2.4 Measures
2.4.1 The screening measures
2.4.1.1 Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001).
This brief questionnaire was used to screen out participants with severe levels of depression (see appendix B). It consists of nine questions that match the criteria in the Diagnostic and Statistical Manual of Mental Disorders-fourth edition (DSM-IV) for a major depressive episode.
For the purposes of the present study, the PHQ-9 was used as a depression severity measure. Total scores range from 0 to 27 and scores for each question range from ‘0’ (not at all) to ‘3’ (nearly every day). A total score of 1-4 represents minimal depression; 5-9, mild depression; 10-14, moderate depression; 15-19, moderately severe depression; and 20-27, severe depression.
The PHQ-9 is routinely used in primary care including the Wellbeing Services that the present study recruited from and typically takes two-three minutes to complete. It was therefore a suitable depression measure to use.
The authors of the PHQ-9 report various psychometric findings (Kroenke et al., 2001). Internal reliability was found to be excellent when it was completed by 6,000 patients in multiple primary care and obstetrics-gynaecology clinics, demonstrating Cronbach α of 0.89 and 0.86 respectively. At the severe level threshold, the specificity of the PHQ-9 for diagnosing major depression was 95% with a likelihood ratio of 13.6. It was also found to have good construct validity with severe levels associated with worst
48 functioning on all six domains of the Study Short Form General Health Survey (SF-20; Stewart, Hays, & Ware, 1988). Cameron, Crawford, Lawton, and Reid (2008) also validated the PHQ-9 using primary care patients and found this measure to demonstrate discriminant validity when compared to an established anxiety measure, internal
consistency (e.g., a Cronbach α of 0.89) and robustness of factor structure.
2.4.1.2 Brief Symptom Inventory (BSI; Derogatis, 1975).
This 53-item questionnaire assesses the psychological symptom status of patients. In the present study it was used to gain an overall clinical picture of the participant’s presentation. The BSI consists of nine dimensions relating to general symptoms of mental health, namely somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation and psychotism. The questionnaire typically takes approximately ten minutes to complete. All items are ranked on a 5-point rating scale of ‘0’ (not at all) to ‘5’ (extremely) to reflect perceived distress during the past seven days. Of primary interest to the researcher were items relating to psychotism to check whether further assessment was required to ensure that participants were not floridly psychotic, and therefore ineligible.
Derogatis and Melisaratos (1983) reported good internal consistency reliability for the nine dimensions of the BSI with Cronbach α ranging from 0.71 to 0.85. The BSI manual (Derogatis, 1993) sites factor analyses results confirming the a-priori construction of the different dimensions and correlations with the Symptom Checklist-90-Revised (SCL-90-R; Derogatis, 1977) ranging between 0.92 and 0.99.
2.4.1.3 Generalized Anxiety Disorder Questionnaire (GAD-Q-IV; Newman et al.,
2002).
This self-report diagnostic measure of GAD is closely based on criteria for GAD set out in the DSM-IV (see appendix C). It is described by Newman et al. (2002) as an
49 effective way to screen for the presence or absence of diagnosable GAD and so this was its primary purpose in the present study. It is a much more convenient and quicker method of screening for GAD compared to structured diagnostic interviews and typically takes approximately five minutes to complete. The GAD-Q-IV has been used as a screening tool for GAD in other CBM-I studies (Hayes et al., 2010; Hirsch et al., 2009). In the current study it will also be used as an outcome measure to assess severity of generalised anxiety.
The GAD-Q-IV consists of nine questions that include: five yes/no checklists assessing the occurrence of worry, a DSM-IV symptom check list, a listing of the most frequent topics of worry and two 8-point Likert scales ranging from ‘0’ (none) to ‘8’ (very severely) assessing distress and interference of worry and physical symptoms. The authors recommend a weighted scoring system that provides an overall index of the severity of GAD.
The maximum score is 13 however the authors reported that a score cut-off of 5.7 yields the optimal balance between sensitivity (83%) and specificity (89%). The authors also noted this cut-off to have generated a false positive rate of 11% and a false negative rate of 17% with the Anxiety Disorders Interview Schedule for DSM-IV (ADIS-IV;
Brown, Di Nardo, & Barlow, 1994); that is, it incorrectly classified 11% of cases as having GAD and incorrectly classified 17% of cases as not having GAD.
Newman et al. (2002) found that in a sample of undergraduate students, the GAD-Q-IV successfully discriminated individuals diagnosed with GAD using either the ADIS-IV or the Anxiety Disorders Interview Schedule for DSM-IV, Lifetime Version (ADIS-IV-L; Di Nardo, Brown, & Barlow, 1994) from individuals diagnosed with panic disorder and social phobia - common co-morbid diagnoses of GAD. Newman and colleagues also found the GAD-Q-IV to have good convergent and discriminant validity compared to a battery of other anxiety-related measures. When test-retest reliability was
50 examined using the 5.7 cut-off score, a Kappa agreement between Time 1 and Time 2 of 0.64 was generated. In addition, the authors performed a logistic regression which showed that the GAD-Q-IV score at time 2 was reliably predicted by time 1 score (X2(1, N = 148) = 42.1, p < .001).
In another study where the ADIS-IV-L and the GAD-Q-IV were administered using a cut-off score of 5.7 (Luterek, Turk, Heimberg, Fresco, & Mennin, 2002), the GAD-Q-IV correctly classified 50 of 53 non-anxious community participants as not having GAD (96.2% specificity) and all of the 31 participants with GAD as having GAD (100% specificity).
Newman et al. (2002) also compared their study’s undergraduate sample to a community sample of individuals with GAD and found that they did not differ on the PSWQ and the STAI suggesting that these results are generalisable to community samples.
Beyond these studies, relatively little psychometric data is available on the GAD-Q-IV although it has been tested favorably in some research using older adult samples (Diefenbach, Tolin, Meunier, & Gilliam, 2009; Staples & Mohlman, 2012; Webb et al., 2008). Also some confusion remains over the optimal cut-off to use and optimal scoring system (Rodebaugh, Holaway, & Heimberg, 2008).
Because of these considerations and in the absence of a rigorous diagnostic assessment in the present study, caution was exercised when describing the diagnostic status of eligible participants who met Newman and colleagues’ advised cut-off of 5.7. Participants in the current study are viewed as experiencing clinical levels of generalised anxiety, rather than as having diagnosed GAD.