Clinical Applications of Health-Related Quality-of-Life Research
A preponderant body of clinical research has demonstrated the validity, reliability and responsiveness of health-related quality-of-life (HRQOL) data endpoints. HRQOL information is able to accurately characterize patient status and, as a consequence, it is now beginning to be applied to a variety of important non-research clinical activities, such as patient monitoring, risk prediction and quality assessment. For there to be routine, widespread use of HRQOL measures in clinical applications, advances will be required in information technology infrastructure and biostatistical strategies to enable the interpretation of the large-scale data sets that continually accrue. Also crucial to this will be studies demonstrating that clinical decisions guided by HRQOL data can yield sufficiently improved medical and financial outcomes to justify the cost and effort of HRQOL systems. As real-time health status tracking over entire patient populations becomes routine, each patient encounter will be guided by—and in turn will help shape—dynamically derived definitions of clinical guidelines. This fusion of clinical research and practice will finally enable the healthcare industry to monitor, define and improve the quality of its services with the accuracy that other industries have achieved for years.
A recap of HRQOL research
Health-related quality-of-life (HRQOL) data capture self-reported information from patients about their perceptions of function, symptoms, mental state and interpersonal relations. HRQOL instruments (questionnaires statistically validated against other standard measures during development) deliver measures comparable in accuracy to other types of clinical test results. Disease-specific instruments reveal detailed patient characteristics that are relevant to longitudinal time-based monitoring of particular conditions. General health surveys canvas the overall state of patients across organ systems and diagnostic categories, and yield useful population-wide assessments (1). Utility measures extract an understanding of patient preferences by leading the patient through a series of theoretical choices about treatment alternatives and experiential outcomes. Properly performed, these instruments generate quantitative evidence of the impact of illness on patients’ lives, as well as the impact of therapeutic interventions. Operational and logistical constraints can impact the deployment of HRQOL instruments in challenging study settings, but various emerging information technologies promise to reduce the cost and complexity of large-scale HRQOL data collection. Though it is still a proportionally small niche within medical research, the number and scope of HRQOL investigations have recently rapidly increased, and this is shown by a burgeoning collection of publications in the medical literature.
From research to clinical applications
Transferring HRQOL measures from research settings—where they perform a descriptive function—to clinical care settings—where they can play a prescriptive function—is a logical but not necessarily simple proposition. This table outlines some of the salient differences between these two contexts:
|Objectives||Subject characterization||Patient/population characterization; Clinical course prediction; Therapeutic decision guidance|
|Current status of HRQOL use in this setting||The descriptive capabilities (validity, reliability, responsiveness) of HRQOL measures are well established||The value (clinical and financial outcomes) of HRQOL data fed back to clinicians has not yet been demonstrated in large trials; return on investment case not yet made|
|Context||Controlled clinical trial protocols||Uncontrolled, changing factors in routine patient care|
|Target population||Sampled/randomized/matched groups from the target population as appropriate to the study hypotheses and required power of the study||The total unselected population within the care system|
|HRQOL instruments used||Small, focused set determined by the trial protocol||All instruments that are relevant to each individual patient’s clinical condition|
|Data collection interval||Circumscribed per protocol||Open-ended, longitudinal|
|Required information infrastructure||Simple manual tools are adequate||Highly automated information technologies are needed to scale up to large populations and high-volume data|
|Analytic strategies||Regression and other standard biostatistical measures of association||Causal models, statistical control chart methodologies; others to be determined|
At a philosophical level, the fact that a measure is accurate does not necessarily mean that knowledge of that measure will make a significant difference to the outcome of a decision. For instance, a disease-specific instrument like the Kansas City Cardiomyopathy Questionnaire (KCCQ) can, with exquisite sensitivity, track the clinical course of heart failure patients (2). However, supplying attending physicians with knowledge of their patients’ KCCQ scores will not necessarily influence medication choices, dosing adjustments, decisions regarding hospitalization and, ultimately, the patients’ subsequent clinical course. While HRQOL data seem likely to impact the process of care delivery beneficially, evaluating this impact constitutes a critical new focus for health services research. The relative contribution of HRQOL data, compared with other types of clinical data, will require careful, comprehensive assessment.
A complex, interesting and important conceptual question about the real-time assessment of continually accumulating HRQOL data regards the statistical inference strategies that are appropriate to analyze these data. Standard statistical approaches—typified by regression techniques—predict future events by inferring associations from within a target population, assuming that the experimental conditions remain the same. However, these restrictive conditions do not occur in real-life clinical settings, as patients and patient populations experience continually changing circumstances caused by treatments, exposures and multiple other uncontrolled, interacting factors (3). Thus the statistical tools conventionally used in research contexts could prove inadequate for use with large-scale clinical HRQOL data. Approaches such as causal inference models (3,4), G-estimation algorithms (5,6) and statistical process control charts (7,8) appear to offer important advantages.
Furthermore, there are important practical differences in how clinicians will manage HRQOL data compared with researchers. In clinical research, investigators select from a population of interest-matched subsets, using a typically randomized sampling method, and then collect data from the groups for a defined period. Apart from a limited number of interim evaluations (such as for safety monitoring purposes), the data are stored until the study’s conclusion. Only at that point will statistical analysis, assessment of results and publication occur. On the other hand, in clinical practice, physicians will collect HRQOL data from their entire, unselected, patient populations on a longitudinal, open-ended basis. Data are saved directly into databases and will undergo immediate scrutiny, generating realtime assessments that are delivered back to the physician to factor into specific patient management decisions. Therefore, a metaphor for the change from using HRQOL data in the research setting to using them in the clinical care setting would be a change from credit card receipts tossed in a shoe-box to real-time, color Doppler radar.
There are considerable practical challenges implicit in this transition. To scale up to the large patient populations and high volumes of data handled by large provider organizations, efficient automated information technology systems are necessary. These systems must extend into the patients’ home and work settings to capture the HRQOL data at the most appropriate time and place. They must also provide busy clinicians with easy-to-use interfaces to integrated data (such as wireless handheld computer access to a complete electronic medical record) so that they can make quick and accurate patient decisions while maintaining efficient progress through their workday. The sophistication and efficiency of information systems and databases must be orders of magnitude greater than the fairly limited systems that typically suffice for research efforts.
Prototype clinical applications
Despite the challenge of transposing HRQOL measures into routine clinical practice, leaders in HRQOL research have long recognized that substantial benefits could result once these issues are resolved (9–11). This section reviews the clinical applications that become possible if a healthcare provider organization has systems in place to collect and analyze HRQOL data routinely and prospectively. As this figure shows, clinicians can apply HRQOL information about their patients before, during and after specific clinical encounters in order to accomplish different clinical management purposes.
Point-of-care decision support
The application of patient-specific HRQOL data at the point of care is the most obvious clinical use. Point and trend data—conditioned by the appropriate norm comparators— could assist physician–patient deliberations in the following types of scenarios:
Problem list prioritization
A general health survey might reveal new or worsening problems in a certain physical domain, leading the physician to focus his/her history taking and physical examination on these. Appropriate disease-specific HRQOL instruments could then be deployed to enable further monitoring to take place (12).
Medication selection and dosing
A disease-specific instrument’s score trends could reveal suboptimal symptom control, indicating a need for medication. A general survey might uncover the development of a new adverse effect, leading to the discontinuation of the current medication and the beginning of a new alternative.
Pharmacological versus procedural management
A disease-specific instrument might document deteriorating scores despite maximal medical therapy, thereby providing a compelling quantitative rationale to justify a more expensive and risky procedure. Spertus et al. have described such ‘appropriateness criteria’ for coronary angioplasty and have demonstrated that they can use these criteria to guide procedural case selection in order to optimize the delivered benefit (13).
Informed consent for procedures
Utility measures can help patients weigh up complex alternative concepts of benefit and risk when planning for potentially risky procedures (14,15).
Incorporation of patient preference into clinical guidelines
Treatment algorithms can be augmented at appropriate decision points by objective measures of patient preferences evaluated by utilities (16).
An automated system that routinely obtains HRQOL information from patients could generate a constant stream of quantitative intelligence that provider organizations would then scrutinize during resource planning, staffing decisions and patient scheduling.
Chronic disease management
Regular, frequent monitoring of patient status permits the earlier detection of clinical deterioration, allowing easier and more cost-effective interventions. Of equal importance is the fact that routine monitoring can also document stable and improved patient status, which provides objective validation of decisions not to intervene. (This can be particularly important in capitated payment situations where the provider bears the financial risk for treatment decisions and so must avoid expending unnecessary resources on patients, while at the same time being sure to provide the required level of care.) The scheduling of decisions about which patients to see and when to see them could therefore be driven by an objective assessment of need derived from HRQOL surveillance. Benefits of such HRQOL monitoring have been demonstrated in several clinical areas: cancer (17), angina (18,19) and geriatric health (20).
General population health status
The surveillance of healthy populations and, more particularly, at-risk populations, can enable the early detection of significant conditions at a time when cost-effective interventions are possible (21).
Predicting patients’ future clinical courses, use of resources and costs of care is extremely complex and fraught with uncertainty. Risk assessment typically requires the factoring in of co-morbid conditions, age and family history. Instead, HRQOL assessment could be more valuable than reviewing medical records when generating these assessments (22). Indeed, HRQOL data have been shown to predict in-house mortality risks (23). Risk assessments from HRQOL data can also be used to adjust outcomes measures statistically to account for case-mix differences among different patient populations. When comparing the clinical outcomes of patients cared for by individual providers (through different treatment regimens), as well as by entire provider organizations, such adjustments are critical to ensure that the data are comparable and that fair conclusions can be drawn (24,25).
It is increasingly important to assess and document the quality of provided healthcare services, although considerable controversy remains as how best to accomplish these tasks. HRQOL measures could augment and perhaps replace many of the ‘process’ measures now used in quality benchmarks. For instance, evaluations of the care provided by cardiologists could derive at least in part from their patients’ mean (and/or trend) Seattle Angina Questionnaire scores. Benchmarks—with appropriate risk-adjustments as discussed above—could be generated from large-scale HRQOL databases to yield institutional ‘report cards’ that document the quality of services provided. Consumers, employers and insurers could consider these quality reports when selecting specific physicians, as well as whole provider groups, during benefits-plan decisions. Risk-adjusted outcomes measures would provide more concrete information for setting appropriate capitation rates during managed-care contract negotiations between purchasers and providers (26). Conversely, the ability to document the quality of delivered care could significantly empower provider groups during contract negotiations with managed-care organizations, self-insured employers and government purchasers.
This figure illustrates a prototype ‘report card’ created by John Spertus at the Mid-America Heart Institute in Kansas City, MO, USA. This diagram encodes in one figure a wealth of data—how many patients have particular scores along two important SAQ scales and which physician is caring for them. This encoding is accomplished by the use of different icons for different physicians and different sized icons for different numbers of patients. The numbers of patients falling below target functional levels can be noted at a glance, as can the numbers of patients doing well. This particular illustration remains somewhat confusing and is undergoing continual refinement by Spertus, but it demonstrates the explanatory power that the routine collection of HRQOL data will soon be able to provide.
Definition of clinical guidelines
Ultimately, the universal routine monitoring of patient health status will be able to redefine how healthcare practice standards are set. When patient outcomes—measured by HRQOL data—can be correlated in realtime with all treatment decisions, providers could define ‘best practice’ guidelines directly from real data about their patients’ outcomes instead of attempting to extrapolate from published trials. Because patient characteristics (and also physician skills and available facilities) regularly fail to match those in controlled trial settings, systems enabling providers to determine what ‘works best’ within their local experience would prove more beneficial to patient outcomes.
As HRQOL measures gain traction in clinical settings, various organizational entities will emerge as primary advocates for, and consumers of, HRQOL data.
Disease management companies
Commercial groups that deliver tailor-made services for specific clinical problems need to generate return-on-investment evidence for their marketing messages. Patient outcomes measured through HRQOL instruments could provide the elusive effectiveness portion of their cost-effectiveness claims. Successful disease management companies with large patient populations under their care could also generate important normative data for quality comparisons.
The utilization-review and quality-measurement industry has focused on various process measures (derived largely from claims data), primarily because these data have been readily available, not because they were the most informative. Innovative quality-assessment consulting companies could provide outsourced services for collecting, managing and interpreting HRQOL data for provider organizations who lack the information technology expertise themselves.
Pharmaceutical and device manufacturers
Post-market Phase IIIB/IV surveillance studies are important for detecting late-appearing adverse effects that guide dosing recommendation refinement, investigating opportunities for new indications and generating evidence for competitive marketing claims against rival products. HRQOL evidence typically plays a large role in these investigations. Once such data are routinely generated during patient care, the speed, scope and cost-effectiveness of these studies will greatly improve.
Regulatory and watchdog groups
When evaluating the skills and capabilities of providers, accreditation and licensure decisions could eventually incorporate summary data about the functional status of patients under the care of those providers.
Several decades of HRQOL research have produced tools that generate high-fidelity data about patient health status. As information technology solutions arise that make the widespread deployment of HRQOL measures feasible during the provision of routine care, clinicians will be able to apply HRQOL data to important management questions. Ongoing research will quantify the impact that HRQOL-guided decisions have on clinical and financial outcomes and will validate the return on investment that HRQOL data systems yield. Further development of these clinical HRQOL applications will derive from the efforts of important stakeholder organizations within the healthcare industry. Ultimately, HRQOL measures have the power to play a crucial role in the creation of real-time quality assessment systems that simultaneously monitor and guide the standards of healthcare service delivery.
- Guyatt GH. A taxonomy of health status instruments. J Rheumatol 1995;22:1188–90.
- Green CP, Porter CB, Bresnahan DR et al. Development and evaluation of the Kansas City Cardiomyopathy Questionnaire: A new health status measure for heart failure. J Am Coll Cardiol 2000;35:1245–55.
- Pearl J. Causal inference in the health sciences: A conceptual introduction. Los Angeles (CA): University of California, Los Angeles; 2001 Feb. Report No.: R-282. Available from: URL:ftp://ftp.cs.ucla.edu/pub/stat_ser/R282.pdf
- Pearl J. Causality: Models, reasoning and inference. Cambridge (UK): Cambridge University Press, 2000.
- Robins JM, Greenland S. Identifiability and exchangeability for direct and indirect effects. Epidemiology 1992;3:143–55.
- Witteman JC, D’Agostino RB, Stijnen T et al. G-estimation of causal effects: Isolated systolic hypertension and cardiovascular death in the Framingham Heart Study. Am J Epidemiol 1998;148:390–401.
- Carey RG. Measuring health care quality. How do you know your care has improved? Eval Health Prof 2000;23:43–57.
- Amin SG. Control charts 101: A guide to health care applications. Qual Manag Health Care 2001;9:1–27.
- Bergner M, Barry MJ, Bowman MA et al. Where do we go from here? Opportunities for applying health status assessment measures in clinical settings. Med Care 1992;30(Suppl.):MS219–30.
- Deyo RA, Patrick DL. Barriers to the use of health status measures in clinical investigation, patient care, and policy research. Med Care 1989;27(Suppl.):S254–68.
- Deyo RA, Carter WB. Strategies for improving and expanding the application of health status measures in clinical settings. A researcher–developer viewpoint (see discussion). Med Care 1992;30(Suppl.):MS176–86. Discussion in: Med Care 1992;30(Suppl.):MS196–209.
- Rubenstein LV, McCoy JM, Cope DW et al. Improving patient quality of life with feedback to physicians about functional status. J Gen Intern Med 1995;10:607–14.
- Tripuraneni R, Spertus J, Johnson W et al. The MAHI appropriateness criteria: A new method for quantifying the indications for PCI. J Am Coll Cardiol 2001;37(Suppl. A):506A.
- Zug KA, Littenberg B, Baughman RD et al. Assessing the preferences of patients with psoriasis. A quantitative, utility approach. Arch Dermatol 1995;131:561–8.
- Hazen GB, Hopp WJ, Pellissier JM. Continuous-risk utility assessment in medical decision making. Med Decis Making 1991;11:294–304.
- Nease RF Jr, Owens DK. A method for estimating the cost-effectiveness of incorporating patient preferences into practice guidelines. Med Decis Making 1994;14:382–92.
- Christ G, Siegel K. Monitoring quality-of-life needs of cancer patients. Cancer 1990;65(Suppl.):760–5.
- Spertus JA, Bliven BD, Farner M et al. Integrating baseline health status data collection into the process of care. Jt Comm J Qual Improv 2001;27:369–80.
- Spertus J, Dewhurst T, Dougherty C et al. Benefits of an ‘angina clinic’ for patients with coronary artery disease: A demonstration of health status measures as markers of health care quality. Am Heart J 2001 (in press).
- DeVore PA. Ability of a computerized geriatric assessment to predict need for change in living status among elderly living at home. South Med J 1994;87:743–8.
- Hennessy CH, Moriarty DG, Zack MM et al. Measuring health-related quality of life for public health surveillance. Public Health Rep 1994;109:665–72.
- Katz JN, Chang LC, Sangha O et al. Can comorbidity be measured by questionnaire rather than medical record review? Med Care 1996;34:73–84.
- Davis RB, Iezzoni LI, Phillips RS et al. Predicting in-hospital mortality. The importance of functional status information. Med Care 1995;33:906–21.
- Parkerson GR Jr, Broadhead WE, Tse CK. Health status and severity of illness as predictors of outcomes in primary care. Med Care 1995;33:53–66.
- Fowles JB, Weiner JP, Knutson D et al. Taking health status into account when setting capitation rates: A comparison of risk-adjustment methods. JAMA 1996;276:1316–21.
- Petryshen P, Pallas LL, Shamian J. Outcomes monitoring: Adjusting for risk factors, severity of illness, and complexity of care. J Am Med Inform Assoc 1995;2:243–9.
Created: February 06, 2009 15:49; Last updated: February 06, 2009 16:11