Study selection
The search generated 8,154 hits. After removing duplicates and irrelevant publications based on the title and abstract screening, we assessed 367 full-text articles for eligibility, six of which originated from the additional hand and citation searching. After the exclusion of 353 articles (see Fig. 1 for the PRISMA flow chart), a total of 14 studies (seven main project reports and seven process evaluation reports) were included in the synthesis.
Characteristics of included studies
One out of the seven included main project reports (14,3%) was a RCT [22], the other six (85,7%) were cluster RCTs (cRCTs) [23–28]. Two out of these seven studies (28,6%) included nested process evaluation components in the main report [23, 24]. For the remaining five main project reports (71,4%), additional process evaluation reports were published separately, which we considered within this analysis. Among those, one used qualitative methods [29], two used quantitative methods [30, 31] and four used a mixed-methods approach [32–35]. Details on the characteristics and results of the included studies can be found in Additional file 2.
The studies were published between 2008 and 2017 and took place in PC settings in three different countries: five out of seven (71,4%) in the Netherlands [24–28], one (14,3%) in the UK [23] and one (14,3%) in Canada [22].
The included projects comprised 5,822 participants (3,634 patients in intervention groups; 2,188 patients in control groups).
The mean ages in the intervention groups ranged from 67.1 to 81.7 years and from 66.0 to 82.8 years in the control groups. One study only reported overall age range, which was 60 to 75 years, and did not report mean age [23].
All projects compared CPWs with usual care to assess their effectiveness. Three out of seven projects (42,9%) tested a CPW for persons with specific health conditions, which were type 2 diabetes [24], chronic obstructive pulmonary disease (COPD) [27], and heart failure [22]. The other projects (n = 4; 57,1%) targeted on community-dwelling people [23, 25, 26, 28]. More detailed information about the study characteristics and the results of single studies can be found in Additional file 2.
Despite the general diversity of the seven CPWs, there were commonalities with regard to the development and structure of the interventions. The development of all interventions was evidence-based, and four out of seven studies (57,1%) reported the involvement of clinicians. Four projects (57,1%) undertook a previous pilot/feasibility study. A total of five CPWs (71,4%) started with an patient assessment, six provided an individually tailored treatment (85,7%) and one the application of locally adapted recommendations (14,3%). Six out of the seven CPWs (85,7%) included scheduled evaluation or monitoring of patient outcomes on a regular basis. Education and training for health care providers was included in six CPWs (85,7%). More detailed information about the structure of the interventions is displayed in Table 3. No project provided a clear and comprehensive distinction between intervention components and used implementation strategy. For details of the components of the seven CPWs, see Additional file 2.
Table 3
Main components of the interventions reported in the included main project reports
Source, year | Development and piloting | Components of the intervention: recipient | Components of intervention: provider |
Evidence-based | Involvement of clinicians | Previous feasibility/ pilot study | Assessment | Individually tailored treatment | Locally adapted recommendations | Regular evaluation/ monitoring | Training activities |
Azad et al., 2008 [22] | ⎫ | not reported | ⎫ | ⎫ | ⎫ | X | ⎫ | ⎫ |
Bleijenberg et al., 2016a [26] | ⎫ | ⎫ | ⎫ | ⎫ | ⎫ | X | ⎫ | ⎫ |
Harris et al., 2015 [23] | ⎫ | not reported | not reported | not reported | ⎫ | X | ⎫ | ⎫ |
Melis et al., 2008 [28] | ⎫ | not reported | ⎫ | ⎫ | ⎫ | X | ⎫ | not reported |
Metzelthin et al., 2013b [25] | ⎫ | ⎫ | ⎫ | ⎫ | ⎫ | X | ⎫ | ⎫ |
van Bruggen et al. 2008 [24] | ⎫ | ⎫ | not reported | not reported | not reported | ⎫ | not reported | ⎫ |
Weldam et al., 2017b [35] | ⎫ | ⎫ | not reported | ⎫ | ⎫ | X | ⎫ | ⎫ |
⎫=Yes; X = NO |
Detailed information about characteristics of excluded studies and reasons for exclusion are available from the authors upon reasonable request.
Outcome measures
Five out of seven projects (71,4%) used patient-relevant primary outcomes, such as disability [25], daily functioning [26], functional performance in activities of daily living and mental well-being [28], quality of life and functional capacity for older females living with heart failure [22] and health status of COPD patients [27]. Two of seven studies (28,6%) investigated surrogate endpoints, such as changes in average daily step count [23] and the percentage of people with poor glycaemic control [24].
Quality of evidence
Details of the judgements about each RoB item in the included (cluster-)randomized controlled studies and across these trials are shown in Additional file 3, Fig. 2 and Fig. 3. We judged the RoB in 85,7% (n = 6) of included trials in generation of the allocation sequence, in 71,4% (n = 5) in incomplete outcome data, in 42,9% (n = 3) each in blinding of outcome assessment, in selective reporting and in cluster randomized trials and in 14,3% (n = 1) in allocation concealment as low. We assessed 42,9% of included studies (n = 3) as being at high RoB in blinding of participants and personnel, 28,6% (n = 2) each in cluster randomized trials and in other bias like a small sample size, and 14,3% (n = 1) each in blinding of outcome assessment, incomplete outcome data and selective reporting. Due to a lack of information in almost all studies, the authors judged a total of 43,6% (n = 24/55) of RoB domains as being unclear (38,2% as low risk: n = 21/55; 18,2% as high risk: n = 10/55).
The problem of poor reporting was also relevant in the quality assessment of the process evaluation reports (see Tables 4 for CASP and Table 5 for MMAT). None of the studies that use qualitative methods adequately described the relationship and interaction between the participants and the researcher. This also applies to qualitative parts of mixed-methods studies. One qualitative study (33,3%) did not report approval of an ethics committee or institutional review board.
Table 4
Quality assessment results of aspects of the qualitative studies (CASP Checklist)
Quality assessment question | Bleijenberg et al., 2015 [29] | Harris et al., 2015 [23] | van Bruggen et al., 2008 [24] |
Was there a clear statement of the aims of the research? | ⎫ | ⎫ | can’t tell |
Is a qualitative methodology appropriate? | ⎫ | ⎫ | ⎫ |
Was the research design appropriate to address the aims of the research? | ⎫ | ⎫ | can’t tell |
Was the recruitment strategy appropriate to the aims of the research? | can’t tell | can’t tell | can’t tell |
Was the data collected in a way that addressed the research issue? | ⎫ | ⎫ | can’t tell |
Has the relationship between researcher and participants been adequately considered? | can’t tell | can’t tell | can’t tell |
Have ethical issues been taken into consideration? | can’t tell | ⎫ | ⎫ |
Was the data analysis sufficiently rigorous? | ⎫ | can’t tell | can’t tell |
Is there a clear statement of findings? | ⎫ | can’t tell | ⎫ |
⎫=Yes; [X = NO] |
Table 5
Quality assessment results of aspects of the mixed-method studies (MMAT)
Quality assessment question | Bleijenberg et al., 2013b [33] | Bleijenberg et al., 2016b [34] | Weldam et al., 2017b [35] | Metzelthin et al., 2013a [32] |
Screening Questions (for all types) | | | | |
Are there clear qualitative and quantitative research questions (or objectives), or a clear mixed methods question (or objective)? | ⎫ | ⎫ | ⎫ | ⎫ |
Do the collected data allow address the research question (objective)? E.g., consider whether the follow-up period is long enough for the outcome to occur (for longitudinal studies or study components). | can’t tell | ⎫ | ⎫ | ⎫ |
Qualitative | | | | |
Are the sources of qualitative data (archives, documents, informants, observations) relevant to address the research question (objective)? | ⎫ | ⎫ | ⎫ | ⎫ |
Is the process for analyzing qualitative data relevant to address the research question (objective)? | ⎫ | ⎫ | ⎫ | ⎫ |
Is appropriate consideration given to how findings relate to the context, e.g., the setting, in which the data were collected? | ⎫ | ⎫ | ⎫ | ⎫ |
Is appropriate consideration given to how findings relate to researchers’ influence, e.g., through their interactions with participants? | can’t tell | can’t tell | can’t tell | can’t tell |
Quantitative descriptive | | | | |
Is the sampling strategy relevant to address the quantitative research question (quantitative aspect of the mixed methods question)? | ⎫ | ⎫ | ⎫ | ⎫ |
Is the sample representative of the population understudy? | ⎫ | ⎫ | ⎫ | ⎫ |
Are measurements appropriate (clear origin, or validity known, or standard instrument)? | can’t tell | can’t tell | can’t tell | can’t tell |
Is there an acceptable response rate (60% or above)? | ⎫ | ⎫ | ⎫ | ⎫ |
Mixed methods | | | | |
Is the mixed methods research design relevant to address the qualitative and quantitative research questions (or objectives), or the qualitative and quantitative aspects of the mixed methods question (or objective)? | ⎫ | ⎫ | ⎫ | ⎫ |
Is the integration of qualitative and quantitative data (or results) relevant to address the research question (objective)? | ⎫ | ⎫ | ⎫ | ⎫ |
Is appropriate consideration given to the limitations associated with this integration, e.g., the divergence of qualitative and quantitative data (or results) in a triangulation design? | ⎫ | can’t tell | can’t tell | ⎫ |
⎫=Yes; [X = NO] |
Factors influencing the success of implementation
The classification of barriers to and facilitators of successful implementation of CPWs in PC was based on the context, implementation and setting dimensions of the CICI framework [15].
An overview of barriers and facilitators in the individual studies is shown Table 7. Barriers were most frequently identified within the dimensions of implementation agents (n = 7) and setting (n = 4). Facilitators were most frequently determined within the implementation agents (n = 6) and implementation strategies (n = 4) (see Table 6).
Table 6
Distribution of barriers and facilitators
Source of main project report, year | Barriers | Facilitators |
Context | Implementation | Setting | Context | Implementation | Setting |
Geographical context | Epidemiological context | Socio-cultural context | Socio-economic context | Ethical context | Legal context | Political context | Implementation theory | Implementation process | Implementation strategies | Implementation agents | Implementation outcomes | | Geographical context | Epidemiological context | Socio-cultural context | Socio-economic context | Ethical context | Legal context | Political context | Implementation theory | Implementation process | Implementation strategies | Implementation agents | Implementation outcomes | |
Azad et al., 2008 [22] | | X | | | | | | | | | X | | X | | | | | | | | | | | | | |
Bleijenberg et al., 2016a [26] | | X | X | X | | | X | | | | X | X | X | | | | | | | | | | X | X | | X |
Harris et al., 2015 [23] | | | | | | | | | | | X | | | | | | | | | | | | X | X | | |
Melis et al., 2008 [28] | | | | | | | | | | | X | | | | | | | | | | | | | X | | |
Metzelthin et al., 2013b [25] | | | | | | | | | | X | X | X | | | | | | | | | | | | X | | |
van Bruggen et al. 2008 [24] | | | | | | | X | | | | X | | X | | | | | | | | | | X | X | | |
Weldam et al., 2017a [27] | | X | X | X | | | X | | | | X | | X | | | | | | | | | | X | X | | |
Context
Three out of seven CPWs (42,9%) considered aspects of the epidemiological context such as multi-morbid [31, 33, 35] patients aged at least 85 years [33] with mental health problems [34] as barriers to applying an intervention.
28,6% (n = 2) of the CPWs reported the cultural background [33, 35], a low health literacy [35] and gender [33, 35] as potential barriers that could be attributed to the domain of socio-cultural context. Such patient-related characteristics can lead to a time lag in the application of an intervention. Additionally, the frequency of general practice visits [33, 35] have been reported to have a negative impact by two CPWs (28,6%) and could therefore be seen as barrier according to two CPWs.
Additionally, two out of seven CPWs (28,6%) considered a low socio-economic status [33, 35] within the domain of socio-economic context as barriers to applying an intervention.
Furthermore, aspects related to the political context, such as a lack of an incentive systems [24] or adequate reimbursement models [35] or absent monetary compensations [33], were reported in three out of seven CPWs (42,9%) as potential barriers for the effective implementation of an intervention.
No barriers or facilitators within the domains geographical, ethical and legal context could be identified. None of the CPWs described facilitators in any of the dimensions of the domain context.
Implementation
Within the domain of implementation strategies the involved HPs of three out of seven CPWs (42,9%) emphasized the importance of training activities and reported appropriate training and education in applying an intervention [23, 24, 33] as facilitator. One CPW (14,3%) considered an overload of information during training activities as potential barrier [32]. According to the results of one CPW, a handbook as facilitator can serve as a clear guideline for HPs to promote a structured application of intervention [35].
The domain of implementation agents can be divided into the two areas of HPs and patients.
On the one hand, HPs’ insufficient or even lack of knowledge about how to perform intervention components such as assessments or tests [24, 32, 33], their lack of competence in general [32] and their insufficient experience and job training [32] were considered barriers regarding knowledge and skills in three out of seven CPWs (42,9%). On the other hand, 42,9% (n = 3) of included CPWs identified knowledge and skills such as professional [32, 33, 35], organizational [32] and communication skills [32] and empathic capacity [32] as serving as facilitators to the implementation of the approach. The behaviour-related factors of attitude and awareness, such as a lack of motivation of end-users [24] (14,3%; n = 1) and initial difficulties in implementation due to changes in routines [32, 35] (28,6%; n = 2) were reported as barrieres, which can reduce the success of intervention. Further barriers were negative attitude towards the intervention, such as doubts about the expected results [33] in one out of seven CPWs (14,3%), and reluctance regarding an intervention component due to a lack of agreement [24, 35] in 28,6% (n = 2) of included CPWs, e.g., the prescription of multiple drug regimes [24]. In contrast, a positive attitude towards the effectiveness of the intervention [33, 35] is reported to be a facilitator according to two out of seven CPWs (28,6%). One CPW (14,3%) stated that interventions that provide recommendations to both patients and GPs increased adherence among HPs and affected patients and are therefore facilitators [30].
Interaction-related factors were identified in five out of seven CPWs (71,4%) as influencing aspects. In this regard, HPs named communication and collaboration issues [33] and difficulties in organizing team meetings [32] as barriers. HPs considered good interdisciplinary communication and cooperation [32–34] in 28,6% (n = 2) of included CPWs as well as clear roles and task definition [32, 33] in two out of seven CPWs (28,6%) as facilitators. In addition to the consideration of the multi-professional team, the positive impact of intradisciplinary communication and cooperation was identified in 28,6% (n = 2) of included CPWs as a facilitator [24, 33], e.g., by making comparisons with peers [24]. The integration of family caregivers into the intervention, if possible, was identified as facilitator in one CPW (14,3%) [29], whereas insufficient involvement of single professions was mentioned as barrier in one CPW (14,3%) [33]. According to three out of seven CPWs (42,9%), further barriers in application of the CPW arise due to the extent of intervention, such as time-consuming parts [32, 33, 35] and overly complex intervention components [32, 33]. Two out of seven CPWs (28,6%) reported an individual, flexible, tailored intervention customized to patients’ needs, wishes and preferences providing the HPs as major facilitator in application [33, 35]. Another facilitator in implementation is a good fit of the intervention to the day-to-day work of the delivery agents [35]. A practicable layout of the intervention can ease adoption in daily practice [35] as facilitator sccording to one included CPW (14,3%).
In addition to HPs, patients as consumers of the intervention, were also considered to affect implementation success. Aspects in this domain were partly identified by the patients themselves (self-assessments) and partly by HPs based on their experiences with affected patients (external assessments): regarding behaviour-related factors, HPs in three out of seven CPWs (42,9%) assumed patients’ motivational issues to be a reason for their low treatment adherence and therefore as barrier [30, 33, 35]. Furthermore, external factors such as transportation issues, sometimes due to adverse weather conditions or scheduling conflicts with other appointments, affected the adherence of intervention recipients and serve as barriers [31]. Similar to HPs, patients in two out of seven studies (28,6%) also indicated that positive expectations regarding interventions [32, 33] were a facilitator. The delivery was also affected by the structure of the intervention components. Participants of one CPW (14,3%) perceived high temporal expenditure due to time-consuming participation to be a barrier [32]. Recipients of each one CPW (14,3%) classified high bureaucratic effort [23] and difficulties in distinguishing the involved disciplines [32] as barriers. On the other hand, two out of seven CPWs (28,6%) reported tailored interventions meeting patients’ current needs [23, 29, 33]; one CPW (14,3%) the possibility for adaptations to avoid excessively restricting their own decision making, e.g., through self-management approaches [32]; and one CPW (14,3%) close monitoring of changing situations, which transmits a sense of security [29], as facilitators. Furthermore, in one CPW (14,3%) the provision of written advice such as a handbook [23] and the use of technical devices for outcome measurement [23] were seen as facilitators by consumers. In addition, patients considered interactions with HPs through personal meetings [23, 32] in two out of seven CPWs (28,6%), good professional-patient relationships [29, 32, 33] in 28,6% (n = 2) of CPWs and good internal exchange between HPs [29] in one CPW (14,3%) to be facilitators.
Within the domain of implementation outcomes two CPWs (28,6%) reported a barrier in problems occurred during the identification of the appropriate target group as the first step of the intervention [32, 33], e.g., due to dysfunctional screening methods [32].
No barriers or facilitators within the domains implementation theory and implementation process were reported. In addition, no facilitators within the domain of implementation outcomes were mentioned by included CPWs.
Setting
Barriers reported in four out of seven CPWs (57,1%) within the work environment in the dimension of setting are inadequate staffing due to the general lack of available staff [31, 33], e.g., due to illness or part-time employment [31] and lack of sufficiently educated staff [33]. Structural conditions lead to time pressure [24, 33–35], e.g., due to excessive workload in daily practice [34, 35], which negatively affects the situational performance of intervention components. Additionally, two CPWs (28,6%) mentioned a lack of space as barrier [31, 35]. Also, one CPW (14,3%) cited discontinuity problems in GPs as a barrier [29]. Transparency about referral possibilities promoting the familiarity of HPs with these options was identified as a facilitator [33].