Bias in research - Evidence-Based Nursing

1 bias in researchJoanna Smith,1 Helen Noble2 The aim of this article is to outline types of bias acrossresearch designs, and consider strategies to minimisebias. Evidence-Based Nursing , defined as the process bywhich evidence, Nursing theory, and clinical expertiseare critically evaluated and considered, in conjunctionwith patient involvement, to provide the delivery ofoptimum Nursing care, 1is central to the continueddevelopment of the Nursing professional. Implementingevidence into practice requires nurses to critically evalu-ate research , in particular assessing the rigour in whichmethods were undertaken and factors that may is bias in relation to research and whyis understanding bias important?

2 bias is defined by the Oxford Dictionary as: an inclin-ation or prejudice for or against one person or group,especially in a way considered to be unfair ; a concen-tration on an interest in one particular area or subject ; a systematic distortion of statistical results due to afactor not allowed for in their derivation ( ). Understanding research bias isimportant for several reasons:first, bias exists in allresearch, across research designs and is difficult to elim-inate; second, bias can occur at each stage of theresearch process; third, bias impacts on the validity andreliability of studyfindings and misinterpretation ofdata can have important consequences for practice.

3 Thecontroversial study that suggested a link between themeasles-mumps-rubella vaccine and autism in children2resulted in a rare retraction of the published studybecause of media reports that highlighted significantbias in the research occurred on severallevels: the process of selecting participants was misre-presented; the sample size was too small to infer anyfirm conclusion from the data analysis and the resultswere overstated which suggested caution against wide-spread vaccination and an urgent need for furtherresearch. However, in the time between the original pub-lication, and later research refuting the originalfindings,the uptake of measles-mumps-rubella vaccine in Britaindeclined, resulting in a 25-fold increases in measles inthe 10-year period following the original different study designs have specific meth-odological challenges and constraints, bias can occur ateach stage of the research process (table 1).

4 In quantita-tive research , the validity and reliability are assessedusing statistical tests that estimate the size of error insamples and calculating the significance offindings(typically p values or CIs). The tests and measures usedto establish the validity and reliability of quantitativeresearch cannot be applied to qualitative , in the broadest context, these terms are applic-able, with validity referring to the integrity and applica-tion of the methods and the precision in which thefindings accurately reflect the data, and reliability refer-ring to the consistency within the analytical is bias minimised when undertakenresearch?

5 bias exists in all study designs, and although researchersshould attempt to minimise bias , outlining potentialsources of bias enables greater critical evaluation of theresearchfindings and conclusions. Researchers bring toeach study their experiences, ideas, prejudices and per-sonal philosophies, which if accounted for in advance ofthe study, enhance the transparency of possible researchbias. Clearly articulating the rationale for and choosingan appropriate research design to meet the study aimscan reduce common pitfalls in relation to bias . Ethicscommittees have an important role in consideringwhether the research design and methodologicalapproaches are biased, and suitable to address theproblem being explored.

6 Feedback from peers, fundingbodies and ethics committees is an essential part ofdesigning research studies, and often provides valuablepractical guidance in developing robust quantitative studies, selection bias is oftenreduced by the random selection of participants, and inthe case of clinical trials randomisation of participantsinto comparison groups. However, not accounting forparticipants who withdraw from the study or are lost tofollow-up can result in sample bias or change thecharacteristics of participants in comparison research , purposeful sampling has advan-tages when compared with convenience sampling inthat bias is reduced because the sample is constantlyrefined to meet the study aims.

7 Premature closure of theselection of participants before analysis is complete canthreaten the validity of a qualitative study. This can beovercome by continuing to recruit new participants intothe study during data analysis until no new informationemerges, known as data quantitative studies having a well-designedresearch protocol explicitly outlining data collection andanalysis can assist in reducing bias . Feasibility studiesare often undertaken to refine protocols and can be reduced by maximising follow-up andwhere appropriate in randomised control trials analysisshould be based on the intention-to-treat principle, astrategy that assesses clinical effectiveness because noteveryone complies with treatment and the treatmentpeople receive may be changed according to how theyrespond.

8 Qualitative research has been criticised forlacking transparency in relation to the analytical pro-cesses researchers must demon-strate rigour, associated with openness, relevance topractice and congruence of the methodologicalapproach. Although other researchers may interpret thedata differently, appreciating and understanding howthe themes were developed is an essential part of dem-onstrating the robustness of thefindings. Reducing biascan include respondent validation, constant compari-sons across participant accounts, representing deviantcases and outliers, prolonged involvement or persistentEditor s choiceScan to access morefree of Human and HealthSciences, University ofHuddersfield, Huddersfield, UK2 School of Nursing andMidwifery, Queens s UniversityBelfast, Belfast, UKCorrespondence to:Dr Joanna Smith,School of Human and HealthSciences, University ofHuddersfield, HuddersfieldHD1 3DH, Based NursOctober 2014| volume 17| number 4| research made simple on June 29, 2022 by guest.

9 Protected by Based Nurs: first published as on 5 August 2014. Downloaded from observation of participants, independent analysis of thedata by other researchers and summary, minimising bias is a key considerationwhen designing and undertaking research . Researchershave an ethical duty to outline the limitations of studiesand account for potential sources of bias . This willenable health professionals and policymakers to evalu-ate and scrutinise studyfindings, and consider thesewhen applyingfindings to practice or K,McSherry R. Evidence-Based Nursing : clarifying theconcepts for nurses in Crit Care2009;3:67 71, AJ,Murch SH, Anthony A,et al.

10 Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasivedevelopmental disorder in ;351:637 ileal-lymphoid-nodular hyperplasia,non-specific colitis, and pervasive developmental disorder ;375 JM,et al. Verification strategies for establishingreliability validity in qualitative J Qual Res2002;1:1 PJ,Berlin JA, Gopalan R,et al. Publication bias ;337:867 M,Egan M, Thomson H,et al. Publication bias inqualitative research : what becomes of qualitative researchpresented at conferences?J Epidemiol Community Health2008;62:552 in research ;3:780 J,Johnston M, Robertson C,et al.