i need someone who has ability to do a research paper in the Law of Health care
Defining quality of care
· S.M Campbell , ,
· M.O Roland ,
This paper defines quality of health care. We suggest that there are two principal dimensions of quality of care for individual patients; access and effectiveness. In essence, do users get the care they need, and is the care effective when they get it? Within effectiveness, we define two key components — effectiveness of clinical care and effectiveness of inter-personal care. These elements are discussed in terms of the structure of the health care system, processes of care, and outcomes resulting from care. The framework relates quality of care to individual patients and we suggest that quality of care is a concept that is at its most meaningful when applied to the individual user of health care. However, care for individuals must placed in the context of providing health care for populations which introduces additional notions of equity and efficiency. We show how this framework can be of practical value by applying the concepts to a set of quality indicators contained within the UK National Performance Assessment Framework and to a set of widely used indicators in the US (HEDIS). In so doing we emphasise the differences between US and UK measures of quality. Using a conceptual framework to describe the totality of quality of care shows which aspects of care any set of quality indicators actually includes and measures and, and which are not included.
Keywords
· Quality;
· Care;
· HEDIS;
· NHS
Introduction
Growing demand for health care, rising costs, constrained resources, and evidence of variations in clinical practice have increased interest in measuring and improving the quality of health care in many countries of the world. Quality improvement is high on the national agenda both in the UK ( Roland, Holden & Campbell, 1999 ) and in the USA (Schuster, McGlynn & Brooks, 1998). There has been a move away from assessing costs and activity to assessing quality with an emphasis on both efficient use of resources and on the effectiveness of health care. In the UK this trend has led to an emphasis on both performance measurement and quality improvement for the NHS, and the development of a national performance framework ( Department of Health, 1997 and NHS Executive, 1999 ). Within this, clinical governance will form a framework through which ‘NHS organisations are accountable for monitoring and improving the quality of their services’, with the aim of promoting ‘an environment where excellence of clinical care will flourish’ ( NHS Executive, 1998a and NHS Executive, 1998b ).
The increasing focus by governments on improving quality of care requires that the concept is clearly understood. In this paper we define quality of care. We focus specifically upon care received by individuals from formal institutional health care systems which individuals or carers have chosen to access. Care in this context refers to care provided by any health care professional. We suggest that there are two dimensions of quality in this context; access and effectiveness. In essence, do users get the care they need, and is the care effective when they get it? There are two key elements of effectiveness — clinical effectiveness and the effectiveness of inter-personal care — and both of these should be related to need. These elements are discussed with reference to the structure of the health care system, processes of care and outcomes resulting from care. The aim of accessing effective health care for individuals is to maximise health benefit according to need. We suggest that quality of care is a concept that is at its most meaningful when applied to the individual user of health care.
However, care for individuals must be placed within the context of providing health care for populations. We also define quality of care for populations by reference to a separate but overlapping set of criteria reflecting the opportunity cost of providing care for any individual patient or group of patients. We show how our framework can be of practical value by applying the concepts outlined to the set of performance indicators proposed for use in the UK ( NHS Executive, 1999 ) and a widely used set of US quality indicators, HEDIS 3.0 ( National Committee for Quality Assurance, 1998 ). A framework is important as it both increases understanding of what such sets of indicators mean and also those aspects of care that should be covered and which are, and are not, actually being covered.
Many symptoms experienced by individuals are not presented to formal health care systems ( Rogers, Hassell & Nicholaas, 1999 ). We recognise that pathways to formal care are complex and multidimensional ( Campbell and Roland, 1996 and Rogers et al., 1999 ). They are dependent on factors which include the sociodemographic characteristics of the population ( Ben-Shlomo, White & McKeigue, 1992 ; MacIntyre, McIver & Sooman, 1993 ), health need ( Evandrou, Falkingham, Le Grand & Winter, 1992 ; Feinstein, 1993 ) and factors such as lay support ( Robinson & Granfield, 1986 ; Oakley, 1994 ), frequent attendance ( Neal, Heywood, Morley, Clayden & Dowell, 1998 ) and health beliefs ( Egan and Beaton, 1987 , Murray and Corney, 1990 and Van der Kar et al., 1992 ).
However, this paper does not focus on the attributes of health care systems which might influence consulting behaviour. Rather, we focus specifically upon care received by individuals from formal institutional health care systems starting from where an individual or carer has decided to access care. We first define the components of care, then the attributes of quality. We bring these together to provide a framework to describe quality of care.
Defining health care
Definitions of care are varied and contested ( Heyman, 1995 , Webb, 1996 and McCance et al., 1997 ). For the purposes of this paper, we define health care as being composed of health care systems and actions taken within them designed to improve health or well-being. Donabedian, 1966 and Donabedian, 1988 ) first proposed a systems-based framework of structure, process and outcome. These have also been described as categories of care ( Ellis & Whittington, 1993 ). Donabedian’s model has been used previously as a basis for defining quality ( Steffen, 1988 , Tarlov et al., 1989 , Baker, 1995 and Irvine and Donaldson, 1993 ).
It is necessary to distinguish between the structure of health care, actual care given (process) and the consequences of the interaction between individuals and a health care system (outcome). We suggest that outcome is not a component of care but a consequence of care. Similarly, structure is not a component of care but the conduit through which care is delivered and received. Fig. 1 shows the key domains underlying structure, process and outcome for care received by individuals from a health care system and the dimensions of each of these domains.
Fig. 1.
A systems based model for assessing care.
Structure
Structure refers to the organisational factors that define the health system under which care is provided ( Donabedian, 1980 ). We identify two domains of structure: physical characteristics and staff characteristics, and Fig. 1 shows the dimensions of each of these domains. Components of the dimension of resources include, for example, personnel, equipment and buildings. The ways in which those services are organised would include, for example, opening hours and the existence or otherwise of a booking system for appointments. These are both structural elements of the health care system. Structural features of health care provide the opportunity for individuals to receive care but do not guarantee it. On the whole structures, whilst being able to increase or decrease the likelihood of receiving high quality care, are indirect and contingent influences on care. However, structural features within a systems based model of care can have a direct impact on processes and outcomes, e.g. if necessary equipment or skills are not available to undertake an effective examination or if all appointments are booked-up prohibiting a patient accessing care.
Process
Processes of care involve interactions between users and the health care structure; in essence, what is done to or with users. Process is the actual delivery and receipt of care. Two key processes of care have often been identified: technical interventions and inter-personal interactions between users and members of a healthcare system ( Blumenthal, 1996 , Donabedian, 1988 , Donabedian, 1992 , Irvine, 1990 , Tarlov et al., 1989 and Steffen, 1988 ).
Technical care refers to the application of clinical medicine to a personal health problem ( Donabedian, 1980 ) and is based upon a theory of function which can be evaluated for efficacy and generally standardised. Care should be appropriate and necessary ( Kahan et al., 1994 and Brook, 1994 ). In practice, care is often overused, i.e. provided when inappropriate, and underused, i.e. not provided when necessary ( Brook et al., 1996 and Schuster et al., 1998 ). Both necessary and appropriate care must be seen from both ends of the scale; for example, appropriateness is used as much to define what is inappropriate as appropriate.
Both ‘clinical care’ and ‘technical care’ have been used to describe the more bio-medically oriented aspects of health professional’s behaviour. We suggest that clinical care is the more appropriate term to use as there are also technical aspects to inter-personal care, e.g. specific skills in relation to giving information to patients. We therefore define the process of care in terms of clinical and interpersonal aspects of care.
Interpersonal care describes the interaction of health care professionals and users or their carers. This includes “the management of the social and psychological interaction between client and practitioner” ( Donabedian, 1980 ). A number of skills underlie good inter-personal skills including: communication, the ability to build a relationship of trust, understanding and empathy with the patient ( Blumenthal, 1996 ) and to show humanism, sensitivity and responsiveness ( Carmel & Glick, 1996 ). Patients want explanation and discussion about their symptoms ( Woloshynowych, Valori & Salmon, 1998 ), and to be involved in decisions about their management.
Both clinical and inter-personal care processes involve the definition and communication of problems or needs, diagnoses, their management and co-ordination by the patient and professional concerned. In addition to actions taken by an health professional, identified as elements of process in Fig. 1 , the process of care may be further classified into preventive care, care for chronic disorders, and care for acute illness. This classification is particularly relevant to quality measurement, since quality indicators abound for the first, are patchy for the second, and almost absent for the third. The lack of quality indicators for acute minor illness has important implications for assessing the quality of general/family practice where they form the bulk of all presenting problems. Clinical and inter-personal aspects of care are relevant for each consultation and apply equally to chronic, acute or preventive care.
Outcome
Outcomes are consequences of care. Structure as well as processes may influence outcome, indirectly or directly. For example, a patient may die from cervical cancer either because a screening service is not available (structure) or because her cytology report is misread (process). However, the relative importance of each of these components will vary in different situations and the relationships between them are not necessarily linear.
The effectiveness of structure and processes (both clinical and inter-personal) can be defined in terms of their capacity to result in two principal domains of outcome: health status and user evaluation — incorporating non-health as well as health related outcomes. There may be feedback loops with, for example, an individual’s user evaluation (outcome) influencing their subsequent consulting behaviour, or care negotiated in one consultation affecting subsequent decisions ( Rogers et al., 1999 ).
Our framework focuses upon care for individual users so outcome in Fig. 1 refers to health status and user evaluation, e.g. satisfaction, enablement ( Howie, Heaney & Maxwell, 1996 ) and health related quality of life. These must also be related to patient expectations and to the needs of that individual patient ( Stott et al., 1997 ). User evaluation may include assessment of outcome (symptom resolution) and/or processes of care (e.g. communication skills of the health professional). Although outcomes may be explicitly related to process for both health status (e.g. Antiplatelet Trialists’ Collaboration, 1994 ) and user evaluation (e.g. Schmittdiel, Selby, Grumbach & Quesenberry, 1997 ), this is not always the case ( Long & Fairfield, 1996 ). This is because outcomes are often causal rather than logical consequences of processes and their activation is contingent upon a wide variety of factors. Outcome C will not always follow logically from process B using structure A. Care processes may have little or no consequence on disease outcomes if, for example, an illness has an immutable natural history.
There has been considerable debate about whether process or outcome should be assessed as measures of quality of care ( Davies and Crombie, 1995 and Brook et al., 1996 ). In general, process measures are better indicators of quality of care if the purpose of measurement is to influence the behaviour of the health care system: processes are common, under the control of health professionals, and may more rapidly be altered. Outcomes are often rare, may follow a change in process by up to 10 years (e.g. management of hypertension), and may be dependent on factors outside the control of the individual health professional ( Giuffrida, Gravelle & Roland, 1999 ). However, process measures suitable for measuring quality should be clearly linked to evidence of improved outcomes.
We have defined care as a systems based model composed of health structures and two processes of care (clinical and inter-personal) which result in consequent outcomes. However, whilst systems and processes increase or decrease the likelihood of individuals receiving the care they need, they do not guarantee quality care. We now consider the concept of quality.
Defining quality
Quality has been defined in a number of different ways ( Crosby, 1979 , Donabedian, 1980 , Maxwell, 1984 , Peters, 1987 and Juran, 1988 ). Definitions of quality are either generic or disaggregated — approaches which are not inherently incompatible but can be seen as opposite ends of a continuum. Generic definitions of quality include excellence ( RCGP, 1994 ), expectations or goals which have been met (Ellis & Whittingham, 1993; Steffen, 1988 ), ‘zero defects’ ( Crosby, 1979 ) or fitness for use ( Juran, 1988 ). Other generic definitions are more complex. For example, the Institute of Medicine ( Lohr, 1992 ) has defined quality as the ‘degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge’. Generic definitions are not easily operationalisable and trade both sensitivity and specificity for generalisability.
Disaggregated approaches, on the other hand, recognise that quality is complex and multidimensional ( Donabedian, 1980 , Maxwell, 1984 and HSRG, 1992 ; Winefield, Murrell & Clifford, 1995 ). They define quality according to individual dimensions or components. Some examples of previous definitions of quality of care using disaggregated approaches are included in Fig. 2 .
Fig. 2.
Examples of previously described dimensions of quality of care.
Each individual component of quality provides a partial picture of quality when viewed on its own, but offer more specificity in defining quality when viewed in combination. However, we propose that there are only two domains of quality — access and effectiveness. For individual patients, all dimensions of quality suggested previously, including those listed in Fig. 2 , can be subsumed by and appropriately regarded, as elements of access and effectiveness. Quality of care may, in this way be reduced to two questions. First, can an individual get the care they need when they need it? Second, when they get the care, is it effective both in terms of clinical effectiveness and inter-personal relationships? We now develop this proposition by applying access and effectiveness to the systems based model of care (structure, process and outcome) described above.
Defining quality of care for individual patients
We propose a combined generic and disaggregated approach to defining quality of care. Our definition of quality of care for individuals is:
whether individuals can access the health structures and processes of care which they need and whether the care received is effective.
Fig. 3 shows how access and effectiveness can be related to health care structures, to processes of care, and outcomes, to produce a taxonomy of quality of care for individual patients. We now develop this argument by considering separately access and effectiveness.
Fig. 3.
Dimensions of quality of care for individual patients (dimensions of quality of care for populations are considered separately in the text).
Access
The first part of our definition of quality of care emphasises “whether individuals can access health structures and processes of care which they need”. The most basic dimension of access to a health structure is geographic/physical access ( Haynes, 1991 ). This may relate to geographic barriers to getting to health care facilities such as rurality ( Watt et al., 1993 and Cox, 1994 ), or use of premises themselves such as access by disabled or elderly people ( Bentham & Haynes, 1985 ). Availability is the extent to which the health care system provides facilities (structures) and services (process) which meet the needs of individuals. Examples include access to a female general practitioner, a specialist clinic, or a member of a health care team other than a doctor (e.g. nurse practitioner, physiotherapist or counsellor). The latter is increasingly important given the emergence of a far more diverse range of skills among health care providers as part of attempts to increase the cost–effectiveness of care ( Pritchard & Pritchard, 1994 ).
Organisational access is a sub-component of availability. If people are physically able to access a health facility, they may still face barriers to accessing care in terms, for example, of the length and availability of appointments, or whether the health professional can speak their language. In a system where the primary care practitioner acts as a gatekeeper, access to secondary care or other members of the PHCT is determined by access to the primary care practitioner and by his or her referral behaviour.
We suggest that three of the four attributes of quality identified as being of particular relevance to primary care ( Starfield, 1994 ) are sub-components of access. First contact, comprehensiveness — a range of services and care broad enough to meet all common needs as they occur — and provider continuity or longitudinality are all sub-components of the structural aspects of availability, rather than quality attributes in their own right. Our view is that continuity of care is not a unique dimension of quality but a structural characteristic that may produce benefits, though is not guaranteed to do so ( Freeman & Hjortdahl, 1997 ). For example, continuity of care from an ineffective professional with poor communication skills does not constitute high quality care, even though such a professional could nevertheless score well on access and patient evaluation dimensions.
In this paper, we do not include equity as a component of quality of care for individuals as equity requires a element of comparison of individuals within a population; however, we do regard equity as a key element of quality of care at the population level, which we discuss later.
Monetary costs of access at the individual level are only relevant to the payer. This may be the patient where services are not free at the point of delivery. So affordability is a key component of access in countries where health care is not provided by the state. Financial barriers may also be important in other systems, either in terms of the material cost of attending an appointment (transport costs, childcare, prescription charges) or the opportunity cost incurred, for example, in loss of earnings. Other barriers include having to accommodate competing social role commitments into the daily routine which may prevent, for example, mothers with young children accessing health care on their own behalf ( Young, 1996 ).
Health status and user evaluation are the principal outcomes of quality of care for individuals relating to access ( Fig. 3 ). User attitudes cannot be separated from nor properly understood without reference to the experiences with which they are connected ( Cornwell, 1984 ). These evaluations may influence future decisions about accessing care; e.g. an individual may feel disinclined to consult if the last time they consulted they were treated unsympathetically ( Murray & Corney, 1990 ). User evaluation thus forms part of a feed back loop with acceptability a dimension of outcome within user evaluation which may influence whether an individual seeks to access formal care in the future.
Effectiveness
The second part of our definition of quality of care relates to whether, when accessed, the care received by an individual is effective. Effectiveness is the extent to which care delivers its intended outcome or results in a desired process, in response to need. Fig. 3 shows the two key elements of effectiveness — clinical care and inter-personal care.
The effectiveness of clinical care depends on the effective application of knowledge based care. Knowledge-based care refers to both evidence-based medicine ( Sackett, Rosenberg, Gray, Haynes & Richardson, 1996 ) and care which is regarded as legitimate ( Donabedian, 1990 ). The latter relates to aspects of care that may be widely accepted without necessarily having scientific evidence of effectiveness. Knowledge-based care incorporates the extent to which a treatment or service is consistent with patients’ reasonable expectations and contemporary professional standards of care, reflecting both societal and professional norms. For example, primary care counselling has been found to be popular with patients ( Priest et al., 1996 and Scott and Freeman, 1992 ) and is widely implemented within primary care in the UK ( Sibbald, Addlington-Hall, Brenneman & Freeling, 1993 ) but as yet there is limited evidence for its effectiveness ( Roth & Fonagy, 1996 ). Care is only described as evidence-based when there is good scientific evidence of a link between process and outcome; for example, the role of aspirin in secondary prevention of coronary heart disease (Antiplatelet Trialists Collaboration, 1994).
Once again, processes and outcomes are context specific for each individual person because quality of care for individual users requires that individual circumstances and the complexity of individual patients must be considered. Effective care, therefore, requires appreciation of the patient’s personal experience of illness ( Stewart, Brown, Weston, McWhinney, McWilliam & Freeman, 1995 ) and must align the agendas of the professional and the patient ( Tuckett, Boulton, Olson & Williams, 1985 ). Care should be planned for and with individual patients ( HSRG, 1992 , p. 2154; Stewart et al., 1995 ) through negotiation between doctor and patient ( Buetow, 1999 ) and shared responsibility for care ( McGlynn, 1997 ). Focusing on the effectiveness of patient centred processes reduces the risk of over-emphasising scientific/technical aspects of care at the expense of the experiences of the individual patient. Originating in the psychodynamic study of professional patient interactions ( Balint, 1964 ), patient centredness is increasingly regarded as a set of skills which can be taught ( Stewart et al., 1995 ) and which are related to improved outcome ( Howie, Hopton, Heaney & Porter, 1992 ). Patient-centredness is concerned with the meaning of care for individuals ( Stewart et al., 1995 and HSRG, 1992 ) rather than care which is generalisable to all.
Co-ordination or integration of care for individual patients is also an important attribute of quality (i.e. effectiveness) of care, and is a further attribute of particular relevance to primary care ( Starfield, 1994 ). We suggest that co-ordination is a sub-component of both clinical effectiveness and the effectiveness of inter-personal skills. Co-ordination refers to the effectiveness with which health professionals deal with those other organisations, or other professionals within the same organisation, which impact directly or indirectly upon the health or health related quality of life of the patient (e.g. social care, education or housing).
We suggest that effectiveness, when related to individual patients, should refer to maximising care and desired processes and outcomes based upon need. We consider efficacy ( Donabedian, 1990 and O’Leary and O’Leary, 1992 ) to be a redundant component of quality of care because we are unable to conceptualise high quality care which involves delivering treatments which are efficacious (in ideal settings), but ineffective in routine healthcare settings. We also reject acceptability as a basic component of quality or as an attribute of care: it is more appropriately regarded as an outcome or consequence of care: it is a component of users’ evaluation.
Stakeholder perspectives of quality of care
Much has been written about the fact that different stakeholders (e.g. users, professionals and payers) define and perceive quality of care differently ( McGlynn, 1997 , Annandale and Hunt, 1998 , Ovretveit, 1992 , Joss and Kogan, 1995 and Blumenthal, 1996 ; RCGP, 1994). The contingent nature of quality means that there will be different perspectives upon the dimensions of access and effectiveness and on what constitutes a desired process or outcome. For example, efficiency and cost–effectiveness are principally concerns of managers (reflecting a more societal perspective) whereas users emphasise patient-centredness in response to their (individual) needs, e.g. humaneness, kindness, communication and access ( Wensing, Grol & Smits, 1994 ). Differences between users and health professionals ( Van der Waal, Casparie & Lako, 1996 ), for example when an evidence-based intervention designed to produce a outcome is not wanted by the patient ( Mulley, 1995 ), will provide examples of the effectiveness of the patient-centredness of care.
For some stakeholders, e.g. a national healthcare system, it is the outcome for a population which is of greatest importance, rather than the outcome for individuals. We now discuss how quality of healthcare may be defined for populations, since this differs in some important respects from quality of healthcare for individuals.
Defining quality of care for populations
We have suggested that quality of care has most meaning when applied to the level of the individual user of health care. After all, it is individuals who are the recipients of care. Questions about quality of care for individuals cannot be answered by reference to populations, only put in to the context of a social construction of quality of care. Borrowing from economics we recognise that care is not given as part of a closed model ceteris paribus. For example, most economic evaluations of health care adopt a societal perspective and focus on the optimisation of health benefit for populations ( Craig & Sutton, 2000 ). Whilst quality of care for individual patients may therefore be understood with reference to the framework outlined in Fig. 3 , it also needs to be understood in terms of, and may be contingent upon, care for populations.
Care for whole populations may potentially conflict with care for individual patients, for example, because of prioritisation of resources towards particular health needs. We have argued that consideration of quality of care for an individual requires information which is context specific to that user. However, a societal perspective requires information about all potential users of health care ( Birch & Gafni, 1992 ) and this requires consideration of opportunity costs. What is done to an individual user of health care (in terms of resources, including time as well as costs) may, therefore, be contingent upon what is done to all other users of the health care; particularly within systems with resource constraints. This divergence reflects Donabedian’s distinction between individual and social quality ( Donabedian, Wheeler & Wyszewianski, 1982 ), with the former based upon the individual and the latter emphasising aggregate net benefit (or net utility) for the entire population ( Donabedian, 1980 ). At the population level we propose that there are three additional key factors: equity, efficiency and cost which contribute to our model of quality of care for populations. We therefore define quality of care for populations as:
the ability to access effective care on an efficient and equitable basis for the optimisation of health benefit/well-being for the whole population.
Access
Quality of care can only be understood within the overall context in which health care is provided. Local or national factors help define the context in which health care is provided in organisations — and these are often outside the control of health professionals ( Heath, 1995 and Goldstein and Speigelhalter, 1996 ). Examples of such factors include the availability of primary and secondary health care services, such as community psychiatric services and the availability of social services or local authority services. Theses are dimensions of structural availability at a societal level. Supply factors have been found to be important determinants of utilisation ( Evandrou et al., 1992 ).
In our framework, we define equity as a sub-component of access relevant to structure and process; it is defined as the extent to which all individuals in a population access the care they need. Definitions of access must emphasise a notion of access according to need and the timely use of services ( Evandrou et al., 1992 , Maxwell, 1992 , Field and Lohr, 1992 and IOM, 1993 ); both in terms of horizontal equity (equally accessible to effective care for all users), and vertical equity (greater access to effective care for those with more need). The test of equity are systematic differences in use and outcomes ( IOM, 1993 ). We support Maxwell’s (1992) concept of relevance as whether ‘the overall pattern and balance of services (is) the best that could be achieved, taking into account the needs and wants of the population as a whole’ but suggest it is a sub-component of availability. In essence, services should be available on an equitable basis according to need. As with care for individual patients, health status and user evaluation are key outcomes of access for populations, in addition to equity. However, equitable access does not automatically translate into equitable effectiveness.
Effectiveness
Within the domain of effectiveness, we again define equity as an important component of quality of care for populations. In addition, we define efficiency as an attribute of quality of care for populations. Efficiency is to the ratio between benefit and costs. In essence, the most efficient use of care to maximise outcome, which can involve allocative efficiency (focusing on procedures which produce maximum benefit) and technical efficiency (employing procedures in the most technically competent manner). Definitions of efficiency which are output-specific concentrate upon maximising desired outcomes for individual users ( Ryan & Shakley, 1995 ). However, achieving maximum desired outcomes for all individual users may not be sustainable or affordable at the population level. We recognise the importance of cost–effectiveness when referring to health care for populations where choices have to be made between differing priorities for the allocation of scarce resources. At the population level, inputs are as important as outputs and resources should be distributed efficiently to where there is the most potential for health gain ( Blaxter, 1996 ). It is up to society to choose whether resources should be allocated to particular sub groups within a population ( Donabedian et al., 1982 ) based, for example, on decisions about need and equity ( Bevan, 1998 ).
There is a need to balance effectiveness and efficiency to gain the highest net benefit to individuals and society ( Donabedian, 1980 ). Efficiency and quality are not mutually exclusive — the challenge is to fuse economic and clinical incentives ( Brook, 1997 ). However, whether related to either individuals or populations, the ratio of cost to benefit (process or outcome benefit) will increase when care is ineffective. Ineffective care is inefficient ( Ovretveit, 1992 and Crosby, 1979 ).
Affordability at societal level is a sub-component of availability and relates for example to the rationing or prioritising of treatment. At population level the efficient use of care for the maximum desired outcomes (health status and user evaluation) within costs is paramount. This is, in essence, a ‘welfarist’ approach to health care with an emphasis upon the sum of (health) benefit received by each individual in society (Ryan & Shackley, 1995). However, efficient care must be provided on an equitable basis.
We suggest, therefore, that at the societal level there are four outcomes of effectiveness; user evaluation and health status and, in addition, cost and equity.
Practical application of the framework
In this section, we demonstrate the utility of the framework by applying it to two sets quality indicators, the NHS National Performance Framework in the UK, and the HEDIS set developed for the assessment of health plans in the United States by the National Committee for Quality Assurance ( NCQA, 1998 ). We show how the framework can be used to demonstrate what aspects of care are being measured and what are not, and we also show how it can be used as a framework to permit critical analysis of quality indicators.
The UK government has recently published a National Performance Assessment for the National Health Service (National Health Service Executive, 1999). This framework is intended to be applied at health authority level (populations 200,000–500,000). The HEDIS set of measures provides a common dataset to enable purchasers and consumers of health care to choose between health plans (NCQA, 1998). Annual reporting of HEDIS measures us required of health plans for accreditation by the NCQA. At present there are over 100 indicators, with many more being developed, particularly in the area of clinical effectiveness.
The indicators in the two sets are summarised in Table 1 . In some cases, the distinction between access and effectiveness is artificial. For example, in the UK indicators, teenage pregnancy rate is specifically advanced both as an indicator of availability of services and of the effectiveness of those services. The following critique is not intended to be comprehensive, but rather to demonstrate how a conceptual framework can be used critically to examine measures of quality of care.
Table 1.
HEDIS and the National Performance Framework compared
|
Quality of care dimensions for individual patients (from Fig. 3 ) |
National Performance Framework indicators proposed for use by the National Health Service Executive (1999) |
Performance indicators in HEDIS 3.0, 1999 dataset ( NCQA, 1999 ) |
|
Structure |
||
|
Geographic access |
None |
None (though could be included in patient survey) |
|
Physical access |
None |
None (though could be included in patient survey) |
|
Affordability |
None |
Cost/member/month |
|
Availability |
|
Rate of use of ambulatory health services |
|
|
% Adults registered with dentist |
No. of providers (primary care, mental health, obstetric and dentists) accepting new patients |
|
|
% Children registered with dentist |
|
|
|
Size of in-patient waiting list |
Co-ordination with education and social services |
|
|
Outpatient (within 13 weeks of referral) and inpatient (within 3 months of decision to admit) waiting times |
|
|
Effectiveness |
|
Qualifications of physicians |
|
|
|
No. of doctors who left in previous year |
|
|
|
Physicians’ incentives to overtreat or undertreat |
|
|
|
Financial stability of health plan |
|
Process |
||
|
Affordability |
None |
None |
|
Availability |
|
% Follow up after admission for mental health problem |
|
|
|
% Dental checks in children in the last year |
|
|
|
Availability of interpreters |
|
|
|
No. of well child/well adolescent visits |
|
|
Length of stay (fractured neck of femur, stroke) |
Length of stay (elective admissions, maternity care) |
|
|
Day surgery rate |
Elective surgery rate |
|
|
|
Admission rates (hospital, nursing home) |
|
|
|
Outpatient utilisation rate (primary care, emergency room, outpatient surgery, mental health, chemical dependency) |
|
Effectiveness of clinical care |
% Children immunised |
% Children and adolescents immunised |
|
|
|
% Smokers receiving advice |
|
|
|
%>65 receiving influenza vaccine |
|
|
% Mammography in target population |
% Mammography in target population |
|
|
% Cervical screening in target population |
% Cervical screening in target population |
|
|
|
% Pregnant women starting prenatal care by 13 weeks |
|
|
|
% Recommended prenatal and post-partum visits |
|
|
|
% Post MI patients receiving beta blocker |
|
|
|
% Diabetic patients with eye exam in previous year |
|
|
Elective surgery rate (e.g. appropriate=CABG, hip and knee replacement, cataract replacement, inappropriate=D&C, grommet insertion) |
|
|
|
Hospital readmission rates within 28 days (emergency, and mental health) |
Hospital readmission rates (mental health and chemical dependency) |
|
|
Avoidable admission rate (ENT and urinary infection, heart failure, epilepsy, diabetes, asthma) |
|
|
|
Rate of discharge to usual residence following stroke (within 56 days) and fractured neck of femur (within 28 days) |
|
|
|
Delayed discharge over age 75 |
|
|
|
Case-mix adjusted length of stay |
|
|
|
Emergency admission rate for >75 s |
|
|
|
Operation rate for recurrent hernia |
|
|
|
% Orchidopexies carried out below age 5 |
|
|
|
Psychotropic prescribing (volume of benzodiazepine prescribing) |
|
|
Effectiveness of interpersonal care |
None |
Included in patient satisfaction survey — see patient evaluation below |
|
Outcome |
||
|
Health status |
Under 16 conception rate |
% Elderly receiving functional status questionnaire |
|
|
Mortality (all cause age 15–64, 65–74, deaths from cancer, circulatory disease, suicide and accidents) |
|
|
|
30 Day mortality post myocardial infarction and post surgery |
|
|
|
Cancer registration |
|
|
|
5 Year survival rate (breast and cervical cancer) |
|
|
|
Infant mortality and stillbirth rate |
|
|
|
Decayed/missing/filled teeth in 5 year olds |
|
|
Notification rate for measles, pertussis and TB |
|
|
|
|
Episode rate for fractured femur |
|
|
User evaluation/experience |
Rate of operations cancelled for non-medical reasons |
Patient satisfaction survey |
|
|
Patients waiting >2 hours in A&E |
No. of patients who disenrolled in previous year |
|
|
DNA first outpatient attendance |
|
|
Costs of care |
Generic prescribing rate |
Drug prescription costs |
|
|
Unit cost (mental health) |
High cost DRGs |
|
|
|
Costs per member per month |
Access and availability
Inspection of Table 1 shows that neither set of indicators contains information on structural barriers to seeking care (e.g. physical or geographical access). This is potentially a serious omission, especially for some population groups such as the elderly or disabled. To some extent this is rectified (especially in HEDIS) by regarding utilisation as a measure of access: the UK set contains no information on primary care utilisation. However, as HEDIS is designed for the evaluation of care provided to those enrolled with health plans, it does not address a key equity issue in the US — a large uninsured population.
Effectiveness
The greatest differences between the two sets of indicators are seen in the areas of clinical effectiveness and outcome measurement. These are large and reflect major differences between the two health care systems, and in particular differences in the information systems in the two countries.
UK data on effectiveness relies almost entirely on population aggregated rates of procedures, e.g. admission rate for diabetes, volume of benzodiazepine prescribing. This makes the assumption that there are correct normative levels for these and (in these examples) ‘high is bad’. In the UK data, screening of children, women of cervical screening age and women of breast screening age are the only ones where an element of ‘need’ is built into the denominator. While data on utilisation rates are collected in the HEDIS set, it is much more specific about procedures having been carried out in specific population groups — e.g. smokers advised to stop, and percentage of patients receiving beta blockers post myocardial infarction. This type of indicator is also prominent among indicators being developed by NCQA, with the 1999 set (NCQA, 1999) containing cholesterol management after cardiovascular events, and details of comprehensive diabetes care. Differences in types of data available for quality indicators in the two countries relates largely to the system of billing in the US which has led to the development of detailed clinical databases and which are now being used to assess quality. The absence of such databases in the UK and the consequent reliance on information that is currently routinely available and which can only be collected at population level is a severe limitation of the UK indicators ( Shekelle & Roland, 1998 ).
Eight out of the ten current clinical effectiveness indicators in HEDIS 3.0 relate to preventive care, reflecting the prominence traditionally given to screening in US healthcare. However, this balance will be redressed with the inclusion of the types of new indicators outlined above: the HEDIS indicators are gradually being expanded into areas where health gain is most likely to result from a reduction in variability of performance. This emphasis is not seen in the UK indicators, in large part because of the limited nature of the data available for constructing indicators in the UK.
Health outcomes
However, while the HEDIS set appears replete with information on process measures of effectiveness, it is severely deficient in terms of clinical outcomes: these are limited to the percentage of elderly receiving an annual functional status questionnaire. In contrast, the proposed UK set contains detailed mortality data, and some morbidity data. This reflects two major differences between the health care systems. The NHS provides universal coverage and has a very strong public health focus. So, mortality data figures prominently in the UK set.
The focus of the HEDIS indicators is on health plan members, and not on the population. Since health plans can choose who they enrol, inclusion of health status and mortality might increase cream skimming by health plans, and could perversely lead to worse care for the population overall if health plans selectively discouraged enrolment by those who were likely to have poor health outcomes. To a limited extent, the HEDIS set tries to address the issue of cream-skimming, but only by recording the characteristics, including ethnic group, of enrolled Medicaid (welfare assisted) patients.
However, although health outcomes are required to assess the magnitude of a health problem in a population, they are not necessarily good performance indicators for the behaviour of individual health care professionals unless sophisticated case mix adjustment can be made. Health outcomes are often determined by population characteristics and behaviours outside the control of individual health professionals, e.g. socio-demographic population characteristics ( Giuffrida et al., 1999 ). They follow indirectly from individual health encounters, or may be separated from these by many years (e.g. as in the care of diabetic or hypertensive patients). They are therefore more suitable for use in the UK, where the government in concerned with the performance of the whole system, but less suitable in the US context, where year on year improvement is sought against published benchmarks.
Inter-personal care
One of the key important elements of quality of care which we have identified is the quality of inter-personal care and communication. The authors of both UK and US sets of indicators are gradually incorporating users’ experiences as a measure of the performance of the health care system. Direct assessment of interpersonal care is omitted entirely from the published NHS indicators, although the NHS started at the end of 1998 to conduct annual patient surveys, and it is planned that these will be reported at health authority level. Reports of patient experiences have recently become an significant part of HEDIS with the inclusion of a mandatory standard patient survey (CAHPS) in the most recent version of HEDIS (NCQA, 1999). The HEDIS set also includes information on patients who disenrolled in the previous year, which may in part relate to deficiencies in inter-personal aspects of care. Inclusion of this indicator reflects the greater ability of US patients to move between different types of health care: this choice is available in the UK for primary care (though rarely exercised), but scarcely exists for secondary care within the NHS.
Discussion
There is no universally accepted definition of care, quality or quality of care. In this paper we suggest that quality of care for individual patients is defined by their ability to access effective care with the aim of maximising health benefit in relation to need. This is not to say that other dimensions of quality identified previously in Fig. 2 are not important, but that their role in delivering quality of care for individuals is dependent upon users actually getting the care they need and it being effective. Both access and effectiveness, and the structure and processes of health care, are measured by their capacity to result in desired outcomes and/or processes as negotiated by user and health professional.
We have suggested that quality of care is a different concept when considered for individual users or whole populations. At the societal level quality of care is the ability to access effective care on an efficient and equitable basis for the optimisation of health benefit/well-being for the whole population. However, we regard quality of care to be a concept that has most meaning when related to individual users. The complexities and needs of individual patients or consultations are only discernible at the level of the individual and by focusing upon what is required for an individual in their interaction with a health care provider. People presenting to a health professional expect good individual care ( Stott et al., 1997 ) which they will evaluate in relation to how it meets their individual needs ( McGlynn, 1997 ). Population based measures of quality of care should, in part, be judged in terms of their ability to reflect the aggregation of individual encounters. However, this is not to say that people should necessarily get what they want — the health needs of individuals must be balanced against optimising health outcomes for society.
There is a discrepancy between defining the various aspects of quality of care and the methods currently available for collecting the data and/or the political will, by organisations and governments, to collect data using different methods. We have emphasised how different stakeholders value different components of quality of care. Moreover, our conceptual framework places inter-personal care equally with clinical care as a key process of quality of care. Importantly, the methods of measuring different elements of defined care are not standardised and different methods requires different methodologies for measuring quality of care ( Blumenthal, 1996 ). For example, methods for measuring patient-centredness and communication skills are more time consuming and expensive than computer searches of hospital clinical databases to determine admission rates. Strict reliance on explicitly measuring data quantitatively contained in administrative datasets can only measure those elements of quality of care which are consistently and reliably recorded therein. It is unlikely that data collected for administrative of financial purposes will meet the requirements of data to be used for quality assessment or quality assurance. Whether policy-makers give equal weight to other perspectives and methods is a political rather than conceptual issue. The potential expenditure on comprehensive quality assessment is very considerable, and choices will need to be made between spending money on quality assessment or on direct patient care. The assessment of quality on its own is not sufficient: it needs to be linked to effective (and cost effective) methods of improving quality.
Moreover, if quality indicators are developed as quantitative statistical measures (as with NPF and HEDIS) it is important to consider statistical and practical issues. These include the reliability and validity of the data collected, the feasibility (ease and cost) of collecting the data and differences in availability and comparability of data between and within countries.
We have not considered how the various components of quality of care might be weighted or compared with one another. We have defined quality of care as accessing effective care according to need. However, is a user, who receives good access but ineffective care, receiving worse or better care compared to a user who has had difficulties accessing effective care? Moreover, is a procedure which is rated as effective by a health professional, but rated poorly as an outcome by a user, good or bad care? Such questions are important for measuring, as opposed to defining, quality of care and whilst these highlight often potentially irreconcilable differences between different stakeholder perspectives, they will form the test of our ability to measure patient-centred care. Ultimately a national framework must assess quality of care from perspectives important to professionals, payers and patients ( McGlynn, 1997 ).
We have applied the framework to a widely used set of indicators in the US and a set of indicators proposed for use in the UK. Both the UK National Performance Framework and HEDIS attempt to move beyond simple activity measures to include indicators based on evidence and user evaluations. However, most of the indicators rely solely upon readily collectable data and on standardised population measures. Whilst this is perhaps inevitable, there is a risk that they will perversely reduce quality for other elements of care, e.g. those which have wide professional and public legitimacy (though limited evidence of effectiveness), and inter-personal aspects of care. There are dangers that by defining explicitly what is to be measured, this encourages health care organisations/professionals to “jump through the hoop”, safe in the knowledge that it is only performance in these aspects of care that will be measured. Moreover, inevitably, discussion of quality of care and cost often go hand-in-hand on both sides of the Atlantic. In the UK others have emphasised how quality of care forms part of a wider UK government strategy to manage public services ( Hoggett, 1996 and Traynor, 1996 ) driven by a desire to get “more for less” ( Hood, 1991 ). In the USA there is a growing movement to ensure that the new organisational structures for measuring quality do not sacrifice quality at the expense of costs ( Schuster et al., 1998 ). Importantly, quality of care is both an end in itself, but can also be used as a means of addressing other agenda (e.g. control of health care expenditure). Any discussion of quality of health care must be set within the context that health care does not operate in a political vacuum.
We believe that both sets of indicators are to be welcomed as important advances in measuring quality of care; in what is essentially a new and very rapidly expanding discipline. The HEDIS set is getting more comprehensive with every edition and the NHS set is the first of its kind in the UK. The challenge for the future is to focus on all the important aspects of quality of care (using the most valid methodology possible and whether data are readily available using existing systems at present or not) whilst trying to minimise perverse incentives. Both sets incorporate a range of elements of quality of care and, in so doing, also emphasise that quality of care can not, and should not, be measured by one global indicator.
Our models explicitly acknowledge the potential conflicts between focusing upon individuals or populations. Providing a framework which defines quality in relation to care allows some of the strengths and weaknesses of any set of quality indicators to be assessed. It exposes those areas of quality of care infrequently addressed, e.g. acute care, co-ordination between professional groups looking after an individual patient and inter-personal aspects of care, and also access for ‘hard to reach groups’. Enthusiasm for any individual set of indicators needs to be tempered with understanding about what care is not being measured (and subsequently potentially improved). The framework outlined in this paper offers a basis for such analysis.
Acknowledgements
The authors would like to thank Peter Bower, Toby Gosden and Ruth Young for helpful comments on earlier drafts of this paper.