Dealing with anxiety: A pilot cognitive behavioural therapy program for diabetic clinic outpatient attendees

2.1. Recruitment of subjects 

People who attended the Hunter Area Diabetes Services RNH Diabetes Outpatient Clinic were offered written information about participating in the Dealing with Anxiety [DWA] Group CBT Program. Those who consented to phone contact were sent information packages by mail. Consenting volunteers were then randomly allocated to an immediate intervention CBT group or a CBT group in 3months time [delayed intervention group]. There was no formal screening test for co-morbid psychiatric disorder but all patients were evaluated by a family practitioner experienced in psychiatry. One patient was excluded because of a borderline personality. This study was approved by the Hunter New England Research Ethics Committee clinical trial registration number NCT00659932.

2.2. Intervention 

The DWA CBT Program comprised seven group sessions: an initial 5h session followed by six 3h sessions over a 3 months period. The core program was cognitive behavioural therapy adapted from the Generalised Anxiety Disorder Patient Treatment Manual, of St. Vincent’s Hospital, Sydney with the addition, of diabetes specific resources. The intervention was delivered by one person.

Participants were asked to complete individual homework tasks between group sessions, based on the session content. Each participant also had an individual consultation, ranging from 1 to 2h duration, in between sessions 1 and 2 or sessions 2 and 3. Individuals with Depression Anxiety Stress Scale [DASS] [9] scores in the severe or extremely severe ranges were encouraged to consult their family General Practitioner for additional support.

2.3. Data collection 

There were five points of data collection at three monthly intervals, timed to coincide with routine HbA1C collections and the 14weeks duration of the CBT program. The data collected included haemoglobin A1C [HbA1C, the percentage of glycosylated haemoglobin in the blood], DASS and Diabetes Quality of Life [ADDQoL] [10] questionnaires at each timepoint. Demographic details were collected and thyroid stimulating hormone [TSH] levels measured at study entry. Of the 61 patients enrolled, one was excluded because of a personality disorder and complete data were collected from 55 at pre-treatment, 35 post-CBT 29 at 3months post-treatment and 31 at 6months post-treatment. Patients were excluded if they missed two or more sessions.

2.4. Measures 

Demographics included the basic descriptive data of age, gender, BMI, age ceased full time education, years since diagnosis of diabetes, type of treatment [lifestyle, oral medication, insulin], number of diabetes related complications and number of diabetes associated co-morbidities.

Evaluation of satisfaction with the conduct of the sessions was completed after each of the seven sessions, using a participant feedback questionnaire with eight items to assess process [e.g., comfort with conduct of group] and content [e.g., appropriateness/value of session content, identification of content needing improvement]. General comments were also invited.

Psychological instruments used were the DASS and the ADDQoL. The DASS is a set of three 14 item self-report scales designed to measure the negative emotional states of depression, anxiety and stress [nine Subjects are asked to use a 4 point severity/frequency scale to rate the extent to which they have experienced each state over the past week. As the essential development of the DASS was carried out with non-clinical samples, it is suitable for screening normal adults. The principal clinical value of the DASS is to clarify the locus of emotional disturbance and to assess the severity of the core symptoms of depression, anxiety and stress. It also gives recommended cut-offs for conventional severity levels [normal, mild, moderate, severe, extremely severe]. It does not measure suicidality.

The ADDQoL is a 20 item questionnaire that measures perceived quality of life in general [QI, two items] and in relation to living with diabetes [QII, 18 items] [10]. It weights each score by the person’s rating of importance [from −9 to +9], reflecting the most negative to the most positive impact of diabetes on that domain. A zero score means no impact of diabetes on the person for that domain and/or the domain is not important to that person. An average of these scores measures the overall impact of diabetes on an individual’s perceived quality of life.

2.5. Statistical analysis 

Data analysis was performed by the second author [TL], using Excel and SPSS software [version 14.0, SPSS Inc., Chicago, Ill, USA], and comprised two main elements: firstly, an analysis of the initial treatment effect [using data from time 1 and time 2]; and, secondly, an examination of the persistence of treatment effects and the consequences of delaying treatment [using data that were re-aligned to reflect the same treatment and follow-up phases]. Initial treatment effects were assessed using 2×[2] repeated measures ANOVAs [treatment status by time], where the “treatment group” received immediate treatment [IM] [i.e., pre- to post-treatment changes] and the “control group” consisted of the delayed treatment group [DL] [i.e., baseline to pre-treatment changes]. In order to demonstrate whether the treatment effect persisted, the data were re-aligned into the four main phases [pre- and post-treatment, and 3- and 6-month follow-ups]. In view of the fluctuations in sample size across these phases, planned comparisons were examined between all pairs of phases [from repeated measures ANOVAs].

As a partial control for the number of statistical tests, the threshold for significance was set at P<0.01. Given the pilot nature of this study, we also report as “possible trends” [i.e., for consideration in future studies] statistical tests in the range P>0.01 and <0.10. To facilitate discussion, differences are also expressed in standardised [effect-size] units, which were obtained by dividing the raw differences by the relevant grand standard deviation for the outcome of interest [based on all available data]. We also conducted a parallel series of traditional intention-to-treat [ITT] [11]. For these analyses, missing follow-up data were imputed by carrying forward the last available observation.

3.1. Sample characteristics 

The characteristics of the patients were as follows: age [n=60], mean 57.1, range 22–84; gender [n=60], 35 male, 25 female; BMI [n=45], mean 34.5, range 19.5–62.3; age ceased full time education [n=56], mean 16.1, range 11–28; years since diagnosis of diabetes [n=56], mean 14.3, range 1–45; number of diabetes related complications [including eyes, kidneys, blood vessels, feet, limbs, sexual dysfunction] [n=57], mean 1.77 range 0–5; number of diabetes associated co-morbidities [including hypertension, obesity, lipids disorders] [n=57], mean 1.84, range 0–3. For the majority of people who failed to complete the programme either ill health or conflicting appointments led to automatic exclusion for missing two or more sessions.

3.2. Initial treatment effect [Table 1] 

3.2.1. HbA1C 

Haemoglobin A1C rose significantly from time 1 to time 2 [P=0.003], reflecting a progression of the disease over time.

Table 1. Analysis of initial treatment effects: CBT program [immediate, IM] vs. waiting list [delayed, DL].

	Outcome measure	Group [G]	Time [T]		Statistically significant differences [estimated effect size]
			Time 1 [T1]	Time 2 [T2]
			Mean [SD]	Mean [SD]
	HbA1C	IM [n=13]	8.33 [1.44]	8.97 [1.92]	T: F[1,31]=10.00, P=0.003∗
		DL [n=20]	7.41 [1.64]	8.02 [1.83]
	General quality of life	IM [n=18]	0.61 [0.61]	0.67 [0.97]
	[ADDQoL – QI]	DL [n=25]	0.60 [1.04]	0.48 [1.12]
	Diabetes specific quality of life	IM [n=18]	−1.22 [1.17]	−0.94 [0.94]	T: F[1,41]=8.58, P=0.006∗
	[ADDQoL – QII]	DL [n=25]	−1.64 [0.95]	−1.20 [1.19]	[−0.36]
	Depression	IM [n=18]	13.50 [12.37]	8.94 [12.22]	G×T: F[1,40]=3.40, P=0.073#
	[DASS]	DL [n=24]	14.13 [11.38]	15.83 [12.49]	[IM: 0.39 vs. DL: −0.15]
	Anxiety	IM [n=18]	9.72 [9.90]	8.56 [10.82]	G x T: F[1,40]=2.98, P=0.092#
	[DASS]	DL [n=24]	10.38 [8.07]	12.04 [9.68]	[IM: 0.13 vs. DL: −0.19]
	Stress	IM [n=18]	15.06 [11.63]	13.06 [12.54]	G x T: F[1,40]=4.11, P=0.049#
	[DASS]	DL [n=24]	15.96 [10.20]	18.21 [11.68]	[IM: 0.17 vs. DL: −0.20]

HbA1C=glycosylated haemoglobin; ADDQoL=Diabetes Quality of Life; DASS=Depression Anxiety Stress Scale. Effects from Group [G] by Time [T] repeated measures analyses of variance [ANOVAs] are reported: ∗P<0.01; #possible trend [P>0.01 and <0.10].

3.3. Duration of treatment effect [Table 2] 

The “phase effects” column in Table 2 refers to overall differences between pairs of study phases in the delayed and immediate intervention group.

Table 2. Analysis of differences between treatment phases: CBT program [immediate, IM] vs. waiting list [delayed, DL].

	Outcome measure	Group [G]	Phase [all available data]				Statistically significant differences [Estimated effect size]
			Pre [PR]	Post [PO]	3months [3M]	6months [6M]	Phase effects	Interaction effects
			Mean [SD, n]	Mean [SD, n]	Mean [SD, n]	Mean [SD, n]
	HbA1C	IM	8.23 [1.64, 24]	8.69 [1.95, 15]	8.98 [2.06, 8]	8.78 [1.29, 12]		G×PR vs. PO: F[1,24]=9.06, P=0.002∗
		DL	7.90 [1.74, 23]	6.97 [1.24, 14]	6.90 [1.01, 13]	6.99 [0.93, 10]
		OV	8.06 [1.68, 47]	7.86 [1.84, 29]	7.69 [1.78, 21]	7.97 [1.44, 22]
	General quality of life [ADDQoL – QI]	IM	0.55 [0.69, 29]	0.67 [0.97, 18]	0.38 [1.19, 13]	0.67 [1.05, 15]	PR<PO: F[1,32]=2.89, P=0.099# [−0.26]
		DL	0.38 [1.20, 26]	1.00 [1.03, 16]	0.87 [0.83, 15]	0.92 [1.12, 13]
		OV	0.47 [0.96, 55]	0.82 [1.00, 34]	0.64 [1.03, 28]	0.79 [1.07, 28]
	Diabetes specific quality of life [ADDQoL – QII]	IM	−1.45 [1.02, 29]	−0.94 [0.94, 18]	−1.31 [1.03, 13]	−1.19 [0.75, 16]	PO>3M: F[1,23]=9.11, P=0.006∗ [0.23]	G×PO vs. 3M: F[1,23]=4.31, P=0.049# [IM: 0.41 vs. DL: 0.07]
		DL	−1.23 [1.18, 26]	−1.19 [1.11, 16]	−1.27 [1.10, 15]	−1.23 [1.09, 13]	PO>6M: F[1,26]=3.00, P=0.095# [0.22]
		OV	−1.35 [1.09, 55]	−1.06 [1.01, 34]	−1.29 [1.05, 28]	−1.21 [0.90, 29]
	Depression [DASS]	IM	15.31 [11.98, 29]	8.94 [12.22, 18]	13.29 [13.62, 14]	9.94 [11.09, 16]	PR>PO: F[1,31]=10.27, P=0.003∗ [0.39]
		DL	16.24 [12.39, 25]	8.25 [8.43, 16]	7.47 [7.72, 15]	10.00 [11.61, 13]	PR>3M: F[1,27]=6.24, P=0.019# [0.34]
		OV	15.74 [12.07, 54]	8.62 [10.46, 34]	10.28 [11.16, 29]	9.97 [11.12, 29]	PO<6M: F[1,26]=3.27, P=0.082# [−0.12]
	Anxiety [DASS]	IM	11.66 [9.48, 29]	8.56 [10.82, 18]	10.86 [10.44, 14]	8.44 [10.00, 16]	PR>PO: F[1,31]=12.26, P=0.001∗∗ [0.34]	G×PR vs. PO: F[1,31]=5.04, P=0.032# [IM: 0.13 vs. DL: 0.60]
		DL	12.00 [9.47, 25]	4.56 [3.74, 16]	6.33 [7.35, 15]	9.46 [9.13, 13]	PR>3M: F[1,27]=6.34, P=0.018# [0.25]	G×PR vs. 3M: F[1,27]=4.06, P=0.054# [IM: 0.05 vs. DL: 0.43]
		OV	11.81 [9.39, 54]	6.68 [8.41, 34]	8.52 [9.11, 29]	8.90 [9.47, 29]	PO<6M: F[1,26]=4.01, P=0.056# [−0.18]	G×PO vs. 6M: F[1,26]=4.01, P=0.056# [IM: 0.00 vs. DL: −0.43]
								G×3M vs. 6M: F[1,22]=3.66, P=0.069# [IM: 0.09 vs. DL: −0.32]
	Stress [DASS]	IM	17.14 [11.20, 29]	13.06 [12.54, 18]	15.57 [12.81, 14]	13.31 [11.30, 16]	PR>PO: F[1,31]=9.95, P=0.004∗ [0.37]	G×PR vs. PO: F[1,31]=3.00, P=0.093# [IM: 0.17 vs. DL: 0.60]
		DL	18.48 [11.52, 25]	10.44 [10.11, 16]	12.07 [12.52, 15]	13.46 [11.49, 13]	PR>3M: F[1,27]=7.29, P=0.012# [0.33]
		OV	17.76 [11.26, 54]	11.82 [11.37, 34]	13.76 [12.56, 29]	13.38 [11.18, 29]

OV=overall [IM+DL]; HbA1C=glycosylated haemoglobin; ADDQoL=Diabetes Quality of Life; DASS=Depression Anxiety Stress Scale. Effects from planned comparisons analysis of [pair-wise] differences between phases by Group [G] are reported [from repeated measures ANOVAs]: ∗P<0.01; ∗∗P<0.001; #possible trend [P>0.01 and <0.10].

3.4. Effect of delaying treatment [Table 2] 

The “interaction effects” column in Table 2 documents the extent to which delaying treatment impacted on the differences between pairs of study phases.