Skip navigation

Understanding why patients are readmitted ICUs


Improving the clinical effectiveness and cost efficiency of Intensive Care Units (ICUs) is increasingly important as the use of advanced medical technologies and increased numbers of older more acutely ill patients push up costs. Discharge criteria could shorten lengths of stay while not compromising the quality of care. But, there is concern that patients are being discharged "quicker and sicker'. As efforts to reduce ICU utilisation and length of stay become more common, can the causes and risk factors associated with unexpected readmission to ICU be identified? This question prompted a systematic review of the literature on unexpected ICU readmission.


AL Rosenberg and C Watts. Patients Readmitted to ICUs. A systematic review of risk factors and outcomes. Critical Care Reviews. Chest 2000 118: 492 - 502.

How was the question tackled?

Medline databases were searched for citations between January 1966 to June 1999. The search focused on adult medical and surgical ICU patients and excluded work related to coronary care units and paediatric ICUs. It was limited to work reported in the English language.

The review identified eight primary studies focusing on ICU readmission and eight multi-institutional ICU outcome studies for inclusion in the review. The eight primary studies were examined to determine whether they explicitly defined:

The review also considered whether the studies evaluated the quality of care that readmitted and non-readmitted patients received.

What did the review find?

1. ICU readmission rates.

Pooled data from the multi-institutional studies of ICU outcomes resulted in a mean ICU readmission rate of 7%. A similar rate (6%) was reported in some of the primary studies. These overall figures cover significant variations across hospitals - ranging from about 3% to about 9%. The studies found that teaching hospitals had higher readmission rates than non-teaching hospitals. No consistent change in ICU readmission rate over time appeared. The variability in rates was probably due to institutional factors, patient mix and perhaps different definitions of an ICU readmission.

2. Reasons for ICU readmission

Patients were readmitted to the ICU for a diagnosis or problem similar or identical to the initial ICU admission from19% to 53% of the time. (The lower rate probably results from the use of a different definition). Planned readmission accounted for 5% to 14% of the readmission and involved staged operations or post procedure monitoring.

The initial ICU admission most frequently associated with readmission included hypoxic respiratory failure, inadequate ventilatory or pulmonary toilet, upper-GI bleeding, neurologic impairment and sepsis. The most common reasons for ICU readmission were pulmonary problems, including hypoxia and inadequate pulmonary toilet. A variety of cardiac conditions (arrhythmia, congestive heart failure and cardiac arrest) followed by upper-GI bleeding and neurologic impairments were also common causes of ICU readmission.

Although the reasons for both admission and readmission varied by the type of ICU three points merit note. First, pulmonary disorders were consistently the leading or second leading cause of readmission. Second, recurrence of GI bleeding was the most common cause of readmission related to GI diagnosis. Third, several conditions commonly seen in ICUs (including toxic ingestion or drug overdoses and metabolic or electrolyte disorders) are rarely associated with ICU readmission.

Early readmission is potentially of particular interest because it may indicate patients discharged prematurely. Early readmission occurred in 22 to 30% of patients. Premature discharge was thought to have occurred in 22 to 42% of all readmitted patients. Only one primary study evaluated early readmission but found no significant differences in the cause of readmission or the quality of care.

3. Predictors of readmission

These were examined in six of the studies.

Not surprisingly, readmitted patients had significantly higher illness scores at the time of their initial admission and discharge from their first ICU course, compared to non- readmitted patients. Variables most commonly associated with ICU readmission included, fever, hypoxia, an elevated respiratory rate or heart rate, an admission diagnosis of upper GI bleeding and increasing age. Variables also associated with readmission were positive fluid balance, elevated PCO2, discharge hematocrit and positive blood cultures. The variables were measured throughout the course of the first ICU stay, and none were strongly correlated or consistently found to be associated with readmission. Medical patients were 10 to 40% more likely to be readmitted than surgical patients were.

4. Outcomes associated with ICU readmission.

Length of Stay for readmitted patients was at least twice as long as that for patients discharged from the ICU but not readmitted. Average hospital length of stay ranged from 35 to 47 days for readmitted patients, compared to 16 to 21 days for patients not readmitted. Length of stay was significantly longer for readmitted patients.

Mortality was significantly higher for patients readmitted to the ICU. Hospital death rates were 1.5 to almost 10 times higher among ICU readmission. Even when adjusting for severity and disease category two studies found that the odds of death remained six and seven times higher among readmitted patients.

=Several factors explain the higher death rates.

  1. First, readmitted patients appear to be sicker and thus have a higher risk of death.
  2. Second, the higher mortality among patients readmitted to an ICU may reflect a higher prevalence of patients receiving a form of ineffective care - or who may not respond to therapy.
  3. Third, increased mortality may reflect poor quality of care such as premature discharge, defined by clinical instability at the time of discharge.

5. ICU readmission and quality indicators

As efforts to reduce ICU utilisation and length of stay become more common it is increasingly important to identify patients at high risk of returning to ICU. For many patients, readmission may be a function of their failure to respond to treatment - rather than a reflection of poor quality of care or premature discharge. On the other hand low readmission rates may represent a failure to discharge patients resulting in prolonged ICU stays.

Unplanned readmission is often due to recurrent problems associated with a patient's specific disease and the inherent instability of a severely ill patient. However there has been no theoretical or experimental evaluation of the factors that might separate appropriate ICU readmission from those resulting from poor quality care. Other work has found that unplanned hospital readmission can be an indicator of poor quality care only when considered at the level of specific medical and surgical diagnoses. No evidence was found that hospital readmission were correlated with the overall quality of the hospital.

There is no agreed standard for an appropriate readmission rate to an ICU. It may be reasonable to assume that when a significant number of patients are readmitted within 48 hours the quality of care may be sub optimal. However no evidence indicates that a longer ICU stay would prevent readmission, nor can poor care outside the ICU be ruled out as a cause for readmission. Prompt return might indicate high quality of care: many readmitted patients are among the sickest in the ICU and a readmission may be a necessary. It is also possible that a low readmission rate may be an indicator of patients who are having inappropriately long stays.

What has the review told us?

The decision to discharge a patient from ICU is complex and frequently influenced by resource issues such as bed capacity (and length of stay), care alternatives (such as intermediate care units), and clinical guidelines (advocating new approaches to care). The review confirms that ICU readmission is associated with higher hospital mortality. Unstable vital signs at the time of ICU discharge are the most consistent predictor of ICU readmission. But there are no consistent data supporting the use of readmission rates as a measure of the quality of ICU care.

Few studies have evaluated the ICU discharge decision making process. It is worthy of further investigation as part of ongoing questions about quality control and resource utilisation. Development of ICU predictive tools to support clinicians in their decisions making could lead to reductions in readmission rates.