Inter-Organizational Coordination to Improve Patient Outcomes in Multimorbid Older Patients Following Hospital Discharge – a Systematic Review

Older patients with multimorbidity during hospital to home transitions

The complex needs of older patients often includes a combination of multimorbidity, such as multiple chronic diseases, physical and/or mental impairment and inadequate ability of self-care [1011]. These health-related vulnerabilities result in a state described as frailty, where minor stressors can cause disproportionate increases in adverse outcomes [1213]. The prevalence of frailty depends on the definition, but is reported to be from 4% to 59% among community-dwelling individuals older than 65 years of age [14]. The concept of frailty includes symptoms such as generalized weakness, poor balance, exhaustion and cognitive impairment, and clinical frailty scales are often used to determine patients care needs in hospitals [13]. This group of patients is particularly vulnerable during the transition from hospital-to-home, where the responsibility of care shifts from hospitals to outpatient health and non-healthcare providers [15].

In particular, older patients with complex needs are often referred to primary care services which in comparison to specialized care give simultaneous attention to the range of diseases affecting the patient as well addressing the patients’ functional ability [1617]. Several studies have also shown that medication reconciliation by pharmacists and interventions by occupational- and/or physiotherapists are important post-discharge [1819]. Furthermore, home care and other non-health care services have vital roles in ensuring the patients quality of life (QoL) and capability of performing activities of daily living (ADL) through e.g. home adaptation, transportation services, personal hygiene and meal delivery [620].

Studies from different countries show that care coordination, herein defined as arrangement of roles, responsibilities and tasks between different health- and social care services, is often insufficient after hospital discharge [2122232425]. For older patients, inadequate care coordination often leads to missed or delayed care and medical errors, factors that are frequently cited as causes of avoidable readmissions and prolonged length of hospital stays (LoS) [2627]. From a health systems perspective, improving post-discharge coordination is seen as crucial for enhancing overall system efficiency [6]. However, research at the system level indicates that coordination is challenging, often attributed to communication breakdown, resource shortages, unclear delegation of responsibility, and professional and cultural barriers [28293031].

Coordinated care for older patients with multimorbidity

Coordination between health and non-health care organizations can be achieved at different levels, for instance at the systemic level where integration is sometimes accomplished by merging health- and non-health care providers into a single entity [32]. Such formal integration of service providers have been suggested to improve patient outcomes and system efficiency through i.e. centralizing management, facilitating communication (e.g. through harmonizing journal systems and facilitating sharing of information) and by aligning incentives and priorities (e.g. through goals and reimbursement) [3233343536]. However, formal integration of multiple service providers is practically unfeasible in countries where health- and social care is provided by distinct and autonomous organizations. In for instance Sweden, health- and social care is organized by autonomous agencies (i.e. regions and municipalities) that are governed by different legislation and have separate commissioning and funding structures. In such non-hierarchical contexts where no single centralized actor is ‘calling the shots’, coordination between providers is primarily achieved by voluntary agreements or partnerships at the organizational and clinical (service delivery) levels, hereon referred to as inter-organizational coordination, IOC [37]. In practice, important aspects of integration and coordination, such as alignment of reimbursement systems and harmonization of electronic health records, can be difficult to achieve at the organizational level. Instead, coordination efforts are often focused on the service level, for example through multi- or inter-professional teams, facilitation of communication (e.g. notification systems), establishing structures for shared decision making (e.g. care conferences) and addressing cultural barriers among providers [36].

Interventions of integrated and coordinated care for older patients during hospital discharge are often interventions of high complexity. Such interventions include multiple components simultaneously and the individual components are often assumed to have synergistic effects when the intervention is tailored to specific study populations and local contexts [3839404142434445] Commonly reported components of these interventions include geriatric risk assessments, multidisciplinary teams, and designated care coordinators [4246].

Furthermore, many studies incorporate program theory that outlines the goals and proposed mechanisms for why a specific intervention is expected to be effective, both regarding health-related outcomes as well as system efficiency, for instance through shifting inpatient to outpatient care. [384748].

In terms of effects, previous umbrella reviews report that integrated care can potentially reduce all cause readmissions by 10–30%, decrease LoS by 1–7 days and improve quality of life (QoL) [4649]. However, positive effects are more commonly observed in disease-specific interventions, such as heart failure, which may not be adequately reflected in the needs of older patients with multimorbidity or be practically feasible in several health care systems [50].

In addition, interventions rarely show unequivocal effects [564651], which is likely related to heterogeneity entailed by diverse intervention components [40], study populations [46], intervention fidelity [40], intervention level (micro-, meso-, macro-), context (health care system) [63840], and clinical setting (e.g. preventing hospitalization or readmissions) [39465253]. As a consequence, identifying consistently effective interventions, or configurations of effective components, has been described as challenging [5455].

Search strategy

A systematic search for English language publications published between 1^st of January 2010 and 22^nd of December 2023 was performed in Cinahl, PubMed, Scopus and Web of Science based on PRESS-guidelines [57]. The search was restricted to 2010–2023 based on a bibliometric analysis of integrated- and coordinated care that showed a large increase in publications starting in 2010. Search terms were developed together with an information specialist at Uppsala University library based on the PICO framework with the addition of context [5859]. The accuracy of the literature search was validated by verifying that key papers previously known to the authors were found [6061]. Additional records were obtained from reference lists of included studies and systematic reviews identified in the systematic search. An outline of the search strategy is presented in the supplementary material A (Supplementary table 1).

Eligibility Criteria

Studies investigating Inter-organizational coordination for patients aged 65 and older with multimorbidity during the transition from hospital to ordinary care or nursing home care were identified. IOC was defined as interventions where a hospital and at least one non-hospital service provider both actively participated in IOC and coordinated their activities in order to achieve health goals. Coordination was broadly defined as “the arranging of roles and tasks into an organized whole” [21]. Furthermore, the coordinated activities had to include both health- and non-health care services, e.g. social care, home care and home adaptation services [821]. Hospital length of stay (LoS) and readmissions were the primary outcomes and in the final sample, additional outcomes relevant to the aims of coordinated care, e.g. health- care utilization, patient health and patient experiences, were included and analyzed. A summary of the eligibility criteria is presented in Table 1.

Study selection

After removing duplicates, the remaining records were screened for eligibility by three researchers working independently (WL, RS, and JM). Records were considered for full text review if two of the assessors agreed that the record could not be excluded based on the abstract. In the full text review, studies were randomly assigned to four authors working independently (WL, RS, JM and IF) with each study undergoing full text-review from two authors before being included or excluded. Disagreements regarding inclusion were resolved by discussion between the two reviewing authors and a third researcher (UW). Measures of agreement (Cohen’s Kappa) were calculated for abstract screening and full text review separately [62]. In a few instances, authors of potentially eligible articles were contacted to confirm the context of at least two distinct organizations. Screening and study selection were performed using software Covidence.

Data extraction, risk of bias assessment and data synthesis

Data from the included studies were extracted by three authors (WL, JM, and IF) using a template from the Cochrane Public Health Group Data Extraction and Assessment Template. Reported outcomes were extracted based on adjusted effects and standardized when possible [63].

Studies meeting eligibility criteria were assessed for risk of bias by three authors (WL, RS and IF). Risk of bias assessments included randomization, allocation concealment, similar baseline outcome measurements, similar baseline characteristics, incomplete outcome data, prevention of knowledge of allocation during recruitment, protection against contamination, selective outcome reporting and other biases [64]. The Risk of Bias assessment template was applied together by the three authors to two studies. The remaining studies were divided between the authors with each study being assessed by two authors working independently. Disagreements were mainly related to the domain protection against contamination and were resolved through discussion between the two responsible authors.

Qualitative evidence regarding interventions and quantitative evidence regarding effects were analyzed using a narrative synthesis. [65]. No meta-analysis was planned as the studies were expected to be interventions of high complexity [45].

Results of the literature search

A flowchart of the study selection process is presented in Figure 1. The literature search yielded 4380 records. After adding 168 additional records and removing duplicates, 4152 records were screened for eligibility with 252 records identified as potentially eligible (Cohen’s Kappa ranging between 0.43–0.62). In total, 12 studies met the eligibility criteria and underwent data extraction and risk of bias assessments (Cohens Kappa ranging between 0.21 and 0.26). A brief summary of excluded articles is presented in supplementary material A.

Figure 1

Description of the studies identified

Intervention components

Seven studies reported descriptions of program theory and included theoretical reasoning as to why the individual components of the interventions would be effective [60666770727375]. Across studies, interventional components were divided into non-clinical, pre-discharge, transitional and post-discharge components, see Table 3. The interventions, components and their effects on primary outcomes are presented in Supplementary material B where it is possible for readers to filter based on specific components.

Intervention effects

All twelve studies reported readmission rates with a follow-up period ranging from 30 to 180 days with eight studies reporting LoS (see Figures 2 and 3 respectively). Five studies reported mortality within 30–365 days, see Figure 4. Tables with QoL, ADL and other additional outcomes are presented in supplementary material A (Supplementary tables 2 and 3).

Figure 2

Figure 3

Figure 4

Strengths and limitations

The main strength of this review lies in its incorporation of contextual factors, in addition to the traditional PICO framework, which expands our understanding of how interventions operate in real world settings for older patients with complex needs. Previous systematic reviews have often overlooked the critical role of context, particularly for complex interventions that must interact with the surrounding health- and social care systems. Without considering these contextual elements, interventions may prove impractical or unfeasible to implement.

A related strength of the review is the inclusion of studies reporting both health care utilization and health related outcomes, generating a broad picture of the effects of IOC in comparison to reviews focusing on single outcomes as well being in better alignment with the goals of coordinated care [4047]. Furthermore, this systematic review has the advantage of including articles with relatively strong study designs with comparatively low risk of bias in comparison to studies included in other reviews [6]. Indeed, the Risk of Bias assessments in this review indicates that the overall risk of bias in the included studies is rather low given the complexity of the interventions.

In order to avoid evidence selection bias it important that systematic reviews include all relevant studies [78]. Although several steps were taken to minimize selection bias in this review, e.g. searches in several databases informed by a trained information specialist and screening of relevant systematic reviews, the possibility of selection bias cannot be excluded. First of all, the search strategy may have missed relevant search terms as there is no consensus regarding definitions and technical terms in the field of integrated and coordinated care [79]. Secondly, the inclusion criteria regarding study design may have excluded relevant IOC-interventions, as these are sometimes evaluated using less reliable study designs. Furthermore, some studies that could be relevant for older patients may have been excluded because of the age criteria (i.e. patients older than 65 years of age).

Third, the authors consisted of a team with different backgrounds (physicians, health economists and health services researchers) with different views on what constitutes coordinated care. As such, the screening process and inclusion of full text articles required continuous discussions regarding which papers to include which is reflected in the Cohen’s kappa statistics. The main reasons for the relatively low agreement between authors were due to poor descriptions of context and uncertainty about the active participation of both in- and outpatient service providers. Although several sources were consulted (e.g. study protocols, process evaluations and corresponding authors) relevant articles may have been excluded [55].

Another concern in systematic reviews is related to publication bias where the tendency of not publishing negative or non-findings causes systematic reviews to overestimate the effects of interventions [80]. However, this does not seem to be a concern in this systematic review as unfavorable and non-significant effects were more common than beneficial effects.

Despite this, there are still some concerns regarding the included studies that could the interpretation of the results. For instance, even though several studies reported implementation activities [666870737475], few studies described intervention fidelity. As such, it is not known whether the lack of desired effects in some studies were due to failures of the interventions or failures of their implementation [8081]. Furthermore, some studies reported LoS as average number of days in hospital, not specifying if the analysis was made based on days in hospital per hospital visit or per patient, changes in LoS may therefore be indicative of changes in readmissions rather than length of hospital stay [61676871].