Screening for problem gambling within mental health services: a comparison of the classification accuracy of brief instruments

• DOI: 10.1111/add.14150
• Publication Date: 2017-12-24T10:30:24Z
• Authors: Nicki A. Dowling; Stephanie S. Merk...; Victorian Manning; Rachel Volberg; Stuart J. Lee; Simone N. Rodda; Dan I. Lubman
• Keywords: Adult; Male; Mental Health Services; 330; Victoria; Mental Disorders; 613; Middle Aged; Sensitivity and Specificity; Severity of Illness Index; 3. Good health; 03 medical and health sciences; 0302 clinical medicine; Surveys and Questionnaires; Gambling; Humans; Mass Screening; Female

AbstractBackground and aimsDespite the over‐representation of people with gambling problems in mental health populations, there is limited information available to guide the selection of brief screening instruments within mental health services. The primary aim was to compare the classification accuracy of nine brief problem gambling screening instruments (two to five items) with a reference standard among patients accessing mental health services.DesignThe classification accuracy of nine brief screening instruments was compared with multiple cut‐off scores on a reference standard.SettingEight mental health services in Victoria, Australia.ParticipantsA total of 837 patients were recruited consecutively between June 2015 and January 2016.MeasurementsThe brief screening instruments were the Lie/Bet Questionnaire, Brief Problem Gambling Screen (BPGS) (two‐ to five‐item versions), NODS‐CLiP, NODS‐CLiP2, Brief Biosocial Gambling Screen (BBGS) and NODS‐PERC. The Problem Gambling Severity Index (PGSI) was the reference standard.FindingsThe five‐item BPGS was the only instrument displaying satisfactory classification accuracy in detecting any level of gambling problem (low‐risk, moderate‐risk or problem gambling) (sensitivity = 0.803, specificity = 0.982, diagnostic efficiency = 0.943). Several shorter instruments adequately detected both problem and moderate‐risk, but not low‐risk, gambling: two three‐item instruments (NODS‐CLiP, three‐item BPGS) and two four‐item instruments (NODS‐PERC, four‐item BPGS) (sensitivity = 0.854–0.966, specificity = 0.901–0.954, diagnostic efficiency = 0.908–0.941). The four‐item instruments, however, did not provide any considerable advantage over the three‐item instruments. Similarly, the very brief (two‐item) instruments (Lie/Bet and two‐item BPGS) adequately detected problem gambling (sensitivity = 0.811–0.868, specificity = 0.938–0.943, diagnostic efficiency = 0.933–0.934), but not moderate‐risk or low‐risk gambling.ConclusionsThe optimal brief screening instrument for mental health services wanting to screen for any level of gambling problem is the five‐item Brief Problem Gambling Screen (BPGS). Services wanting to employ a shorter instrument or to screen only for more severe gambling problems (moderate‐risk/problem gambling) can employ the NODS‐CLiP or the three‐item BPGS. Services that are only able to accommodate a very brief instrument can employ the Lie/Bet Questionnaire or the two‐item BPGS.