Thinking about Accountability

Abstract

Accountability is a key component of healthcare reforms, in Canada and internationally, but there is increasing recognition that one size does not fit all. A more nuanced understanding begins with clarifying what is meant by accountability, including specifying for what, by whom, to whom and how.

These papers arise from a Partnership for Health System Improvement (PHSI), funded by the Canadian Institutes of Health Research (CIHR), on approaches to accountability that examined accountability across multiple healthcare subsectors in Ontario. The partnership features collaboration among an interdisciplinary team, working with senior policy makers, to clarify what is known about best practices to achieve accountability under various circumstances. This paper presents our conceptual framework. It examines potential approaches (policy instruments) and postulates that their outcomes may vary by subsector depending upon (a) the policy goals being pursued, (b) governance/ownership structures and relationships and (c) the types of goods and services being delivered, and their production characteristics (e.g., contestability, measurability and complexity).

Accountability is a key component of many current healthcare reform efforts, both in Canada and internationally (Canadian Healthcare Association 2001; Leo and Canadian Healthcare Association 2006; Marchildon 2013). But there is increasing recognition that one size does not fit all, and that it is important to “unpack” the concept and to clarify both what the term accountability means, and which approaches to achieving it might work where. As Brown and colleagues (2006) have noted, “strengthening accountability is central to the recommendations made in all recent studies on the future of health care.” Yet there is insufficient information about best practices, and a sense that badly designed or implemented approaches may have unintended negative consequences.

This paper presents the conceptual framework that has guided the research reported in this Special Issue of Healthcare Policy/Politiques de Santé. The research emerged from a Partnership for Health System Improvement (PHSI), funded by the Canadian Institutes of Health Research (CIHR), on approaches to accountability (see also http://www.approachestoaccountability.ca). We use the term “approaches” to refer to the “big picture” that includes such elements as how (the instruments, tools or mechanisms being employed), to whom these elements apply, and the consequences of success or failure. The full project features collaboration among an interdisciplinary team, working in partnership with senior policy makers across multiple healthcare subsectors, to clarify what is known about best practices to achieve accountability under various circumstances. The research design has employed a series of case studies, using a common data collection template, to allow comparison across subsectors and across jurisdictions, to ascertain the impact of three categories of key independent variables – goals, governance/ownership and services provided – on how accountability is defined and the advantages and disadvantages of the possible approaches.

The Policy Issue

Accountability has multiple definitions (Mulgan 2000). Most simply, it means having to be answerable to someone, for meeting defined objectives (Emanuel and Emanuel 1996). This can be done in a variety of ways, using a variety of policy instruments. In practice, however, accountability has often proven difficult to achieve, and there are lurking suspicions that approaches suitable under certain circumstances may be suboptimal or counterproductive in other settings. Clarification about the best ways to achieve accountability has been identified as a major priority by governments, providers and recipients of healthcare services, both in Canada and internationally.

The literature suggests that accountability has financial, performance and political/democratic dimensions (Brinkerhoff 2004) and can be ex ante or ex post. Within healthcare, these three dimensions may translate into fiscal accountability to payers, clinical accountability to a variety of actors for quality of care (Dobrow et al. 2008) and accountability to the public. The actors involved may include various combinations of providers (public and private), patients/service recipients, payers (including insurers and the legislative and executive branches of government) and regulators (governmental, professional) who are connected in various ways (Shortt and Macdonald 2002). The ways to establish and enforce accountability are similarly varied. A more nuanced understanding necessarily begins with clarifying what is meant by accountability, including specifying for what, by whom, to whom and how. A related set of concepts are linked to rewards and punishments, including what happens when outcomes are not achieved. Under those circumstances, being answerable may translate into efforts to censure, shame and blame those who are seen as “accountable” for the failure. In these studies, we concentrate on the potential consequences (both intended and unintended) of using various approaches to ensure that goals are met and performance is being improved.

We accordingly constructed an analytical framework drawing from several literatures that have not previously, to our knowledge, been used to analyze the factors affecting the strengths and weaknesses of various approaches to accountability. Rather than focus on “models” that attempt to force an intrinsically variable concept into boxes, we instead focused on dimensions that might affect performance, policy development or both. One dimension of the framework classified potential approaches to accountability in terms of the political science concept often referred to as “policy instruments” or “governing instruments.” We concentrated on four, which were evident in our review of currently used approaches to accountability: financial incentives, regulations, information directed towards patients/payers and professionalism/stewardship – which represented variations on the “expenditure,” “regulation” and “exhortation” governing instruments. Our framework postulated that these approaches would have differing success when applied to various categories of services, and within various subsectors, with the likely outcomes depending upon three additional dimensions: (a) the policy goals being pursued (which affect the question of “accountability for what”); (b) the governance/ownership structures and relationships in place, which in turn affect who will be accountable and to whom; and (c) the types of goods and services being delivered and their “production characteristics.”

To the extent that different subsectors represented different configurations of governance, ownership and service mix, some of these dimensions will not vary within a particular sub-sector, while others may vary within and between settings. There is also scope for variation depending on the characteristics of those receiving the services, and their ability to monitor performance. The research reported here thus allowed us to compare and contrast across a series of related substudies.

Approaches: Policy Instruments

Political scientists have noted that decision-makers have available to them a series of what are termed “policy instruments” or “governing instruments,” from which they can select to carry out preferred policy directions (Baxter-Moore 1987; Doern and Phidd 1992; Hood 1983; Howlett and Ramesh 1993).

There are a number of different ways of classifying such instruments. Doern and Phidd (1992) stress increasing government involvement/control, ranging from encouraging cooperation to taking over and directly running the activity; they use the terms exhortation, expenditure, regulation, taxation and public ownership. Another formulation, which makes similar points in slightly different language, is Hood's NATO, which uses the terms nodality (information), authority, treasure and organization (Hood 1983).

These typologies stress that policy instruments vary considerably in how coercive (or intrusive) they are. At the extreme non-intrusive end of the scale, decision-makers may choose not to act at all. Moving slightly up the scale, they can choose symbolic responses to encourage people to act in a particular way. This may involve information/education, symbolic gestures or both. Doern and Phidd (1992) term this approach “exhortation.” Somewhat more intrusively, decision-makers may choose to intervene indirectly by using incentives for action, ranging from attempts to secure voluntary compliance with their objectives without accompanying threats or inducements, through to “expenditures” and/or “taxation” policies. A still more intrusive set of instruments may be termed directives (what Doern and Phidd call “regulation”); these often shift compliance costs from regulators to those being regulated. Although most writing on policy instruments conceptualizes coercion as the extent to which government directly intrudes on private decision-making, this analysis can be extended to examining the potential intrusiveness of one level of decision-makers upon others. The framework also includes the literature on ways of enforcing these agreements, including information, licensure/accreditation, payment and legal sanctions. This literature has been linked to the literature on the new public management (Hood 2000) and pays particular attention to interactions between public and private forces, as well as the implications of the type of policy network for selection of policy instrument (Bressers and O'Toole 1998). Although these concepts have been applied to the field of environmental regulation, particularly in the European Union (Jordan et al. 2005; Zito et al. 2003) as well as, in a more limited way, to healthcare – e.g., the governance of primary care in Switzerland (Braun and Etienne 2004) and social services, such as child health policy in Australia (Leggat 2004) – we are not aware of any efforts to apply them to accountability.

In these studies, we have focused on the following four major approaches to accountability currently being employed in the health sector in Canada and internationally:

1. Financial incentives, which adjust payments to induce providers to behave in a desired manner (Donaldson et al. 2005; Evans 1984; Robinson 2001). These employ the “expenditure” governing instrument (also called “treasure”). One example is the family of pay-for-performance (P4P) experiments for physician services underway in such jurisdictions as the United Kingdom, United States, Australia and Ontario (Devlin et al. 2006; Doran et al. 2006; Epstein 2006; Pink et al. 2006; Rosenthal et al. 2005), which often tie funds to performance of desired activities (Marks et al. 2011). Another involves changes in the financial incentives for hospitals, including moves to activity-based funding (Chalkley and Malcomson 2000; Sutherland 2011; Sutherland et al. 2011). Similar initiatives can be found in other subsectors, including home care.

2. Regulations, which by definition employ the “regulation” governing instrument (Walshe 2003), play a major role in healthcare. These require providers to behave (or not behave) in a certain way. Although regulations can be backed up through signing binding agreements, they may also rely on agency theory (Eisenhardt 1989; McGuire 2000) and be enforced using professional regulatory bodies. The literature notes the ongoing tension of balancing market forces and regulation (Chinitz et al. 1998; Saltman et al. 2002) and the implications of regulatory and medico-legal barriers for achieving such goals as interprofessional practice (Lahey and Currie 2005).

3. Information directed towards potential users (patients, public and private payers) within a context of allowing market forces to work more effectively by encouraging rational choice of the “best” care (Howells 2005). These are one variant of the “exhortation” governing instrument. This instrument may work both directly and indirectly (e.g., interest groups and media may affect the reputation of various providers, which in turn affects the willingness of patients and payers to purchase their services). Examples of the use of information include ongoing activities in performance measurement and improvement (Barnsley et al. 2005; Hurst and Jee-Hughes 2001; Shaw 2003; Smith 2002; Veillard et al. 2005). Issues in using this approach include who establishes these measures and who enforces them. In healthcare, common examples include the use of report cards (e.g., for hospitals), audit reports, publicly available inspection reports (e.g., for nursing homes) and quality indicators, including adverse events (Baker et al. 2004). The Health Council of Canada and the Canadian Institute for Health Information have been involved in developing, collecting and publicizing various informational mechanisms.

4. Reliance on professionalism and stewardship (Saltman and Ferroussier-Davis 2000; Shortell et al. 1998). This approach employs a second variant of the exhortation governing instrument, but directs the information to providers rather than to payers or consumers (Lemieux-Charles and Champagne 2004). It relies on high trust and the expectation that providers – as a group – wish to do the right thing, but may need support in clarifying best practices as well as exposing poor practice. Clinical guidelines and other forms of evidence-based practice often fall within this category if compliance is voluntary; they are currently being used in various subsectors, including hospitals, nursing homes and primary healthcare. Under some circumstances, report cards may also fall into this category of approach, depending upon the indicators used and the dissemination approach adopted. Note that this approach is often backed up by regulatory approaches (e.g., through self-regulation of the professions).

An additional nuance is the extent to which blended models may be used. This is particularly evident in the use of additional policy instruments for enforcement, which may include various combinations of information (e.g., efforts to evaluate and improve the quality of information, citizen engagement to widen the scope of inputs) (Abelson and Guavin 2004), expenditure (e.g., fiscal penalties), taxation (e.g., tax breaks to encourage desired activities) and regulation (e.g., audit, accreditation, professional self-regulation and legal sanctions). Some jurisdictions have established formal appeal mechanisms for patients (e.g., Norway has a Patient's Bill of Rights; Ontario has the various review procedures for getting out-of-province coverage and a public complaints process for nursing homes). Other enforcement mechanisms may rely upon litigation (e.g., malpractice, human rights). Auditor General reports fall within the information category, but may often catalyze additional actions.

What Affects the Impact of Accountability Approaches?

The second component of this framework focuses on three dimensions likely to affect the success of various approaches: goals, governance/ownership and services.

Policy goals may encompass both outcomes and processes. Policy goals for healthcare traditionally include combinations of access, quality (including safety), cost control/cost-effectiveness and customer satisfaction. Behn (2001) suggests three objectives for accountability: improved performance, fairness and financial stewardship. Often, policy goals may clash. For example, hospital effectiveness may vary if measured in terms of doing more (increasing the number of admissions, market share, occupancy rate), financial performance (net profit, cash flow), meeting the needs of the community (satisfaction of patients, providers) or delivering high-quality care (better outcomes). Ideally, there should be congruence between the policy goals being pursued and what organizations are being held accountable for, although this is not always the case.

The governance/ownership structures in place also vary across jurisdictions and across subsectors; they affect who will be accountable, to whom and for what (Denis 2004; Jordan et al. 2005; van Kersbergen and van Waarden 2004). One key element is the balance between and rationale for public versus private provision and how such factors affect governance and accountability (Deber 2004; Horwitz 2006; Osborne and Gaebler 1992; Sloan 2000). For example, for-profit organizations also have a fiduciary duty to maximize the return to their investors, which in turn may affect the services they choose to provide and the populations they choose to serve.

As Denis and colleagues (2005) have noted, “governance deals principally with the adaptation of organizations to new contingencies” and deals with “the roles of all regulatory, administrative, professional and clinical authorities in the pursuit of collective goals.” In that connection, structures where actors are accountable to more than one authority can add complexity (Rhodes 1997). An additional complexity is that Canada is what the Organisation for Economic Co-operation and Development (OECD) refers to as a “public-contract model,” in which public payers contract with private healthcare providers (Docteur and Oxley 2003).

In turn, this means that many accountability arrangements are between government and the “third sector” (“civil society”), a situation that presents additional complications (Mayne and Wilkins 2005; Schwartz 2001, 2002, 2003).

Ownership presents further issues; one US study found that the effects of the governance/ownership configuration were more pronounced in freestanding and public not-for-profit hospitals compared with system-affiliated and private not-for-profit hospitals. The board structure, including whether corporate-style or philanthropic-style models were used, was also important (Alexander and Lee 2006). Some have suggested that this governance/ownership category of variables may be the most amenable to policy change (Preker and Harding 2003). One potentially useful concept is the “soft governance” approach, in which government “relies less on hierarchy than on information to steer local organizations” (Brandsen et al. 2006). There is literature suggesting a relationship between governance/ownership and the ability to achieve and monitor such goals as quality improvement (Baker et al. 2006; Thomas 2006). A helpful resource is the framework developed by Denis and colleagues (2005) for Accreditation Canada, which identifies three governance models (agency, stakeholder and stewardship).

We also postulate that what economists call the production characteristics of the goods and services being delivered – specifically, what the literature refers to as contestability, measurability and complexity – will affect performance measurement and monitoring (Deber 2004; Jakab et al. 2002; Preker and Harding 2000, 2003; Preker et al. 2000; Rico and Puig-Junoy 2002; Vining and Globerman 1999). Preker and Harding (2000) define them as follows: “Contestable goods are characterized by low barriers to entry and exit from the market, whereas non-contestable goods have high barriers such as sunk cost, monopoly market power, geographic advantages and ‘asset specificity.’” These authors define measurability as relating to “the precision with which inputs, processes, outputs and outcomes of a good or service can be measured.” Monitoring performance is easiest when measurability is high. For example, it is relatively simple to specify the performance desired for conducting a laboratory test or collecting municipal garbage. In contrast, it would be more difficult to specify the activities to be expected of a general practitioner, and hence more difficult to monitor the physician's performance and ensure quality. Complexity refers not to how complex the particular goods and services are, but to whether the goods and services stand alone or require coordination with other providers. For example, laboratory tests are highly measurable, but gain much of their value by being embedded within a system of care in which providers order tests appropriately and are aided in interpreting and acting upon their results. A related concept that emerged from our research is observability, which we define as the extent to which the activities are easily monitored by others beyond those directly involved in providing or receiving a service. For example, services delivered in a home setting would be less observable than those delivered in a hospital operating room. Additional insights arise from the theory of transaction costs and monitoring costs, which are also addressed in economics theories relating to incomplete contracting (Williamson 1981, 1985, 1999), defined by Williamson (1985) as “the comparative costs of planning, adapting and monitoring task completion under alternative governance structures.”

Substudies and Decision-Making Partners

The substudies examined in this Special Issue encompass a variety of subsectors, including hospitals (Kraetschmer et al. 2014; Kromm et al. 2014), cancer (Bytautas et al. 2014), community services (Steele Gray et al. 2014), laboratories (Gamble et al. 2014), public health (Schwartz et al. 2014), primary healthcare (Mukhi et al. 2014), long-term care (Berta et al. 2014; Wyers et al. 2014) and professional regulation (Baumann et al. 2014; Zelisko et al. 2014), and brief commentaries looking at international comparisons (Kirsch 2014; Peckham 2014) and the role of accreditation (Mitchell et al. 2014). These substudies represent different combinations of governance/ownership and services, and should assist in clarifying the advantages and disadvantages of various approaches to accountability. We are extremely fortunate to have the collaboration of such highly qualified decision-making partners and researchers (including graduate students).

The final paper of this Special Issue (Deber 2014) summarizes what we have learned from the substudies.