A comparison of the accuracy of clinical decisions based on full-text articles and on journal abstracts alone: a study among residents in a tertiary care hospital.

2013·0.15 MB·English

by  Marcelo, Alvin

Original EBM research
A comparison of the accuracy of clinical decisions based on
full-text articles and on journal abstracts alone: a study
among residents in a tertiary care hospital
Alvin Marcelo,1,2 Alex Gavino,3 Iris Thiele Isip-Tan,4
Leilanie Apostol-Nicodemus,5 Faith Joan Mesa-Gaerlan,6
Paul Nimrod Firaza,1 John Francis Faustorilla Jr,1 Fiona
M Callaghan,3 Paul Fontelo3
10.1136/eb-2012-100537 Abstract questions.3–8 Journal abstracts may have become the de
Background Many clinicians depend solely on journal facto resource for health professionals wanting to prac-
abstractstoguideclinicaldecisions. tice EBM because they are easy to read and are easily
1NationalTelehealthCenter, Objectives Thisstudyaimstodetermineiftherearedif- accessibleanywhere.2579
UniversityofthePhilippines,
ferences in the accuracyof responsesto simulated cases Althoughabstractsarecommonlyutilisedforclinical
Manila,Philippines
2DepartmentofSurgery, between resident physicians provided with an abstract decisions, caution should be made in using them
onlyandthosewithfull-textarticles.Italsoattemptsto because they may not completely reflect the entire
PhilippineGeneralHospital,
Manila,Philippines describetheirinformation-seekingbehaviour. article.7 Studies by Pitkin et al10 11 and Peacock et al12
3NationalLibraryofMedicine, Methods Seventy-seven resident physicians from four identified abstracts that contained data which were dif-
NationalInstitutesofHealth, specialty departments of a tertiary care hospital com- ferent or missing in the full-text. High-impact factor
Bethesda,Maryland,USA pletedapaper-basedquestionnairewithclinicalsimula- journals had abstracts that failed to include harm
4DepartmentofMedicine, tion cases, then randomly assigned to two intervention despite being mentioned in the main article.13 A study
PhilippineGeneralHospital, groups—access to abstracts-only and access to both by Berwanger et al8 found that the abstracts of rando-
Manila,Philippines
abstracts and full-text. While having access to medical mised controlled trials from major journals were
5DepartmentofFamilyand
literature, theycompleted anonline versionofthe same reported with suboptimal quality. Moreover, abstracts
CommunityMedicine,Philippine
questionnaire. are also subject to authors’ biases which may mislead
GeneralHospital,Manila,
Findings The average improvement across departments thereaders.14
Philippines
6DepartmentofEmergency was not significantly different between the abstracts- Efforts have been made to improve the quality and
Medicine,PhilippineGeneral only group and the full-text group (p=0.44), but when accuracy of journal abstracts since they are often the
Hospital,Manila,Philippines accounting for an interaction between intervention and most commonly read part of an article—if not the only
department, the effect was significant (p=0.049) with part read.10 11 15 In 1987, the Ad Hoc Working Group
improvement greater with full-text in the surgery for Critical Appraisal of the Medical Literature intro-
Correspondenceto:
department. Overall, the accuracy of responses was duced a seven-heading format (Objectives, Design,
DrAlexGavino,
greater after the provision of either abstracts-only or Setting, Patients, Interventions, Measurements and
OfficeofHighPerformance
full-text (p<0.0001). Although some residents indicated Conclusion) for structured abstracts.16 Variations in
Computingand
that ‘accumulated knowledge’ was sufficient to respond structured abstracts include the eight-heading format
Communications,ListerHill
NationalCenterforBiomedical to the patient management questions, in most instances proposed by Haynes et al,17 IMRAD18–20 (Introduction,
Communications,National (83%ofcases)theystillsoughtmedicalliterature. Methods, Results and Discussion), and more recently,
LibraryofMedicine,8600 Conclusions Our findings support studies that doctors BMJ’s pico format21 (Patient, Intervention, Comparison
RockvillePike,Bethesda, willuseevidencewhenconvenientandcurrentevidence and Outcome). Structured abstracts tend to be longer
MD20894,USA; improved clinical decisions. The accuracy of decisions than traditional ones but they also tend to have better
[email protected] improved after the provision of evidence. Clinical deci- content, readability, recall and retrieval.14 18 22–25 Aside
sions guided by full-text articles were more accurate from structuring, ‘quality criteria’ and guidelines have
than those guided by abstracts alone, but the results been developed to assist authors in preparing
seem to be driven by a significant difference in one abstracts.2627
department. Most of the research on journal abstracts focus on
their quality compared to the full-text,10 11 13 22 23 or
based on their structure.14 20 24 Given the tendency of
Introduction physicians to use abstracts for evidence, there is a need
Background to evaluate their reliability in clinical decision making.
Broad estimates show that a specialist would need A study by Barry et al5 looked at the effect of abstract
almosttwomillionpiecesofdatatopracticegoodmedi- format on physicians’ management decisions. However,
cine.1 To keep updated and applyevidence-based medi- weareunabletofindstudiesthatcompareclinicaldeci-
cine (EBM) in practice, a physician must critically sions between those with access to abstract-only or
appraise full-text articlesto guide theirclinical decision full-text.
making.2 However, owing to limited access to full-text Theprimaryobjectiveofthisstudywastodetermine
articles, inadequate critical appraisal skills or lack of whether there is a significant difference in the accuracy
OpenAccess time to read the entire article, many clinicians depend of the clinical decisions made on simulated cases by
Scantoaccessmore
freecontent solely on journal abstracts to answer clinical residents with accessto full-text articles and thosewith
48
Evidence-BasedMedicineApril2013|volume18|number2|
Original EBM research
access to abstract-only. The specific objectives were: preintervention phase, the residents were randomly
(1) to compare the effect of access to abstracts-only or assigned to one of two groups—access to ‘full-text’ or
full-textarticlesontheclinicaldecision-makingofresi- ‘abstracts-only,’stratifiedbydepartment.Thesameclin-
dents; (2) to determine whether providing either the ical cases and questions in the preintervention phase
abstractorfull-textarticleincreasedtheaccuracyofclin- were presented to the residents using the online version
ical decisions and (3) to characterise the information- of the questionnaire to simulate real-time access to
seeking behaviour and use of information resources by medical literature. A 20-min time limit was allotted for
residentsoffourdepartmentsinatertiarycarehospital. eachquestionbothforthepaper-basedandonlineques-
tionnaire. The journal material provided, whether
Methods abstracts-only or full-text, was dependent on their
Ethicsreview assigned group. If the resident assigned to the
Theresearchprotocolwassubmittedfortechnicalreview abstracts-onlygroupclickedonthelinktothefull-text,a
to the Research Grants Administration Office of the promptsaying,‘Sorry,full-textisunavailable’appeared.
University of the Philippines Manila and for ethical Although access to either journal abstracts or full-text
evaluation to the Institutional Review Board, both of articles on the online version was available to all resi-
whichapprovedthestudy. dents, they had the option of not using any resource at
all.Theresidents’actionsregardingtheuseornon-useof
Prestudyclinicalcasedevelopment medical literaturewere recorded. Mouseclicks related to
A physician consultant from each of four clinical theresidents’requestforthearticles’abstractsorfull-text
departments (Surgery, Internal Medicine, Emergency wereloggedintheserver.Theaccuracyofresponsewasa
Medicine and Family and Community Medicine) pre- measure of correctness of residents’ answer when com-
paredfivesimulatedclinicalcasesofvaryingcomplexity paredwiththeanswers(‘goldstandard’)providedbythe
and the corresponding clinical questions to assess the consultants.Thesameconsultantswhopreparedtheclin-
residents’ management decisions. They searched ical cases and questions evaluated the accuracy of the
PubMed for at least three recent (from 2007 onwards) residents’ answers. A correct response was scored as ‘1’
journalarticlesthatweredeemedrelevantforeachcase. and an incorrect response scored ‘0’. Incomplete
‘Gold standard’ answers to the clinical questions were responses were rated as inaccurate and scored ‘0’.
based on the journal articles and other relevant infor- Resident responses were anonymised in both the paper
mation(applicabilityandappropriatenesstolocalcondi- andonlineversions.
tions, available resources and practice environments). A
paper-basedquestionnairewasusedforthepreinterven- Dataanalysis
tion assessment while an online version was used Inordertoaccountfortherepeatedmeasuresnatureof
duringthe intervention phase toallowaccesstojournal the data (physicians answered multiple questions), we
abstractsorfull-textarticles. fitmixedeffectslogisticregressionmodelswithdepart-
ment type, intervention, and the interaction between
Studyparticipantsandsetting department and intervention as independent variables,
Seventy-sevenresidentphysiciansfrom the fourclinical andaccuracyoftheresponseasthedependentvariable.
departments (above) at the Philippine General Hospital Unless otherwise stated, all results were based on this
participated in the study. The Philippine General model. Resident year level was also considered as a
Hospital is a 1500-bed tertiary care, state-owned, refer- predictorinthemodelbutwasnotfoundtobesignifi-
ral centerand teaching hospital of the Universityof the cant and was dropped. We also fit a model with an
Philippines College of Medicine, College of Nursing, interaction between intervention and department. For
College of Dentistry and allied colleges. It is the largest univariate analysis, we used the nonparametric
government hospital with a yearly patient load of Wilcoxon two-sample tests and Fisher exact tests, as
600000, mostly indigent patients. Fourteen clinical appropriate.29 All analyses were performed using R
departmentsofferresidencyandfellowshiptraining. StatisticalSoftware.30
Studydesign Results
Duringtheprestudybriefing,theresidentswereinformed Participantprofile
that they were to answer questions related to the case Seventy-seven residents from the departments of
simulationsandthattheycouldaccessreferencearticles Surgery (n=20), Internal Medicine (n=20), Emergency
if needed during the online phase of the study. Written Medicine (n=20) and Family and Community Medicine
consent was obtained and paper-based case simulations (n=17) participated in this study. Table 1 shows the
weregiventoeachresidenttoreplicatethehospitalscen- descriptionofthestudyparticipantsbydepartment.
arioofpaperpatientrecords.Afterreadingthecasesimu-
lations, they were asked to respond to five clinical Comparingtheeffectofabstract-onlyandfull-text
questionsand indicatewhether theyconsidered alitera- accessontheaccuracyofclinicaldecisionmaking
ture search was needed to answer the questions or Thefirstobjectivewastoanswerthequestion:Istherea
accumulated knowledge28 was adequate. Accumulated significant difference in the accuracy of responses of
knowledgewasdefinedinthisstudyastheresidents’per- residentsintheabstract-onlygroupandfull-textgroup?
sonal knowledge base accumulated through years of Overall, there was no significant difference between the
formaleducation,training,researchofthemedicallitera- interventions (p=0.44). Post-hoc power of the experi-
ture and clinical experience. Immediately after the ment to detect an overall difference between the
49
Evidence-BasedMedicineApril2013|volume18|number2|
Original EBM research
Table1 Characteristicsofparticipatingresidenttraineesbydepartment
Surgery, Internalmedicine, Emergencymedicine, Familyandcommunitymedicine,
Characteristics n=20 n=20 n=20 n=17
Meanage,years(SD*) 29.2(2.7) 28.1(1.6) 30.8(3.3) 30.5(2.4)
Gender,n(%)
Female 5(25) 3(15) 6(30) 13(76)
Male 15(75) 17(85) 14(70) 4(24)
Yearsinresidencytraining,n(%)
1 6(30) 7(35) 7(35) 11(65)
2 2(10) 6(30) 7(35) 1(6)
>3 12(60) 7(35) 6(30) 5(29)
*Standarddeviation.
interventions was low and varied from approximately interventionsweresignificant (p<0.0001).Table3shows
44–58%,31 depending on the level of correlation the comparison ofthesepercentagesby departmentand
between the answers within each department. In a thetestsofsignificance.
modelfittoincludeaninteractionbetweenintervention When given full-text articles, the departments of
anddepartment,theinteractionwassignificantsuggest- Surgery,InternalMedicineandFamilyMedicineshowed
ing that intervention effects differed by department significant improvements (p=0.003, 0.03 and <0.0001,
(p=0.03). In that model, accessto full-text was signifi- respectively), while therewas no change for the depart-
cantlybetterthanaccesstoabstracts-only(p=0.049). ment of Emergency Medicine (p=1.0). The differences
We then compared the effect of the interventions amongthedepartmentsweresignificant (p<0.0001)for
withindepartmentsinordertoinvestigatewhichdepart- full-text intervention group. This suggests that full-text
ments seemed to respond differently to the others with was more effective for Surgery, Internal Medicine and
respect to the effect of the interventions. We found no Family Medicine, but not in the Emergency Medicine
significant difference between the interventions for the department. However, the sample size was small (n=10
Internal Medicine, Emergency Medicine and Family or less) at this level. The effect of the abstract-only
Medicine departments (p=0.73, 0.13 and 0.37, respect- intervention seems to have been in a similar direction
ively), but there was a difference between the interven- forall the departments, and no significant difference in
tions for the Surgery department (p=0.02). The OR for effectsacrossdepartmentswasdetected.
each department is given in table 2. The full-text group
had3.6timestheprobabilityofgettingacorrectanswer Information-seekingtrendsoftheresidents
on a case simulation compared to the abstract-only Themajorityoftheresidents(86%)indicatedthatthearti-
group.NotethattheCIforSurgerydoesnotinclude1.0, cles provided in the online version were adequate to
which indicates a significant difference. There were no answer their questions and 77% indicated that they had
differences found between the interventions for any actuallyreadthearticles.Whenaskedwhethertheyused
other department. Power to detect a difference between abstracts-onlyorfull-textarticlestoanswerclinicalques-
the interventions in a specific department was low tionsinactualpractice,53ofthe77residents(69%)indi-
(approx. 17%) because of the reduced sample size in catedthattheyreliedonabstractsmostofthetime,while
each group (n=10). We also investigated whether resi- only24(31%)saidtheywouldreadthefull-textarticle.
dentyearwasasignificantpredictorofclinicaldecision Residents were asked whether or not they felt they
accuracy,butitwasnotsignificantinanymodel. needed extra information in order to answer the ques-
tion correctly. We recorded whether they clicked on the
Accuracyofclinicaldecisionsbeforeandafteraccess links for the abstract or the full-text. We wanted to
toliteraturesearch answer the question: does a perceived need for more
We calculated the mean percentage of accurate information correlate with how often the physicians
responses to the simulated clinical questions before and actuallyaccessedthelinksforabstract-onlyorfull-text?
after each intervention. Overall, mean accuracy For the 157caseswherethe resident indicated that they
increased from 42% to 68% for the abstract-only inter- did not require additional information, there were 131
vention, and 48–75% for the full-text intervention. The (83%) instances where literature was actually accessed
differences between the scores before and after the two (95% CI 77% to 88%). In contrast, out of the 228 cases
where residents indicated that they needed additional
information, there were only 12 (5%) cases where they
Table2 Estimatesoftheoddsofgettinganaccurate
did not actually access literature (95% CI 3% to 9%).
responseafterfull-textinterventioncomparedto
Table 4 shows a summary of whether the resident
abstract-onlyintervention
requestedadditionalinformationandwhethertheyactu-
Department OR(95%CI)
allyaccessedliterature.
Surgery 3.6(1.3to10.3)
Internalmedicine 0.8(0.3to2.2)
Discussion
Emergencymedicine 0.5(0.2to1.3)
The main question we wanted to address in this study
Familymedicine 1.6(0.6to4.7) was whether there is a significant difference in the
50
Evidence-BasedMedicineApril2013|volume18|number2|
Original EBM research
Table3 Comparisonoftheaveragepercentageofaccurateofresponsesbeforeandafterinterventionsandtestsof
significance
Average% Average% Testof Doesthe
ofaccurate ofaccurate significance amountof
responses responses Differencein foreach improvement
before after %before department differby
intervention intervention andafter andoverall department?
N (SD) (SD) (95%CI)* (pvalue)† (pvalue)†
Abstract-only(intervention1)
Department
Surgery 10 48(27) 62(15) 14(0to40) 0.16
Internalmedicine 10 36(18) 66(16) 30(20to60) 0.003 0.06
Emergencymedicine 10 54(21) 72(14) 18(0to40) 0.06
Familymedicine 8 25(14) 75(14) 50(40to60) <0.0001
Overall 38 42(23) 68(15) 27(20to40) <0.0001
Full-text(intervention2)
Department
Surgery 10 60(13) 86(0.16) 26(20to40) 0.003
Internalmedicine 10 46(16) 68(0.1) 22(0to40) 0.03 0.0001
Emergencymedicine 10 64(25) 64(0.23) 0(−20to20) 1.0
Familymedicine 9 20(14) 82(0.21) 62(40to80) <0.0001
Overall 39 48(24) 75(0.2) 27(20to40) <0.0001
*95%CIbasedonWilcoxontest.
†Basedonvaluesfromthelogisticregressionmixedmodel.
clinical decisions between residents who have access to feasible.2However,clinicianswhowanttopracticeEBM
abstracts-onlyandthosewithaccesstofull-textarticles. will also find online manysummaries, reviews and pre-
Overall,ourresultsseemtodemonstratenodifferencein appraised free resources (TRIP Database, ACP Journal
the accuracy of responses between residents provided Club, Cochrane Library, etc) or by subscription
with full-text articles and those with abstracts-only (UpToDate, 5-Minute Clinical Consult, etc). EBM web-
(p=0.4415). When we consider the clustering of physi- sites will have links to these. Many of these resources
ciansbydepartment,wefoundadifferencebetweenthe willhaveapplicationsformobiledevicesliketheiPhone
two interventions (p=0.0494) but further analysis or Android devices. Our observations set the stage for
showed that this difference was observed only in the further research on the role of using abstracts in
department of Surgery (p=0.016). The effects of evidence-basedpractice.Futurestudiesmayincluderan-
abstracts-only and full-text were not significantly dif- domised controlled trials with real-time clinical
ferent for the Internal Medicine, Emergency Medicine decision-makingencounteredatthebedside.
and Family Medicine departments. However, the study EBM encourages the use of timely and relevant
had low power to detect differences between the inter- informationtocomplementtheclinicalacumenofclini-
ventionswithinadepartment. cians.32 We found that the average improvement in the
Our study provides preliminary but useful informa- accuracy of responses across all the departments when
tion related to the use of journal abstracts in evidence- either abstracts or full-text articles were provided was
based practice. We believe this to be the first report significant (p<0.0001 for both interventions). This
involving physicians, that attempted to evaluate how finding supports previous research regarding the role of
abstractsmeasureuptofull-textarticlesinguidingclin- medical literature in improving clinical decisions.33–36
icaldecisions.Thisfindingofferssupportforusing‘con- However,whenindividual departments wereconsidered,
sensusabstracts’(concurringandcorroboratingabstracts there seems to be a significant difference between the
from independently conducted randomised clinical departments in the full-text intervention group
studies and systematic research from meta-analysis and (p=0.0001). This difference in the effect of full-text
systematic reviews that form the basis of clinical evi- between the departments appears to be due to the fact
dence) as a possible alternativewhen access to full-text thattherewasnochangeintheaccuracyofresponsesof
is limited or in other circumstances when it is not Emergency Medicine residents compared to the increase
inscoresfortheotherresidentswhenfull-textwaspro-
vided.Thismaymeanthatfull-textarticleswerebenefi-
Table4 Residents’perceivedneedforadditional cial to Surgery, Family Medicine and Internal Medicine
informationandactualaccessofliterature residents but did not benefit Emergency Medicine resi-
Wasliteratureactually dents. A possible explanation for this is that the
accessed?
Emergency Medicine department is fast paced and resi-
No Yes dentsmaynothavethetimetoreadthefull-textarticle.
Perceivedneedformore No 26(17%) 131(83%) This hypothesis was further supported by data for the
information? Yes 12(5%) 216(95%) abstract-only group where we found no significant dif-
Rowpercentagesaredisplayed. ference between the departments on how the
51
Evidence-BasedMedicineApril2013|volume18|number2|
Original EBM research
intervention improved the accuracy of the responses by accumulatedknowledgewasadequatetoanswerclinical
residents. questions,accessedevidenceanyway.Thisconfirmspre-
Our study also demonstrated some trends in vious findings that easy availability of evidence
information-seeking and utilisation of evidence by resi- encourages the practice of evidence-based medicine.
dents when presented with clinical questions. We Whenclusteredbydepartment,clinicaldecisionsguided
observedthatalthoughresidentsindicatedthataccumu- by full-text articles were more accurate than those
lated knowledge was sufficient to answer the questions, guided by abstracts alone, but this difference can be
in most instances (83.4%), they still accessed the largely attributed to a significant difference in Surgery.
medical literature provided. This observation supports It may be less or not at all in the other three depart-
earlierstudiesthathealthprofessionalswilluseevidence ments but the analysis is not conclusive because of the
fromtheliteraturewhentheyareeasilyaccessibleatthe limited power of this study. Without departmental clus-
timethequestionarises.37 tering, the findings seem to show that they may not be
Morethanathirdoftheresidents(68.8%)whoparti- significantlydifferent.
cipated in this study claimed that they commonly used
abstracts in seeking answers to their clinical dilemma. Funding ThisresearchwassupportedbytheIntramural
Other studies have reported similar observations. A ResearchProgramoftheNationalInstitutesofHealth
study by Haynes et al3 found that two-thirds of clinical (NIH),NationalLibraryofMedicine(NLM)andLister
decisions were influenced by literature even if the full- HillNationalCenterforBiomedicalCommunications
text was not read. Moreover, internists reported that in (LHNCBC).
63% of the articlesthey come across, only the abstracts
were read.4 These findings may even be higher among Disclaimer Theviewsandopinionsoftheauthors
physicians in low- and middle-income countries expressedhereindonotnecessarilystateorreflectthose
because of even more limited availability of full-text oftheNationalLibraryofMedicine,NationalInstitutes
articles. ofHealthortheUSDepartmentofHealthandHuman
Services.
Limitations
The small sample of residents from a tertiary govern- OpenAccess ThisisanOpenAccessarticledistributed
menthospitalinthePhilippineslimitsthegeneralisabil- inaccordancewiththeCreativeCommonsAttribution
ity of the study to the larger medical community. NonCommercial(CCBY-NC3.0)license,whichpermits
Simulatedclinicalcaseswereusedassurrogatetoactual otherstodistribute,remix,adapt,builduponthiswork
clinical encounters that a resident may be presented non-commercially,andlicensetheirderivativeworkson
with. The clinical questions were specific within the differentterms,providedtheoriginalworkisproperly
realmofthedisciplinesandarenotnecessarilycompar- citedandtheuseisnon-commercial.See:http://
able to each other. The residents only answered five creativecommons.org/licenses/by-nc/3.0/
questions which reduced the variation in the study. A
‘learning effect’ was considered to explain the higher
References
score during the intervention phase but was deemed
unlikely because of the short interval period—the resi- 1. PaukerSG,GorryGA,KassirerJP,etal.Towardsthesimulation
ofclinicalcognition:takingapresentillnessbycomputer.AmJ
dents took the online version questions immediately
Med1976;60:981–96.
afterthepreinterventionsession.Furthermore,thisstudy
2. FonteloP.Consensusabstractsforevidence-basedmedicine.
doesnotaddresswhetheraccesstofull-textwould have EvidBasedMed2011;16:36–8.
more impact than access to the abstract in a complex 3. HaynesRB,McKibbonKA,WalkerCJ,etal.Onlineaccessto
case, a case in which the details of a treatment or MEDLINEinclinicalsettings.Astudyofuseandusefulness.
outcomeormagnitudeorsignificancemightaffectprac- AnnInternMed1990;112:78–84.
tice. It also does not address the impact on standard or 4. SaintS,ChristakisDA,SahaS,etal.Journalreadinghabitsof
routine or long-term practice. Finally, although there internists.JGenInternMed2000;15:881–4.
wasreasonablepowertodetectadifferencebetweenthe 5. BarryHC,EbellMH,ShaughnessyAF,etal.Familyphysicians’
useofmedicalabstractstoguidedecisionmaking:styleor
interventionsoverall, therewaslowpowertodetectdif-
substance?JAmBoardFamPract2001;14:437–42.
ferences within a department. It is possible that there
6. ThePLoSMedicineEditors.Theimpactofopenaccessupon
were differences between the interventions for each
publichealth.PLoSMed2006;3:e252.
department but our study did not have enough sample
7. Editorial.ReadMEDLINEabstractswithapinchofsalt.Lancet
size to investigate the effect of the intervention at the 2006;368:1394.
departmentlevel. 8. BerwangerO,RibeiroRA,FinkelsztejnA,etal.Thequalityof
reportingoftrialabstractsissuboptimal:surveyofmajor
Conclusion generalmedicaljournals.JClinEpidemiol2009;62:387–92.
In this study, we demonstrated that clinical decisions 9. HaynesRB,RamsdenMF,McKibbonKA,etal.Onlineaccessto
made by residents improved when evidence, either MEDLINEinclinicalsettings:impactofuserfees.BullMedLibr
Assoc1991;79:377–81.
abstracts or full-text articles were provided. However,
10. PitkinRM,BranaganMA.Cantheaccuracyofabstractsbe
this study also indicates that some clinical questions
improvedbyprovidingspecificinstructions?Arandomized
maybesimpleenough;answeredquicklyusingaccumu- controlledtrial.JAMA1998;280:267–9.
lated knowledge, but accumulated knowledge was
11. PitkinRM,BranaganMA,BurmeisterLF.Accuracyofdatain
enhancedbytheuseofappropriatemedicalinformation. abstractsofpublishedresearcharticles.JAMA
The residents, in spite of initially stating that 1999;281:1110–11.
52
Evidence-BasedMedicineApril2013|volume18|number2|
Original EBM research
12. PeacockPJ,PetersTJ,PeacockJL.Howwelldostructured 25. TheEditors.Addressingthelimitationsofstructuredabstracts.
abstractsreflectthearticlestheysummarize?EurSciEditing AnnInternMed2004;140:480–1.
2009;35:3–5. 26. WinkerMA.Theneedforconcreteimprovementinabstract
13. Bernal-DelgadoE,FisherES.Abstractsinhighprofilejournals quality.JAMA1999;281:1129–30.
oftenfailtoreportharm.BMCMedResMethodol2008;8:14. 27. HopewellS,ClarkeM,MoherD,etal.CONSORTforreporting
14. TaddioA,PainT,FassosFF,etal.Qualityofnonstructuredand randomizedcontrolledtrialsinjournalandconferenceabstracts:
structuredabstractsoforiginalresearcharticlesintheBritish explanationandelaboration.PLoSMed2008;5:e20.
MedicalJournal,theCanadianMedicalAssociationJournaland 28. ChoudhryNK,FletcherRH,SoumeraiSB.Systematicreview:
theJournaloftheAmericanMedicalAssociation.CMAJ therelationshipbetweenclinicalexperienceandqualityof
1994;150:1611–15. healthcare.AnnInternMed2005;142:260–73.
15. PitkinRM.Theimportanceoftheabstract.ObstetGynecol 29. RosnerB.FundamentalsofBiostatistics.5thedn.PacificGrove,
1987;70:267. CA:Duxbury,2000.
16. AdHocWorkingGroupforCriticalAppraisaloftheMedical 30. RDevelopmentCoreTeam.R:alanguageandenvironmentfor
Literature.Aproposalformoreinformativeabstractsofclinical statisticalcomputing.RFoundationforStatisticalComputing,2011.
articles.AnnInternMed1987;106:598–604. Vienna,Austria.http://www.r-project.org/(accessed7May2012).
17. HaynesRB,MulrowCD,HuthEJ,etal.Moreinformative 31. DiggleP,HeargertyP,LiangK,etal.Analysisoflongitudinal
abstractsrevisited.AnnInternMed1990;113:69–76. data.2ndedn.OxfordUniversityPress,Oxford,UK:2002.
18. HuthEJ.Structuredabstractsforpapersreportingclinicaltrials. 32. SackettDL,RosenbergWM,GrayJA,etal.Evidencebased
AnnInternMed1987;106:626–7. medicine:whatitisandwhatitisn’t.BMJ1996;312:71–2.
19. MacAuleyD.Criticalappraisalofmedicalliterature:anaidto 33. McGowanJ,HoggW,CampbellC,etal.Just-in-time
rationaldecisionmaking.FamPract1995;12:98–103. informationimproveddecision-makinginprimarycare:a
20. NakayamaT,HiraiN,YamazakiS,etal.Adoptionofstructured randomizedcontrolledtrial.PLoSONE2008;3:e3785.
abstractsbygeneralmedicaljournalsandformatfora 34. CrowleySD,OwensTA,SchardtCM,etal.AWeb-based
structuredabstract.JMedLibrAssoc2005;93:237–42. compendiumofclinicalquestionsandmedicalevidenceto
21. GrovesT,GodleeF.InnovationsinpublishingBMJresearch. educateinternalmedicineresidents.AcadMed2003;78:
BMJ2008;337:a3123. 270–4.
22. NarineL,YeeDS,EinarsonTR,etal.Qualityofabstractsof 35. WestbrookJI,CoieraEW,GoslingAS.Doonlineinformation
originalresearcharticlesinCMAJin1989.CMAJ retrievalsystemshelpexperiencedcliniciansanswerclinical
1991;144:449–53. questions?JAmMedInformAssoc2005;12:315–21.
23. DupuyA,KhosrotehraniK,LebbeC,etal.Qualityofabstractsin3 36. LeonSA,FonteloP.MedlinePlusenEspañoland
clinicaldermatologyjournals.ArchDermatol2003;139:589–93. Spanish-speakers.AMIAAnnuSympProc2007;1028.
24. HartleyJ.Currentfindingsfromresearchonstructured 37. SackettDL,StrausSE.Findingandapplyingevidenceduring
abstracts.JMedLibrAssoc2004;92:368–71. clinicalrounds:the‘evidencecart’.JAMA1998;280:1336–8.
53
Evidence-BasedMedicineApril2013|volume18|number2|