Design and setting

This study is part of a prospective, observational, cohort study that during the 3 years 2011–2014, enrolled 961 consecutive patients 70 years or older after acute admissions to the two hospitals in Bergen. The design, setting and patients have been described more extensively in two other studies [11, 12].

After a short stay of median 5 days in the hospital for establishing the diagnoses and start of therapy, the patients were transferred to the nursing home 19-bed IC unit. The staffing was approximately to the level of a community hospital with increased multidisciplinary personnel (two fulltime physicians, one of them being a geriatrician, 15 nurses, 1.2 positions for physiotherapists, and 0.8 positions for an occupational therapist). If the patient could not return home within 14 days, transfer to an ordinary lower-cost, skilled nursing facility should occur. In these premises, the multidisciplinary staffing was approximately 1/3 of the staffing in the IC unit.

Patients

The patients that were transferred to the IC unit comprised approximately 20 % of all admissions from hospital to nursing homes in the municipality of Bergen. Both medical patients (from the departments of internal medicine, including cardiology and pulmonology) and orthopaedic patients were admitted. Most of the orthopaedic patients had suffered a fall, and none were admitted after elective surgery.

Subdivision of patients into rapid, slow and poor recovery groups

Patients were divided into 3 groups based on their ability to return home after acute hospitalization and nursing home stay [12].

The rapid recovery group included the patients who were able to return directly home from the short term stay (median 14 days, range 2–31) in the IC unit. Follow-up of these patients demonstrated that 87 % of them were living at home 6 months after the acute hospitalization [12].

The slow recovery group included patients who were not able to return home after a short term stay in the IC unit, and had to be transferred to an ordinary skilled nursing facility. However, 2 months after the acute hospitalization, these patients had returned to their own home, and 6 months after hospitalization, 87 % of them were still living at home [12].

The poor recovery group included patients who were not able to return home after short-term nursing home IC and were transferred to an ordinary skilled nursing facility. Two months after hospitalization these patients were either dead or still living in a nursing home. Six months after hospitalization, only 20 % of them were living at home [12].

Data collection and subgrouping of Barthel Index and diagnoses into 3 different groups

The data on patient’s demographic and baseline clinical characteristics were obtained from hospital records.

In 99 % of the patients, functional capacity was assessed by BI at admission by nurses in the IC unit observing the patients [17]. BI scores 10 different ADL-items (feeding, bathing, grooming, dressing, defecation, bladder function, ability to use the toilet, transfer, mobility and climbing stairs). The score range is 0–100, the highest score indicates the best function. Patients with different BI scores were divided into three meaningful clinical groups: Severely decreased ADL: BI < 50, moderate reduced ADL: BI 50–79, and independent ADL: BI 80–100, according to another study reporting on 5087 geriatric patients [15].

Patients with different diagnoses were divided into three meaningful clinical groups depending on the rate of recovery, as indicated in Fig. 1. One subgroup included the patients from the medical, pulmonology and cardiac departments. A second subgroup included patients from both the medical and orthopaedic department with minor trauma/contusions and vertebrae compression fractures. The third subgroup included patients from the orthopaedic department with all other fractures, including 76 hip fractures.

In addition to BI, further geriatric assessment was performed during the first week on >90 % of the patients by the following geriatric screening tests: 1) The Norwegian version of the Mini Mental Status Examination, MMSE [18, 19]. (Score range 0–30, higher scores indicate better cognitive status and score < 24 is considered a sign of cognitive impairment. 2) Geriatric Depression Scale, GDS [20] score range 0–30, higher score indicates increasing symptoms of depression) and 3) Mini Nutritional Assessment - Short Form; MNA-SF [21]. Lower score indicates malnutrition and a score < 8 gives suspicion of malnutrition. Patients with suspected delirium on admission to the IC unit were evaluated by the nursing home geriatrician using the Confusion assessment method (CAM).

Information on whether the patients returned home after transfer to an ordinary nursing home, residence status and survival, was obtained from the patient administrative system in the municipality.

Statistical analyses

For identifying the clinical characteristics that were associated with a rapid, slow and poor recovery, odds ratios (ORs) with 95 % confidence intervals (CIs) were estimated using logistic regression models. The patients with a rapid recovery were compared to the rest of the patients having a slow or poor recovery. Each of the patient groups with slow or poor recovery was compared with the patients that had a rapid recovery.

The characteristics associated with p < 0.25 in univariate analysis were noted as likely predictors and included in multivariate, adjusted logistic regression models. The characteristics associated with p ≤ 0.05 was considered statistically significant in the multivariate models. Two logistic multivariate regression models were analyzed; 1) a model including BI and further geriatric tests; MMSE, GDS and MNA (geriatric model) and 2) a simpler model not including MMSE, GDS and MNA (Table 2).

The analyses were performed using the Statistical Package for Social Science (IBM SPSS), version 20 for Windows.

Patient characteristics

Of 1085 patients who were admitted to the IC unit during the period 2011–2014, 112 were not asked to participate in the study at times when the geriatrician in charge was absent, 5 patients refused to participate, 4 patients had delirium and 3 patients had language problems. Thus, altogether 961 patients were included in the study.

The different rate of recovery according to the ten different diagnoses groups are shown in Fig. 1. The figure indicates that the majority of patients with a medical diagnosis and contusion/injury diagnosis had rapid recovery. Slow recovery was more often observed for patients with non-vertebral fractures, while the influence of diagnoses was not so pronounced for the groups of patients that had a poor recovery.

Variables associated with slow recovery

Variables associated with poor recovery

As shown in Table 2, in the multivariate analysis, only a severely reduced ADL (BI < 50), together with receiving homecare, diagnosis of a non-vertebral fracture, increasing age, and in the model with geriatric assessment, cognitive impairment, were significantly associated with poor recovery. No particular fracture subgroup or medical diagnosis was associated with a higher risk of poor recovery when this was analyzed in a separate multivariate analysis (p > 0.05). This is also indicated in Fig. 1.

The effect of Barthel Index, diagnoses and cognitive impairment on the likelihood of slow and poor recovery, in multivariate regression analysis

As shown in Table 2, patients with a severe BI reduction (<50), were 6.9 and 8.3 times more likely to have poor recovery (models without and with geriatric assessment), and 6.3 and 4.2 more likely to have a slow recovery. Patients with a moderate BI reduction (50–79) were 3.7 and 3.6 more likely to have a slow recovery, but no more likely to have a poor recovery.

Patients with diagnosis of non-vertebral fractures were 4.5 and 6.6 times more likely to have a slow recovery but less likely (3.2 times) to have a poor recovery.

Cognitive impairment (MMSE <24) was most strongly associated with poor recovery (3 times increased odds), but also with slow recovery (2 times increased odds).