Solutions Database
(Sept. 28, 2008; rev. Jan. 5, 2009)
The Access database that was originally placed on this site on September, 28, 2008, and revised on October 1, 2008, has been revised to allow one also to conduct an analysis that adjusts for an irreducible minimum rate, as discussed on the Irreducible Minimum sub-page of the Scanlan’s Rule page. The material previously describing the database that did not allow for such adjustment is now maintained, slightly modified, under Section A. The instructions regarding the adjustment for absolute minimums now set out under Section B. The last paragraph of Section A applies to the entire database.
A. Standard Analysis
The Access database made accessible below allows one to appraise the difference between the rates at which two groups experience some outcome in terms of the difference between means (in terms of percentage of a standard deviation) of underlying normal distributions of continuously scaled risks of experiencing the outcome (as is done in the tables in references B13-B18, D43, D45, D46, D46a, D48, D52, D53, D55, D56, D58, D60, D61 of the Measuring Health Disparities page of jpscanlan.com (MHD) and as discussed generally on the Solutions page that page (Solutions Page).
The database enables one to determine, for example, that where the white and black coronary angiogram rates are 0.86% and 0.43% at one point in time and 2.28% and the 1.61% at a later point in time, the EES is .25 in the former case and .14 in the latter case (as discussed in D48 (Comment on Escarce). It also enable one to determine that (as in the illustration in the database itself) when the white and black selection rates are 14% and 6% in one case and 28% and 12% in another, the EES is .47 in the former case and .60 in the latter case.
In the database, the query styled “1 Focus on Data” is based on Model Data Table, which currently serves as a sort of placeholder. The figures AG and DG are adverse outcome rates for the advantaged and disadvantaged groups. The Model Data Table should be replaced with a table or query with the adverse outcome rates of interest. The database allows analysis by up to four categorization levels. Level is determined by insertion of field from an underlying table/query into fields A1 to A4 of Query 1 Focus on Data. AGR is the advantaged group’s rate as a percent and DGR is the disadvantaged group’s rates as a percent. For example, if in the underlying table/query (as in the Model Data Table), the analysis is only by year, the year field is “Year,” the advantaged group’s rate field is “AR,” and the disadvantaged group’s rate field is “DR,” the fields in query 1 would look like this.
A1:”E”
A2:”E”
A3:”E”
A4: Year
AGR:AR
DGR: DR.
Section B discusses the role of the IM field in Query 1 Focus on Data. For an analysis that does not consider the role of the absolute minimum, the field can be left out of the query.
Anyone familiar with Access will know how to make the necessary adjustments. The query styled “9d Results” provides the results. Query 9d Results will yield the same figures regardless of whether Query 1 Focus on Data includes an IM field or the value in the IM field.
The database is copyrighted. But until otherwise notified, anyone affiliated with a not-for-profit organization or agency of any government may download and use the database for activities related to the activities of such organization or agency (not including compensated outside consulting work) upon completion of the form below. Others interested in use of the database should contact James P. Scanlan at jps@jpscanlan.com.
B. Adjusted Analysis
The database has been augmented to conduct the same analysis as described in Section A with an adjustment for irreducible minimums. The adjusted analysis merely requires that the data table/query include a value for the irreducible minimum, styled “IM,” and that such field be brought into the query 1 Focus on Data (not Query 1 Focus on Data Adjusted. Query 9d Results Adjusted provides the adjusted results. Query 9e Compare Adjusted and Unadjusted Results compares the adjusted and unadjusted results.
(If Query 1 Focus on Data contains an IM field with a zero value, the Query 9d Results and Query 9d Results Adjusted will yield equal figures. But if Query 1 Focus on Data does not contain an IM field, neither Query 9d Results Adjusted nor Query 9e will run.)
Note: The deriving of an EES figure involves the matching of an advantaged group’s actual rate with published data on normal distributions. Since published values will not invariably match an actual rate, the program identifies the published rate that is the closest match to the actual rate, selecting the higher published rate where the actual rate is equidistant from two published rates. In the Solutions Database as originally published on this site, for purposes of matching the disadvantaged group’s rate with a published figure and thereby identifying the EES figure associated with a combination of advantaged and disadvantaged group rates, the higher published rate was also chosen, which in effect rounded the EES up to the higher value (of the available one-hundredths of a standard deviation). But, whereas it was essential to choose either the higher or lower value for purposes of matching of the advantaged group’s rate with a published figure, it was not necessary to make such a choice in the case of the match for the disadvantaged group’s rate. Therefore in the revised database, in cases where the disadvantaged group’s rate fall equidistant from two published values, the EES is simply carried out to the third decimal place. That is, instead of yielding an EES of either .38 or .39 (which would have been .39 if rounded up as in the original program), the database yields an EES of .385. As I noted in initially describing this approach with results carried to the third decimal place (in D23 of MHD), the use of the third decimal place should not be interpreted as suggesting a high level of precision in the approach. I nevertheless have modified the approach in the manner described simply because it is preferable not to arbitrarily round anything up or down when doing so is unnecessary.
|