High Intensive Care readmission rates, risk factors and possible solutions

High Intensive Care readmission rates, risk factors and possible solutions

A Quality improvement Project held at Lancashire teaching hospital

 

Abstract

Background

It is well known that early readmission to intensive care during the same hospital admission is associated with increased mortality, morbidity, and increased length of stay in intensive care and overall hospital stay. A review of the readmission rate at Royal Preston Hospital Intensive Care found it to be always been below the national average apart from August 2020 to August 2021, which may be due to the effect of Covid. Reducing the early readmissions to Preston critical care is part of the Trust Big Plan strategy for Consistently Delivering Excellent care.

 

Method

I collected the number of patients who were readmitted within 48 hours to intensive care between August 2020 and August 2021 from the ICNARC database. After that, we looked at the clinical information on the QuadraMed computer system. For each patient I deeply investigated the documentation behind the decision-making process and reason for the first admission, clinical interventions, and communications with the other specialties during the intensive care admission and after discharge in addition to the cause of readmission.

 

Results

There were 35 patients found to be readmitted within 48 hours. Most patients (46%) who were readmitted to intensive care during that period were aged 50 year or less, compared to 17% aged 70 years old or above and the rest 37% aged between 50 and 70 years old. Patients admitted in an emergency setting were found to have high risk of readmission compared to the elective admitted patient, 94 % vs 6 % respectively. The percentage of patients who had a high NEWs score (6 or above), one hour before discharge was approximately 70% compared to 40% when the patient was seen earlier by a consultant during the ward round. Another important correlation was the high WBCs and CRP in patients likely to be discharged within 24 hours – approximately 30% had high levels of CRP and WBCs 24 hours before being discharged to the ward. There was poor communication and handover to the ward, as 86% of the discharged patients were not handed over to the ward doctor, but they had a written discharge summary. The main cause of readmission was respiratory failure including Covid-related patients, but the Covid was not the leading cause of readmission despite it was being the main cause of the primary intensive care admission in this group. Many of those patients were discharged successfully from intensive care (86%) compared to 14% who died in ICU. After those findings significant changes were made in the clinical pathway and communication tools aiming to decrease the readmission rates to intensive care.

 

Conclusion

Early unplanned intensive care readmission is associated with high mortality rate, long hospital and intensive care stay and increased financial burden on the trust. Some important simple and achievable interventions like: NEWs score 1 hour before discharge, 24 hours and inflammatory markers checks before discharge and proper hand over using the SmartPage can decrease the rate of intensive care readmission but cannot totally prevent it.

 

Introduction and problem

Readmission to intensive care is a marker of poor outcome. The aim of the Quality improvement Project was to answer the following question: Why are the intensive care readmission rates in Royal Preston Hospital higher than the national rates? The simplistic answer would be to simply state that the readmission rates have always been increasing at Preston Hospital Intensive Care Unit due to increased workload and high patient flow! However, my primary objective was to look for a deeper understanding; the factors associated with this alarming rate of readmission and to prevent the preventable.

Royal Preston Hospital Intensive care is the largest Critical Care in Lancashire and South Cumbria, with a total capacity of 34 beds, with a plan to expand to 50 beds. Royal Preston Hospital intensive care provides tertiary care for different specialties as neurocritical care, major trauma, vascular and oncology services for a population of 1.5 million people in addition to providing the intensive care support for the medical patients in this catchment area. Royal Preston Intensive Care has a large team of more than 300 experienced doctors, nurses, physiotherapists, radiographers, pharmacists and other allied health professionals (AHPs). In our unit there are 19 intensive care consultants who together ensure we provide a consultant-led service 365 days a year, supported by over 30 trainees, specialty doctors and international medical graduates.

 

Background

Readmission to intensive care is not only a burgeoning risk to the health of the patient but also increased stress on the families involved. Since 2010, when the Patient Protection Affordable Care Act created the Hospital Readmission Reduction Programme to penalise hospitals with high readmission rates in America; enormous efforts, trials and studies were done to alleviate the problems that high readmission rates caused (2). As it is well known, intensive care stay is more costly compared to ward care so, most of these efforts assessed the readmission rates to intensive care in different ways using a plethora of approaches and tolls to decrease the Intensive care readmission rates.

The majority of the hospitals have introduced the outreach and acute care team services aiming to support the acutely unwell patients and to monitor patients who have been recently discharged from ICU to decrease the intensive care readmission rates. Some recent studies found that the introduction of the outreach team decreases the intensive care readmission rates from 5.1% to 3.3% p 0.05)(9). On the other hand, other studies were found to confirm that outreach services appear not to affect re-admission rates (10). Thus, determining the main risk factors associated with preventable ICU readmission as well as developing tools to alert the intensive care physician and avoid premature discharge and lowering readmission rates has been mainstay of ICU research over the past decade (6).

Al-Jaghbeer et al in their study of potentially preventable ICU readmissions, categorised readmissions as preventable or non-preventable. The percentage of potentially preventable readmitted patients to intensive care with obvious precipitating factors was 11.8%. According to their findings, Preventable readmissions were caused by system error, management error, procedure events, diagnostic errors and medication errors (7). These preventable readmissions had shorter ICU lengths of stay compared to the non-preventable readmissions and most of them were ultimately discharged to the ward (7). Despite lots of work done to evaluate the ICU readmission as a potential quality marker and indicator for intensive care performance, there have been few studies examining whether ICU readmissions are preventable or not. Indeed, the Al-Jaghbeer et al study was one of the few pioneering studies to do so (7). Other retrospective analysis and review of ICU readmission rates have found the percentages of preventable intensive care readmission rates were approximately similar to Al-Jabbeer et la study (5).

Other studies have used more complex algorithmic approaches to predict and identify the risky patients for readmission. Using across-sectional machine learning approach algorithm achieved good discrimination better and above that of the treating clinicians. Moreover, it was found to be superior to Stability and Workload Index for Transfer (SWIFT) score with Area under the receiver operating characteristic curve (AUROC) of 0.7095 (SE 0.0260) vs (AUROC) of 0.6082, (SE 0.0249), respectively (11).

Most of the studies that assessed intensive care readmission rates relied only on the information available at the time of discharge. Recently, M Loreto et al performed an interesting database analysis of 9926 intensive care patients of them, 658 patients (6.6%) were readmitted during the same hospital stay(5). The conclusion of their findings was that” basal characteristics and information collected at the time of patients` admission can enable accurate predictions of intensive care readmission” (5). As a consequence, prediction of   patients who are at high risk of intensive care readmission should start early once the patient was admitted to intensive care and should not depend only on the discharge information.

In their literature review of 13 articles, McNeill & Khairate found the Intensive care readmission rates varies between 1.5% and 13.4%(3) compared to 4% and 11% in other studies (5). The United Kingdom intensive care readmission rate is approximately 1.2% (1). The main identified reason for intensive care readmission was respiratory failure (18%-59%) and 31% of those patients required intubation and ventilation (3). In their review, McNeil& khairat found that cardiac related readmissions were the second most common cause of readmission after respiratory failure(3). In their retrospective analysis of readmission and mortality in a surgical intensive care unit, S Utzonilo et al found Respiratory failure as a reason for readmission implied a 44% risk of death (P < .001(4). Other important risk factors for readmissions are, age, male sex and decreased GCS (3). Furthermore, multiple comorbidities and high APACHE score were found to be important risk factors (12).

Using a noncomplex practical tool to identify high risk patient for readmission to intensive care like NEWs score was found to be very effective and carries a sensitivity of 93.6% and specificity of 82.2 %( 3). Maximilian et al developed RISC Score as a tool to predict patients who carries readmission risk in surgical intensive care (13).

 

Method and design

The data was collected retrogradely and the numbers of patients who were readmitted to intensive care between August 2020 and August 2021 were collected from the ICNARC.

Inclusion and exclusion criteria. All readmitted patients within 48 hours were included and there were no exclusion criteria.

The information regarding each patient was reviewed on the QuadraMed electronic information system. A specific Performa was created which included:

  1. Admission data: Patient age, sex, comorbidities, type of admission (emergency or elective) causes of first admission, physiotherapy interventions and length of stay.
  2. Before discharge data, we looked at blood investigation results 24 hours before discharge, decision to discharge the patient by the consultant ward round, was the patient discharged within 4 hours or not and NEWs score before the actual time of discharge. Communication and handover to the ward team was involved as well.
  3. Readmission data: length of stay in the ward, how fast was the response of the ward team once the patient had deteriorated, cause of deterioration and the time that was taken before referring the patient to ICU or outreach. The ICU interventions, length of stay and out come after readmission were involved as well.

After data collection, statistical analysis was done to compare the results to other studies and to find the main factors associated with high readmission rates and to prevent the preventable ones.

 

Findings

Most of our findings from the data analysis corroborated with data from similar projects and systematic reviews. The main difference that I found is the age of the readmitted group. Generally, the literature suggests that increased age is associated with high risk of readmission, but in Preston Intensive care unit the majority of the readmitted patients are aged 50 years or less (46%), compared to 37% and 17% for patients aged 50-70 and more than 70 respectively (figure 1). This may be due to COVID which affected the admission criteria in favour of younger patients. An interesting finding was that approximately one third (34%) of the re-admitted patients have been fit and well before the ICU admission and two thirds (66%) have had different comorbidities (figure 2). The most common comorbidity was HTN (47%), respiratory related problems (COPD and Asthma) in 43%, patients who had DM represented 30% and 26% had CKD. Approximately 21.7% of the readmitted patients with comorbidities had psychological problems, 17% had spinal cord pathology (stenosis and spina bifida ) and 1 patient had myasthenia gravis.

Through the readmitted group, COVID was the main cause of the first admission which represented 37%, while both medical and surgical related admissions were equal and represented 23 %. only 17% of the readmitted group was admitted primarily due to neurosurgical related pathologies. The main cause of readmission was respiratory failure (collectively 80%) with 57% non COVID related and only 23% was related to COVID (figure 3). The other causes for readmission (hypotension,pain management and agitation) represented only 20 %(figure 4).

The cause for the first admission and readmission was found to be the same in 37% of the readmitted patients. Interestingly, COVID represented 61.5% as the same cause of the first admission and readmission, (Figure 5).

63% of patients who were readmitted to intensive care during that time were men while the female represented 17% only.

All the readmitted patients had clear escalation plans during the first admission with 94% found to be for full escalation compared to 6% as level 2. This was consistent across patients whether they were originally elective or emergency cases (94%vs 6%) respectively.

Approximately, 91% of the readmitted patients were discharged between 07:00 and 21:59 and 9% discharged after 22:00 which was found to be higher than the national rate (1.9%). On the other hand, 60% were readmitted between 08:00 and 19:59 and only 40% were readmitted after 20:00. Only 31% of the readmitted patients were discharged from ICU within the 4 hours after the decision for discharge was taken during the first admission which may reflect how busy was the wards due to COVID patients.

Length of stay was found to be less than 7 days in 71% during the first admission compared to 66% of the readmitted group. Only 14% of the readmitted patients stayed more than 14 days (Figure 6).

In patients who were readmitted WBC and CRP trends were worsening in 35% and 23% of patients respectively 24 hours before discharge.

NEWs score was also found to be dramatically different between time of discharge and during the decision-making process. The NEWs score was found to be more than 6 in 71% of patients at the time of discharge compared to 43% at the time of the Consultant ward round. Only 36% of the high NEWs scores were escalated to the intensive care doctor before discharge and 64% were not subsequently reviewed. Only 14% of the readmitted group was handed over to the ward doctor leaving 86% discharged to the ward without proper handover.

In the ward, 46% of the readmitted group between August 2020 and August 2021 deteriorated in the first 12 hours of ward stay. Further analysis of those patients who deteriorated in the first 12 hours, 69% deteriorated in the first 8 hours which may reflect premature discharge (Figure 7).

More than 90% of the readmitted patients had been reviewed by-ward doctors of those, 25% were reviewed before deterioration and 75% reviewed after deterioration. Approximately 85.7% of those patients were referred to ICU within 2-4hours of deterioration, but the rest was admitted with no clear documentation about the time of referral and discussion with ICU team.

As the main cause of the readmission was related to respiratory problems, 83% required respiratory support (46% NIV, 23% intubation and 14% HFNC). Only 8% were readmitted for vasopressor support while patient s who were readmitted for pain, agitation and suction represented a similar percentage of 3% for each sector.

86% of the readmitted group were discharged to the ward and only 14% died in ICU after readmission, 5.7% of them died within 24 hours with EOL after readmission.

 

Limitation and difficulties

Between August 2020 and August 2021, the Covid peaks may affect the discharge and readmission rates to ICU. Also this may affect the documentation in the ward and ICU.

 

Discussion.

Early readmission within 48 hours to ICU may reflect early and premature discharge (14). Despite the literature describing readmission to ICU as being heavily associated with high mortally rates and prolonged length of stay (15), our QIP showed that Length of stay after readmission was actually reasonably short and mainly less than 7 days (66%), approximately similar to the first admission stay.

Due to the high selective criteria for patient admission to Preston ICU, the majority of the readmitted patients were young and 34% were fit and well. This may be influenced by COVID as well.

As has been found in similar studies and research articles, the most common comorbidities were respiratory, DM, hypertension and chronic kidney diseases. Similarly, the most common cause of readmission was related to respiratory problems. Despite Covid infection being the leading cause of the primary admission, it was not the main cause of readmission related respiratory conditions (hospital acquired pneumonia or the residual of ventilator associated pneumonia were stronger predictors of readmission). As a consequence, 83% of the readmitted patients required respiratory support – the majority of which was managed with HFNC and NIV and only 23% required intubation and ventilation. In addition to this, approximately one third of the readmitted patients had a worsening trend of WBCs and CRP 24 hours before discharge. Furthermore, 71% of patients NEWs were scoring higher than 6 just before discharge. Considering all of the previous factors, this group of patients may have preventable causes and their admission to ICU may be prevented if the patient was assessed well before discharge and if appropriate actions were taken (17).

The 4-hour target to discharge the patient from ICU was not achieved in approximately 69% of the readmitted patients (We did not look at the data of patients who have been discharged but not readmitted) which may reflect the intensive care unit and hospital in general being overwhelmed by the plethora of Covid-related admissions.

Surprisingly, most of the readmissions happened during the daytime, not during the night-time and all readmissions were discussed with the intensive care consultants.

Human factors were found to be an important element in the mystery of increased readmission rates. In 64% of the readmitted patients, it could be said that the communication between the Intensive care team and the ward team was not optimal and there was no appropriate handover. If there was a proper discussion and handover of the discharged patients, the readmission rates would be markedly decreased (16). Conversely, there was adequate and timely communication of patient deterioration from the ward team to the outreach and intensive care teams (85.7%) regarding escalation. This discrepancy in communication may be due to the ease of accessing the ICU from other departments compared to the difficulty in communicating with other teams for discharge of an ICU patient especially after 17:00.

The escalation plan regarding readmissions to ICU is another important factor the readmission rates as if proper decisions was taken before discharging the patient to the ward, some of the readmitted patients may not be for readmission.

 

Changes and interventions

These important findings were discussed in the audit meeting and a committee, which contained an intensive care consultant, intensive care trainee representative, discharge co-coordinator and staff nurses. The committee recommendations were discussed with the IT analyst and the changes in the discharge pathway were sent to the QuadraMed changing board. The changing board has discussed the changes and approved the following changes to occur in the discharge pathway

  • Nursing discharge check list needs to move to appropriate place on the QuadraMed (as it can be missed after prolonged admission).
  •  A box in the nurse discharge check list that asks if a MCA and DOLS is required and if yes tick if completed
  • NEWs score within 1 hour before discharge the patient from ICU
  • A box in the nursing discharging check list regarding the action that was taken if the NEWs were found high (escalated to the doctor/band 6-7 nurse).
  • NEWs score box to be added to the doctors’ discharge pathway and an action plan should be documented according to the trust guidelines “If the NEWS is 5 or above this must be discussed with the ward team, a patient with a NEWS >7 must be reviewed by a senior clinician within critical care (trust guidelines”.
  • A separate box in the doctors` discharge pathway about the WBCs and CRP trends, if high, an action plan should be clear regarding repeating the bloods in the ward (CRP, WBCs, PCT) and the length of ABX course/micro discussion.
  • A separate box to be added to the discharging pathway for readmission plan (is the patient for re-admission or not)

To improve the communication between the intensive care team and the ward teams, we had a meeting with the SmartPage clinical trainer lead. This was followed by a training session to the Intensive Care team and the SmartPage was identified as a good communication tool to decrease the high readmission rates to intensive care.

 

Conclusion

Readmission to intensive care is associated with increased mortality, prolonged length of stay and increased financial burden. Our QIP data showed that premature discharge and human factors can play a vital role and if both were managed properly, it can decrease the readmission rates. Importantly, NEWs score is a very useful tool which is applicable and not associated with increased work load or financial requirements. Using the news score before discharge to identify patients who carry high risk for readmission is sensitive and specific. Usage of new technology tools (smart Pages) in communication may decrease the readmission rates.

 

References

  1. Guidelines for the Provision of Intensive Care Services (GPICS, Faculty of Intensive Care medicine &Intensive care society, 2019.
  2. Lin Y-W, Zhou Y, Faghri F, Shaw MJ, Campbell RH (2019) Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory. PLoS ONE 14(7): e0218942, 2019.
  3. Heidi Mcneill , Saif Khairat. mpact of Intensive Care Unit Readmissions on Patient Outcomes and the Evaluation of the National Early Warning Score to Prevent Readmissions: Literature Review, (JMIR Perioper Med 2020;3(1):e13782) doi: 10.2196/13782.
  4. Utzolino S, Kaffarnik M, Keck T, Berlet M, Hopt UT. Unplanned discharges from a surgical intensive care unit: readmissions and mortality. J Crit Care 2010 Sep;25:375-381. doi: 10.1016/j.jcrc.2009.09.009
  5. Melina Loreto , Thiago Lisboa , Viviane P. Moreira. Early prediction of ICU readmissions using classification algorithms, Computers in Biology and Medicine 118 (2020) 103636.
  6. Carolina R. Ponzoni, Thiago D. Correa, Roberto R. Filho, Ary Serpa Neto, Murillo S. C. Assunção, Andreia Pardini, and ˆ Guilherme P. P. Schettino. Readmission to the Intensive Care Unit: Incidence, Risk Factors, Resource Use, and Outcomes, A Retrospective Cohort Study. Ann Am Thorac Soc Vol 14, No 8, pp 1312–1319, Aug 2017
  7. Al-Jaghbeer MJ, Tekwani SS, Gunn SR, et al. Incidence and etiology of potentially preventable intensive care unit readmissions. Society of Critical Care Medicine journal, September 2016 , Volume 44, Number 9.
  8. Thomas Bice, ICU Readmissions: Good for Reflection on Performance, but not a Reflection of Quality, Crit Care Med. 2016 September ; 44(9): 1790–1791. doi:10.1097/CCM.0000000000001809.
  9. Alison J. Pittard. Out of our reach? Assessing the impact of introducing a critical care outreach service . Anaesthesia 2003;58:882–5.
  10. Leary, S. Ridley Impact of an Outreach team on re-admissions to a critical care unit, 18 March Anesthesia 2003 doi.org/10.1046/j.1365-2044.2003.03077
  11. Thomas Desautels,1 Ritankar Das,1 Jacob Calvert,1 Monica Trivedi,2 Charlotte Summers,3 David J Wales,4 Ari Ercole. Prediction of early unplanned intensive care unit readmission in a UK tertiary care hospital: a cross-sectional machine learning approach. BMJ Open 2017; 7:e017199. doi:10.1136/ bmjopen-2017-017199
  12. Malcolm Elliott, Linda Worrall-Carter, Karen Page. Intensive care readmission: A contemporary review of the intensive and critical care nursing http://dx.doi.org/10.1016/j.iccn. 2013.10.005
  13. Maximilian HammerStephanie D Grabitz https://pubmed.ncbi.nlm.nih.gov/32840427/ – affiliation-1Bijan Teja , Karuna Wongtangman , Marjorie Serrano , Sara Neves , Shahla Siddiqui , Xinling Xu , Matthias Eikermann. A Tool to Predict Readmission to the Intensive Care Unit in Surgical Critical Care Patients-The RISC Score.  Intensive Care Med. 2021 Nov;36(11):1296-1304.  doi: 10.1177/0885066620949164. Epub 2020 Aug 25.
  14. Pınar Öngel Çayören, Kerem Erkalp, Mehmet Salih Sevdi, Serdar Demirgan, Hüseyin Çayören, Ayşin Selcan. Evaluation of Readmitted Patients After Intensive Care Unit Discharge (Retrospective Study), Bagcilar Med Bull 2019;4(1):10-15 DOI: 10.4274/BMB.galenos.2019.19884.
  15. Nayana AminJV DivatiaVandana Agarwaland AP Kulkarni. Readmissions in a surgical intensive care unit: patient characteristics and outcome. : Indian Journal of Critical Care Medicine (Vol. 7, Issue 1) January-March 2003.
  16. Nelleke van Sluisveld, Gijs Hesselink, Johannes Gerardus van der Hoeven, Gert Westert, Hub Wollersheim, Marieke Zegers. Improving clinical handover between intensive care unit and general ward professionals at intensive care unit discharge, systematic review, Intensive Care Med (2015) 41:589–604.
  17. Andrew Gosling, Admission and discharge to adult critical care procedure, Lancashire Teaching Hospital guidelines, 2018

 

Authors

Mohammed Hatab, Peter Frank, Jasbir Chhabra, Marco Parolin and Alecia Wegstape

Correspondence to Mohammed Hatab mhatab@doctors.org.uk.

Categories: ARTICLES