Identifying individuals with virologic failure after initiating effective antiretroviral therapy: The surprising value of mean corpuscular hemoglobin in a cross-sectional study

Table 2 Results of logistic model after screening for variables by the random forest approach***

	Beta Coefficient	Odds Ratio	Odds Ratio 95% Confidence Interval	p-value
Intercept	7.27	**		<0.0001
MCH (pg/cell)	-0.19	0.83	0.77, 0.89	<0.0001
Change in MCH (per pg/cell)*	-0.22	0.81	0.74, 0.88	<0.0001
CD4 (per 100 cells/mm³)	-0.32	0.73	0.65, 0.81	<0.0001
Change in Percent CD4 (per percent)*	-0.05	0.95	0.91, 0.99	0.008

* Change is relative to the pre-HAART value for an individual, thus a positive value for change in MCH indicates an increase in MCH for an individual from their pre-HAART value.
** The intercept has no OR interpretation.
*** To determine the predicted probability of having an HIV RNA > 500 copies/ml for an individual, take the value of each variable and multiply it by the corresponding Beta Coefficient. Take the sum of the resulting values and add the Intercept Beta Coefficient. This is the log(odds) that an individual has an HIV RNA value above 500 copies/ml. The predicted probability is then 1/(1+e(-log(odds)).