Patient Safety Tip of the Week

May 10, 2011

Preventing Preventable Readmissions: Not As Easy As It Sounds



Preventing hospital readmission has been a hot topic since CMS announced there will be financial penalties for hospitals having poor track records on readmissions for certain conditions. The first batch of targeted conditions are congestive heart failure, MI and pneumonia and the clock starts ticking this October. The potential penalties apply not just to those three conditions but to all Medicare discharges. And the list of targeted diagnoses will expand in the future. In addition, those organizations forming Accountable Care Organizations (ACO’s) will also want to reduce avoidable readmissions as a major means of improving both cost and quality.


It’s been a while since we did a column focusing on readmissions. Our February 24, 2009 Patient Safety Tip of the Week “Discharge Planning: Finally Something That Works!” began to address the rehospitalization problem and focus on the hospital discharge process and we followed that up with Patient Safety Tips of the Week for April 7, 2009 “Project RED” and April 14, 2009 “More on Rehospitalization After Discharge”. In those columns, we discussed the New England Journal of Medicine (Jencks et al 2009) study on Medicare data that showed 19.6% of all patients discharged from an acute care hospital are rehospitalized within 30 days and the randomized study (Jack et al. 2009) from Project RED that documented considerable improvement in rehospitalization rates using a structured hospital discharge program. Recall that that study showed about a 30% reduction in rehospitalizations or ER visits after hospital discharge and saved about $412 per patient. We encourage you to go click the links to those columns to see the numerous strategies that are being employed to prevent readmissions.


Interventions utilized include the following:



But it’s a lot easier said than done. A few recent papers have highlighted the difficulties.


Though there are some predictors of rehospitalizations (certain DRG’s, presence of ESRD, long LOS, many prior hospitalizations), some have noted that the data strongly support the need to re-engineer the process for all patients, not just targeted ones. And in this age of limited resources it would be useful to be able to risk stratify patients by their likelihood of readmission. A much smaller fraction of the patients accounts for the bulk of the readmissions. Unfortunately, identifying them is not so easy.


One group (Allaudeen 2010) performed multivariate analysis on both clinical and nonclinical data on over 10,000 consecutive admissions to the medicine services at UCSF and identified a handful of risk factors that appeared to predict readmission. These included nonclinical factors like race and Medicaid as payor plus clinical factors like inpatient use of narcotics or steroids, cancer (with or without metastases), renal failure, congestive heart failure, iron deficiency anemia and weight loss. They note that they were not trying to put together a risk prediction rule and note that several prior attempts by multiple groups to do so have not been successful.


However, that same group (Allaudeen 2011) just published another paper noting the inability of providers to predict unplanned readmissions. They did a study over a 5-week period where they asked providers (physicians, nurses, case managers) to predict which patients were likely to be readmitted within 30 days. They also calculated the P(ra) score for the same patients. The latter is a tool the “Probability for Readmission” score which is widely used in managed care to predict subsequent utilization of resources. It turns out that none of the groups nor the P(ra) score performed very well in predicting readmissions. Physicians were the closest to the actual readmission rates but all the others significantly overestimated the risk. The other major finding was that the actual readmission rate was over 32%, far higher than one would expect from the literature (perhaps because their followup methodology was more thorough).


So are there any tools that are good at predicting readmission? Many are using the Risk Assessment Tool: the 8P’s from Project BOOST. We previously mentioned the P(ra) and in the UK they utilize the PARR++ and Combined Predictive Model. All are reasonable at predicting future use of healthcare resources but may not be specific for predicting readmission. Other groups (eg. EQ●Health Solutions, the Medicare QIO in Louisiana) are working with tools to assess readmission risk but there are no published results yet. And yet others (Amarasingham) have developed condition-specific readmission risk prediction tools. The latter for CHF has been followed by a 40% reduction in Medicare readmissions at Parkland Health & Hospital System (Betbeze 2011). It may also turn out that simpler tools such as the Frailty Index (Fried 2001) or the more extensive frailty index (Rockwood 2005) are good predictors of readmission, though they have not specifically been studied for that. A different form of Elder Risk Assessment (Takahashi 2011) was not able to predict readmissions from SNF’s to hospitals.



Another recent paper (Hansen 2011) has some equally disturbing findings about the discharge process. They did a study of over 1000 patients who were readmitted and compared them to about 1000 patients who were otherwise closely matched but not readmitted. Surprisingly, they found no association between readmission and documentation of most of the elements we consider to be key to the discharge process (eg. medication reconciliation, timely transmission of discharge summaries, specific discharge instructions, etc.). It’s not clear whether the problems lie with the documentation or whether those components are not as effective as we think.


Another study (Showalter 2011) looked at another “in vogue” intervention – provision of standardized electronic discharge instructions. They found that these failed to reduce readmissions or post-discharge ED visits within 30 days. A recent systematic review of the efficacy of computer-enabled discharge communication interventions (Motamedi 2011) found improvements in timeliness and physician/patient satisfaction and a perceived reduction in adverse events but noted that reporting on more important outcomes like readmission and mortality were inconsistent.



As an aside, there is also a good article (van Walraven 2011) and accompanying editorial (Goldfield 2011) in a recent Canadian Medical Association Journal issue about “avoidable” readmissions.






Links to some of our prior columns on readmissions:


February 24, 2009 “Discharge Planning: Finally Something That Works!

April 7, 2009 “Project RED

April 14, 2009 “More on Rehospitalization After Discharge”.







Jencks SF, Williams MV, Coleman EA.. Rehospitalizations among Patients in the Medicare Fee-for-Service Program. NEJM 2009; 360: 1418-1428



Jack BW, Chetty VK, Anthony D et al. A Reengineered Hospital Discharge Program to Decrease Rehospitalization: A Randomized Trial. Annals of Internal Medicine 2009; 150(3): 178-187



Allaudeen N, Vidyarthi A, Maselli J, Auerbach A. Redefining readmission risk factors for general medicine patients. J Hosp Med 2010; 6(2): 54-60



Allaudeen N, Schnipper JL, Orav EJ, et al. Inability of Providers to Predict Unplanned Readmissions. J Gen Intern Med 2011; Online First March 11, 2011



Hansen LO, Strater A, Smith L, et al. Hospital discharge documentation and risk of rehospitalisation. BMJ Qual Saf 2011; Published Online First: 22 April 2011



Showalter JW, Rafferty CM, Swallow NA, et al. Effect of Standardized Electronic Discharge Instructions on Post-Discharge Hospital Utilization. J Gen Intern Med 2011; Online First April 16, 2011



Motamedi SM, Posadas-Calleja J, Straus S, et al. The efficacy of computer-enabled discharge communication interventions: a systematic review. BMJ Qual Saf 2011; 20: 403-415



Project BOOST. Tool for Addressing Risk: A Geriatric Evaluation for Transitions.



The Kings Fund. Predicting and Reducing Re-admission to Hospital. Updated 1/04/10

PARR++ and Combined Predictive Model



Medicare QIO for Louisiana (EQ●Health Solutions). Discharge Risk Assessment.



Betbeze P. Know Your Readmission Risk Score. HealthLeaders Media, March 16, 2011



Amarasingham R. National Association of Public Hospitals and Health Systems. Parkland Health and Hospital System. Using an Automated Model to Identify Heart Failure Patients at Risk for 30-Day Readmission or Death Using EMR Data.



Fried LP, Tangen CM, Walston J; et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001; 56(3): M146-M156



Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fitness and frailty in elderly people. CMAJ. 2005 August 30; 173(5): 489–495



Takahashi PY, Chandra A, Cha S, Borrud A. The Relationship Between Elder Risk Assessment Index Score and 30-Day Readmission From the Nursing Home. Hospital Practice 2011; 39(1):   DOI: 10.3810/hp.2011.02.379     March 21, 2011



van Walraven C, Bennett C, Jennings A, et al. Proportion of hospital readmissions deemed avoidable: a systematic review. CMAJ 2011 183: E391-E402



Goldfield N. How important is it to identify avoidable hospital readmissions with certainty? CMAJ 2011 183: E368-E369













Tip of the Week Archive


What’s New in the Patient Safety World Archive