SPEECH & LANGUAGE THERAPY IN PRACTICE
JOURNAL CLUB 5
Journal club 5:
READ THIS SERIESIF YOU WANT TO
BE MOREEVIDENCE-BASEDIN YOURPRACTICE
FEELMOTIVATED TO READJOURNALARTICLES
Jennier Reid’sseries aims to help you access the speech and languagetherapy literature, assess its credibility and decide how to act on yourndings. Each instalment takes the mystery out o critically appraising adiferent type o journal article. Here, she looks at observational studies.
esearch is undamentally a quest orexplanations. Explanations go beyondsimple description in order to provide anaccount o causal relationships, or example,between events, human belies, behaviour,experiences and ill-health. It is really importantto keep the notion o causalityin mind as wetry to get our heads round the observationaldesigns used in health-related research. Causalreasoning, in a nutshell, requires us to:• know what conditions preceded thephenomenon o interest,• assess which o these antecedentconditions are candidates as causalagents, and then• organise this knowledge into a plausible,causal chain o events. There are a number o observational researchdesigns, and not all provide robust evidence o causality. It is not enough to demonstrate thatthere is anassociationbetween two actors.I you nd that children living in high atshave poorer health, this is not good evidencethat living in a at causes ill-health. The twoactors may be related, with economic orsocial circumstances perhaps much bettercandidates or an underlying cause.
Experimental or observational?
When an article talks about an intervention,how do I know i this is an experimental studyor an observational one?Group intervention studies, especially thoseusing randomisation to groups, are consideredsuperior to observational designs oranswering causal questions about healthcareinterventions, since there is better control o the efects o any unoreseen (conounding)actors. However, such anexperimentaldesign is not always practical or possible,especially where the participants’ commoncharacteristic cannot be manipulated (such ashaving a genetic condition) or would not beethical because o likely negative efects (orexample, not talking to your baby).Observationalstudies examine aspectso people’s past and/or present lie in orderto identiy relevant inormation throughobservation rather than experimental man-ipulation. They do not ofer a particularintervention and measure
its efect.However, inormation may be collected aboutintervention(s) the participants have received. The goal o observational studies is usuallyto identiy actors which
be causallyrelated, and thus may be incorporated intointerventions which then produce betteroutcomes in the uture.Sometimes people who have received aparticular intervention are ollowed up andtheir outcomes compared to the outcomes o others who did not receive the intervention.An observational study appraisal tool is likelyto be the one to choose or such a study. Theexception is i the intervention was oferedas a core part o the research design andparticipants were allocated or selected toreceive one intervention or another accordingto some preset criteria.
There are a number o observational designsand some jargon to deal with.1. Cohort studyA cohort study ollows over time o a groupo people who have something in common.(The term cohort was also used to reer to agroup o Roman soldiers, which could be auseul aide-mémoire.) The group may havea common characteristic, such as wherethe participants were born, how they wereeducated or an aspect o their health or well-being. Alternatively they may have all beenexposed to a risk or challenging circumstanceo some kind, or have received a particularhealth intervention. The comparison group or cohort studiesmay be the general population rom which thecohort is drawn, or another cohort o personsthought to be similar except or the commoncharacteristic under investigation. Alternatively,subgroups within the cohort may be comparedwith each other. This is commonly the case inbirth cohortstudies, where all children bornin particular years in one geographical areaare studied. Results are analysed to detect acohort efect. This means nding out whethermembership o the cohort, and thereorehaving the common characteristic, appears tomake a diference to the outcome.In research designed to investigate riskso adverse health outcomes, a cohort isidentied
the appearance o thecondition(s) under investigation. For example,Conti-Ramsden & Botting (2007), in a studyo emotional health in adolescents, describetheir young people with specic languageimpairment as, “originally recruited at 7 yearso age as part o a wider study … The original
NHS Fie’s Dunermline cluster Journal Club. Author Jen Reid has her back to the camera.