Should symptom frequency be factored into scalar measures of alcohol use disorder severity?

To evaluate whether weighting counts of alcohol use disorder (AUD) criteria or symptoms by their frequency of occurrence improves their association with correlates of AUD.

Data were collected in personal interviews with a representative sample of US adults interviewed in 1991-92. Analyses were conducted among past-year drinkers (12+ drinks, n = 18 352) and individuals with past-year DSM-IV AUD (n = 2770).

Thirty-one symptom item indicators, whose frequency of occurrence was measured in eight categories, were used to create unweighted and frequency-weighted counts of DSM-IV past-year AUD symptoms and criteria. Correlates included density of familial alcoholism and past-year volume of ethanol intake, frequency of intoxication and utilization of alcohol treatment.

Although the AUD correlates were associated strongly and positively with the frequency of AUD symptom occurrence, weighting for symptom frequency did not strengthen their association consistently with AUD severity scores. Improved performance of the weighted scores was observed primarily among AUD correlates linked closely with the frequency of heavy drinking and among individuals with AUD. Criterion counts were correlated nearly as strongly as symptom counts with the AUD correlates.

Frequency weighting may add somewhat to the validity of AUD severity measures, especially those that are intended for use among individuals with AUD, e.g. in clinical settings. For studying the etiology and course of AUD in the general population, an equally effective and less time-consuming alternative to obtaining symptom frequency may be the use of unweighted criterion counts accompanied by independent measures of frequency of heavy drinking.