In Emilia-Romagna region hospitals, during the index period, 8% of patients experienced a delay in discharge, which amounts to about 590 (10%) beds per day occupied inappropriately. This figure is close to the lower boundaries of the prevalence ranges reported in two systematic reviews on inappropriate use of acute care beds (15%–50%  and 8%–66% ).
Measures to optimise hospital care resources have been introduced in the region in the last fifteen years, starting from the implementation of hospital networks based on the hub and spoke model, that have been followed more recently by interventions aimed to improve patient flow logistic according to clinical severity and complexity of care. Strategies for discharge planning have also been implemented, supported by nurses and other staff in the role of discharge coordinators. These policies and strategies had the objective to foster care coordination for older people with complex needs and to reduce healthcare costs. It is possible that the low prevalence of DHD in the region is partially accounted by these organisational efforts, in line with a recent study  showing that medical staff involvement and discharge planning significantly reduced DHDs in a teaching hospital.
However, the need to cut hospital budget and reduce the number of bed requires further efforts to rationalise the use of beds. In this sense, the study of the delayed discharge phenomenon, and specifically of its organisational and individual determinants, may help stakeholders to develop informed strategies and policies.
In the present study, a remarkable variability was observed in the prevalence of DHDs among the five disciplines considered, from 3% in general surgery to 15% in long-term/rehabilitation units, that proved to be the main bottleneck. In these units, almost half of DHD patients (45.3%) (Table 3) were waiting to be transferred to residential facilities and 22.6% did not have a caregiver at home. Conversely, 42.1% of DHD patients in internal medicine and 40.0% in geriatric wards were waiting for a bed in long-term/rehabilitation units. Our results are consistent with Seymour et al. and Coid et al.[22, 23], who identified a major risk of DHD for medical specialties and very few bed-blocking patients in surgical wards. Several factors may account for this observation. The younger age of patients treated in surgical wards indicates that this patient group is less likely to pose the multiplicity of problems which characterise elderly patients. Another explanation could be that surgery units discharge patients or transfer them elsewhere as soon as possible to make their bed available .
As to the demographic and clinical determinants of DHDs, we found that increasing age, the number of comorbidities, a primary diagnosis of dementia, fracture or tumour, and the provision of intensive care were associated with an increased likelihood of DHD. Specifically, each decade of life increased the risk of DHD by 24% (p < 0.001). These results altogether support the evidence that elderly patients are more prone to have protracted hospital stay [3, 24] because they present with multimorbidity or with specific problems like cognitive impairment or orthopaedic conditions that may require rehabilitation, domiciliary services or some form of institutional care that may be not immediately available at discharge [25, 26].
Of note, we found that Hospital Trusts were more likely to have DHD patients than Local Health Authority Hospitals. A possible explanation is that Hospital Trusts have weaker links with the primary care sector, that is managed by Local Health Authorities.
The evidence provided by our study supports the need to set up service delivery models meant to reduce the hospital stay (especially for older patients hospitalised in long-term/rehabilitation units or waiting to access them) through the provision of enhanced health and social care arrangements. In Australia, Canada and the US, a number of programmes were implemented in the hospital units, including the Acute Care for Elders (ACE) , the Hospital Elder Life Program (HELP)  and orthopaedic-geriatric medicine cocare . These programmes were associated with significant reductions in morbidity and mortality, and increases in optimal postoperative care [28, 29], and proved to prevent significantly functional decline, reduce cost of care and length of hospital stay, and increase home discharge . However, complex care models such as the ACE programme (including patient-centred care, frequent medical review, early rehabilitation, and early discharge planning) are difficult to implement in acute hospital wards on a routine basis .
In the community setting, service delivery models aimed to facilitate rehabilitation and hospital discharge, or more holistically care close to home, have been denominated in various ways like “hospital at home”, “nursing led in-patient units”, “general practitioner run community hospitals”, “intermediate care in nursing homes” and “community care centres” [32, 33], and in the UK they have been named “intermediate care”. A recent realist review has been published by Pearson et al. with the aim to build up a conceptual framework for intermediate care to investigate under which circumstances it is likely to be feasible and effective and, most important, cost-effective. However, evidence on this topic is still lacking.
One strength of the present study is that it included a large number of units from different medical specialties, and a relatively large number of patients. Despite this significant strength, this study also has an important limitation. Patients were considered eligible for inclusion in the study at any point during their hospital stay: this could have accounted for the low prevalence of DHD and confounded results, as patients who are at the end of their stay are more likely to be designated DHDs compared with those that are newly admitted. However, since the one-day and two-week prevalence rates reported in the present study are very similar, the cross-sectional assessment did not lead to an underestimation of the phenomenon.