Journal of Behavioral Medicine

, Volume 36, Issue 6, pp 556–566

Effects of a cognitive behavioural treatment in patients with type 2 diabetes when added to managed care; a randomised controlled trial

Authors

    • Department of General Practice, EMGO Institute for Health and Care ResearchVU University Medical Center
  • Patricia van Oppen
    • Department of General Practice, EMGO Institute for Health and Care ResearchVU University Medical Center
    • Department of Psychiatry, EMGO Institute for Health and Care ResearchVU University Medical Center
  • Sandra D. M. Bot
    • Department of General Practice, EMGO Institute for Health and Care ResearchVU University Medical Center
    • Department of Epidemiology and Biostatistics, EMGO Institute for Health and Care ResearchVU University Medical Center
  • Piet J. Kostense
    • Department of Epidemiology and Biostatistics, EMGO Institute for Health and Care ResearchVU University Medical Center
  • Jacqueline M. Dekker
    • Department of Epidemiology and Biostatistics, EMGO Institute for Health and Care ResearchVU University Medical Center
  • Giel Nijpels
    • Department of General Practice, EMGO Institute for Health and Care ResearchVU University Medical Center
Article

DOI: 10.1007/s10865-012-9451-z

Cite this article as:
Welschen, L.M.C., van Oppen, P., Bot, S.D.M. et al. J Behav Med (2013) 36: 556. doi:10.1007/s10865-012-9451-z

Abstract

Effects of a cognitive behavioural treatment (CBT) in type 2 diabetes patients were studied in a randomised controlled trial. Patients were recruited from a diabetes care system (DCS). The intervention group (n = 76) received managed care from the DCS and CBT. The control group (n = 78) received managed care only. Effects on risk of developing coronary heart disease (CHD), clinical characteristics, lifestyle, determinants of behaviour change, quality of life, and depression were assessed after 6 and 12 months. The intervention did not result in a significant reduction of CHD risk (difference between intervention and control group was −0.32 % (95 % CI: −2.27; 1.63). The amount of heavy physical activity increased significantly in the intervention group at 6 months [intervention versus control group was 20.14 min/day (95 % CI: 4.6; 35.70)]. Quality of life and level of depression improved as well. All effects disappeared after 6 months. No effects were found on clinical characteristics.

Keywords

Type 2 diabetesLifestyle interventionCognitive behavioural treatmentCardiovascular disease riskRandomised controlled trial

Introduction

Behavioural interventions focused on lifestyle in patients with type 2 diabetes (T2DM) have shown to be effective in improvements of clinical outcomes. (Ismail et al., 2004; Norris et al., 2004). It is still unclear what kind of intervention is most effective (Ismail et al., 2004), but it is generally acknowledged that a behavioural intervention should be based on a theoretical framework (Norris et al., 2001; Peyrot & Rubin, 2007; Hardeman et al., 2005) and that such interventions should focus on the increase of self-management of the patient (Glasgow et al., 2004; Peyrot & Rubin, 2007).

In order to improve patients’ clinical characteristics and finally the 10-year risk of a coronary heart disease (CHD), we have added a cognitive behavioural treatment (CBT), which has been supported in the literature to facilitate behaviour change, to the standard diabetes care (Spahn et al., 2010).

The CBT was guided by techniques of Problem Solving Treatment (PST), a practical skill building treatment (D’Zurilla & Nezu, 2001; Mynors-Wallis, 2005). PST is focused to teach patients to use their own skills to resolve problems and improve their symptoms, a key element for successful self-management of diabetes (Franke et al., 2007; Glasgow et al., 2004).

We hypothesized that PST might increase patient’s attitude, social influences and self-efficacy, according to the theoretical framework of the Attitude, Social influences and self-Efficacy model (ASE-model) which attribute to a patients’ intention to change behaviour, resulting in improvement of patients’ characteristics and in a reduction of CHD risk (Brug et al., 1995). The intervention aimed to increase physical activity, change eating behaviour and/or quit smoking. It is believed that this contributes to a more likely increase in self-management of patients than an intervention focused on one domain, because patients are encouraged to make choices (Clark et al., 2004). In contrast to other studies this theoretical driven study was implemented in clinical care and patients were allowed to choose the behaviour they wanted to change.

The present study describes the effects of adding a CBT aimed at changing lifestyle for T2DM patients on the estimated risk of developing a CHD event in the next 10 years (Stevens et al., 2001). Effects on lifestyle, patients’ clinical characteristics, and determinants of behaviour change were also assessed. In addition, we compared quality of life, and the level of depression between the intervention and control group.

Methods

Study design

A randomised controlled trial was conducted with T2DM patients that were included in the Diabetes Care System (DCS), a disease management system in the Netherlands (Welschen et al., 2006). The design of the study has been described in detail previously (Welschen et al., 2007).

Patients were invited from 13 general practices participating in the DCS for a recruitment visit by means of an invitation letter, including information on the study. The inclusion period lasted for 1 year. Patients were considered eligible for the study based on the following inclusion criteria: 18–75 years old; able to understand the Dutch language; at high risk of developing cardiovascular disease and diabetes complications (HbA1c ≥ 52 mmol/mol (7.0 %) and/or body-mass index ≥ 27.0 kg/m2 and/or smoking). Eligible patients were randomised at an individual level into an intervention and a control group by means of block randomization within general practices, to avoid contamination of the intervention by the general practitioner. Randomisation was performed by a manager of the DCS, not involved in the patients’ care by means of a list drawn up by a computer program (Random Allocation Software version 1.0.0). Patients, diabetes nurses and dieticians could not be blinded to the intervention. Study participants were seen by different diabetes nurses and dieticians if in control or intervention group. The medical assistants had contact with the patients only prior to randomisation and were not involved in the intervention. Therefore for them it was not necessary to be blinded. The principal investigator remains blinded during the entire intervention. All participating patients gave their written informed consent. The Medical Ethics Committee of the VU University Medical Center in Amsterdam approved the study design.

Control group

The control group received the managed diabetes care from the DCS. The diabetes care in the DCS is organized in yearly assessments consisting of a measurement visit followed by a visit to a dietician and a diabetes nurse in order to receive information and education. A standardized protocol according to the Guidelines of the Dutch College of General Practitioners was used (Bouma et al., 2006).

Intervention group

The intervention group was planned to receive 3–6 CBT sessions of 30 min, which was dependent on the need of the patient. During the first session, the most important behavioural domain was assessed by the diabetes nurse. The intervention was transmitted to a dietician if the component was related to dietary intake. In case of smoking or physical activity, a diabetes nurse continued with the intervention. During all sessions, PST was used to set achievable goals for behaviour change. PST consists of several steps including: problem definition, goal setting, generating a solution, implementing the solution, and evaluation of the outcome of the implementation (Mynors-Wallis, 2005).

The dieticians (n = 6) and diabetes nurses (n = 4) had received a training in performing the CBT of 2 days, followed by two instruction days on how to implement the intervention in the DCS. A treatment manual was used during the intervention. In addition, all sessions were tape recorded in order to be able to assess treatment fidelity. Each 4 weeks, a supervision meeting, guided by a specialized CBT psychologist (PvO), was organized.

Measurements

Outcome measurements were extracted from self-reported questionnaires and physical examinations.

CHD risk

The 10-year risk of developing a CHD event was calculated at baseline and at 12 months using the UK Prospective Diabetes Study (UKPDS) risk engine (Stevens et al., 2001). Variables included in this algorithm are: age at diagnosis, duration of diabetes, sex, ethnicity, smoking status, systolic blood pressure, HbA1c, total and HDL-cholesterol.

Clinical characteristics

Measurements were taken at the DCS by research assistants according to a standardized protocol. Weight and height were measured while patients were barefoot and wearing only light clothes. Systolic and diastolic blood pressure were measured after 5 min of rest in a seated position using a oscillimetric device (Colin Press Mate BP-8800, Komaki City, Japan). HbA1c was measured with high performance liquid chromatography. Fasting plasma glucose was measured by means of a hexokinase method (Roche Diagnostics GmbH, Mannheim, Germany). Levels of total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were measured using enzymatic techniques (Boehringer-Mannheim, Mannheim, Germany). All measurements were performed at baseline, after 6 and 12 months, except for total and HDL cholesterol and triglycerides which were only measured at baseline and after 12 months.

Lifestyle: physical activity, eating behaviour and smoking

The SQUASH questionnaire (Short Questionnaire to Assess Health Enhancing Physical Activity) was used to assess physical activity (Wendel-Vos et al., 2003). The total amount of minutes per day that a patient was performing light (2–4 MET), moderate (4–6.5 MET), or heavy physical activity (≥6.5 MET) was calculated [MET = unit of metabolic equivalent, which is the ratio of the energy cost of a given activity to resting metabolic rate and was derived from published tables (Ainsworth et al., 2000)].

Eating behaviour was assessed by the Dutch Eating Behaviour Questionnaire (DEBQ) (Van Strien et al., 1986). This questionnaire assessed whether a patient is a restraint eater (overeating after a period of slimming), an external eater (eating in response to external foot cues), or an emotional eater (eating in response to emotional arousal states). Classification into one of these three domains was achieved by dividing the sum of the corresponding items for a specific domain by the number of items.

Smoking was assessed as a dichotomous outcome measure asking if a patient was a smoker or a non-smoker.

Determinants of behaviour change

Determinants of behaviour change were assessed by means of a questionnaire developed according to the ASE model (de Vries et al., 1995). There were 3 separate questionnaires for all 3 behavioural domains (increase physical activity, eat healthier, and quit smoking), each containing the same items. There were 5 items concerning attitude, 2 items on self-efficacy, 2 items on social influences (1 item for influences from the partner and 1 from friends) and 2 items on intention to change behaviour. All items were measured on a 7-point Likert scale.

The mean of the items of each behavioural determinant was taken as a measure for the specific determinant. Low scores indicate a positive attitude towards change of behaviour, a high self-efficacy to perform the behaviour, a positive social influence from the partner and/or friends, and a strong intention to change behaviour.

Cronbach’s α were all >0.60, indicating sufficient internal consistency. Except for the Cronbach’s α of smoking (between 0.50 and 0.60).

Quality of life

The EuroQol was used to assess quality of life (Brooks, 1996). This questionnaire consists of a visual analogue scale on which patients had to indicate their health status (scale 0–100) and five questions on different domains, each with a scale of three levels: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. A mean weighted health status was calculated with a range of 0–1.

Depression

An important co-morbidity of patients with diabetes and also an important covariate in intervention studies is depression (Lustman & Clouse, 2005). We used the CES-D (Center for Epidemiological Studies Depression scale) to assess if depression was present (Radloff, 1977). An overall score was calculated by summing up all scores, resulting in an overall score between 0 and 60. A patient with a score ≥16 was considered as a possible depressive case.

Statistical analyses

Two way analyses of variance were used to calculate differences in the changes between 0–6 and 0–12 months between the two groups. General practices were added to the model as a fixed factor, to account for differences between them. Mantel–Haenszel statistic was used to assess differences between the two groups for dichotomous outcome measures. In additional analyses was adjusted for age, gender, and diabetes duration. All analyses were performed according to the intention-to-treat principle.

The following subgroup analyses were performed to investigate if there were specific patients who would benefit from the intervention: high education level (college/university), per-protocol analysis including patients with ≥3 CBT sessions, patients without depression (CES-D score < 16).

Missing data were not imputated. P values below 0.05 were considered statistically significant. All statistical analyses were performed using SPSS for Windows (version 14.0, SPSS Inc., Chicago, IL).

Results

At baseline, 76 patients were randomised to the intervention group and 78 patients to the control group. A flow chart showing follow-up of patients can be found in Fig. 1. The mean amount of CBT sessions patients in the intervention group attended was 3 (SD 1.7).
https://static-content.springer.com/image/art%3A10.1007%2Fs10865-012-9451-z/MediaObjects/10865_2012_9451_Fig1_HTML.gif
Fig. 1

Design of the RCT

More men than women were included in both groups (Table 1). Groups were comparable at baseline. No differences were found in characteristics of patients who dropped out of the study and those who completed the study.
Table 1

Baseline characteristics

 

Intervention group (n = 76)

Control group (n = 78)

Age (years)

60.5 ± 9.4

61.2 ± 8.8

Gender (% male)

59.5

64.2

Ethnicity (% Caucasian)

97.3

94.9

Marital/cohabiting status (% with partner)

84.8

83.3

Work status (% employed)

30.3

38.2

Level of education (%)

 Primary

50

42.9

 Secondary

34.8

44.2

 College/university

15.2

12.9

Smoking (% smokers)

28.3

23.3

Diabetes duration (years)

7.6 ± 5.0

7.8 ± 6.1

Body-mass index (kg/m2)

31.6 ± 5.7

31.5 ± 5.2

Systolic blood pressure (mmHg)

144.6 ± 20.3

144.6 ± 18.3

Diastolic blood pressure (mmHg)

77.8 ± 7.9

77.0 ± 9.0

HbA1c (mmol/l)

51 ± 11

49 ± 11

HbA1c (%)

6.8 ± 1.0

6.7 ± 1.0

Fasting blood glucose (mmol/l)

7.8 ± 2.2

7.7 ± 1.5

Total cholesterol (mmol/l)

4.4 ± 1.0

4.4 ± 1.0

HDL cholesterol (mmol/l)

1.2 ± 0.3

1.2 ± 0.3

Triglycerides (mmol/l)

1.9 ± 1.1

2.0 ± 1.0

Risk for CVD (%)

10.4 ± 7.4

11.0 ± 8.7

PHQ score depression (% patients)

 No depression (score = 0)

19.7

24.2

 Minimal (1–4)

36.1

39.4

 Mild (5–9)

36.1

31.8

 Moderate (10–14)

4.9

1.5

 Moderately severe (15–19)

3.3

1.5

 Severe (20–27)

0

1.5

CES-D score ≥ 16 (% depressive patients)

23.4

17.8

Data are mean ± SD, or % of patients

CHD risk

The risk of developing CHD in the next 10 years decreased from 10.6 to 9.9 % in the intervention group and increased from 11.1 to 11.2 % in the control group (Table 2). The difference in change was not statistically significantly [95 % BI: −2.27; 1.63].
Table 2

Differences in CHD risk and clinical characteristics between patients in the intervention group and the control group at baseline, after 6 and 12 months

 

Baseline

6 months

12 months

P value Δ 0–6 months (95 % CI)

P value Δ 0–12 months (95 % CI)

CHD risk (%, UKPDS risk engine)a

Intervention

10.6 ± 7.2

NA

9.9 ± 5.9

NA

P = 0.75 (−2.27; 1.63)

Control

11.1 ± 8.8

NA

11.2 ± 7.0

 

Weight (kg)a

Intervention

94.4 ± 19.8

95.3 ± 20.8

94.1 ± 22.1

P = 0.78 (−0.90; 1.19)

P = 0.73 (−1.33; 0.93)

Control

95.8 ± 17.0

96.0 ± 17.2

95.8 ± 16.6

Body-mass indexa(kg/m2)

Intervention

31.6 ± 5.7

31.8 ± 6.0

31.5 ± 6.1

P = 0.75 (−0.29; 0.40)

P = 0.77 (−0.50; 0.37)

Control

31.6 ± 5.2

31.6 ± 5.3

31.8 ± 5.2

Systolic blood pressure (mmHg)a

Intervention

144.5 ± 20.3

143.3 ± 21.3

143.7 ± 19.9

P = 0.53 (−7.08; 3.69)

P = 0.80 (−5.89; 4.58)

Control

144.6 ± 18.3

145.1 ± 19.3

142.4 ± 18.0

Diastolic blood pressure (mmHg)a

Intervention

77.8 ± 7.9

75.4 ± 8.2

76.4 ± 9.1

P = 0.08 (−5.50; 0.31)

P = 0.22 (−4.22; 0.96)

Control

77.0 ± 9.0

77.0 ± 8.3

77.0 ± 8.3

HbA1c (mmol/mol)a

Intervention

51 ± 11

49 ± 12

51 ± 12

P = 0.65 (−4.27; 2.67)

P = 0.71 (−3.55; 2.42)

Control

49 ± 11

50 ± 10

51 ± 10

HbA1c (%)a

Intervention

6.8 ± 1.0

6.7 ± 1.1

6.8 ± 1.1

P = 0.65 (−0.39; 0.24)

P = 0.71 (−0.33; 0.22)

Control

6.7 ± 1.0

6.7 ± 0.9

6.8 ± 0.9

Fasting blood glucose (mmol/l)a

Intervention

7.8 ± 2.2

7.9 ± 2.6

8.1 ± 2.2

P = 0.25 (−0.29; 1.11)

P = 0.37 (−0.36; 0.97)

Control

7.7 ± 1.5

7.7 ± 1.5

7.6 ± 1.8

Total cholesterol (mmol/l)a

Intervention

4.4 ± 1.0

NA

4.1 ± 0.7

NA

P = 0.24 (−0.46; 0.11)

Control

4.4 ± 1.0

NA

4.3 ± 0.9

HDL cholesterol (mmol/l)a

Intervention

1.2 ± 0.3

NA

1.1 ± 0.3

NA

P = 0.49 (−0.07; 0.03)

Control

1.2 ± 0.3

NA

1.2 ± 0.3

Triglycerides (mmol/l)a

Intervention

2.0 ± 1.1

NA

1.9 ± 1.3

NA

P = 0.55 (−0.45; 0.24)

Control

1.9 ± 0.9

NA

2.0 ± 1.1

Data are mean ± SD, NA not applicable

aTwo way analysis of variance, adjusted for general practice

Clinical characteristics

Clinical characteristics are shown in Table 2. We only found small differences between the intervention and the control group after both 6 and 12 months of follow-up, but these were considered not of clinical importance.

Lifestyle: physical activity, eating behaviour and smoking

We found statistically significant differences (Table 3) on heavy physical activity at 6 months but these had disappeared at 12 months. There were no statistically significant differences between the three time points on eating behaviour. The percentage of smokers decreased from 25 to 18.3 % in the intervention group, compared with an increase from 22.4 to 22.9 % in the control group, although this did not reach statistical significance.
Table 3

Differences in lifestyle (physical activity, eating behaviour and smoking) and determinants of behaviour change (ASE model: attitude, social influences and self-efficacy model) between patients in the intervention and control group at baseline, after 6 and 12 months

 

Baseline

6 months

12 months

P value Δ 0–6 months (95 % CI)

P value Δ 0–12 months (95 % CI)

Lifestyle

Physical activity SQUASHa (min/day)d

 Light activities

  Intervention

104 (38–251)

101 (41–197)

135 (30–291)

P = 0.95 (−40.63; 43.45)

P = 0.28 (−29.85; 103.15)

  Control

135 (0–274)

120 (19–255)

171 (17–332)

 Moderate activities

  Intervention

40 (17–134)

51 (12–130)

69 (13–171)

P = 0.47 (−36.62; 17.03)

P = 0.97 (−42.84; 41.45)

  Control

41 (17–80)

30 (16–99)

34 (10–151)

 Heavy activities

  Intervention

17 (0–46)

23 (0–55)

29 (0–79)

P = 0.01* (4.6; 35.70)

P = 0.38 (−14.73; 36.08)

  Control

17 (0–58)

17 (0–43)

14 (0–73)

Eating behaviour DEBQb,d

 Emotional

  Intervention

1.87 ± 0.90

1.84 ± 0.87

1.82 ± 0.80

P = 0.61 (−0.14; 0.24)

P = 0.17 (−0.05; 0.67)

  Control

1.82 ± 0.75

1.74 ± 0.81

1.66 ± 0.72

 External

  Intervention

2.28 ± 0.59

2.19 ± 0.63

2.16 ± 0.64

P = 0.87 (−0.15; 0.13)

P = 0.21 (−0.27; 0.06)

  Control

2.30 ± 0.58

2.23 ± 0.63

2.22 ± 0.64

 Restraint

  Intervention

2.87 ± 0.81

2.94 ± 0.81

2.83 ± 0.80

P = 0.59 (−0.15; 0.26)

P = 0.25 (−0.33; 0.09)

  Control

2.94 ± 0.82

2.95 ± 0.85

2.92 ± 0.83

 Smoking (n (%))e

  Intervention

17/68 (25 %)

11/60 (18.3 %)

9/62 (14.5 %)

P = 0.94

P = 0.96

  Control

17/76 (22.4 %)

16/70 (22.9 %)

14/69 (20.3%)

Determinants of behaviour change

ASE physical activityc,d

 Attitude

  Intervention

2.7 ± 0.2

2.5 ± 0.1

2.6 ± 0.1

P = 0.25 (−0.20; 0.77)

P = 0.16 (−0.13; 0.76)

  Control

3.1 ± 0.2

2.7 ± 0.1

2.8 ± 0.1

 Self-efficacy

  Intervention

4.3 ± 0.2

4.4 ± 0.2

4.1 ± 0.2

P = 0.50 (−0.73; 0.36)

P = 0.27 (−0.83; 0.24)

  Control

4.3 ± 0.2

4.5 ± 0.2

4.3 ± 0.2

 Social influences partner

  Intervention

2.6 ± 0.2

2.7 ± 0.2

2.7 ± 0.2

P = 0.95 (−0.54; 0.50)

P = 0.75 (−0.49; 0.67)

  Control

2.8 ± 0.2

2.9 ± 0.2

2.8 ± 0.2

 Social influences friends

  Intervention

4.3 ± 0.2

4.1 ± 0.2

4.4 ± 0.2

P = 0.57 (−0.64; 0.36)

P = 0.10 (−0.83; 0.24)

  Control

4.5 ± 0.2

4.4 ± 0.1

4.1 ± 0.1

 Intention

  Intervention

3.8 ± 0.2

3.7 ± 0.2

4.0 ± 0.2

P = 0.09 (−1.28; 0.09)

P = 0.98 (−0.73; 0.75)

  Control

4.1 ± 0.2

4.4 ± 0.2

4.2 ± 0.2

ASE dietary intakec,d

 Attitude

  Intervention

3.3 ± 0.1

3.3 ± 0.1

3.4 ± 0.1

P = 0.29 (−0.52; 0.16)

P = 0.72 (−0.51; 0.35)

  Control

3.2 ± 0.1

3.4 ± 0.1

3.5 ± 0.1

 Self-efficacy

  Intervention

4.0 ± 0.3

3.4 ± 0.2

3.9 ± 0.2

P = 0.14 (−1.23; 0.18)

P = 0.60 (−1.02; 0.60)

  Control

3.6 ± 0.2

3.8 ± 0.2

3.6 ± 0.2

 Social influences partner

  Intervention

3.2 ± 0.3

3.5 ± 0.2

3.0 ± 0.2

P = 0.63 (−0.42; 0.69)

P = 0.80 (−0.96; 0.74)

  Control

3.4 ± 0.2

3.2 ± 0.2

3.0 ± 0.2

 Social influences friends

  Intervention

4.6 ± 0.1

4.2 ± 0.1

4.4 ± 0.1

P = 0.72 (−0.41; 0.58)

P = 0.42 (−0.25; 0.59)

  Control

4.5 ± 0.1

4.3 ± 0.1

4.3 ± 0.1

 Intention

  Intervention

4.1 ± 0.3

4.3 ± 0.2

4.5 ± 0.2

P = 0.66 (−0.56; 0.88)

P = 0.35 (−0.37; 1.05)

  Control

4.5 ± 0.3

4.5 ± 0.2

4.6 ± 0.2

ASE Smokingc,d

 Attitude

  Intervention

3.7 ± 0.4

2.7 ± 0.2

3.2 ± 0.2

P = 0.53 (−0.95; 1.75)

P = 0.21 (−0.62; 2.56)

  Control

3.4 ± 0.4

2.5 ± 0.2

2.6 ± 0.3

 Self-efficacy

  Intervention

3.9 ± 0.6

4.0 ± 0.6

4.7 ± 0.6

P = 0.35 (−1.54; 0.58)

P = 0.98 (−0.74; 0.76)

  Control

5.4 ± 0.4

4.9 ± 0.5

4.9 ± 0.5

 Social influences partner

  Intervention

2.9 ± 0.5

2.4 ± 0.5

3.3 ± 0.6

P = 0.87 (−1.27; 1.08)

P = 0.29 (−0.58; 1.84)

  Control

1.7 ± 0.4

1.8 ± 0.5

1.8 ± 0.4

 Social influences friends

  Intervention

4.9 ± 0.3

4.0 ± 0.4

4.5 ± 0.3

P = 0.78 (−1.53; 2.01)

P = 0.92 (−1.40; 1.27)

  Control

4.8 ± 0.3

4.3 ± 0.3

4.7 ± 0.3

 Intention

  Intervention

4.0 ± 0.5

3.8 ± 0.4

4.2 ± 0.6

P = 0.55 (−1.52; 2.75)

P = 0.34 (−1.02; 2.79)

  Control

4.4 ± 0.4

3.7 ± 0.4

3.7 ± 0.4

P < 0.05

aSQUASH = Short questionnaire to assess health enhancing physical activity. Values are median (interquartile range). Metabolic equivalent of task (MET) in minutes per day, representing the time engaged in specified physical activities multiplied by the metabolic equivalent value of each activity. Light activities are rated as 2.0 to <4.0 METs, moderate activities are rated as ≥4.0 to <6.5 METs, heavy activities are rated as ≥6.5 METs

bDEBQ = Dutch eating behaviour questionnaire. Values are mean ± SD, measured on a 5-point scale. Higher scores indicate that the patient is likely to be a emotional, external and/or restraint eater

cASE model = Attitude, social influences and self-efficacy model. Values are mean ± SD, measured on a 7-point scale. Low scores indicate a positive attitude towards change of behaviour, a high self-efficacy to perform the behaviour, a positive social influence from the partner and/or friends, and a strong intention to change behaviour

dTwo way analysis of variance, adjusted for general practice

eMantel–Haenszel statistic, adjusted for general practice

Determinants of behaviour

We found no statistically significant differences between the two groups on the components of the ASE-model (Table 3). Self-efficacy to change behaviour was neither high nor low, in both groups at all time points. Patients did not have a strong intention to change behaviour, and neither a weak one.

Quality of life

Quality of life, as measured by the EuroQoL questions, improved little in both groups (Table 4). The intervention group showed a statistically significant increase in quality of life between baseline and 6 months on the VAS scale, whereas the control group showed a decrease.
Table 4

Differences in quality of life and depression between patients in the intervention and control group at baseline, after 6 and 12 months

 

Baseline

6 months

12 months

P value Δ 0–6 months (95 % CI)

P value Δ 0–12 months (95 % CI)

Quality of lifea

Score EuroQoLd

 Intervention

0.71 ± 0.28

0.73 ± 0.27

0.73 ± 0.25

P = 0.45 (−0.05; 0.10)

P = 0.40 (−0.04; 0.10)

 Control

0.76 ± 0.24

0.77 ± 0.19

0.74 ± 0.25

VAS scale EuroQoLd

 Intervention

65.2 ± 18.2

68.4 ± 17.5

65.1 ± 20.5

P = 0.03* (0.54; 11.08)

P = 0.59 (−4.65; 8.12)

 Control

65.6 ± 18.4

63.6 ± 17.2

61.5 ± 19.2

Depression: CES-Db

Score CES-Dd

 Intervention

11.1 ± 8.1

9.9 ± 7.7

11.3 ± 9.9

P = 0.01* (−4.70; −0.81)

P = 0.70 (−2.38; 1.60)

 Control

9.6 ± 8.2

10.3 ± 8.6

11.0 ± 9.4

Depressive patients (%)c,e

 Intervention

23.4

21.7

26.2

  

 Control

17.8

14.9

17.9

P = 0.63

P = 0.90

P < 0.05

aEuroQoL = Values indicate a mean weighted health status ± SD, with a range of 0–1. VAS = Visual analogue scale EuroQoL (scale 0–100). Higher scores indicate a more favourable quality of life

bCES-D = Center for epidemiological studies depression scale. Values are mean ± SD on a scale of 0–60. A score ≥16 indicated depression

cDepressive patients = score of ≥16 on the CES-D

dTwo way analysis of variance, adjusted for general practice

eMantel–Haenszel statistic, adjusted for general practice

Depression

Between baseline and 6 months of follow-up, we found a statistically significant difference (P = 0.01) between the intervention group, in which patients became less depressive, and the control group, in which patients’ level of depression increased (Table 4).

Covariates and subgroups

Age, gender and diabetes duration did not influence the results (data not shown). In addition, subgroups with either a high education, or more than 3 CBT sessions, or without depression did not have better improvements in all outcome measures than patients who did not fulfil any of these criteria.

Discussion

In the present study, we found that a cognitive behavioural treatment did not significantly improve CHD risk, when adding to managed care for T2DM patients. The amount of heavy physical activity increased significantly in the intervention group but only on the short term. We found no significant effects between the two groups on clinical characteristics and eating behaviour. Quality of life improved as a result of the intervention, and the level of depression decreased. The statistically significant effects all occurred between 0 and 6 months and disappeared between 6 and 12 months, indicating that the intervention was not effective on the long term.

This study was implemented in managed care. It is likely that the diabetes care had already provided adequate care, with the consequence of a ceiling effect that the CBT is not of additional value to decrease CVD risk. This is indicated by the low baseline HbA1c levels of 6.8 % in our study, compared to HbA1c values of about 8.5 % in other intervention studies (Steed et al., 2005; Gaede et al., 2003; Thoolen et al., 2007).

The strengths of our intervention were that (a) it was theoretically driven by the ASE-model; (b) the key element was PST which was carefully implemented and controlled by means of a training for the diabetes nurses and dieticians. We showed that general diabetes health professionals are able to acquire psychological skills and incorporate them into daily health care; (c) the use of a treatment manual, which encouraged diabetes nurses and dieticians to give a standardized intervention. Supervision meetings and tape recordings consolidated this (data not shown).

We found four other similar intervention studies, theoretically driven and based on techniques like problem solving and goal setting. These studies found small effects on clinical characteristics and often not sustainable on the long term, like we also found in our study. This might be due to the general acknowledged issue that behaviour change is difficult and people are tended to return to their usual habits and are not able to incorporate new behaviours in their daily life. However, these studies, as well as a recent systematic review on lifestyle interventions of Angermayr et al. (2010) concluded that they may be effective in reducing severe complications in patients with T2DM and therefore it is important to continue developing lifestyle interventions.

This study also had some methodological limitations that we have to address. Firstly, it is known that people have the tendency to overestimate physical activity. This might have happened in both groups but it is likely that patients in the intervention group, who are encouraged to change, overestimate physical activity more than patients in the control group. Secondly, we were not able to find a validated questionnaire to assess determinants of behaviour. A questionnaire was developed by using several questionnaires of colleagues in the field but we are not able to say that results were valid. Thirdly, we aimed for 97 patients in each group, but 76 and 78 patients were included in the intervention and control group, respectively. Fourthly, the mean number of sessions that patients in the intervention group attended was 3 (SD 1.7), indicating that not all patients attended the minimal intended number of 3 sessions, probably due to time restrictions of the patients to attend more sessions or difficulties in maintaining motivated. It is shown that more intensive interventions are successful in improving CHD risk factors (Alam et al., 2009). This might be a reason for the small effects that we found. However, a more intensive intervention would be even more difficult to incorporate in real diabetes care. Finally, the baseline CHD risk was relatively low. It is possible that targeting a higher risk group would have more effect. However, in that case not many patients would be eligible for this study which would limit the relevance of this study.

Conclusion

In conclusion, this study showed no overall effect on the risk of developing CHD, calculated with the UKPDS risk engine, although it showed improvements in heavy physical activity, quality of life and depression score at 6 months, but these disappeared by 12 months. We have provided an extensively described study design and we hope that the results of our study will not discourage other researchers to continue performing lifestyle interventions.

Acknowledgments

We would like to thank all diabetes nurses, dieticians, and research assistants that were involved in the study. We also would like to thank Tootje Hoovers and Jolanda Bosman for taking care of the organization of the study within the Diabetes Care System West-Friesland. In addition, we would like to thank Wendy Hardeman for her comments on the study design during the development of the intervention. The study was funded by the EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, the Netherlands.

Copyright information

© Springer Science+Business Media, LLC 2012