Difference between revisions of "Benefits and Risks to Secondary Use of Clinical Data from Electronic Medical Records"
m (Improve readability with some subheadings) |
|||
| Line 1: | Line 1: | ||
The huge increase in coded health data generated by electronic medical records has an enormous potential to increase our ability to do clinical research. Compared to traditional research methods, there are many potential benefits and detriments to secondary use of clinical data. | The huge increase in coded health data generated by electronic medical records has an enormous potential to increase our ability to do clinical research. Compared to traditional research methods, there are many potential benefits and detriments to secondary use of clinical data. | ||
| − | + | ==Benefits== | |
Benefits include increased simultaneous entry of clinical and research data, increasing the number of subjects and the amount of data collected, ultimately increasing the generalizability of research. In traditional clinical research, data is collected on paper forms separately from clinical documentation, requiring double entry of data. This adds cost and may limit accuracy if data collection is performed by non-clinical personnel. The number of subjects enrolled and the amount of data collected is limited by cost and time. If data can be collected simultaneously with clinical documentation, costs can be reduced and data may be more accurate when entered immediately by the clinician caring for the patient. More patients can participate in studies and research findings become more generalizable to a diverse population. | Benefits include increased simultaneous entry of clinical and research data, increasing the number of subjects and the amount of data collected, ultimately increasing the generalizability of research. In traditional clinical research, data is collected on paper forms separately from clinical documentation, requiring double entry of data. This adds cost and may limit accuracy if data collection is performed by non-clinical personnel. The number of subjects enrolled and the amount of data collected is limited by cost and time. If data can be collected simultaneously with clinical documentation, costs can be reduced and data may be more accurate when entered immediately by the clinician caring for the patient. More patients can participate in studies and research findings become more generalizable to a diverse population. | ||
| − | + | ==Limitations== | |
Limitations of secondary use of clinical data are that of retrospective research, and which is inherently subject to many sources of bias and error. With prospective study design, a research question is posed and the study designed to be able to accurately measure and analyze the data required to answer the question. Inconsistencies of medical terminology are a recognized challenge to research validity (misclassification bias), and each research plan requires careful attention to the definitions necessary to answer the question. The conditions to be studied, the treatments rendered, and the outcomes to measure are carefully defined. Templates are designed to enhance accuracy and minimize missing data. Potential sources of bias and confounding are considered and managed. | Limitations of secondary use of clinical data are that of retrospective research, and which is inherently subject to many sources of bias and error. With prospective study design, a research question is posed and the study designed to be able to accurately measure and analyze the data required to answer the question. Inconsistencies of medical terminology are a recognized challenge to research validity (misclassification bias), and each research plan requires careful attention to the definitions necessary to answer the question. The conditions to be studied, the treatments rendered, and the outcomes to measure are carefully defined. Templates are designed to enhance accuracy and minimize missing data. Potential sources of bias and confounding are considered and managed. | ||
| − | + | ==Case study== | |
The studies of estrogen therapy after menopause are an excellent example of bias and erroneous findings in retrospective studies. Briefly, before the Women’s Health Initiative (WHI) results were published 2002 (Rossouw JE, JAMA, 2002) (a randomized trial of estrogen therapy in menopausal women), there were numerous retrospective studies indicating that women who used menopausal hormone therapy had a 50% reduction in death from heart disease. Women were encouraged to take estrogen by clinicians as a strategy for reducing heart disease. A question posed by many researchers was “Did this finding occur because 1) estrogen improves cardiovascular function or 2) healthier women choose estrogen more often than less healthy women?” (selection bias). The randomized trial found that estrogen did NOT confer a cardiac benefit, and now estrogen is NOT recommended as a strategy for reducing heart disease. This story emphasizes the magnitude and impact of potential errors that may result from retrospective research. Note--this illustration is simplified, and does not represent the complexities of an individual woman’s benefit or risk of taking hormone therapy.) | The studies of estrogen therapy after menopause are an excellent example of bias and erroneous findings in retrospective studies. Briefly, before the Women’s Health Initiative (WHI) results were published 2002 (Rossouw JE, JAMA, 2002) (a randomized trial of estrogen therapy in menopausal women), there were numerous retrospective studies indicating that women who used menopausal hormone therapy had a 50% reduction in death from heart disease. Women were encouraged to take estrogen by clinicians as a strategy for reducing heart disease. A question posed by many researchers was “Did this finding occur because 1) estrogen improves cardiovascular function or 2) healthier women choose estrogen more often than less healthy women?” (selection bias). The randomized trial found that estrogen did NOT confer a cardiac benefit, and now estrogen is NOT recommended as a strategy for reducing heart disease. This story emphasizes the magnitude and impact of potential errors that may result from retrospective research. Note--this illustration is simplified, and does not represent the complexities of an individual woman’s benefit or risk of taking hormone therapy.) | ||
Revision as of 09:52, 16 September 2007
The huge increase in coded health data generated by electronic medical records has an enormous potential to increase our ability to do clinical research. Compared to traditional research methods, there are many potential benefits and detriments to secondary use of clinical data.
Benefits
Benefits include increased simultaneous entry of clinical and research data, increasing the number of subjects and the amount of data collected, ultimately increasing the generalizability of research. In traditional clinical research, data is collected on paper forms separately from clinical documentation, requiring double entry of data. This adds cost and may limit accuracy if data collection is performed by non-clinical personnel. The number of subjects enrolled and the amount of data collected is limited by cost and time. If data can be collected simultaneously with clinical documentation, costs can be reduced and data may be more accurate when entered immediately by the clinician caring for the patient. More patients can participate in studies and research findings become more generalizable to a diverse population.
Limitations
Limitations of secondary use of clinical data are that of retrospective research, and which is inherently subject to many sources of bias and error. With prospective study design, a research question is posed and the study designed to be able to accurately measure and analyze the data required to answer the question. Inconsistencies of medical terminology are a recognized challenge to research validity (misclassification bias), and each research plan requires careful attention to the definitions necessary to answer the question. The conditions to be studied, the treatments rendered, and the outcomes to measure are carefully defined. Templates are designed to enhance accuracy and minimize missing data. Potential sources of bias and confounding are considered and managed.
Case study
The studies of estrogen therapy after menopause are an excellent example of bias and erroneous findings in retrospective studies. Briefly, before the Women’s Health Initiative (WHI) results were published 2002 (Rossouw JE, JAMA, 2002) (a randomized trial of estrogen therapy in menopausal women), there were numerous retrospective studies indicating that women who used menopausal hormone therapy had a 50% reduction in death from heart disease. Women were encouraged to take estrogen by clinicians as a strategy for reducing heart disease. A question posed by many researchers was “Did this finding occur because 1) estrogen improves cardiovascular function or 2) healthier women choose estrogen more often than less healthy women?” (selection bias). The randomized trial found that estrogen did NOT confer a cardiac benefit, and now estrogen is NOT recommended as a strategy for reducing heart disease. This story emphasizes the magnitude and impact of potential errors that may result from retrospective research. Note--this illustration is simplified, and does not represent the complexities of an individual woman’s benefit or risk of taking hormone therapy.)
Understanding the potential limitations of secondary use of data will facilitate changes to mitigate the risks. Already there is much emphasis on improving the clarity of medical terminology. Research questions can be “designed in” to EMR’s to accurately capture the data needed to answer the question, using templates, drop-down menus with definitions provided (research decision support). Currently the risks of secondary use of data are large, but it is within the realm of EMR design to mitigate these risks. When such design changes have been implemented, EMR’s will be able to provide a rich source of data for analysis for clinical research to enhance human health.
