Tobacco Documents Online

The New England Absttuts in lhe advenising a:C(iOns ®~n~. of Medicine Established in 1812 as The NEW ENGLAND JOURNAL OF MEDICINE AND SURGERY I VOLL'\1}: 3'_>1 Ji:11' 9. 1991 NNUMBER 19 Original Articles Treatmet.t of Aplastic Anemia with Antilym- phocyte Globulin and Methylpredniso- lone with or without Cyclosporine..... 1297 NokBt.kT 1-kU.IQltttt.V \XU UtUt.kS .. Renal Hemodynamics and the Renin-Angio- tensin-Aldosterone System in Normo- tensive Subjects with Hypertensive arid Normotensive Parents ............ 1305 INGktu \l.S. \•.\% 1{tK,t-r .\nu <)ntt.ks Preoperative Transcatheter Closure of Con- genital Muscular Ventricular Septal Defects .............................. 1312 N.\Nt.v 1). I;ktuth.S .\.D UIJft.k, Clinical and Laboratory Findings in the Ocu- locerebrorenal Syndrome of Lowe, with Special Reference to Growth and Renal Function .................. 1318 L.\\vkt.]c..t. R. (:n.\kV..s.\.U OttnkN Special Article Risk Factors for Falls as a Cause of Hip Frac- ture in Women ....................... .,\V (-ik1stiU .\\U OI11tk5 1326 Review Articles Current Concepts: Mycobacterium avium Complex Infection in the Acquired Immunodeficiency Syndrome ......... 1332 (:. Itcutt.kr Hltkait kc.n.,Jk. Drug Therapy: Erythropoietin ............ 1339 .\t.t,aV ./. t:k.,t t v Case Records of the Massachusetts General Hospital An 80-Year-Old Woman with Increasing Dyspnea and Extensive Pulmonary Opacities while Receiving a De- creasing Prednisone Dose for Poly- myalgia Rheumatica .................. 1345 i nlul l.v R. \\'t \\u Et c.l.r./. \I \kti Editorials Immunosuppression in Aplastic Anemia - Postponing the Inevitable? ............ 1358 \!.\t c.ut.+t .\.S. MtNrkt. \ND 1{c'ut, (' \. t kn-\i \t..\tirl~.\ Biotechnology - The Enormous Cost of Success .............................. 1360 Rc„ttt. 1', l)titttttt.t Sounding Board Avoiding Bias in the Conduct and Reporting of Cost-Effectiveness Research Spon- sored by Pharmaceutical Companies ... 1362 .1t,.\. 1.. tiu O t nt ks Massachusetts Medical Society ...... 1365 Correspondence Protecting Severely Abused and Neglected Children ............................. 1366 Joint-ReplacementSurgery .................. 1367 Successful Treatment of Aplastic Anemia with High-Dose Immunoglobulin ............. 1368 Erythropoieiin for Anemia of Renal Failure in Sickle Cell Disease ..................... 1369 More on the Association of Glucose-6-Phosphate Dehydrogenase Deficiency with Hairy-Cell lf.~ukemia ............................. 1370 Iaparoscopic Nephrectomy .................. 1370 Case 2-1990 ............................... 1371 Frequency and Cost of Diagnostic Imaging in Of- fice Practice - A Comparison of Self-Refer- ring and Radiologist-Referring Physicians . 1371 Walking through Plate Glass: A Trial of One ... 1373 Book Reviews .......................... 1373 Books Received ........................ 1375 Notices ................................. 1377 Corrections Itigh•Dose Intramuscular Triamcinolone in Se- vere, Chronic, Life-Threatening Asthma ... 1380 Abnormalities of Lipoprotein Metabolism in the Nephrotic Syndronte ................... 1380 OHned. Published, and , Coprrightid, 1991. bl• the \tassachutiettc \tedical Sucicty I ut N't t, %a' 1. ~~ t;. ~t ~~t \II ut. t\t 111\ f:'+{ t, lndhh.ln ~I uc~ i,h Ntt.- t. tL„t.,n. \1 \ n!111.taril "ul- t il.ti.m ln l~vat. ".-n.L. L- p.ud,t IS~..I.~u "ml et .nlliu.~nal ni.niuil~ .~lh. r.. i'().I \1 \, I LR 1. mi -6!c... .li.a ;, t,. 1' () IS,.\ ibt t, \\Ai6.'m. \I \ u:!'.I u"n i

t3r>> •rttr: NEW t•:N<a.:.ND,lcyU.u`At. OF Ntt:Iyua`t: May. <.1, 1991 7. Bchbach 3W. Egric JC. Downing D;R, Brownc JK, Aduason JW. Corrm:- tion of thc ancmia of cnd•ua^c rcnal discasc with rccombin:mt human crylh- ropoictin: results of a combined Phcw I and 11 clinical trial. N Engl J blcd 1987: 316:73•8. 8. Enkv AL Erythtc>poictin. N Engl J Atcd 1991; 324:1339-34. 9. United Statcs Renal Data System. The magnitudc of the problcm. Annual data rcport. Washington. D.C.: Department of Health and Human Scrviccs. 1990:5-10. Itl. Cx~cshon D. Amgen scorcs a knockout. Nature 1991: 350:99. 11. Stavro B. lnvcctors tracking in.idcrs' .tock .alcs for clucs on Amgcn. t.os Angctcs7imcs. March 27. 1991:Dl. SOUNDiNG BOARD AVOIDING BIAS IN THE CONDUCT AND REPORTING OF COST-EFFECTIVENESS RESEARCH SPONSORED BY PHARMACEUTICAL COMPANIES BECAUSE of the growing focus on containing health care costs, pharmaceutical companies are trying to demonstrate the cost e[Tectiveness of their products relative to alternatives. In Europe and Australia, eco- nomic analy: es are often required for government ap- proval and pricing of new pharmaceuticals. In the United States, such analyses are increasingly being used for marketing and to obtain formulary approval. Because of the corporate need for timely results and confidentiality about a new drug in the period before marketing begins and because other finan- cial support is limited, pharmaceutical companies themselves sponsor most academic research into the cost effectiveness of pharmaceuticals. The relation- ship is symbiotic. Academic researchers - especially young, unestablished investigators - are eager for new sources of funding.''3 At the same time, pharma- ceutical marketing benefits from the imprimatur of research published by respected independent academ- ic researchers. We have performed 33 economic analyses for 15 pharmaceutical companies since 1978. Our experience indicates that the partnership between academia and industry in t}•is research is often exemplary, generat- ing important and valid new information. Sometimes, however, we have encountered difficulties that threat- ened to bias the conduct and results of an analysis. "I'hese problems represented potential conflicts of in- terest, because the sponsors of the studies benefit most from favorable results. In this article Nve explore the conduct of industry- sponsored economic analysis. We offer examples of some speciSc biases that may occur because of the behavior of the ;nvestigator, the sponsoring company, or both. We ex- -e whether there is a role for regttla- tion by the Fo.,- and I)rug Administration. Finallv, we ofTer suggestions for structuring the relatiunship .'.d,lrc., reprint rcyuc.ts to Dr. Alen L. H llmen at tius A'l. HtrPnal ot thc Univcrnity of PcnnsNMnia. 331t0 Spruce St.. l'hiludelphia. PA 191tF3 :'83. between industry and academia so that bias is mini- mized in economic analyses of pharmaceuticals. ETHICAL CONDUCT IN INDUSTRY-SPONSORED ECONOMIC ANALYSIS The respos sibilitics of investigators in clinical re- search have been identified, evaluated, and debated in the literature.{'s Investigators' fundamental responsi- bilities are to conduct research and present results in an accurate and unbiased fashion.j Several sets of guidelines regarding fraud,''" informed consent,1z bias,'3''' randomization procedures,11'" and other methodologic issues have been devcloped ' Ethical is- sues in industry-sponsored research include potential conflicts of interest between researchers and their funding sources,'-'"-t7 the usefulness of industry-spon- sored supplements to medical journals in which peer review is scant;'1 the responsibilities of academic cen- ters in their partnerships with industry to develop pro- prietary patents and products,'-' and equity interests held by researchers in the companies sponsoring their rescarch:=' There has also been scrutiny of ~ roader encounters between physicians and the pharmareuti- cal industry, including investigation of the con:ent of continuing-education programs sponsored by phar- maceutical companies,21.1f the use of company repre- sentatives as a primary source of information about new pharmaceutical agents,s"'33 and the ethics of pro- motional efforts by pharmaceutical companies, such as offering providers gifts, vacations, or free prod- ucts. r'•31.a5 So far, however, the special case of industry-spon- sored economic analysis of pharmaceuticals has es- caped attention. Economic analysis is difTerent from the clinical trials conducted at academic centers sub- ject to FDA regulations. In contrast to clinical trials, economic studies generally use data and analytic methods of varying degrees of precision and power that are usually tnlstandardized, rather than stand- ardized designs and analytic techniques used under the strict scrutiny of an external regulatory agency; involve subjective opinion and interpretation about what the results demonstrate~ rather than limiting themsclves to narrow conclusions about safety and eC ficacy; look at products selected by the pharmaceuti- cal firms from among agents already (or nearly) in clinical use, rathcr than at all drugs for which FI)A approval is sought; may make selective use of propri- etary data gathered after marketing has begun, rather than disclosing all data by FI)A requirement; and are funded and overseen most comntonly by marketing departments, rather than medical or scientific divi- sions. Pharmaceutical companies conStde r economic anal- yses to be marketing devices more than scientific or clinical research. '1'11is view raises problems rrlated to dilTerences in the missions of the sponsoring corporate -departments and to the orient:uion and b,tckground of the people who oversee the rescarch. Scientific person- ncl in pharmaceutical comp;inics are accustomed to 0 U 0 e

I I Vo1.3`L4 \o.1tl SOUNDING BOARI) the ethics aad issues of resea: ch methodology, where- as marketing personnel are concerned with the ex:gen- cies of the marketplace and the potential efTect of economic findings on sales volume. Economic analyses by academic investigators are funded by both grants to universities and contracts with them. Although some universities require that investigators follow strict rules to prevent conflicts of in~erest and protect academic freedom, such policies are neither universal nor uniform. Many economic analyses are performed without this oversight, by pri- vate research and consulting firms under a variety of contractual arrangements. Furthermore, economic at• lysis is a new, evolving, complex, and unstand- ardized field with which many editors, reviewers, poli- cy makers, physicians, and universities are unfamiliar. Because economic analysis is poorly understood and unsupervised by an external agency, existing universi- ty policies to protect academic integrity are less able to detect and prevent bias in such studies than in other forms of research. Although the supervision of all types of clinical research may need strengthening, ad- ditional explicit gaidelines are especially needed for economic analysis. THE NEED FOR GUIDELINES In economic analysis, the choice of which agents and interventions to compare, which data to include, which costs to measure, which perspective to adopt, which outcomes to assess, which assumptions to make, and how to present the results may produce important dill'erences in a study's conclusions. For ex- ample, excluding subjects who do not respond, using charges rather than costs or payments, or not dis- counting outcomes can improve the calculated eco- nomic effect of an agent. When precise clinical data and information on price and use are unavailable, they must be estimated with statistical modeling tech- niques or expert opinion. Many of the assumptions made in economic analyses are not easily recognized by the readers of a study report. In the absence of standardized methods, investiga- tors may (consciously or subconsciously) design eco- nomic analys:-s to produce favorable findings, which may enhance future opportunities for corporate fund- ing.l'.'9 Invcstigators also prefer positive findings be- cause they are easier to have published and command more attention than negative or marginal results among peers and the public.13•"' Pharmaceutical com- panies, which generally prefer cooperative investiga- tors to skeptical ones," sometimes intimate (or sug- gest) the desirability of making the dollar impact of their product appear as favorable as possible. Inexpe- rienced investigators with a limited perspective on re- search methods, ethics, and standards of practice and investigators who depend heavily on pharmaceutical funding may be most susceptible to these pressures. Pharmaceutical companies' main interest in eco- nomic research is to promote sales. They the,-efore fund projects with a high likelihood of producing Iia- 1.3G3 vorable results. Limited comparisons betweeft prod- ucts are sometimes authorized, but they exclude prod- ucts that may compare favorably with the sponsor's own. Sometimes, only favorable clinical data are re- leased to investigators. Pilot studies are commonly performed to assess the likelihood of favorable results before academic researchers are engaged to perform publishable independent studies. Some projects are funded in steps, so that losses can be cut if the initial results are not favorable. Viewed as part of a company's marketing efforts, these represent good and well-accepted business prac- tices. They systematically distort the balance of com- pleted and published studies, however, since negative studies may be terminated before they are ready for publication. Indeed, published economic findings that are distinctly unfavorable to the sponsoring firm's product are rare, although companies are often disap- pointed when the answers are less favorable than an- ticipated. Some information about older, less cost eC fective pharmaceuticals is published because these are the agents with which companies generally want to compare their own products. Many researchers, funding agencies, and journals prefer positive findings in any form of research."-w Bias in economic analysis, however, leads to the sys- tematic suppression of information about already available clinical agents with documented medical benefit but less favorable economic profiles. This bias is different from simply not undertaking the expense of a scientific investigation into a compound that is un- likely to have medical benefit in the first place. With economic analysis, any valid finding can potentially help decision makers determine the most appropriate use of an approved pharmaceutical (e.g., the sub- groups of patients who should use the drug) or allo- cate resources more effectively. Pharmaceutical companies, like other public or pri- vate organizations that contract for research, usually retain the right to review and comment on results and drafts of manuscripts before publication. Such reviews take various forms, with various degrees and styles of input. Input can often be helpful - for example, in facilitating the investigator's access to additional data, or in clarifying how a clinical trial was performed. In extreme cases, however, corporate personnel may seek to control the content and use of the final report, in- cluding the decision to publish. Investigators may be threatened with the withdrawal of current or future funding unless specific changes in methods, presenta- tion, or results are made. Investigators may be with- out recourse if the company has stipulated that confi- dential data cannot be published when a contract has been terminated. Bias may also result from an overemphasis on cost savings. A drug that increases costs may be more de- sirabie than anothcr treatment if the additional cost produces increased bencfit and value. For marketing purposes, however, such a result is less compclling than re.cults showing net cost savings. Investigators

1361 THF. NENV F,NGLANI),JOL'RNAL OF.MEDICINE May 9, 1991 frequently lose funding for potential projects when the prospects for cost savings (rather than mere cost elTec- tiveness) are not promising. Fina!ly, pharmaceutical companies seek to sponsor economic analyses at a competitive price. The absence of regulatory oversight allows firms and investiga- tors to compromise excessively in the inevitable trade- off between the cost and quality of a study. For example, the sample size needed to demonstrate an economic difference among pharmaceutical inter- ventions is often larger than that needed to demon- strate clinical differences. Nonetheless, funding for economic analyses is often defined by the sample sizes involved in clinical projects designed to detect clinical differences. In our experience, flagrant, intentional attempts by researchers or companies to manipulate results are rare, but other practices are not. For example, review of manuscripts before publication is usual, attempts to fund only positive studies are the rule, and incentives for investigators to make favorable assumptions arc ever present. WOULD RF.GULATION HELP? Economic analyses are not currently required or re- viewed by the FDA or any other regulatory body. Furthermore, although the few economic analyses sponsored bv federal agencies (e.g., the Agency for Health Care Policy and Research, the Depart- ment of Veterans Affairs, and the National Insti- tutes of Health) do include peer review of proposed methods before funding, projects sponsored by phar- maceutical companies are not subject to peer review until the material is submitted for publication. FDA oversight could help standardize the relationship be- tween companies and investigators and reduce the po- tential for bias. Companies would be required to pro- vide explicit funding from scientific departments for economic analysis, rather than including it in market- ing budgets on a discretionary basis. The lack of a standardized approach to economic analysis would make additional FDA regulation cumbersome and controversial at this time, however, especially given the cost and delay it has introduced into the areas it alreadv covers. One type of bias is actually fostered by current FDA rules that restrict advertising to FDA-approved indications, which are based on the results of con- trolled clinical trials. Often, these trials do not reflect the ways doctors actually practice, because patient selection and monitoring in the trials are driven by protoco'.s. When data (from studies undertaken after marketing has begun, for example) exist to indi- cate the effectiveness of a pharmaceutical in actual clinical experience, it may be desirable to assess the economic effect of the drug as it actually is used, not the effect of the FDA-appro.•cd uses alone. Some companies authorize economic components as part of clinical studies; in such cases, one arm as- sesses the behavior of patients and doctors outside protocol-driven constraints. Permitting the use of such results in marketing could help to ensure that economic analyses reflect both clinical realities and available clinical data. RECOMMENDATIONS: A PROTOCOL FOR ECONOMIC ANALYSIS Pressure toward bias is to be expected in a competi- tive marketplace. The selective producti.,n and pres- entation of results by one firm will provoke a com- petitive response from other firms. This impedes scientific progress by casting doubt on the objectivity and validity of results, whether favorable or un- favorable.l' Identifying studies that have followed pro- tocols designed to reduce bias would allow readers to recognize the likelihood of bias in studies not so desig- nated. As a result, competitive pressures would be channeled toward less biased, more useful research. Adherence to these standards would generate a seal of appra,~al that could in :tself be valuable in selling pharmaceuticals in an increasingly skeptical and com- petitive marketplace. We offer the following eight rec- ommendations. First, written agreements between pharmaceutical companies and investigators should be in the form of research grants to universities, rather than contracts with individual in.estigators or universities. The agreements should stipulate that the researcher may publish findings regardless of their nature and retains sufficient access to proprietarv data to allow publica- tion even if funding is withdrawn. Such an arrange- ment will also ensure that questions raised by readers and other investigators can be addressed effectively. Private research and consulting firms should establish similar rules. Second, economic analyses are by their nature com- parative. The selection of alternatives to be compared should be based on their clinical relevance, not on the potential favorability of the results. At a minimum, comparisons required by the FDA for controlled clini- cal trials should be included. Research reports should be explicit about the comparisons chosen ind those omitted. Data from all relevant studies, ~:,ther than a selected subset, should be made available to the inves- tigator. Third, investigators should be allowed to expand the company's study design to include additional types of costs, economic perspectives, and comparison drugs as data permit. If the funding group has placed constraints on an investigator, this information must be clearly disclosed to editors, reviewers, and readers. Fourth, if projects are funded in a series of steps rather than by one large grant, results should not be provided to the sponsor until publication is guaran- teed and funded. (Initial steps can include, for exam- ple, a review of the literature, general advice, and the planning of the economic modcl.) Fifth, investigators should be vigilant to the tcmpta- tion to produce favorable findings. The assumptions required in a study should be conservative (i.e., biased against the results sought by the funding company), clearly identified, and supported with formal sensitiv- ~W

Vol. 321 No. 19 \iASSACHUSE'I"I'S MEDICAL SQCIE'PY 1365 ity analyses. Editors should encourage the publication of these details despite current limits on the length of articles in many journals. Sixth, investigators should publish valid results re- gardless of their promotional value to the sponsoring company, and journal editors should try to avoid a bias against publishing negative results. Furthermore, r,:searchers who receive a grant should not act as consultants on projects related to the study medication during the active period of the grant. Journals should require that all financial rela- tions between authors and their sponsoring companies (as well as their direct competitors) he disclosed, in- cluding equity interests other than the ownership of shares in mutual funds. Finally, researchers should take all reasonable steps to ensure that the level of funding permits methodo- logically sound, clinically relevant results with enough statistical power to detect important differPnces among the alternatives compared. Projects should avoid a "good enough for marketing" mentality. All published studies should include a statement that their authors have adhered to this protocol. Uni- versities and nonacademic researchers should develop and enforce standard agreements that meet these con- flict-of-interest guidelines. Also, academic research societies, journal editors, pharmaceutical compa- nies, and federal funding agencies should adopt these guidelines, perhaps after modification by a joint task force. Such actions will help ensure that this burgeon- ing neti• field of academic endeavor grows to fulfill its promise as a tool to help decision makers allocate re- sources effectively. ALAN L. HILLMAN, M.D., M.B.A. JOHN M. EISENBERC, M.D., M.B.A. Leonard Davis Irtstitute of Health Economics University of Pennsylvania Philadelphia. PA 19104 MARK V. PAULY, PH.D. BLRNARD S. BLOOAt, PH.D. HENRY GLICK, M.A. BRUCE KIVOSIAN, M.D. J. SANFORD SCHWARTZ, M.D. We are indebted to \is. Danita Joell for administrative as- sistance. REFERENCES 1. Healy B. Innovators for the 21st century: will we face a crisis in biomedical- research brainpowcr? N Engi l Med 1988; 319:I058-64. 2. Healy B. Keyworth GA 11. The NIH and numbers: a vital concern's concern. N Eng! J Med 1985; 312:1450-2. 3. Kahn CR. A proposed new role for the insurance industry in biomedical research funding. N Engl J Med 19€4: 310257-8. 4. Americart Federation for Clinical Research National Council. American Federation for Clinical Research guidelines for avoiding conflict of interest. Clia Res 1990; 38:23940. 5. Ad Hoc Committee on Medical Ethics, American College of Physicians. American College of Physicians ethics manual. Part 11. Research. other ethical isstKS: rrcornrnended reading. Ann Intern Ucd 1984; 101:263- 74. 6. Freedman B. Equipoi;e and the ethics of clinical research. N Engl J Stcct 1987; 317:141-5. 7. Angell M. Retman AS. Fraud in biomedical research: a time for Con~n:s- sional restraint. N Engl 1 tilcd 1988; 318:1-162-3. 8. Angell M. Ethical imperialism? Ethics in international collatwrativc clinical research. N Engl I Stcd 1988: 319:1081-3. 9. Engler RL. Cov-11 JW, Friedman PJ. Kitchcr PS, Pcter. RM. \tiureprc.cn- tation and rcspcnsibilin in mc fical rescarch. ti Eng11 S1c.1 :c;87: 317:132i3- 9. 10. Nobel 11. Compar'son of research quality guidelines in nrademic and non- academic cnvironmcntc. JAMA 1990; 263:1435-7. 1 t. Woolf PK. Ensuring integrity in biomedical publication. JAMA 1987• 258:3424-7. 12. Chalmers T. 'Mc ethics of randomization as a decision-making technique and the problem of informed consent. In: Bcauchamp TL. Walters L. eds. Contemporary issues in biocthics. Encino. Calif.: Dickenson. 1978:426- 9. 13. Chalmcrs TC, Frank CS. Reitman D. Minimizing the three stages of publi- cation bias. JAMA 1990; 263:1392-5. 14. Sharp DW. What can and should be done to reduce publication bias? The perspective of an editor. JAMA 1990; 263:1390-1. 15. Dickersin K. The existence of publication bias and risk factors for its occur- rcnce. JAMA 1990: 263:1385-9. 16. Chalmers 1. Undcnepcxting research is scientific misconduct. JAMA 1990; 263:1405-8. 17. Schafcr A. The ethics of the randomized clinical trial. N Eng11 Mcd 1982; 307:719-24. 18. P^art WS. The pharmaceutical industry: research and responsibility. Lancet 1980; 2:465-6. 19. Pctsonk EL. Conflicts of interest in drug research. N Engl l Mcd 1979; 301:335. 20. Rawlins MD. Doctors and the drug makers. Iancet 1984; 2:276•8. 21. Sivarajan M. Tarnishing the medical professiori s image. N Engl J Mcd 1985:312:1132. 22. Goldfingcr SE. A matter of influcncc. N Engt J Mcd 1987; 316:1408-9. 23. The relationship b:twcen physicians and the pharntxeutical industry: a report of the Royal College of Physicians. J R Coll Physicians Lond 1986; 20:235-42. 24. GoldfingerSE. Physicians and the pharmaceutical industry. Ann Intc:n Med 1990; 112:624-6. 25. Council on Scientific Affairs and Council on Ethical and Judicial Affairs. Conflicts of interest in medical centerlindustry research relationships. JAMA 1990; 263:2790-3. 26. Bricker EM. Industrial marketing and medical ethics. N Engl J Mcd 1989; 320:1690-2. 27. Kippen A. Doctored results: how drug companies bribe doctors and medical journals. Washington Monthly. October 1990:38-42. 28. Blumenthal D. Epstein S. Maxwell l. Commercializing university research: lessons from the experience of the Wisconsin Alumni Research Foundation. N Engl I Med 1986; 314:1621-6. 29. Relman AS. Dealing with conflicts of interest. N Engl J Mcd 1984; 310:1182-3. 30. McKinney WP. Schicdermayer DL. Lurie N, Simpson DE, Goodman JL, Rich EC. Attitudes of internal medicine faculty and residents toward profcs sional interaction with pharmaccutical sales representatives. JAMA 1990; 264:i693-7. 31. Avorn J, Chen M. Hartley R. Scientific versus commcrcial sourccs of influ- ence on the prescribing behavior of physicians. Am J Med 1982: 73:4-8. 32. Musher DM, Young EI, Hamill R1. The ethics of pharmaceutical promo- tion. N Engl J Med 1986; 315:590. 33. Stross JK. Information sources and clinical decisions. J Gen I tem Med 1987; 2:155-9. 34. Chren M-M. Idndcfeld CS. Murray TH. Doctors, drug cornpanics, and gtlts. JAMA 1989; 262:3-i48-51. 35. Miller K. Gou•,eia WA. Barza M, ct al. Undesirable marketing practices in th^ pharmaceutical industy. N Engl 1 Med 1985: 313:54. MASSACHUSETTS MEDICAL SOCIETY DEATHS B,vt:En - Philip \VcHtdell Baker, \I.I).. of Chaz)•, N.\ ., died on Fcbruan• 21 at the age of 85. Dr. Baker reccivrd his de>;rec from Hanard Mctiical Sc'tcwl iu 1931. \Lv - David I. \1a, M.I).. uf \\'cstun, dicd on \i: rc h 8. Iic N\r:ts '31. Dr. \Ia recc:ved his dcs;rcc frum tht• Priti.krr S; heKd uf \lt•dicine of thc L'nivcrsity of Chicat;o.