Upload
jin-booker
View
22
Download
1
Tags:
Embed Size (px)
DESCRIPTION
RESEARCH METHODS Quantitative Research by Dr Viv Rolfe. Evidence Based Practice The University of Nottingham School of Nursing. Dr Vivien Rolfe BSc PhD. 20 Year research career Degree - Physiology PhD - Diarrhoeal diseases Research - Inflammatory bowel disease (Great Ormond Street) - PowerPoint PPT Presentation
Citation preview
RESEARCH METHODS Quantitative Research
byDr Viv Rolfe
Evidence Based PracticeThe University of Nottingham
School of Nursing
Dr Vivien Rolfe BSc PhD
20 Year research career•Degree - Physiology
•PhD - Diarrhoeal diseases•Research - Inflammatory bowel disease (Great
Ormond Street)•Research - dog and cat nutrition (Mars UK)
•Market research – qual and quant (Mars UK)•Lecturer - University of Nottingham
Ice Breaker + Needs from today (5mins)
• Do they know eachother?• Everyone say your name, where
from and something about yourself.
• What do they want out of today?• Write on board.
Plan of Today
• Introduction to research• Qual versus quant• Types of quantitative research
Research is…..
• inquiring into, finding out• a systematic investigation to establish
facts• a search for knowledge
• “original investigation undertaken in order to gain knowledge and understanding.”Dept. For Employment and Learning & HEFCE (Higher Education
Funding Council for England)
Health and Medical Research
• Research is fundamental to advancing healthcare and medicine.
– Produces new treatments and drugs.– Advances understanding of new diseases and
problems (obesity, aging, CJD, MRSA…all topical).
– Can help determine best type of care.– Can help evolve patient services.– Development of diagnostic tests.– Development of prognostic markers.
UK Medical Research Priorities
• DOH - £540m in 2002-2003
• cancer, mental health, coronary heart disease (CHD)
• ageing and older people, public health, genetics , diabetes.
Think about research? (5 minutes in pairs)
• List some different types of research you are aware of.
• Who does it?• What is it?• Write ideas on post it notes and
cluster (hopefully into QUAL and QUANT).
Research – from dream to reality
IDEAHypothesis
Literature review
RESEARCH PLANWhat? Which? Who?
What am I looking at?Which methods to use?Who am I investigating?
FundingEthicsImplementation
Conduct ResearchData analysisConclusions
Dissemination
Qualitative versus Quantitative
Quantify - measure
QUAL
Subjective (opento personal interpretation)
Words “verbatims”
Generates ideas and theories
Opinions, feelings
Big picture
QUANT
Numbers and statistics
Objective
Focus
Tests ideas and theories
Quantitative Research
“a formal, objective, systematic process in which numerical data are utilised to obtain information about the world“
Non-experimental
Case studyCross sectionalCohort/longitudinalCase controlCorrelationalDescriptiveSurveyEpidemiologyNeeds assessment
Information gatheringabout populations.
Types of QUANT
Experimental
Human clinical trialVeterinary clinical trialAnimal experimentationIn vitro laboratory
All test a hypothesisabout an intervention,technique, practice,service.
Types of QUANT Study Design
4 main types:
• Clinical trials (from crap to RCT)• Cohort studies• Case-control studies• Cross-sectional studies
Clinical Trials
Clinical Trial
• Experimental research testing the effectiveness of an intervention on a sample of human subjects.
• Called intervention trials.• Clinical research usually stems from
laboratory-derived ideas and hypotheses:
– In vivo – whole body– In vitro – test tube
What happens in a Clinical Trial?
• People usually do have a disease or illness at the study beginning, they are observed to see the treatment/intervention is having an EFFECT.
No Treatment
Measure and compare: e.g.number of fillings over a year
Fred, Harry, Mavis, Bert, Daisy Tom, Betty, David, Edna, Bill
Treatmente.g. brushing teeth using fluoride toothpaste
“I’m designing a clinical trial and I’ve picked….”
•A and B subjects are hand-picked•small numbers (10’s)•the groups are different•the study is not blind•there is no placebo•it is one phase (single leg)
What do you notice about this study?
Pitfalls to Overcome
THE PEOPLE• A and B subjects are hand-picked - researcher could manipulate
the outcome of the research so introduce BIAS (unfairness)• The group numbers are small – you must complete a POWER
CALCULATION to see how many subjects to use, otherwise statistical analysis will be problematic.
• The groups are different - one contains children and the other doesn’t, so we are not comparing like with like. Groups must be MATCHED.
THE DESIGN• The study is not blind - means the clinical staff and patients
know who is in what group, which causes BIAS.• There is no placebo - or “dummy”, so no adequate control.• The design has only one leg (ONE PHASE) so may be
susceptible to differences due to TIME.
Creating a Robust Trial
A and B subjects are hand-pickedsmall numbers; not matched, not blind, no placebo; one phase (single leg)
A and B are randomly chosenGroups are matched (age, sex, other)
Single blind (patient doesn’t know)
Placebo controlled
Double blind (patient and clinician don’t know)
Cross-over
Multi-centre (UK, International)
•Pilot study•Not robust evidence•Used to get funding/interest for a largerstudy
•Large-scale study•Evidence to informclinical/healthcare decisions
How could you improve the study?
• 10 minutes in groups• What do you want to look at?• Who would you use?• What would you measure?
Ideal Study
Randomly chosenLarger numbers
MatchedDouble blind
Toothpaste + placebo Toothpaste + fluoride
measurement measurement
Toothpaste + placeboToothpaste + fluoride Cross-over
measurement measurement
e.g. if studydone at Easteror Xmas, timewill be aninfluence!.
Multicentre!
Conclusions
• Randomisation - the random assignment of subjects to treatment groups
• Double-blind - neither the researcher or the subjects know which are the experimental group or control group
• Placebo-controlled• Cross-over (may be constrained by ethics,
but better if each subject acts as their own control which can reduce sample size – butneed a “washout phase” and not appropriate if treatment gives permanent effects)
• Multi-centre (may be costly and timely)
What to Look for in a Good Trial…..
• Look for these words in the methods section…
• Randomised-controlled trial (RCT)• Double-blind randomised-placebo-
controlled trial – THE BEST.• RCT’s follow standardised
procedures to ensure good scientific standard.
Sources of RCT evidence
• www.nelh.co.uk• Cochrane Database of Systematic
Reviews
BREAK?10.30 – 10.45
The 3 C’s
Cross-sectionCohortCase-control
Effectiveness of diagnosis or screening
Aetiology or prognosis
Aetiology or prognosis for a rare condition
Non-experimental QUANT
Cross-Sectional Study
Cross-sectional Studies
• Look at a relationship in a defined group at ONE POINT IN TIME (not over a period of months/years).
• Used for testing the effectiveness of diagnostics or screening, or tests for the prevalence of a disease or parameter.
• May measure a physiological parameter/do a scientific test.
• Give valuable insight into whether a practice/test needs to be modified to improve health care.
Example 1 Tanning is associated with optimal vitamin D status and higher bone
mineral density.
The Methods
• Why – Vitamin D is made in the skin in sunlight, so do people who use sunbeds have higher concentrations of vitamin D?
• What – a cross-sectional analysis of sunbed and non-sunbed users.
• Who – 50 people who use sunbeds weekly and 106 control non-users.
• How – compare blood Vitamin D levels, and bone mineral density testing.
The Results
• Tanning bed users had 90% higher blood vitamin D levels compared to controls.
• Tanners had significantly higher bone mineral density scores.
The Conclusions
• Regular use of a tanning bed can benefit the skeleton.
• OK, there are obvious known additional risks of skin cancer, but this study indicates that sun beds may be of therapeutic use for some conditions.
Strengths and weaknesses
•Cheap and simple •Ethically safe
BUT •Establishes an association/relationship between things (UV rays and Vitamin D), but but not causes
•If it relies on questionnaires, patient recall of events may introduce bias susceptibility
Cohort Study
Cohort Studies
•“Cohort” means group of individuals with similar traits…
– e.g. same age– support same football team (West
Ham of course)– who all have eczema
Why do a cohort study?
• Often carried out to study harmful interventions (like smoking) which would not be ethically permitted in a clinical study.
• To see what risk factors contribute to a disease, or see what their outcome or PROGNOSIS is in the future.
• These are longitudinal studies – over a period of time.
Design
• Can be prospective - look forward, to look at if a disease develops, or the outlook/prognosis if the disease is already there.
• Can be retrospective - look backward, to examine the history of a disease, and find risk factors associated with it.
Retrospective cohort study
JANUARY 2005
Clinical RecordsOr
Questionnaires1995 - 2005
Disease
What were they exposed to?
Risk factorsPossible causes
Example 2 A high morbidity outbreak of MRSA among players on a college football
team.
The Methods
• Why - MRSA infections in football are an increasing problem.
• What - A retrospective cohort study was carried out to see what factors were contributing to the infections.
• Who – cohort of 200 football players with skin abscesses or infection were studied.
• How – their habits/ behaviours/ recorded.
The Results – the risk factors
• The incidence of infection differed depending on the players position (and whether he had more knocks and bumps).
• Turf burns enhanced the risk.• Body shaving enhanced the risk.• Sharing the Jacuzzi at the end of the
match showed a small increase in risk of developing infection.
The Conclusions
• Educate players to reduce body shaving before a game.
• Improve padding to prevent turf burns.• Introduce means of sterilising Jacuzzi
water, just as an additional precaution.• Probably can’t do much about playing
positions.
Prospective cohort study
Assessment of risk factors (environment, lifestyle)Can be clinical measure or a questionnaire
No Disease DiseasePrognosis/outcome
JAN 2005
JAN 2005 – JAN 2006
Example 3 Female-male infectivity of HIV among circumcised and uncircumcised
Kenyan men.
The Methods
• Why - HIV infection leading to AIDS is a major cause of mortality in Africa (and other countries). Observations suggest that circumcision reduces HIV infectivity.
• What – a prospective cohort study was carried out.
• Who - 745 Kenyan truck drivers.• How – sexual habits were recorded at
interview and follow up interviews.
The Results
• Prognosis - after intercourse the probability of contorting aids was high.
• Infectivity was higher for uncircumcised men compared to circumcised men.
The Conclusions
• Lack of circumcision does appear to be a risk factor in AIDS transmission.
• Encourage circumcision if culturally possible, or promote awareness of this fact to modify sexual behaviour.
Strengths of Cohort Studies
•Looks at more than one influence or factor compared to clinical trial that just studies one (e.g., using the Jacuzzi, turf burns, player’s game position).•Provides rich information about lifestyle and health associated RISK FACTORS.•Ethically safe (i.e. not inflicting treatments on people).•Easier and cheaper than an RCT.
Weaknesses of Cohort Studies
• In order to observe any statistical and biological effects it is often necessary to study large groups.
• Large studies are costly and timely, particularly the prospective ones.
• Blinding is difficult – i.e. the subjects know what group they belong to, so is often best not to tell them the nature of the study.
• No good for rare diseases where large groups don’t exist.
• Often relies on people’s recollections.
Case Control Study
Case-control studies
• Similar to cohort in being able to identify risks.
• Rather than looking at random groups, smaller numbers of patients (cases) are chosen along with a CONTROL.
• Because patients are chosen, this can introduce bias either from who is selected, and how the measurements are taken.
• However, you can focus in on rare cases or situations.
Case control studies
• Answers the question “what makes a group of individuals different in terms of the causes of a disease”?
• Design: Can be prospective or retrospective looking back at medical history.
• Needs 2 groups - Compares those with the disease with a control group of people with similar characteristics (but no disease).
Example 4 Diet and stomach cancer: a case-control study in South India.
The Methods
• Why - the prevalence (occurrence) of stomach cancers in some parts of India is increasing. It is not known what factors are associated with this.
• What – a prospective case-control study monitored the risks of developing stomach cancer.
• Who - 194 patients and a matched (age, sex, religion) healthy control group for 3 years.
• How – interviews using questionnaire.
The Results
• A questionnaire collected data on food intake, socio-demographic factors and other behaviours such as smoking.
• Risk of cancer was high for those eating spicy food, consuming chilli, and consuming high temperature food.
The Conclusions
• Some interesting associations have been found between diet and cancer.
• Further studies are needed to see whether there is a definite link between spicy food and stomach cancer (ie now go back to the lab and test this hypothesis).
Strengths and weaknesses
•Quick and cheap
•The only feasible method for very rare disorders fewer subjects needed than cross-sectional studies
BUT
•Often relies on recall to determine result
•Difficult to select a control group which may introduce bias
Types of Quantitative Research
4 main types:
• Clinical trials (from crap to RCT)• Cohort studies• Case Control study• Cross-sectional studies
Activity – in groups
• From the handout distinguish which of the following are:– 1. Clinical trial (is it a “good” or “bad”
one?– 2. Cohort studies– 3. Case control studies– 4. Cross-sectional studies
Additional Resources
Studies Cited in this Course
Tangpricha V, Turner A, Spina C et al. Tanning is associated with optimal vitamin D status (serum 25-hydroxyvitamin D concentration) and higher bone mineral density. Am J Clin Nutr, 2004, 80(6), 1645.
Begier EM, Frenette K, Barrett NL et al. A high-morbidity outbreak of MRSA among players on a college football team, facilitated by cosmetic body shaving and turf burns. Clin Infect Dis, 2004, 15;39(10), 1446.
Mathew A, Gandadharan P, Varghese C et al. Diet and stomach cancer: a case-control study in South India. Eur J Cancer Prev, 2000, 9(2), 89.
Baeten JM, Richardson BA, Lavreys L et al. Female-to-male infectivity of HIV-1 among circumcised and uncircumcised Kenyan men. J Infect Dis, 2005, 15;191(4), 546.
Medline – the one-stop resource for all global medical and health research.For example to find a type of study type in “case control” “cohort”, etc. You can alsouse the “limits” button to search for an RCT or a particular population group.
http://www.cebm.net/study_designs.asp
http://symptomresearch.nih.gov/tablecontents.htm
http://www-phm.umds.ac.uk/teaching/ClinEpid/ObservS.htm#case-control
http://www.mja.com.au/public/issues/174_05_050301/craig/craig.html
http://www.clinicalevidenceonline.org/
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi
http://medicine.ucsf.edu/resources/guidelines/users.html
http://www.shef.ac.uk/scharr/ir/netting/
Useful Weblinks
For quick reference to research terms such as prognosis, prevalence, subjective, cohort, cross-section, statistics, then go to:
http://www.mori.com/rmu/glossary.shtml
http://www.fortunecity.com/greenfield/grizzly/432/rra2.htm#experiment
http://www.cebm.net/glossary.asp
Glossaries of Research Terms
Prevalence
• Measure of a condition in a population– At a point in time– Over time (e.g. 1 year)
• Gives an indication of the extent of a problem and health care requirements.
Prevalence = No. of cases
Population
Incidence
• Number of new occurances of a condition• over time (e.g. 1 year, 10 years.)• Gives an indication of the extent of a problem
and health care requirements.
Incidence = No. of NEW cases
Population or subpopulation
1995 - 2005
Prevalence 4 out of 20 = 20% in the last 10 years.
Incidence = 2 out of 22 = 9%
Experimental Non-experimental
Intervention of drugor practice
Diagnosis orScreening test
Aetiology(causes)
Prognosis(outlook)
Clinical trialRCT
Cross-sectional
study
Cohort(case control
for rare groups)
Cohort(case control
for rare groups)
Summarise your own learning here…..
Session Feedback for Viv Rolfe, EBL Course, Monday 24th January
Content
a) The course content reflected the aims andthe expected learning outcomes.
c) The strong points of the course were (please indicate)
d) The weak points of the course were (please indicate)
Lecturer
a) She was knowledgeable
b) She could answer questions satisfactorily
c) She/the course kept my attention.
b) The pace was about right.
1 2 3 4 5 Stronglyagree
Stronglydisagree
1 2 3 4 5 Stronglyagree
Stronglydisagree
d) I felt able to participate.
Please tick which box you think (1 = strongly agree to 5 = strongly disagree)
Session Feedback for Viv Rolfe, EBL Course, Monday 24th January
Event Materials
a) The handouts complemented the course.
c) I shall use the website.
Overall
a) The course was useful.
b) Please rate you understanding/level of knowledge on the subject of research methods:
b) The handouts aided understanding.
1 2 3 4 5 Stronglyagree
Stronglydisagree
1 2 3 4 5 Stronglyagree
Stronglydisagree
c) Any other comments
Please tick which box you think (1 = strongly agree to 5 = strongly disagree)
Before today
Low 1 2 3 4 5 High
Low 1 2 3 4 5 HighAfter today
Website and contact details
Additional information, links to the glossariesand other websites mentioned can be found on Viv’s website
www.nottingham.ac.uk/~nqzvr/teaching.htm
Password is einstein
You can also contact me by email if you have any further questions orcomments