Abstract
This study compares nationwide medical and surgical readmission rates in patients with and without serious mental illness.
People with serious mental illness (SMI)—for example, schizophrenia, bipolar disorder, and major depression—experience pronounced challenges in psychological and social functioning that often co-occur with physical health conditions. They receive inferior quality of medical care compared with patients without SMI.1 Risk-adjusted 30-day readmissions are an important indicator of quality care used by the Centers for Medicare & Medicaid Services to guide pay for performance.2 Research using hospital and state data suggests that medical and surgical readmission rates for patients with SMI may be higher than for those without SMI, but the differences in nationwide estimates of 30-day all-cause readmissions are unknown.3,4,5 The objective of our study was to compare nationwide medical and surgical readmission rates in patients with and without SMI.
Methods
We used the 2014 Nationwide Readmissions Database (NRD), which was designed to generate national estimates of readmissions by aggregating acute care hospitalizations from 21 states. The NRD represents inpatient use for 49% of the US population.6 In 2014, the NRD contained approximately 14 million unweighted discharges and 36 million nationally representative weighted discharges. The study was deemed exempt from review by the institutional review board of Northeastern University and did not require informed consent for use of the publicly available database.
We identified index medical and surgical admissions for reasons not considered elective among adults 18 years and older.2 We excluded index admissions if patients were not discharged alive or were transferred to another acute care hospital. The patients with comorbid SMI were identified by any secondary discharge diagnosis with codes for bipolar disorder, major depression, and schizophrenia from the International Classification of Diseases, Ninth Revision, Clinical Modification. All-cause unplanned 30-day readmission rates were calculated for patients admitted from January 1 through November 30, 2014 (to allow for 30 days of follow-up).
Data were analyzed from August 7, 2017, through March 23, 2018. We used χ2 and 2-tailed t tests to compare demographic, clinical, and hospital characteristics of patients with and without a comorbid SMI diagnosis. We also used logistic regression models controlling for demographic, clinical, and hospital characteristics to estimate the effect of SMI diagnosis on the odds of being readmitted within 30 days. We added an interaction term for substance use disorder and mental illness.5 As a sensitivity analysis, we also examined the summary Elixhauser comorbidity score,7 controlling for demographic, hospital, and other clinical characteristics. We used survey procedures to adjust for the stratified cluster design of the NRD with domain, strata, cluster, and weight statements.6
Results
The 2014 NRD contained 15 769 527 weighted index admissions for medical (1 825 405 weighted readmissions) and surgical (590 121 weighted readmissions) reasons (Table 1). The all-cause 30-day medical readmission rate was 23.1% for patients with SMI and 13.8% for patients without SMI. The all-cause 30-day surgical readmission rate was 19.3% for patients with SMI and 9.4% for patients without SMI. In the multivariate logistic regression models, even after controlling for clinical, demographic, and hospital characteristics, the odds of a patient with SMI being readmitted were greater than for patients without SMI (adjusted odds ratios, 1.80 [95% CI, 1.77-1.83] for medical and 1.95 [95% CI, 1.90-1.99] for surgical readmissions; P < .001) (Table 2).
Table 1. Index Admission Characteristics for Patients Without and With a Serious Mental Illnessa.
| Characteristic | Patient Group | P Valueb | |
|---|---|---|---|
| No Comorbid SMI Diagnosis (n = 5 767 218) | Comorbid SMI Diagnosis (n = 561 395) | ||
| Demographic Characteristics | |||
| Age, y, No. (%) | |||
| 18-35 | 498 948 (8.6) | 71 362 (12.7) | <.001 |
| 36-50 | 847 357 (14.7) | 125 450 (22.3) | <.001 |
| 51-65 | 1 728 310 (30.0) | 206 257 (36.7) | <.001 |
| 66-85 | 2 225 417 (38.6) | 139 305 (24.8) | <.001 |
| ≥86 | 467 186 (8.1) | 19 021 (3.4) | <.001 |
| Sex, No. (%) | |||
| Male | 2 742 164 (47.5) | 331 514 (59.1) | <.001 |
| Female | 3 025 054 (52.4) | 229 881 (40.9) | <.001 |
| Primary insurance, No. (%) | |||
| Private | 1 504 810 (26.1) | 93 000 (16.6) | <.001 |
| Medicaid | 720 282 (12.5) | 129 923 (23.1) | <.001 |
| Medicare | 3 063 096 (53.1) | 291 569 (51.9) | <.001 |
| Self-pay | 244 349 (4.2) | 25 286 (4.5) | <.001 |
| Other | 186 712 (3.2) | 16 440 (2.9) | <.001 |
| Hospital location, No. (%)c | |||
| Large urban (≥1 million) | 1 668 615 (28.9) | 185 863 (33.1) | <.001 |
| Small urban (<1 million) | 3 277 937 (56.8) | 306 031 (54.5) | <.001 |
| Micropolitan (10 000-50 000) | 442 284 (7.7) | 38 976 (6.9) | <.001 |
| Rural (<10 000) | 378 382 (6.6) | 30 525 (5.4) | <.001 |
| Income quartile, No. (%)d | |||
| 0-25th percentile ($1-$39 999) | 1 598 978 (28.2) | 176 403 (32.0) | <.001 |
| 26th-50th percentile ($40 000-$50 999) | 1 523 306 (26.9) | 148 474 (26.9) | .26 |
| 51st-75th percentile ($51 000-$65 999) | 1 318 574 (23.3) | 122 895 (22.3) | <.001 |
| 76th-100th percentile (≥$66 000) | 1 227 098 (21.7) | 103 559 (18.8) | <.001 |
| Clinical Characteristics | |||
| Medical admission, No. (%) | 3 568 081 (61.9) | 430 620 (76.7) | <.001 |
| Surgical admission, No. (%) | 2 199 137 (38.1) | 130 775 (23.3) | <.001 |
| Discharge disposition, No. (%) | |||
| Routine | 3 719 694 (64.5) | 334 515 (59.6) | <.001 |
| Transfer to short term hospital | 49 086 (0.8) | 6 212 (1.1) | <.001 |
| Transfer to other or skilled nursing facility | 876 989 (15.2) | 117 397 (20.9) | <.001 |
| Home health care | 1 072 864 (18.6) | 90 355 (16.1) | <.001 |
| Against medical advice | 45 599 (0.8) | 12 521 (2.2) | <.001 |
| Unknown | 2 986 (0.1) | 395 (0.1) | <.001 |
| No. of Elixhauser comorbidities, mean (SD) | 2.71 (1.9) | 3.83 (2.0) | <.001 |
| Elixhauser comorbidities, No. (%)e | |||
| Hypertension | 3 376 903 (58.6) | 303 749 (54.1) | <.001 |
| Fluid and electrolyte disorders | 1 630 551 (28.3) | 187 291 (33.4) | <.001 |
| Diabetes, uncomplicated | 1 245 148 (21.6) | 115 083 (20.5) | <.001 |
| Chronic pulmonary disease | 1 146 204 (19.9) | 155 456 (27.7) | <.001 |
| Deficiency anemias | 1 063 448 (18.4) | 115 965 (20.6) | <.001 |
| Obesity | 927 117 (16.1) | 105 158 (18.7) | <.001 |
| Renal failure | 815 062 (14.1) | 71 057 (12.6) | <.001 |
| Hypothyroidism | 718 815 (12.5) | 82 684 (14.7) | <.001 |
| Hospital Characteristics | |||
| Bed size, No. (%) | |||
| Small | 830 648 (14.4) | 80 221 (14.3) | .02 |
| Medium | 1 652 176 (28.6) | 156 897 (27.9) | <.001 |
| Large | 3 284 394 (56.9) | 324 277 (57.8) | <.001 |
| Location, No. (%)b | |||
| Large urban (>1 million) | 3 340 377 (57.9) | 340 522 (60.6) | <.001 |
| Small urban (<1 million) | 2 009 655 (34.8) | 179 884 (32.0) | <.001 |
| Micropolitan (10 000-50 000) | 300 115 (5.2) | 29 127 (5.2) | .62 |
| Rural (<10 000) | 117 071 (2.0) | 11 832 (2.1) | <.001 |
| Teaching status, No. (%) | |||
| Metropolitan nonteaching | 1 752 819 (30.4) | 171 350 (30.5) | .04 |
| Metropolitan teaching | 3 597 213 (62.4) | 349 086 (62.2) | .01 |
| Nonmetropolitan | 417 186 (7.2) | 40 959 (7.3) | .09 |
| Ownership, No. (%) | |||
| Nonfederal government | 750 440 (13.0) | 72 642 (12.9) | .12 |
| Private, not for profit | 4 097 681 (71.1) | 396 600 (70.6) | <.001 |
| Private, for profit | 919 097 (15.9) | 92 153 (16.4) | <.001 |
Abbreviation: SMI, serious mental illness.
Includes 6 328 613 participants. Percentages have been rounded and may not total 100.
P < .05 was considered significant (calculated using χ2 and 2-tailed t tests).
From the American Hospital Association’s urban-rural designation of the hospital and derived from the 12 categories of Urban Influence Codes.
Data were missing for some participants.
Includes the most common 8 comorbidities.
Table 2. Association of SMI Diagnosis With Odds of Readmission.
| Readmissions | Weighted No. of Participantsa | OR (95% CI)b | |
|---|---|---|---|
| Unadjusted | Adjustedc | ||
| 30-d Medical | 1 825 408 | 1.88 (1.85-1.90) | 1.80 (1.77-1.83) |
| 30-d Surgical | 590 121 | 2.33 (2.28-2.38) | 1.95 (1.90-1.99) |
Abbreviations: OR, odds ratio; SMI, serious mental illness.
Includes 6 328 613 participants.
All ORs are significant at P < .001.
Controlled for demographic (age, sex, primary insurance, income quartile, residential location), clinical (primary admitting diagnosis [all patient refined diagnosis related group], discharge disposition [routine, transfer to short-term hospital, transfer to skilled nursing facility, or home health care], Elixhauser comorbidity [excluding depression and psychosis]), and hospital characteristics (bed size, location, teaching status, ownership).
Discussion
In this first study, to our knowledge, of a nationally representative sample of adult readmissions, we found the odds of an unplanned 30-day readmission were nearly 2 times greater for patients with SMI. These results are slightly greater than other smaller state or hospital studies,4,5 which have estimated 1.60 greater odds of medical readmission for patients with SMI and 1.24 greater odds of readmission for patients with diabetes and SMI. These findings suggest that the Centers for Medicare & Medicaid Services Hospital Readmissions Reduction Program should attend to the population of patients with SMI to reduce unplanned readmissions. Further research is needed to identify interventions such as transitional care to prevent readmissions for these patients. This study is limited in that the NRD uses state-specific identifiers that cannot follow up patients across states, which may underestimate readmission rates. Our findings warrant further investigation of hospital, system, and community factors influencing variation in readmission rates for patients with SMI.
References
- 1.Mitchell AJ, Malone D, Doebbeling CC. Quality of medical care for people with and without comorbid mental illness and substance misuse: systematic review of comparative studies. Br J Psychiatry. 2009;194(6):491-499. doi: 10.1192/bjp.bp.107.045732 [DOI] [PubMed] [Google Scholar]
- 2.Centers for Medicare & Medicaid Services 2015. measure information about the 30-day all-cause hospital readmission measure, calculated for the value-based payment modifier program https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeedbackProgram/Downloads/2015-ACR-MIF.pdf. Updated March 2017. Accessed March 6, 2018.
- 3.Germack H, Caron A, Solomon R, Hanrahan N Systematic review of medical-surgical readmissions in patients with co-occurring serious mental illness. Poster presented at: AcademyHealth Annual Research Meeting; June 24-26, 2018; Seattle, WA. [DOI] [PubMed] [Google Scholar]
- 4.Hanrahan NP, Bressi S, Marcus SC, Solomon P. Examining the impact of comorbid serious mental illness on rehospitalization among medical and surgical inpatients. Gen Hosp Psychiatry. 2016;42:36-40. doi: 10.1016/j.genhosppsych.2016.06.002 [DOI] [PubMed] [Google Scholar]
- 5.Chwastiak LA, Davydow DS, McKibbin CL, et al. The effect of serious mental illness on the risk of rehospitalization among patients with diabetes. Psychosomatics. 2014;55(2):134-143. doi: 10.1016/j.psym.2013.08.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Agency for Healthcare Research and Quality Overview of the Nationwide Readmissions Database. https://www.hcup-us.ahrq.gov/nrdoverview.jsp. 2017. Accessed June 29, 2017.
- 7.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. [DOI] [PubMed] [Google Scholar]
