This study assessed the perceived extrinsic barriers and their mediating effect on the association between knowledge and intended practice of GPs in community MCI detection and management. Our study revealed that social stigma and a lack of confidence in GPs are major barriers of patient engagement as perceived by the GP respondents, while resource constraints and a lack of policy, financial and policy support are major working environment and system barriers. The perceived extrinsic barriers have a negative mediating effect on the association between knowledge and intended behaviour, hence, hypothesis one is supported. Training has a positive moderating effect on the association between knowledge and intended behaviour, and the effect is less powerful when GPs have a higher level of knowledge, hence, hypothesis two is supported. However, past experience did not show a significant effect on perceived extrinsic barriers, hence, hypothesis three is not supported.
This study indicated that both intrinsic drivers (such as knowledge and attitudes of physicians) and extrinsic drivers (patient engagement, working environment, and system context) have shaped the behaviour of primary care physicians in detecting and managing MCI. These results are consistent with the findings reported in a recent systematic review (23), which categorised barriers of dementia care into patient, provider, and system related. The influence of system context on health practice has been widely acknowledged in health policy documents (54, 55). The organisational factors have been identified as influencing the motivation of healthcare providers from both the perspectives of financial and non-financial incentives (35). The average barrier score for patient engagement was found to be 65.23 out of a maximal of 100 in this study, which is the highest among the three barrier domains. A cross-sectional survey of 703 GPs in the Netherlands (ref) indicated that the most perceived barriers to implementing clinical guidelines came from external factors, especially patient preferences, needs and abilities (56). However, findings on improving physician guideline adherence behaviour may not be generalisable, since barriers in one setting may not be present in another.
Our SEM results showed that perceived extrinsic barriers had a negative mediating impact (β=-0.012, p=0.025) on the association between knowledge and intended behaviour, accounting for 15.4% of all indirect effects. A partial mediating effect was confirmed, which suggests that the K-A-P pathway remains to be a major pathway for translating knowledge into practice, and perceived extrinsic barriers have a weak but non-negligible effect on the intended behaviours of GPs in MCI detection and management. Therefore, there is no doubt that the intention of GPs to detect and manage MCI can be compromised when they perceive high levels of extrinsic barriers. Accordingly, it could be understandable why GPs rarely detected MCI in practice (10) even though the majority of primary care physicians acknowledged the value of assessing cognitive impairment in primary care (11).
The association between higher knowledge and higher perceived extrinsic barriers is concerning. It is likely that the GPs with a high level of MCI-related knowledge are more likely to notice potential extrinsic barriers in their practice compared with those with a low level of knowledge. Those knowledgeable physicians may perceive more challenges when the process of implementing approaches impacts their routines and workflow, and requires them to work in new ways. Similarly, a systematic review (57) found that workload and time constraints are dominant barriers to implementing evidence-based dementia care.
The moderation analyses showed that training can potentially improve compliance to practice guidelines. However, the effect of training is less powerful when GPs have already had a higher level of knowledge. It is important to note that knowledge is often acquired through training, but high knowledge is also associated with high perceived barriers. A systematic review of six studies (58) concludes that education alone would not increase adherence of primary care to dementia care guidelines. However, a targeted physician practice-based educational intervention along with community services support is more effective for improving the dementia care competency of clinicians according to a cohort study (36).
In this study, we did not find a significant moderation effect of past experience on the association between knowledge and perceived extrinsic barriers. It may be, at least partly, due to the fact that only 14.8% of the GP respondents reported experience of MCI detection and management. Community-based MCI management is still in its initial development stage in China. Some GPs may have obtained the experience through research or experimental studies. However, implementation or incorporation of the services into routine practice is a different matter. The additional resources available to a research project are likely to disappear. The patients receiving services may become more diversified. The widespread participation of GPs in the new initiative would require some additional incentives. Unfortunately, those who are prepared to practice are more likely to be aware of the barriers in working environment according to the findings of this study. This result is consistent with the results of a qualitive study that explored a “disconnect” between perceptions of GPs and other providers regarding the need for implementation of a chronic disease prevention program in primary care settings. GPs are likely to be more concerned about the lack of a supportive environment than their colleagues (59).
The findings of this study have some implications for policy and management as well as educational activities. Extrinsic barriers in relation to patient engagement, working environment, and system context should be addressed to provide support to GPs in community detection and management of MCI. According to Herzberg’s motivation theory (60), those extrinsic barriers are deemed as hygiene factors which can result in staff dissatisfaction if not addressed properly, even though they do not in themselves motivate employees. They may even deter the efforts of some intrinsically motivated GPs. Training remains critical given that the overall knowledge level of GPs in MCI detection and management is low. Training is particularly powerful when knowledge is low. Meanwhile, however, GPs need to be equipped with skills to adequately cope with the challenging environment. This should include, but not be limited to, more proactive engagement in patient and public education campaigns and advocacy for increasing policy and management support to community detection and management of MCI (61).
Strengths and limitations
The SEM-PLS method was adopted to explore the complex exploratory structural equation model with both formative and reflective measures. This study tested the mediating effect of perceived extrinsic barriers and the moderation effects of training and past experience on the association between MCI knowledge and intended practice in a large sample of GPs. The findings have both policy/management and training implications for developing the programs in relation to community detection and management of MCI in response to the challenges of an aging society.
Like any other study, this study also has some limitations. Firstly, although this survey included a large sample size, it did not represent the entire GP workforce. Attempts to generalise findings of this study should be undertaken cautiously. Secondly, the concept of perceived barriers measured in this study is not equivalent to the objective existence of extrinsic barriers. Data were collected through self-reporting, which is subject to reporting bias. However, perceived extrinsic barriers bear a more direct connection with behavioural intentions (13). Finally, we took intended behaviour to be an outcome measurement, since there exist limited actual MCI detection and management activities including those originating from research and experimental projects. However, intended behaviour has been regarded as the most immediate predictor of actual behaviour, even though there exists an intention-behaviour gap (62). A meta-analysis of 10 meta-analyses concluded that intention accounts for almost one-third of the variance in behaviour (63).