Discussion articles

Order Description

provided three articles; you will be required to discuss each article in separate and to provide 175 words for each. These are problem-based cases where you are required to review a journal article from pre-1990 then discuss the current climate. You should discuss if the findings of the article still hold or if medical advances have changed anything. What is the current epidemiology in Australia, worldwide? Are there sub-populations with factors you should take into account in Australia (e.g. indigenous Australians, CALD)? Internationally? How would someone be treated today and what would the prognosis be compared to when the article was produced. For each discussion you must follow the following:
€¢ Only address one article per discussion €“ all your discussion writing should relate to that article
€¢ Each writing discussion you make can only be a maximum of 175 words (not including citations and references).
You will provide in text citation and referencing list in each writing discussion separately as each article has different marking of each.

1) Article one provided (AIDS); separately with its own discussion and in text citation and referencing list provide 3 references at least

Discussion 175 words exceled (In texnt citation) Referencing list

2) Article 2 Aspirin (Cardiovascular)
Separately with its own discussion and in text citation and referencing list provide 3 references at least

Discussion 175 words exceled (Intent citation) Referencing list

3) Article 3 Circumescion

Separately with its own discussion and in text citation and referencing list provide 3 references at least

C H Hennekens, L K Karlson and B Rosner
A case-control study of regular aspirin use and coronary deaths.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
Copyright © 1978 American Heart Association, Inc. All rights reserved.
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
doi: 10.1161/01.CIR.58.1.35
Circulation. 1978;58:35-38

http://circ.ahajournals.org/content/58/1/35

the World Wide Web at:
The online version of this article, along with updated information and services, is located on

http://circ.ahajournals.org//subscriptions/

Subscriptions: Information about subscribing to Circulation is online at:

http://www.lww.com/reprints

Reprints: Information about reprints can be found online at:
information about this process is available in the Permissions and Rights Question and Answer document.
located, click Request Permissions in the middle column of the Web page under Services. Further
Editorial Office. Once the online version of the published article for which permission is being requested is
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
Downloaded from http://circ.ahajournals.org/ by guest on December 26, 2012
ORIGINAL ARTICLES
A Case-Control Study
of Regular Aspirin Use
and Coronary Deaths
CHARLES H. HENNEKENS, M.D., LYNNE K. KARLSON, B.A.,
AND BERNARD ROSNER, PH.D.
SUMMARY Information was collected for a large number of coronary risk factors on a series of 568 married, white men,
aged 30-70 years, who died from coronary heart disease. Information on the same risk factors was collected on an equal
number of living controls matched on age, sex, marital status and neighborhood. For regular aspirin users (i.e., _ 4 days per
week) compared with non-users, the crude matched pair risk ratio estimate was 1.0 (95% confidence limits 0.9-1.1). Even
after controlling for possible confounding effects of other variables using a paired multiple logistic regression analysis, there
was no evidence of association. These data provide no evidence for a preventive role of regular aspirin intake in coronary
deaths.
IN PHARMACOLOGIC STUDIES, 1,2 the finding
of a reduction of platelet aggregation by aspirin has
suggested its possible preventive role in coronary heart
disease (CHD). However, conflicting results have been
reported in epidemiologic studies of this association.3 6
In a hospital-based case-control study of non-fatal
myocardial infarction (MI), the risk ratio of MI
among regular aspirin users (defined as taking aspirin
four or more days per week) was 0.53 that of nonusers,
a finding which indicated a protective effect of
regular aspirin intake upon MI. In a randomized
clinical trial of aspirin therapy among survivors of
MI,5 the results were not statistically significant but
were compatible with a small protective effect of
aspirin which increased with duration of use. In contrast,
the results of a large prospective cohort study of
deaths due to coronary heart disease6 indicated no
association with aspirin use.
This report evaluates from retrospective data
whether there is an association between regular aspirin
intake and coronary deaths. The design uses male subjects
who died as a result of CHD. Neighbors of these
men are used as controls.
Methods
Study Population
The study population of cases and controls was
restricted to married white men age 30-70 residing in
two Florida counties. Subjects were identified by
From Channing Laboratory, Departments of Medicine and
Preventive and Social Medicine, Harvard Medical School, and the
Peter Bent Brigham Hospital, Boston, Massachusetts.
This study was supported by research grants HL 14141 and HL
05998 from the National Heart, Lung and Blood Institute.
Dr. Hennekens is recipient of a Research Career Development
Award (HL 00286) from the National Heart, Lung and Blood
Institute.
Address for reprints: Charles H. Hennekens, M.D., 180
Longwood Avenue, Boston, Massachusetts 02115.
Received September 6, 1977; revision accepted January 20, 1978.
weekly reviews of death certificates during a 16 month
period. Only men whose deaths were attributed to
CHD within 24 hours after onset of symptoms were
included.
One living control, individually matched as to age
within the same decade (i.e., 30-39, 40-49, etc.) and
neighborhood of residence, was selected for each case
by a systematic household survey.
Of 1,019 wives of eligible patients, 174 were nonrespondents
and an additional 196 had to be excluded
due to non-cooperation. The final study population
consisted of 649 case-control pairs. Of these 649 pairs,
81 were excluded because of missing or unknown
values, so that the data analysis are based on 568 casecontrol
pairs.
Procedures
Letters of introduction were sent to the wives of
eligible patients. An interviewer telephoned the wives
to obtain an interview, which was conducted in the
home between two weeks to two months after the
death of the patient. For each wife of a case interviewed,
a wife of a control was also interviewed. The
interviewer asked questions concerning the husband’s
aspirin consumption during the three months before
death for the case and for the control, as well as
questions about other coronary risk factors. More
detailed descriptions have appeared in a previous
publication.7
Methods of Data Analysis
The aspirin variable was measured by asking
whether aspirin was consumed daily or weekly and, if
so, how many tablets were taken. Regular aspirin
users were defined as those who took aspirin at least
four days per week. Initially, we calculated the
matched pair risk ratio to quantitate the crude
association between regular aspirin intake and coronary
deaths.
35
Downloaded from http://circ.ahajournals.org/ by guest on December 26, 2012
VOL 58, No 1, JULY 1978
To control for possible additional confounding
variables, we used a recently-developed method of
analysis of matched pair studies, which maintains the
matching and allows for the control of additional confounding
variables.8€² In this method, a multiple
logistic regression analysis within pair differences for
all available variables is consecutively entered into the
logistic equation in decreasing order of strength of the
association with the dependent variable, in this case
use or non-use of aspirin. When these other variables
are included in the logistic equation, the residual
association between aspirin use and coronary deaths is
evaluated according to whether the regression constant
significantly differs from zero. Unlike most multiple
regression models where the chief aim is to estimate
the regression coefficients and test them for
statistical significance, in this method we estimate the
regression constant and its standard error. To describe
the association between regular aspirin intake and coronary
deaths, we calculated an adjusted risk ratio. We
defined the risk ratio as the ratio of the likelihood that
a case took aspirin and the control did not, to the
likelihood that the control took aspirin and the case
did not.8€² 9 For each risk ratio estimate we calculated
95% confidence limits.
Results
Table 1 shows the results of various crude matched
pair risk ratio estimates and their 95% confidence
limits. For the total series of 568 case-control pairs,
there were 94 in which the case was an aspirin user and
the control was not, and 92 in which the control was
an aspirin user and the case was not. The risk ratio estimate
was, therefore, 1.0 (95% two-sided confidence
limits 0.9 to 1.1).
To determine whether this result was affected by the
presence or absence of prior CHD among cases, we
obtained additional crude matched pair risk ratio estimates,
the first among the 312 pairs without prior
CHD, and the second among the 233 pairs with a prior
history ofCHD (23 pairs where the control had a prior
history of CHD and the case had no such history were
excluded). These estimates were 0.9 (0.6-1.3) and
1.3 (0.8-2.2), respectively.
Table 2 shows the results of the paired multiple
logistic regression analyses for the 568 pairs,
specifically, the variables significantly associated with
the within pair differences for aspirin intake. These
variables are first, use of additives (i.e. milk, cream,
non-dairy creamer) with coffee or tea, followed in
order by current cigarette smoking, history of elevated
TABLE 1. Matched Pair Risk Ratio Estimates for Coronary
Deaths Among Regular Aspirin Users Versus Non-users
Total Number of 95%
number of discordant Risk Confidence
pairs pairs ratio limits
Total series 568 186 1.0 0.7-1.3
No prior CHD 233 58 1.3 0.8-2.2
Prior CHD 312 125 0.9 0.6-1.3
Abbreviation: CHD = coronary heart disease.
TABLE 2. Variables Significantly Associated with Aspirin
Intake Among 568 Case-control Pairs
Regression Standard
Variable coefficient error P value
Constant +0.352 0.197 NS
Additives to coffee or tea -0.841 0.250 P <0.001
Current cigarettes +0.719 0.237 P = 0.002
Elevated cholesterol +1.007 0.480 P = 0.006
Diabetes mellitus -1.290 0.356 P <0.001
Relative weight +1.010 0.516 P = 0.05
Abbreviation: NS = not significant.
cholesterol, history of diabetes mellitus and relative
weight (based on the Framingham classification10).
Table 3 shows various adjusted risk ratio estimates
and their 95% confidence limits. After controlling for
other variables, the adjusted matched pair risk ratio
estimate for coronary deaths among aspirin users in
the total series was 1.4 (0.9-2.1). For those without
prior CHD, the risk ratio estimate was 1.1 (0.7-1.8),
whereas for those where the case had prior CHD, the
estimate was 1.4 (0.9-2.2).
Discussion
These findings show no evidence of association
between regular aspirin intake and coronary death
and, therefore, do not support the hypothesis of a
preventive role of aspirin in CHD. Furthermore, as
shown in table 3, the adjusted risk ratio estimates are
all greater than one, indicating that observed
differences are not even in the direction of possible
benefit.
These results from a case-control study are consistent
with those of Hammond and Garfinkel6 from
their prospective cohort study. In that survey of over
one million people, CHD rates were no lower among
people who took aspirin €œoften€ than among those
who did not.
The data from the present investigation are based
on deaths due to CHD within 24 hours after the onset
of symptoms as indicated on death certificates. In contrast,
Jick et al.4 found a protective effect of aspirin,
using cases of non-fatal MI and hospitalized controls.
Questions may be raised about the suitability of the
selection of both the patients and controls in that
study. An unknown fraction of patients hospitalized
less than 72 hours were missed because of time lags
between patient admissions and interviews. This selection
is reflected in the finding that only 4% of patients
studied died in the hospital, as compared with frequencies
of 20-30% for hospitalized acute MI in general.
TABLE 3. Adjusted Risk Ratio Estimates for Coronary Deaths
Among Regular Aspirin Users Versus Non-users
95%
Number of Risk Confidence
pairs ratio limits
Total series 568 1.4 0.9-2.1
No prior CHD 312 1.1 0.7-1.8
Prior CHD 233 1.4 0.9-3.3
Abbreviation: CHD = coronary heart disease.
36 CIRCULATION
Downloaded from http://circ.ahajournals.org/ by guest on December 26, 2012
ASPIRIN AND CORONARY DEATHS/Hennekens et al.
An additional selection occurred before admission,
since about 60% of MI patients die before getting to a
hospital.’1 These patients, therefore, are a highly
selected subgroup of the total number of patients who
experience MI. It is possible that aspirin intake is unrelated
to the onset of MI, but favors survival following
the event.
The controls were patients admitted to the hospitals
for a variety of conditions, any of which may have had
positive associations with aspirin. Aspirin may cause
some conditions, such as gastrointestinal disturbances.
The chronic persistence of other conditions,
such as arthritis or headache, may have led some controls
to take aspirin. Nevertheless, in the Jick study,
subjects with a secondary diagnosis of these conditions
were not excluded. This previous study, therefore, is
an example of the selection bias of hospital data
(Berkson’s fallacy).€2 The underlying concept is that in
the general population in which disease is incident, the
variable €œhospitalization€ has the associations with
exposure and outcome characteristic of a confounding
variable, but it is not a confounding variable because it
is an effect of exposure. It should not, therefore, be
€œcontrolled€ by restriction, that is, by limiting the
study to the hospitalized group.
The results of the randomized case-control study of
Elwood et al.5 showed a small protective, though nonsignificant,
effect of aspirin on the recurrence of MI.
In the present investigation, there is no evidence of a
protective effect of aspirin, even when restricting our
analyses to those pairs where the case had a prior
history of CHD and the control did not. This finding is
compatible with the results of a clinical investigation
by Frishman€ et al., who found no effect of aspirin on
the reduction of anginal pain during exercise.
The validity of the findings of the present study may
be affected by at least two possible sources of bias,
namely, selection and observation.
Selection Bias
A potential source of bias exists in the selection of
wives for interview, in that the wives of patients may
have been more or less available for interview than the
wives of controls. Insofar as availability is associated
with the husband’s aspirin consumption, a systematic
error in either direction could result.
It is also possible that selection bias may have occurred
in this design due to the use of death certificate
diagnoses of death from coronary disease. A previous
paper showed that in a special autopsy series of
sudden (within one hour after onset), unexpected
deaths in white males, the medicological diagnosis of
CHD was justified in 90% of the cases.’4 In a more
general group of deaths, Moriyama et al. found that
for about 80% of diagnoses relating to sudden death,
the certified diagnoses should be classified as
€œreasonable inference€ or better, although in many of
these cases information was sketchy, and that for all
deaths classified as CHD, 70-75% of the diagnoses
should be similarly classified. 15 More recently,
Moriyama et al. concluded, €œBecause there are few
diseases other than CHD which are frequent causes of
sudden, unexpected death, it is reasonable to attribute
such deaths to CHD.’6 With respect to the present investigation,
a selection error, if unbiased for aspirin
intake, would not dilute the true association. A biased
selection error would dilute the true effect only if
patients incorrectly included in the study were more
similar to controls than patients with true CHD. To
explore the possibility of bias and its direction in the
data, we compared the frequency of regular aspirin intake
of 365 death certified patients without autopsy
confirmation (19.7%), 203 patients with confirmation
(17.2%) and 568 controls (18.5%). The difference is in
the direction of diluting the true association, but it is
too small to alter the estimate of the risk ratio.
Observation Bias
A second potential source of bias is that wives may
not accurately report the aspirin intake of their
husbands. This inaccuracy may have been
systematically different for cases and controls, since
subjects had died and controls were alive. Depending
on whether wives thought that using aspirin was either
unhealthy or healthy, they may have either over- or
underestimated their husband’s aspirin consumption.
Wives of controls might have also over- or underestimated
their husband’s aspirin consumption.
This systematic error would lead to either an under- or
overestimate of the protective effect of aspirin. A
systematic inaccuracy in the opposite direction could
also be supposed. Since we collected information from
the husbands of 48 control wives about their own
aspirin intake, we were able to determine whether this
potential source of bias was present among controls.
The husbands and wives reported very similar frequencies
of regular aspirin intake (20.8% and 18.8%,
respectively). It was, of course, not possible to
evaluate this potential source of bias among cases.
In summary, these findings from retrospective data
provide no evidence of a protective role of regular
aspirin use in coronary deaths either among patients
with or without a history of CHD. To further clarify
this relationship, it will be important to compare
results from several investigations using different
study designs. In that regard, the outcome of the
ongoing Aspirin Myocardial Infarction Study
(AMIS),1€³ a cooperative, randomized clinical trial,
will be particularly informative.
Acknowledgment
We are indebted to George B. Hutchison, M.D., Harvard School
of Public Health, and John E. Davies, M.D., University of Miami
School of Medicine, for advice and help, and to Martin VanDenburgh,
B.S., for performing the computer analyses.
References
1. Weiss HJ, Aledort LM: Impaired platelet connective-tissue
reaction in man after aspirin ingestion. Lancet 2: 495, 1967
2. O’Brien JR: Effects of salicylates on human platelets. Lancet 1:
779, 1968
3. Boston Collaborative Drug Surveillance Group: Regular
37
Downloaded from http://circ.ahajournals.org/ by guest on December 26, 2012
CIRCULATION
aspirin intake and acute myocardial infarction. Br Med J 1:
440, 1974
4. Jick H, Miettinen OS: Regular aspirin use and myocardial infarction.
Br Med J 1: 1057, 1976
5. Elwood PC, Cochrane AL, Burr ML, Sweetman PM, Williams
G, Welsby E, Hughes SJ, Renton R: A randomized controlled
trial of acetylsalicylic acid in the secondary prevention of mortality
from myocardial infarction. Br Med J 1: 436, 1974
6. Hammond EC, Garfinkel L: Aspirin and coronary heart disease:
findings of a prospective study. Br Med J 2: 269, 1975
7. Hennekens CH, Drolette ME, Jesse MJ, Davies JE, Hutchison
GB: Coffee drinking and death due to coronary heart disease. N
Engl J Med 294: 633, 1976
8. Rosner B, Hennekens CH: Analytic methods in matched pair
epidemiologic studies. Int J Epidemiol, in press
9. Prentice R: Use of the logistic model in retrospective studies.
Biometrics 32: 599, 1976
10. Shurtleff D: The Framingham Study: an epidemiological investigation
of cardiovascular disease, Sec. 26. Bethesda, MD,
National Heart Institute, 1970
VOL 58, No 1, JULY 1978
11. Gordon T, Kannel WB: Premature mortality from coronary
heart disease. JAMA 215: 1617, 1971
12. Berkson J: Limitations of the application of fourfold table
analysis to hospital data. Biometrics 2: 47, 1946
13. Frishman WH, Christodoulou J, Weksler B, Smithen C, Killip
T, Scheidt S: Aspirin therapy in angina pectoris: effects on
platelet aggregation, exercise tolerance and electrocardiographic
manifestations of ischemia. Am Heart J 92: 3, 1976
14. Spain DM, Dradess VA, Mohr C: Coronary atherosclerosis as
a cause of unexpected and unexplained death. JAMA 174:
384-388, 1960
15. Moriyama IM, Dawber TR, Kannel WB: Evaluation of
diagnostic information supporting medical certification of
deaths from cardiovascular disease. Natl Cancer Inst Monogr
19: 405-419, 1966
16. Moriyama IM, Krueger DE, Stamler J: Cardiovascular diseases
in the United States. Harvard University Press, Cambridge.
1971, p 53
17. Marx JL: The AMIS trial: Can aspirin prevent heart attacks?
Science 196: 1075, 1977
38
Downloaded from http://circ.ahajournals.org/ by guest on December 26, 2012