Epidemiology, as the name suggests is the study of diseases by analysing their trigger factors and their causes and rate of spreading. Since it is a composite study of diseases and health problems and focuses on solving these problems by research and understanding the various factors. Analysts therefore utilize various methods dependent on the disease or the health issue in question and the desired outcome of the study. The term “disease’ here covers a wide range of health related issues including their causes, controls and outcomes at various stages of the disease such as disability and eventual death. There are many different kinds of epidemiological study designs that are employed by analysts and researchers to study the issues mentioned above. Each study design possesses specific advantages as well as disadvantages and selected to suit the parameters set for the study and the desired results. The various study designs prevalently used by analysts are
Descriptive Study
Ecological Study
Cross-Sectional Study
Retrospective Cohort Study
Prospective Cohort Study
Randomised Controlled Trial
Community Trial
This article briefly broaches the topic of Epidemiology and certain study designs by answering some relevant questions. We will also review in the same manner the research done by Annlia Paganini Hill and her associates Ronald K Ross and Brian E Henderson (1988) in evaluating postmenopausal Oestrogen treatment and its effect in reducing death by stroke.
Evaluation of Study Design:
What is the most appropriate study design for the following scenarios?
a) A study design employed to investigate whether taking vitamin C supplements can reduce a man’s risk of developing prostate cancer.
In a study to investigate a case as mentioned above, analysts would need to observe the participants in the study over a long period of time in order to analyse the effects or controls of the supplements. Therefore a prospective cohort study would be the most effective way to conduct this kind of a study. With this kind of study it is possible to study individuals under various groupings in their surroundings and analyse results based on their lifestyle, habits and other factors that might be considered trigger factors. The main disadvantage of such a study would be the expense factor owing to the long time-span of the study. But since it is involved in determining controls of a disease where research data may be easily available to analysts as compared to other diseases, there is a certain amount of definitive result that can be guaranteed. The nature of the disease studied also has a long incidence and latency period, therefore analysts have the advantage of time to analyse the various factors and ascertain results. There is also a better chance of studying the influence of other diseases or health factors in studying causation thereby minimizing recall errors. (Porta, 2008).
b) A study compared changes in ovarian cancer mortality rates (Standardised Mortality Ratios) and average family size over time in England and Wales and the results are shown in the graph below.
A study such as the above that involves comparison of two groups, those with the disease and those who do not (the controls), a case-control study is more effective and also less expensive. Case-control study is more effective in this case because it is a retrospective study. As Lewallen and Courtright (1998) emphasize, it is better to study a population who are known to be at threat of contracting a certain disease rather than the contrary where the researchers have to wait for symptoms to develop, thereby being unable to set parameters for the study (para. 3). Since this study is based on a specific aspect as a cause or control, a study such as this will also help in yielding results in a short span of time. A study design such as this also gives options of studying a case based on several exposures, which is applicable in this kind of communal research study. In a case control study design however it is important to guard against information bias.
Post Menopausal Oestrogen treatment and stroke: a prospective study
a) What is the research question for this study?
The research question in the study conducted by Pagnani-Hill et all (1988) was focussed on determining if Oestrogen replacement treatments administered to post menopausal women reduced the instance of death by stroke.
b) What is the outcome factor and how was this measured?
The outcome factor was studied based on the responses given by the participants in the initial questionnaire and also on periodical monitoring of the respondents over a period of 4 to 5 years. The results obtained in the study showed that there was a higher risk of death by stroke for post-menopausal women who had not been undergoing any kind of oestrogen replacement treatment. While the instance of death by stroke on Oestrogen users was nominal there was almost a 14% evidence of death by stroke on those who had not undergone any kind of treatment. These results were based on a study on 4952 women who had undergone treatment and 3845 who, had not. (Pagnini-Hill et al. 1988. P.520). The authors also studied the low incidence of stroke on people who exercised more often among those who had not had any Oestrogen replacement treatments. The effect of smoking on cardio-vascular disease and being a trigger for stroke-induced death was apparent but was not conclusively studied as, it was not “statistically apparent” (Pagnini-Hill et al, 1988). P.520). These, results were obtained by biennial mailings to the facility, monitoring of obituary columns and also having access to hospital visit and discharge records.
c) What is/are study factor(s) and how was this/were they measured?
The study factor as stated above was to ascertain the effects of Oestrogen replacement treatments in post-menopausal women in reducing the risk of death by stroke.
It was measured by conducting a survey on a selection of women in a retirement community based in California. The participants were selected based on a questionnaire that was mailed to the facility. The study was based on previous knowledge of stroke and cardiovascular disease being the highest causes for death in the United States. The entire study was conducted based on respondents of an initial questionnaire which gave the analysts detailed information of their health status, history of diseases and also their life style choices and personal habits. The research started in June 1981 and questionnaires were mailed out subsequently in the years 1982 and 1985 to include new inmates to the facility. A death, of the respondents within the study time was monitored through the obituary pages and also by Biennial electronic mailings to the facility. This email also helped monitor the instance of any new health conditions in the participants. There appears to have been a meticulous follow up because only 13 out of the 13986 respondents missed in the follow up list.
d) Does the main finding presented in this paper suggest any significant association between the study factor and the outcome factor? What did the result suggest?
As mentioned above, evaluating the reduced number of deaths by stroke suggests that the use of Oestrogen replacement treatments may have had a hand in it. The research also proved the theory, that Oestrogen replacement treatments to treat hypertension in post- menopausal women leads to stroke, wrong. Their findings showed that respondents who were given similar treatments for hypertension exhibited less risk of death by stroke. It was also found that those with higher body mass index and thereby with healthy oestrogen levels for their age also were protected against death by stroke. From the above we can see that the study factor had close relation to the outcome factor.
e) Could information bias have affected the results in any way? Why (or why not)? Did the authors discuss this possibility adequately?
We do see possibility of information bias in the study but we also see that the way the study was handled eliminated most of the risks of bias.
The main investigation was also based on previously obtained substantiated evidence through similar research on cardio-vascular disease and its relation to Oestrogen treatments. Research material was also available from prior case control and cohort studies on the effects of Oestrogen replacement therapy in preventing death due to cardio-vascular disease. Therefore based on that information the analysts proceeded to determine its effect on other heart –related diseases, namely stroke. There is a possibility that the analysts were already looking for a certain outcome to their analysis. However they expanded the study group to eliminate any personal habits or lifestyle preferences, which may interfere with the supposed positive effects of Oestrogen replacement treatments. The authors discuss a similar test conducted in the same facility in the years 1964-73 in the same facility provided contrasting results. The previous research found no relation between Oestrogen and stroke (Hill et al. 1988, p. 521).
The population in the chosen facility also consisted of well-educated members of society who volunteered information based on a questionnaire given to them. Therefore we can assume that they understood the importance of such research for future generations. Pagnini-Hill et al (1988) also mention that they were careful not to rely on only pharmaceutical records to determine dosage and usage. Careful study of the kind of supplements used was also done and dosage was supposed based on the colour of the pills, these were further substantiated by pharmaceutical records and by the respondents themselves. They were also careful to eliminate respondents who were not clear about the usage of Oestrogen supplements or having undergone any kind of Oestrogen replacement. There we can see that there was no ambiguity in that respect.
f) Name another common type of bias that may occur in this type of study design. Is it likely to have influenced the results here? Explain your reasoning briefly.
Another kind of bias in such kind of study would likely be medical surveillance bias. In the case of the research in question the research started in June 1981 and questionnaires were mailed out subsequently in the years 1982 and 1985 to include new inmates to the facility. We have seen above that death and hospital visits were monitored periodically. We cannot however ignore the fact that some of the respondents could have been on regular check ups for other symptoms and ailments and this in turn could have affected the diagnosis of stroke related symptoms. This can come in the way of accurate analysis of the study.
g) What are the potential confounders identified by the authors? Have they influenced the association between the study factor and the outcome factor? Explain your view briefly
When studying the mortality rate the study also took into note the relation between hypertension and high instance of stroke. We already have clinical evidence of the effect of Oestrogen replacement treatments to combat hypertension in pre and post-menopausal women. The statistically lower rate of death by stroke in women with hypertension but under oestrogen treatments also showed evidence to prove this research. Hill et al (1988) also evaluated the effect on the basis of the dosage and years of usage of Oestrogen supplements and found that recent users, thus classified based on the initial questionnaire showed a relatively low risk of stroke that those who had been administered some kind of Oestrogen treatment for a few years before the questionnaire. The median years here were set at a minimum of 8 years to ascertain the effectiveness of the treatment. (p. 521). As we can see that the analysis had covered most grounds needed to substantiate their findings eliminating potential compounders in their analysis.
h) With regard to Table 1, do you agree with the authors’ comments on the effects of age on the oestrogen-stroke mortality association? Explain your view briefly.
We do have prior scientific evidence to the fact that the Oestrogen levels of women decline to almost 60% during pre and post-menopausal stage for women. We also know that this decline in Oestrogen levels leads to hypertension which are symptomatic to stroke in post-menopausal women. Therefore the effects of age on the oestrogen-stroke mortality association hold credence.
i) In Table III, the authors analysed the mortality rates from stroke by different duration and dose of oestrogen. Did the results indicate significant differences in mortality rates? Please interpret the table briefly (Hints: compare with the major finding and check the 95% CI column)
In Table III Pagnini-Hill et al (1988) the mortality rate calculation was based on women who had been administered all kinds of Oestrogen replacement treatment. However dosage could only be calculated for those who had taken oral treatment. In this analysis, the researchers found that when they calculated based on the duration of use between two groups of women, there were no significant results. However when they narrowed it down to dosage they found that those who were recent users of Oestrogen also seemed to be taking higher dosage of conjugated equine Oestrogen (p.520) and therefore showed greater protection from death by stroke. Therefore the duration was complemented by higher dosage. These results thus calculated and tabulated showed that recent users who started dosage at the time of the initial questionnaire showed a risk level of 0-21 as against the risk level of 0-67 of those who had been using for a much longer period of time. (p.521) Therefore analysis of mortality rates from stroke by different duration and dose of Oestrogen holds credibility.
j) In discussing their findings, the authors have ignored a major threat to the validity of their estimates of the ‘protective’ effect of oestrogens. What alternative study design could overcome this problem better and why?
We see evidence to the fact that Pagnini-Hill et al (1988) have focussed too much on the protective effect of Oestrogen and thereby may have ignored the harmful effects of it. We know that the use of Oestrogen replacement as a means to combat many health factors in ageing women is a topic that is still under research. The prolonged usage of such treatments and its ill effects has also been studied and are no longer recommended as a means of treatment. Oestrogen replacement treatments have also been found to increase the risk of cancer although no conclusive evidence has been found yet. While the research by Pagnini-Hill et al (1988) in no way recommends any particular treatment as a preventive method we have to be careful to evaluate certain research findings by looking at it from a broader spectrum and just as a means of identifying or elimination of specific factors which can then be applied to a larger research. Oestrogen replacement treatments may be the answer to reducing death by stroke but it may be necessary to analyse the need for such treatments based on symptoms and also by evaluating other risk factors.
It may be possible to eliminate this by combining the study with a community trial design and widen the prospects. With the knowledge obtained from this study it may be possible to substantiate with clinical trials while incorporating all the variants included in the prior prospective study and thereby use it to filter the findings and arrive at a broader consensus.
k) To whom may the results be generalised?
The results from the research cannot also be taken as conclusive and applicable over a broad basis as it targeted only, a specific stratum, of society, white, affluent and well educated (Pagnini-Hill et al, 1988. p. 519) with a median age of 73. We can therefore see that this research is applicable to the white population with similar history and lifestyle influences and experiences. It is also generalized amongst those who are undergoing regular checkups due to hypertension and other symptoms and thereby diagnosable.
References
World Health Organization. Epidemiology. 2012. May 03, 2012.
http://www.who.int/topics/epidemiology/en/
Porta M (ed) (2008). A dictionary of epidemiology. 5th. edition. New York: Oxford
University Press.
Susan Lewallen, Paul Courtright (1998). Epidemiology in Practice: Case-Control Studies.
Journal of Community Eye Health. May 03, 2012.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1706071/
Dr Trevor A. Howlett. Hormone Replacement Therapy / Oestrogen Replacement
Benefits and Risks. Leicester Royal Infirmary. January 2003. May 03, 2012.
http://www.lnrmed.org.uk/Endocrinology/Endo_PDF/HRT%20Information%20Sheet.pdf
Lynda Lisabeth, Cheryl Bushnell (2009). Stroke risk in women: the role of menopause and
hormone therapy. The Lancet Neurology, Volume 11, Issue 1, Pages 82-91
Rogerio A. Lobo (2007). Treatment of the Postmenopausal Woman: Basic And Clinical
Aspects.(3). Surendra Kumar
