Population analysis

Top Previous Next 
To access the population analysis function:
Each time you perform serum level analysis, the calculated pk parameters are saved. This data is a virtual gold mine of information about your patient population. With this tool you can derive a model better fitted to your patients.
Howto First, select a model to analyze from the drop down list, then select a date range. The remainder of the criteria are optional. You may use these optional criteria to further narrow down the population analysis: Area This keys on the "location" field in your patient data, which may not be applicable to your practice situation. For example, if all your ICU beds start with ICU, then select "Specific" and enter ICU in the input box. This will limit the analysis to all patients whose locations *start* with ICU. The important point to remember is, the program matches your criteria with the start of the location field. Age You may choose to select an age range to analyze. Gender You may choose to select a specific gender to analyze. Creatinine clearance You may choose to select a creatinine clearance range to analyze.
Analyze Click this button to analyze the data and print a summary report. Information included in this report are:
The volume of distribution analysis calculates the population Mean Vd, standard deviation, range, and the following descriptive statistics: Skewness Skewness is a measure of the asymmetry of the data around the sample mean. If skewness is negative, the data are spread out more to the left of the mean than to the right. If skewness is positive, the data are spread out more to the right. The skewness of the normal distribution (or any perfectly symmetric distribution) is zero. As a general rule of thumb:
Kurtosis Kurtosis is a measure of how outlierprone a distribution is. The kurtosis of the normal distribution is 0. Distributions that are more outlierprone than the normal distribution have kurtosis greater than 0. A higher kurtosis means more of the variance is the result of infrequent extreme deviation, i.e., extreme values are more likely.
Least squares linear regression is used to analyze CrCl vs Kel. The yintercept of this regression line is the Nonrenal K, the slope of the line is the Renal K. The following descriptive statistics are also calculated: Standard error of estimate The standard error of estimate is analogous to standard deviation, indicating data spread. It is an estimate of the accuracy of the regression equation. Coefficient of determination (Rsquared) The coefficient of determination quantifies the overall quality of the fit of the regression line. The closer the value is to 1, the better the linear fit. A correlation coefficient value of zero would indicate that the data are randomly scattered and have no pattern or correlation in relation to the regression line model. Regression plot A plot of this regression analysis displays the computed regression line and all data points.
Export Click this button to save the data to a CSV file which can be read by a spreadsheet program such as Excel. This allows you to further manipulate and analyze the data within a spreadsheet.
Error messages Yintercept less than zero is displayed when the regression line intercepts the Y axis below zero.
Regression analysis yielded a negative slope is displayed when the regression line is inverted.
The regression line is a plot of X vs Y where X is creatinine clearance (the independent variable) and Y is the elimination rate (the dependent variable). The Y intercept is the Nonrenal Kel, the slope of the regression line is the Renal Kel for the following equation: Kel = NonRenalK + [CrCl x RenalK] It would be impossible (of course) for there to be a negative NonRenal Kel, that would mean you are somehow putting drug back into the system instead of excreting it. Likewise, a negative slope would indicate an inverse relationship with creatinine clearance.
Both of these "errors" are caused by a widely scattered data set for which a regression line cannot be accurately calculated. This would be expected with a drug like Vancomycin which has a wide variability in pk parameters. Usually it takes a higher N in your data set before the data becomes analyzable. Keep saving your consults and eventually you will acquire a data set that will give usable results.
