Production of scientific articles on health in six Latin American ...

Production of scientific articles on health insix Latin American countries, 1973–19921

Alberto Pellegrini Filho,2 Moisés Goldbaum,3 and John Silvi 4

The production of articles resulting from biomedical, clinical, and public health studies thatoriginated in Argentina, Brazil, Chile, Cuba, Mexico, and Venezuela from 1973 through 1992was analyzed to discover trends in health research in Latin America. From the database of theInstitute for Scientific Information (ISI), 41 238 articles with first authors who resided in thosecountries were extracted. These articles were analyzed by subject area, type of study, country,number of authors and institutions that participated in the investigation, and citations re-ceived by each article. Also analyzed were 95 articles in epidemiology selected from a pool of570 published by authors from the six countries in 11 public health journals that enjoy inter-national prestige.

The results showed that the number of published works increased by 117% between the firstand last five-year periods within the study period. Clinical research was distributed the mostevenly among the countries, and public health research was the most concentrated (60.7%originated in Brazil). The numbers of biomedical and public health research articles showedrelatively more growth than those reporting on clinical research throughout the period. A rel-ative decrease was found in articles by only one author, which suggests a greater frequency ofteam efforts, and an increase was seen in articles with authors tied to two or more national orforeign institutions, which indicates greater cooperation between institutions and countries.The average number of citations received by each article was three, which was less than halfthe number received per article in the ISI database (7.78). Regarding the subset of 95 articlesin epidemiology, the great majority (96%) dealt with infectious diseases or maternal and childhealth, while in the international literature 78% of such articles were about chronic diseases.This group of articles gave evidence of more cooperation with international institutions andhad a citation index of 4.36 per article.

It is concluded that, despite the inherent limitations, this type of study reveals some generaltrends in the development of research in the six Latin American countries with the greatest sci-entific production and makes it possible to formulate hypotheses on the factors that influencethese trends. Taken with the proper caution, the results of studies like this one can be of greatvalue in defining health science and technology policies.

ABSTRACT

This article examines a number of re-search trends by analyzing publicationof scientific works in the health field inArgentina, Brazil, Chile, Cuba, Mexico,and Venezuela during the period1973–1992. The analyses relate to arti-cles published in journals with an inter-national scope by authors residing inthese six countries during the period inquestion. Previous studies (1, 2) haveanalyzed health research trends by

considering ongoing projects and in-formation obtained from a number ofdatabases—including a preliminaryanalysis of published scientific worksincluded in the database of the Institutefor Scientific Information (ISI) in Phila-delphia, Pennsylvania, for the period1972–1982. The study presented herewas based on a compilation of workspublished in internationally recognizedjournals indexed in the ISI database.

Rev Panam Salud Publica/Pan Am J Public Health 2(2), 1997 121

1 Published in Spanish in this journal (Vol. 1, No. 1,1997, pp. 23–34) with the title “Producción deartículos científicos sobre salud en seis países deAmérica Latina, 1973 a 1992.”

2 Program of Research Coordination, Division ofHealth and Human Development, Pan AmericanHealth Organization, Washington, D.C., U.S.A.Mailing address: PAHO; 525 Twenty-third Street,NW; Washington, DC 20037, USA.

3 Department of Preventive and Social Medicine,University of São Paulo School of Medicine, SãoPaulo, Brazil.

4 Health Situation Analysis Program, Division ofHealth and Human Development, Pan AmericanHealth Organization, Washington, D.C.

THE NATURE ANDLIMITATIONS OF THE STUDY

The type of study we are concernedwith here falls within the ambit of sci-entometrics, a discipline developed inthe early 1970s by sociologists and his-torians of science interested in con-ducting quantitative analyses in thisfield. In the mid-1970s, concern aboutresearch costs and the need to evaluateresearch productivity led planners ofscientific and technologic activities torecognize scientometric studies’ po-tential for supporting the planningprocess, which to that point was basedalmost exclusively on expert opinions.

This acknowledged potential not-withstanding, scientometric studieshave been criticized for both their conceptual approach and their meth-odology (3). As regards the former, sci-entometrics has been criticized for conceiving of science in terms of aninput-output model and for devotingitself exclusively to measuring the in-puts and outputs of scientific activity,with no concern for analyzing scienceas a process that is influenced by bothintrinsic and extrinsic factors and thatin turn exercises an influence on thosefactors. Such critics point out that anarticle published in one of the major,widely recognized journals is virtuallythe only product of scientific researchacknowledged by these studies.

With regard to methodologic issues,there is a tendency to criticize sciento-metrics’ lack of parameters and criteriafor assessing both scientific productionand productivity. Scientometric stud-ies generally limit their scope to mak-ing comparisons between countries orto analyzing historical series in a singlecountry in order to identify trends.Comparisons between countries arecriticized for not taking into accountcultural differences influencing publi-cation; and similarly, comparisons be-tween production and productivity indifferent areas or different types of re-search (basic and applied) are criti-cized for not considering different pub-lication patterns. Moreover, not allarticles are equal; nor do they repre-sent equivalent efforts that would

allow them to be compared withoutconsidering such differences.

The citation index (the relative num-ber of times one source is cited by oth-ers) that is often used in scientometricstudies appears to offer a typical exam-ple of an indicator that was establishedbefore an a priori definition of the thingto be indicated (quality, impact, orsome other attribute) was arrived at.Moreover, there is generally no way toeliminate the distortions caused by self-citations, citations based on “cronyism”among group members, indirect cita-tions, negative citations, etc.

The most frequently used source forscientometric studies, the ISI database,is not considered appropriate for ana-lyzing scientific work in developingcountries, as it includes only articlespublished by authors from those coun-tries in the industrialized world’s most prestigious journals. The ISIdatabase also exhibits a tendency tofavor English-language publications.

To overcome these limitations in thefuture, it will be necessary to satisfynumerous methodologic needs: theneed to combine quantitative and qual-itative approaches; the need to developlocal parameters for studying scientificproduction in developing countries;and the need to establish new indica-tors that will respond to well-definedquestions and take account of the di-versity existing between areas, disci-plines, types of research, etc.

Meanwhile, with regard to currentcriticism of scientometric studies,there appear to be two possible op-tions. These are (1) abandon this typeof study, because of its lack of utility,until appropriate methodologies be-come available; or (2) continue con-ducting studies of this sort withproper appreciation of their limita-tions and using the results with care todefine policies. As the existence of thepresent study suggests, we favor thesecond position. However, we do rec-ognize the study’s superficial and par-tial nature—as well as its potential forhelping to improve understanding aswell as to describe in a limited mannerthe evolution experienced by healthresearch in our region.

THE STUDY PERIOD AND THE ROLE OF GOVERNMENT

The period covered, 1973–1992, isrelatively long for studies of this type.This length makes it easier to observewith some clarity a number of aspectsof the evolution of health-related sci-entific activity in the six study coun-tries. The period also appears uniquein the history of Latin America, be-cause scientific production underwenta dramatic upsurge in terms of bothquality and quantity, and recognitionof the importance of science and tech-nology for socioeconomic develop-ment was further consolidated. Thisrecognition was manifested in theearly part of the study period by cre-ation in most countries of entities forplanning scientific and technologic de-velopment. Also, in those countriesthat already had such entities, theirdesignations were generally changedfrom “research boards” to “scientific-technologic development boards,” in-dicating a new and broader sphere ofaction.

During this period a number of im-portant economic, social, political, andinstitutional transformations tookplace. The region experienced eco-nomic crises, adjustments, and re-newal—as well as deterioration and,ultimately, recovery of democratic in-stitutions. All of these macro processeshave, in the final analysis, affected de-velopment of science and technologyin the region and can be related, atleast hypothetically, to some of theresults of this study.5

122 Pellegrini et al. • Production of scientific articles on health in six Latin American countries

5 While the roles of a number of internal and exter-nal factors are perpetually debated (4 ), it is gener-ally recognized that both internal and external fac-tors play a role in the development of scientificactivity (5). The former include factors of a logical-cognitive nature, because in all historical periodsthe ultimate progress of science has dependedupon the cognitive material accumulated duringthe preceding period. Factors of an extrinsic natureinclude those related to the material life of soci-eties, such as levels of socioeconomic develop-ment, education, and culture. To more accuratelyanalyze the extent to which each of these factorsexerts an influence, it will be necessary to conductmore in-depth studies designed specifically forthat purpose. By way of contrast, the current studyis intended to describe general trends and to for-mulate a number of hypotheses regarding factorsthat may be correlated with those trends.

The same period also saw a processof health research “demedicalization.”That is, health research ceased to beconducted primarily by physiciansworking within a health service settingand broadened its scope in terms ofproblems researched, disciplines in-volved, approaches employed, typesof research performed, professionalbackground of the researchers, and in-stitutional settings where research wasconducted.

It is also relevant to note that at pres-ent the status of science and technol-ogy in the health field is beset by anumber of contradictions. On the onehand, there is a strong demand fornew knowledge that will contribute togreater understanding of the complexhealth situation and support processesof health sector reform currently un-derway in most countries of the re-gion. At the same time, biomedicine’sdynamic development requires thatgroups working actively in this fieldmake an enormous effort to keep theircommand of strategically importantknowledge and technologies current,not only to resolve health problemsbut also to advance other sectorswhere such technologies may be ap-plied. On the other hand, preciselywhen these needs are becoming mostcritical, it is possible to observe a sig-nificant decline in the resources avail-able for research in a number of coun-tries. In large measure, this is the resultof reductions in public expenditures,which constitute the primary source offunds for health research.

There is widespread recognition ofthe pressing need for government torefrain from limiting its role to that offunder and implementor of researchand to expand its regulatory role in thefield of scientific and technologic ac-tivity, spearheading the process ofcalling together the various actors in-volved in such activities in order to de-fine goals in a coordinated manner.However, we see on the contrary thatscience and technology boards, thegovernment agencies charged withtaking on such responsibilities in thecountries of the region, are experienc-ing reductions not only in their human

and financial resources but in theirpolitical strength.

This study is one in a group ofPAHO activities directed at cooperat-ing with countries of the Americas inthe process of overcoming these chal-lenges and contradictions. Specifically,it seeks to call attention to some of thecharacteristics of scientific productionduring the study period, thus con-tributing to improved understandingof the current situation and historictrends, and thereby helping to providea stronger base for policies in the fieldof health science and technology.

METHODOLOGY

This study employed the ISI data-base referred to above. This databaseis quite selective and includes approx-imately 3 500 journals, including thosemost important in diverse scientificfields. According to its creator, E.Garfield, its selective coverage is duenot only to economic limitations butalso to the fact that in any field thenumber of truly important journalsthat are frequently read and cited isrelatively small, reflecting a preceptknown as “Bradford’s law” (6). In theISI database itself, which is alreadyquite selective, more than 50% of thearticles and 80% of the citations appearin only 500 journals. Accordingly, thisdatabase does not permit one to de-velop an inventory of the sum total ofscientific production in the six studycountries, but instead provides a basisfor inventorying a portion of that pro-duction circulating in the internationalliterature that is deemed of the highestquality by those responsible for thisdatabase.6

The ISI database used, known for-mally as the ISI Database Science Indi-cators File, includes the Science Cita-tion Index, the Social Sciences CitationIndex, and the Arts and Humanities

Citation Index. To cover all sciencefields, this database uses 89 categorycodes (“CatCodes”). Each of the codesrelates to a group of journals, and it isassumed that all of the articles pub-lished in those journals pertain to thatcode. We selected 38 of these codes inorder to delimit the area of science per-taining to human health. We then sub-divided this set into category codesdealing primarily with biomedical,clinical, and public health research (7).During the period in question, the 38codes included 1 975 journals, fromwhich data on the articles analyzed inthis study were obtained.

The six countries selected—Argen-tina, Brazil, Chile, Cuba, Mexico, andVenezuela—have in the past ac-counted for almost 90% of the litera-ture relating to scientific production inLatin America and the Caribbean,with regard to both science in generaland health science in particular (6). Itshould be noted that Cuban scientificproduction recorded in the database isquantitatively smaller than that of fourother countries not included in thisstudy (Colombia, Jamaica, Peru, andUruguay). However, Cuba’s inclusionis justified because of the particularcharacteristics of science and technol-ogy policies in that country.7 Over the20-year study period, 32 Latin Ameri-can health journal titles (29 from thesix study countries) were included inthe ISI database, which annually in-cluded between 17 and 21 such titles(Annex 1).

For the entire period of 1972–1992,72 899 published entries by primary(first-listed) authors residing in one ofthe six countries were obtained. A de-cision was then made to work exclu-sively with research articles, to the ex-clusion of editorials, notes, letters,reviews, meeting summaries, etc.,which reduced the number of entries


6 The countries of Latin America and the Caribbeanthat in the 1970s contributed approximately 1% ofthe articles to the database increased their partici-pation to 1.5% during the 1980s (6).

7 During the study period, Cuba consolidated a sci-ence and technology planning model based on theproblems faced at various levels (national, sectoral,local, etc.), one defined by corresponding govern-ment agencies under the coordination of the Acad-emy of Science.

to 42 292. The first year of the ISI data-base (1972) was excluded for reasonsof consistency and uniformity, andthis further reduced the number ofarticles to a total of 41 238.

The data recorded for each articleincluded its year of publication, title,authors’ names, authors’ respective in-stitutions and countries of residence,journal in which the article was pub-lished, and (for articles published after1982) the number of citations receivedfrom other sources in each year fol-lowing the article’s publication.

For the most part, the analysis pre-sented below should be regarded as aninitial approximation limited to identi-fying major trends of a general naturefor the entire health research field overa period of 20 years. However, to helpexplore the potential for conducting astudy in greater depth, trends of scien-tific production in the field of epidemi-ology were also assessed. From the ISIdatabase list of nearly 40 journals deal-ing with public health research we se-

lected 11 international journals 8 thatwe considered to be of greatest impor-tance. A review was conducted of thedata registered on the 570 articles pub-lished by those journals in 1973–1992whose first author resided in one ofthe six study countries. In this way, 95articles on epidemiology were selectedfor more detailed analysis.

PRINCIPAL FINDINGS ANDDISCUSSION

As noted above, in 1973–1992 pri-mary authors residing in the six studycountries published a total of 41 238

articles listed in the journals selectedfrom the ISI database. Of these, 38.3%were published in the first 10-year period (1973–1982) and 61.7% in the second. In comparing production by 5-year periods (Table 1), the followingincreases were found: in 1978–1982 atotal of 9 417 articles appeared (48%more than the 6 377 published in1973–1977); in 1983–1987 the numberrose to 11 576 (23% more than in1978–1982); and in 1988–1992 it rose to13 868 (20% more than in 1983–1987).

It is interesting to note that this evo-lution bears no relation to the numberof health-related journals published inthe six study countries that were in-dexed in the ISI database, a numberthat remained more or less stablethroughout the study period and thatrespectively stood at 17, 21, 20, and 18journals in each of the successive 5-yearportions of that period.

As may be seen, from the first to the last 5-year period the number ofpublished articles rose by 117% (from


TABLE 1. The 41 238 selected articles published in 1973–1992 whose first authors resided in one of the six study countries, by country, 5-year period (within the 20-year study period), and type of research involved

Number of articles

Type ofCountry research 1973–1977 1978–1982 1983–1987 1988–1992 1973–1992

Argentina Biomedical 1 028 1 050 1 622 1 668 5 368Clinical 1 052 1 576 1 687 1 751 6 066Public health 41 19 48 41 149

Brazil Biomedical 749 1 366 1 946 2 947 7 008Clinical 628 1 540 1 615 2 085 5 868Public health 47 80 264 281 672

Chile Biomedical 237 516 554 692 1 999Clinical 794 821 946 1 228 3 789Public health 18 11 32 49 110

Cuba Biomedical 37 59 97 145 338Clinical 43 77 75 94 289Public health 3 0 2 7 12

Mexico Biomedical 289 446 669 905 2 309Clinical 1 012 1 264 1 392 1 372 5 040Public health 23 30 31 40 124

Venezuela Biomedical 215 277 330 264 1 086Clinical 144 283 256 288 971Public health 17 2 10 11 40

All Biomedical 2 555 3 714 5 218 6 621 18 108Clinical 3 673 5 561 5 971 6 818 22 023Public health 149 142 387 429 1 107

Total 6 377 9 417 11 576 13 868 41 238

8 The 11 journals selected: American Journal ofEpidemiology, American Journal of Public Health,American Journal of Tropical Medicine and Hygiene,Bulletin of the World Health Organization, Epidemiol-ogy and Infection, International Journal of Epidemiol-ogy, Journal of Clinical Epidemiology, Medical Care,Preventive Medicine, Public Health Reports, andTransactions of the Royal Society of Tropical Medicineand Hygiene.

6 377 to 13 868), with variations occur-ring in the rate of growth. These varia-tions could be related to the instabilityof resources available for research thatwas observed during the period. Thegreatest increase in the number of arti-cles, which occurred from the first tothe second 5-year period, coincidedwith a rise in economic indicators anda growth of expenditures for scienceand technology within the region as awhole. Subsequently, during the 1980sand particularly beginning in 1983,most countries saw these expendituresreduced (8), a reduction that was fol-lowed by a period of recovery at theend of the decade.

Country data

Disaggregating the data by country,as in Figure 1, shows that the increasein the number of articles was not con-sistent from one country to another. Inparticular, there was a sharp 110% in-crease in the number of articles byBrazilian authors from the first to thesecond 5-year period, a reduction inthis growth rate to 28% from the sec-ond to the third period, and a slight re-covery to 39% from the third to thefourth period. Chile and Cuba fol-lowed a similar pattern, though the in-crease from the first to the second pe-riod was not so marked.

Also worthy of mention is the smallgrowth observed in Argentina, Mexico,and Venezuela from the third to thefourth period. This pattern was espe-cially marked in the case of Venezuela,which recorded a major increase fromthe first to the second period, followedby a very slight increase from the sec-ond to the third period and negativegrowth from the third to the fourthperiod. It is possible that this pattern isrelated to the stringent policies of eco-nomic adjustment and a decline in pub-lic expenditures that were observed inthese countries in the late 1980s.

As already indicated, authors in thesix study countries selected producedclose to 90% of the health-related arti-cles from Latin America and theCaribbean that were recorded in theISI. Within this group, a large concen-

tration of articles (61%) was found tobe produced by authors from one oftwo countries; that is, 28.1% of the pri-mary authors resided in Argentinaand 32.9% lived in Brazil. This concen-tration exhibited an upward trendduring the study period. For example,if the two delimiting years are consid-ered, 57% of the articles in 1973 wereby authors from Argentina and Brazil,as were 68% of the articles in 1992.

It is also interesting to note a changethat has occurred in the relative num-bers of articles published by Argentineand Brazilian authors. Argentines ac-counted for over a third of all the 1973articles (38% of the total), twice theBrazilian share (19%). By 1992, how-ever, these positions had been re-versed, with Brazilians accounting foralmost half the total number of articles(45%) and Argentines for only 23%.This change began to emerge from thefirst to the second five-year period,when Brazilians went from contribut-ing 22% of the total number of articlesproduced by the six countries to 32%,while the number of articles originat-ing in Argentina dropped from 33% to 28%. This reversal was attributable

both to the sizable 110% increase inBrazilian article production from thefirst to the second period and to theArgentines’ poor performance (an in-crease of only 25%, which was lowerthan the average 48% growth recordedby residents of the six study coun-tries), a situation that could be linkedto political and institutional instabilityand to emigration of scientists fromArgentina during the mid-1970s.

Production by type of research

As Table 2 shows, published re-search articles in the three areas of bio-medicine, clinical medicine, and pub-lic health were not evenly distributedamong the six study countries. Clinicalresearch, the type of research mostcommonly practiced in Latin Americaand the Caribbean, was fairly evenlydistributed; but biomedical and publichealth research were concentrated pri-marily in Brazil and Argentina, whoseauthors accounted for almost 70% ofthe articles published in those areas.Moreover, authors in Brazil accountedfor 60.7% of the 1 107 public health


FIGURE 1. Percentage growth (by five-year period over the years 1973–1992) in the numberof articles appearing in the selected health-related journals whose primary authors residedin one of the six study countries

Per

cent

120

100

80

60

40

20

0

–20Argentina TotalBrazil Chile Cuba Mexico Venezuela

Percent growth from one five-year period to the next:From the first to the secondFrom the second to the thirdFrom the third to the fourth

Source: Institute for Scientific Information, Philadelphia, Pennsylvania, U.S.A.

articles, a share that exhibited an up-ward trend during the study period.9

The latter figure is of some concern,as public health research (i.e., researchon health systems and health status)should be more broadly represented inall countries, it being an element es-sential for decision-making and devel-opment of health policies affecting awide range of health problems andpopulation groups.

As Table 3 shows, the vast majorityof the research articles were in theareas of biomedicine or clinical medi-cine rather than public health, indicat-ing that the individual focus continuesto prevail over the population ap-proach. Over the entire study period,only 2.7% of the articles were classifiedas being in the area of public health re-search, as compared to 53.4% in thearea of clinical research and 43.9% inthe area of biomedical research. How-ever, over the 20-year period an up-ward trend was noted in the share ofarticles devoted to public health andbiomedicine, matched by a corre-sponding decline in the share devotedto clinical research.

This tendency appears to parallel,with some delay, world trends previ-

ously observed in the international liter-ature. In 1982 PAHO conducted a studyon trends in scientific production, com-paring the content of world medical lit-erature with that of Latin Americanmedical literature (9). For the former,the Index Medicus (IM) of the UnitedStates National Library of Medicine wasused, while for the latter the IndexMedicus Latinoamericano (IMLA) ofthe Latin American and Caribbean Cen-ter for Information on the Health Sci-ences (BIREME) was used.

In the world literature surveyed, thesubject areas reflecting the greatest de-gree of expansion between 1966 and1980 were health-related physical sci-ences (+193%), biological sciences(+143%), and public health (+148%),while growth of all the scientific litera-ture examined in this period was 51%.Comparing the profile of productioncatalogued in the IM for 1980 and inthe IMLA for 1979–1982, it was ob-served that, of the 14 MeSH (medicalsubject headings), the large majority(42%) of the articles from the IMLAwere included under the heading “ill-nesses,” while in the IM only 15% ofthe articles were classified under thatheading. This could indicate a greater

relative development of clinical re-search, i.e., research on diseases, inLatin America as opposed to study ofunderlying biologic phenomena. Infact, the share of production accountedfor by health-related articles in the bi-ologic and physical sciences wasfound to be greater in the world litera-ture (respectively 12% and 5% of theIM entries, as compared to 9% and 1%of the entries in the IMLA).

The study cited (9) concludes thatthe pattern of health-related publica-tion in Latin America over the 1979–1982 period came closer to resemblingthe pattern observed in the world liter-ature in the mid-1970s than that foundin 1980. What was observed in thisstudy cannot be compared to the find-ings of the current study, however, asthe former analyzed what was beingpublished in the world and LatinAmerican literature, with each article’scountry of origin being regarded as thecountry where the journal was pub-lished rather than the primary author’splace of residence, while the currentstudy analyzes the publication patternsof authors residing in the six studycountries. However, the upward trendobserved in articles on biomedical re-search and the decrease in those deal-ing with clinical research, as seen inthe current study, indicates a move-ment of health-related scientific pro-duction by authors in the six studycountries toward the pattern found inthe world literature of 1980.

Although this trend was observed inall six study countries, the analyses ofproduction curves by country, type ofresearch, and five-year periods revealinteresting country-by-country varia-tions (see Table 1). Thus, in Argentinathe share of research articles classified


TABLE 3. Percentage distribution by type of research (biomedical, clinical, or public health)of study articles published in 1973–1982, 1983–1992, and 1973–1992

Years

Type of research 1973–1982 1983–1992 1973–1992

Biomedical 39.7 46.5 43.9Clinical 58.5 50.3 53.4Public health 1.8 3.2 2.7


TABLE 2. Percentage distribution by country of the study articles published in the areas ofbiomedical, clinical, and public health research, 1973–1992

Type of research

Country Biomedical Clinical Public health

Argentina 29.6 27.5 13.5Brazil 38.7 26.6 60.7Chile 11.0 17.2 9.9Cuba 1.9 1.3 1.1Mexico 12.7 22.9 11.2Venezuela 6.0 4.4 3.6


9 This concentration bears no relation to the Brazil-ian origins of the only two public health journalsproduced in the six countries and indexed by theISI between 1972 and 1992—these being the Revistadel Instituto de Medicina Tropical and the Revista deSalud Pública [Journal of the Institute of Tropical Med-icine and Journal of Public Health]. During the periodstudied, these two journals published 951 articlesby authors from the six countries, 429 (45%) ofthem by Brazilian authors. As will be seen, of thetotal number of articles on epidemiology pub-lished in the 11 selected international journals, 66%(63/95) were by Brazilian authors.

as biomedical (48%) was nearly thesame as the share classified as clinical(50%) during the first 5-year period.However, the next 5-year periodshowed a major rise in the number ofclinical research articles while thenumber of biomedical research articlesremained constant, although the latterrecovered its previous share in subse-quent periods. This profile suggeststhat developments in the mid-1970smay have led to an expansion of clini-cal research that was not matched bybiomedical research until the 1980s.

In the cases of Brazil and Cuba, therate of growth in biomedical researchliterature was greater than thatrecorded for clinical research litera-ture, and the period ended with cleardomination by the former. In Cuba,the study data indicate a major growthof biomedical research over the lastdecade, most likely owing to effortsmade by that country to promote de-velopment of biotechnology.

In contrast, Chile and Mexico (espe-cially the latter) exhibited more endur-ing traditions of clinical research. How-ever, even though our data indicateclinical research remained predomi-nant in both countries, the numbers of biomedical research articles grewslightly faster over the 20-year studyperiod than did the numbers of clinicalresearch articles.

Venezuela was the only countrywhere biomedical research appearedto predominate in the first five-yearperiod, but where the number of clini-cal research articles by local authorsexceeded the number of biomedicalresearch articles by such authors in1988–1992. In any case, the sharp de-cline of biomedical research articlesfrom 1983–1987 to 1988–1992 is per-haps indicative of this discipline’s vul-nerability to the economic crisis and toadjustment policies implemented dur-ing the period involved.

Collaborative scientific work

One development reflecting a certain“modernization” of scientific activityin the study countries was an increasein collaborative work. Specifically, a

relative decrease was found in thenumber of articles published by re-searchers working independently.That is, in the first decade of the studyperiod articles by a single author ac-counted for 17% of the total, but thatshare dropped to 10% in the seconddecade, while articles by six or moreauthors increased from 8% to 15%(Table 4).

This increased tendency toward col-laborative scientific activity was alsoindicated by a tendency toward in-creased collaboration between peopleat different institutions, both domesti-cally and abroad. That is, there was anincrease in the number of articleswhose first author was from an insti-tution in one of the countries selectedand whose remaining authors werefrom other national or international in-stitutions. Overall, during the entire1973–1992 period, 73.4% of the articleswere produced by authors from a sin-gle institution, while 18.8% involvedauthors from national institutionsother than that of the first author (in-ternal interinstitutional collaboration)

and 7.7% included authors from insti-tutions in other countries (external in-terinstitutional collaboration). How-ever, from the first to the second10-year period, the percentage of arti-cles by authors from a single institu-tion declined from 77.6% to 70.9%,while the percentage of articles involv-ing internal interinstitutional collabo-ration increased from 16.2% to 20.4%and the percentage involving exter-nal interinstitutional collaboration in-creased from 6.3% to 8.6%.

In comparing data for the entireperiod by research area and type ofcollaboration (Table 5), a relativelyhigh percentage of biomedical re-search articles (10.1%) appear to haveinvolved external interinstitutionalcollaboration, as compared to 5.8% ofthe clinical research articles and 8.4%of the public health research articles.On the other hand, a higher percent-age of public health research articlesappeared to involve internal interin-stitutional collaboration (25.7%) thandid either the biomedical research arti-cles (18.5%) or the clinical research


TABLE 5. Percentage distribution of the study articles in biomedical, clinical, and publichealth research according to the institutional affiliations of their authors (one institution,more than one institution in the same country, or more than one institution in differentcountries), 1973–1992

Type ofInstitutions

research One local Several local Foreign

Biomedical 71.4 18.5 10.1Clinical 75.5 18.6 5.8Public health 65.9 25.7 8.4


TABLE 4. Percentage distribution of the study articles publishedin 1973–1982 and 1983 –1992 that had one, two, three to five, orsix or more authors

Number of authorsPeriod

per article 1973–1982 1983–1992

1 17 102 25 213–5 50 54≥6 8 15


articles (18.6%). Meanwhile, a slightlyhigher percentage of clinical researcharticles were produced by authorsfrom a single institution (75.5% versus71.4% for biomedical research articlesand 65.9% for public health researcharticles).

Origins of authors and journals

The health-related scientific articlesof the six selected countries do not ap-pear to have been produced by a rela-tively small number of authors. On thecontrary, the 41 238 articles had 20 095primary authors, with most (63%) ofthe latter contributing only one articleand 17% contributing only two. Only6% of the primary authors contributedsix articles or more.

With regard to the countries of ori-gin of the selected journals, a numberof interesting points were noted. Forone thing, almost 30% of the study ar-ticles produced by authors residing in Argentina, Brazil, and Mexico andrecorded in the ISI database appearedin journals published in the primaryauthor’s country of residence. In thecase of Chile, this figure reached alevel of 50%, while for Cuba andVenezuela it dropped to 10%.

Of all the articles published abroadby authors residing in Argentina,Brazil, and Chile, almost 35% ap-peared in journals published in theUnited States, while 60% appeared inEuropean journals (20% in British jour-nals). The U.S. journals were consider-ably more important for Mexican andVenezuelan authors, who respectivelypublished 50% and 45% of their arti-cles in them, as compared to roughly45% in European journals (20% inBritish journals). Cuban authors fol-lowed a different pattern, with almost80% of their articles in the ISI databaseappearing in European journals (14%in British journals and 65% in otherEuropean journals), while a scant 14%appeared in U.S. journals. Annex 2lists the journals publishing the great-est numbers of articles whose primaryauthors resided in each of the six studycountries.

As the foregoing data suggest, rela-tively few of the articles appeared inLatin American or Caribbean journalsthat were published in countries wherethe primary author did not reside. Thisfact can be attributed to the low degreeof exchange existing among thesecountries, at least with regard to ISI-indexed journals. In terms of residence,the authors with the least propensity topublish in Latin American journalsoutside their country were those inMexico (1.6%), Argentina (1.7%), andBrazil (2.7%). Somewhat higher per-centages of authors residing in Chile(5.4%), Venezuela (6.1%), and Cuba(7.8%) published their articles in jour-nals produced in other Latin Americancountries.

Citations per article

The citations received by a particu-lar article are traditionally used as ameasure of that article’s impact. Natu-rally, just as the scientific article is notthe sole product of scientific activity,the citations received by such an arti-cle are likewise not the only measureof its impact, much less of its quality.Bearing in mind the limitations of thisindicator, but recognizing its broad ac-ceptance as an approximate measureof the reception accorded to a study bythe scientific community, an analysiswas made of the citations received byarticles published in 1982–1992 whosefirst authors resided in one of the sixstudy countries.

The ratio of the number of citationsreceived to the number of scientific ar-ticles published in a given period isused as an indicator of the articles’ im-pact in that period. Garfield (6) foundthat in 1981–1993 the average numberof citations per scientific article world-wide for all of the sciences was 7.78,while for scientific articles publishedin Latin America and the Caribbean itwas 3.52—suggesting a relative impact(citations per article from Latin Amer-ica and the Caribbean relative to cita-tions per article worldwide) of 0.45.

In the health sciences, the averagenumber of citations per article au-

thored by residents of the six studycountries was 3.10, a figure approxi-mating that for scientific articles (allsciences) appearing in the region’sjournals. When an effort was made tomeasure the relative impact of thestudy articles by country (the ratio ofthe average number of citations re-ceived by articles from one country tothe average number of citations re-ceived by articles from the six coun-tries), the following results were ob-tained: Argentina, 1.00; Brazil, 0.91;Chile, 0.84; Cuba, 0.50; Mexico, 1.20;and Venezuela, 1.47.

There were also significant differ-ences between the apparent impacts ofarticles in the research areas of bio-medicine, clinical medicine, and pub-lic health. Overall, the apparent im-pact of the biomedical research articleswas considerably higher, the averagenumber of citations received being4.03, as compared to averages of 2.29and 1.61 for the clinical and publichealth research articles, respectively.In terms of relative impact, biomedicalresearch scored 1.30, while clinical re-search scored 0.74 and public healthresearch scored 0.52. Of the 90 995 ci-tations received during the period,60.5% cited biomedical articles, 38%cited clinical articles, and 1.5% citedpublic health articles, while the respec-tive percentages of published articlesin these fields were 46.3%, 50.6%, and 3.1%. All this indicates, at least interms of citations, that articles in thefield of biomedicine had the greatestimpact. It should be noted that publichealth research, with a percentage ofpublished articles far lower than theother two areas, registered an evenlower relative participation with re-gard to citations received.

Production in epidemiology

As noted in the section on method-ology, in order to analyze epidemio-logic research production, 11 publichealth journals considered of great im-portance were selected. The ensuingstudy of articles by residents of the sixcountries that appeared in these jour-


nals is intended to serve only as a pre-liminary and partial analysis of pro-duction in the field of epidemiology,as it is limited to articles published inpublic health journals, to the exclusionof articles on epidemiology publishedoccasionally by journals such asLancet, the Journal of the American Med-ical Association, etc. In the journals se-lected there were 570 articles whosefirst authors resided in one of the sixcountries; of these, 95 were in the fieldof epidemiology.

Brazilian authors produced thegreat majority (66%) of these 95 arti-cles, a situation similar to that foundfor articles on public health in thebroader study (see Production byType of Research, above).

With regard to number of authors,the percentage of articles with a singleauthor (31%) was considerably greaterthan that observed in the broaderstudy, where only 12.8% of the health-related articles had a single author.

Despite this, the findings of theepidemiologic study highlighted theimportance of interinstitutional collab-oration, especially international collab-oration, in this field. In all, only 32% ofthe articles (nearly all by single au-thors) were produced by one or moreauthors affiliated with a single institu-

tion; 26% listed authors affiliated withtwo or more institutions within a par-ticular country (internal interinstitu-tional collaboration); and 42% listedauthors affiliated with institutions indifferent countries (external interinsti-tutional collaboration). In contrast, thebroader study found the respectivepercentages to be 73.4%, 18.8%, and7.7%, indicating that the epidemiolo-gists who published in the 11 selectedjournals had more contact with the in-ternational scientific community thandid the general group of authors pro-ducing health-related scientific articlesassessed in the broader study.

It is also noteworthy that the largemajority of epidemiology articles(83%) by authors from the six studycountries dealt with the epidemiologyof infectious diseases, while only 4%dealt with chronic diseases and 13%dealt with other subjects (especiallytopics related to maternal and childhealth). In order to compare this sub-ject profile to the corresponding profileworldwide, two epidemiology jour-nals generally considered to be leadersin the field were consulted, these beingthe International Journal of Epidemiologyand the American Journal of Epidemiol-ogy. The former journal published atotal of 381 articles in 1987, 1989, and

1991, of which only 98 (25.7%) dealtwith infectious diseases or problemsrelating to maternal and child health.The latter journal published a total of344 articles during the second semes-ters of these years, of which only 66(19.2%) dealt with infectious diseasesor maternal and child health problems.These data indicate that while thestudy of chronic diseases predomi-nates in the world’s epidemiologic lit-erature, epidemiologists within LatinAmerica who successfully publish injournals enjoying wide circulation atthe international level continue to dealalmost exclusively with infectious dis-eases. Of the 381 articles published bythe International Journal of Epidemiology,only three (0.78%) had primary au-thors residing in one of the six studycountries.

With regard to citations received byarticles from the study countries onepidemiology, the 77 study articlespublished between 1981 and 1992 re-ceived 336 citations, giving them anaverage impact index of 4.36 citationsper article, considerably greater thanthat found by the broader study for ar-ticles in the area of public health (1.61),and also greater than that found by thebroader study for all health-relatedscientific articles (3.1).


1. Organización Panamericana de la Salud. La in-vestigación en salud en América Latina: estudio depaíses seleccionados. Washington, DC: OPS;1992. (Scientific publication 543).

2. Pellegrini A. La investigación en salud en cincopaíses de América Latina. Bol Oficina SanitPanam 1993;114(2):142–157.

3. Velho L. Indicadores científicos: aspectos teóri-cos y metodológicos. In: Martínez E, ed. Cien-cia, tecnología y desarrollo: interrelaciones teóricasy metodológicas. Caracas: Editorial Nueva So-ciedad, UNESCO; 1994.

4. Bunge M. Sociología de la ciencia. Buenos Aires:Siglo XXI; 1993.

5. Kedrov B. Acerca de las leyes del desarrollo de lasciencias. Havana: Editorial de Ciencias Sociales;1977.

6. Garfield E. Quantitative analysis of the scien-tific literature and its implications for sciencepolicymaking in Latin America and theCaribbean. Bull Pan Am Health Organ 1995;29(1):87–95.

7. Frenk J, Bobadilla JL, Sepúlveda J, Rosenthal J,Ruelas E. Un modelo conceptual para la inves-tigación en salud pública. Bol Oficina SanitPanam 1986;101(5):477–489.

8. Sagasti FR, Cook C. Tiempos difíciles: ciencia ytecnología en América Latina durante el dece-

nio de 80. Lima: GRADE; December 1985. (Photocopied document.)

9. García JC. Nuevas tendencias de la investi-gación biomédica y social. Washington, DC:Organización Panamericana de la Salud; 1982.(Photocopied document.)

Manuscript received on 30 November 1995. Revised ver-sion accepted for publication on 20 March 1996.

REFERENCES


ANNEX 2. Journals publishing the largest numbers of study articles in 1973–1992 whose primary au-thors resided in the indicated countries.

ArgentinaJournal No. %

Prensa Médica Argentina 1 645 14.2Medicina 1 297 11.2Acta Physiologica Latinoamericana 348 3.0Biochimica et Biophysica Acta (Amsterdam) 233 2.0Acta Physiologica et Pharmacologica Latinoamericana 195 1.7Biochemical and Biophysical Research Communications 101 0.9FEBS Letters (Amsterdam) 98 0.8Comparative Biochemistry and Physiology [B] (Oxford) 94 0.8Biochemistry Journal (London) 87 0.7Molecular and Cellular Biochemistry (The Hague) 85 0.7Neuroendocrinology (Basel) 85 0.7Others (1 110) 7 303 63.1

Total 11 571 100.0

ANNEX 1. The 32 Latin American health journals included in the ISI database during the 1972–1992period.

Country Journal title Articles published (No.)

Argentina Acta Physiologica et Pharmacologica Latinoamericana 238Acta Physiologica Latinoamericana 459Acta Psiquiátrica y Psicológica de América Latina 94Medicina (Buenos Aires) 1 312Microscopía Electrónica y Biología Celular 70Prensa Médica Argentina 1 657

Brazil Arquivos Brasileiros de Psicologia 325Arquivos Brasileiros de Psicologia Aplicada 42Arquivos de Neuro-Psiquiatria 33Brazilian Journal of Medical and Biological Research 686Memórias do Instituto Oswaldo Cruz 345Revista Brasileira de Genética 726Revista Brasileira de Medicina 1 057Revista Brasileira de Pesquisas Médicas e Biológicas 170Revista do Instituto de Medicina Tropical de São Paulo 584Revista de Saúde Pública 449

Chile Archivos de Biología y Medicina Experimentales 208Revista Médica de Chile 2 759

Colombia Archivos de la Sociedad Americana de Oftalmología y Optometría 1Revista Latinoamericana de Psicología 97

Cuba Interferón y Biotecnología 73

Mexico Archivos del Instituto de Cardiología de México 156Archivos de Investigación Médica 701Investigación Médica Internacional 477Medicina (Mexico) 89Patología (Mexico) 150Revista de Investigación Clínica 938Revista Mexicana de Radiología 215Salud Mental 175

Peru Acta Médica Peruana 2

Venezuela Acta Gastroenterológica Latinoamericana 10Investigación Clínica 208


ANNEX 2. (Continued)

BrazilJournal No. %

Revista Brasileira de Medicina 1 049 7.7Revista Brasileira de Genética 688 5.1Brazilian Journal of Medical and Biological Research 679 5.0Revista do Instituto de Medicina Tropical de São Paulo 522 3.9Revista de Saúde Pública 429 3.2Arquivos Brasileiros de Psicologia 316 2.3Memórias do Instituto Oswaldo Cruz 307 2.3Comparative Biochemistry and Physiology [B] (Oxford) 173 1.3Revista Brasileira de Pesquisas Médicas e Biológicas 167 1.2Biochimica et Biophysica Acta (Amsterdam) 134 1.0Others (1 329) 9 084 67.1

Total 13 548 100.0

ChileJournal No. %

Revista Médica de Chile 2 742 46.5Archivos de Biología y Medicina Experimentales 197 3.3IRCS Medical Science Biochemistry 162 2.7FEBS Letters (Amsterdam) 65 1.1Comparative Biochemistry and Physiology [B] (Oxford) 57 1.0Contraception (Stoneham, Massachusetts) 52 0.9General Pharmacology (Oxford) 48 0.8Biochemistry International (Sidney) 40 0.7Biochimica et Biophysica Acta (Amsterdam) 39 0.7Cellular and Molecular Biology (Elmsford, New York) 37 0.6Neuroscience Letters (Limerick) 37 0.6Others (691) 2 420 41.0

Total 5 896 100.0

CubaJournal No. %

Interferón y Biotecnología 68 10.6Neoplasma (Bratislava) 36 5.6Allergologia et Immunopathologia (Madrid) 35 5.5Revista Clínica Española (Madrid) 24 3.8Biotechnology Letters 17 2.7Folia Microbiologica (Prague) 14 2.2Acta Biotechnologica 11 1.7Acta Paediatrica Hungarica (Budapest) 10 1.6Haematologia (Budapest) 10 1.6Revista do Instituto de Medicina Tropical de São Paulo 9 1.4Sangre (Barcelona) 9 1.4Acta Diabetologica Latina (Milan) 8 1.3Activitas Nervosa Superior 8 1.3American Journal of Tropical Medicine and Hygiene 8 1.3Revista de Investigación Clínica (Mexico) 8 1.3Others (191) 364 57.0

Total 639 100.0

MexicoJournal No. %

Revista de Investigación Clínica 920 12.3Archivos de Investigación Médica (Mexico) 701 9.4Investigación Médica Internacional 394 5.3Revista Mexicana de Radiología 212 2.8Salud Mental (Mexico) 166 2.2Archivos del Instituto de Cardiología de México 145 1.9Patología (Mexico) 107 1.4Journal of Rheumatology (Toronto) 85 1.1Biochimica et Biophysica Acta (Amsterdam) 81 1.1Medicina — Revista Mexicana 79 1.1Others (934) 4 583 61.3

Total 7 473 100.0

Con objeto de observar las tendencias de la investigación sobre salud en América La-tina, se analizó la producción de artículos resultantes de investigaciones biomédicas,clínicas y de salud pública originadas en Argentina, Brasil, Chile, Cuba, México y Ve-nezuela de 1973 a 1992, inclusive. De la base de datos del Institute for Scientific Infor-mation (ISI) se recuperaron 41 238 artículos con primeros autores residentes en esospaíses. Esta producción se analizó por área de estudio, tipo de investigación, país, nú-mero de autores e instituciones que participaron en la investigación, y citaciones reci-bidas por cada artículo. Se hizo también un análisis de 95 artículos en epidemiologíaseleccionados de 570 publicados por autores de los seis países en 11 revistas de saludpública de gran prestigio internacional.

Los resultados muestran que el número de trabajos publicados aumentó 117% entreel primero y último quinquenios. La investigación clínica tuvo la distribución más ho-mogénea entre los países y la de salud pública, la más concentrada (60,7% se originóen el Brasil). Los artículos de investigación biomédica y de salud pública tuvieron unincremento relativamente mayor que los de investigación clínica a lo largo del período. Se observó una disminución relativa de artículos de un solo autor —lo quesugiere trabajo en equipo más frecuente— y un aumento de artículos con autores vin-culados a dos o más instituciones nacionales o extranjeras, lo cual indica mayor co-operación entre instituciones y países. En promedio, las citaciones recibidas por ar-tículo fueron tres, lo que corresponde a menos de la mitad de las que reciben los artículos de la base del ISI (7,78). En cuanto al subconjunto de los 95 artículos deepidemiología, la gran mayoría (96%) trataban de enfermedades infecciosas o del área maternoinfantil, mientras que en la literatura internacional 78% de los artículosestaban dedicados a las enfermedades crónicas. En este conjunto se observó máscooperación con instituciones extranjeras y un índice de 4,36 citaciones por artículo.

Se concluye que, pese a las limitaciones inherentes a este tipo de estudio, se pudie-ron notar algunas tendencias generales del desarrollo de la investigación en los seispaíses con mayor producción científica de América Latina y formular hipótesis sobrelos factores que determinan esas tendencias. Tomados con cautela, los resultados deestudios de este tipo pueden ser de gran valor al definir políticas de ciencia y tecno-logía de salud.

RESUMEN

Producción de artículoscientíficos sobre salud en seis

países de América Latina,1973 a 1992


ANNEX 2. (Continued)

VenezuelaJournal No. %

Investigación Clínica 205 9.8Biochimica et Biophysica Acta (Amsterdam) 67 3.2Current Therapeutic Research, Clinical and Experimental (New York) 47 2.2American Journal of Tropical Medicine and Hygiene (Lawrence, Kansas) 28 1.3Investigación Médica Internacional 27 1.3American Journal of Physiology (Bethesda, Maryland) 25 1.2Comparative Biochemistry and Physiology [A] (Oxford) 24 1.1Molecular and Biochemical Parasitology (Amsterdam) 23 1.1Biochemical and Biophysical Research Communications (New York) 22 1.0Mycopathologia (Dordrecht) 22 1.0Biotropica 20 1.0Journal of General Physiology (New York) 20 1.0Others (525) 1 566 74.7

Total 2 096 100.0

Production of scientific articles on health in six Latin American ...

Documents

Transcript of Production of scientific articles on health in six Latin American ...