Introduction

Efficient use of intensive care units (ICUs) has become a top priority of hospitals worldwide as a result of the increased pressure on hospital budgets [1, 2]. An optimal patient flow is critical to ensure a high quality of care, given that ICUs are often subject to forward pressure from various internal sources, such as emergency departments or operating theaters [3, 4]. Early discharge from the ICU to the general ward is one strategy that can be used to relieve this pressure, but the successful implementation of this strategy requires close cooperation between a variety of healthcare professionals across different clinical settings [4–7].

A patient’s discharge from one specialty to another is a high-risk event in the care process and one where poor clinical handover between healthcare professionals leads to preventable errors and adverse events [8, 9]. Patients discharged from the ICU are particularly vulnerable to poor handovers due to the complex physiology of their health condition and the significant decrease in monitoring which occurs upon the transfer of these patients to a general ward [10]. These factors are particularly relevant for patients subjected to early discharge policies [7, 11].

Despite the availability of professional guidelines for ICU discharge [12–14], the quality of clinical handover practices varies between ICUs [15]. Several studies have identified deficits in the communication, coordination of care and information exchange between ICU and ward healthcare professionals [16–20]. These factors increase the risk of suboptimal ICU discharge and may result in severe adverse events, ICU readmissions, and increased mortality [6, 21, 22]. In a study conducted in the USA in 2003, Nishi and colleagues reported that 37.3 % of the ICU readmissions within 48 h were potentially preventable [16, 23]. Based on available data, it is estimated that a reduction of the readmission rate by 1 %, incorporating an overall mean ICU stay of 6.6 days, could save the U.S. government $1.4 billion per year [1, 23–25].

There are several strategies to improve clinical handovers between ambulance crew and emergency department [26, 27], between shifts [28, 29], and between the hospital and community setting [20, 30], as well as postoperative handovers [31]. However, a comprehensive evaluation of the effectiveness of interventions with the aim to improve inter-specialty handovers from the ICU to a general hospital ward is lacking. Niven and colleagues recently reviewed the effect of transition programs for patients discharged from an ICU which focused on post-ICU discharge interventions and excluded studies with a neonatal or pediatric population [9]. Better insight into effective interventions could guide healthcare professionals and policy-makers in the development and implementation of polices aimed at reducing patient mortality rates and costly readmissions [32–34].

The purpose of the study reported here was to systematically review interventions with the aim to improve the quality of patient handover between ICU and general ward healthcare professionals at ICU discharge and to evaluate the overall effects of these interventions.

Methods

The criteria for article inclusion and data analysis were pre-specified [35], and the protocol followed is given in Electronic Supplementary Material (ESM) 1. We followed the recommendations outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [36].

Data sources and searches

Using specific search terms (for details, see ESM Table 1), we searched for full-text intervention studies in the following databases: PubMed (including MEDLINE), CINAHL, PsycINFO, EMBASE, Web of Science, and the Cochrane Library. There were no restrictions based on publication date or language, but the presence of an English abstract was considered to be important. The authors’ personal files, references from included studies, and bibliographies of previously published related reviews were also searched to identify additional relevant studies (snowballing) [9, 10, 37].

Study selection

Two researchers (NS and GH) independently screened the titles and abstracts of all studies identified by the search strategy for their eligibility. A study had to meet all of the following inclusion criteria to be included in the review:

  1. 1.

    Inclusion of patients or healthcare professionals involved in the handover from the ICU to a step-down unit or ward.

  2. 2.

    Inclusion of an intervention explicitly describing one or more components that aimed to improve the handover of care between healthcare professionals from the ICU and those of a step-down unit or general ward.

  3. 3.

    Study design was experimental or quasi-experimental, such as a (cluster) randomized controlled trial, cohort study, or a non-controlled before–after study.

  4. 4.

    There was at least one process or outcome measure addressing the quality or safety of the discharge.

Studies not available in full-text format were excluded. When the title and abstract did not clearly indicate whether the inclusion criteria were met, a full-text copy was retained and reviewed.

The full text of the potentially relevant studies were retrieved and reviewed by two researchers (NS, GH). The inclusion criteria were applied a second time, and a final set of studies was identified for data extraction. Disagreement on inclusion was resolved by discussion; when no consensus could be reached, a third researcher (MZ) made the final decision.

Data extraction

Data from each study meeting the inclusion criteria were independently extracted by two researchers (NS, GH) using a pre-designed form modified from a checklist developed by Grimshaw and colleagues [38]. The extracted data described the objectives, underlying theory-based concepts, setting, study population, intervention characteristics, implementation activities, process evaluation, and outcome measures. Outcomes were divided into four pre-specified groups by the two researchers separately as: (1) use of care (e.g., ICU readmissions), (2) continuity of care (e.g., information accuracy), (3) adverse events, and (4) mortality. Any disagreement between the two researchers was resolved by discussion.

Quality assessment

The methodological quality was assessed by two researchers (NS, GH) independently. To ensure standardized scoring, we used a standardized form adapted from the Cochrane Effective Practice and Organization of Care (EPOC) Group’s Risk of Bias Criteria [39]. Methodological quality was assessed on 11 criteria, including (1) whether studies used random and concealed allocation, (2) whether the studies documented similar baseline characteristics and outcomes between the intervention and control group, (3) whether the studies described a strategy for handling missing data, (4) the likelihood of contamination between study groups, and (5) whether the criteria were free from selective outcome reporting. The decision on whether the criteria were fulfilled was resolved by discussion, or a final decision was made by the third researcher (MZ). Studies were given 1 point for each fulfilled criterion, with a maximum of 11 points. If information was inadequate or missing, the criterion was labeled ‘unknown’ and no point was given.

Data synthesis and analysis

The study outcomes, such as sample size, intervention characteristics, outcome measures, statistical significance, and direction of the effects observed, were assessed by two researchers (NS, GH) and organized in a tabular form.

The interventions were classified by two researchers (NS and GH) based on the definition of continuity of care by Hellesø and colleagues [17]. This classification consists of three elements: (1) the quality of information that is exchanged between healthcare professionals in terms of completeness, accuracy, and clarity; (2) the coordination of care between healthcare professionals in terms of the quality of assessment, planning, and organization of follow-up services and needs; (3) the communication between healthcare professionals in terms of personal and direct contact, accessibility, and timeliness [17, 20].

Results

Search results

Our initial search identified 6,591 records (Fig. 1), of which 5,268 remained following the exclusion of duplicates. Subsequent screening by title and abstract excluded 5,231 records. The remaining 37 full-text studies were retrieved and reviewed, of which 29 were excluded. Three articles were identified through snowballing. The final set of articles included in the review of consisted of 11 published studies that had undergone full-text extraction [40–50].

Fig. 1
figure 1

Summary of evidence search and selection

Full size image

Characteristics of included studies

A summary of the characteristics of the 11 studies included in this review is presented in Table 1. The study popul ation included neonatal, pediatric, and adult patients and their healthcare professionals. Patients were treated in neonatal, pediatric, general, medical, and/or surgical ICUs situated in various types of hospitals (tertiary, regional, metropolitan, teaching, university/academic, community-teaching, tertiary-referral). Ten studies were single-center studies, and one was conducted across multiple (n = 3) hospitals [47]. The sample size per study ranged from 46 to 4,951 participants for the intervention group and from 53 to 1,872 participants for the control group.

Table 1 Characteristics of the 11 studies included in the review
Full size table

The studies reported various outcomes (Table 2), although most studies reported an outcome related to use of care [40, 42–45, 47, 49, 50], mortality [40, 44, 45, 47, 49, 50], and continuity of care [41, 48–50]. One study reported adverse events as an outcome measure [46]. Statistical significant improvements were observed in two categories: continuity of care (reduced discharge delay, increased perceived accuracy of information, reduced time to finalize discharge letter) [41, 48, 49], and preventable adverse events [46]. Inconsistent effects were observed for various aspects of the use of care, namely, reduction of step-down unit length of stay (LOS), increase of transfers to higher care and increase of surgical procedures required [44, 45].

Table 2 Outcome measures and statistical significance of effects reported in the 11 studies included in the review
Full size table

Methodological quality

The overall methodological quality of the studies was relatively low, with an average score of 4.5 points out of 11 possible (ESM Table 2). In none of the studies was the allocation sequence randomly assigned and the allocation concealed. Six studies did not report similar baseline characteristics [40, 42, 43, 48–50], eight studies did not perform a sample size calculation [40–42, 44, 45, 48–50], and nine studies had no plan for handling missing data [40, 42–46, 48–50].

Classification and effects of interventions

Table 3 provides an overview of the five types of interventions we identified in the 11 studies, namely, handover forms (n = 3 studies), a redesigned discharge process (n = 1), medication reconciliation (n = 1), liaison nurses (n = 4), and outreach services (n = 2).

Table 3 Overview of interventions reported in the 11 studies included in the review
Full size table

In three of the 11 studies, handover forms as a tool for improving the information transferred between ICU and ward were evaluated [46, 48, 49]. All three studies found a statistically significant improvement in the reduction of adverse events [46] or in continuity of care [48, 49]. Williams and colleagues investigated the efficacy of a multidisciplinary form completed predominantly by nurses in combination with a discharge checklist completed by the medical staff. They found that the proportion of preventable adverse events was significantly reduced from 65 to 42 % (p < 0.001) [46]. Palma and colleagues implemented the use of a printed sign-out document and sign-out data entry form and reported that the staff perceived these new discharge tools to be significantly more accurate in terms of improving the transfer of information than those used previously (p = 0.0025) [48]. Medlock and colleagues investigated the implementation of an electronic discharge letter with a template to support content decisions. The median time to finalize the discharge letter was significantly reduced from 23 [interquartile range (IQR) 9–41) to 4 days (IQR 2–9; p < 0.0001) [49].

Four studies examined the effects having an ICU liaison nurse in place to coordinate care and communication between ICU and general ward healthcare professionals [41, 42, 44, 45]. Three of these studies found that a liaison nurse had a statistically significant effect on use of care [44, 45] or continuity of care [41]. Chaboyer and colleagues evaluated the effects of liaison nurses who were involved in assessing patients for ICU discharge, coordinating transfer to other wards, and communicating with ward staff [41]. In their study, the liaison nurse assessed ward staff skill-mix and resources, prepared both the ICU and ward for transfer, assessed bed status, and provided clinical support and resources to ward nurses. The authors found that the proportion of patients with a discharge delay of >2 h decreased significantly from 49 to 22 % [odds ratio (OR) 3.3, 95 % confidence interval (CI) 1.7–6.2, p < 0.001] and that the proportion with a discharge delay of >4 h decreased significantly from 29 to 14 % (OR 2.5, 95 % CI 1.2–5.2, p < 0.05) [41]. Elliott and colleagues implemented the use of liaison nurses who supported the management of discharged patients with complex care needs. The service involved communicating with ward staff and providing support and bedside education. These authors reported a significant reduction in mean step-down unit LOS from 71 to 37 days [44]. Endacott and colleagues investigated the role of a liaison nurse who visited patients at least daily for the first 3 days after ICU discharge [45]. In their study, the liaison nurse clinically assessed each patient, reviewed the charts, and provided support and informal education to ward staff. The proportion of patients discharged from the ICU who needed transfer to higher care was significantly increased from 14 to 23 % (adjusted OR 1.82, 95 % CI 1.07–3.09, p = 0.014), and the proportion of patients requiring a surgical procedure significantly increased from 16 to 26 % (adjusted OR 1.85, 95 % CI 1.09–3.12, p = 0.022) [45].

Two studies evaluated outreach services, in which activities were used which focused mainly on the follow-up of discharged ICU patients and supporting ward staff. Both studies found that the intervention did not have a statistically significant effect [40, 47].

Chaboyer and colleagues implemented a redesigned ICU discharge process, including a handover form to facilitate face-to-face or phone communication between ICU and ward healthcare professionals, a notification from the ward to their ICU counterparts of a specific time they were able to receive the patient, and a daily update to the ward staff summarizing all likely patient discharges (‘ICU discharge alert sheet’) to better plan patient transfers and coordinate appropriate follow-up [50]. However, these authors reported that their changes to the ICU discharge process did have any statistically significant effects [50].

Zeigler and colleagues examined the use of medication reconciliation [43]. Upon ICU discharge, medication profiles were printed and reviewed by the primary physician and either discontinued or resumed. No statistically significant effects were found on the study outcome, namely, prolonged use of stress ulcer prophylaxis (SUP) [43].

Implementation activities

All of the 11 studies included in the review incorporated specific activities to facilitate the process of implementation of the intervention (Table 3). In terms of the implementation of handover forms, activities assessed to be effective were informal instructional sessions, the automatic filling of the handover form with data from the electronic medical record, development of software by the ICU staff, electronic reminders, a top–down directive, and involvement of healthcare professionals in the decision-making process [46, 48, 49]. Regarding the implementation of liaison nurses, activities assessed to be effective were a clear task description (based on the literature, formats of other hospitals, and experiences of patients, their families, and ICU and ward nurses), correct qualifications (experienced ICU or critical care nurse), provision of training to standardize the tasks carried out by liaison nurses, and encouraging ward staff to consult a liaison nurse if in doubt [41, 44, 45].

Discussion

In this review we have reported the effects of interventions focusing on improving clinical handovers between ICU and ward healthcare professionals at the time of patient discharge from the ICU. After an extensive search process of six databases and subsequent selection of relevant reports, we ultimately only included 11 studies in our review, which indicates that very few studies on this specific topic have been performed. A statistically significant effect on quality of handover was observed in six of these studies (55 %). Effective interventions included: (1) implementation of liaison nurses to improve the communication and coordination of care between ICU and ward healthcare professionals and (2) handover forms to facilitate the timely handover of complete and accurate clinical information from ICU to ward healthcare professionals. Interventions were effective in improving the continuity of care and reducing preventable adverse events. The effects found for the use of care were inconsistent; a decrease in step-down unit LOS was observed [44], as well as an increase in transfers to higher care and in the requirement for surgical procedures [45],

In accordance with our review, two recent studies report that liaison nurses can be a useful tool for bridging coordination gaps between healthcare settings [20, 30]. In the studies included in this review, factors facilitating the implementation of a liaison nurse were a clear task description, ‘casting’ the right person based on experience, and encouragement of ward staff to consult a liaison nurse [41, 44, 45]. The literature also highlights a number of factors considered to be important for proper functioning of a liaison nurse: (1) that the liaison nurse be able to personalize his/her role as an ICU liaison nurse; (2) that the liaison nurse be able to gain the respect of ICU and general ward colleagues; (3) that ward staff view the implementation of a liaison nurse as a collaborative and supportive effort—and not as an intrusion in their ward [5].

Published studies also show that poor information transfer is a common patient safety issue in all types of handover settings [51–53]. Various reviews have reported the effectiveness of standardizing tools (e.g., standardized handoff tools, computerized handoff tools) to improve information transfer [30, 54, 55] and, possibly, quality of care as well. In accordance with these studies, we found that the use of an ICU discharge form is an effective intervention by which to standardize information transfer and communication between ICU and ward healthcare professionals. It is interesting that the aims of the studies on information transfer focused on improving written communication [e.g., improving the situation background assessment recommendation (SBAR) checklist or collaboration between ICU and ward healthcare staff with team training], even though culture, team climate and verbal communication have been identified as important factors for inadequate patient handover [56].

Assessment of the 11 studies included in our systematic review reveals that a timely transfer of the patient together with accurate and complete information on the patient being transferred from the ICU to the ward are the specific aspects of handover which most readily show a change following interventions. Continuity of care in terms of reduced discharge delay and improved accuracy of discharge information was improved in three out of four studies investigating this outcome measure. However, whether this resulted in any beneficial clinical outcome beyond a better recording of data is unclear. Mortality rates were not improved in any of the seven studies evaluating mortality as an outcome measure. Evidence showing a reduction in ICU readmission or ICU LOS was limited; only one study found evidence for a reduction in step-down unit LOS [44].

The limitations of our review relate to the nature of the interventions and the study designs used. Similar to other reviews on patient handover, most interventions consisted of a complex set of activities. Most studies contain specific activities that have not been studied outside the set of activities used in the intervention. These aspects hinder an appropriate and direct evaluation of the interventions [20, 57]. Second, the poor methodological quality of most of the included studies makes it difficult to draw firm conclusions on the effectiveness of individual interventions. Single-institution evaluations with an observational design, i.e., non-controlled before–after design, dominated the studies we identified. In general, observational studies overestimate the effect. Third, the studies were characterized by significant heterogeneity for both interventions and outcome, making it impossible to perform a meta-analysis. Heterogeneity has been acknowledged to be a common limitation in the clinical handover literature [9, 20, 54, 57]. Fourth, the classification of interventions into information, coordination, and communication categories was strictly based on the description of the intervention provided in the studies. Although interventions were independently classified by two researchers, the classification may be subject to bias due to minimal or unclear intervention descriptions. Fifth, we excluded non-published studies and non-full-text studies, which may have increased the risk of publication bias, i.e., the risk that this review overestimates or underestimates the true intervention effects. Moreover, we could not assess the risk of publication bias using a funnel plot due to the heterogeneity in outcome measures and the small number of studies found [39].

Despite handover being an important topic for the World Health Organization [58], and national government agencies, such as the Joint Commission, this systematic review highlights the absence of evidence on how to improve patient handovers between the ICU and general wards. Several reasons for the lack of effects have been described: use of an inappropriate intervention in relation to the underlying healthcare problem [43], measurement of inappropriate outcomes [50], and suboptimal research population, such as low mortality rate at baseline [47]. The lack of effects may also be influenced by limited actual exposure of healthcare professionals to the intervention and implementation activities [59]. These reasons reflect the difficulty of demonstrating the effectiveness of complex quality improvement interventions, as has been mentioned in several publications [60–62].

Our hope is that this systematic review will act as a stimulus to gather more evidence on the interventions described in the 11 studies included in the review, as well as interventions evaluated in other settings, such as a shared electronic information exchange system to improve handover between hospital and primary healthcare providers [63]. The implementation of interventions for which insufficient evidence is available carries the burden of potentially wasting valuable resources, which may increase the reluctance of clinicians to implement other quality improvement initiatives [64]. We recommend that researchers and clinicians considering to conduct an evaluation of an improved handover process use robust designs to strengthen the quality of evidence on this topic. Randomized controlled trials are often impossible to conduct due to difficulties in blinding and concealment of allocation. Cluster randomized controlled trials pose difficulties in terms of sample size and obtaining a uniform control group. Other rigorous study designs, such as an interrupted time-series or a controlled before-and-after study, are good alternatives and are more feasible in practice [65]; however, they are associated with a greater risk of bias. Objective outcome or performance measures, such as readmission rate or mortality rate, are the ideal parameters for measuring effectiveness, but due to low incidence, it is hard to reach statistical significance. Process measures can be used to gain more insight in the processes leading to improvement in the outcome measures [61].

In conclusion, liaison nurses and handover forms are promising interventions to improve the quality of clinical handover between ICU and ward healthcare professionals at patient discharge. Due to the limited number of studies identified in this review and the weak methodological quality of the included studies, more robust evidence is needed.