For each scenario below identify 1) the ideal study design, 2) other potential study designs and 3) the advantages and disadvantages of the nominated study designs.
a) A dietician is interested in identifying the risk factors present in infancy associated with obesity in primary school aged children.
The ideal study design to answer this question is?
Cohort study, this type of study will look at the associations between the risk factors
and the outcome of interest. The defined population (infants), that have not yet
experienced the outcome and who have a known or measurable exposure to a
potential risk factor. The population is observed overtime and there subsequent
outcomes identified (obesity in primary school aged children).
(Doust & Sanders, 2005)
Other potential study designs to answer this question are?
Case controlled study. This type of study looks at a population with and without a
certain disease or problem (Obesity in primary school aged children) and looks back
to see what they were exposed to (risk factors in infancy that contributed to obesity in
primary school) (Doust & Sanders, 2005).
What are the advantages and disadvantages of the different study designs to
answer this question?
The advantages of a cohort study include they are ethically less challenging than
randomised controlled trials (RCT’s), they can sometimes be cheaper and easier than
a RCT, eligibility and outcome criteria can be standardised and the author can
establish the timing between factor and event. The disadvantages include potential for
losses to follow-up, large sample sizes and long follow-up is needed, there is no
randomisation so there is an increased potential for confounding and selection bias.
In this case a long follow up is needed with close monitoring of the child’s lifestyle.
(Doust & Sanders, 2005).
The advantages of case controlled studies are they are reasonably fast and cheap, they
are the only feasible method for very rare disorders, or those with long lags between
exposure and outcome. The disadvantages include having to rely on recall or records
to determine exposure status, this may lead to measurement bias and inevitable
confounding. The control group selection can be difficult and may lead to selection
bias. In this case it may be reasonably fast and cheap to question the parents of obese
primary school children but the authors of the study will have to rely on the parents
memories, as well as the risk of the parents changing some of their answers if they
think it will make them look better (confounding as participants not blinded). (Doust
& Sanders, 2005).
b) The health department is considering the usefulness of needle exchange programs for preventing transmission of blood borne diseases among intravenous drug users.
The ideal study design to answer this question is?
Randomised Controlled Trials, (RCT’s) because the health department wants to find if
the intervention of a needle exchange programme will be effective in preventing the
transmission of blood borne disease among IV drug users an RCT will be able to
assess the effectiveness of therapy or prevention, it is the only study that can
adequately minimise confounding and therefore validly estimate the effectiveness of
an intervention (Doust & Sanders, 2005).
Other potential study designs to answer this question are?
Cohort Study, this type of study will look at the associations between the risk factors
and the outcome of interest. The defined population (IV drug users) that have not yet
experienced the outcome and who have a known or measurable exposure to a
potential risk factor (blood borne diseases). The population is observed overtime and
there subsequent outcomes identified (did the needle exchange programme reduce the
transmission of blood borne diseases). Another study design for this question is a
case controlled study. This type of study looks at a population with and without a
certain disease or problem and looks back to see what they were exposed to. In this
case did the population using the needle exchange programme have a lower incidence
of blood borne diseases than those who were not involved in the programme? (Doust
& Sanders, 2005).
What are the advantages and disadvantages of the different study designs to answer this question?
The advantages of randomised controlled trials (RCT’s) include successful blinding of
subject and/or the investigator if possible, randomisation facilitates unbiased
distribution of confounders. Eligibility and outcome criteria can be standardised and
the authors are able to establish the timing between cause and effect. The
disadvantages of RCT’s is they can be ethically problematic at times, RCT’s can be
expensive both in time and place, it can be difficult to enrol a large enough sample,
generalisablity may be limited due to strict eligibility criteria, there is potential for
loses to follow-up which threatens the validity of the study results. In this case it may
be difficult to keep track of a large group of IV drug users, due to movement, jail,
death and difficultly in obtaining consent, however due to the lack of confounders the
establishment of timing between the cause and effect this study is a good way to see if
needle exchange programmes are an effective intervention (Doust & Sanders, 2005).
The advantages of a cohort study include they are ethically less challenging than
randomised controlled trials (RCT’s), they can sometimes be cheaper and easier than
a RCT, eligibility and outcome criteria can be standardised, the author can establish
the timing between factor and event. The disadvantages include potential for losses to
follow-up, large sample sizes and long follow-up is needed, there is no randomisation
so there is an increased potential for confounding and selection bias. May be difficult
to follow-up due to movement, death, imprisonment, lack of interest by participants.
(Doust & Sanders, 2005).
The advantages of case controlled studies are they are reasonably fast and cheap; they
are the only feasible method for very rare disorders, or those with long lag between
exposure and outcome. The disadvantages include having to rely on recall or records
to determine exposure status may lead to measurement bias, inevitable confounding,
and the control group selection can be difficult and may lead to selection bias. This
may be quite difficult due to the unreliability of most drug users in terms of memory
(Doust & Sanders, 2005).
c) A woman whose partner has been hospitalised with influenza is prescribed an antiviral and is advised to take it prophylactically. The drug (a neuraminidase inhibitor) is expensive and she wonders how effective it is at preventing influenza.
The ideal study design to answer this question is?
RCT’s. In this type of study the purpose of randomisation is to evenly distribute the
individual characteristics or variables (confounding factors) making both groups as
similar as possible, if one group was sicker than the other at the start of the study it
might make one treatment look better or worse when in reality there is no real
difference between the treatments. A RCT is able to give the best evidence for the
effectiveness of a therapy or in this case the antiviral prescribed above (Doust &
Sanders, 2005).
Other potential study designs to answer this question are?
Cohort Study, this type of study will look at the associations between the risk factors
and the outcome of interest. The defined population that have not yet experienced the
outcome and who have a known or measurable exposure to a potential risk factor.
The population is observed overtime and there subsequent outcomes identified (did
the antiviral reduce the incidence of influenza?). Another study design for this
question is a case controlled study. This type of study looks at a population with and
without a certain disease or problem and looks back to see what they were exposed to
(Doust & Sanders, 2005).
What are the advantages and disadvantages of the different study designs to answer this question?
The advantages of randomised controlled trials (RCT’s) include successful blinding of
subject and/or the investigator if possible, randomisation facilitates unbiased
distribution of confounders, eligibility and outcome criteria can be standardised and
the authors are able to establish the timing between cause and effect. The
disadvantages of RCT’s is they can be ethically problematic at times, RCT’s can be
expensive both in time and place, it can be difficult to enrol a large enough sample,
generalisablity may be limited due to strict eligibility criteria, there is potential for
loses to follow-up which threatens the validity of the study results (Doust & Sanders,
2005).
The advantages of a cohort study include they are ethically less challenging than
randomised controlled trials (RCT’s), they can sometimes be cheaper and easier than
a RCT, eligibility and outcome criteria can be standardised, the author can establish
the timing between factor and event. The disadvantages include potential for losses to
follow-up, large sample sizes and long follow-up is needed, there is no randomisation
so there is an increased potential for confounding and selection bias (Doust &
Sanders, 2005).
The advantages of case controlled studies are they are reasonably fast and cheap; they
are the only feasible method for very rare disorders, or those with long lag between
exposure and outcome. The disadvantages include having to rely on recall or records
to determine exposure status may lead to measurement bias, inevitable confounding,
and the control group selection can be difficult and may lead to selection bias (Doust
& Sanders, 2005).
d) A child health nurse is pregnant with her second child and has maternal hypothyroidism (an under active thyroid). She has heard that this can have an affect on her child’s development and wonders what the available evidence suggests.
The ideal study design to answer this question is?
Cohort study, this type of study will look at the associations between the risk factors
and the outcome of interest. The defined population that have not yet experienced the
outcome and who have a known or measurable exposure to a potential risk factor
(Doust & Sanders, 2005).
Other potential study designs to answer this question are?
Case- Control study, this type of study looks at a population with and without a
certain disease or problem and looks back to see what they were exposed to (Doust &
Sanders, 2005).
What are the advantages and disadvantages of the different study designs to answer this question?
The advantages of a cohort study include they are ethically less challenging than
randomised controlled trials (RCT’s), they can sometimes be cheaper and easier than
a RCT, eligibility and outcome criteria can be standardised and the author can
establish the timing between factor and event. The disadvantages include potential for
losses to follow-up, large sample sizes and long follow-up is needed, there is no
randomisation so there is an increased potential for confounding and selection bias
(Doust & Sanders, 2005).
The advantages of case controlled studies are they are reasonably fast and cheap; they
are the only feasible method for very rare disorders, or those with long lag between
exposure and outcome. The disadvantages include having to rely on recall or records
to determine exposure status may lead to measurement bias, inevitable confounding,
and the control group selection can be difficult and may lead to selection bias (Doust
& Sanders, 2005).
e) A GP is wondering whether waist circumference is a useful measure for identifying patients who need weight management programs.
The ideal study design to answer this question is?
Cross-sectional study, a cross-sectional measures both factors of interest and
outcomes at one point in time (Doust & Sanders, 2005).
Other potential study designs to answer this question are?
Case-control study this type of study looks at a population with and without a certain
disease or problem and looks back to see what they were exposed to (Doust &
Sanders, 2005).
What are the advantages and disadvantages of the different study designs to answer this question?
The advantages of a cross sectional study include they are relatively fast and cheap,
the authors can collect accurate baseline data, may be able to suggest an association
between exposure and outcome. The disadvantages include the group sizes will be
unequal in a ‘free living’ population, confounders will be unequally distributed, the
study does not readily establish causality (Doust & Sanders, 2005).
The advantages of case controlled studies are they are reasonably fast and cheap; they
are the only feasible method for very rare disorders, or those with long lag between
exposure and outcome. The disadvantages include having to rely on recall or records
to determine exposure status may lead to measurement bias, inevitable confounding,
and the control group selection can be difficult and may lead to selection bias (Doust
& Sanders, 2005).
Consider each study excerpt below and identify the study design (2 marks for each question).
a) Excerpt 1
Method: Subjects were 39 adolescents who stutter, between ages 13 and 18 years (mean AGE=14.4 years), and 39 adolescents who did not stutter (mean AGE=14.6 years), all of whom were currently enrolled in school. Control subjects were recruited from local school districts and matched for same grade, ethnicity, gender, and approximate age. Information about the demographic characteristics of the subjects and the onset, duration, perceptions, and chronicity of stuttering were obtained. Information was also collected about familial history of stuttering, types and duration of treatment, and concomitant speech, language, or associated problems. Significantly higher levels of communication apprehension and poorer scores on self-perceived communication competence were found in adolescents who stutter when compared with adolescents who do not stutter. Subscore test data revealed that adolescents who stutter had significantly greater fears about speaking in Group Discussions and Interpersonal Conversations than they had about Public Speaking and talking during Meetings, when compared with students who do not stutter. They also had significantly poorer perceptions about their own communication competence on the Talking to Strangers subscore test when compared with students who do not stutter. A significant positive relationship among stuttering severity, communication apprehension, and self-perceived communication competence total scores was found. Students who stutter severely had greater fears about speaking in group discussions and interpersonal conversations.
This is a case controlled study because the authors gathered subjects (not randomly)
who stutter and matched them to a control case (no stuttering). The authors then
looked back at exposure to see what the outcome of stuttering had on people (Doust &
Sanders, 2005).
b) Excerpt 2
SUBJECTS: Twenty-seven people with 42 chronic leg ulcers participated in the study. METHODS: The subjects were separated into subgroups according to primary etiology of the wound (diabetes, arterial insufficiency, venous insufficiency) and then randomly assigned to receive either HVPC (100 microseconds, 150 V, 100 Hz) or a sham treatment for 45 minutes, 3 times weekly, for 4 weeks. Wound surface area and wound appearance were assessed during an initial examination, following a 1- to 2-week period during which subjects received only conventional wound therapy, after 4 weeks of sham or HVPC treatment, and at 1 month following treatments.
This study is a Randomised controlled study, as the patients were randomly assigned
to a treatment (Doust & Sanders, 2005).
c) Excerpt 3
METHOD: The authors reviewed the charts of 59 female patients with anorexia nervosa who were transferred from 24-hour inpatient care to an eating disorder day hospital program. They evaluated the prognostic significance of a variety of anthropometric, demographic, illness history, and psychometric measures. RESULTS: Greater risk of day hospital program treatment failure and inpatient readmission was associated with longer duration of illness (for patients who had been ill for more than 6 years, risk ratio = 2.7), amenorrhea (for patients who had this symptom for more than 2.5 years, risk ratio = 5.7), or lower body mass index at the time of inpatient admission (for patients with a body mass index of 16.5 or less, risk ratio = 9.6; for those with a body mass index 75% or less than normal, risk ratio = 7.2) or at the time of transition to the day hospital program (for patients with a body mass index of 19 or less, risk ratio = 3.9; for those with a body mass index 90% or less than normal, risk ratio = 11.7). CONCLUSIONS: Inpatients with anorexia nervosa who have the poor prognostic indicators found in this study are in need of continued inpatient care to avoid immediate relapse and higher cost and longer duration of treatment.
Cross-sectional study because the authors measured both the factors of interest
and the outcomes at the same time (Doust & Sanders, 2005).
d) Excerpt 4
An advertisement for an over the counter product has data of the effect of the oil on the appearance of scars. Please go to the following website and determine what type of study this is: "http://www.bio-oil.info/australia/user_trials.php"
This is a Cohort study because the identified population was observed over a period of
time to see what the effects of the exposure (Bio-oil) had on the appearance of scars
(Doust & Sanders, 2005, http://www.bio-oil.info/australia/users_trials.php)
Consider the questions below. What evidence source/s would you look in first? (Assume, for the purposes of this assignment, that you have access to all the evidence sources discussed in the course material). (5 marks for each question)
a) In an outpatient setting, what is the accuracy of a brief screening instrument that uses self-report for detecting migraine headaches?
The best study design would be cross sectional or cohort. Synopses would be a great
place to start; a synopsis contains the minimum information needed to make a clinical
decision without having to access the full article. The Cochrane library uses DARE
which has access to nearly 3000 systematic reviews. Other evidence-based medical
and nursing resources are a good place to look such as Cinahl or online journals such
as evidence based nursing available through the UQ Cybrary. If the question cannot
be answered this way the next place to look is syntheses, they collate, appraise and
summarise the available primary studies to answer a specific question. A great
resource for these is the Cochrane Database of systematic Reviews as well as DARE
in the Cochrane library. If still unable to find the answer to the question the next step
is to access the primary research study, this can be done through search engines such
as PubMed, Medline either Webspires or Ovid, Central (Cochrane library) or the
internet. (Doust & Sanders, 2005; http://www.thecochranelibrary.com.au,
b) Among adults with an acute sore throat, how effective are antibiotics at reducing symptoms?
A systematic review of randomised trials is the best place to start this can be done
through the Cochrane Library; other research sources include Clinical Evidence,
available through UQ Cybrary and Bandolier and Up To Date, available through the
Internet. Because this is a relatively new system not every question may be answered
this way. If this is the case Synopsis (They contain the minimum information needed
to make a clinical decision) is the next best place to look they can be found in online
evidence based journals available through UQ Cybrary or DARE in the Cochrane
Library. If the question still cannot be answered this way the next place to look is
syntheses, they collate, appraise and summarise the available primary studies to
answer a specific question. A great resource for these is the Cochrane Database of
Systematic Reviews as well as DARE in the Cochrane library. If still unable to find
the answer to the question the next step is to access the primary research study, this
can be done through search engines such as PubMed, Medline either Webspires or
Ovid, Central (Cochrane library) or the Internet. (Doust & Sanders, 2005).
Conduct a search in the Cochrane Library for evidence to answer the following questions. Report the exact search terms used and present the reference of the best study to answer each question.
a) What effect does attendance at childcare during infancy and early childhood have on a child’s performance and behaviour at school?
As this question type is ‘prognosis’ the best type of study to answer this question is a
Systematic review of cohort studies, this can be done by accessing the Cochrane
library of systematic reviews (systematic reviews locate and evaluate all available
evidence). The first step is to look at the question using the PICO formula: P = early
childhood, I = attendance at childcare, C = no attendance at childcare, O = child’s
behaviour and performance at school. After accessing The Cochrane library, click on
Cochrane advanced search, this brings up a list of all the search types you can conduct
on Cochrane including, DARE (abstracts of systematic reviews), Central (abstracts of
trials) and Cochrane systematic reviews. Click on Cochrane Systematic Reviews to
limit the search to this section of the library. In the first search box type in ‘early
childhood’ ensuring ‘search all text’ is marked next to it, in the second search box
ensuring ‘AND’ was marked not ‘OR’ or ‘NOT’ type in ‘child care attendance’
ensuring ‘search all text’ is marked and not ‘author’ etcetera, in the third search box
once again ensuring ‘AND’ and ‘search all text’ is marked type in performance and
behaviour and clicked on search. 11 articles appeared such as ‘thioridazine for
schizophrenia’, ‘screening children in the first four years of life to undergo early
treatment for otitis media with effusion’ and ‘male circumcision for prevention of
heterosexual acquisition of HIV in men’. There was one article called ‘Day care for
pre-school children’ (Zoritch B, Roberts I, Oakley A. Day care for pre-school
children. The Cochrane Database of Systematic Reviews 2000, issue 3. Art. No.:
CD000564.DOI:10.1002/14651858.CD000564) that answered the above question. It
is a systematic review conducted to assess the effects of day-care on children and
families. The results of the review show that children who attend day care have better
behaviour when older than those children who didn’t. There is even evidence to
suggest that children who attend day-care are less likely to develop criminal
behaviour in later years. The review also discusses other advantages and
disadvantages of sending children to day care. Interestingly when ‘attendance at
childcare’ ‘AND’ ‘early child care’ ‘AND’ ‘performance and behaviour at school’
was typed into the search boxes no articles appeared in the results list. (Doust &
sanders, 2005; http://www.thecochranelibrary.com.au,
b) In people with chronic, low back pain, does sleeping on a firm mattress help to reduce pain and disability?
This question is looking at treatment; the best way to answer this question is with a
systematic review of randomised trials. The first step is to look at the question using
the PICO formula, P = people with chronic low back pain, I = firm mattress, C = other
type of mattress, O = reduced pain and disability. After accessing the Cochrane
Library click on advanced search and request a search of all Cochrane Library as
when the results are brought up each system is represented with the number of results
it contains for a particular search (DARE, CENTRAL, MeSH, Systematic Reviews
etc). This is particularly handy in saving time. Ensuring that the search boxes had
‘search all text’ and ‘AND’ marked next to them type in ‘back pain’ ‘AND’ ‘mattress’
this brings up 6 Cochrane Reviews (systematic reviews) such as Position in second
stage labour for women without epidural anaesthesia and repositioning for pressure
ulcer prevention and 11 articles in the CENTRAL database (containing abstracts of
trials) including Short term outcomes of chronic back pain patients on an air mattress
verus a innerspring mattress and Associations between back pain, quality of sleep and
quality of mattress, double blind pilot study. Some of these articles talked about
sleeping on various mattresses and the effects on back pain. To refine the search
further to see if there were articles that specifically talked about firm mattresses and
lower back pain, type in ‘low back pain’ ‘AND’ ‘firm mattress’ into the search engine
using the same limiters as before and the search produced three Cochrane reviews and
two CENTRAL articles I looked at the Cochrane reviews first but this search
produced articles such as Support surfaces for pressure ulcers. Under Central there
were two articles both answering the question and both suggesting that a more firm
mattress will reduce back pain, however the article entitled ‘effect of firmness of
mattress on chronic non-specific low back pain: randomised, double blind, controlled,
multicentre trial (Kovacs FM, Abraira V, Pena A, Martin-Rodriguez JG, Sanchez-
Vera M, Ferrer E, Ruano D, Gullen P, Gestoso M, Muriel A, Zamora J, Gil del Real
MT, Mufraggi N, 2003. Vol362, 936: 1599-1604), answers the question the best. It
has a wider group of people, longer time frame and it is a randomised controlled trial.
The article suggests that a medium firm mattress is effective in reducing non-specific
low back pain (Doust & Sanders, 2005: http://www.thecochranelibrary.com.au
In recent years, there has been considerable controversy over the possible association between non-steroidal anti-inflammatory drugs (both Cox II and non-selective NSAIDs) and the risk of myocardial infarction. Below are extracts from a study trying to identify the association between non-selective NSAIDs and myocardial infarction. Read the excerpt and answer the questions below:
Excerpt from study of NSAIDs and risk of myocardial infarction
We identified, through a computer search of the General Practice Research Database, potential cases of a first-time acute myocardial infarction between January 1995 and April 2001 in patients aged 89 years or younger. Subjects with a database history of less than 3 years before the index date (the date of the acute myocardial infarction) were excluded. We reviewed the computer records of all potential cases, blinded to any information on NSAID exposure.
According to previous validation studies, a high percentage (> 90%) of selected potential cases with acute myocardial infarction can be confirmed by the presence of specific diagnostic criteria in hospital discharge letters.27–29 Because of this fact, we included all potential cases after identifying them by manual review of patient profiles.
Control Subjects
Four control subjects (i.e., patients without acute myocardial infarction) were identified at random and matched to a case patient on age (± 1 yr), sex, general practice attended, number of years of recorded history in the database, and calendar time (by using the same index date— i.e., the date of the acute myocardial infarction diagnosis in the corresponding case patient). Control subjects with a history of less than 3 years in the database were excluded.
What type of study design is this?
Case control study because the authors found people with and with out the outcome,
in this case myocardial infarction (MI) and looked back to see if they had been
exposed to a NSAID or not and if it made a difference to the risk of having an MI.
Even though the authors were blinded to what NSAID the subjects were taking they
were not randomly allocated to the exposure – NSAID, they were already taking when
the study commenced.
b) Do the results in Table 2 indicate that NSAID use is associated with an increased or a reduced risk of myocardial infarction?
According to the study the use of NSAID is associated with an increased risk of
myocardial infarct, although as can be seen by the odds ratio not all NSAID show a
high risk of contributing to MI. Napoxen, Piroxican, Ketoprofen, Nabumetone,
Flurbiprofen and Tiaprofenic acid show odds of less than one indicating protection.
c) Are there any differences between the different types of NSAID?
Yes, most of the NSAID listed above work by producing anti-inflammatory and
analgesic effects possibly inhibiting the prostaglandin synthesis, however they have
different side effects for example Diclofenic, and Indomethacin, can cause heart
failure, oedema, and hypertension, where as Naproxen and Ketoprofen have no direct
cardio vascular adverse reactions, each drug may also have different drug interactions
(Dwyer, Ed, 1999).
d) What types of factors might be used to adjust the odds ratio in the last column?
Whether the person was taking other medications such as aspirin, their predisposing
factors, such as general health, smoking, family history and obesity.
e) The adjusted odds ratio for patients using aspirin and without any other NSAID exposure was 0.87 (95% CI 0.75-1.00). The adjusted odds ratio for patients using both aspirin and a NSAID was 0.74 (95% CI 0.57-0.97).
What do these numbers imply?
The odds ratio refers to the relative risk, that is the likelihood of an event in the
intervention group compared with the control group. A score of more than one
indicates risk, a score of less than one usually indicates protection. People who are
taking aspirin and a NSAID have a slightly more reduced risk of having an MI than
those just taking aspirin (Doust & Sanders, 2005)
f) Based on this data, do you believe that NSAIDs increase or decrease the risk of myocardial infarction? What else might you like to know about the study? Would you like more data? If so, what type of data would you like?
I believe some NSAIDs increase the chance of having an MI and some don’t, as can
be seen by the odds ratio. I could not make a decision without seeing the whole study,
what are the confounders if any, what is the background, is there more information on
the people in the trials, are they likely to have an MI regardless of NSAID use? More
information on NSAIDs used in the study is needed. I would like to see a RCT on
this subject, were there any confounders or selection bias.
Think of a health care question related to your current working environment. Formulate the question using the PICO format. Where might you search for answers to this question? What type of study design would be ideal to answer your question?
What is the best treatment for chronic otitis media of aboriginal children living
in outback communities?
P = Aboriginal children with chronic otitis media
I = Best treatment
C = No treatment or the current treatment of amoxicillin
O = Effective treatment of chronic otitis media
The type of study design ideal for this question on treatment is a randomized control
trial; the best source of these is the Cochrane library. If there are currently no
systematic reviews then other sources include CENTRAL and DARE or other
databases such as Medline and cinahl. Cohort and case-control studies will also be
able to answer the question.
Doust & Sanders, 2005. Evidence Based Health Care. The University of Queensland.
Dwyer (Ed), 1999.
Australasia Nursing Drug Handbook. Springhouse Corporation, Pennsylvania. 
No comments:
Post a Comment