Original Contributions Trends in Alternative Use in the United States, 1990-1997 Results of a Follow-up National Survey

David M. Eisenberg, MD; Roger B. Davis, ScD; Susan L. Ettner, PhD; Scott Appel, MS; Sonja Wilkey; Maria Van Rompay; Ronald C. Kessler, PhD

Context.—A prior national survey documented the high prevalence and costs of native medicine prevalence, costs, and use in the United States in 1990. patterns of use1 demonstrated that al- Objective.—To document trends in alternative medicine use in the United States ternative medicine has a substantial between 1990 and 1997. presence in the US health care system. 2 Design.—Nationally representative random household telephone surveys using Data from a survey in 1994 and a public opinion poll in 19973 confirmed the ex- comparable key questions were conducted in 1991 and 1997 measuring utilization tensive use of alternative medical thera- in 1990 and 1997, respectively. pies in the United States. An increasing Participants.—A total of 1539 adults in 1991 and 2055 in 1997. number of US insurers and managed Main Outcomes Measures.—Prevalence, estimated costs, and disclosure of care organizations now offer alternative alternative to physicians. medicine programs and benefits.4 The Results.—Use of at least 1 of 16 alternative therapies during the previous year majority of US medical schools now of- increased from 33.8% in 1990 to 42.1% in 1997 (PՅ.001). The therapies increas- fer courses on alternative medicine.5 ing the most included , , megavitamins, self-help groups, National surveys performed outside folk remedies, energy healing, and homeopathy. The probability of users visiting an the United States suggest that alterna- alternative medicine practitioner increased from 36.3% to 46.3% (P = .002). In both tive medicine is popular throughout the industrialized world.6 The percentage of surveys alternative therapies were used most frequently for chronic conditions, in- the population who used alternative cluding back problems, anxiety, depression, and headaches. There was no signifi- therapies during the prior 12 months has cant change in disclosure rates between the 2 survey years; 39.8% of alternative been estimated to be 10% in Denmark therapies were disclosed to physicians in 1990 vs 38.5% in 1997. The percentage (1987),7 33% in Finland (1982),8 and 49% of users paying entirely out-of-pocket for services provided by alternative medicine in Australia (1993).9 Public opinion polls practitioners did not change significantly between 1990 (64.0%) and 1997 (58.3%) and consumers’ association surveys sug- (P = .36). Extrapolations to the US population suggest a 47.3% increase in total gest high prevalence rates throughout visits to alternative medicine practitioners, from 427 million in 1990 to 629 million Europe and the United Kingdom.10-13 in 1997, thereby exceeding total visits to all US primary care physicians. An esti- The percentage of the Canadian popula- mated 15 million adults in 1997 took prescription medications concurrently with tionwhosawanalternativetherapyprac- titioner during the previous 12 months herbal remedies and/or high-dose vitamins (18.4% of all prescription users). Esti- has been estimated at 15% (1995).14 The mated expenditures for alternative medicine professional services increased 45.2% wide range of utilization rates can be between 1990 and 1997 and were conservatively estimated at $21.2 billion in 1997, explained, in part, by the disparity in with at least $12.2 billion paid out-of-pocket. This exceeds the 1997 out-of-pocket definitions of alternative and the expenditures for all US hospitalizations. Total 1997 out-of-pocket expenditures re- selection of therapies assessed. lating to alternative therapies were conservatively estimated at $27.0 billion, which The presumption is that alternative is comparable with the projected 1997 out-of-pocket expenditures for all US phy- medicine use in the United States has sician services. increased at a considerable pace in re- Conclusions.—Alternative medicine use and expenditures increased substan- cent years. The purpose of this follow-up tially between 1990 and 1997, attributable primarily to an increase in the proportion national survey was to investigate this presumption and document trends in al- of the population seeking alternative therapies, rather than increased visits per ternative medicine prevalence, costs, patient. disclosure of use to physicians, and cor- JAMA. 1998;280:1569-1575 relates of use since 1990.

From the Center for Alternative Medicine Research ALTERNATIVE medical therapies, METHODS and Education, Department of Medicine, Beth Israel functionally defined as interventions Deaconess Medical Center (Drs Eisenberg and Davis, neither taught widely in medical schools Sample Mr Appel, and Mss Wilkey and Van Rompay), and the 1 Department of Health Care Policy, Harvard Medical nor generally available in US hospitals, We conducted parallel nationally rep- School (Drs Ettner and Kessler), Boston, Mass. have attracted increased national atten- resentative telephone surveys in 1991 Reprints: David M. Eisenberg, MD, Center for Alter- tion from the media, the medical com- and 1997. Survey methods were ap- native Medicine Research and Education, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, munity, governmental agencies, and the proved by the Beth Israel Deaconess In- MA 02215. public. A 1990 national survey of alter- stitutional Review Board, Boston, Mass.

JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al 1569 ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 Both surveys used random-digit dialing conversion rate to all of the refusers and scribed by a doctor. Prayer or spiritual to select households and random selec- weighting the data for the undersam- healing by others was asked about sepa- tionof1householdresident,aged18years pling of initial refusers, we obtained a 60% rately from prayer or spiritual practice or older, as the respondent. Eligibility (41.3% + [31.4% × (100% − 41.3%)]) for individual health concern. Commer- was limited to English speakers in whom weighted overall response rate among eli- cial diet programs were described as cognitive or physical impairment did not gible respondents. “the kind you have to pay for, but not prevent completion of the interview. We including trying to lose or gain weight on asked respondents about their use of al- Interview your own.” A lifestyle diet included ex- ternative therapies during the prior 12 In both years, the interview was pre- amples like vegetarianism or macrobiot- months. We consider the results of the sented as a survey conducted about the ics. Questions regarding energy healing 1991 survey, fielded between January health care practices of Americans by included examples of magnets, energy- and March of that year, representative of investigators from Harvard Medical emitting machines, or the “laying on of 1990, and the results of the 1997 survey, School. No mention was made of alter- hands,” and use of relaxation techniques fielded between November 1997 and native or complementary therapies. The was explained using the examples of February 1998, representative of 1997. substantive questions began by asking meditation or the relaxation response. The sampling scheme was designed about perceived health, health worries, The remaining 9 therapies were asked with a target sample of 1500 in 1990 and days spent in bed, and functional impair- about without interviewer clarification. 2000 in 1997. The latter sample size was ment due to health problems. We then The 1997 survey was longer (average, chosentoprovidepowerinexcessof80% asked respondents about their interac- 30 minutes) than the 1990 survey (aver- to detect an increase from 34% to 39% in tions with a medical doctor, defined as “a age, 25 minutes) because we sought to theproportionofadultswhousedatleast medical doctor (MD) or a doctor of os- explore a number of areas in more depth. 1 form of alternative therapy during the teopathic medicine (DO), not a chiro- All the important questions in the 1990 prior 12 months. The actual numbers of practororothernonmedicaldoctor.”The survey were repeated in 1997. These completed interviews were 1539 in 1990 term medical doctor was used through- replicated questions are the focus of the (67% response rate) and 2055 in 1997 out the remainder of the interview. current report. One major change in the (60% weighted response rate). A secular To document trends we explored the 1997 survey involved replicated ques- trend in lower survey response required following: (1) Respondents in both sur- tions: respondents who reported using us to offer a $20 financial incentive for veys were presented with a list of com- more than 3 alternative therapies were participation in the 1997 survey to main- mon medical conditions and asked if they asked in-depth questions (eg, use of a tain a response rate near the one had experienced each of these conditions practitioner of alternative therapies, achieved in 1990. No financial incentive during the previous 12 months. (2) Re- number of visits, out-of-pocket ex- was used in the 1990 survey. spondents who reported more than 3 penses, reasons for use) for all such The data in each survey were sepa- conditions were asked to identify their 3 therapies in 1990 but only for a random rately weighted to adjust for geographic most bothersome or serious medical con- sample of 3 such therapies in 1997. This variation in cooperation (eg, by region of ditions and were then asked about seeing was required because of expansion in countryandurbanicity)andforhousehold a medical doctor for these principal medi- both the number of alternative thera- variationinprobabilityofselection(ie,the cal conditions and about the perceived pies we assessed in 1997 and questions inverse relationship between size of quality of these interactions. (3) Respon- about each therapy. The 1997 data were household and probability of selection be- dents were asked about their lifetime and weighted to adjust for this sampling in cause only 1 interview was completed in past12-monthuseof16alternativethera- making comparisons with the 1990 data. each sample household). The data were pies and whether each of these therapies then weighted in parallel on sociodemo- was used for each of the principal medical Insurance Coverage graphic variables to adjust for aggregate conditions. The 1997 survey also asked For each therapy for which respon- discrepancies between the sample distri- about use for a representative sample of dents said they used services of an alter- butions and population distributions pro- other medical conditions and expanded native medicine practitioner, we asked vided by the US Census Bureau. This last the list of therapies beyond the original whether insurance helped pay for any of stage of weighting was based on the 1997 16 assessed in 1990. (4) We distinguished the costs of the therapy and whether the Current Population Survey data15 and between use under the supervision of a respondent paid any of the costs out-of- was done in parallel across the 2 surveys practitioner of alternative therapy and pocket. Based on the answers to these to remove any between-survey discrep- use without such supervision. Respon- questions, we calculated the proportion of ancies of weighted sociodemographic dis- dents who reported supervised use were users of each therapy who had complete, tributions. asked about their number of visits in the partial, or no insurance coverage for that Of the initial sample of 9750 telephone past 12 months to practitioners of each therapy. We also calculated the overall fre- numbers in 1997, 26% were nonworking, therapy.(5)Allusersofalternativethera- quency of insurance coverage by weight- 17% were not assigned to households, and pies in 1997 who acknowledged seeing a ing the insurance frequencies within each 9% were unavailable (ie, despite 6 at- medical doctor during the past year were therapy by the proportion of all user thera- tempted follow-up contacts). We de- then asked if they had discussed their use pies accounted for by that therapy. clared 481 households ineligible because ofeachtherapywithamedicaldoctorand, respondents did not speak English or if not, why not. Construction of Cost Measures because of cognitive or physical incapac- Prior use of 16 targeted therapies was The total cost of visits to alternative ity. Among the remaining 4167 eligible explored using a computer-assisted in- medicine practitioners was calculated by respondents, 1720 (41.3%) completed the terview transcript, which included the multiplying the number of visits for each interview on initial request. Attempts following clarifications in both 1990 and therapy by a per-visit price and adding were then made to convert a random sub- 1997: When asking about high-dose vi- the prices of the following therapies: re- sample of 1066 refusers by offering them tamin or megavitamin therapies, inter- laxation techniques, herbal medicine, an increased stipend ($50). A total of 335 viewers made clear that the survey massage therapy, care, (31.4%) of the 1066 contacted were con- sought information on vitamins not in- megavitamins, self-help groups, imag- verted in this manner. Extrapolating this cluding a daily vitamin or vitamin pre- ery techniques, commercial diet, folk rem-

1570 JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 edies, lifestyle diet, energy healing, ho- mates through the years. As the data in Table 1.—Characteristics of the 1997 (N = 2055) meopathy, hypnosis, biofeedback, and both surveys are weighted, the Taylor and 1990 (N = 1539) Subjects Interviewed Com- . Out-of-pocket costs were series method was used to compute sig- pared With the US Population* constructed for each therapy by multi- nificance tests using SUDAAN soft- 1997 US plying each user’s visits by the full price ware.18 ␹2 Tests of independence were 1997 Bureau of the 1990 of the visit if the user had no insurance used for comparing proportions, while t Characteristic Survey, % Census,17 % Survey, % coverage, by 0.2 if the user had partial tests were used for continuous mea- Sex Female 52 52 48 insurance coverage, and by zero if insur- sures. Extrapolations of survey esti- Male 48 48 52 ance paid the full price of the visit. The mates to the total population were based Age, y assumption of a 20% coinsurance rate on the assumption that there were 180 18-24 10 13 16 among users with partial insurance cov- million adults living in the US household 25-34 22 20 23 erage should yield a conservative esti- population in 1990 and 198 million in 35-49 33 32 27 15 mate of out-of-pocket costs, because it ig- 1997. Ն50 35 35 34 nores deductibles and benefit caps and Race/ethnicity assumes that insurance benefits for al- RESULTS White 77 73 82 ternative therapy are similar to medical African American 8 12 9 Characteristics of Respondents coverage. Hispanic 10 11 6 We calculated costs based on per-visit The characteristics of the subjects we Asian 1 4 1 prices chosen from typical prices paid for interviewed are shown in Table 1. The Other 4 1 2 such services by private insurers using a sociodemographic characteristics of the Education Resource-Based Relative Value Scale survey sample are similar to the popu- ϽHigh school 14 18 24 16 (RBRVS) system in selected states. We lation distributions published by the US High school 37 34 35 then recalculated costs using a second set Bureau of the Census.15 graduate of prices chosen partly to reflect empirical College or more 49 48 40 dataontheout-of-pocketcostspaidbythe Patterns of Use Annual income, $ Ͻ respondents, but primarily to represent Use of alternative therapies in 1997 20 000 27 33 30 conservative estimates of the per-visit was not confined to any narrow segment 20 000-49 999 45 41 53 cost of alternative therapies. Total costs of society. Rates of use ranged from 32% Ն50 000 27 26 18 based on this second set of prices should to 54% in the wide range of sociodemo- Region Northeast 21 19 22 represent a lower bound on true expen- graphicgroupsexamined.Usewasmore North central 24 24 32 ditures. common among women (48.9%) than Out-of-pocket costs of herbs, megavi- men (37.8%) (P = .001) and less common South 35 35 26 tamin supplements, and commercial diet among African Americans (33.1%) than West 20 22 19 products were calculated by multiplying members of other racial groups (44.5%) *Due to rounding, percentages do not always total 100. the total population of users by the aver- (P = .004). People aged 35 to 49 years re- age out-of-pocket expenditures reported ported higher rates of use (50.1%) than total number of therapies used, from 577 by respondents who used each of these people either older (39.1%) (P = .001) or therapies per 1000 population in 1990 to products. In 1997, each respondent who younger (41.8%) (P = .003). Use was 953 per 1000 in 1997. used an alternative therapy was also higher among those who had some col- Several categories of alternative asked, “Did you spend any additional lege education (50.6%) than with no col- therapy warrant clarification about the money on things like books, classes, lege education (36.4%) (P = .001) and actual modalities used. Three quarters of equipment, or any other items related to morecommonamongpeoplewithannual respondents who acknowledged use of [the alternative therapy] in the past 12 incomes above $50 000 (48.1%) than with relaxation techniques said they used months?” Out-of-pocket expenditures on lower incomes (42.6%) (P = .03). Use was meditation. Among those who reported these other items were calculated follow- more common among those in the West usingenergyhealing,themostfrequently ing the same procedures used for herbs, (50.1%) than elsewhere in the United cited technique involved the use of mag- megavitamins,andcommercialdietprod- States (42.1%) (P = .004). With the ex- nets. Other modalities common to this ucts. Out-of-pocket expenditures on ception of observed sex differences in category included Therapeutic Touch, herbs, megavitamins, commercial diet 1997, these patterns are consistent with Reiki, and energy healing by religious products, and related items were based those identified in 1990. groups. The use of self-prayer, in contrast on actual dollar amounts reported, so Populationprevalenceestimatesofal- to spiritual or energy healing performed changes between 1990 and 1997 include ternative medicine use in 1990 and 1997 by others, was investigated in terms of inflation. To isolate the increase in the are shown in Table 2. The 1990 survey prevalence of use but not in terms of cost of practitioner visits between 1990 estimated that 33.8% of the US adult costs, referral patterns, or insurance re- and 1997 solely because of the increase in population (60 million people) used at imbursement. All analyses in this article the use of alternative therapies, we cal- least 1 of the 16 alternative therapies exclude data involving self-prayer. culated1990practitionercostsusing1997 listed, while the 1997 survey estimated Table 2 (second column) shows that a prices. The differences between the 1990 that this proportion increased signifi- significantly higher proportion of alter- and 1997 costs of practitioner services re- cantly to 42.1% (83 million people). A native therapy users saw an alternative ported are understated because they do comparison of specific therapies in the medicine practitioner in 1997 (46.3%, not take into account inflation, estimated first column shows increases in 15 of the equivalent to 39 million people) than in at 44% by the medical component of the 16 therapies; 10 of these were statisti- 1990 (36.3%, equivalent to 22 million Consumer Price Index.17 cally significant (P Յ .05). The largest people). Of the 15 therapies for which increases were in the use of herbal medi- the question was asked, the proportion Statistical Analysis cine, massage, megavitamins, self-help of users who saw a practitioner in- Analyses reported herein consist of groups, folk remedies, energy healing, creased for 11. However, even in 1997 computation of prevalence and mean es- andhomeopathy.SummingTable2(first there were only 5 therapies in which timates and comparisons of these esti- column) data shows a 65% increase in a majority of users consulted a practi-

JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al 1571 ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 Table 2.—Comparison of Prevalence and Frequency of Use of Alternative Therapies Among Adult Respondents, 1997 vs 1990*

Saw a Mean No. of Practitioner Visits per Used in Past in Past User in Past No. of Visits per 12 mo, % 12 mo, % 12 mo 1000 Population Estimated Total No. of Visits in 1997 Total Type of Therapy 1997 1990 1997 1990 1997 1990 1997 1990 (in Thousands)† Visits, %‡§ Relaxation techniques 16.3¶ 13.1 15.3 9.0 20.9 18.6 521.2 219.3 103 203 16.4 Herbal medicine 12.1** 2.5 15.1 10.2 2.9 8.1 53.0 20.7 10 491 1.7 Massage 11.1** 6.9 61.6# 41.4 8.4 14.8 574.4 422.8 113 723 18.1 Chiropractic 11.0 10.1 89.9** 71.1 9.8 12.6 969.1¶ 904.8 191 886 30.5 Spiritual healing by others࿣ 7.0# 4.2 . . . 9.2 . . . 14.2 . . . 54.9 ...... Megavitamins 5.5** 2.4 23.7 11.8 8.6 12.6 112.1 35.7 22 196 3.5 Self-help group 4.8** 2.3 44.4 38.3 18.9 20.5 402.8 180.6 79 754 12.7 Imagery 4.5 4.2 23.1 15.1 11.0 14.2 114.3 90.1 22 640 3.6 Commercial diet 4.4 3.9 43.2 24.0 7.3 20.7 138.8 193.8 27 474 4.4 Folk remedies 4.2** 0.2 6.2 0.0 1.0 . . . 2.6 . . . 516 0.1 Lifestyle diet 4.0 3.6 8.0 12.5 2.8 8.1 9.0 36.5 1774 0.3 Energy healing 3.8** 1.3 26.3 32.2 20.2# 8.3 201.9¶ 34.7 39 972 6.4 Homeopathy 3.4** 0.7 16.5 31.7 1.6 6.1 9.0 13.5 1777 0.3 Hypnosis 1.2 0.9 62.7 51.8 2.8 2.6 21.1 12.1 4171 0.7 Biofeedback 1.0 1.0 54.3 20.8 3.6 6.4 19.5 13.3 3871 0.6 Acupuncture 1.0è¶ 0.4 87.6 91.3 3.1 38.4 27.2 140.2 5377 0.9 Ն1 of 16 alternative therapies 42.1** 33.8 46.3# 36.3 16.3 19.2 3176.0 2373.0 628 825 ... SE 1.2 1.4 1.9 2.5 1.8 4.5 378.7 599.7 74 997 . . . Self-prayer࿣ 35.1** 25.2 ......

*Percentages are of those who used that type of therapy. Ellipses indicate data not applicable. †Estimate based on 1997 population estimate of 198 million. ‡Percentage of total visits of the 16 therapies (ie, excluding self-prayer). §Because of rounding, percentages do not total 100. ࿣Respondents who received spiritual healing by others were not asked for details of visits in 1997, nor were those who used self-prayer in either year. ¶PՅ.05; #PՅ.01; **PՅ.001. 47.3% increase in total visits is largely for principal condition(s) (second col-

700 1997 1990 because of increases in visits for relax- umn) occurred for back problems, aller- 628 825 000 ation therapy, massage, chiropractic, gies, arthritis, and digestive problems. 600 self-help, and energy healing. The visits The highest condition-specific rates of to practitioners of alternative therapy in alternative therapy use in 1997 were for 500 1997 exceeded the projected number of neck (57.0%) and back (47.6%) problems. 427 120 000 visits to all primary care physicians in The proportion of respondents with 1 or 400 385 919 000 387 558 000 the United States by an estimated 243 more medical conditions who reported million(Figure1).19,20 Visitstochiroprac- use of an alternative therapy for at least 300 tors and massage therapists accounted 1 of those conditions increased signifi- for nearly half of all visits to practition- cantly from 22.9% in 1990 to 33.7% in 200 ers of alternative therapies. 1997 (PՅ.001). The weighted condition- Total No. of Visits, in Millions Visits, of No. Total Prevalence estimates for selected ad- specific proportion who saw an alterna- 100 ditional therapies assessed in 1997 but not tive medicine practitioner for a given 1990 include: (5.6%), neu- condition also increased significantly 0 Visits to Visits to All Visits to Visits to All ral therapy (1.7%), naturopathy (0.7%), from 6.8% in 1990 to 11.4% in 1997 Practitioners Primary Practitioners Primary and chelation therapy (0.13%) (data not (PՅ.001). of Alternative Care of Alternative Care Therapies Physicians 20 Therapies Physicians 19 shown). Comparisons of total visits and Table 3 also summarizes the probabil- costs for 1990 and 1997 were performed ity that individuals who saw a medical Figure 1.—Trends in annual visits to practitioners of without inclusion of these data. Preva- doctor for a particular condition also used alternative therapies vs visits to primary care phy- lence estimates for the simultaneous use an alternative therapy (fourth column) or sicians, United States, 1997 vs 1990. Data are from of prescription medications with herbs, also saw a practitioner of alternative the National Ambulatory Medical Care Survey from 199620 and 1990.19 with high-dose vitamins, or with both were therapy (fifth column) for that same con- obtained. Among the 44% of adults who dition during the same year. A generally tioner: massage, chiropractic, hypnosis, said they regularly take prescription medi- increasing pattern of alternative medi- biofeedback, and acupuncture. Unsuper- cations, nearly 1 (18.4%) in 5 reported the cine use can be seen across the range of viseduse(ie,aformofexpandedself-care) concurrent use of at least 1 herbal prod- conditions studied. In 1990, an estimated remains the usual method of use for all uct, a high-dose vitamin, or both. 1 (19.9%) in 5 individuals seeing a medical other alternative therapies. Table 3 summarizes results regarding doctor for a principal condition also used Table 2 (third column) reveals no con- use of alternative therapies for the most an alternative therapy. This percentage sistent change in the average number of commonly reported principal medical increased to nearly 1 (31.8%) in 3 in 1997 visitsamongrespondentswhoconsulted conditions in either survey. In each year, (PՅ.001). The percentage who saw a practitioners of alternative therapy be- a majority of respondents reported 1 or medical doctor and also sought the ser- tween 1990 (19.2%) and 1997 (16.3%). more principal medical conditions. The vices of an alternative practitioner in- However, because of the increase in the list of conditions was expanded in 1997 creased significantly from 8.3% in 1990 to proportion of people using these thera- (37 conditions) compared with 1990 (24 13.7% in 1997 (PՅ.01). In both 1990 and pies, the total number of visits increased conditions). Significant increases in the 1997, chiropractic, relaxation techniques, substantially from 1990 to 1997. This proportion using alternative therapies and massage therapy were among the al-

1572 JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 Table 3.—Comparison of Use of Alternative Therapies for the Most Frequently Reported Principal Medical Conditions, 1997 vs 1990

Saw Medical Saw Medical Used Doctor and Doctor and Alternative Saw Alternative Used Alternative Alternative Percentage Therapy for Practitioner for Therapy Practitioner for Reporting Condition in Condition in for Condition in Condition in Condition Past 12 mo, % Past 12 mo, % Past 12 mo, % Past 12 mo, % Therapies Most Condition 1997 1990 1997 1990 1997 1990 1997 1990 1997 1990 Commonly Used in 1997 Back problems 24.0# 19.9 47.6# 35.9 30.1# 19.5 58.8** 36.1 39.1# 23.0 Chiropractic, massage Allergies 20.7# 16.0 16.6# 8.7 4.2 3.3 28.0¶ 15.7 6.4 5.0 Herbal, relaxation Fatigue* 16.7 . . . 27.0 . . . 6.3 . . . 51.6 . . . 13.1 . . . Relaxation, massage Arthritis 16.6 15.9 26.7¶ 17.5 10.0 7.6 38.5¶ 23.8 15.9 13.8 Relaxation, chiropractic Headaches 12.9 13.2 32.2 26.5 13.3¶ 6.3 42.0 31.8 20.0 12.1 Relaxation, chiropractic Neck problems* 12.1 . . . 57.0 . . . 37.5 . . . 66.6 . . . 47.5 . . . Chiropractic, massage High blood pressure 10.9 11.0 11.7 11.0 0.9 2.9 11.9 11.6 1.1 3.5 Megavitamins, relaxation Sprains or strains 10.8 13.4 23.6 22.3 10.3 9.6 29.4 24.7 15.9 13.6 Chiropractic, relaxation Insomnia 9.3# 13.6 26.4 20.4 7.6 4.0 48.4 19.8 13.3 10.9 Relaxation, herbal Lung problems 8.7 7.3 13.2 8.8 2.5 0.5 17.9 11.1 3.4 0.6 Relaxation, spiritual healing, herbal Skin problems 8.6 8.0 6.7 6.0 2.2 1.6 6.8 6.9 0.0 2.5 Imagery, energy healing Digestive problems 8.2 10.1 27.3# 13.2 9.7¶ 3.6 34.1¶ 15.3 10.7 5.8 Relaxation, herbal Depression† 5.6 8.4 40.9 20.2 15.6 7.0 40.9 35.2 26.9 14.0 Relaxation, spiritual healing Anxiety‡ 5.5 9.5 42.7 27.9 11.6 6.5 42.7 45.4 21.0 10.4 Relaxation, spiritual healing Weighted average across ...... 28.2** 19.1 11.4** 6.8 31.8** 19.9 13.7# 8.3 . . . all conditions§ People with Ն1 condition࿣ 77.8¶ 81.5 33.7** 22.9 15.3** 6.9 ......

*Not included as a separate question in 1990 survey. Ellipses indicate data not applicable. †The 1997 question asked about severe depression, which is not directly comparable with the 1990 question that asked about depression. ‡The 1997 question asked about anxiety attacks, which is not directly comparable with the 1990 question that asked about anxiety. §The weighted averages are calculated based on all 37 conditions studied in 1997 and all 24 conditions studied in 1990, ie, condition is unit of analysis. ࿣This row shows percentage of respondents who reported 1 or more principal medical conditions, along with the percentage of these respondents who reported use of therapy or practitioners for at least 1 of these conditions, ie, person is the unit of analysis. ¶PՅ.05; #PՅ.01; **PϽ.001. ternative therapies used most commonly Using conservative assumptions about pocket payments to alternative medicine to treat principal medical conditions. the fees charged by practitioners of practitioners for the 15 therapies studied. As in 1990, 96% of 1997 respondents who alternative therapies and assuming no Adding the estimates of $5.1 billion for saw a practitioner of alternative therapy changes in visit prices, Americans spent herbal therapies, $3.3 billion for megavi- for a principal condition also saw a medi- anestimated$14.6billiononvisitstothese tamins, $1.7 billion for diet products, and cal doctor during the prior 12 months, and practitioners in 1990 and $21.2 billion in $4.7 billion on alternative therapy–spe- only a minority of alternative therapies 1997 (Table 5). Using less conservative cific books, classes, and equipment, the used were discussed with a medical doc- (RVRBS) price figures, the amount spent totalout-of-pocketexpendituresforalter- tor. Among the 618 respondents in 1997 on services of practitioners of alternative native medicine are conservatively esti- who used 1 or more alternative therapies therapies was estimated at $22.6 billion in mated to be $27.0 billion. Using the aver- and had a medical doctor, only 377 (38.5%) 1990 and $32.7 billion in 1997. Regardless age per-visit prices derived from an of the 979 therapies used were discussed of which set of prices is used, total expen- RBRVS system16 rather than our conser- with the respondent’s medical doctor. This ditures for practitioners of alternative vative estimates (Table 5), the estimated is not significantly different from the 353 therapiesareestimatedtohaveincreased total out-of-pocket expense is approxi- (39.8%) of the 886 therapies discussed by by approximately 45% between 1990 and mately $34.4 billion, which is comparable the comparable group of respondents 1997 exclusive of inflation. with the projected 1997 out-of-pocket ex- (n = 501) in the 1990 survey. Given that Estimated out-of-pocket expenditures penditures for all physician services.21 most alternative therapy is used without for high-dose vitamins increased from These estimates exclude out-of-pocket the supervision of an alternative practi- $0.9 billion in 1990 to $3.3 billion in 1997. expenditures associated with therapies tioner, a substantial portion of alterna- Smallerincreaseswereobservedforcom- unique to the 1997 survey (eg, naturopa- tive therapy use for principal medical con- mercial diet products ($1.3 billion vs $1.7 thy, aromatherapy, neural therapy, and ditions (46.0% in 1997 and 51.3% in 1990) billion). Unlike the 1990 survey, the 1997 chelation therapy). was done without input from either a survey included questions about expen- medical doctor or practitioner of alterna- ditures for herbal products ($5.1 billion) COMMENT tive therapy. and respondents’ alternative therapy– The results of our study are limited by specific books, classes, or equipment the restriction of the sampling frame to Payment for Alternative Therapy ($4.7 billion). people who speak English and have tele- Data on insurance coverage of expen- The estimated total out-of-pocket phonesandbythelowresponserate.The ditures for alternative therapy services component of the alternative medicine decrease in overall response rate from are shown in Table 4. The majority of market in 1997 is shown in Figure 2. Pro- 67% in 1990 to 60% in 1997 is consistent people who saw alternative therapy jected out-of-pocket expenditures for all with secular trends for US telephone in- practitioners paid all the costs out-of- hospitalizations in 1997 in the United terviews in recent years.22 It is difficult pocket in both 1990 (64.0%) and 1997 Statestotaled$9.1billion,whileprojected toknowwhat,ifany,biaswasintroduced (58.3%). None of the changes in insur- out-of-pocket expenses for all US physi- or whether trend estimates are biased ance coverage between 1990 and 1997 cian services in the same year were $29.3 by the fact that financial incentives were were statistically significant, probably billion.21 This compares to a conserva- used in 1997 but not 1990. Furthermore, due in part to small sample sizes. tively estimated $12.2 billion in out-of- we have no data on the accuracy of self-

JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al 1573 ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 Table 4.—Insurance Coverage of Alternative Medicine Services in the United States, 1997 vs 1990*

Percentage of Users of Services Herbal Products Diet Products High-Dose Vitamins Professional Services Therapy-Specific Books, Classes, Etc Coverage, 1997 Coverage, 1990 40 Type of Therapy Complete Partial None Complete Partial None 35 $34.4 Relaxation techniques 28.8 6.6 64.7 5.3 25.9 68.7 $29.3 Herbal medicine 8.6 11.2 80.2 30.7 15.5 53.8 5.1 30 Massage 11.8 16.7 71.5 19.1 18.3 62.6 3.3 $27.0 Chiropractic 17.6 38.1 44.3 11.5 32.8 55.9 25 4.7 5.1 Spiritual healing by others† ...... 0.0 0.0 100.0 Megavitamins 2.7 53.3 44.0 0.0 100.0 0.0 20 1.7 3.3

Self-help group 11.7 36.9 51.5 2.8 17.4 79.8 4.7 Imagery 51.5 3.5 45.0 16.1 0.0 83.9 15 1.7 Commercial diet 5.0 40.1 54.9 0.0 5.1 94.9 $19.6 $9.1 10 Folk remedies 0.0 0.0 100.0 ...... $12.2 Lifestyle diet 0.0 44.9 55.1 62.3 0.0 37.7 5 Energy healing 30.8 8.2 61.1 0.0 19.1 80.9 Annual Out-of-Pocket Expenditures, $ in Billions Expenditures, Out-of-Pocket Annual Homeopathy 0.0 0.0 100.0 0.0 24.7 75.3 0 Upper Lower Out-of-Pocket Out-of-Pocket Hypnosis 5.1 0.0 94.9 7.0 0.0 93.0 Bound Bound All US All US Biofeedback 30.5 43.7 26.0 14.1 19.9 66.0 (RBRVS (Conservative Physician Hospital- Prices) Prices) Services21 izations21 Acupuncture 0.0 40.7 59.3 21.6 23.0 55.4 Weighted average across all therapies 15.3 26.4 58.3 12.3 23.7 64.0 Figure 2.—Estimated annual out-of-pocket expen- *Data are percentage of users of alternative therapies provided by practitioners. Ellipses indicate data not applicable. ditures for alternative therapies vs conventional †Reimbursement patterns not explored in 1997. medical services, United States, 1997. Data are from the Health Care Financing Administration, Table 5.—National Projections of Expenditures for Alternative Therapies in the United States, 1997 vs 1990* United States.21 RBRVS indicates Resource-Based Relative Value Scale. 1997 1990 Change (%), (Billions of Dollars) (Billions of Dollars) 1997 vs 1990 (Billions of Dollars) pared with 3 in 10 in 1990. For adults aged Conservative RBRVS Conservative RBRVS Category of Expenditure (SE) (SE) (SE) (SE) Conservative RBRVS 35 to 49 years in 1997, it is estimated that Total expenditures on 21.2 (2.4) 32.7 (3.8) 14.6 (4.0) 22.6 (6.1) 6.6 (45.2) 10.1 (44.7) 1 of every 2 persons used at least 1 alter- professional services for 15 native therapy. Overall prevalence of use alternative therapies† increased by 25%, total visits by an esti- Out-of-pocket expenditures mated 47%, and expenditures on services Professional services, 15 12.2 (1.7) 19.6 (3.3) 7.2 (1.3) 11.0 (2.1) 5.0 (69.4)§ 8.6 (78.2)§ therapies†‡ provided by practitioners of alternative Megavitamins 3.3 (0.4) 0.9 (0.3) 2.4 (266.7)࿣ therapies by an estimated 45% exclusive Commercial diet products 1.7 (0.3) 1.3 (0.3) 0.4 (30.8) of inflation. Moreover, the use of alterna- Subtotal of out-of-pocket 17.2 24.6 9.4 13.2 7.8 (83.0) 11.4 (86.4) tivetherapiesisdistributedwidelyacross expenditures assessed in all sociodemographic groups. 1997 and 1990† It is possible to arrange the 16 princi- Out-of-pocket expenditures pal therapies common to the 1990 and assessed only in 1997 Herbal medicine 5.1 (0.5) ...... 1997 surveys along a spectrum that var- Therapy-specific books, 4.7 (0.8) ...... ies from “more alternative” to “less alter- classes, and equipment native” in relationship to existing medi- Total out-of-pocket 27.0 34.4 ...... cal school curricula, clinical training, and expenditures for alternative practice.Arguably,therapiessuchasbio- therapies in 1997† feedback, hypnosis, guided imagery, re- *The 1990 and 1997 cost measures are based on 1990 and 1997 population estimates, respectively (180 million vs laxation techniques that involve elicita- 198 million). Both used 1997 per-visit price estimates as follows (conservative price estimate is followed by Resource- tion of the relaxation response (Ͻ1% of Based Relative Value Scale [RBRVS] estimate for each therapy): relaxation techniques ($20, $50), herbal medicine ($40, $60), massage therapy ($40, $60), chiropractic care ($40, $65), megavitamins ($40, $50), self-help groups ($20, $20), the sample), lifestyle diet, and (possibly) imagery techniques ($45, $50), commercial diet ($20, $20), folk remedies ($20, $50), lifestyle diet ($20, $60), energy vitamintherapycanbeconsideredasrep- healing ($40, $50), homeopathy ($45, $60), hypnosis ($60, $80), biofeedback ($60, $80), and acupuncture ($40, $60). (Price estimates for spiritual healing by others were not included because respondents reporting use were not asked resentative of the more conventional (ie, for details of professional visits). Ellipses indicate data not applicable. less alternative) side of the spectrum. †These figures reflect the range in out-of-pocket expenditures for conservative vs RBRVS-derived visit prices. ‡Assumes a 20% copayment for users with partial insurance coverage. Visits associated with these 6 categories §PՅ.05.; ࿣PՅ.001. accounted for less than 10% of total visits to alternative medicine practitioners; the reports concerning recollections of num- Withinthecontextoftheselimitations, remainderwereassociatedwiththemore ber of visits and amounts spent on books, theresultsofthese2surveyssuggestthat alternative therapies. classes, relevant equipment, herbs, or the prevalence and expenditures associ- In light of the observed 380% increase supplements. To the extent possible, we ated with alternative medical therapies in the use of herbal remedies and the adjusted by weighting data on sociode- in the United States have increased sub- 130% increase in high-dose vitamin use, mographic variables associated with al- stantiallyfrom1990to1997.Thisincrease it is not surprising to find that nearly 1 in ternative therapy use (eg, income, edu- appears to be primarily due to increases 5 individuals taking prescription medi- cation,age,region).Itisconceivablethat in the prevalence of use and in the fre- cations also was taking herbs, high-dose the estimated prevalence and costs of al- quency with which users of alternative vitamin supplements, or both. Extrapo- ternative therapy use would have been therapy sought professional services. In lations to the total US population sug- lower if it were possible to correct for 1997,anestimated4in10Americansused gest that an estimated 15 million adults those limitations. at least 1 alternative therapy as com- are at risk for potential adverse interac-

1574 JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007 tions involving prescription medications have a sufficiently large database to pro- 6. Goldbeck-Wood S, Dorozynski A, Lie LG, et al. and herbs or high-dose vitamin supple- vide precise estimates of the patterns of Complementary medicine is booming worldwide. BMJ. 1996;313:131-133. ments. This figure includes nearly 3 alternative therapy use among African 7. Rasmussen NK, Morgall JM. The use of alterna- million adults aged 65 years or older. Ad- Americans, Hispanic Americans, Asian tive treatments in the Danish adult population. verse interactions of this nature, includ- Americans, or other minority groups. Complementary Med Res. 1990;4:16-22. ing alterations of drug bioavailability Parallel surveys, modified to include 8. Vaskilampi T, Merila¨inen P, Sinkkonen S, et al. 23-27 The use of alternative treatments in the Finnish or efficacy, are known to occur and therapiesuniquetominoritypopulations adult population. In: Lewith GT, Aldridge D, eds. are more likely among individuals with andtranslatedwhenappropriate,should Clinical Research Methodology for Complemen- chronic medical illness, especially those be conducted using necessary sampling tary Therapies. London, England: Hodder & with liver or kidney abnormalities. No strategies. Only then can we compare Stoughton; 1993:204-229. adequate mechanism currently is in patterns across ethnic groups and pri- 9. MacLennan AH, Wilson DH, Taylor AW. Preva- lence and cost of alternative medicine in Australia. place to collect relevant surveillance oritize research agendas for individual Lancet. 1996;347:569-573. datatodocumenttheextenttowhichthe populations. As alternative medicine is 10. Fisher P, Ward A. Complementary medicine in potential for drug-herb and drug-vita- introduced by third-party payers as an Europe. BMJ. 1994;309:107-111. min interaction is real or imaginary. attractiveinsuranceproduct,itwouldbe 11. Sermeus G. Alternative health care in Belgium. Complementary Med Res. 1990;4:9-13. The magnitude of the demand for al- unfair for individuals without health in- 12. Bouchayer F. Alternative . Comple- ternative therapy is noteworthy, in light surance and those with less expendable mentary Med Res. 1990;4:4-8. of the relatively low rates of insurance income to be excluded from useful alter- 13. Piel E. Erfahrungen mit Naturheilmitteln-Um- coverage for these services. Unlike hos- native medical services or consultation frageergebnisse aus West-und Ostdeutschland. Therapeutikon. 1991;5:549-551. pitalizations and physician services, al- (eg, professional advice on use or avoid- 14. Millar WJ. Use of alternative health care prac- ternative therapies are only infrequently ance of alternative therapies). titioners by Canadians. Can J Public Health. 1997; included in insurance benefits. Even In conclusion, our survey confirms 88:154-158. when alternative therapies are covered, thatalternativemedicineuseandexpen- 15. US Bureau of the Census. United States popu- lation estimates, by age, sex, race, and Hispanic ori- they tend to have high deductibles and ditures have increased dramatically gin, 1990 to 1997. Available at: http://www.census co-payments and tend to be subject to from 1990 to 1997. In light of these ob- .gov/population/estimates/nation/intfile2-1.txt. stringent limits on the number of visits or servations, we suggest that federal 16. Resource-Based Relative Value Scale Con- total dollar coverage. Because the de- agencies, private corporations, founda- verter. Customized Report for Medical Billing and mand for health care (and presumably al- tions, and academic institutions adopt a Collection of All CPT Codes. Salt Lake City, Utah: Medicode Inc; 1996. ternative therapies) is sensitive to how more proactive posture concerning the 17. US Bureau of Labor Statistics Data. Consumer much patients must pay out-of-pocket,28 implementation of clinical and basic sci- Price Index—All Urban Consumers. US Bureau of current use is likely to underrepresent ence research, the development of rel- Labor Statistics Web site. Available at: http:// utilization patterns if insurance coverage evant educational curricula, credential- 146.142.4.24/cgi-bin/surveymost?cu. 1998. Series ID: CUUR0000SAM. foralternativetherapiesincreasesinthe ing and referral guidelines, improved 18. SUDAAN: Professional Software for Survey future. quality control of dietary supplements, Data Analysis [computer program]. Version 7.5. In1990,afullthirdofrespondentswho and the establishment of postmarket Research Triangle Park, NC: Research Triangle In- used alternative therapy did not use it surveillance of drug-herb (and drug- stitute; 1997. 1 19. Schappert SM. National Ambulatory Medical for any principal medical condition. supplement) interactions. Care Survey: 1990 Summary. Hyattsville, Md: Na- From these data, we inferred that a sub- tional Center for Health Statistics, 1992. Advance stantial amount of alternative therapy This study was supported in part by National In- DataFromVitalandHealthStatistics,No.213:1-11. was used for health promotion or disease stitutes of Health grant U24 AR43441, Bethesda, 20. Woodwell DA. National Ambulatory Medical Md, the John E. Fetzer Institute, Kalamazoo, Mich, Care Survey: 1996 Summary. Hyattsville, Md: prevention. In 1997, 42% of all alterna- The American Society of Actuaries, Schaumburg, National Center for Health Statistics; 1997. Ad- tive therapies used were exclusively at- Ill, the Friends of Beth Israel Deaconess Medical vance Data From Vital and Health Statistics, No. tributed to treatment of existing illness, Center, and the Kenneth J. Germeshausen Foun- 295:1-25. whereas 58% were used, at least in part, dation, Boston, Mass, and the J. E. and Z. B. Butler 21. Health Care Financing Administration, Office Foundation, New York, NY. of the Actuary, National Health Statistics Group. to “prevent future illness from occur- The authors thank the staff of DataStat, Inc, Ann Nationalhealthcareexpenditureprojectionstables. ring or to maintain health and vitality.” Arbor, Mich, for their assistance with telephone Available at: http://www.hcfa.gov/stats/NHE-Proj Despite the dramatic increases in use data collection, Linda Bedell-Logan for assistance /tables/. with RBRVS data analyses, Dan Cherkin, PhD, 22. Groves RM, Couper MP. Societal environmen- and expenditures associated with alter- Murray Mittleman, MD, Ted Kaptchuk, OMD, and native medical care, the extent to which tal influences on survey participation. In: Nonre- Thomas Delbanco, MD, for their review of the sponse in Household Interview Surveys. New York, patients disclose their use of alternative manuscript, and Debora Lane, Marcia Rich, and NY: John Wiley & Sons Inc; 1998:159. therapiestotheirphysiciansremainslow. Robb Scholten for their technical assistance. 23. Ernst E. Harmless herbs? Am J Med. 1998;104: Less than 40% of the alternative thera- 170-178. 24. D’Arcy PF. Adverse reactions and interactions pies used were disclosed to a physician in References with herbal medicines, part 1: adverse reactions. both 1990 and 1997. It would be overly 1. Eisenberg DM, Kessler RC, Foster C, et al. Un- Adverse Drug React Toxicol Rev. 1991;10:189-208. simplistic to blame either the patient or conventional medicine in the United States. N Engl 25. D’Arcy PF. Adverse reactions and interactions their physician for this inadequacy in pa- J Med. 1993;328:246-252. with herbal medicines, part 2: drug interactions. tient-physician communication. The cur- 2. Paramore LC. Use of alternative therapies. Adverse Drug React Toxicol Rev. 1993;12:147-162. J Pain Symptom Manage. 1997;13:83-89. 26. De Smet PAGM, D’Arcy PF. Drug interactions rent status quo, which can be described 3. LandmarkHealthcare.TheLandmarkReporton with herbal and other non-orthodox remedies. In: as “don’t ask and don’t tell,” needs to be Public Perceptions of Alternative Care. Sacra- D’Arcy PF, McElnay JC, Welling PG, eds. Mecha- abandoned.29 Professional strategies for mento, Calif: Landmark Healthcare; 1998. nisms of Drug Interactions. New York, NY: responsible dialogue in this area need to 4. Pelletier KR, Marie A, Krasner M, et al. Current Springer Publishing Co Inc; 1996:327-352. trends in the integration and reimbursement of 27. De Smet PA. Health risks of herbal remedies. be further developed and refined. complementary and alternative medicine by man- Drug Saf. 1995;13:81-93. Data from this survey, reflective of aged care, insurance carriers, and hospital provid- 28. Shekelle PG, Rogers WH, Newhouse JP. The the US population, are representative of ers. Am J Health Promot. 1997;12:112-122. effect of cost sharing on the use of chiropractic ser- a predominantly white population. Even 5. Wetzel MS, Eisenberg DM, Kaptchuk TJ. vices. Med Care. 1996;34:863-872. Courses involving complementary and alternative 29. Eisenberg DM. Advising patients who seek al- if we were to combine data sets from the medicine at US medical schools. JAMA. 1998;280: ternative medical therapies. Ann Intern Med. 1997; 1990 and 1997 surveys, we would not 784-787. 127:61-69.

JAMA, November 11, 1998—Vol 280, No. 18 Trends in Alternative Medicine Use in the United States—Eisenberg et al 1575 ©1998 American Medical Association. All rights reserved. Downloaded from www.jama.com by TedNissen, on February 28, 2007