1. JoAnn Sandra Koslowsky, RPh. 11/16/10
Journal Club
1). The introduction is appropriate in length. It is not too short or too long. It contains
relevant information to the study and what is described in the article itself.
2). The study objectives and hypothesis are clearly stated. They tell you that the
comparison of thiazolidinediones with other oral glucose-lowering agents is not known.
These are meaningful and important, as they state in the introduction, in diabetics to help
prevent complications of type 2 diabetes.
3). The study was experimental in design, as it was a randomized, double-blind controlled
trial. This design is appropriate for the question being investigated.
4). The strengths of an experimental study are that there is a minimal chance of outside
factors being responsible for the outcomes. Accurate comparisons and conclusions can be
made because investigators can control the exposure of treatment and interventions. The
weaknesses of an experimental study are that the study can be expensive and patient
enrollment and long term follow-up can be difficult.
5). The study population was well defined and included patients who had not previously
received medication for their recently diagnosed (within 3 years) type 2 diabetes. They
were predominately white, middle-aged males and females who were obese.
6). The inclusion data was patient age between 30-75, fasting glucose levels between
126-180 mg/dl, with their only treatment being lifestyle management. The exclusion data
included significant hepatic disease, renal impairment, CHF, uncontrolled hypertension,
history of lactic acidosis and unstable or severe angina. Both inclusion and exclusion data
are clear and appropriate.
7). The patients were selected based on age, fasting plasma glucose levels and current
treatment of the type 2 diabetes.
8). Patients were classified as newly diagnosed (within 3 years) type 2 diabetes.
9). Selection bias is possible but was minimized by allowing everyone to have an equal
chance in being placed in any of the three groups. There are no differences between the
three groups, except what intervention they were given.
10). The patients were centrally randomized, which was concealed and stratified
according to the sex of the patients, in blocks of six. This process was clearly explained
and is appropriate.

2. 11). Patient disease and risk factors were considered during the randomization process.
Patients were excluded based on hepatic disease, renal impairment, history of lactic
acidosis, CHF and unstable or severe angina. I do agree that these patients should have
been excluded from the study.
12). The initial intervention was either 4mg of rosiglitazone, 500mg of metformin or
2.5mg of glyburide, all given once daily. Each drug was increased, per protocol, to the
maximum effective dose of 4mg rosiglitazone twice daily, 1000mg metformin twice daily
and 7.5mg glyburide twice daily. The drug, dose and route are all mentioned and they are
appropriate drugs and doses.
13). There were two active control groups which were drugs (metformin and glyburide)
that had established efficacy and the results of these two groups were compared to the
study group (rosiglitazone).
14). Compliance is assessed and noted that 9 patients were excluded who did not receive
the drug and 224 patients who withdrew before the first scheduled evaluation of efficacy.
15). It was a double-blind study so both the patients and investigators were unaware of
which patients received what treatment. Preparation was done at a central location and all
drugs were prepared in identical capsules, so no one could tell them apart.
16). The primary outcome was the time to monotherapy failure on the basis of glucose
levels of greater than 180 mg/dl, after an overnight fast. I do believe that all important
outcomes were considered. Secondary outcomes were fasting glucose levels, glycated
hemoglobin levels, weight, insulin sensitivity and beta-cell function.
17). The measures used to evaluate the outcomes were adequately described. The primary
outcome measurement was fasting plasma glucose levels. The secondary measurements
were glycated hemoglobin, beta-cell function, weight, waist circumference, hip
circumference and waist-to-hip ratio.
18). The patients were followed every two months in the first year and every three
months thereafter. The blood draws were in person. I believe the frequency was
appropriate. If the blood draws were more often, patients may have dropped out due to
inconvenience.
19). The length of the study was a median period of four years, which was an appropriate
amount of time to answer the study’s question.
20). The measures are appropriate. The measurements used are the standard of care for
patients with type 2 diabetes. The data used to measure the outcomes was objective and
not subjective. I believe the measures are reproducible.

3. 21). There were steps taken in the study to reduce observer bias. Using objective data to
measure the outcomes is one example. Another was the use of an independent
adjudication committee, whose members were unaware of treatment assignments, to
determine if the primary outcome was reached. They used prespecified criteria to
determine this.
22). There is no information bias in this study.
23). Recall bias is not an issue in this study because the patients were not used as sources
or information.
24). There are sources of confounding bias. The study took measures to minimize these
such as having exclusion criteria, having patients be equal at baseline, by double-blinding
and by randomization that was concealed and stratified.
25). The statistical test, ANOVA, was described adequately and the appropriate data,
which was continuous, was used in this study.
26). The sample size, power and level of significance have been described. The initial
enrollment of 3600 patients was amended and increased to 4182 patients, to compensate
for a higher anticipated rate of withdrawal. This number of 4182 patients was needed to
achieve power. Risk reduction, cumulative index and P values were reported for all three
groups and they were all appropriate.
27). The data was analyzed on an intent-to-treat basis. The analysis was strong because it
accounted for the question of why patients dropped out of the study. There were not any
negatives associated with this analysis because the investigators increased the patient
enrollment due to the initial high drop out rate.
28). All the patients enrolled in the study have been accounted for. The number of
dropouts, which was 224, was accounted for by the investigators for a primary reason of
adverse events.
29). The study evaluated 4351 patients which is an adequate number of patients studied.
30). The patient demographics represented in this study were middle-aged, obese and
predominately white males and females. There demographics are representative of those
patients who would have type 2 diabetes. The groups did not have any significant
differences in baseline variables.
31). The primary outcome of the study showed that the number of patients who failed
monotherapy at five years was least with rosiglitazone. All outcomes are discussed such
as monotherapy failure, risk (incidence) reduction, treatment effect among older patients
and those with larger waist circumference, rate of progression to a confirmed fasting
glucose level greater than 140 mg/dl and glycemic measures.

4. 32). The results are presented appropriately. The authors used four tables and one graph
to represent baseline characteristics, primary outcomes, secondary outcomes and adverse
events.
33). The safety of the intervention was discussed. The authors mentioned CHF events,
edema, GI effects, alanine aminotransferase levels, hematocrit levels and LDL levels.
34). The authors do discuss limitations of the study, such as the high rate of withdrawal
of patients. They do a good job of discussing this limitation and its implications. I do not
see any other limitations of the study.
35). The author’s conclusions are consistent with the results. They concluded that
rosiglitazone delayed monotherapy failure better than metformin and gluburide. The
results of patients who failed monotherapy support this conclusion.
36). I do feel that the abstract does support the article, although there was no mention of
the safety concerns of the thiazolidinedione, as was mentioned in the discussion part of
the article.
37). This article was published in the New England Journal of Medicine, which is a
reputable and peer-reviewed journal.
38). The study was funded by grants from GlaxoSmithKline (GSK). The role of GSK in
this study was discussed. Two of the seven members of the study oversight steering
committee were from GSK. The sponsor, GSK, imposed no restrictions in the decision to
publish the article, which did minimize their role.
39). The study does have internal validity. The results given in the primary outcome
section of the article support the authors conclusion in the discussion that rosiglitazone
slowed progression to monotherapy failure better than metformin and glyburide.
40). The study does have external validity. I believe there are many patient’s in the
general population, with type 2 diabetes, who have similar baseline characteristics to
patients in this study. Therefore, there are numerous people who can benefit from using a
thiazolidinedione as first line choice monotherapy in the treatment of type 2 diabetes.
41). The results are meaningful and important. The study shows that the use of
rosiglitazone can delay loss of glycemic control better than metformin or glyburide. This
would change my practice when consulted about monotherapy for patients with type 2
diabetes, who fell into the same category as patients in this study. I would recommend a
thiazolidinedione as a first choice of therapy.

5. I believe this study was strong and well designed. The investigators posed and researched
an intervention that can be considered critical in the care of patients with type 2 diabetes.
They adequately assessed the information revealed in the study for relevance, validity
and importance. They used and experimental design which is considered the “Cadillac”
of studies, although difficult and costly. I feel that the number of years the patients were
followed was certainly a long enough period of time. The investigators used a large
number of patients in their study and the selection and allocation process was perfectly
fine. Randomization and double-blinding is something we like to see in a study and was
used here. There was limited bias throughout the study and although it was sponsored by
GlaxoSmithKline, the drug company had limited input into the publication of the study. I
believe the investigators did an excellent job in following through on a well-planned and
though out study. Overall, I think the study was very good and reflective of an
intervention that can be very helpful to the current population, for which type 2 diabetes
is very common.