Generic care pathway for elderly patients in need of home care services after discharge from hospital: a cluster randomised controlled trial

For elderly patients characterised by multimorbidity, functional decline and complex medical regimens, the transition between general hospitals and primary care is associated with a risk of adverse events; especially concerning medication discrepancies [1, 2], insufficient information transfer [2, 3] and inadequate follow-up in primary care [2]. Three systematic reviews from 2012 to 2014 [4‐6] found that several types of interventions may improve transition across care settings; multicomponent interventions incorporating both pre-discharge and post-discharge interventions seems to be most effective in reducing post-discharge adverse events [5]. Although there is broad consensus that more care must shift from hospital to primary care [7, 8], only few of the papers included in the reviews were studies of post-discharge intervention performed by primary care. However, two Scandinavian studies have shown that post-discharge interventions run by primary care can reduce readmissions [9, 10], dependence on municipal care [10] and mortality [11].

The care pathway ‘Patient Trajectory for Home-dwelling elders’ (PaTH) was developed [12] and introduced [13] in Central Norway region in 2009 to ensure adequate pre-discharge planning and coordination between general hospitals and primary care for elderly patients in need of home care services after hospital discharge. Furthermore, post-discharge follow-up by home care professionals and general practitioners (GPs) was structured to ensure adequate care of medical conditions, prevent functional decline and ensure sufficient social support by introducing checklists at defined stages in the patient trajectory (Fig. 1). PaTH was developed by health care professionals from six municipalities and three hospitals who decided on a generic care pathway—suitable for patients with most diagnoses—in contrast to care pathways developed and used in hospitals targeting a defined group of patients with a specific medical condition [14].

During a qualitative process evaluation of the implementation of PaTH [13], home care professionals expressed that they were better prepared before discharge and experienced improved collaboration and exchange of information with the GPs. They also reported that the systematised observations and measures provided by using PaTH resulted in services of higher quality. The home care leaders valued PaTH as a management tool that served to facilitate change in coordination and provision of health care. These effects on the process level became gradually more apparent with time in the municipalities where the home care professionals found a way to incorporate PaTH into daily working routines.

The aim of this study was to establish the effect of PaTH on patient level—compared to usual care − for elderly patients in need of home care services after discharge from a general hospital, regarding primarily the patients’ functional level and readmissions, secondarily use of health care services, mortality and quality of life.

This study of the effect of the recently developed PaTH [12], had an unblinded, cluster randomised controlled trial (cRCT) design and enrolled patients consecutively in the period October 2009–March 2011. It was registered in Clinical Trials.gov (NCT01107119).

Twelve home care clusters from five Norwegian municipalities included a total of 304 patients (Fig. 3). The intervention (PaTH) group and control group were comparable with respect to baseline characteristics for the patients, apart from lower functional level in the PaTH group (p < 0.002) (Table 2).

There were no dropouts except for deaths during the trial. Register data from hospitals and municipal care systems were complete. Printouts from the GPs’ electronic health records were missing for six patients in the PaTH group and five patients in the control group. The response rate was similar in the groups ranging from 99 to 80% for NEADL questionnaires and 59 to 76% for the SF-36 questionnaires (Table 3).

All PaTH clusters used the PaTH checklists, but to varying degrees (Table 4); 79% of the patients (129 patients) were assessed by at least one, 63% (103 patients) by at least 2 and 36% (59 patients) by at least three checklists. Use of checklists improved over time; 52% of the first half of included patients in the intervention group and 75% of the last half were assessed by at least two checklists.

The PaTH and control groups did not differ with respect to the primary outcomes: functional level (Table 3) and readmissions (Table 5), or the secondary outcomes: quality of life (Table 3), mortality, and health care utilisation apart from more GP consultations (p = 0.04) in the PaTH group (Table 5). Moreover, there were no statistical significant differences between the groups in the subgroup analyses.

Lack of adherence to PaTH rendered the study inconclusive regarding the elderly’s functional level and the number of readmissions after hospital discharge. The effect of intended PaTH use could not be adequately tested because most patients (64%) were assessed by either none of the main PaTH checklists (21%) or only one or two (44%) of them (Table 4). Furthermore, we had to base the sample size estimation on a proxy as we only had access to normal decrease in ADL in the target population from the Norwegian IPLOS scoring system. The sample size estimation was not adjusted for the cluster design, which generally requires more participants than individual-controlled trials [26]. However, no cluster effects were found in this study—except for a statistical non-significant cluster effect on deaths—thus minimising this limitation.

The cluster design represented otherwise a strength of the study. This was chosen because PaTH implied a new way—involving all home care staff—to provide and organise the daily services of home care in the post-discharge period. An individual randomised design would have implied a substantial risk of contamination of the control patients, reducing the possibility of detecting effect of the intervention [27]. Other strengths are high response rate to NEADL and completeness of registry data (from hospitals and municipal care) on health care utilisation. The real-life multicentre setting is a strength, as we could test the feasibility of the intervention in several locations in a city as well as in larger and smaller rural communities. However, multicentre settings increase the complexity and can reduce the possibilities of detecting effect in a trial [28].

A promising finding in our study was a higher number of GP consultations in the PaTH group than in the control group. In two Scandinavian studies with documented effect of post-discharge interventions performed by primary care [9‐11], the physicians took an active part in the medical follow-up in the early post-discharge period. In our study, the GPs and home care nurses performed independent assessments of common patients, did not share patient records, and communicated mainly by using phone or fax (Fig. 2). The PaTH procedures implied that home care staff performed the main follow-up and initiated the GP consultations, leaving the GPs in a more passive role [13]; they were merely responding to the request from the home care nurses in line with normal procedure rather than actively set the agenda for further medical follow-up as in the other Scandinavian studies [9‐11]. Considering these findings, we believe that more active involvement of GPs and closer integration of follow-up by home care services and GPs in the early post-discharge phase is important to improve outcomes for this patient group characterised by multimorbidity and functional decline (Table 2).

Even if it was not possible to document other effects of PaTH on patient level in this cRCT, our former study of the implementation process [13] indicated positive effect of PaTH on the process level. Several studies have documented that care pathways can have an effect, both on the quality of health care provision and on patient level outcomes, but the effect is dependent on how the implementation process is carried out [29, 30]. A systematic review from 2012 by Smith et al. on interventions in primary care, concluded that multicomponent complex organisational interventions, such as changes in delivery of care, seem less effective on patient level outcomes than interventions directed primarily at the patients, e.g. training directed on improving activities in daily living [31]. One reason can be that implementation of complex interventions, requiring behavioural changes for the health personnel involved, takes time and require targeted efforts before the intervention is settled in the organisation [32]. This became evident in our previous implementation study [13] where we found that enthusiasm and positive attitude to the intervention was not sufficient for a successful implementation. Considerable efforts were needed to make the home care professionals understand their responsibilities in PaTH and how to assess the patients by using the checklists. A strong managerial focus on creating commitments and engagement to PaTH and practical facilitation of work processes further characterised the implementation process in the municipalities that succeeded in integrating PaTH into daily working processes.

In our cRCT, the implementation challenges were underestimated as the home care professionals had been sitting in the driver’s seat during the development process and the care pathway was developed according to their own needs of information transfer and structured follow-up [12]. Furthermore, there were no indications of implementation challenges during the pilot period. During the research period, the researcher had monthly conference calls to the contact persons in the home care services in every municipality [13]. These calls did not reveal any implementation challenges. A more targeted exploration in the pre-trial phase would have been necessary to disclose these challenges. We used the implementation theory ‘Normalisation process theory’ (NPT) [33] when exploring the implementation process [13] and see, in retrospect, that NPT could have been useful in the pilot period as well for more targeted testing of the prerequisites for implementation.

For future studies of complex interventions, we suggest, in line with the UK Medical Research Council [27], a prolonged pilot period for testing of acceptability and understanding of the intervention and furthermore, to explore the potential for implementation into daily working practices by applying an implementation theory. We further suggest that adherence to the intervention is closely recorded before preceding to the main trial to ensure that full effect can be observed within the study period.

In some of the municipalities participating in the cRCT, PATH is now in common use. Further studies are needed to evaluate the long-term sustainability as well as exploration of potential adverse events to service provision and patients.

Lack of adherence to PaTH rendered the study inconclusive regarding the elderly’s functional level, number of readmissions after hospital discharge and health care utilisation except for more consultations with the GPs. A targeted exploration of prerequisites for implementation, supported by relevant implementation theories, is recommended in the pre-trial phase of complex intervention studies.