3.3 Results and discussion
3.3.4 Conclusions
As the guideline on evaluating complex interventions by the Medical Research Council (2008) pointed out, “[f]ew interventions are truly simple, but there is a wide range of complexity”. The guideline listed five dimensions that make an intervention complex, and all dimensions seem to apply to interventions on the balance between primary and hospital care:
“·Number of and interactions between components within the experimental and control interventions
·Number and difficulty of behaviours required by those delivering or receiving the intervention
·Number of groups or organisational levels targeted by the intervention
·Number and variability of outcomes
·Degree of flexibility or tailoring of the intervention permitted”
In response, the guideline suggested that 1) good theoretical understanding be developed about how the intervention causes changes, in order to address potential
“weak links in the causal chain”, 2) the process evaluation be used to identify implementation problems (which could explain lack of impact); 3) sample size might need to be increased to be able to account for variability, while cluster- rather than individually-randomized designs might need to be considered; 4) studies should employ a range of measures and to identify unintended consequences; 5) interventions might need to adapt to local setting instead of strictly following protocols. The positions suggest a need to clarify complexity issues and reflect them in research approaches, though no example was found which used these guidelines in studying the balance between primary and hospital care.
This review has identified some complexity issues involved in shifting the balance
between primary and hospital care (gatekeeping in particular) and their implications for studying strategies to shift the balance between primary and hospital care.4 First, interventions under similar categorization are implemented in varied ways. All gatekeeping studies have involved some other changes in what could be seen as complementary arrangements. The gatekeeping study from Egypt, for example, involved increased charges and co-payment for patients bypassing primary care providers, a marketing campaign to raise awareness, and an exemption policy for the poor who received care at the primary care level or received care at hospitals with referral from primary care (Ward, 2010). Indeed, the findings of the Organisation for Economic Co-operation and Development (OECD) Health System Characteristics Survey 2012 showed varied arrangements of gatekeeping and cost-sharing policies for accessing outpatient specialist care across rich countries(Organization for Economic Co-operation and Development, 2013). Studies on pro-coordination reforms including gatekeeping also found that coordination reforms were not self-sustaining but required supporting conditions at systems, professional, and practices levels (Rico et al., 2003;
Saltman et al., 2005).
Second, there are issues related to the context within and beyond health systems.
A study in the Netherlands suggested that GPs use a “demand-satisfying” attitude when it comes to gatekeeping, even though they think patients are receiving unnecessary care (Wammes et al., 2014). Important contextual changes that drive policy changes include the increasing demand on coordination caused by rising prevalence of non-communicable diseases, as well as pressures related to cost-containment (Rico et al., 2003).
Third, there are issues related to the various and interrelated impacts of the
4 The review draws from Greenfield, Foley, and Majeed (2016) and Rico, Saltman, and Boerma (2003).
gatekeeping policies. Financial incentives for general practitioners in the United Kingdom (UK) to reduce specialist referrals caused concerns about conflicting roles and ethical issues (Lauridsen, 2009; Matthews-King, 2016). While pro-gatekeeping financial incentives seemed to reduce access to specialist care in France, particularly for poor and uninsured people (Dourgnon & Naiditch, 2010), gatekeeping also reduced overall health care inequities in European countries (Biro, 2013; Reibling & Wendt, 2013; Schnitzer et al., 2011), provided disadvantaged groups with assistance in decision making, and reduced specialist use by advantaged populations, who were the more frequent users of specialist care (Reibling et al., 2013). There have been different opinions regarding the influence of gatekeeping on the traditional divide between GPs and specialists, particularly in relation to the flow of information between the two groups (Bjornsson et al., 2010). Restrictions on patient choice are associated with lower patient satisfaction (Bjornsson et al., 2010; Dusheiko et al., 2007; Greenfield et al., 2012), though not everywhere (Gervas, Ferna, & Starfield, 1994), while low patient satisfaction has been shown to be associated with compromised outcome and compliance (Roter & Hall, 2006). Studies have also indicated that primary care gatekeeping may actually be associated with poorer cancer survival potentially due to delayed diagnosis (Crawford, 2014). In short, the results of policies to shift the balance between primary and hospital care (particularly gatekeeping) appear to include a range of interrelated consequences, some of which were not necessarily the intended outcomes of the interventions, highlighting the complexity of the interventions.
Fourth, history and institutions also influenced the implementation of strategies to shift the balance between hospitals and primary care providers. A comparative study (Rico et al., 2003) referenced above analysed trajectories of development in relation to strengthening the role of primary care both in service provision and coordination (including gatekeeping) across health systems in European countries during the 1990s.
The study showed that health systems in which primary care historically had strong
informal political power (particularly private ownership) and state had monopsonic control of health services tended to find it easier to implement primary care-focused pro-coordination reforms. The distribution of informal political power and state control were historically formed, and affected by policy feedback and societal pressures (cost-containment, market competition and increasing prevalence of chronic illnesses). In other words, this study indicated the importance of power dynamics in determining the outcome of policy interventions related to the balance between primary and hospital care. Such power appeared to come from historically shaped institutional factors that determined distribution of power and formation of coalitions among groups, as well as the political structure that affected the influence of these coalitions in driving or preventing policy changes.
2.5.3 Commentary
Section 2.5 has reviewed findings regarding strategies to shift the balance between primary and hospital care with a focus on gatekeeping. Research about gatekeeping has been especially rare, particularly in low- and middle-income settings.
Studies on effectiveness of strategies to shift the balance in China have been rare and of low quality. The point about gatekeeping involving some concurrent arrangements deserves particular attention.
The review has also identified complexity issues related to shifting the balance between primary and hospital care. Table 2-5 summarizes these complexity issues and extends to include their corresponding methodological implications.
Table 2-5. Complexity issues and methodological challenges No. Complexity issues Methodological implications
1 Varied implementation of interventions under the similar categories
Research on effective strategies needs to be able to isolate the impact of the component of interest, and be complemented with understanding of implementation details.
2 Variations in the context both within and beyond health systems
Research needs to deal with the variations within the context, including concurrent changes and unobservable factors.
Challenges of isolating the effectiveness of interventions from complex contexts
3 Multiple consequences, including interrelating and unintended consequences
Research needs to address not only the immediately intended outcomes, but also to examine the consequences not directly intended and the interrelationships between consequences of the intervention.
Research needs to understand how historically formed institutional factors influence the power dynamics affecting policy and system changes.
The issue of variations in implementation of intervention programmes suggests the need to go beyond the generic brands of certain intervention programmes and to identify the specificity of implementation. Methodologically, this means more ground-level understanding of implementation details, including the barriers to and facilitators of implementation in the context of interest. The issue of variations in the context both within and beyond health systems also requires research designs to explore/control for the variations within the context, which may include concurring changes and unobservable factors.
The issues about multiple consequences demand research methods to reveal the multitudes of interrelating consequences, and to be sensitive to interrelationships and unintended consequences. The important influences of historical and institutional factors on implementation of interventions also require a research approach to conceptualize and study such factors.
To sum up, it is difficult if not impossible to isolate effects of one intervention (or multiple interventions under one brand) embedded in multiple dimensions of complexity. Such complexity issues as discussed above pose methodological challenges that deserve attention to each one of them.