Click for next page ( 153


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 152
1 ~ Manufacturers' Responses to the Increased Demand for Outcomes Research ANN K. M. MARSHALL The title of this chapter would imply that health care manufacturers face a well-defined demand for outcomes research. This is not so. Rather, they con- front an evolving and heterogeneous set of market dynamics that may collective- ly be referred to as the technology assessment trend. Simply stated, the technolo- gy assessment trend emphasizes the importance of considering comparative effectiveness and relative value in evaluating medical technologies. But the application of this concept varies considerably among providers, payers, and regulators and over time-creating uncertainty on the part of manufacturers as to precisely what sorts of information will be required as the technology assessment trend matures. Manufacturers' activities in the area of outcomes research reflect the nascent stage of this trend. To gain perspective into manufacturers' outcomes research activities, it is important to highlight the factors that have driven the emphasis on technology assessment and their connection with outcomes research. First, however, it is useful to provide working definitions of medical technology assessment and out- comes research. TECHNOLOGY ASSESSMENT AND OUTCOMES RESEARCH Medical technology assessment can be defined as the careful evaluation of a medical technology for evidence of the health, economic, social, and ethical consequences of its technical applications, both in absolute terms and in compari- son with other competing technologies (Office of Technology Assessment, U.S. Congress, 1982; Perry, 1988~. As discussed elsewhere in this volume, real-life 152

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 153 Assessment Activities Evaluative Research 1 . ~ ~ Literature Review Prudent Purchaser Analysis ~ , ~ Meta-Analysis Expert Consensus Panels Prospective Studies Retrospective Analysis Economic Modeling Decision Analytic Modeling FIGURE 12-1 The spectrum of technology assessment. OUTCOMES RESEARCH technology assessments are less comprehensive Wan the ideal suggested by the above definition. Moreover, the focus of a technology assessment depends on the technology assessed (e.g., drug, device, or procedure) and the type of organi- zation conducting the assessment (e.g., hospital, managed care organization, third-palty payer, government agency, academe, or a manufacturer). The spec- trum of technology assessment ranges from more or less informed technology appraisals to controlled clinical studies, as depicted in Figure 12-1. Outcomes research can be viewed as a subset of technology assessment activities. It encompasses prospective clinical studies, retrospective analyses of large health databases, and economic and decision-analytic modeling. In addi- tion to relevant areas of clinical expertise, outcomes research draws upon the fields of epidemiology, economics, and health services research. Outcomes re- search can be roughly divided into two categories: economic outcomes research (cost-benefit, cost-effectiveness, and cost-utility analyses) and patient outcomes research (measurement of treatment outcomes such as functional status, quality of life, and long-term survival). Patient outcomes research extends beyond the safety and efficacy studies traditionally conducted by pharmaceutical companies. First, whereas traditional efficacy research often studies intermediate or surro- gate endpoints, patient outcomes research focuses on "outcomes of real interest to patients," such as death and disability (Eddy, 19901. Second, patient outcomes studies generally incorporate comparisons of competing treatments and may even compare different treatment modalities (e.g., drug versus procedure).] 1 Strictly speaking, the traditional safety and efficacy studies conducted by pharmaceutical compa- nies to support New Drug Applications filed with the Food and Drug Administration are also out- comes research. However, the term outcomes research is often used to refer to the study of outcomes other than classical safety and efficacy outcomes and will be so used in this chapter.

OCR for page 152
154 ANN K. M. MARSHALL Although health care providers and third-party payers use outcomes research findings in their technology assessments, they rarely conduct outcomes research themselves. The Agency for Health Care Policy and Research (AHCPR) devel- ops rigorously systematic syntheses of the literature and expert opinion, but it produces little in the way of outcomes research, per se. In the future, more AHCPR and third-party payer activities in meta-analysis, decision-analytic mod- eling, and retrospective health database analysis can be expected. For the most part, however, experimental research on patient and economic outcomes will continue to be conducted by manufacturers arid academic or pr~vate-sector re- searchers, the latter frequently being at least partly funded by manufacturers. THE TECHNOLOGY ASSESSMENT TREND: WHY NOW? Although "technology assessment" and "outcomes research" have only re- cently become household phrases, the origins of the technology assessment trend may be traced back almost 20 years. In the early 1970s, John Wennberg and others documented significant geographic differences in the rates at which cer- tain medical procedures were being performed that could not be explained by differences in the incidence or severity of disease among the populations being compared (Wennberg and Gittelsohn, 1973; Wennberg et al., 1987,19891. These differences naturally raised questions about appropriateness of care and optimal treatment pattems, questions that triggered a surge of interest in studying patient outcomes. The challenges of patient outcomes research lead clinical researchers to ex- pand their scope of clinical methodologies and techniques. For example, the limitations of randomized, controlled clinical trials (RCTs) in assessing the rela- tive effectiveness of treatment options in clinical practice encouraged researchers to turn to epidemiology a field in which researchers have expertise in such areas as observational study design, analysis of large databases, and knowledge of the natural history of diseases. Within the field of health status measurement, new instruments were developed to measure disease-specific parameters. In ad- dition, more responsive instruments were developed to measure more discrete, but clinically relevant, changes in health status and quality of life.2 These meth- odological advances helped fuel what Wennberg has called the "outcomes agen- da," namely, the "systematic evaluation of all of the outcomes that are relevant to patients- mortality, morbidity, complications, symptom reduction, and function- al status improvement" (Wennberg, 1990, p. 45~. 2 See Medical Innovation at the Crossroads. Vol. 1, Modern Methods of Clinical Investigation (Institute of Medicine, 1990) for an overview of the techniques and methods that have been employed in outcomes research, including observational methodologies, meta-analysis, decision analysis, and health status measurement. See Freund and Dittus (1992) for a concise summary of the techniques and methods specific to economic outcomes research.

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 155 Despite its potential to significantly advance the practice of medicine, pa- tient outcomes research probably would have remained in the domain of academ- ic researchers were it not for the intractable escalation of health care costs. As the 1980s progressed and the oft-recited waves of belt-tightening and cost-shift- ing rippled through the U.S. health care system, policymakers noted the potential of outcomes research concepts and methodologies to help rationalize health care expenditures. Patient outcomes research held out the hope that, by eliminating inappropriate and suboptimal health care, costs could be controlled without di- minishing quality of or access to care. Faith in the promise of health care ration- alization was the impetus behind the U.S. Congress' creation and aggressive funding of AHCPR in 1989-1990. The emphasis on economic outcomes research has been fueled by public- sector "prudent purchaser" initiatives and the growth of managed care. Third- party payers, health maintenance organizations (HMOs), and formulary commit- tees are now important decisionmakers in the process of technology adoption and diffusion. Unlike physicians, these new decisionmakers have a direct interest in cost containment, and they have the consolidated market clout to demand evi- dence of the comparative effectiveness and relative value of new drugs, devices, and procedures. Hospital, HMO, and Medicaid formulary committees are in- creasingly considering outcomes research particularly economic outcomes re- search in their drug formulary decisions. Third-party payers, both public and private, now often require evidence of the comparative effectiveness of new therapeutic or diagnostic procedures before granting coverage or adjusting reim- bursement levels for those procedures. These market trends are creating new informational requirements for pharmaceutical companies and high-tech, proce- dure-related device companies. MANUFACTURERS' RESPONSE TO THE TECHNOLOGY ASSESSMENT TRENDS Device Companies The U.S. medical device industry comprises roughly 7,000 manufacturers whose varied outputs include mundane commodity products such as tongue de- pressors and surgical masks, expensive capital equipment such as gammaradia- tion sterilizers and magnetic resonance imaging scanners, high-technology single- use products such as percutaneous transluminal coronary angioplasty catheters 3 This chapter utilizes information obtained through 23 interviews with representatives of 15 man- ufacturers and 4 consulting or research organizations. To ensure the anonymity of interviewees and their companies, only publicly disclosed examples of manufacturer outcomes research activities are cited specifically. The comments and opinions expressed in this chapter are the author's and do not represent the views or policies of Abbott Laboratories.

OCR for page 152
156 ANN K. M. MARSHALL and biodegradable sutures, and equipment ranging from ventilators to endoscopic staplers. It should be acknowledged from the start that only a small percentage of these companies view outcomes research as critical to the commercial success of their products. For the most part, only manufacturers of high-cost and innovative technologies that have a direct impact on patient therapeutic or diagnostic out- comes are likely to consider conducting outcomes research. Historically, device companies' introduction to outcomes research arose out of their struggles with coverage and reimbursement in the early to mid-1980s. At that time, a number of essentially new, medically complex devices and technolo- gies had been approved by the Food and Drug Administration (FDA) but were not immediately granted coverage by the Health Care Financing Administration (HCFA), the Blue Cross and Blue Shield carriers (the Blues), or private insurers. Instead, third-party payers conducted their own technology assessments of such FDA-approved devices before making coverage decisions, often consuming two or three years in the process. (Examples of devices subjected to additional third- party payer scrutiny include the implantable infusion pump, cardiac defibrillator, magnetic resonance imaging scanners, extracorporeal shock wave lithotripter, and cochlear implant.) The delay of third-party payer coverage created an ex- tremely costly post-FDA hurdle that had to be crossed before a manufacturer could accomplish full commercialization.4 As mentioned above, payers' newly activist stance in questioning costly new technologies was prompted by the cost-containment imperative. When a new medical technology threatened to significantly increase expenditures, payers wanted to know whether and to what extent it improved the quality of patient care in comparison with existing treatment options. Payers were obliged, howev- er, to pose their questions in terms of the regulatory and contractual language that governed their coverage obligations: Was the new device or procedure "reason- able and necessary" (Medicare) or "medically standard and acceptable"/"not in- vestigational" (private payers and the Blues). The underlying question, however, was one of value: Will the additional cost of this new technology yield improved outcomes (or outcomes at least equivalent to those from existing technologies) for the beneficiary? The fact that payers answered these questions by conducting their own tech- nology assessments sent two clear messages to the involved device manufactur- ers. First, FDA's process of approval for devices did not leave payers sufficient- ly comfortable about the effectiveness of novel devices. Second, even granting the effectiveness issue, the fact that FDA approval could be obtained without presenting comparative data left open the question of whether a new technology 4 See Bucci et al. (1985), Kane and Manoukian (1989), McGivney (1991), and VanAntwerp (1985) for background on the impact of technology assessment activities of third-party payers on the adop- tion and diffusion of new device and procedure technologies. Bucci et al., Kane and Manoukian, and McGivney also discuss the related but additional impact of inadequate reimbursement.

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 157 was worth the (usually) higher cost. These concerns were evident in the negative recommendations of the technology assessments conducted by the Office of Health Technology Assessment (OHTA), as commissioned by the HCFA, during this pivotal period of the mid-1980s. OHTA cited the lack of well-controlled clinical trials demonstrating safety and effectiveness as the basis for its negative recommendations although it is worth noting that few of the devices receiving positive OHTA recommendations had been supported by well-controlled clinical trials showing safety and effectiveness (Bucci et al., 1985~. Nonetheless, by the mid-1980s manufacturers began to realize that they could significantly reduce the risk of coverage delay or denial if they produced well-controlled clinical trials that showed the comparative effectiveness of their devices. It should be mentioned that, since the late 1980s, trends in the FDA device approval process have been consistent with the technology assessment trend. Whereas previously the FDA tended to focus on whether a device was safe and performed as intended, by the late 1980s reviewers also emphasized that "the 'clinical utility' of all devices undergoing review in a . . . Epre-marketing approv- al application] must be established prior to approval" (Office of Drug Evaluation, Food and Drug Administration, 1991~. For example, the issue of clinical utility was central to the FDA's unwillingness to approve Healthdyne's home uterine monitor for use in preterm labor. Although the FDA was satisfied that the device accurately monitored the intensity and timing of contractions, it required evi- dence that such monitoring made a difference in the outcomes of deliveries. More broadly, device manufacturers noted that the FDA had come to require more rigorous biostatistical analyses to support effectiveness claims in premar- keting approval applications, and by the end of the 1980s often required control arms and even comparative analysis in cases in which it would not have done so in the past. This new rigor is at the root of the protracted difficulties experienced by the companies Dornier and Medstone in trying to show the clinical effective- ness of their biliary lithotripter devices (Stern, 19901. In this environment, device companies have become much more sensitized to the need to conduct patient outcomes research. However, the decision to conduct such research is very much product specific and depends on affirmative answers to such questions as: Is this an essentially new product (as opposed to a "follow on") that will increase expenditures by the health care system? Is the reimbursement structure for procedures in which this product is used likely to pose difficulties for users of the product? Are payers likely to try to restrict the use of this product to a limited set of indications? Because a limited number of new medical devices face these issues, the pool of device manufacturers who conduct outcomes research is rather small and quite sophisticated compared with the broader population of device manufacturers. These "high-tech" device manufacturers have tended to sponsor prospective con- trolled clinical trials, typically focusing on patient outcomes. The types of end- points that have been studied vary by technology, but include survival, disease,

OCR for page 152
158 ANN K. M. MARSHALL functional status, and quality-of-life measures. Some medical devices are partic- ularly well suited to quality-of-life studies (e.g., artificial hips and-incontinence devices), and a fair amount of work has been done in that area. In a number of cases, device manufacturers have undertaken considerable work in support of their outcomes research capabilities. For example, they have developed databas- es (e.g., using coding and cost data), they have contracted with outside consult- ants to do meta-analyses of device-relevant therapeutic areas, and they have de- veloped or validated instruments to measure quality of life in patient outcomes research. Most high-tech device manufacturers employ a professional dedicated to outcomes research activities, usually housed in the "medical" area. Device com- panies almost always commission outside researchers to run the clinical trials associated with their outcomes research, largely because they do not have the extensive clinical research infrastructures found in pharmaceutical companies. But device companies also emphasize research credibility as an important reason for commissioning independent researchers. On the whole, device companies supporting prospective outcomes research seem to be extremely concerned that the research be conducted to the highest standards, indicating, for example, "We want our clinicals to be run like the highest caliber pharmaceutical trials" or "We follow National Institute of Health standards for clinical trials in contracting with outside researchers." This strong emphasis on credibility makes sense when one recalls that, historically, the need to do outcomes research arose out of payers' skepticism about the effectiveness of devices and procedures. Most device manufacturers engaged in outcomes research are very clear about why they are doing the research and who their audience is: They do outcomes research to secure coverage and reimbursement for the procedures in which their devices are used. Their primary audience is the HCFA, the Blues, and the large commercial payers. Because third party payers have historically questioned comparative effectiveness, device manufacturers have tended to focus on patient outcomes. However, this is beginning to change. Many now include an economic component in their studies, although patient outcomes still domi- nate the focus of study designs. Retrospective analyses do not play a large role, because studies are typically conducted prior to widespread coverage. It is worth reemphasizing that the above remarks refer to a rather small subset of device manufacturers that have developed a fairly sophisticated ap- proach to outcomes research. Most device manufacturers do not conduct multi- arm clinical trials, develop economic models, or employ rigorous syntheses of observational and experimental data. This is not to deny the increasing efforts of many device companies to develop cost data pertaining to the use of their prod- ucts. But these data are typically used to develop accounting-oriented cost analy- ses, as opposed to cost-benefit or cost-effectiveness analyses based on patient outcomes. It is important to note that accounting-oriented cost analyses can be quite useful, particularly for direct purchasers of devices such as hospitals. As a

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 159 matter of fact, such cost analyses are better suited to the ways that most hospitals actually make their financial decisions than are true economic outcomes analy- ses-which is one very good reason why a broad array of device manufacturers produce them. Pharmaceutical Companies Pharmaceutical firms did not suffer from the sorts of coverage difficulties experienced by device manufacturers and described above. Throughout the early 1980s, the HCFA and virtually all other third-party payers automatically covered FDA-approved drugs (to the extent that they covered drugs at all). Nevertheless, a few pharmaceutical firms recognized the importance of the burgeoning technol- ogy assessment trend. Those firms supported the development of in-house ex- pertise in the evolving methods of clinical investigation. Equally important, they perceived the receptivity of the market toward outcomes information and under- stood the marketing opportunity inherent in that receptivity (Jack, 1991~. these early efforts In outcomes research began to near fruit In the mid- 1980s. The two classic studies of that period, on auranofin and captopril, were both published in 1986 (Bombardier et al., 1986; Croog et al., 1986~. The study of auranofin (for adult rheumatoid arthritis) used well-developed, multidimen- sional health status measures that provided a consolidated health status score for each patient. The study of captopril (for essential hypertension) employed nine unrelated functional status and quality of life instruments, some of which were newly developed for the study. Despite their methodological differences, both studies used nontraditional methods of clinical investigation to measure clinical endpoints that were too subjective to be measured by traditional approaches. In addition, both companies undertook these studies with the physician audience in mind, to show the impact of their drugs on dimensions of life that are meaningful and important to patients. This contrasts sharply to the payer orientation of . device companies. However, most pharmaceutical firms did not begin to think seriously about outcomes research until the 1987-1989 time frame. By then the increasingly restrictive practices of hospital, HMO, and Medicaid formulary committees had forced pharmaceutical companies to address directly the issue of value with their customers. In the U.S. Congress, the pharmaceutical industry faced escalating criticism of its prices. In addition, the HCFA and some managed care organiza- tions were beginning to voice an interest in new concepts like therapeutic substi- tution and drug utilization review. It became increasingly evident that key con- stituencies were narrowly focusing on drug prices. To show the value of their products, pharmaceutical companies had to move the discussion to the topic of cost-effectiveness and then provide credible evidence of that cost-effective- ness. Outcomes research, particularly economic outcomes research, was well- suited to this task. ~. . .

OCR for page 152
160 ANN K. M. MARSHALL In most pharmaceutical firms these market dynamics were initially observed in public policy or managed care marketing functions, and those functional areas often took the lead in espousing the importance of outcomes research. Typically, such early efforts met with widespread skepticism. The business side expressed doubts. After all, outcomes research is costly and one can legitimately question its cost-effectiveness. The clinical research and development (R&D) areas also registered resistance. This resistance was partly due to a lack of confidence in (or familiarity with) the clinical methodologies used in outcomes research, but it also arose from the perception that outcomes research draws resources away from conducting the studies required to support New Drug Application filings. Thus, most pharmaceutical firms went through a period of "conversion" before com- mitting to the importance of outcomes research. Relatively few pharmaceutical firms established departments, or even individual positions, fully dedicated to outcomes research until the l990s. The companies that did have early involve- ment of their R&D and marketing divisions in outcomes research have more advanced outcomes research capabilities today. With few exceptions, pharmaceutical companies are still grappling with the central questions of how to structure their outcomes research activities. Exper- tise is situated in such widely disparate organizational locations as marketing, public policy, corporate planning, R&D, and medical affairs. In broad terms, there appears to be a general trend to locate the primary outcomes research func- tion in clinical R&D, with communication linkages to the marketing and new product development divisions. That said, however, few companies have satis- factorily integrated outcomes research into their clinical development processes. In most companies, the outcomes research function is still quite protean and will continue to evolve over the coming years (Freeman, 1991; Steward, l991a,b, 1992; The Zitter Group and Technology Assessment Group, 1992~. Although the very early outcomes research efforts of pharmaceutical compa- nies focused on patient outcomes, recent emphasis has shifted to economic out- comes. In fact, cost-e~ectiveness seems to be the dominant buzzword in pharma- ceutical industry conversations about outcomes research and technology assessment. A number of companies report that patient outcomes (e.g., function- al status and quality of life) are studied in the course of conducting economic studies, but are rarely the sole subject of a study. Even those companies that follow a more balanced approach to outcomes research acknowledge a bias to- ward economic outcomes. This is not surprising when one recalls that aggressive cost consciousness has driven most of the market dynamics that lead the majority of pharmaceutical firms to do outcomes research in the first place. Nonetheless, the more balanced approach will likely prevail over time, as indicated in a per- sonal communication from the director of the outcomes research function at a large pharmaceutical firm: Given the current market trends, there is increasing emphasis on economic stud ies. You don't make the market; the market makes you. But, ultimately, good

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 161. medicine is good economics. In the end, if you can show patient benefit, the economics will follow. On the other hand, no amount of pharmacoeconomics studies will support a product that doesn't offer a meaningful clinical advan tage unless you were to price it at a fraction of the price of competitive thera p~es. Pharmaceutical firms employ a wide variety of outcomes research methodol- ogies. This variety is as indicative of the infant stage of the discipline as it is of differences among companies. There is a significant degree of disagreement among knowledgeable parties regarding the usefulness, validity, and limits of the investigational methodologies and instruments that are being employed in the service of outcomes research. Most pharmaceutical firms produce a mix of prospective studies, retrospec- tive analyses, and economic and decision-analytic modeling. Not all of these studies are used as "end products." A significant amount of outcomes research and related work is conducted to support and guide study design for prospective studies. The following are examples of the types of support work conducted by pharmaceutical companies. Historical cohort studies may be conducted to develop information on the cost and medical resource use associated with a disease and its treatment in populations. Such analyses can also provide limited information on the natural history of a treated disease. Baseline data on the direct and indirect costs of a specific type of illness may be developed by prospectively or retrospectively collecting cost data. In some cases, companies have invested significant resources to develop cost-of- illness information to provide valid and consistent inputs into cost-effectiveness studies. . Lconomic and decision-analytic models are often used to identify the most important parameters that should be examined in a prospective study. It is useful to do this sort of preliminary work, because only a limited number of parameters can be studied in any given prospective trial. The researcher wants to study those parameters that are most sensitive to the disease state in question (and to its standard treatment options) to ensure that the results of the study will provide relevant and meaningful data on the compound studied. Similarly, it is not feasible to collect data on every disease- and treatment-related cost when conducting a prospective study. Economic modeling is used to identify which factors account for the bulk of the costs associated with a disease and its treat- ment; a prospective study can then focus on the major contributors to cost. In some cases there is no valid and reliable instrument available for mea- suring the set of quality-of-life or functional parameters appropriate to the dis- ease state being studied. Researchers may do extensive work to develop and validate new instruments or to validate existing instruments. These efforts can, on occasion, consume years.

OCR for page 152
162 ANN K. M. MARSHALL Most firms also use economic modeling and, less frequently, retrospective studies as end products. On the whole, there appears to be a growing reluctance to use retrospective analysis in stand-alone studies. This is because most of the available databases are claims databases-designed to track financial transac- tions, not clinical data. This raises validity issues with respect to the results, severely limiting a company's ability to communicate those results to customers. Even when it is possible to use existing databases, the adjustments and supple- mental work required are usually quite costly. Economic modeling is much better accepted as a stand-alone study approach. Most companies noted that, as long as a model's assumptions are reasonable and clearly stated, the results can be useful and informative. Companies often follow a three-step process in developing modeling studies: (1) develop the model struc- ture and preliminary assumptions; (2) present the study concept, model, and pre- liminary assumptions to an expert panel for consensus review; and (3) revise the model and assumptions as appropriate. Modeling is particularly useful when the endpoints being studied are rare events or take a long time to become manifest (e.g., progression of degenerative disease or death from a chronic condition). In such cases, prospective studies could take 10 to 15 years and require tens of thousands of patients. Modeling allows the researcher to project economic or patient outcomes using available data on the disease and its relevant treatment options within a time frame that is practical for real-life decisionmaking. In addition, modeling has been used as a decisionmaking tool in the early stages of product development. For example, one company reported using deci- sion-analytic modeling to determine the efficacy rate that its investigational com- pound would have to achieve to be cost-effective. This information was then used in developing the target profile of the product. In some pharmaceutical firms, those who are involved in the outcomes re- search function operate primarily as consultants to those who are involved in the clinical R&D functions, who then run the studies (particularly when studies are of investigational compounds and are incorporated into the clinical development program). In other companies, those involved in the outcomes research function operate primarily in a project manager role, and most studies are contracted to outside researchers, including both academe and independent firms. In the ma- jority of companies, however, a mix of internal and external personnel conduct outcomes research. Virtually all pharmaceutical companies use the consulting services of out- side experts in their outcomes research activities, even those companies commit- ted to developing substantial internal capabilities. In the case of quality-of-life studies, companies sometimes access the expertise of consultants on the technical aspects of instrument development or selection, but they typically conduct the studies in-house. Companies' clinical development groups generally feel fairly comfortable in their ability to develop and run a quality-of-life study and tend to view such studies as relatively uncomplicated. In some cases, this betrays an

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 163 oversimplified view of the methodological issues involved in these studies, as evidenced by the growing practice of fairly routinely inserting "short-form" or abbreviated quality-of-life measures into clinical trials. By contrast, pharmaceutical firms are more likely to commission outside researchers to conduct economic outcomes studies, whether they be prospective, retrospective, modeled, or a combination thereof. Companies are less comfort- able with the methodological issues involved in cost-benefit and cost-effective- ness studies because of the lack of standardized approaches to doing those stud- ies.5 This uneasiness is accentuated by the fact that many companies have not yet developed sufficient in-house expertise to grapple with the methodological issues involved in designing pharmacoeconomic studies although several com- panies stand out as notable exceptions to this general rule. It is interesting to note that there is considerable difference of opinion re- garding whether or not it is optimal to commission outside researchers to perform outcomes research. Some pharmaceutical firms believe that outside researchers enhance the perceived credibility of outcomes research by mitigating the appear- ance of conflict of interest. These companies tend to view their internal out- comes research function primarily as a knowledgeable and proactive sponsor of externally conducted studies. Other pharmaceutical firms are of the opinion that outcomes research is simply too important for a company to allow itself to be heavily dependent on outside expertise. These companies are committed to ex- panding their internal expertise in health economics, pharmacoepidemiology, and health services research to increase the proportion of outcomes research conduct- ed in-house. Moreover, they are confident that by producing high-quality studies they will be able to defuse any lingering skepticism about the objectivity of manufacturer-conducted outcomes research. All companies interviewed stated that their studies, whether conducted internally or externally, are designed to be published in peer-reviewed journals. Most companies reported that when they sponsor the work of outside researchers, those researchers are free to publish the results of those studies. In such cases, the sponsoring company generally does not have the right to edit or censor the researcher's publication, although it usual- ly has the right to read the manuscript prior to submission for publication. Despite some differences in how pharmaceutical firms approach outcomes research, a number of clear trends can be identified. Pharmaceutical companies are initiating outcomes research earlier in the product development cycle, con- ducting it for more compounds, and using prospective study designs more often. Increasingly, quality-of-life and economic outcomes studies are conducted si- multaneously. Although the trend in economic outcomes studies is to collect cost 5 Most companies reported that they had not done any cost-utility studies, primarily because of methodological concerns. There are very real difficulties with the generalizability of the "health state preferences" used in such studies, because there is little basis for claiming that the health state preferences of the study population represent those of the broader population.

OCR for page 152
164 ANN K. M. MARSHALL data prospectively, economic modeling studies are still used but often as a preliminary tool for identifying appropriate prospective study parameters. Pharmaceutical companies are developing a keener sense of the importance of producing timely outcomes information about new products. It is becoming routine for companies to ask, by phase I or II of the clinical development pro- gram for an investigational compound, what sorties) of outcomes research should be conducted. Prospective studies are now frequently initiated in phase II of a clinical development and evaluation program. At the same time, companies are developing a longer-term approach to outcomes research. A significant number of companies have initiated long-term patient functionality studies. This high- lights companies' recognition that the market will require outcomes data, not only when a pharmaceutical product is launched but also as it achieves broader use. Biotechnology Pharmaceutical Companies Given the high costs associated with genetically engineered drugs, the mar- ket forces driving outcomes research should apply even more convincingly to biotechnology pharmaceutical (biotech) companies. In an increasingly price- sensitive market, biotech companies need to show that their products are cost- effective, even though they are expensive in absolute terms. In comparison with traditional pharmaceutical firms, however, biotech com- panies have not been very active in outcomes research. This may be because biotech firms are typically more technology driven and less market oriented than traditional pharmaceutical firms. As a group, biotech companies have few mar- keted pharmaceutical products. Their efforts have largely been focused on devel- oping promising compounds and raising the capital to fuel those development efforts. As a consequence, many biotech companies have been better attuned to the capital markets than to the pharmaceutical marketplace. To date, only a few biotech firms have invested in outcomes research. These are firms that have developed a stronger market orientation either through the experience of market- ing a pharmaceutical product or in anticipation of a product launch. Among biotechnology companies, Amgen was a pioneer in outcomes re- search, sponsoring both economic and quality-of-life studies of its first marketed product, recombinant human erythropoietin (indicated for anemia associated with chronic renal disease), in 1989.6 Those studies were undertaken because the low level of Medicare reimbursement for erythropoietin therapy from the HCFA was 6 Although a number of early economic modeling studies comparing various thrombolytic therapy options, including Genentech's recombinant tissue plasminogen activator, were conducted, those studies were not sponsored by Genentech (Iones-Grizzle and Bootman, 1992). Genentech did pro- vide partial support for the ISIS-3 study of thrombolysis patient outcomes, sponsored by the Depart- ment of Veterans Affairs and reported in 1992 (ISIS-3 Collaborative Group, 1992).

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 165 prompting providers to underdose patients with chronic renal disease (CRD). By showing that proper dosages of erythropoietin significantly improved the well- being of patients with CRD, Amgen persuaded the HCFA to alter its reimburse- ment structure for erythropoietin. Building on the success of its erythropoietin studies, Amgen initiated outcomes studies for its next product, filgrastim (recom- binant methionyl human granulocyte colony-stimulating factor) during phase III clinical trials. These studies showed that filgrastim substantially reduced the need for postchemotherapy hospitalization and provided a basis for the positive differentiation of filgrastim from other granulocyte colony-stimulating factor products. Despite the evident success of Amgen's outcomes research efforts, other biotech companies remained inactive for some time. In 1991 Centocor conduct- ed a retrospective cost-effectiveness study of its first therapeutic product, Cen- toxin7 (then awaiting FDA approval for the treatment of septic shock). But the study was initiated late in the product development process- well into FDA's review period and, even more important, well after the hospital community had reached a widely publicized consensus about the "devastating" financial impact of a $3,700-per-dose septic shock drug. It is unclear whether the FDA's failure to approve Centoxin without further clinical trials was at all influenced by the tumult surrounding its price. What is clear is that Centocor's response to the hospitals' sticker shock was too little and too late. In the wake of the Centoxin experience, biotech companies will likely devel- op a keener appreciation of the importance of timely outcomes research. At least one biotech company, Synergen, has already begun to integrate outcomes re- search into its clinical development process (Longman, 1992~. Indeed, Syner- gen's efforts rival those of the large, traditional pharmaceutical firms and will probably serve as a model for other biotech companies. Summary Observations It is evident from the foregoing discussion that manufacturers' outcomes research is market driven. Manufacturers produce outcomes data for more or less defined audiences in response to the perceived informational requirements of those audiences and to address specific marketing concerns. The audience can include any or all of those decisionmakers who influence the approval, diffusion, pricing, or utilization of medical products. The marketing concerns include the various pricing, reimbursement, coverage, registration, and formulary access is- sues faced by manufacturers. Most device companies engaged in outcomes research have identified a pri- mary audience and are pursuing fairly straightforward reimbursement or regula 7 Centoxin is the registered trademark.

OCR for page 152
166 ANN K. M. MARSHALL tory goals. Biotech companies also have a well-defined rationale for conducting outcomes research. In a cost-constrained environment, therapies that add sub- stantially to the cost of health care must provide evidence of sufficient value to warrant that cost-even when the therapies are "breakthrough" technologies. The fact that many biotech companies are only now beginning to recognize this ratio- nale for outcomes research does not make it any less clear. By contrast, tradi- tional pharmaceutical firms, whose products rarely embody the biotech extremes of high cost and radical innovation, face more subtle and diffuse market forces.8 It is interesting to note, therefore, that traditional pharmaceutical firms have taken the lead in outcomes research. Most pharmaceutical firms are committed to the importance of outcomes research, even though many are uncertain about how the outcomes research movement will evolve. This uncertainty is largely due to the early stage of outcomes research. Most of the studies undertaken to date are still in progress and, until the results become available, one can only guess how diverse decisionmakers will respond to the range of information being generated. Granted, isolated studies have been completed, but the market re- sponse to those studies provides little guidance on the long-term role of outcomes research in decisions about the availability, utilization, and pricing of medical technologies. Simply stated, it is not clear what sort of market dynamic will emerge as outcomes research becomes more widespread. Because the technology assess- ment trend is young, open questions abound: lIow extensively will physicians incorporate information from outcomes research in their therapeutic choices? Will patients become a significant audience for information on outcomes? Will formulary committees fall back on simple cost-minimization strategies in the face of tight budgets? Will regulatory bodies become involved in monitoring outcomes research? These issues are of vital importance not only to manufactur- ers but also to future medical innovation. IMPLICATIONS FOR THE HEALTH CARE INDUSTRY AND MEDICAL INNOVATION Outcomes research provides answers to questions about what therapies work best, under what conditions, to promote the outcomes that patients care about. As mentioned above, the answers to these questions should provide a basis for rationalizing health care expenditures and improving patient care and it is for ~ The audience for outcomes research is far more diverse in the United States than elsewhere, particularly with respect to pharmaceuticals. Outside of the United States, pharmaceutical prices and reimbursement levels are often controlled by government bodies that, in recent years, have placed increasing importance on outcomes data, particularly economic outcomes data. In some countries, the pricing determination process is closely linked to the approval process. At the extreme, Australia and Canada are moving toward requiring cost-effectiveness data as part of the registration package for pharmaceuticals.

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 167 these reasons that outcomes research has deservedly attracted significant atten- tion in recent years. However, at least two major issues must be addressed if the promise of outcomes research is to be realized. First, although manufacturer-sponsored outcomes research generates impor- tant and useful information, that information alone will not support the broader, systemic promise of outcomes research. Manufacturers appropriately focus their outcomes research activities on their own products and on the procedures in which their products are used. Manufacturers do not sponsor outcomes research on medical procedures or therapies that are independent of their products unless a procedure or therapy serves as the established comparator to an alternate proce- dure or therapy for which a manufacturer's product is used. Moreover, although manufacturer-sponsored outcomes research is almost always comparative, such research rarely addresses every therapeutic option relevant to the disease state in question nor is it realistic to expect manufacturers to support such broad-based research. Yet, many of the most striking opportunities for improving patient care and rationalizing health care expenditures depend on conducting outcomes research on (1) procedures that are not product driven (e.g., prostatectomies, hysterecto- mies, and cesarean deliveries) or (2) clinical conditions whose therapeutic op- tions are only partially product driven (e.g., coronary artery disease, cataracts, lower back pain, and peripheral vascular disease). In the former cases, manufac- turers have no rationale to conduct outcomes research. In the latter cases, manu- facturers may have a rationale to conduct outcomes research, but most likely it would not be inclusive of all clinically relevant therapeutic modalities. It is important to note that third-party payers have begun to sponsor outcomes re- search on investigational procedures that are not essentially product driven, such as autologous bone marrow transplantation for breast cancer. However, with the exception of studies conducted by AHCPR-sponsored Patient Outcomes Re- search Teams, there has been little comprehensive outcomes research conducted on clinical conditions or on noninvestigational, non-product-driven medical pro- cedures. Much more work of this sort is needed if the emerging body of out- comes research is to provide an adequate basis for improving patient care and . . . . rationalizing expenditures. The second issue is more fundamental. Even supposing that an adequate body of outcomes research information is being generated, there remains the critical challenge of appropriately integrating that information into actual medi- cal practice. As indicated in other chapters of this volume, current incentive structures do little to encourage physicians to modify their practice patterns in response to outcomes research, particularly when such modification requires giv- ing up an established and familiar therapeutic approach or a lucrative procedure. Even today, few medical schools incorporate courses on outcomes research or epidemiology in their core curricula, and correspondingly few physicians are inclined to seriously consider outcomes research findings in their day-to-day

OCR for page 152
68 ANN K. M. MARSHALL practice decisions. Recognizing the importance of this issue, AHCPR has under- taken research on the relative effectiveness of various means of disseminating its technology assessment and outcomes research findings to relevant audiences. Dissemination strategies are only part of the picture, however. If outcomes re- search is to have a meaningfully positive impact on the U.S. health care system, then the systemic factors that affect the ability and willingness of health care professionals to act on outcomes research information must be confronted. It is important to recognize and address those factors that limit the potential benefits of outcomes research, because substantial resources are being devoted to it. The cost of a single prospective study can easily exceed $1 million. Meta- analyses and retrospective studies are less expensive, but these approaches are usually feasible only after a drug or a device has achieved relatively broad use, which typically happens only after it has been marketed. Since key decisionmak- ers are increasingly demanding that manufacturers provide outcomes research when a product is launched, manufacturers are increasingly conducting prospec- tive studies, the costliest type of outcomes research. Thus, an important consequence of heightened market demand for outcomes research is that the investment required to bring a new drug or high-tech device to market has increased. Because the outcomes research trend is still so new, one can only speculate about the ultimate implications of this shift. However, the classic theory of competitive dynamics is suggestive. Generally, as an industry moves toward requiring larger investments to bring new products to market, the number of players in that industry will decrease and the size of each player will increase. This is because larger up-front product development costs require that firms have a larger revenue base (critical mass) to sustain the increased cash outflow during the development stage of product life cycles. This would imply that the increased demand for outcomes research constitutes a competitive advan- tage for larger firms and will tend to reduce the number of smaller firms through acquisitions, mergers, or failures. This dynamic is likely to affect high-tech device companies more profound- ly than pharmaceutical companies, because (1) the increased expense of out- comes research, viewed as a percentage of total product development costs, is considerably higher for high-tech device companies than for pharmaceutical com- panies, and (2) unlike the pharmaceutical industry, the high-tech device industry includes a significant number of relatively small companies, which makes the device industry more sensitive to competitive shifts in the requisite critical mass. Indeed, looking at the change in the average size of companies receiving FDA approval for class III devices between 1981 and 1988 a period during which class III device approval times lengthened and the FDA placed increasing em- phasis, for devices for which premarketing approval was being sought, on evi- dence of clinical utility and rigor in biostatistical analysis, it appears that the high-tech device industry responded predictably to the escalating cost of bringing new products to market. In 1981, the average size of companies who received

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 169 approval for class III devices was $300 million, whereas by 1988 the average size had increased fourfold to $1.2 billion (in constant 1980 dollars; Bucci- et al., 1990). The competitive dynamics of the pharmaceutical industry are also likely to be affected by the increased demand for outcomes research, but differently. In- vestment in outcomes research can be used to create barriers to entry in specific markets, particularly in markets for pharmaceutical therapies that treat chronic conditions. This is because key decisionmakers show keen interest in outcomes research on the comparative impact of pharmaceutical therapies on endpoints such as survival and disease progression in patients with chronic diseases; but such studies can take 5 to 15 years, require thousands of patients, and cost mil- lions of dollars. When firms that are established in a chronic therapy market (e.g., for hypertension) invest in such research, they create a twofold barrier to entry. First, they raise the ante for any newcomer wishing to enter the market. Second, they create a "time-buffered" competitive advantage, because the new entrant will have to wait years before the results of outcomes research on its new compound are available. Predictably, this sort of competitive positioning is al- ready under way in some of the markets for therapies that treat chronic condi- tions such as asthma and hypertension. The strength of such barriers to entry will be directly proportional to the market's insistence on long-term outcomes re- search information. The potential impact of these competitive dynamics on innovation is sober- ing. Clearly, if the above speculations about demand for outcomes research and barriers to entry in the pharmaceutical industry were to prove correct, the effect would be to dampen pharmaceutical innovation in certain markets. Of greater concern is the predicted impact on the high-tech device industry. As Alan Kahn has described, innovation in the medical device industry is characterized by "smaller companies taking the lead, twith] a more fluid innovation process" that can respond to an ill-defined and rapidly changing market (Kahn, 1991, p. 89~. Smaller companies are better able to stay close to their customers, approach unmet needs in an entrepreneurial fashion, and respond quickly to market input about product performance and features-qualities that have driven much of the innovation in high-tech devices for the past 20 years. If the high-tech device industry evolves toward fewer and larger companies, the United States can ex- pect a decline in radical medical product innovations. CONCLUSION Rarely in life does one encounter an unequivocally positive trend, and the trend toward increasing demand for outcomes research is no exception to this cheerless observation. The growing market demand for outcomes research, par- ticularly as a condition for registration, reimbursement, or formulary acceptance, will likely slow the pace of innovation in drugs and devices to some degree. It

OCR for page 152
170 ANN K. M. MARSHALL would be a mistake, however, to conclude that the outcomes research trend is therefore ill advised. The information generated by outcomes research has real potential to improve patient care and help rationalize health care expenditures. Of course, as discussed above, realizing this potential requires that (1) sufficient outcomes research be conducted on procedures and clinical conditions that are not product driven and (2) outcomes information be appropriately incorporated into actual medical practice. Given the cooperative efforts by the venous stake- holders in the U.S. health care system to address these issues, the promise of outcomes research may warrant a limited trade-off with the pace of medical product innovation. REFERENCES Bombardier, C., Russell, I. J., Ware, J., et al. 1986. Auranofin therapy and quality of life in patients with rheumatoid arthritis: Results of a multicenter trial. American Journal of Medicine 81:565-578. Bucci, V. A., Hall, N. C., and Reiss, J. B. 1985. New obstacles in the path of marketing new medical devices: The stream of regulation. Journal of Health Care Technology 2:81-96. Bucci, V. A., Hall, N. C. and Norcia, L. A. 1990. FDA Decision-Making on Class III PMA Products: An Analysis of 124 PMA Approvals. Unpublished manuscript. Croog, S. H., Levine, S., Testa, M.A., et al. 1986. The effects of antihypertensive therapy on the quality of life. New England Journal of Medicine 314:1657-1664. Eddy, D. M. 1990. Should we change the rules for evaluating medical technologies? In: Institute of Medicine. Medical Innovation at the Crossroads. Vol. 1, Modern Meth- ods of Clinical Investigation. A. C. Gelijns, ed. Washington, D.C.: National Acad- emy Press, pp. 117-134. Freeman, R. A. 1991. Organizational issues in outcomes management research. Journal of Pharmacoepidemiology 2~2~:59~7. Freund, D. A., and Dittus, R. S. 1992. Principles of pharmacoeconomic analysis of drug therapy. PharmacoEconomics 1~1~:2~29. Institute of Medicine. 1990. Medical Innovation at the Crossroads. Vol. 1, Modern Methods of Clinical Investigation. A. C. Gelijns, ed. Washington, D.C.: National Academy Press. ISIS-3 Collaborative Group. 1992. ISIS-3: A randomized comparison of streptokinase vs tissue plasminogen activator vs anistreplase and of aspirin plus heparin vs aspirin alone among 41,299 cases of suspected acute myocardial infarction. Lancet 339:753- 770. Jack, W. 1991. Pharmaceutical differentiation through quality of life measurement: A case study. Journal of Pharmaceutical Marketing and Management 6~1~:65-85. Jones-Grizzle, A. J., and Bootman, I. L. 1992. Pharmacoeconomics of genetically engi- neered drugs. PharmacoEconomics 1~1~:45-53. Kahn, A. 1991. The dynamics of medical device innovation: An innovator's perspective. In: Institute of Medicine. Medical Innovation at the Crossroads. Vol. 2, The Chang

OCR for page 152
MANUFACTURERS' RESPONSES TO DEMAND FOR OUTCOMES RESEARCH 171 ing Economics of Medical Technology. A. C. Gelijns and E. A. Halm, eds. Wash- ington, D.C.: National Academy Press, pp. 89-95. Kane, N. M., and Manoukian, P. D. 1989. The effect of the Medicare prospective payment system on the adoption of new technology. New England Journal of Medi- cine 321:1378-1383. Longman, R. 1992. Synergen rewrites the pricing debate. In Vivo 10~7~:20-24. McGivney, W. T. 1991. Hurdles to technology diffusion: What are expectations for PET? Journal of Nuclear Medicine 32:660~64. Office of Device Evaluation, Food and Drug Administration. May 3, 1991. Clinical utility and premarket approval. Internal memorandum from the director, Office of Drug Evaluation, to the Office of Drug Evaluation review staff. Office of Technology Assessment, U.S. Congress. 1982. Strategies for Medical Technol- ogy Assessment. Washington, D.C.: U.S. Government Printing Office. Perry, S. 1988. Technology assessment in health care: The U.S. perspective. Health Policy 9:317-324. Stern, W. R. 1990. Meeting of the Food and Drug Administration gastroenterology- urology device section advisory panel on extracorporeal shock-wave lithotripsy for gallbladder stones, October 19, 1989. American Journal of Gastroenterology 85: 238-240. Steward, S. 1991a. Part 1: Building bridges with cost-effectiveness. Pharmaceutical Executive 11~91:58-70. Steward, S. l991b. Part 2: Taking the lead with cost-effectiveness. Pharmaceutical Executive 11~111:46~1. Steward, S. 1992. Part 3: Developing strategies with cost-effectiveness. Pharmaceutical Executive 12~2J:7~84, 109. VanAntwerp, M. 1985. Diffusion of new technologies: Can Medicare payment decisions be made faster and more efficiently? Journal of Health Care Technology 2:97-103. Wennberg, J. E. 1990. What is outcomes research? In: Institute of Medicine. Medical Innovation at the Crossroads. Vol. 1, Modern Methods of Clinical Investigation. A. C. Gelijns, ed. Washington, D.C.: National Academy Press, pp. 33~6. Wennberg, I. E., and Gittelsohn, A. 1973. Small area variations in health care delivery. Science 182:1102-1108. Wennberg, J. E., Freeman, J. L., and Culp, W. J. 1987. Are hospital services rationed in New Haven or over-utilised in Boston? Lancet 1:1185-1189. Wennberg, J. E., Freeman, J. L., Shelton, R. M., and Bubolz, T. A. 1989. Hospital use and mortality among Medicare beneficiaries in Boston and New Haven. New En- gland Journal of Medicine 321:1168-1173. The Zitter Group and Technology Assessment Group. 1992. Pharmaceutical Outcomes Activities. The Zitter Group and Technology Assessment Group.