Stepped care for depression and anxiety in visually impaired older adults: multicentre randomised controlled trial

Study design

This study was a single masked international multicentre randomised controlled trial, exactly as described in the original protocol.18 Participants were individually randomised in the ratio 1:1 to one of two parallel groups—that is, to usual care or stepped care plus usual care.

Participants

Between July 2012 and April 2013, 3000 patients aged ≥50 from outpatient low vision rehabilitation organisations in the Netherlands and Belgium were contacted by letter and telephone and asked to participate. Of these, 914 provided written informed consent (response rate 30%). Participants were allowed to withdraw their consent for any reason at any time during the study. Baseline interviews with responders were performed to determine eligibility.

The organisations follow the World Health Organization criteria for eligibility, which are described in the Dutch guideline “Vision disorders, rehabilitation and referral.”19 This guideline dictates that all patients should have a decimal visual acuity of ≤0.3 and/or a visual field of ≤30° around the central point of fixation and/or an evident request for help for which options in regular ophthalmic practice are not adequate, such as contrast sensitivity or glare. Additional inclusion criteria were having subthreshold depression and/or anxiety—that is, a score of ≥8 on the hospital anxiety and depression scale-anxiety subscale (HADS-A)20 21 and/or ≥16 on the Centre for Epidemiologic Studies depression scale (CES-D)22 23; not meeting the diagnostic criteria of a major depressive, dysthymic, and/or anxiety disorder according to the DSM-IV (measured with the Mini International Neuropsychiatric Interview (MINI))24 25; being able to speak the Dutch language adequately; and not being severely cognitively impaired (measured with a six item screen, a short version of the mini-mental state examination, MMSE).26 Additional details on inclusion criteria and protocol design are described elsewhere.18

Patient involvement

Patients (n=8) from low vision rehabilitation organisations were involved in the development and implementation of the stepped care programme based on two focus group meetings in the Netherlands and Belgium. Patients were not involved in determining study conduct, recruitment, and design. The burden of the intervention and participation in the study in general was assessed by a panel of patient representatives, which was assigned by the funding agency. The burden of the intervention was not assessed as such by participating patients but satisfaction with the intervention was. Results of the study will be disseminated by letter to all participants by the end of 2015.

Randomisation and masking

Our prespecified power calculation was based on the study of van ‘t Veer and colleagues,27 who found the proportion of people developing a disorder to be 0.4 in the control group and 0.2 in the intervention group, with a relative risk of 0.5, leading to an effect size of 2*arcsine(√0.2)−2*arcsine(√0.4)=0.44. In addition, we used α≤0.05 (two sided), power 0.85, and dropout rate 20%, which showed that we needed a minimum of 230 patients (115 in each arm).18 As dropout rates observed at the start of the trial were higher than expected, we recruited more patients (n=265). Patients were assigned to usual care or the stepped care programme in addition to usual care. A computerised random number generator produced the allocation scheme. The scheme was based on blocks of two and stratified by 17 locations of three outpatient low vision rehabilitation organisations in the Netherlands and Belgium. An independent researcher carried out the randomisation after the baseline measurement. Patients were registered as being a participant of this study in their records at the rehabilitation centres. Only when guidance needed to be offered in step two or three of the programme were the clinical staff directly informed by the independent researcher as to which patient to treat.

Data were collected from September 2012 to July 2015, during which seven measurements took place (at baseline and at 3, 6, 9, 12, 18, and 24 months) by means of telephone interviews. These were performed at the VU University Medical Centre by masked research assistants, who were trained to diagnose depressive and anxiety disorders and follow a prespecified protocol. At the outset of the study and at the start of each telephone interview participants were told not to divulge the nature of their treatment during the telephone interviews. We checked if masking was maintained by asking research assistants to guess which treatment arm was offered. They were right in 38% of the cases, indicating that masking was effective. To minimise the possibility of data entry errors, the research assistants used specially designed data entry software (Blaise) to record all measurements. Because of the nature of the intervention, the participants and therapists could not be masked.

Interventions

The stepped care programme was based on a model similar to that previously used in the general elderly population and shown to be effective.27 28 The programme was altered and tailored to the needs of people with vision impairment based on a focus group with social workers and psychologists from the low vision rehabilitation organisations (n=12) and two focus groups with patient representatives (n=8). Specific attention was given to the difficulty of adjusting to vision loss and the physical and psychological consequences of this impairment (such as bereavement, fatigue, psychosocial adjustment) that could lead to feelings of depression and anxiety. Exercises and examples were altered and added based on direct input of patients and professionals. Specific attention was also given to the manner in which the programme was offered (for example, audio and Braille version of written documents). Additional information on programme development is provided elsewhere.18

The final programme contained four consecutive steps that took about three months each: watchful waiting, guided self help based on cognitive behavioural therapy, problem solving treatment, and referral to the general practitioner (GP) (box). All treatments were offered individually. Only when patients still had increased symptoms of depression and/or anxiety (score of ≥8 on the HADS-A and/or ≥16 on the CES-D) could they move on to the next step. A score below the cut-off point resulted in a (longer) period of watchful waiting until an increased score indicated the need for the next step of the programme. Therefore, not all patients of the stepped care group completed all steps of the intervention. Patients were seen at the rehabilitation centre or at home, based on the patient’s preference. Patients in both the stepped care and usual care group who developed a major depressive, dysthymic, and/or anxiety disorder, were directly referred to their GP to discuss further treatment. Usual care in both the treatment and control group included outpatient low vision rehabilitation care and/or care that was provided by other healthcare providers.

Outcome measures

The primary outcome measure of this study was the incidence of major depressive, dysthymic, and/or anxiety disorders (panic disorder, agoraphobia, social phobia, and/or general anxiety disorder) according to the DSM-IV, for which the Dutch MINI Plus (5.0.0), developed in clinician rated format (MINI-CR), was used at baseline and at 3, 6, 9, 12, 18, and 24 months in both groups. The MINI is a brief structured interview developed to diagnose psychiatric disorders according to DSM-IV criteria. It is considered a valid and reliable tool to define mental disorders based on a 20 minute telephone interview.24 25 The MINI shows moderate to high κ coefficients for all diagnoses, except for generalised anxiety disorder for which the κ is just below 0.5.24 25 Although a dysthymic disorder requires a depressed mood for over two years (not interrupted by more than two months at a time), it was included in the outcome measure because participants who were not diagnosed with a dysthymic disorder at one time point (for example, they had experienced a depressed mood only for the past 1.5 years) could be diagnosed with this disorder by the next time point. History of major depressive, dysthymic, and panic disorder at baseline was also determined with the MINI.

Secondary outcome measures were symptoms of depression and anxiety measured with the CES-D and HADS-A at baseline and at 3, 6, 9, 12, 18, and 24 months. The CES-D is a 20 item scale with a total score ranging from 0-60 and a cut-off score for subthreshold depression and/or anxiety of ≥16. It is a widely used scale and considered a valid and reliable instrument to measure symptoms of depression and anxiety in older adults.22 23 The HADS-A, also used to measure symptoms of anxiety, has seven items, with a total score ranging from 0-21 and a cut-off score for subthreshold anxiety of ≥8. Its reliability is reported to be “good to very good” in older adults.20 21 In addition, vision related quality of life was measured with the low vision quality of life questionnaire (LVQOL, with 21 questions on a 6 point Likert scale, measuring the disability experienced by patients in daily life),29 30 and adaptation to vision loss was measured with the adaptation to vision loss (AVL) scale (adapted from the AVL-12,31 with nine questions on a 4 point Likert scale, measuring intrapersonal and interpersonal acceptance of vision loss) at baseline and after 12 and 24 months. Psychometric properties of these questionnaires were investigated with item response theory models. We found no evidence of multidimensionality, local dependence, or differential item functioning, and all scales showed good fit to the model (that is, graded response model), except the HADS-A. Three items from the LVQOL were deleted to resolve local dependence, leading to the unidimensional LVQOL-18.

Health related quality of life was measured at baseline and after 12 and 24 months with the Euroqol-5 Dimensions (EQ-5D, which consists of five dimensions of functional impairment: mobility, self care, usual activities, pain/discomfort, and depression/anxiety).32 We used utility scores based on the Dutch tariff, where 1 denotes full health and 0 means a health state comparable with death (range −0.58 to 1, where negative utilities are valued as worse than death).32

For the process evaluation, firstly we measured compliance with treatment in step two and three of the programme based on the number of patients who rejected the intervention and the number and duration of appointments. Secondly, we reviewed therapist adherence to the problem solving treatment protocol based on audiotapes of a random selection of sessions (n=13). Thirdly, we determined adoption of the interventions based on therapists’ experiences, measured with two questions (“Are you satisfied with the results of the intervention?” and “Do you think the intervention suited the needs of the patient?”) and patients’ evaluation of the services, measured with the Dutch mental healthcare thermometer of satisfaction, a widely used 20 item questionnaire.33

We used the Trimbos/iMTA questionnaire for costs associated with psychiatric illness (TicP)34 to measure usual care at 6, 12, 18, and 24 months. This questionnaire measured self reported use of healthcare based on the number of contacts with a GP, company physician, medical specialist, physiotherapist or occupational therapist, social worker, psychologist or psychiatrist, alternative healer, homecare, guided group based peer support, admission to hospital, and use of drugs in the past six months.34 At baseline we used the perceived need for care questionnaire (PNCQ) to determine mental health services received in three months before the start of the study. This measured received information about mental illness and treatment possibilities, practical support, skills training, counselling/therapy, and drugs.35

Decimal visual acuity was retrieved from patient files at the low vision rehabilitation centres; missing values (n=22) were supplemented with estimates of visual acuity provided by self report based on recent ophthalmic diagnostics. To enable meaningful computations, these values were transformed to logMAR values (−log₁₀ visual acuity), where a visual acuity of 0.00-0.29 represents normal vision, 0.30-0.51 mild vision loss, and 0.52-2.00 low vision or blindness.

Patients were asked about comorbidity based on eight major condition groups: peripheral arterial disease, asthma or chronic obstructive pulmonary disease, diabetes mellitus, osteoarthritis and rheumatoid arthritis, cerebrovascular event or stroke, cardiac disease, cancer, and other chronic conditions. Compared with GP information, the accuracy of the self reports of these diseases was shown to be adequate and independent of cognitive impairment.36

Data analysis

We used SPSS for Windows version 20 (SPSS IBM, New York, USA) to perform an intention to treat analysis. Firstly, we tested differences in patients’ characteristics in the stepped care and usual care group, and in patients who dropped out and those who completed the follow-up period, for consistency based on independent sample t tests and χ² tests. Secondly, we determined the absolute and relative risk of developing a depressive and/or anxiety disorder in the usual care versus the stepped care plus usual care group and the number needed to treat as the inverse of the risk difference. Thirdly, we carried out a survival analysis based on a Kaplan-Meier curve, log rank test, and (adjusted) Cox proportional hazard regression analysis to compare differences between the stepped care and usual care groups in time to the onset of a depressive and/or anxiety disorder. We chose survival analysis because time played an important role in the present study as the programme aimed to delay or prevent the onset of a depressive and/or anxiety disorder. Fourthly, to investigate the effect of the intervention on the secondary outcomes, we performed linear mixed models using maximum likelihood estimation. Follow-up measurements of the secondary outcomes were adjusted for their baseline value. The intervention effect was defined as the interaction of treatment allocation (stepped care v usual care) by time (follow-up until 24 months).