Metodologia cercetării ştiinţifice
Ipoteza de lucru
Organizarea studiului
Tipuri de studii
Corin Badiu, 2007
Obiective
• Dobandirea abilitatilor de cercetare
• Redactarea unui protocol de 10 pagini
pentru un studiu real
• Analiza altor studii (peer review)
Tipuri de Studii
• Nu sunt cea mai buna alegere pentru acest curs
 Studii fundamentale: molecule, celule, soareci
 Cost-eficienta, meta-analize
 Cercetare calitativa
• Ideal

Un studiu observational (sau experiment) ce
implica pacienti
Sectiuni ale protocolului de studiu
•
•
•
•
Ipoteza de lucru
Semnificatia (fondul)
Designul studiului
Populatia de studiu si recoltarea datelor;
controlul de calitate al datelor
• Variabile si masuratori
• Analiza statistica a datelor
• Etica cercetarii medicale
Ipoteza de lucru
• Toate studiile incep cu o problema, intrebare
fara raspuns
• Scopul este alegerea unei intrebari de
cercetare relevante, ce poate fi transformata in
proiect
Problema studiului
• Format pentru un studiu descriptiv

“Intr-o populatie de [populatie de studiu ],
care este prevalenta (sau media, mediana,
etc.) [variabila dependenta]?”
• Majoritatea studiilor au mai mult decat o
problema de evaluat si rezolvat
• Protocolul de cercetare deriva din
problema studiului
Problema studiului
• Format uzual (studii analitice):
“Intr-o populatie de [populatie de studiu],
exista o asociere intre [variabila
dependenta] cu [variabila independenta]?”
Identificati problema studiului
enuntata in rezumat
Acute respiratory tract infections caused by
Streptococcus pneumoniae are a leading cause of
morbidity and mortality in young children. We
evaluated the efficacy of a 9-valent pneumococcal
conjugate vaccine in a randomized, double-blind
study in Soweto, South Africa. At 6, 10 and 14 weeks
of age, 19,922 children received the 9-valent
pneumococcal polysaccharide vaccine conjugated to
a noncatalytic cross-reacting mutant of diphtheria
toxin (CRM197), and 19,914 received placebo…
Klugman KP, Madhi SA, Huebner RE, et al. A trial of 9-valent pneumococcal
conjugate vaccine in children with and without HIV infection.
N Engl J Med 2003; 349:1341-8.
Intr-o populatie de [populatie de studiu], exista o asociere
intre [variabila dependenta] si [variabila independenta]?”
In a population of South African infants is
immunization with a 9-valent pneumococcal
conjugate vaccine associated with invasive
pneumococcal disease?
Originile unei probleme de cercetare
• Cercetarea literaturii

Discutia problemelor din literatura
• A fi deschis la idei si tehnici noi

Importanta invatarii
• Imaginatia
Caracteristicile unui studiu
FINER
•
•
•
•
•
Fezabil
Interesant pentru investigator
Nou
Etic
Relevant
Caracteristicile FINER
Criterii de Fezabilitate
•
•
•
•
Numar adecvat de subiecti
Expertiza technica adecvata
Realizabil ca resurse de timp si bani
Resurse umane calificate adecvat
Caracteristicile FINER
Criterii de Interes
• Studii cu impact populational
• Detalierea a noi mecanisme de boala
• Studii de eficienta terapeutica pentru
metode noi
Caracteristicile FINER
Criterii de Noutate
• Confirma sau anuleaza cercetari anterioare
• Extinde cercetari anterioare
• Furnizeaza noi date
Caracteristicile FINER
Criterii de Etica
• Consimtamant informat !!!
• Protejeaza pacientii sau NU ii expune unui risc
suplimentar
• Excluderea imediata / oprirea studiului in cazul
unor efecte adverse severe
Caracteristicile FINER
Criterii de Relevanta
• Pentru cunoasterea stiintifica
• Pentru clinicieni, sanatatea publica sau
politicile de sanatate
• Pentru directii de cercetare viitoare
Probleme si solutii:
Planul nu este FINER
• Nefezabil




Prea larg
Prea putini pacienti disponibili
Metode sofisticate, peste abilitatile
dovedite de investigator
Prea scump
• Fara interes, noutate sau relevanta
• Evaluare etica inadecvata
Probleme si solutii:
Planul nu este Fezabil
• Prea larg


Un set mai mic de variabile
Se delimiteaza mai mult domeniul
• Insuficienti subiecti disponibili





Extinde criteriile de includere
Schimba criteriile de excludere
Adauga alte surse de subiecti
Extinde durata de inrolare
Foloseste strategii de scadere a lotului
Probleme si solutii:
Planul nu este fezabil
• Metode in afara abilitatilor investigatorului



Colaborare cu colegi cu aptitudini
Consultarea de experti si reluarea literaturii pentru
metode alternative
Invata si practica aptutidinile cerute de studiu
• Prea scump

Scade costurile designului de studiu
• Mai putini subiecti si masuratori
• Detaliu mai mic al masuratorii
• Mai putine vizite de urmarire
Probleme si solutii:
Planul nu este FINER
• Neinteresant, vechi sau irelevant


Consulta mentorul
Modifica problema cercetata
• Abordare etica nesigura


Consulta comitetul de etica
Modify the research question
Probleme si solutii
• Planul de studiu este vag

Rescrie planul de cercetare mai amanuntit

Detaliaza din planul de studiu
• Cum vor fi selectati subiectii si lotul martor
• Cum vor fi masurate variabilele
Probleme si solutii: exemple
• Ce relatie exista intre depresie si
sanatate?
• Consumul de carne rosie determina
cancer?
• Scaderea colesterolului seric previne
boala cadiaca ischemica?
Anatomia Cercetarii Clinice
Fiziologia cercetarii clinice:
Cum are loc?
Se folosesc masuratorile intr-o proba /
esantion /lot de studiu
pentru a extrage inferente despre
fenomene (variabile) intr-o populatie
Tipuri de variabile
Variabile de
confuzie*
Predictor*
Rezultat
Modificatori ai efectului*
*Considerate general ca expunere la factori de risc
Problema de studiu
• In studii descriptive:
“Intr-o populatie de [study population], care
este prevalenta [outcome variable]?”
• In studii analitice :
“Intr-o populatie de [study population], este
[predictor variable] asociata cu [outcome
variable]?”
Tipuri de studii clinice
• Studii fara variabile

Studii de caz, serii de cazuri, editoriale,
opinii / comentarii, rapoarte review
• Studii cu o singura variabila

Studii descriptive
• Studii cu ≥2 variabile



Experimente
Studii observationale
Meta-analize si review-uri sistematice
Ierarhia tipurilor de studii clinice
Descriptive studies
Experimental studies
Cohort
Analytic studies
Observational studies
Case-control
Cross-sectional
Conceptie (study design) legata de obiectivele studiului.
Conceptie inainte de colectarea si analiza datelor
Concepta studiului este importanta intrucat permite o crestere a validitatii si
reducerea erorilor.
Variabile
Variabila predictor
(independenta)
Variabila rezultat
(dependenta)
Factori care influenteaza conceptia studiului
1. Obiectivele studiului – intrebarile cercetarii
2. Epidemiologia bolii / expunerii - rara / comuna
3. Aspecte etice
4. Resurse- bani, oameni, infrastructura
Ce tip de studiu este mai
adecvat?
• Telefonia mobila determina cancer cerebral?
• Dieta afecteaza riscul de cancer mamar?
• Fumatul pasiv determina cancer pulmonar?
Tipuri de study design
1. Experimental/ Quasi-experimental sau
observational
2. Directia studiului in relatie cu expunerea si
aparitia bolii
3. Cantitativ /Calitativ
Women's Health Initiative (1991)
http://www.nhlbi.nih.gov/whi/
Ar trebui sa primeasca substitutie cu estrogeni
femeile postmenopauza?
Women's Health Initiative (1991)
Ar trebui sa primeasca substitutie cu estrogeni
femeile postmenopauza?
Subiecti: femei postmenopauza
Predictor: “estrogeni”
Rezultat: ?
Ameliorarea problemei de cercetare
Tratamentul cu estrogeni previne
accidentul coronarian la femei postmenopauza?
Subiecti:femei postmenopauza
Predictor:
tratament cu estrogeni vs no treat
Rezultat:
accident coronarian
FINER?
Fezabil
Interesant
Nou
Etic
Relevant
7 tipuri de studii
Randomized controlled trial
2. Cohort study
3. Case-control study
4. Sample survey or cross sectional
study
5. Ecologic or trend study
6. Case report or series
7. Qualitative research
1.
Ncesitatea desing-ului
• Trial clinic randomizat


Puncte finale
Interes primar
• Studiu observational



Cross-sectional
Case control
Cohorta
1. Studii Experimental / Interventionale Randomized controlled trial (RCT)
• Evaluate new forms of therapy and prevention
 New drugs and other treatment of disease
 New medical and health care technology
 New methods of primary prevention
 New programs for screening and early detection of
disease
 New ways of organizing and delivering health services
( community trial )
 Impact of new policies in health care and health care
financing ( community trial )
Randomized controlled trial
• Cel mai bun design- randomizare, eroare de selectie minima
(dar nu zero);
• Nu este etic: daca interventie poate dauna, rezultat clinic slab
• Nu este practic – boli rare, efect rar;
• Participarea subiectilor la studiu este cruciala.
Randomized Controlled Trial
Elements
Design
Selection of subjects
-similar features
-inclusion/exclusion criteria
Defined population
Randomization
Allocation of subjects
Intervention
Outcome
New treatment
Improved
Data collection
Masking (blinding)
Not
improved
Current treatment
/no treatment
Improved
Not
improved
RCT
•
Blinding



•
Double- subjects & investigators
Single – subjects (placebo effect)
Triple blind- subjects & investigators & statisticians
Outcome/endpoint

improvement ( desired effect) and side effects

must be explicitly defined

measured comparably in all study groups
•
Multi-centre trial
•
Comply with GCP requirements
•
Clinical trials/ community trials
•
Results of RCT- benchmark- good clinical practice (GCP) ,
clinical governance, treatment protocol.
Types of RCT:
• Superiority or equivalence trials
A
• Parallel-most common
B
• Cross over
-
Planned –washout
period
A
B
randomized
B
A
- Unplanned
randomized
Surgical
Analyze outcome by intention to Rx
Medical
Refuse surgery
Require surgery
Surgery
No surgery
Limite
1. ? effective in uncontrolled community
2. consent – refusal- automatic selection
(people who participate are different from those who do not)
3. Non compliance ( people who are compliant are very different from
those who are not)
- drop-outs: no adherence to experimental regimen, loss to f/u
- drops-in: no adherence to control regimen
4. Compliance -Need monitoring
5. Most costly
Studii observationale
1.
Cohort study
2.
Case-control study
3.
Cross sectional study
4.
Ecologic or trend study
5.
Case report or series
6.
Qualitative research
Direction, timing , +/- control
Studiu de cohorta
Subiecti

5000 femei cu varsta 55+ ani
Factor predictor:

estrogeni post-menopausal?
Outcome:

Incidenta accidentelor coronariene dupa 5 ani
de terapie
Cand este bun un studiu de cohorta?
• When there is good evidence of an association of the
disease with a certain exposure
• When exposure is rare and incidence of disease among
exposed is frequent
• When the time between exposure and disease is short
• When the investigator has a long life expectancy!
Cohorta
• Cohort: “any designated group of individuals who
are followed or traced over a period of time”
• Free of the outcome at the beginning of follow-up
(population at risk)
• Usually defined or divided by exposure status
Populatia la risc
• All subjects within a cohort must be free of the
outcome at the start of the follow-up period
• All subjects must be at risk for developing the
disease (population at risk)
Studii de cohorta Prospective vs.
Retrospective
• Prospective
Cohort Study

Better control of the
quality and quantity of
the data
• Retrosepective Cohort
Study

• greater potential for bias
• less potential for bias


More time consuming
More expensive
Poorer control of the
quality and quantity of
the data


Less time consuming
Less expensive
Pierderea din urmarire
(Loss to Follow-up)
• Problem, especially if lost subjects are at
higher/ lower risk for developing the outcome
compared to other subjects

even worse if this problems is different with
comparison groups
• < 60% follow-up is generally regarded with
skepticism
Analiza datelor – studii de cohorta
Start with exposure, then follow to see whether
Incidence rate
total
Disease + Disease of disease
Exposed +
a
b
Exposed -
c
d
a+b
c+d
a/a+b
c/c+d
Analiza datelor – studii de cohorta
• Absolute risk = incidence of disease cases
in exposed group / total population
incidence of dis among E+
• Relative risk=
incidence of dis among E-
• Taria asociatiei (Odds ratio) = ad/bc
Limite
• Some exposure can change over time e.g. aging, life style
(diet, smoking pattern), exposure to pharmaceutical agents,
air pollution
• Changes on method over time- disease identification
• Prospective - follow-up, thus long period
• Costly
• Potential bias





selection bias ( who has been exposed and who has not )
information bias – historical data
bias in outcome assessment
non-response and loss to f/u
analytical bias
Cross-sectional design
Subiecti

2000 femei varsta 55+ ani
Predictor:

Tratament post-menopausal cu estrogeni?
Rezultat:

Istoric de accident coronarian?
Sample survey or cross sectional study :
• It is a survey of the frequency of disease and
other characteristics in a defined population
at one particular time.
• Focus on describing a state or process at a
fixed time.
• Suitable for describing prevalence and
distribution of heath problem in a community.
(prevalence or health survey).
Sample survey or cross sectional study : - cont
• In clinical research, used in :



Describing disease presentation (spectral description
study)
Diagnostic test accuracy study
Quality of care assessment.
• Generally efficient but large scale community
survey can be expensive.
• Sometimes survey data given causal
interpretation (pseudo-longitudinal) but liable to
error, caution.
Advantages of Cross Sectional
Study Design
• One stop shopping
• Less expensive
• Potentially can detect effect of exposure
that do not vary over time
Disadvantages of Cross
Sectional Study Design
• May not detect associations that for
diseases with short duration that have
cyclic or seasonal trends (e.g. some
infectious diseases)
• May require over-sampling of low
prevalence exposure or disease groups
Design Case-control
Subiecti / femei la menopauza


Cazuri: 100 femei cu accident coronarian
Controale: 100 femei fara patologie cardiaca
Predictor:

Au luat estrogeni post-menopauza?
Rezultate:

Cazuri vs controale
Controls
• Case Control Studies compare
exposure history of cases and controls
• Controls provide exposure distribution
of exposure (exposed and unexposed
cohorts) in source population that gave
rise to the cases in the study
• Biggest Problem: Identifying an
appropriate control group
Sources for Controls
•
•
•
•
Population Controls
Neighborhood Controls
Random-Digit Dialing
Hospital-Based Controls
Nested Case Control Studies
• Every Case Control Study can be thought as
being conducted within a cohort study


Cases represent the outcomes of the cohort
Controls provide estimate of exposure distribution
of cohort
• Traditionally, Nested Case Control Study
usually refers to study that are conducted
within the experience of a well-defined cohort
study
2x2 Table
Exposure +
Exposure Total
Case
a
c
M1
OR = (a/c) / (b/d)
 (a/c) / (K1/K0)
 IRR
Control
b
d
M0
Hospital-Based Controls
• Most appropriate for hospital-based
cases with ill-defined catchment area as
source population
• Often chosen from diagnostic groups
not thought to be related to the
exposure


better to choose more than one diagnostic
group for controls
minimizes impact of a poor choice
Hospital-Based Controls
• History of diagnosis for different
condition related to exposure is not
grounds for exclusion

Some cases may have the same history
• Assumption:


catchment area for control diagnosis is
same as catchment area for cases
control reflect exposure distribution for
source population of cases
Comparison of Case Control
and Cohort Studies
• Primary Difference:


Cohort Study involved complete
enumeration of the source population
Case Control Study involved a sample
(controls) of the source population
• Case Control Studies are modified Cohort
Studies
When is a case control study
warranted?
• When there is good evidence of an
association between a certain exposure
and the disease
• When disease is rare and exposure is
frequent among exposed
• Advantage:


Cost- relatively inexpensive
Sample size –relatively small
Analysis of data
-case control study
Begin with cases with disease and controls without
disease, then measure past exposure
Disease + Disease total
Exposed +
Exposed Proportion of
exposed
a
b
a+b
c
d
c+d
a/a+c
b/b+d
Odd ratio= a/a+c / b/b+d= ad/bc
Limitation
• Selection of appropriate controls
• Recall bias
- Human ability to recall
- Recall bias (Case may remember their
exposure more than controls do)
Etic?
• Echivalenta (nu este sigur daca domina
beneficii sau riscuri)

Beneficii ale terapiei de substitutie
• Scad simpotomele de climax
• ? Preventia fracturilor
• ? Preventia BCI
• ? Preventia Alzheimer
• ? Cresterea calitatii vietii

Riscuri
• ? trombo-embolism pulmonar
• ? Cancer mamar
Relevant?
• Estrogeni / estroprogestative
• Decizia afecteaza o mare parte din
populatie
TRECUT
PREZENT
A primit HRT?
No
Cancer mamar
Yes
Cazuri
No
Yes
Controale
Fara cancer de san
Case report or series :
• Description of one or several cases in
which no attempt is made to answer
specific hypotheses or to compare
results with another group of cases.

classical clinical study; its value probably
under rated (Pickering Lancet 19##)
Case report or series
• Description of one or several cases in which
no attempt is made to answer specific
hypotheses or to compare results with
another group of cases
• No control
• Classical clinical study
• Its value probably under rated
Case Reports and Case Series
Example: investigators at a single institution
note that a small number of patients treated
with a alternative (complementary) therapy exhibit
longer survival and better survival than "expected"
Example: investigators notice a higher than expected
incidence of a cancer in a community
Example: investigators note a common occupational
exposure in a series of patients with a rare cancer
Case Reports and Case Series
• Problems: lack of a control group
no measure of effect
 no formal statistical analysis

• Advantages: hypotheses
generators
Case Series With Literature
Controls
• Compare outcomes of case series to
"expected" results based on existing
records (e.g. published results, national
incidence/mortality rates)
• Problems: lack of comparability

Confounding
• Healthy Worker Effect

Bias
5. Ecologic or trend study :
• Group based study
• Unit of observation is a group eg
country, state etc.

Observed effect difficult to interpret at
individual level (ecologic fallacy).

Otherwise efficient to do.
5. Ecologic or trend study :
• Group based study, unit of observation is a group
e.g. country, state etc.
• Ignore variability between individuals
• May ascribe to members characteristics they do
not possess as individuals
Age standardized
death rate per 1000
Quantity of salt sold at
Henan province in China
Qualitative research :
• Development of concepts which help us to
understand social phenomena in natural
settings, giving due emphasis to the
meanings, experiences, and views of all the
participants
• It does not primarily seek to provide
quantified answers to research questions
• Particularly useful for explaining complex
phenomena not amenable to quantitative
research
Qualitative research : - cont
• It has an important role in medical research
because health care deals with people and
people are more complex than the subjects of
the natural sciences.
• There is a whole set of questions about
human interaction that we need answers to.
• Obvious areas of application are doctorpatient relationship, treatment compliance,
clinical decision making process, issues on
health service organisation and policy issues.
Qualitative research-cont’d:
• Development of concepts which help us to understand
social phenomena in natural settings, giving due emphasis
to the meanings, experiences, and views of all the
participants
• Useful for explaining complex phenomena not amenable to
quantitative research
• Approaches- Focus group, observation, interview, diary
• Obvious areas of application are doctor-patient relationship,
treatment compliance, clinical decision making process,
issues on health service organization and policy issues
Conceptia unui plan de cercetare
1. Definitia problemei
2. Dezvoltarea planului de cercetare
•
Lista elementelor studiului
•
Identificarea referintelor publicate
3. Introducerea (fondul problemei) si
semnificatia sa (istoric al cercetarilor
anterioare, motivatie, importanta)
Plan de studiu
•
•
•
•
Problema cercetata
Semnificatie (background)
Designul studiului
Populatia de studiu si
esantionarea
• Variabile si masuratori
• Elemente statistice
• Aspecte etice
• Controlul calitatii si
evaluarea datelor
• Problema principala a studiului?
• De ce este important?
• Cum se va face studiul ?
• Care sunt subiectii si cum sunt
selectati?
• Ce masuri vor fi luate?
• Dimensiunea lotului si analiza?
• Exista probleme etice?
• Cum se asigura controlul calitatii
datelor si analizei acestora?
Fond si semnificatie
• Bazat pe experienta
• Atentie la idei noi
• Atentie la tehnologii noi
• Sa fim constienti despre punctele forte si slabe ale
cercetarilor anterioare similare, proprii sau de
literatura
• Sa stim ce dorim sa comunicam / confirmam la final
Fondul si semnificatia problemei /
importanta pentru protocol
•
•
•
•
•
•
Asemanator cu introducerea la un articol
Introduce cercetarea propusa, in context
Descrie motivatia studiului
Ce se stie despre topic
De ce este problema propusa, importanta?
Ce tip de raspuns asteptam de la studiu?
Fond si semnificatie
• Citarea cercetarilor relevante anterioare (inclusiv
propriile date)
• Indica punctele forte si slabe ale studiilor anterioare,
ca si intrebarile ramase fara raspuns.
• Arata cum vor fi folosite rezultatele studiului propus
pentru rezolvarea neclaritatilor si influentarea politicii
de sanatate sau practicii clinice
Gasirea literaturii adecvate
• End note, Reference manager
• Verifica formatul referintelor conform
instructiunilor pentru autori si sunt incluse pe
masura ce se scrie protocolul
• Se pastreaza copii ale tuturor referintelor
• PubMed: www.ncbi.nlm.nih.gov
• Comunicari personale (cine si cand)
• Date nepublicate (Cel mai bine se evita)
Gasirea literaturii adecvate
• End note, Reference manager
• Verifica formatul referintelor conform
instructiunilor pentru autori si sunt incluse pe
masura ce se scrie protocolul
• Se pastreaza copii ale tuturor referintelor
• PubMed: www.ncbi.nlm.nih.gov
• Comunicari personale (cine si cand)
• Date nepublicate (Cel mai bine se evita)
Fundamentele Stiintifice ale Cercetarii
• mai multe principii
fundamentale sunt utilizate in
cercetarea stiintifica:
1.Ordinea
2.Deductia si Sansa
3.Evaluarea si Probabilitatea
4.Ipoteza
Fundamentele Stiintifice ale Cercetarii
• Evaluarea si Probabilitatea(1)
– Cerinta critica a efectuarii
proiectului in cercetare, aceea
care asigura validitatea, este
evaluarea probabilitatii de la
inceput pana la capat
Fundamentele Stiintifice ale Cercetarii
Ipoteza
– ipotezele sunt afirmatii atent construite referitoare la un fenomen
intalnit intr-o populatie
– ipotezele pot fi generate fie prin:
• rationament deductiv
• rationament inductiv rezultat din observatii anterioare
Concluzii
Les sciences sont faites par additions.
Nul ne commence ou acheve.
Nous sommes montes sur les epaules d’un geant. Nous
voyons tot ce que voit le geant, et quelque chose
d’avantage.
Guy de Chauliac, 1684