Estimation of menstrual blood loss volume based on menstrual diary and laboratory data
1 Medical Affairs Support, Jenapharm GmbH und Co KG 07743, Jena, Germany
2 Center for Clinical Studies, Universitätsklinikum Jena, 07747 Jena, Germany
3 Institute of Stochastics, University of Jena, 07737, Jena, Germany
4 Global Clinical Statistics, Bayer Pharma AG, 13342, Berlin, Germany
5 Institute for Medical Epidemiology, Biometrics and Computer Science, Universitätsklinikum Halle, 06112, Halle, Germany
6 Global Medical Affairs, Bayer Pharma AG, 13342, Berlin, Germany
7 Department of OBGYN, Saarland University, 66421, Homburg/Saar, Germany
BMC Women's Health 2012, 12:24 doi:10.1186/1472-6874-12-24Published: 20 August 2012
Abnormal uterine bleeding is often investigated in clinical studies and critical to identify during gynecological consultation. The current standard for quantification of menstrual blood loss is the alkaline-hematin-method. However, this method is expensive and inconvenient for patients. Bleeding diaries, although widely used, provide only qualitative information on menstrual blood loss. Other methods have been developed, but still do not provide reliable quantitative data.
We estimated blood loss volume using data from two clinical studies in women suffering abnormal menstrual bleeding. These estimations were derived from mixed linear models based on diary data, hematological parameters and age. To validate the models, we applied our results to data from a third study with a similar patient population.
The resulting best fitting model uses diary entries on bleeding intensity at a particular day, information on occurrence and frequency of single bleeding intensities in defined time windows, hemoglobin and ferritin values and age of the patient all as predictors of menstrual blood loss volume. Sensitivity and specificity for the diagnosis of excessive bleeding were 87% and 70%, respectively. Our model-based estimates reflect the subjective assessment by physicians and patients in the same way as the measured values do.
When applying the model to an independent study, we found a correlation of 0.73 between estimated and measured values for the blood loss in a single day. Further models with reduced number of parameters (simplified for easier practical use) still showed correlation values between 0.69 and 0.73.
We present a method for estimating menstrual blood loss volume in women suffering from prolonged or excessive menstrual bleeding. Our statistical model includes entries from bleeding diaries, laboratory parameters and age and produces results which correlate well with data derived by the alkaline-hematin-method. Therefore, this model may be used to estimate menstrual blood loss volume in both routine gynecological counseling and clinical studies.