Clinical Research Team
Clinicians / Epidemiologists
We cannot tell whether or not someone has Alzheimer's disease by a simple blood test. Instead, a trained clinician must apply a set of criteria having to do with the clinical history, neurological examination, laboratory studies, and neuropsychological testing. Clinicians can also identify special features that subdivide AD patients into important groups for research purposes. Clinicians manage the needs of the patients during the research study. Clinicians often generate the ideas about what to study based upon their experience in taking care of AD patients. The ability to formulate these ideas in a way that can be tested depends upon the science of epidemiology: “the study of the distribution and determinants of health-related states or events in specified populations.”
Epidemiological studies involve surveillance, observation, hypothesis testing, and experiment. Most clinician researchers have training in epidemiology, but for large and complex studies they work with Ph.D. epidemiologists to generate new ways to answer increasingly complex questions. Working together, clinicians and epidemiologists will decide in advance what data to collect in order to answer a particular question, how many subjects are needed, how to ensure safety and confidentiality, how much random variability will exist between subjects, and what instructions will be given to the data team regarding coding and arrangement of the electronic data.
In order to do research, everything has to be measured and turned into numbers. For example, IQ gives a numerical estimate of intelligence, something that can change over time in AD. Through elegant research, we have learned to measure language abilities, visual skills, motor skills, judgment, and problem-solving abilities in people with and without AD. Psychometricians have been trained to interact directly with patients and to score their responses on all of these tests. They are also trained to use published data (norms) in order to learn whether or not a particular patient is or is not impaired on a particular test response in comparison to people of the same age and education. A typical battery of tests for an AD patient might take 4 hours to administer, and 8 hours to score and interpret.
Collecting information from human beings is fraught with errors. For example, a patient may misunderstand a question and give a wrong answer, or a caregiver may forget the dose of a medication, or a physician may forget to write something down, or write the wrong thing. Therefore, data personnel are assigned to create data collection forms that minimize errors, to check the forms after they are filled out, and to correct errors. This invisible process is costly and time consuming, but ultimately determines the value of the research results.
Clinical data must be maintained in patient charts by law, but the charts are not useful for research because they are not systematic and because someone would have to read and re-read every chart each time there was a research question. When patients give written permission for the research use of their data, data personnel can construct an electronic database using computers and then enter the data written on specially designed data forms. Data entry can also create new errors (for example, typing in a 2 when it should have been a 20), so the data staff has to design procedures to check the data as they are being entered and, periodically, after groups of data are entered. If the data are not correctly entered, the results of the analysis will be incorrect.
When all of the data collection is complete and it is time to answer a research question, a specially trained database manager is needed to design queries that will give back the needed data in the needed format. For example, the data might be put into tables containing each variable for each subject with a particular feature, and displayed as frequency tables, means and medians, or diagrams.
After the data team organizes all of the data for a particular study, we need to use it to determine the answer to the research question. For example, does the drug Aricept benefit patients who are taking it compared to patients who are not taking any therapy? A biostatistician is trained specifically to apply statistics to biological or medical data. This involves close collaboration with the clinicians and epidemiologists, because the biostatistician does not have personal experience with the data points that are going to be used in the analysis, or with the disease being studied. Working together, the team agrees on an analysis plan. The biostatistician takes the formatted data and imports it into statistical language so that it can be mathematically manipulated. Often he/she must apply several formulas and tests to the data in order to choose the correct procedures for analysis. The analysis will generate pages of statistical comparisons with significance levels (tells whether the finding occurred by plan or by chance) and other mathematical characteristics. The biostatistician, clinicians and epidemiologists will then draw conclusions together from the analysis. In the end, even the simplest research question can lead to new information that benefits patients, but the process is always complex. In order to publish the results of a research study, the team has to convince the scientific community that the question was important, the study was well-designed and able to answer the question, the data procedures were excellent so the data can be relied upon, the statistics were done correctly, and the conclusion is important to human beings. No advance in medicine can benefit patients without clinical research. For example, even if we learn that a disease is caused by a deficiency of a certain chemical, we cannot prove the theory until we design a study in which replacing the chemical (as a pill, for example) reverses the disease.
Ongoing Areas of Research in the BCM ADMDC
- Heterogeneity of AD (how and why the disease affects different patients differently).
- Standardization of testing measures, such as the development of normative scores for the Alzheimer's Disease Assessment Scale-cognitive subscale.
- Development of new assessment tools: Baylor instruments developed for use with patients with profound dementia, for the assessment of neuropsychological function, and the development of tools to measure self-awareness of deficits in AD.
- Lateralization (how the disease is affected by patient handedness and by the predominance of problems on the right or left side of the brain).
- Predicting and modifying progression of disease (studying individuals with age related cognitive changes, several forms of Mild Cognitive Impairment, and Alzheimer's disease to predict and modify progression rates). We are examining how IQ and education affect the progression of AD, as well as how to optimize treatment regimens for the best outcomes.
- Predicting survival in AD, when controlling for other medical illnesses.
- Understanding the factors that lead to development of agitation in AD.
- Understanding factors related to elder abuse and neglect, and frailty in AD patients.