| title: | Long-term effects of hormone replacement therapy on health, quality of life and use of health services among Estonian postmenopausal women: intention-to-treat analysis and the analysis of the causal effect of adherence in a randomized trial |
|---|---|
| reg no: | ETF6570 |
| project type: | Estonian Science Foundation research grant |
| subject: |
3.4. Health Sciences |
| status: | accepted |
| institution: | TU Faculty of Medicine |
| head of project: | Krista Fischer |
| duration: | 01.01.2006 - 31.12.2007 |
| description: | Main purpose of the project is to analyze long-term causal effects of Hormone Repllacement Therapy (HT) in the Estonian randomized study “Menopause and woman’s health” and to develop suitable statistical methodology to analyze such trials with complicated design. The Estonian Postmenopausal Hormone Therapy Trial was carried out in 1999 to 2004 as a randomized preventive trial with 1823 participants aged 50 to 65 years at the age of sampling. The trial consisted of four arms (two arms both in the blind and in the non-blind group) and compared continuous combined oestrogen-progestin therapy (HT) to placebo or non-treatment, in order to study the effects of HT on the sum of major ischaemic heart disease events and of stroke, the sum of major fractures, mortality and incidence of breast cancer and other cancers, and deaths from all causes, on well-being and quality of life; and on the use of health services. The unique design of the study was motivated by the purpose to assess the placebo effect and trial effect on health outcomes, recruitment and adherence. The trial underwent pre-term stopping due to the negative results of large international studies. At the same time the participating women will be followed up to year 2012 for survival and important health outcomes. At the moment, only short-term effects of HT on health and health service utilization have been analyzed. The data on quality of life, as well as large part of other health data that was collected by yearly questionnaires, has not yet been analyzed. The data on various health complaints (bleeding, incontinence) and on certain medical procedures (e.g. hysterectomies) will be studied by doctoral student Fred.Kirss. P.Veerus and K. Fischer will be analyzing the direct and indirect effects of HT. It is known that the HT-related health risks will be notable only in long term, requiring follow-up for several years. Also it is still unknown, how long will last the protective effect of HR on bone fractures and colorectal cancer. There is no information available on quality of life and health service utilization by woman after stopping the Hormone Therapy. Therefore it is of great importance to continue linking the trial data with the Estonian Cancer Registry and Estonian Health Insurance database and to conduct the relevant data analysis. The data analysis in this study is complicated by the fact that the amount of HT actually received by the women differed a lot across woman. One of the reason was early stopping of the trial, another reason was that the women were allowed to stop the trial treatment at any time during the trial (for ethical concerns). Therefore a simple comparison of the groups would not give adequate information on the actual causal effect of HT. The data on the actually received treatment is available, but a simple correlation analysis would not estimate the effect of interest. The reason is that although health is possibly affected by the amount of HT received, also the amount of HT can be affected by health status of the woman. K. Fischer has previously studied methods to assess causal effect of non-compliance in randomized trials, but mainly in settings where the main outcome variable is continuous. In the present study the outcome data is either binary, or in the form of censored failure time. Therefore there is a need to evaluate the existing biostatistical methods and to develop new methodology to account for the specific type of the data. It is also of methodological interest to assess the placebo-effect by comparing blind and open arms of the study. The planned methodological research will be of wide international interest, since up to now there does not exist a unified and generally acknowledged methodology to analyze the effects of adherence in clinical trials and many researchers are still working on it. The statistical methodology for analyzing the effects of adherence will be the main topic for the doctoral student Heti Pisarev (supervised by Krista Fischer). |
| project group | ||||
|---|---|---|---|---|
| no | name | institution | position | |
| 1. | Krista Fischer | Tartu Ülikool | ||
| 2. | Heti Pisarev | |||
| 3. | Piret Veerus | |||