Background
Despite the near-consensus that volume-to-value approaches will be explicit to the reimbursement models that will dominate the healthcare system now and into the future, the empirical transformation of this concept is wrought with complexities. The existence of an underlying tension and competition of interests between providers and patients suggests the need for a statistical tool that will allow the empirical verification of whether patients, insurers or the society as a whole are the beneficiary of the proposed value outcomes.1 As one author states, “… physicians and other health providers respond rationally to existing financial incentives (translation: they do what they get paid to do and generally try to, or have to, minimize those activities and services for which they are not paid).” 2 However, strong proponents of the volume-to-value model make a highly relevant point that has been underdiscussed by analysts during the translation of this conceptual model from theory to healthcare system praxis.3 4 This question is, ‘how can clinicians know with some degree of precision when volume trade-offs are decreasing patient outcomes?’
Consistent with this theme, the entirety of the January 2017 issue of The Lancet implicitly embraces the issue of the measurement of healthcare system overuse and underuse by offering strategies for the prevention of either extreme.5 However, while the centrality of quality measurement and evaluation approaches has been alluded to in the volume-to-value literature and the use of big data recommended as a necessary tool, a paucity of research exists regarding statistical methodologies that may be useful in supporting clinicians and/or their financial managers in ensuring that processes of care services are managed in such a way as to deliver balance within the emerging volume-to-value environment.6
This study introduces a statistical methodology that will allow physicians, their financial managers and/or others to optimise both volume of care and intermediate outcomes. As a result, the risks of financial loss that the volume-to-value paradigm poses for clinicians who treat persons with chronic health conditions can be reduced. Real-world data on providers who treat patients with diabetes are used to demonstrate this approach.
Measuring the relationship between volume of diabetes care and quality: current approaches
The level and volume of care delivered by clinicians plays a major role in the control of diabetes and resultantly in the patient’s outcomes as measured by the prevalence and incidence of adverse diabetes-related medical outcomes.7 Numerous national and local efforts have been launched that seek to improve the care of patients with diabetes.8 Donabedian’s paradigm is often used as the conceptual framework to determine the effectiveness of such interventions.9 10 This paradigm, a model which serves as a dominant one in the area of healthcare management and administration, suggests that a causal pathway exists in which structural characteristics of the environment such as use of the chronic care model affect processes of care such as diabetes-related screening, which in turn affect patient outcomes.11
The approach adopted by most studies that use Donabedian’s paradigm within the context of diabetes is one which uses structural change as the input, while treating both processes and outcomes as outputs. Donabedian’s model implies that structure entails process which, in turn, generates outcomes. Thus, the assumption is that it necessarily follows that structure is also related to outcomes. Moreover, many studies, based on this approach, report significant direct relationships between structures, processes and outcomes.12–18
As one analyses this approach, it is clear that this model also implies that the relationship between structure and outcome may be weaker than originally assumed. Thus, it is less than surprising that another group of studies report significant relationships between changes in structure and changes in care. Yet, empirical research does not confirm significant relationships between changes in structure and improved patient outcomes.19–29 This suggests that a structural shift from a volume-oriented clinical delivery system may not automatically generate adverse quality changes if decreases occur in the volume of processes of care as value-based reimbursement approaches become dominant. Accordingly, this study focuses on the direct relationship between changes in the volume of diabetes processes of care and quality changes as measured by diabetes intermediate outcomes. Few studies have addressed such a relationship and even fewer have corroborated the relationship.
Indeed, a review of the literature revealed no studies that have sought to export findings from a chronicity such as diabetes care into a framework that examines the implications of the findings for the volume-to-value paradigm. Moreover, only a small body of research considers both processes of care and intermediate outcomes as predictors of diabetes long-term value measures such as the rates of death, blindness, cardiovascular disease, amputation and patient’s overall physical health per dollar spent. For example, Kahn found that a change in a health-related quality-of-life score was significantly associated with the process of care composite (eg, exams, lab tests, diagnostic procedures).30 This finding therefore alluded to a relationship between diabetes care and diabetes long-term outcomes. However, the inclusion of diabetes with other chronic diseases, non-diabetic medications and counselling in the definition of care introduces ambiguity into the results.
Rather than using a combination of care and outcomes variable, Harman used separate patient care and intermediate outcome variables to predict the individual Health Outcome Survey physical and mental health scores of Medicare plan enrollees.31 The intermediate outcome measure was associated with changes in both physical and mental health while the process of care measure was only associated with a change in mental health.31 This study also suggests a relationship between volume of care and quality as measured by outcomes. However, the non-significant association between care and physical health weakened this connection.
The medical model of health, which is defined as an absence of disease and/or illness, implicitly hypothesises that intermediate outcomes literally mediate rather than moderate the relationship between process of care and outcomes. However, these studies may suffer from the proximity effect rather than providing an empirically confirmed relationship between volume of care and patient outcomes. Yet, very few studies have tested for a direct relationship between the volume of diabetes process of care variables and diabetes intermediate outcomes (eg, A1C, blood pressure and low-density lipoprotein (LDL) levels). Kirk et al. 32 used a bivariate process of care composite variable equalling ‘yes’ if the patients had received two A1C tests, one blood pressure reading and one LDL measurement in the past year, and ‘no’ if one or more processes of care could not be confirmed. This study makes one of the strongest cases for a volume-to-value transition because it resulted in a non-significant relationship between process of care and the bivariate A1C outcome measures, thereby suggesting that the volume of the processes of care can be decreased without impacting quality.32 Other research also found no significant correlations and suggested that cost savings can be achieved in diabetes care through volume decreases without an impact on quality.21
Indeed, only one single study that modelled the relationship between a diabetes process of care variable and intermediate outcomes found a significant relationship.33 Accordingly, this study suggests that, with an appropriately specified, easy-to-use statistical tool, providers or their financial managers can accurately define volume of care decreases that can be made without quality of care penalties. Such a tool can simultaneously reduce financial risk to providers and health risks to consumers within a volume-to-value healthcare environment.
The purpose of this study was to determine how changes in the ‘volume’ of processes of care services, that is, asking about tobacco use, developing self-management goals and timely testing of feet, eyes, A1C, LDL and nephropathy, affect intermediate outcomes such as controlled blood pressure, A1C and LDL at the practice level in a diabetes patient population. It also explores the implications of the findings for the volume-to-value transition process. Specifically, the study sought to answer the two following questions:
Do practices with a higher volume of processes of care, on average, have ‘better’ patient clinical outcomes, on average, using a chronicity such as diabetes as the case study?
Do practices exceeding or dropping below their average processes of care volume during a particular month also exceed or drop below their average ‘quality’ as measured by patient clinical outcomes in a month?