Integrative Medicine and Systemic Outcomes Research Issues in the Emergence of a New Model for Primary Health Care

SPECIAL ARTICLE Integrative Medicine and Systemic Outcomes Research Issues in the Emergence of a New Model for Primary Health Care

Iris R. Bell, MD, MD(H), PhD; Opher Caspi, MD; Gary E. R. Schwartz, PhD; Kathryn L. Grant, PharmD; Tracy W. Gaudet, MD; David Rychener, PhD; Victoria Maizes, MD; Andrew Weil, MD

linicians and researchers are increasingly using the term integrative medicine to refer to the merging of complementary and alternative medicine (CAM) with conventional biomedicine. However, combination medicine (CAM added to conventional) is not integrative. Integrative medicine represents a higher-order system of systems of care that emphasizesC wellness and healing of the entire person (bio-psycho-socio-spiritual dimensions) as primary goals, drawing on both conventional and CAM approaches in the context of a supportive and effective physician-patient relationship. Using the context of integrative medicine, this article out- lines the relevance of complex systems theory as an approach to health outcomes research. In this view, health is an emergent property of the person as a complex living system. Within this concep- tualization, the whole may exhibit properties that its separate parts do not possess. Thus, unlike biomedical research that typically examines parts of health care and parts of the individual, one at a time, but not the complete system, integrative outcomes research advocates the study of the whole. The whole system includes the patient-provider relationship, multiple conventional and CAM treatments, and the philosophical context of care as the intervention. The systemic outcomes encompass the simultaneous, interactive changes within the whole person. Arch Intern Med. 2002;162:133-140

Any intelligent person can study medical INTEGRATION VS ASSIMILATION literature and understand when or when not to use various treatments. What is so difficult, The term integrative medicine is often used even for a skilled physician, is to apply this to refer to blending the best of conven- knowledge in individual cases. For those who tional (allopathic) and complementary and know nothing about the fundamentals of alternative medicine (CAM). At face value, healing and treat it casually and talk a lot, nothing seems difficult. They don’t think there this goal seems straightforward and prom- is any illness that requires careful deliberation. ises an improved package of medical care The common run of people thinks medicine can for the consumer. However, a closer exami- be learned quite easily, whereas it is really nation of the process of this integration extremely difficult to master even for a raises a number of complex practical and conscientious physician. conceptual issues with which medicine as a field must grapple. This article deals with Maimonides (1135-1204) this process of integration and discusses im- The Regimen of Health Care; 4:7-8 plications for health care outcomes research. We suggest that by adopting a worldview derived from complex systems theory From the Program in Integrative Medicine, Departments of Medicine (Drs Bell, Caspi, in which the whole equals more than the Schwartz, Grant, Gaudet, Rychener, Maizes, and Weil), Psychiatry (Drs Bell and sum of its parts, a new perspective for medi- Schwartz), Psychology (Drs Bell, Caspi, and Schwartz), Neurology (Dr Schwartz), Family & Community Medicine (Drs Bell and Weil), Obstetrics and Gynecology cine and health care research emerges. 1 (Dr Gaudet), and Pharmacy Practice (Dr Grant), The University of Arizona Colleges The dictionary defines integrate as “to of Medicine and Pharmacy, and the Southern Arizona Veterans Affairs Health Care unite with something else,” “to incorpo- System (Dr Bell), Tucson. Dr Gaudet is now with Duke University Center rate into a larger unit.” Although the dic- for Integrative Medicine, Durham, NC. tionary definition of the noun integration

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 133

©2002 American Medical Association. All rights reserved. includes “the incorporation of Within conventional care, some nal contexts to fit a few of these equals into society,”1 it is evident multidisciplinary approaches to pa- smaller pieces into the dominant that the assumption implicit in the tient care have already laid a founda- model of conventional care and merger process for mainstream tion for the more fully integrative medical research. For example, nu- medicine is of an initial inequality medicine that could emerge: for ex- merous studies have investigated the in power and worthiness between ample, (1) in geriatrics, developing efficacy of acupuncture for various conventional and CAM approaches. multidisciplinary special care teams9 Western disorders,22,23 but virtu- That is, the politically dominant and end-of-life programs to optimize ally no studies examine the effec- “larger unit” (ie, conventional quality of life in hospice care10; (2) in tiveness of the sum total of Chinese medicine in the Western world; see chronic pain treatment programs, medicine as practiced. In acupunc- Zollman and Vickers2) carries the applying multidisciplinary care to ture-only research, the effect sizes are values, culture, and conceptual improve individual self-efficacy and often modest.24,25 Yet, traditionally, framework into which it expects the quality of life11; (3) in psychiatry, Chinese medicine uses a coordi- smaller unit (ie, CAM) to assimi- blending social supports, psycho- nated and individualized program late. It operates from the assump- therapy, and medications12 as well that includes various combinations tion that each CAM intervention, as emphasizing the patient’s respon- of diet, botanicals, acupuncture, acu- once tested and proven effective, can sibility for his or her own recovery13; pressure, qi gong, and environmen- be incorporated into conventional (4) in family medicine, valuing good tal interventions to address the care as now practiced.2 physician-patient relationships14 and unique, systemic disturbance pat- preventive interventions15; and (5) in terns in a given patient.18 It is a test- INTEGRATIVE MEDICINE behavioral medicine/health psychol- able hypothesis that the effect sizes IS NOT CAM ogy, using behavioral interventions of the full treatment program could to foster self-care and self-efficacy in be much larger and more clinically However, much of the conventional patients with diabetes16 or arthritis.17 significant if the entire Chinese contemporary practice of physicians, Although these examples of care- medicine treatment program were especially for the treatment of patients beyond-the-lesion offer important studied as used.26 with chronic diseases, has continued models for a more ideal form of medi- Moreover, Western medical re- to focus on a specific somatic disease cine, even these models are not nec- search usually assumes that its ap- process at the end organ rather than essarily widely available and/or acces- proach to diagnosis is the preferred on healing the individual person. Dis- sible within the current health care way to label a patient. It requires ho- affection with how physicians provide system. mogeneous groups of patients with conventional care and rely on phar- As it evolves, truly integrative conventional diagnoses for study. maceutical medicine continues to medicine also depends for its philo- But each system of CAM has its own grow among consumers and physi- sophical foundation and patient- theory-driven method for categoriz- cians alike.3,4 Integrative medicine is centered approach on systems of ing patients. Within a group of pa- acomprehensive,primarycaresystem CAM that emphasize healing the per- tients with asthma, for example, each that emphasizes wellness and healing son as a whole (eg, traditional Chi- CAM system is likely to identify of the whole person5 (bio-psycho- nese medicine, Ayurvedic medi- several different subtypes, each re- socio-spiritual dimensions) as major cine, and classic homeopathy).18 quiring a unique set of interven- goals, above and beyond suppression These CAM systems diverge the most tions for optimal treatment re- of a specific somatic disease.6,7 In the in philosophy, diagnosis, and treat- sponse (Figure 1A). In classic ideal situation, the patient and the in- ment technique from conventional homeopathy, for instance, each of 4 tegrative practitioner are partners in medicine,18 and thus remain margin- different patients with chronic the effort to develop and implement alized. Using philosophy-driven asthma might respond to only 1 rem- a comprehensive treatment plan for frameworks that seek balance, har- edy type chosen on the basis of his issues that extend far beyond the im- mony, and proper flow throughout or her individualized symptom pat- mediate chief complaint and/or con- the patient, they diagnose and treat tern (eg, Arsenicum album or Kali ventional diagnostic category.6 Truly patterns of dysfunction within the en- carbonicum or Medorrhinum or Na- integrative medicine draws from con- tire person as an indivisible sys- trum sulphuricum). In other words, ventional and alternative techniques tem.19 From the perspective of West- homogeneity of sample selection to facilitate healing and to empower ern science, such CAM systems offer from a conventional medical per- the patient because healing is believed (1) virtually no commonly used, re- spective is likely to break down into to originate within the patient rather ductionist, scientific methods by heterogeneous diagnoses from each than from the physician.6 Thus, the which to study them; (2) no obvi- CAM system’s point of view.27 Even philosophy of integrative medicine is ous ways to incorporate them into so, despite different diagnostic la- compatible with the World Health conventional practice; and (3) no bels, these CAM systems share an Organization’s definition of health Western conceptual framework into emphasis on looking for patterns of that equates health with well-being: which they fit, even if aspects of care dysfunction that manifest through- “Health is a state of complete physi- are found useful.20,21 out the individual rather than iso- cal, mental, and social well-being and As a result, clinicians and related problems in separate bodily not merely the absence of disease or searchers often break off parts of subsystems. For homogeneity of infirmity.”8 these CAM systems from their origi- study samples, ideal designs would

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 134

©2002 American Medical Association. All rights reserved. Selection Procedure Asthma Orthodox Chronic Obstructive Patients Diagnostic Bronchitis Pulmonary Disease Criteria First Selection (Orthodox Diagnosis) Emphysema Alternative Diagnostic Criteria “Homogeneous” Group of Patients

Second Selection (Alternative Diagnosis)

“Homogeneous” Group of Patients Acupuncture Homeopathy Naturopathy A Treatment Procedure B A A A H H H H N N N 1 2 3 1 2 3 4 1 2 3 Selected Patients

A1 Earth H1 Silicea N1 Hyper

A2 Water H2 Sulphur N2 Rigid Treatment Placebo

A3 Wood H3 Thuja N3 Paradoxical Registration (and Comparison of Changes) H4 Pulsatilla Figure 1. Double selection procedure for research designs testing complementary and alternative medicine (CAM) techniques for treating patients with a given conventional medical diagnosis that meets criteria for one of multiple, different CAM system diagnoses. Reproduced with permission from Vincent C, Furnham A. Complementary Medicine: A Research Perspective. New York, NY: John Wiley & Sons Ltd; 1997. involve a double selection proce- A B dure: first, for a specific conven- Social Environment tional diagnosis, and second, for a Transcendent specific CAM system diagnosis from Physical Environment among the multiple possibilities Social Environment (Figure 1B). PERSON Research designs that ignore the Physical Environment diagnostic approach of a given CAM PERSON Organs system can achieve only weak tests Organs of the intervention program’s ability Cells to benefit patients. The analogy Cells Molecules would be for conventional researchers to look at the ability of a single antibiotic to treat “infections” as a ho- Molecules mogeneous diagnosis. In a suffi- Figure 2. A, Domains for research in conventional biomedicine. The person is the clinical focus, but ciently large sample of patients with research examines the subsystems of the person as separate, static units. Reductionism limits the infections from various viral, bacte- number of variables under study and progressively eliminates consideration of the context (higher-order rial, and fungal sources, the drug systems and dynamics) in which the studied subsystem functions. This approach permits optimal understanding of each component part of the person. B, Domains for research in integrative medicine. would demonstrate some overall ef- The person is the clinical focus, but the research examines the person as an intact, complex, dynamic ficacy, but the study design would ob- system, composed of lower-order systems and existing within higher-order systems. Integrative research scure the drug’s outstanding capac- includes multiple variables in interaction and emphasizes the evolving context (higher-order systems and ity to kill specific bacteria as opposed dynamics) in which the person as a system functions. This approach permits optimal understanding of the person as a living system within larger systems. to all infectious agents. A tenet of integrative medicine is that the sources of good medical the integrative medical care out- a specific constituent on a bodily practice can be conventional and/or lined above, not just isolated CAM subsystem rather than with an em- CAM.6 Valuing scientific evidence as techniques, is itself economically vi- phasis on the entire simultaneous a method to augment societal under- able (eg, within health care payment spectrum of actions the agent can ex- standing of human life and health, in- systems in the United States or ert on multiple, interdependent sub- tegrative medicine recognizes that England).29 In this consumer-driven systems of the individual as a whole good medicine must always be based process, one challenge for health out- (Figure 2A vs Figure 2B). That is, in good science that is inquiry driven comes researchers is to reevaluate the all agents, both conventional and and open to new paradigms.6 Yet evi- types of outcomes on which to fo- CAM in origin, have in common the dence suggests that the medical cus. Even in conventional research, likelihood of exerting simulta- establishment does not necessarily studies on the cost-effectiveness of neous, multiple, interdependent ac- reciprocate an openness to new para- preventive interventions, for ex- tions. Research on reasons for pa- digms.28 Rather, economic pragma- ample, are difficult and costly.30 tients’ use of CAM indicates that the tism, not idealism or a change to- Moreover, even when research- philosophical orientation toward the ward an integrative philosophy that ers choose to study those treatment entire person attracts them to CAM patients may prefer,5 mainly drives the methods that are most similar to the treatments, despite the need for out- explosive growth of mainstream ser- pharmacological orientation of con- of-pocket expenditures.5 Reduction- vices and research in CAM. At the ventional medical care (eg, botani- ism has an essential place in re- same time, there is currently little evi- cals or nutrients), they usually do it search on the effects of conventional dence that any model for delivering with a focus on a specific action of and CAM techniques on specific tar-

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 135

©2002 American Medical Association. All rights reserved. get subsystems, but it inherently spiritual level as well as on the physi- component parts [organs, cells, mol- misses the full systemic orientation cal plane.18,19 This type of world- ecules].”40 A person is also a dy- of integrative medicine. Other sci- view can lead to the belief that the namic system in that he or she entific approaches are indicated to most effective interventions must evolves over time.41 If we look at gain some objective distance and ex- treat the spiritual disturbance, in medicine itself as a complex, dy- amine the larger picture.31 many cases as a source of the physi- namic system (one whose proper- These considerations raise the cal manifestations. Integrative medi- ties change with time),42 the blend- issue of proper design and meth- cine proposes that the origins of dis- ing of conventional and CAM ods for integrative medical re- ease are multifactorial more than approaches could produce not search. They do not suggest a need hierarchical, and include genetic, merely a new toolbox of tech- for 2 different standards for out- physical, emotional, psychological, niques with which to treat disease, comes research.32 Rather, they indi- and spiritual issues. Thus, an inte- but also an emergent worldview for cate the importance of careful meth- grative medicine approach seeks to the field overall. odologies that include not only the discern multiple perceived origins of An emergent is a property of a same level of scientific rigor used for a disease process and addresses them complex system in which the whole conventional medical research top- all, but without necessarily empha- is more than the sum of its parts.39 ics, but also thoughtful designs hon- sizing spirituality as a root cause. As That is, the larger system exhibits oring the philosophical founda- a result, the scope of consider- properties that are not the result of tions and clinical practices of each ations in an integrative medicine simply summing the properties of its CAM system.31 practice can span the patient’s spiri- component parts. Thus, no part by tual life, relationships, and mind- itself has certain properties that the WORLDVIEWS AND body practices in addition to herbs, larger system possesses. This emer- HEALTH OUTCOMES physical manipulation techniques, gent worldview engenders a unique medications, or surgical proce- form of health care and a new ap- Health care outcomes research is an dures. Integrative medicine as- proach to medical outcomes re- emerging field.33 It contributes sub- sumes that the individual has the po- search, as illustrated in Figure 2A-B. stantially to the knowledge base of tential for healing at the spiritual But what are the implications for in- medicine and health care and pro- level, even when physical healing tegrative medical research? We be- vides the data necessary for health does not take place.6,7 lieve that only through the synthe- policy makers. The problem is that Conventional medicine, while sis of reductionist and systemic health care outcomes may be attrib- respecting patients’ religious and science will outcomes research re- uted to many factors, only some of spiritual values, has confined itself sult in a qualitative transformation which are under the clinician’s con- largely to the belief that the physi- of health care. trol. Whereas it is outside the scope cal manifestations are the disease and of this article to discuss the issue of the primary domain for medical in- IDENTIFYING AND causality in medicine, it is crucially tervention. Consequently, differ- WEIGHTING HEALTH important for the reader to under- ences between the views of conven- OUTCOMES stand that outcomes are essentially tional medicine, various CAM probability statements.33 As in court, systems, and integrative medicine on It is logical to question the value of only outcomes results that are per- the nature of disease can lead to di- the healing-oriented integrative suasive enough will constitute ac- vergent treatment plans and even to medical approach because limited ceptable evidence and will support different goals for healing.19,37 Fu- data exist to suggest that this ap- one health policy over another.34 ture research must specify and com- proach has any advantages over What might be persuasive for pare the outcomes of different treat- other medical worldviews. A criti- one might not be for another, and ment programs derived from the cal step in developing any out- one’s worldview plays a role in one’s different worldviews. comes research is the creation of a ability to relate to new informa- conceptual model that indicates tion.35 Worldviews and the values SYSTEMS THEORY–BASED what is believed to contribute to the placed on different health out- CAM AND INTEGRATIVE outcomes and that includes all the comes are closely related. Thus, the OUTCOMES RESEARCH pertinent variables that are rel- values that underlie medical care evant to the evaluation of the sys- shape the scientific questions that re- Systems theory provides a rational tem under study.33 The classic view searchers ask, the health outcomes conceptual framework within which of quality of health care can gener- they measure, and their interpreta- to evaluate CAM systems, integra- ally be divided into 3 components: tion of the results.36,37 tive medicine, and patient- rather structure (providers’ competency, In health care, for example, one than disease-oriented clinical re- equipment, etc), process (what was convergent theme that many differ- search.36-38 Systems theory and sys- done? how well?), and outcomes ent systems of CAM share (eg, Na- tems science involve the study of the (the results of the intervention).43 tive American, classic homeopa- whole as a whole.39 That is, a com- Thus, optimal integrative out- thy, traditional Chinese medicine, plex system such as a human being comes research designs should be in- and Ayurvedic medicine) is that a is “one whose properties are not fully clusive and comprehensive rather given disease may manifest at the explained by an understanding of its than anecdotal and narrow. Yet each

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 136

©2002 American Medical Association. All rights reserved. of these 3 aspects presents its own comes controversy relates to the core ing systems theory principles as a unique challenges in CAM system of this article and to the frame of ref- framework. Several reasons may un- and integrative outcomes research. erence of the definition of health.It derlie this phenomenon: (1) the un- deals with what matters most. The familiarity of most clinical research- Structure World Health Organization defini- ers with systems constructs and mul- tion of health2 is closer to the world- tivariate scientific methods; (2) the Clinical research generally sets ran- view of integrative medicine as dis- implicit assumption that CAM should domized controlled trials as the gold cussed here than it is to that of provide tools for better conventional standard. While this method may es- conventional medicine. However, by care, not facilitate the transformation tablish strong causality through the considering various viewpoints, we of conventional care into a new, inte- enhancement of internal validity, gen- may shed more light on this issue. grativemedicine;and(3)justasagiven eralizability is sacrificed.44 This Should the primary goal of a CAM system such as traditional Chi- trade-off seems especially challeng- physician be solely to eliminate dis- nese medicine is a complex whole ing in CAM and integrative medi- ease, or should it also be to optimize composed of multiple modalities, so cine research where practices are so well-being? According to Relman, in integrativemedicineisanevenhigher- diverse and practitioner compe- a debate with Weil,50 “Medicine can- order system of multiple systems. The tency is far from being well de- not be expected to make unhappy appropriate scientific methods for fined.45 It is difficult, for example, to people happy, or frightened people assessing the multicausal illnesses, compare the effectiveness of acupunc- calm.” Is it a proper role for a physi- multiple interventions, and multidi- ture treatment for a given conven- cian to assist a patient toward grow- mensional outcomes (bio-psycho-so- tional diagnosis in different settings ing in inner peace and spiritual well- cio-spiritual) claimed in integrative and/or using different techniques. The being, in addition to subduing the medicine require complex multivar- CAM systems and techniques in disease process in the body? What iate design and statistical techniques. practice are indeed highly hetero- outcomes matter to the individual pa- Though familiar to many sociologists geneous.46 As a start, some CAM re- tient, and what differential weights and behavioral scientists, these tools searchers have generated method- do other stakeholders such as phy- are typically unfamiliar to medical re- ological innovations that may sicians, third-party payers, or hospi- searchers.52 improve the likelihood of obtaining tal administrators place on the out- A reductionist approach to sci- reliable, if not generalizable, results comes that the patient desires? Do the ence is valuable. However, to rely ex- (eg, development and dissemination outcomes that matter most to these clusively on this approach to evalu- of research practice manuals based on other stakeholders hold a compa- ate CAM systems and integrative multiple expert practitioners’ ap- rable weight for the individual pa- medicine (1) reveals the bias to- proaches within a given system for a tient? Who chooses the outcome ward the conventional system and given conventional diagnosis).47 goals in the end, and how do re- investigators’ unfamiliarity with the searchers measure success? possibility of emergent properties in Process The worldviews and roles of per- complex systems and (2) in gen- sons within the health care system in- eral fails to reflect the way the real This facet of outcomes research deals fluence the selection of outcome goals world operates.53 In clinical prac- with the appropriateness of the treat- and their relative weighting. Im- tice, physicians seldom recom- ment in relation to its nature and qual- plicit in the worldview of integrative mend single interventions in isola- ity. Problems here arise from 2 medicine, consistent with the patient- tion. Furthermore, one cannot study sources of potential biases and limi- centered approach to health care,6,51 an emergent property without keep- tations: (1) the human factor (ie, Who is the belief that the patient is the most ing the system that generates it in- would do the evaluation? Comple- important stakeholder and that the tact. Our conceptual framework for mentary and alternative medicine rest of the system must give higher systems theory–based outcomes practitioners who are not stakehold- priority to the patient’s needs and val- clinical research thus includes 3 do- ers but who are well skilled in the sci- ues than it does now within conven- mains: outcome design, outcome entific method are hard to find) and tional care. The integrative practi- measures, and outcome analysis. (2) the criteria and measures used (ie, tioner would recognize and act on a allopathic or alternative?). These 2 patient’s reluctance or readiness to use Outcome Design practical problems of integrative re- a particular CAM intervention when search are especially challenging par- appropriate. Comparison among randomized tially because conventional and CAM groups serves as the cornerstone of providers often speak “different lan- INTEGRATIVE VS clinical studies. This is such a well- guages” and value different out- REDUCTIONIST HEALTH accepted design that only seldom do comes.31,48,49 OUTCOMES RESEARCH we debate the appropriateness of these comparison groups. Systems Outcomes Although consensus papers on per- theory rationalizes the inclusion of forming CAM research conclude that at least 1 arm in the design of out- For obvious reasons this is a domain it is entirely possible to apply available comes studies that tests the effect of that is the most emotionally charged; scientific methods to study systems of the total sum of interventions un- the stakeholders are many. The out- care intact,26,31 very few studies are us- der consideration, consistent with

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 137

©2002 American Medical Association. All rights reserved. the assertion that the whole system nitive-behavioral interventions, and Thus, it is essential to carry out may express characteristics not pre- (4) fiber diet plus biofeedback, cog- carefully planned, multifactorial in- viously documented in or pre- nitive-behavioral treatments, and tervention studies to develop an ap- dicted by studies of its component parental support.59 The 3 interven- propriate evidence base toward iden- parts.36,37,39 Other arms of such stud- tions added to the dietary manipu- tifying what constitutes maximally ef- ies would examine the effects of each lation were psychosocial. Notably, fective and safe practice. In practice, intervention by itself as well as in the data suggested that the full com- theresultsfromawholetreatmentsys- comparison with standard conven- bination of the 4 interventions was tem may be different from a simple tional care. Thus, when an integra- not necessarily superior to a com- summing of results expected from tive practitioner recommends mul- bination of fiber plus 1 psychoso- studies of its separate constituents. tiple interventions, conventional cial intervention. Such findings sug- Observational study designs are and/or alternative, we will never gest a need to evaluate the categories a possible strategy for evaluating com- know what the combination pro- and potential synergy, or lack plex interventional systems. All obser- duces in terms of clinical benefit and thereof, among different interven- vational studies have one crucial de- risk unless we study them all to- tions from which a treatment pro- ficiency that may be considered as an gether. Furthermore, to the extent gram emerges, not just the sheer important threat to their validity: the that a provider is involved in the number of interventions. design is not a controlled experimen- treatment, the provider-patient re- Moreover, evidence from con- tal one. Each patient’s treatment is de- lationship is another key factor in ventional research in chronic pain liberatelychosenratherthanrandomly evaluating the outcomes. For self- raises an important question as to how assigned, so there is an unavoidable care treatments and for other treat- to develop an individualized treat- risk of selection bias and systematic ments chosen from the cooperative ment plan for a given patient. Turk,60 differences in outcomes that are not partnership of patient and physi- for example, has found that pain pa- due to the intervention itself.62 Fur- cian, changes in the patient’s sense tients with the same conventional di- thermore, concerns have been raised of self-efficacy may be important.54 agnosis, same demographic charac- that observational studies artificially Given many choices, the patient’s teristics, same physical and laboratory inflate effect size.63 Despite these con- preferences for the course and se- findings, and comparable functional cerns, 2 recent systematic reviews64,65 quencing of action guide the ulti- capacity on testing differ markedly in compareddatafromobservationaland mate decision-making process.55 rehabilitation program outcomes such randomized controlled trial studies At a practical level, multiarm as level of perceived pain and per- across different medical conditions. studies that mix modalities are logis- ceived interference with quality of life, It was discovered that observational tically difficult and costly. Neverthe- on the basis of their concomitant psy- studies gave results similar to those of less, psychiatry researchers have set chosocial/behavioral coping style. He randomized controlled trials.64,65 An- precedents for this type of work in proposes that it is necessary to match other study66 concluded that there are evaluating conventional medications each patient’s psychosocial charac- no systematic biases in observational vs psychotherapy vs combination teristics to a particular type of behav- studies. Thus, not only are observa- therapy.56 Often, but not always, in ioral intervention, in addition to a ba- tional studies justifiable under certain these investigations, combination sic treatment program, to foster higher conditions, they may even have sev- therapy is most effective.56,57 As an ex- levels of rehabilitative outcomes suc- eral advantages over randomized con- ception, in a well-known study of pa- cess in a given medical diagnostic trolled trials, including lower cost, tients with Alzheimer disease, the an- group. In other words, Turk is sug- greatertimeliness,andabroaderrange tioxidant vitamin E alone was more gesting a need for subtyping within of patients.67 Well-validated methods effective in slowing disease progres- conventional diagnoses. like quasi-experimentation should sion than was the combination of vi- This approach may pertain even also be used more frequently because tamin E with the conventional drug beforeapplyingthediagnosticsubtyp- they allow an understanding of how selegiline.58 ing of CAM treatment systems to the our health care interventional world Furthermore, more treat- sameindividuals.Previousstudiessug- operates in actual practice.45 ments may not be better if the cat- gest that pain conditions are among egories of the multiple interven- the leading problems for which pa- Outcome Measures tions overlap excessively (ie, diet tients seek CAM treatments.5,61 How- plus several modalities from the psy- ever, complementary and alternative Systems theory advocates the mea- chosocial realm rather than, for ex- medicine is a large umbrella term cov- surement of not only disease- ample, a package of biochemical/ ering a diverse range of self-help and specific conventional type outcomes dietary, physical manipulation, practitioner-administeredtechniques. such as blood levels of biomarkers (eg, psychological, and energy medi- Integrative medicine outcomes re- glycosylated hemoglobin or prostate- cine options).37 For instance, pedi- searchers may need to meld the sub- specific antigen), cardiac stress tests, atric researchers performed a ran- typing approaches both from the con- or pulmonary function tests, but also domized, 4-arm study of recurrent ventional world (ie, psychosocial/ psychological, social, and spiritual abdominal pain treatment in chil- behavioral coping styles) and from outcomes in the same study. A chal- dren, including (1) a fiber diet only, CAM systems to optimize improve- lenging clinical question to be asked (2) fiber diet plus biofeedback, (3) ments for the largest numbers of is: What should we measure for in- fiber diet plus biofeedback and cog- patients. tegrative quality of life? Most ge-

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 138

©2002 American Medical Association. All rights reserved. neric quality of life outcome tools fo- In other words, the nature of medi- In the 25 years since Engel75 pub- cus on “health-related” aspects (ie, cal research itself can and must ex- lished his seminal article on the bio- forced-choice items on specific types pand beyond the prevailing reduc- psychosocialmodelformedicine,afew of physical, emotional, and social tionist approaches and quantitative theoreticians have tried to point out functioning that may or may not re- study designs to measure the sys- the relevance of dynamic systems flect the functions on which the in- temic effectiveness of integrative theory, chaos, and complexity theory dividual patient places value).33,68 medical practice. for conventional medicine, psychol- These tools generally separate the in- ogy,andCAM.76-79 However,medicine dividual’s values and sense of pur- Outcome Analysis asafieldhasnotyetincorporatedthese pose and meaning in life from ideas on a wide scale. Although clini- “health,” and they overlook the in- Although it is beyond the scope of this cal researchers have dissected down- trusions of the treatment into every- article to describe the specifics of ad- stream effects of change at a higher- day living and its enjoyment within vanced statistical techniques in quan- order level on lower-order levels of a a given individual’s values. What titative research, we advocate the use system (eg, the effects of prayer on dis- would happen if the meaning of the of state-of-the-art, user-friendly ad- tanthealing80 orofpsychosocialstress- problem and impact of the treat- vanced methods of data analysis that ors on neuroimmune markers81; Fig- ment on daily life mattered not only enable investigators to test a CAM sys- ure 2A), they generally have not de- to the individual, but also as a re- tem as a whole, and within its own signed studies to examine the patient search outcome, with weight given in context (Figure 2B). Path analysis (an asacomplex,interactive,dynamicsys- similar proportion to prolongation of extension of multiple regression), temwithinthelargersystemasawhole life and improvement of functional structural equation modeling (analy- (Figure2B).Itisthechallengeofhealth status? Does optimization of well- sis that includes latent variables), and outcomes research to prove or dis- ness, not simply absence of disease, confirmatory factor analysis (a sys- prove the relevance of this integrative, fall within the realm of health- tematic analysis of the pattern of re- systemic worldview to the field of related outcomes? Should all poten- lationships among variables that at- medicine and to test the feasibility of tial outcomes receive equal weight?33 tempts to explain that pattern in terms its emergence as a practical and desir- Are “objective” measures more valu- of a smaller number of underlying hy- able way to provide clinical care. able in this research than are subjec- pothetical factors) represent only a tive measures? For example, geriat- few of the examples applicable to this Accepted for publication May 9, 2001. ric patients’ self-assessments of their type of health care outcomes re- This research was supported in global health status have been shown search.74 These techniques allow us part by grant P50HL61212 from the to be superior to physician assess- to look at the complex relationships National Institutes of Health Pediat- ments and laboratory tests as a pre- among many dependent and inde- ric Center for Complementary and dictor of mortality.69,70 pendent variables at the same time, Alternative Medicine at the Univer- To measure these variables, we consistent with the higher level of or- sity of Arizona, Tucson (Dr Weil), and needtodevelopmultidimensionalout- ganization in a complex systems by Mid-Career Investigator Award comemeasuresthatcanaccommodate theory model.36,37 Furthermore, by K24AT00057 from the National Insti- thistypeofintegrative,patient-oriented specifying paths by which we think tutes of Health, Bethesda, Md (Dr Bell). clinical research. It may also mean that specific variables affect others, we can Corresponding author: Iris R. qualitative research techniques will be strengthen our confidence about Bell, MD, MD(H), PhD, Program in important for developing a greater un- causal inferences.74 Integrative Medicine, The University derstanding of what matters to the pa- of Arizona Health Sciences Center, tient as an individual.71,72 Even assum- CONCLUSIONS AND 1501 N Campbell Ave, Tucson, AZ ing that the patient’s values are priori- IMPLICATIONS 85724 (e-mail: [email protected]). tized over those of other stakeholders, a great deal of initial research must be In conclusion, we propose that com- REFERENCES done to clarify the nature of those in- bination medicine (conventional dividualized values. plus CAM) is not integrative medi- 1. Mish FC, ed in chief. Merriam Webster’s Colle- For research on personal mean- cine. Integrative medicine is a sys- giate Dictionary. 10th ed. Springfield, Mass: Mer- riam-Webster Inc; 1993. ing to the individual, qualitative re- tem of care that considers health (or 2. Zollman C, Vickers A. What is complementary search approaches, with their empha- disease) as an emergent property of medicine? BMJ. 1999;319:693-696. sis on interpretative methods, become the person in an environmental con- 3. Vincent C, Furnham A. Why do patients turn to complementary medicine? an empirical study. Br more useful than quantitative stud- text, conceptualized as an intact, in- J Clin Psychol. 1996;35:37-48. ies.72 Qualitative research, for ex- divisible dynamic system.19,20 Inte- 4. Burdi MD, Baker LC. Physicians’ perceptions of ample, revealed a definition of spiri- grative medicine is a complex, autonomy and satisfaction in California. Health Af- fairs (Millwood). 1999;18:134-145. tual well-being in elderly persons dynamic, higher-order system of sys- 5. Astin JA. Why patients use alternative medicine. whose conventional health ranged tems, conventional and CAM. As JAMA. 1998;279:1548-1553. 6. Gaudet TW. Integrative medicine. Integr Med. from good to terminally ill (ie, “a such, the life domains that medical 1998;1:67-73. sense of harmonious interconnected- care and medical outcomes re- 7. Maizes V, Caspi O. The principles and challenges ness between self, others, nature, and search must address extend far be- of integrative medicine. West J Med. 1999;171: 148-149. Ultimate Other, which exists through- yond clinical laboratory test results 8. World Health Organization. About WHO: Defini- out and beyond time and space”73). or lesions in specific organs. tion of Health. Available at: http://www.who.int

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 139

©2002 American Medical Association. All rights reserved. /aboutwho/en/definition.html. Accessed October 35. Darley JM, Gross PH. A hypothesis-confirming trial of selegiline, alpha-tocopherol, or both as 9, 2001. bias in labeling effects. J Pers Soc Psychol. 1983; treatment for Alzheimer’s disease. N Engl J Med. 9. Burns R, Nichols LO, Martindale-Adams J, Graney 44:20-33. 1997;336:1216-1222. MJ. Interdisciplinary geriatric primary care evalu- 36. Schwartz GE, Russek LG. The challenge of one 59. Humphreys PA, Gevirtz RN. Treatment of recur- ation and management: two-year outcomes. JAm medicine: theories of health and eight “world hy- rent abdominal pain: components analysis of four Geriatr Soc. 2000;48:8-13. potheses.” Adv Mind Body Med. 1997;13:7-23. treatment protocols. J Pediatr Gastroenterol Nutr. 10. Bretscher M, Rummans T, Sloan J, et al. Quality 37. Bell IR, Baldwin CM, Schwartz GER, Russek LGS. 2000;31:47-51. of life in hospice patients: a pilot study. Psycho- Integrating belief systems and therapies in medi- 60. Turk DC. Customizing treatment for chronic pa- somatics. 1999;40:309-313. cine. Integr Med. 1998;1:95-105. tients: who, what, and why. Clin J Pain. 1990;6: 11. Merboth MK, Barnason S. Managing pain. Nurs 38. Bell IR, Walsh M, Russek LGS, Schwartz GER. Pro- 255-270. Clin North Am. 2000;35:375-383. posed applications of conventional research con- 61. Pocock SJ, Elbourne DR. Randomized trials of ob- 12. Rydman RJ, Trybus D, Butki N, Kampe LM, Mar- cepts and tools to homeopathic clinical research. servational tribulations? N Engl J Med. 2000;342: ley JA. Outcome of case management and com- J Am Inst Homeopath. 1999;92:111-128. 1907-1909. prehensive support services following policy 39. Coveney P, Highfield R. Frontiers of Complexity: 62. Eisenberg DM, Davis RB, Ettner SL, et al. Trends changes in mental health care delivery. J Med Syst. The Search for Order in a Chaotic World. New York, in alternative medicine use in the United States, 1999;23:309-323. NY: Ballantine Publishing Group; 1995. 1990-1997: results of a follow-up national sur- 13. Thiels C, Schmidt U, Treasure J, Garthe R, Troop 40. Gallagher R, Appenzeller T. Beyond reduction- vey. JAMA. 1998;280:1569-1575. N. Guided self-change for bulimia nervosa incor- ism. Science. 1999;284:79. 63. Chalmers TC, Celano P, Sacks HS, Smith H Jr. Bias porating use of a self-care manual. Am J Psychia- 41. Goldberger AL. Non-linear dynamics for clini- in treatment assignment in controlled clinical tri- try. 1998;155:947-953. cians: chaos theory, fractals, and complexity at als. N Engl J Med. 1983;309:1358-1361. 14. Marvel MK, Doherty WJ, Weiner E. Medical inter- the bedside. Lancet. 1996;347:1312-1314. 64. Concato J, Shah N, Horwitz RI. Randomized, con- viewing by exemplary family physicians. J Fam 42. MacLean D, MacIntosh R. Health and disease in trolled trials, observational studies, and the hier- Pract. 1998;47:343-348. organizations. J Altern Complement Med. 1998; archy of research designs. N Engl J Med. 2000; 15. Ferguson RS. Preventive care in daily practice. Okla 4:185-188. 342:1887-1892. State Med Assoc. 2000;93:154-160. 43. Donabedian A. Evaluating the quality of medical 65. Benson K, Hartz AJ. A comparison of observa- 16. Anderson RM, Funnell MM, Fitzgerald JT, Mar- care. Milbank Memorial Fund Q. 1966;44:166- tional studies and randomized, controlled trials. rero DG. The Diabetes Empowerment Scale. Dia- 206. N Engl J Med. 2000;342:1878-1886. betes Care. 2000;23:739-743. 44. Figueredo AJ, Sechrest L. Approaches used in con- 66. Britton A, McPherson K, Sanderson C, Black N, 17. Lorig K, Seleznick M, Lubeck D, Ung E, Chastain ducting health outcomes and effectiveness re- Bain C. Choosing between randomized and non- RL, Holman HR. The beneficial outcomes of the search. Eval Program Planning. 2001;24:41-59. randomized studies: a systematic review. Health arthritis self-management course are not ad- 45. Cook TD, Campbell DT. Quasi-Experimentation: Technol Assess. 1998;2:1-124. equately explained by behavior change. Arthritis Design and Analysis for Field Settings. Boston, 67. Feinstein AR. Epidemiologic analyses of causa- Rheum. 1989;32:91-95. Mass: Houghton Mifflin Co; 1979. tion: the unlearned scientific lessons of random- 18. Jonas WB, Levin JS. Essentials of Complemen- 46. National Center for Complementary and Alterna- ized trials. J Clin Epidemiol. 1989;42:481-489. tary and Alternative Medicine. Philadelphia, Pa: Lip- tive Medicine, National Institutes of Health. Ma- 68. Ware JE, Sherbourne CD. The MOS 36-item short- pincott Williams & Wilkins; 1999. jor Domains of Complementary and Alternative form health survey (SF-36). Med Care. 1992;30: 19. Fulder S. The Handbook of Alternative and Comple- Medicine. Available at: http://nccam.nih.gov 473-483. mentary Medicine. 3rd ed. Oxford, England: Ox- /nccam/fcp/classify/. Accessed October 10, 2001. 69. Idler E. Self-assessed health and mortality: a re- ford University Press; 1996. 47. Schnyer R, Allen J. Acupuncture in the Treat- view of studies. Int Rev Health Psychol. 1992;1: 20. Ernst E, Hahn EG, eds. Homoeopathy: A Critical ment of Depression: A Manual for Research and 33-54. Appraisal. Oxford, England: Butterworth Heine- Practice. London, England: Churchill Living- 70. Long MJ, Marshall BS. The relationship between mann; 1998. stone; 2001. self-assessed health status, mortality, service use, 21. Ernst E, White A, eds. Acupuncture: A Scientific 48. Caspi O, Bell IR, Rychener D, Gaudet TW, Weil AT. and cost in a managed care setting. Health Care Appraisal. Oxford, England: Butterworth Heine- The Tower of Babel: communication and medi- Manage Rev. 1999;24:20-27. mann; 1998. cine. Arch Intern Med. 2000;160:3193-3195. 71. Taylor SJ, Bogdan R. Introduction to Qualitative 22. Ezzo J, Berman B, Hadhazy VA, Jadad AR, Lao L, 49. Vickers A, Cassileth B, Ernst E, et al. How should Singh BB. Is acupuncture effective for the treat- Research Methods: The Search for Meanings. 2nd ment of chronic pain? a systematic review. Pain. we research unconventional therapies? Int J Tech- ed. New York, NY: John Wiley & Sons Inc; 1984. 2000;86:217-225. nol Assess Health Care. 1997;13:111-121. 72. Pope C, Mays N, eds. Qualitative Research in 23. Lundeberg T. Acupuncture in headache. Cepha- 50. Relman AS, Weil A. Is integrative medicine the Health Care. London, England: BMJ Books; 2000. lalgia. 1999;19(suppl 25):65-68. medicine of the future? a debate. Presented at: Uni- 73. Hungelmann J, Kenkel-Rossi E, Klassen L, Stol- 24. Ter Riet G, Kleijnen J, Knipshild P. Acupuncture versity of Arizona; April 9, 1999; Tucson, Ariz. [A lenwerk R. Focus on spiritual well-being: harmo- and chronic pain: a criteria-based meta-analysis. short version of the debate can be found in Dalen nious interconnectedness of mind-body-spirit: use J Clin Epidemiol. 1990;43:1191-1199. JE. Is integrative medicine the medicine of the fu- of the JAREL Spiritual Well-Being Scale. Geriatr 25. Ernst E, White AR. Acupuncture for back pain. Arch ture? a debate between Arnold S. Relman, MD, Nurs. 1996;17:262-266. Intern Med. 1998;158:2235-2241. and Andrew Weil, MD. Arch Intern Med. 1999; 74. Norman GR, Streiner DL. Principal components 26. Hammerschlag R, Morris MM. Clinical trials com- 159:2122-2126.] and factor analysis, path analysis and structural paring acupuncture with biomedical standard care. 51. Little P, Everitt H, Williamson I, et al. Prefer- equation modeling. In: Biostatistics: The Bare Es- Complement Ther Med. 1997;5:133-140. ences of patients for patient-centred approach to sentials. Hamilton, Ontario: BC Decker Inc; 2000: 27. Vincent C, Furnham A. Research methods and re- consultation in primary care: observational study. 163-200. search problems in complementary medicine. In: BMJ. 2001;322:468-472. 75. Engel GL. The need for a new medical model: a Complement Medicine: A Research Perspective. 52. Tabachnick BG, Fidell L. Using Multivariate Sta- challenge for biomedicine. Science. 1977;196: West Sussex, England: Wiley & Sons Ltd; 1997: tistics. New York, NY: Harper & Row; 1983. 129-136. 148-174. 53. Green L. On studying effectiveness. J Fam Pract. 76. Butz MR. Chaos and Complexity: Implications for 28. Sampson W. On being a critic. Sci Rev Altern Med. 1999;48:423-424. Psychological Theory and Practice. Washington, 1998;2:4-5. 54. Mitchell A, Cormack M. The Therapeutic Relation- DC: Taylor & Francis; 1997. 29. Coates JR, Jobst KA. Integrated health care. J Altern ship in Complementary Health Care. London, En- 77. Mandell AJ, Selz KA. Nonlinear dynamical pat- Complement Med. 1998;4:209-247. gland: Churchill Livingstone; 1998. terns as personality theory for neurobiology and 30. Brownson RC, Simoes EJ. Measuring the impact 55. McAlister FA, O’Connor AM, Wells G, Grover SA, psychiatry. Psychiatry. 1995;58:371-390. of prevention research on public health practice. Laupacis A. When should hypertension be treated? 78. Coffey DS. Self-organization, complexity, and Am J Prev Med. 1999;16(suppl 3):72-79. the different perspectives of Canadian family phy- chaos: the new biology for medicine. Nat Med. 31. Levin JS, Glass TA, Kuski LH, Schuck JR, Steele sicians and patients. CMAJ. 2000;163:403-408. 1998;4:882-885. L, Jonas WB. Quantitative methods in research on 56. Reynolds CF, Frank E, Perel JM, et al. Nortripty- 79. Shepperd J. Chaos theory: implications for ho- complementary and alternative medicine. Med line and interpersonal psychotherapy as mainte- meopathy. J Am Inst Homeopath. 1994;87:22. Care. 1997;35:1079-1094. nance therapies for recurrent major depression: 80. Harris WS, Gowda M, Kolb JW, et al. A random- 32. Angell M, Kassirer J. Alternative medicine. N Engl a randomized controlled trial in patients older than ized, controlled trial of the effects of remote, in- J Med. 1998;339:839-841. 59 years. JAMA. 1999;281:39-45. tercessory prayer on outcomes in patients admit- 33. Kane RL. Understanding Health Care Outcomes 57. Mynors-Wallis LM, Gath DH, Day A, Baker F. Ran- ted to the coronary care unit. Arch Intern Med. Research. Gaigthersburg, Md: Aspen Publishers domised, controlled trial of problem-solving treat- 1999;159:2273-2278. Inc; 1997. ment, antidepressant medication, and combined 81. Kiecolt-Glaser JK, Glaser R, Cacioppo JT, Malar- 34. Sechrest L, Stickle T, Stewart M. Effective and Per- treatment for major depression in primary care. key WB. Marital stress: immunologic, neuroen- suasive Case Studies. Cambridge, Mass: Human BMJ. 2000;320:26-30. docrine, and autonomic correlates. Ann N Y Acad Services Research Institute; 1996. 58. Sano M, Ernesto C, Thomas R, et al. A controlled Sci. 1998;840:656-663.

(REPRINTED) ARCH INTERN MED/ VOL 162, JAN 28, 2002 WWW.ARCHINTERNMED.COM 140