * | * | * | * | * |
P values are Bonferroni corrected.
Indicates values omitted for a given row cell size <11 per CMS agreement.
DISCUSSION
In this cohort study of 2014 Medicare hospital admissions, we found that nearly a quarter of all patients admitted with a primary diagnosis of SLE were readmitted within 30 days. This rate was the same as the HF cohort and greater than in the comparison Medicare group without either condition. Our findings are consistent with an unadjusted hospital report from 2010 that cites 27% readmission risk for SLE (1), and a report by Yazdany et al. that documented 17% patient-level SLE readmissions in 2008–09 (6). As shown in Figure 1, observed readmissions were as high as 37% among Black patients living in the most disadvantaged neighborhood ADI deciles. Paradoxically, observed 30-day mortality was greatest at 9% among Black patients in the least disadvantaged neighborhoods, which was three-fold higher than White patients in any setting. Our findings underscore the interplay between race and neighborhood context for predicting care quality and outcomes, which has been shown in prior SLE and non-SLE studies (18–22).
Despite identical observed 30-day readmissions in HF and SLE, we found that the multivariable adjusted risk of readmission in SLE was modestly less in HF, suggesting that readmission risk among SLE patients had greater contributions from factors such as age, comorbidities, and social determinants of health than SLE, itself. The strongest multivariable predictors of readmission in SLE included younger age, particularly ages 18–33, and ESRD, thereby identifying potential target populations for transitions-of-care interventions. Yazdany et al. similarly reported renal disease as a predictor of increased readmission risk (6). Young patients and those with ESRD may benefit from complex case management, close post-discharge follow-up (23–28), and could be engaged to co-design SLE-specific interventions.
In contrast to approaches in HF where HF recurrence drives readmission (27, 28), SLE patients had many different readmission diagnoses (Table 5) requiring different strategies. Finding that “injury and poisonings” was the second highest cause of readmission in SLE calls for further investigation regarding possible comorbid opioid use, which was present in one third of SLE patients in one study (29). Given the many causes of readmissions for patients with SLE, the impact of engaging both timely primary care and rheumatology post-hospital follow-up or SLE-specific transitional care (30) should be tested further. The present analysis provides the first step towards designing interventions to improve post-discharge SLE care.
Consideration must also be given to the role and impact of quality, equitable ambulatory SLE care in avoiding admissions, readmissions, and premature mortality. Further research can examine the how readmission risk is impacted by gaps in lupus care continuum steps that collectively lead to disease control: diagnosis, rheumatology linkage, retention in care, and retention on therapy (31). In HIV, measuring similar care continuum steps stratified by race and other social determinants of health has led to a compendium of evidence-based strategies that reduce disparities (32). If, for instance, young adults with SLE have lower retention in outpatient SLE care, this may drive inpatient admissions and readmissions. Interventions to improve retention in outpatient lupus care, with outreach for regular ambulatory SLE visits and labs, could thereby be tested to reduce readmissions and improve outcomes.
Further research should also investigate the observed higher 30-day mortality in Black patients with SLE living in the least disadvantaged neighborhoods. This unexpected finding requires additional mechanistic-focused studies to be fully interpreted. For instance, it is possible that this group suffers greater SLE severity, which was not measured in this study. However, ESRD was lowest in this decile, while median age was no different. Broader investigation may also warrant considering known paradoxes, such as higher peripartum mortality in college-educated Black women (33) suggesting potential implicit bias or systemic racism (34). Overall, our unadjusted mortality findings in SLE merit further investigation of mortality risk in SLE, with emphasis on SLE disease activity, severity, differences in care based on race, and neighborhood context.
Despite the strengths of this study, including a robust national admissions cohort and a direct comparison with HF to assess readmission risk and predictors, we acknowledge limitations. First, Medicare data predominantly include an older and sicker population than the general US SLE population. However, 80% of our SLE cohort was enrolled for disability or ESRD, rather than age, and 55% were younger than 65 years of age. One third of patients with SLE receive public insurance (8, 9) and more than half of SLE inpatient stays are covered by Medicare or Medicaid, which strengthens the generalizability of our study cohort. Moreover, Medicare is a policy-actionable group where readmission prevention (2) and ambulatory post-hospital visit incentives (35) have been pioneered, which could be particularly beneficial in the high-risk SLE group. We also acknowledge that characteristics such as lupus disease activity and damage, functional status and frailty are not captured in claims data and might impact readmission risk. Patients without address data were excluded and while addresses might be missing at random, it could represent patients with high mobility or homelessness. Among the SLE subset, there were no differences by age or sex and missing addresses were more often with White race (p=0.017) and were less often than overall Medicare. Lastly, we acknowledge that the use of Medicare claims may also be limited by coding accuracy and misclassification bias. We have used validated algorithms to define SLE, HF, and other conditions to minimize this bias (6, 11). Overall, we believe that these limitations are offset by the benefits of using a large, nationally representative sample to study 30-day SLE readmissions and mortality.
This analysis is the first step towards designing interventions to improve post-hospital SLE care. Future studies should focus on strategies to prevent readmission in high-risk SLE populations including young adults and those with ESRD. Research should investigate the role of ambulatory care quality (31) and whether more injury and poisoning diagnoses in SLE readmissions relate to opioid use, a known SLE comorbidity (29). Likewise, future studies should investigate complex relationships between SLE severity, race, and neighborhood variations in 30-day SLE mortality.
CONCLUSIONS
Observed 30-day readmissions impact one in four patients hospitalized with SLE, a rate nearly identical to readmissions for patients with HF. Transitional care programs, Medicare policies, and other efforts designed to reduce readmissions should consider including SLE, a currently under-recognized high-risk group for readmissions. Future intervention design should aim to reduce readmissions by focusing on at-risk younger patients with SLE and those with ESRD.
Supplementary Material
ACKNOWLEDGMENTS
Authors would like to thank Monica Messina PhD for help with manuscript preparation as well as the entire University of Wisconsin Health Services and Care Research group.
Funding:
National Institute on Minority Health and Health Disparities Research Award Number R01MD010243. CB also received support from Rheumatology Research Foundation.
Footnotes
Declaration of Interest: CB also receives peer-reviewed institutional grant funding from Independent Grants for Learning and Change (Pfizer) for research unrelated to this study. All other authors declare no conflicts.
REFERENCES
- 1.Elixhauser A, (AHRQ)., Steiner C, (AHRQ). Readmissions to U.S. Hospitals by Diagnosis, 2010: Statistical Brief #153. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD); 2013. [Google Scholar]
- 2.Centers for Medicare & Medicaid Services. Hospital Readmissions Reduction Program (HRRP) Archives. 2013. URL: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/HRRP-Archives.
- 3.Naylor MD, Aiken LH, Kurtzman ET, Olds DM, Hirschman KB. The care span: The importance of transitional care in achieving health reform. Health Aff (Millwood) 2011;30:746–54. [DOI] [PubMed] [Google Scholar]
- 4.Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360:1418–28. [DOI] [PubMed] [Google Scholar]
- 5.Nangit A, Lin C, Ishimori ML, Spiegel BMR, Weisman MH. Causes and Predictors of Early Hospital Readmission in Systemic Lupus Erythematosus. J Rheumatol 2018;45:929–33. [DOI] [PubMed] [Google Scholar]
- 6.Yazdany J, Marafino BJ, Dean ML, Bardach NS, Duseja R, Ward MM, et al. Thirty-day hospital readmissions in systemic lupus erythematosus: predictors and hospital- and state-level variation. Arthritis Rheumatol 2014;66:2828–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yen EY, Shaheen M, Woo JMP, Mercer N, Li N, McCurdy DK, et al. 46-Year Trends in Systemic Lupus Erythematosus Mortality in the United States, 1968 to 2013: A Nationwide Population-Based Study. Ann Intern Med 2017;167:777–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lim SS, Bayakly AR, Helmick CG, Gordon C, Easley KA, Drenkard C. The incidence and prevalence of systemic lupus erythematosus, 2002–2004: The Georgia Lupus Registry. Arthritis Rheumatol 2014;66:357–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.GfK Roper Public Affairs. Lupus Awareness Survey for the Lupus Foundation of America [Executive Summary - October 2012]. Lupus Foundation of America; October, 2012. URL: Available at https://b.3cdn.net/lupus/2489f6ca2bcbde1818_ggm6i6gzi.pdf. [Google Scholar]
- 10.Hanly JG, Thompson K, Skedgel C. Identification of patients with systemic lupus erythematosus in administrative healthcare databases. Lupus 2014;23:1377–82. [DOI] [PubMed] [Google Scholar]
- 11.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27. [DOI] [PubMed] [Google Scholar]
- 12.WWAMI RUCA Rural Health Research Center. Map Classifications. URL: http://depts.washington.edu/uwruca/ruca-maps.php.
- 13.Singh GK. Area deprivation and widening inequalities in US mortality, 1969–1998. Am J Public Health 2003;93:1137–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kind AJ, Jencks S, Brock J, Yu M, Bartels C, Ehlenbach W, et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med 2014;161:765–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pope GC, Kautter J, Ellis RP, Ash AS, Ayanian JZ, Lezzoni LI, et al. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financ Rev 2004;25:119–41. [PMC free article] [PubMed] [Google Scholar]
- 16.Elixhauser A, Steiner C, Palmer L. Clinical Classifcations Software (CCS) for ICD-9-CM. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp: U.S. Agency for Healthcare Research and Quality; 2007. [Google Scholar]
- 17.Horwitz LI, Partovian C, Lin ZQ, Grady JN, Herrin J, Conover M, et al. Development and Use of an Administrative Claims Measure for Profiling Hospital-wide Performance on 30-Day Unplanned Readmission. Annals of Internal Medicine 2014;161:S66–S75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ludwig J, Duncan GJ, Gennetian LA, Katz LF, Kessler RC, Kling JR, et al. Neighborhood effects on the long-term well-being of low-income adults. Science 2012;337:1505–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Christine PJ, Young R, Adar SD, Bertoni AG, Heisler M, Carnethon MR, et al. Individual- and Area-Level SES in Diabetes Risk Prediction: The Multi-Ethnic Study of Atherosclerosis. Am J Prev Med 2017;53:201–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Durfey SNM, Kind AJH, Buckingham WR, DuGoff EH, Trivedi AN. Neighborhood disadvantage and chronic disease management. Health Serv Res 2019;54 Suppl 1:206–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Feldman CH, Hiraki LT, Liu J, Fischer MA, Solomon DH, Alarcon GS, et al. Epidemiology and sociodemographics of systemic lupus erythematosus and lupus nephritis among US adults with Medicaid coverage, 2000–2004. Arthritis Rheum 2013;65:753–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Trupin L, Tonner MC, Yazdany J, Julian LJ, Criswell LA, Katz PP, et al. The role of neighborhood and individual socioeconomic status in outcomes of systemic lupus erythematosus. J Rheumatol 2008;35:1782–8. [PMC free article] [PubMed] [Google Scholar]
- 23.Takeda A, Martin N, Taylor RS, Taylor SJ. Disease management interventions for heart failure. Cochrane Database Syst Rev 2019;1:CD002752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bradley EH, Curry L, Horwitz LI, Sipsma H, Thompson JW, Elma M, et al. Contemporary evidence about hospital strategies for reducing 30-day readmissions: a national study. J Am Coll Cardiol 2012;60:607–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.McKay C, Park C, Chang J, Brackbill M, Choi JY, Lee JH, et al. Systematic Review and Meta-analysis of Pharmacist-Led Transitions of Care Services on the 30-Day All-Cause Readmission Rate of Patients with Congestive Heart Failure. Clin Drug Investig 2019;39:703–12. [DOI] [PubMed] [Google Scholar]
- 26.Kash BA, Baek J, Cheon O, Manzano JM, Jones SL, Paranilam J, et al. How Leading Hospitals Operationalize Evidence-Based Readmission Reduction Strategies: A Mixed-Methods Comparative Study Using Systematic Review and Survey Design. Am J Med Qual 2019;34:529–37. [DOI] [PubMed] [Google Scholar]
- 27.Chapman E, Eastman A, Gilmore-Bykovskyi A, Vogelman B, Kind AJ. Development and preliminary evaluation of the resident coordinated-transitional care (RC-TraC) program: A sustainable option for transitional care education. Gerontol Geriatr Educ 2018;39:160–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kind AJ, Brenny-Fitzpatrick M, Leahy-Gross K, Mirr J, Chapman E, Frey B, et al. Harnessing Protocolized Adaptation in Dissemination: Successful Implementation and Sustainment of the Veterans Affairs Coordinated-Transitional Care Program in a Non-Veterans Affairs Hospital. J Am Geriatr Soc 2016;64:409–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Somers EC, Lee J, Hassett AL, Zick SM, Harlow SD, Helmick CG, et al. Prescription Opioid Use in Patients With and Without Systemic Lupus Erythematosus - Michigan Lupus Epidemiology and Surveillance Program, 2014–2015. MMWR Morb Mortal Wkly Rep 2019;68:819–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Anandarajah AP, McMahon S, Ostronic A, Feng C, Anolik J, Ritchlin CT. Improving Access to Rheumatology Care for High-Risk Lupus Patients Can Help Decrease Hospitalizations [abstract]. Arthritis Rheumatol 2018;70. [Google Scholar]
- 31.Bartels CM, Rosenthal A, Wang X, Ahmad U, Chang I, Ezeh N, et al. Investigating lupus retention in care to inform interventions for disparities reduction: an observational cohort study. Arthritis Res Ther 2020;22:35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Centers for Disease Control and Prevention. Compendium of Evidence-Based Interventions and Best Practices for HIV Prevention. Linkage to, Retention in, and Re-engagement in HIV Care (LRC) Chapter 2019. URL: https://www.cdc.gov/hiv/research/interventionresearch/compendium/lrc/index.html.
- 33.New York Dept of Mental Health and Hygiene. Severe Maternal Morbidity in New York, 2008–2012. New York, NY; 2016. [Google Scholar]
- 34.Martin N, Montagne R. Nothing Protects Black Women From Dying in Pregnancy and Childbirth. ProPublica & NPR News; December. 7, 2017. [Google Scholar]
- 35.Centers for Medicare & Medicaid Services. Frequently Asked Questions about Billing the Medicare Physician Fee Schedule for Transitional Care Management Services. 2016. URL: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/Downloads/FAQ-TCMS.pdf.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.