Mortality Risk Stratification Can Predict Readmissions but Not Length of Stay

February 10, 2021
Anupam Ashutosh Sule, MD, PhD

Rachel Sarah Dada, MD

Jafar Dada, MD

The American Journal of Managed Care, February 2021, Volume 27, Issue 2

Mortality risk stratification can identify patients at higher risk of mortality and readmissions for prevention strategies. Other patients should be the focus of length-of-stay reduction strategies.


Objectives: To determine whether the mortality risk stratification (MORIS) strata can predict outcomes including mortality, readmission, and discharge disposition for specific diagnoses.

Study Design: Retrospective, observational study for hospitalized patients in 2016-2017 at an urban, medium-sized, community tertiary care hospital. All admitted patients with 1 of the following diagnoses were included in this study: acute myocardial infarction, chronic obstructive pulmonary disease, congestive heart failure, pneumonia, and sepsis.

Methods: No interventions were applied in this retrospective study. Data collected from patients admitted under 1 of the 5 diagnoses included mortality, length of stay (LOS), readmission, and discharge disposition.

Results: MORIS strata can predict condition-specific mortality and readmissions but not length of stay or discharge disposition.

Conclusions: Stewardship of resources is necessary to obtain high value in care. A long LOS, discharge to skilled nursing facilities, and unplanned readmissions contribute to a significant utilization of resources. The MORIS strata are useful in predicting disease-specific mortality and readmission, but they are not useful in predicting LOS or discharge disposition.

Am J Manag Care. 2021;27(2):66-71.


Takeaway Points

This study identifies a risk stratification tool that can help categorize hospitalized patients into those at a higher risk of death or readmission and those who are at low risk.

  • Patients at high risk should be monitored closely, with a greater allocation of resources for postdischarge care.
  • Surprisingly, patients who are low risk for death or readmission stayed just as long as those at higher risk.
  • This low-risk group of patients should be targeted for length-of-stay reduction strategies.


The cost of readmissions to the health care system is an estimated $17.4 billion in spending annually by Medicare alone.1 Readmissions are not only costly but also associated with adverse patient outcomes, including death. To drive down rates of hospital readmissions, CMS has publicly reported risk-standardized readmission rates for acute heart failure, pneumonia, and myocardial infarction since 2009.2 Causes of hospital readmission are commonly multivariable, including incorrect discharge planning, improper follow-up with a health care provider, length of stay (LOS), and comorbid conditions.3 Being able to accurately identify patients at highest risk for readmission would be beneficial not only to the patients but also to payers and medical providers. As part of the strategy to address readmissions, it is essential to rapidly identify patients at high risk for readmission to maximize appropriate, timely care and to direct resources to those at highest risk. Research regarding the implementation of a prediction scoring system and its effects on adverse outcomes has been emerging in recent years.1

A commonly used screening tool is the LACE index. The 4 variables included are LOS, acuity of the admission, comorbidities, and prior emergency department visits in the past 6 months. The LACE index is used to predict the risk of death or nonelective 30-day readmission following hospital discharge. A LACE score of 10 or higher is associated with a high risk of readmission or death. One considerable downside to the LACE score is that LOS cannot be factored in until the date of discharge, which limits real-time use.4

Another available screening tool is the HOSPITAL score, which is designed to predict the 30-day risk of potentially avoidable readmissions. The model includes 7 variables, including last available hemoglobin level before discharge (positive if < 12), discharge from an oncology service, last available sodium level before discharge (positive if < 135 mEq/L), any procedure performed during the hospitalization, index admission type, number of admissions within the past 12 months, and LOS (positive if ≥ 5 days). It was intended to allow hospitals to assign extra discharge and care transition services to patients who are most likely to be readmitted. The scoring system ranges from 0 to 13 points. With each additional point awarded, there is an increased risk of 30-day avoidable readmission.5

The mortality risk stratification (MORIS) score is a prediction rule comprising 24 risk factors including demographics (age, sex), past medical history (atrial fibrillation, leukemia, lymphoma, metastatic cancer, any other malignancy, neurological conditions, cognitive disorder, prior hospitalization within the past 365 days), admission characteristics (emergent admission, acute heart failure, acute respiratory failure, sepsis), and laboratory values (blood urea nitrogen, white blood count, platelet count, hemoglobin, lactate, serum albumin, arterial pH, serum troponin, arterial partial pressure of oxygen).6 Each risk factor is weighted, and a score is calculated for each patient at admission. The scores are stratified into 5 categories, with stratum 5 representing a very low risk of 30-day mortality and stratum 1 signifying very high risk of 30-day mortality.6 A follow-up study demonstrated the validity and feasibility of using this stratification at the time of admission.7 The MORIS strata were able to predict additional outcomes, including unplanned intensive care unit admission and rehospitalizations. We investigated the ability of the MORIS strata to predict in-hospital mortality, 30-day readmissions, LOS, and discharge to a skilled nursing facility (SNF) in patients with specific high-value diagnoses rather than in general hospital populations, as done in prior studies.6,7 Our null hypothesis was that the MORIS strata will not correlate with the rate of readmission, LOS, or discharge to an SNF.


Study Design

This was a retrospective, observational study that included all patients admitted between January 1, 2016, and December 31, 2017, with a diagnosis-related group (DRG) code for congestive heart failure (CHF), sepsis, pneumonia, chronic obstructive pulmonary disease (COPD), or acute myocardial infarction (AMI).

Data Extraction

Data for this study were gathered from the electronic medical record of an urban, community-based, moderate-sized, tertiary care hospital. Using a database query for the 5 DRGs, all patients meeting the inclusion criteria were identified. The same query also pulled the following data elements: MORIS stratum, mortality, readmission, disposition, and LOS. The names of patients, demographic information, and electronic medical identifiers were not obtained, and patient charts were not accessed by the research team. The study was approved by the St Joseph Mercy Oakland institutional review board (approval No. 2019006).

Statistical Analysis

Descriptive statistics were calculated. Associations between categorical variables were examined using χ2 tests. Differences between groups on continuous variables were tested using t tests or analysis of variance. P values less than .05 were considered significant. Multivariate logistic regression analysis was performed with MORIS stratum 1 as the baseline for comparison. All analyses were performed using SPSS version 22 software (IBM).


MORIS Strata Can Predict Condition-Specific Mortality

Too few mortality events occurred in patients admitted with pneumonia or COPD to be analyzed. For patients with AMI, CHF, and sepsis, those in MORIS stratum 1 had a significantly higher probability of mortality (Figure 1 and Table 1). For patients with AMI, mortality was 36.4% for those in MORIS stratum 1, which was significantly higher than for those in other categories (P = .001). For patients with CHF, mortality was 23.2% for those in MORIS stratum 1, which was significantly higher than for those in other categories (P < .0005). For patients with sepsis, mortality was 26.1% for those in MORIS stratum 1, which was significantly higher than for those in other categories (P < .0005). Odds ratios of all lower strata compared with MORIS stratum 1 are presented in Table 1. MORIS stratum 5 had too few events to analyze.

MORIS Strata Can Predict Condition-Specific Readmissions

For all 5 diagnoses, those in MORIS stratum 1 had a significantly higher probability of readmission than those in other categories (Figure 2 and Table 2). For patients with AMI, readmission was 42.9% for those in MORIS stratum 1, which was significantly higher than for those in other categories (P = .009). Similarly, MORIS stratum 1 readmission rates for other diseases were 53.5% for CHF (P < .0005), 64.3% for COPD (P < .0005), 65.2% for pneumonia (P < .0005), and 37.4% for sepsis (P < .0005). Odds ratios of all lower strata compared with MORIS stratum 1 are presented in Table 2. MORIS stratum 5 had too few events to analyze.

MORIS Cannot Predict Condition-Specific LOS or Discharge to SNF

For all 5 diagnoses, MORIS strata did not have an association with the LOS (Table 3). MORIS strata had an association with discharge to SNF only in patients with COPD. For patients with COPD, the rate of discharge to SNF was 71.4% for those in MORIS stratum 1, which was significantly higher than for those in other categories (P = .001) (eAppendix Table [available at]).


The MORIS strata were statistically significant for predicting mortality for patients with CHF, AMI, and sepsis. The mortality was too low for statistical analysis in patients with pneumonia and COPD. It is possible that a larger data set would yield significant results with these conditions also. MORIS stratum 1 was a strong predictor of 30-day readmissions for AMI, CHF, COPD, pneumonia, and sepsis. Thus, resources and efforts to reduce mortality and 30-day readmissions should be focused on patients in MORIS strata 1 and 2.

There was no association between the LOS or discharge to SNF and the MORIS strata (except for discharge to SNF in patients with COPD, which was higher in MORIS stratum 1). Thus, patients who were not at increased risk of mortality or readmission were utilizing resources similar to the patients who were at increased risk. Any effort to decrease the LOS or avoid SNF utilization should be focused on patients in MORIS strata 3, 4, and 5.

Although the original LACE study found LOS to be independently associated with readmission within 30 days of discharge, we did not observe any correlation between LOS and readmission except for AMI (unpublished data).8 The original LACE validation study concluded that the LACE index had moderate discrimination for early mortality and readmission.8 Robinson and Hudali demonstrated that the HOSPITAL score was a good indicator for 30-day hospital readmission, but the LACE index was a relatively weaker predictor for 30-day hospital readmission for the same patient population.9 The HOSPITAL score was found to have the capacity to identify patients at high risk for 30-day readmission with moderately high discrimination and have good discriminatory power when predicting potentially avoidable readmissions.5 The MORIS strata should be compared with the HOSPITAL score to determine the ability of both systems to predict outcomes of interest.

Patients in MORIS stratum 1 are at higher risk of mortality and should be placed in closely monitored beds with a hospitalist program and multidisciplinary rounds.10 Those in MORIS strata 1 and 2 have a higher likelihood of readmission and should receive extended nurse-led postdischarge transition of care services, including early postdischarge follow-up, remote monitoring, focused patient education, and provision of social needs via a multidisciplinary team.11 Patients in MORIS strata 3 and 4 have a low risk of adverse outcomes, and efforts should be made to discharge these patients home sooner. Patients in MORIS stratum 5 may be appropriate for observation rather than admission. These goals can be achieved by operations research to improve patient flow, use of data analytics to match capacity and demand, setting of standards to avoid delays in progression, and constant learning to mitigate failures.12


This study was a retrospective, observational study of data from a single hospital. The low mortality rates in the study population resulted in some samples that were too small to analyze for significance (eg, the low mortality rate for patients with COPD and pneumonia). The findings should be validated in a multihospital population to ensure an adequate sample size for the study of diagnosis-specific mortality prediction. This study did not adjust for confounding variables, such as racial and socioeconomic disparities, which may independently affect readmissions.


This study demonstrates that patients can be stratified using MORIS to optimize resource allocation. Patients in MORIS stratum 1 are at higher risk of mortality and should be the target population for in-hospital mortality prevention strategies. Those in MORIS strata 1 and 2 are at higher risk of 30 day-readmission and should be the focus of readmission reduction interventions. Patients in MORIS strata 3 and 4 have a low risk of adverse outcomes, and efforts should be made to discharge these patients home sooner. Patients in MORIS stratum 5 may be appropriate for observation rather than admission. These insights need to be validated in a prospective study with interventions for specific outcomes focused according to patients’ MORIS strata.


The authors thank Karen Hagglund, MS, for statistical analyses.

Author Affiliations: St Joseph Mercy Oakland (AAS, RSD, JD), Pontiac, MI.

Source of Funding: None.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (AAS, RSD, JD); acquisition of data (AAS); analysis and interpretation of data (AAS); drafting of the manuscript (AAS, RSD, JD); critical revision of the manuscript for important intellectual content (AAS, RSD, JD); administrative, technical, or logistic support (AAS); and supervision (AAS).

Address Correspondence to: Anupam Ashutosh Sule, MD, PhD, St Joseph Mercy Oakland, Administration Ste, 44405 Woodward Ave, Pontiac, MI 48341. Email:


1. Auerbach AD, Kripalani S, Vasilevskis EE, et al. Preventability and causes of readmissions in a national cohort of general medicine patients. JAMA Intern Med. 2016;176(4):484-493. doi:10.1001/jamainternmed.2015.7863

2. Hospital quality initiative: outcome measures. CMS. Accessed January 5, 2021.

3. Hasan O, Meltzer DO, Shaykevich SA, et al. Hospital readmission in general medicine patients: a prediction model. J Gen Intern Med. 2010;25(3):211-219. doi:10.1007/s11606-009-1196-1

4. Gruneir A, Dhalla IA, van Walraven C, et al. Unplanned readmissions after hospital discharge among patients identified as being at high risk for readmission using a validated predictive algorithm. Open Med. 2011;5(2):e104-e111.

5. Donzé JD, Williams MV, Robinson EJ, et al. International validity of the HOSPITAL score to predict 30-day potentially avoidable hospital readmissions. JAMA Intern Med. 2016;176(4):496-502. doi:10.1001/jamainternmed.2015.8462

6. Cowen ME, Strawderman RL, Czerwinski JL, Smith MJ, Halasyamani LK. Mortality predictions on admission as a context for organizing care activities. J Hosp Med. 2013;8(5):229-235. doi:10.1002/jhm.1998

7. Cowen ME, Czerwinski JL, Posa PJ, et al. Implementation of a mortality prediction rule for real-time decision making: feasibility and validity. J Hosp Med. 2014;9(11):720-726. doi:10.1002/jhm.2250

8. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ. 2010;182(6):551-557. doi:10.1503/cmaj.091117

9. Robinson R, Hudali T. The HOSPITAL score and LACE index as predictors of 30 day readmission in a retrospective study at a university-affiliated community hospital. PeerJ. 2017;5:e3137. doi:10.7717/peerj.3137

10. Whittington J, Simmonds T, Jacobsen D. Reducing hospital mortality rates (part 2). Institute for Healthcare Improvement. 2005. Accessed August 17, 2018.

11. Boutwell A, Hwu S. Effective interventions to reduce rehospitalizations: a survey of the published evidence. Institute for Healthcare Improvement. 2005. Accessed August 17, 2018.

12. Rutherford PA, Provost LP, Kotagal UR, Luther K, Anderson A. Achieving hospital-wide patient flow. Institute for Healthcare Improvement. 2005. Accessed August 17, 2018.