- Email: [email protected]
Psychotherapy for depression in older veterans via telemedicine: a randomised, open-label, non-inferiority trial
Articles
Psychotherapy for depression in older veterans via telemedicine: a randomised, open-label, non-inferiority trial Leonard E Egede, Ron Acierno, Rebecca G Knapp, Carl Lejuez, Melba Hernandez-Tejada, Elizabeth H Payne, B Christopher Frueh
Summary Background Many older adults with major depression, particularly veterans, do not have access to evidence-based psychotherapy. Telemedicine could increase access to best-practice care for older adults facing barriers of mobility, stigma, and geographical isolation. We aimed to establish non-inferiority of behavioural activation therapy for major depression delivered via telemedicine to same-room care in largely male, older adult veterans. Methods In this randomised, controlled, open-label, non-inferiority trial, we recruited veterans (aged ≥58 years) meeting DSM-IV criteria for major depressive disorder from the Ralph H Johnson Veterans Affairs Medical Center and four associated community outpatient-based clinics in the USA. We excluded actively psychotic or demented people, those with both suicidal ideation and clear intent, and those with substance dependence. The study coordinator randomly assigned participants (1:1; block size 2–6; stratified by race; computer-generated randomisation sequence by RGK) to eight sessions of behavioural activation for depression either via telemedicine or in the same room. The primary outcome was treatment response according to the Geriatric Depression Scale (GDS) and Beck Depression Inventory (BDI; defined as a 50% reduction in symptoms from baseline at 12 months), and Structured Clinical Interview for DSM-IV, clinician version (defined as no longer being diagnosed with major depressive disorder at 12 months follow-up), in the per-protocol population (those who completed at least four treatment sessions and for whom all outcome measurements were done). Those assessing outcomes were masked. The non-inferiority margin was 15%. This trial is registered with ClinicalTrials.gov, number NCT00324701. Findings Between April 1, 2007, and July 31, 2011, we screened 780 patients, and the study coordinator randomly assigned participants to either telemedicine (120 [50%]) or same-room treatment (121 [50%]). We included 100 (83%) patients in the per-protocol analysis in the telemedicine group and 104 (86%) in the same-room group. Treatment response according to GDS did not differ significantly between the telemedicine (22 [22·45%, 90% CI 15·52–29·38] patients) and same-room (21 [20·39%, 90% CI 13·86–26·92]) groups, with an absolute difference of 2·06% (90% CI –7·46 to 11·58). Response according to BDI also did not differ significantly (telemedicine 19 [24·05%, 90% CI 16·14–31·96] patients; same room 19 [23·17%, 90% CI 15·51–30·83]), with an absolute difference of 0·88% (90% CI –10·13 to 11·89). Response on the Structured Clinical Interview for DSM-IV, clinician version, also did not differ significantly (39 [43·33%, 90% CI 34·74–51·93] patients in the telemedicine group and 46 [48·42%, 90% CI 39·99–56·85] in the same-room group), with a difference of –5·09% (–17·13 to 6·95; p=0·487). Results from the intention-to-treat population were similar. MEM analyses showed that no significant differences existed between treatment trajectories over time for BDI and GDS. The criteria for non-inferiority were met. We did not note any adverse events. Interpretation Telemedicine-delivered psychotherapy for older adults with major depression is not inferior to sameroom treatment. This finding shows that evidence-based psychotherapy can be delivered, without modification, via home-based telemedicine, and that this method can be used to overcome barriers to care associated with distance from and difficulty with attendance at in-person sessions in older adults.
Lancet Psychiatry 2015; 2: 693–701 Published Online July 17, 2015 http://dx.doi.org/10.1016/ S2215-0366(15)00122-4 See Comment page 668 Health Equity and Rural Outreach Innovation Center, Ralph H Johnson Veterans Affairs Medical Center, Charleston, SC, USA (Prof L E Egede MD, Prof R Acierno PhD, Prof R G Knapp PhD, Prof C Lejuez PhD, E H Payne MS, Prof B C Frueh PhD); Center for Health Disparities Research, Division of General Internal Medicine (Prof L E Egede, M Hernandez-Tejada DHA), College of Nursing (Prof R Acierno, M Hernandez-Tejada), and Department of Public Health Sciences (Prof R G Knapp, E H Payne), Medical University of South Carolina, Charleston, SC, USA; Department of Psychology, University of Maryland, College Park, MD, USA (Prof C Lejuez); and Department of Psychology, University of Hawaii, Hilo, HI, USA, and The Menninger Clinic, Houston, TX, USA (Prof B C Frueh) Correspondence to: Prof Leonard E Egede, Center for Health Disparities Research, Medical University of South Carolina, Charleston, SC 29425-0593, USA [email protected]
Funding US Department of Veterans Affairs.
Introduction Depression is a serious and debilitating psychological disorder, with a lifetime prevalence of about 10% in the USA for all age groups.1 Prevalence is slightly lower for older adults (defined as ages 60 years and older)2 in the USA (ages 45–64 years 10%; ages 65 years and older 7%); nonetheless, about 6·5 million people older than 65 years have the disorder in the USA.3,4 Furthermore, 20% of older US adults have substantial symptoms of depression that do not meet DSM diagnostic criteria, but might still warrant treatment5,6 because untreated www.thelancet.com/psychiatry Vol 2 August 2015
minor depression is a risk factor for development of the full clinical disorder.7 Depression is particularly problematic for veterans, with substantial depressive symptoms 2–5 times more likely than in their civilian counterparts.8 Depression in old versus young adults seems to be represented by a somewhat different clinical presentation, characterised by increased physical complaints, fatigue, and apathy, and decreased attention and concentration, but fewer reported symptoms of sadness in older adults.9 Causes of these age-based differences might include neurological changes 693
Articles
associated with ageing, effects of acute disease states, effects of chronic disease states such as atherosclerosis10–14 or cardiovascular complications,15,16 and individual predisposition (ie, late onset of depression).17 Age-specific psychosocial factors include increasingly unstable psychosocial environments associated with old age, resulting from death or incapacitation of those giving social support (ie, friends and family die or become too ill to serve in social support roles), reduced financial freedom, and frequent bereavement.9,18 Depression is common in suicidal individuals, and suicide rates in older adults are disproportionally high compared with the rest of the population,19 particularly with respect to older veterans.20 Despite its wide prevalence and potentially increasing lethality in old age, depression is generally poorly detected and hence undertreated.21,22 Effective treatment of depression in older adults includes some specific forms of psychotherapy and pharmacotherapy (most notably the selective serotonin reuptake inhibitors).23 Cognitive behavioural therapies are the most recommended forms of psychotherapy for depression because of their simplicity and costeffectiveness.24 If collaborative care (interprofessional intervention) models are used, reduced costs are often evident secondary to reduced doctor visits and generally improved functioning.25 However, because most geriatric depression is diagnosed and treated in the primary care setting by non-mental-health specialists, pharmacological treatment is the most widely used, although it is slightly less effective than psychotherapy.26,27 Barriers to psychotherapy for elderly people include mobility issues, stigma concerns, and geographical isolation (eg, living in rural areas). Telemedicine service delivery strategies, such as videoteleconferencing, offer a medium to address these concerns and increase access to evidence-based care.28–30 Findings from studies testing use of telemedicine to deliver psychotherapy show that these methods are non-inferior to those obtained by traditional same-room care in treatment of post-traumatic stress disorder,31–35 and are cost efficient.36 Other disorders, including depression, have received substantially less attention. Thus, we designed this study to assess the efficacy of telemedicine in delivery of psychotherapy for depression to older adult veterans in their homes.
Methods Study design and participants In this randomised, controlled, open-label, noninferiority trial, we recruited participants from the Ralph H Johnson Veterans Affairs Medical Center (Charleston, SC, USA) and four associated community outpatientbased clinics (Goose Creek, Beaufort, and Myrtle Beach, SC, and Savannah, GA, USA). Male and female veterans (aged 60 years or older) meeting DSM-IV37 criteria for major depressive disorder were eligible. However, after we started recruitment, veterans from the Vietnam War 694
era already enrolled in the study requested that we lower the age limit to accommodate more veterans from that war era, so we lowered the age eligibility to 58 years to accommodate their request on Nov 1, 2007. We excluded actively psychotic or demented people, those with both suicidal ideation and clear intent, those meeting criteria for substance dependence, and those unable to provide informed consent from participation; however, to increase generalisation of results, presence of other forms of psychopathology (eg, anxiety disorders) were not cause for exclusion. The Veterans Affairs study site (Ralph H Johnson Veterans Affairs Medical Center, SC, USA) maintains a mailing list of all patients other than those who have opted out of this class of mailings, which is available for clinical research recruitment. We mailed postcards from the centre inviting Veterans Affairs patients aged 58 years and older (after the age limit was lowered from 60 years)2 inviting them to contact study personnel if they felt that they were sad or depressed and might be interested in a trial assessing telemedicine. We offered all participants who contacted study staff standard clinic-based assessment and services for depression as an alternative to the study. All study procedures were approved by the Medical University Institutional Review Board and Veterans Affairs Research and Development Committee. All patients gave written informed consent to participate in the study. The study design and methods have been published previously.38
Randomisation and masking The study coordinator verified all eligibility criteria before randomisation. The study coordinator randomly assigned participants (1:1) to one of two study groups delivering behavioural activation for depression—telemedicine and same-room treatment. The randomisation sequence was computer-generated by the senior biostatistician using a permuted-block randomisation scheme, stratified by race. We varied block size to minimise the chance that masking of assignment would be broken (block size 2–6). Patient allocation was provided in individual sealed envelopes to the study coordinator. Once study eligibility and informed consent for a patient were verified, the study coordinator opened the sealed randomisation envelope to ascertain the random treatment assignment for that patient. Once a randomisation assignment was made, the study coordinator entered that patient into the study. Those doing endpoint psychiatric interviews at baseline and 12 months, and those assessing outcomes, were masked to treatment assignment.
Procedures We did not collect data on antidepressant use because medication needed to be stabilised before randomisation. We maintained psychiatric medications at present doses for 3 months at the discretion of the treatment physician. We asked patients meeting major depressive disorder criteria to www.thelancet.com/psychiatry Vol 2 August 2015
Articles
maintain psychiatric medications at present doses when medically possible. Participants who had not initiated new prescription medications in the 4 weeks before enrolment into the study initiated treatment 1 week after randomisation and completion of the baseline assessment. However, potential participants who had recently begun trials of antidepressants or any other prescription medications needed to wait 4 weeks after assessment to ensure medication stabilisation before starting study treatment. All patients received the same individual mental health intervention—behavioural activation for depression—for 8 weeks according to a previously published manual.39,40 We did not modify treatment for the telemedicine group. Behavioural activation is based on the idea that what a patient does plays a part in how he or she feels.40 To increase the likelihood that reinforcing behaviours will be implemented and logistical obstacles to such behaviours overcome, we used daily planners and valued activity lists to schedule positively (eg, attendance at a grandson’s football game) or negatively (eg, completion of a necessary chore resulting in spousal praise) reinforcing behaviours that are of value to the patient. The planner allows obstacles to successful completion of these behaviours to be addressed in advance so that the likelihood that reinforcement is received is maximised. Patients had 60 min sessions about once a week. The intervention group received therapy via videophone, whereas the control group received face-to-face therapy. We did treatment sessions for the telemedicine group using in-home videoconferencing technology. So as not to need participants to have broadband internet access, we used an analogue videophone (KMEA TV500SP; KMEA, CA, USA) that operates via the standard telephone service. Apart from the video screen, this equipment looks and functions much like a basic touchtone telephone. It is a plug-and-play product, with built-in camera, full duplex speakerphone, 4” LCD colour screen (270k pixels), realtime motion display (18 frames per s), and oversized touchtone buttons for easy use by all patients. Therapists were masters-level counsellors with at least 5 years clinical experience. Each received training during a 2 day workshop from CL and participated in subsequent 1 h supervision meetings once a week with RA. We randomly audited (with a computer-generated random list) 20% of session audiotapes for treatment fidelity (adherence to the manualised behavioural activation treatment, ascertained with a checklist). We used previously validated self-report measures to assess depression at baseline, mid-treatment (4 weeks), post-treatment (8 weeks), and 3 months and 12 months of follow-up. We used the Geriatric Depression Scale (GDS)41 and Beck Depression Inventory (BDI)42 to assess depression, both self-administered in the presence of the interviewer. The GDS is a 30 item measure, and we used the generally accepted cutoff score of 11. The BDI is a 21 item self-report scale. We used the Short Portable Mental Status Questionnaire43 to screen for cognitive impairment www.thelancet.com/psychiatry Vol 2 August 2015
(cutoff >7). We administered endpoint psychiatric interviews (ie, Structured Clinical Interview for DSM-IV, clinician version [SCID])44 at baseline and 12 months. We assessed major depressive disorder and other psychopathologies with this structured clinical interview on the basis of the DSM-IV. We specified onset of major depressive disorder. Interviewers were masters-level counsellors trained to 90% agreement in rating scores on the fidelity checklist before doing any study assessments. We audiotaped interviews, and 20% of them were randomly selected (with a computer-generated random list) and rated by an independent masters-level counsellor to assess inter-rater reliability on an item-by-item basis.
Outcomes The primary clinical outcome measure was the proportion of patients who responded to treatment at the end of the 12 months of follow-up. We established this treatment responder status with each of the dependent measures of depression separately (GDS, BDI, and SCID). For the BDI and GDS, treatment responders were those who showed improvement of at least 50% from baseline on the total score ([baseline – post]/baseline ≥50%), as used in similar studies of treatments for depression.45 For the SCID, treatment responders were those who were no longer diagnosed with major depressive disorder at 12 months follow-up. Secondary efficacy outcomes were
780 patients assessed for eligibility
403 did not give consent 131 ineligible 5 did not return for baseline assessment
241 randomised
120 assigned telemedicine
121 assigned same room
10 discontinued treatment before week 4 assessment 8 moved 2 died 2 discontinued treatment before week 8 assessment 2 moved 3 discontinued treatment before 3 month assessment 3 died 5 discontinued treatment before 12 month assessment 5 moved
100 included in per-protocol analysis
14 discontinued treatment before week 4 assessment 12 moved 2 died 2 discontinued treatment before 3 month assessment 2 died 1 lost to follow-up 1 moved
104 included in per-protocol analysis
Figure 1: Trial profile
695
Articles
continuous BDI and GDS total scores over the longitudinal time trajectory, and response status at week 4, week 8, and 3 months.
Statistical analysis The power and sample size calculation showed that about 100 participants per group were needed to detect noninferiority of telemedicine to same-room methods of intervention delivery (power 85%; non-inferiority parameter Δ = 0·15; assumed response proportion for sameroom group 70%) with the per-protocol sample.46 For intention-to-treat analyses, we inflated the sample size by 10% to account for the dilution effect of intention to treat to yield a sample size of 112 patients in each group. We did descriptive analyses (mean and SD) for continuous variables and frequency distributions for categorical variables to describe demographic and baseline clinical characteristics for the total study sample and each treatment group. We carried out the primary analysis using the per-protocol sample, which we defined Total (n=241) Age (years)
Telemedicine (n=120)
Same room (n=121)
63·9 (5·1)
63·5 (4·4)
64·2 (5·6)
235 (98%)
116 (97%)
119 (98%)
White
143 (60%)
68 (58%)
75 (63%)
Black
94 (40%)
49 (42%)
45 (38%)
Married
165 (69%)
83 (70%)
82 (68%)
13·7 (2·6)
13·5 (2·3)
13·8 (2·9)
Male sex Race
Education (years) Insurance coverage Private
67 (28%)
32 (27%)
35 (29%)
Medicaid or Medicare
72 (30%)
41 (34%)
31 (26%)
Private and Medicaid or Medicare
22 (9%)
9 (8%)
13 (11%)
VA only
80 (33%)
38 (32%)
42 (35%)
Employed
50 (21%)
23 (20%)
27 (23%)
Working hours per week if employed
30·3 (14·8)
29·3 (16·1)
31·0 (14·1) 24 (20%)
Income
49 (21%)
25 (21%)
US$15 000 to US$24 999
56 (24%)
24 (20%)
32 (27%)
US$25 000 to
93 (39%)
49 (42%)
44 (37%)
≥$50 000
40 (17%)
20 (17%)
20 (17%)
Health status Better than last year About the same Worse than last year
39 (16%)
20 (17%)
19 (16%)
104 (44%)
61 (51%)
43 (36%)
96 (40%)
38 (32%)
58 (48%)
Smoker Present
49 (21%)
25 (21%)
24 (20%)
Former
132 (56%)
65 (55%)
67 (57%)
Never
56 (24%)
29 (24%)
27 (23%)
1·2 (1·2)
1·1 (1·1)
1·3 (1·2)
Days with at least 20 min moderate or vigorous activities per week Service-connected medical care
45·1 (40·4)
47·8 (39·7)
42·5 (41·1)
Years being a VA patient
16·2 (12·2)
16·1 (12·0)
16·3 (12·4)
(Table 1 continues on next page)
696
as participants who completed at least four treatment sessions and for whom we made all outcome measurements. We decided the cutoff on the basis of the expected minimum number of sessions that would yield a meaningful effect with behavioural activation treatment. We repeated analyses using the intention-to-treat sample, consisting of all randomly allocated participants. We established effect sizes as the difference in response proportions for telemedicine versus same-room methods of delivery, with a positive value for difference in response proportions showing a superior response proportion for telemedicine. We used 90% CIs,47 judging telemedicine to be non-inferior to same-room delivery as long as the response proportion for same-room does not exceed that of telemedicine by more than a clinically meaningful amount, specified by the non-inferiority margin Δ. We set Δ for the primary outcome at an absolute difference in proportions of 15% (Δ = 0·15).38 We used multiple imputation for imputing missing data for the continuous outcome variables to complete the analysis sets at each timepoint using SAS procedures MI (Markov Chain Monte Carlo method) and MIANALYZE (SAS version 9.3, NC, USA). We used age, sex, race and ethnic origin, education, marital status, employment, income, insurance type, and disability status in the imputation models for BDI and GDS. We assumed missing at random. We established percentage change from baseline using the completed (imputed) dataset at 12 months, and we created the primary dichotomous response outcome on the basis of whether the percentage change from baseline was greater than or equal to 50% (responders) or less than 50% (non-responders). We repeated the procedure for establishment of response proportion for the secondary timepoints: 4 weeks, 8 weeks, and 3 months. We used a mixed-effects longitudinal modelling (MEM) approach to compare the longitudinal profile of response for telemedicine with same room. MEM allows for missing data under the assumption that data are missing at random and it is robust to a range of different distributions of the outcome variable. The dependent dichotomous outcome (response status) was based separately on BDI and GDS. The model included treatment, visit, visit-by-treatment (interaction term) as primary (fixed) independent variables. A significant interaction term for these models would provide evidence against a conclusion of non-inferiority of telepsychiatry. We also used the MEM approach, treating BDI and GDS as continuous outcomes, to estimate baseline-adjusted treatment means across the time periods. In a final set of exploratory analyses, we examined differential response to treatment by sex, race, income, and age (putative moderator variables). For these analyses, we added putative moderator variables to the MEM model as treatment-by-moderator interaction terms, and significance of this term shows that treatment response differs by moderator variable. For MEM www.thelancet.com/psychiatry Vol 2 August 2015
Articles
analyses, we used SAS PROC MIXED and PROC GLIMMIX (SAS version 9.3). We report analyses for each measure of treatment response separately, but based on our analysis plan. When results based on each of these analyses do not converge, overall conclusions on treatment response are based on the two measures that concur. Thus, the indicator of treatment response is in terms of each measure and the converging data from at least two different measures. Both BDI and GDS have to support non-inferiority for that conclusion to be drawn, shown by both the per-protocol and intention-to-treat analyses.47 This trial is registered with ClinicalTrials.gov, number NCT00324701.
Total (n=241)
Telemedicine (n=120)
Same room (n=121)
In Vietnam War
209 (87%)
107 (89%)
102 (84%)
Disabled
167 (70%)
82 (69%)
85 (70%)
4·2 (3·4)
4·3 (3·6)
4·2 (3·3)
(Continued from previous page)
Charlson comorbidity score Mental status questionnaire score 0 error, normal mental function
162 (67%)
80 (67%)
82 (68%)
1 error
64 (27%)
34 (28%)
30 (25%)
≥2 errors
15 (6%)
6 (5%)
9 (7%)
Baseline depression severity GDS
20·8 (4·8)
20·9 (4·8)
20·6 (4·8)
BDI
26·8 (10·0)
26·7 (9·8)
26·8 (10·3)
Generalised anxiety disorder
99 (42%)
50 (42%)
49 (42%)
Panic disorder
19 (8%)
9 (8%)
10 (9%)
Alcohol misuse
59 (25%)
30 (25%)
29 (24%)
Alcohol dependence
29 (12%)
13 (11%)
16 (13%)
Cannabis misuse
18 (8%)
6 (5%)
12 (10%)
6 (3%)
4 (3%)
2 (2%)
147 (63%)
75 (65%)
72 (62%)
Symptomatic diagnosis of panic disorder in the past month
12 (17%)
6 (17%)
6 (18%)
Symptomatic diagnosis of PTSD in the past month
143 (79%)
76 (83%)
67 (76%)
Psychiatric comorbidity
Role of the funding source
Lifetime prevalence of:
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. LEE and RGK had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Cannabis dependence
Results Between April 1, 2007, and July 31, 2011, 377 (48%) of the 780 patients who were screened consented to participate in the study (figure 1). Of these, we randomly allocated 241 (64%) to either the telemedicine (120 [50%]) or sameroom (121 [50%]) treatment groups (intention-to-treat sample). In the telemedicine group, 100 (83%) returned for the final assessment and 104 (86%) of same-room participants returned (per-protocol sample). All study therapists achieved a more than 90% protocolspecified fidelity. All ratings exceeded 90% agreement for study assessments. At the week 4 assessment point for the telemedicine group, 8·3% of the data were missing and imputed, at week 8, 10·0% were imputed, at month 3, 12·5% were imputed, and at month 12, 16·7% were imputed. In the same-room group, 11·6% of data were missing at week 4, 11·6% were missing at week 8, 13·2% were missing at month 3, and 14·0% were missing at month 12. We did not find any systematic differences in the missing and not missing outcome value (BDI and GDS at 12 months) between treatment groups in measured demographic and baseline clinical data; the percentage missing GDS 12 month outcome in the per-protocol sample was very small (1·47%), and using GDS as a surrogate for BDI, we found no differences in mean GDS at 12 months for the missing and not missing BDI between treatment groups (appendix). Table 1 shows the characteristics of the intention-to-treat sample at baseline. We did not note any adverse events for any participant in the study. A high proportion of participants were rural residents (170 [71%]). Average session attendance was high. 192 (80%) participants completed all eight sessions (97 [81%] in telemedicine vs 95 [79%] in same room); 35 (15%) participants completed www.thelancet.com/psychiatry Vol 2 August 2015
PTSD
GAF scale* ≤50
4 (2%)
2 (2%)
2 (2%)
51–60
103 (50%)
51 (48%)
52 (52%)
61–75
98 (48%)
51 (49%)
47 (47%)
Data are mean (SD) or n (%). Information was not available for every patient in most categories. VA=Veterans Affairs. GDS=Geriatric Depression Scale. BDI=Beck Depression Inventory. PTSD=post-traumatic stress disorder. GAF=Global Assessment of Functioning.
Table 1: Demographic and baseline characteristics (intention to treat)
Telemedicine
Same room
Difference
1·05% (–8·30 to 10·41)
Intention to treat BDI
27 (22·54% [15·40 to 29·69])
26 (21·49% [14·72 to 28·25])
GDS
25 (20·96% [14·45 to 27·47])
23 (19·30% [13·29 to 25·31])
1·66% (–7·20 to 10·52)
SCID
53 (44·17% [35·78 to 52·55])
58 (47·85% [39·63 to 56·07])
–3·68% (–15·53 to 8·16) 0·88% (–10·13 to 11·89)
Per protocol BDI
19 (24·05% [16·14 to 31·96])
19 (23·17% [15·51 to 30·83])
GDS
22 (22·45% [15·52 to 29·38])
21 (20·39% [13·86 to 26·92])
2·06% (–7·46 to 11·58)
SCID
39 (43·33% [34·74 to 51·93])
46 (48·42% [39·99 to 56·85])
–5·09% (–17·13 to 6·95)
Data are n (% [90% CI]) or % (90% CI). BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Table 2: Treatment response at 12 months (primary outcomes)
fewer than eight sessions (19 [16%] in telemedicine vs 16 [13%] in same room); and 14 (6%) participants did not attend any session (five [4%] in telemedicine vs nine [7%] in same room). Detailed information about session attendance by treatment group is provided in the appendix. Table 2 shows the treatment response proportions for both the BDI and GDS outcomes at the 12 month
See Online for appendix
697
Articles
Telemedicine
Same room
A
Difference
27
4 weeks Intention to treat 10 (8·50% [3·89 to 13·11])
22 (18·47% [11·86 to 25·08])
–9·97% (–18·06 to –1·88)
GDS
6 (4·63% [1·36 to 7·89])
16 (13·14% [7·87 to 18·41])
–8·52% (–14·70 to –2·34)
BDI
7 (8·86% [3·60 to 14·12])
15 (22·06% [13·79 to 30·33])
–13·20% (–23·00 to –3·40)
GDS
5 (5·15% [1·46 to 8·85])
15 (15·46% [9·43 to 21·50])
–10·31% (–17·39 to –3·23)
BDI score
25
BDI Per protocol
23 21 19
8 weeks
0
Intention to treat BDI
25 (20·83% [14·39 to 27·28])
34 (28·22% [20·79 to 35·66])
–7·39% (–17·34 to 2·56)
GDS
21 (17·67% [11·72 to 23·62])
31 (25·45% [18·30 to 32·61])
–7·79% (–17·01 to 1·43)
B 21 20
Per protocol 19 (25·00% [16·83 to 33·17])
23 (33·82% [24·39 to 43·26])
–8·82% (–21·31 to 3·66)
GDS
18 (19·57% [12·76 to 26·37])
26 (27·66% [20·07 to 35·25])
–8·09%(–18·29 to 2·10)
Intention to treat BDI
23 (19·38% [12·46 to 26·29])
28 (23·10% [15·93 to 30·27])
–3·72% (–13·86 to 6·41)
GDS
19 (15·63% [9·95 to 21·30])
23 (18·64% [12·27 to 25·01])
–3·01% (–11·55 to 5·52)
BDI
15 (22·39% [14·01 to 30·76])
19 (26·76% [18·12 to 35·40])
–4·37% (–16·41 to 7·66)
GDS
15 (17·05% [10·45 to 23·64])
19 (20·65% [13·71 to 27·59])
–3·61% (–13·18 to 5·97)
Per protocol
19 GDS score
BDI
3 months
18 17 16 15 0
Data are n (% [90% CI]) or % (90% CI). BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Table 3: Treatment response over 3 months (secondary outcomes)
follow-up point for the per-protocol and intention-to-treat samples (primary outcome). 22 (22·45% [90% CI 15·52–29·38]) patients in the telemedicine group had at least a 50% reduction in symptom severity and were classified as treatment responders at the primary assessment point (12 months follow-up) compared with 21 (20·39% [90% CI 13·86–26·92]) of those in the sameroom group according to GDS in the per-protocol population. Results with BDI (primary outcome) showed a similar pattern, with 19 (24·05% [90% CI 16·14–31·96]) patients classified as responders in the telemedicine group compared with 19 (23·17% [90% CI 15·51–30·83]) in the same-room group. Estimated differences in response proportions were 0·88% (90% CI –10·13 to 11·89) for BDI and 2·06% (90% CI –7·46 to 11·58) for GDS. The CI lower limits for response proportions are within the allowable non-inferiority limit (ie, do not exceed –0·15), showing that telemedicine was noninferior to same-room therapy at the 12 month timepoint. Treatment response on the SCID (primary outcome) at 12 months was 39 (43·33% [90% CI 34·74–51·93]) patients in the telemedicine group and 46 (48·42% [90% CI 39·99–56·85]) in the same-room group, with a difference of –5·09% (90% CI –17·13 to 6·95), which was not significant (p=0·487). Treatment response at 4 weeks, 8 weeks, and 3 months (secondary outcomes) are shown in table 3. At week 8, non-inferiority of telemedicine could not be established because the lower limit of the 90% CI was not less than 698
Actual telemedicine Actual same room Model-derived telemedicine Model-derived same room
s 12
m
on
th
W ee k4 W ee 3m k8 on th s
0
Time
Figure 2: Actual and model-derived mean (A) BDI and (B) GDS scores over time BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Δ and superiority of same-room therapy could also not be established because 0 is contained in the CI. The trend favouring same-room therapy seemed to diminish with time, with telemedicine having a small advantage in response proportion at 12 months, with a conclusion of non-inferiority of telemedicine reached at both the 3 month and 12 month assessment points. The longitudinal trajectories of actual and modelderived BDI and GDS means are shown in figure 2. Both groups improve over time during the treatment phase, with same-room treatment (MEM model: BDI slope –3·90, p<0·0001; GDS slope –2·66, p<0·0001) improving at a slightly faster rate from baseline to week 8 than does telemedicine (BDI slope –3·07, p<0·0001; GDS slope –2·20, p<0·0001), then both get slightly worse at 3 months. This slight worsening is more pronounced in the telemedicine (BDI slope 2·48, p=0·1914; GDS slope 1·65, p=0·1469) than in the same-room (BDI slope 0·94, p=0·6309; GDS slope 0·78, p=0·5056) group, but neither are significant. The telemedicine group then improves between 3 months and 12 months (BDI slope –1·37, p=0·4818; GDS slope –1·19, p=0·3156), whereas the same-room group gets marginally worse (BDI slope 0·36, p=0·8523; GDS slope 0·36, p=0·7486), but neither are significant. MEM analyses showed that no significant differences existed between treatment trajectories over time for BDI and GDS for either the dichotomous response (interaction term for BDI: F value=1·52; p=0·2107; interaction term for www.thelancet.com/psychiatry Vol 2 August 2015
Articles
GDS: F value=2·36; p=0·0729) or continuous (interaction term for BDI: F value=0·76; p=0·5169; interaction term for GDS: F value=0·97; p=0·4058) outcome. In subgroup analyses exploring the effects of putative moderator variables on treatment response, we noted no significant moderator-by-treatment interaction terms for sex, race, income, and age.
Discussion To our knowledge, this is the first randomised controlled trial of manualised evidence-based psychotherapy for depression in older adults via telemedicine (panel). We have shown that this method is feasible and produces outcomes that are no worse than in-person delivery 12 months after treatment. Participants in both groups tolerated and clinically benefitted from behavioural activation for depression. The magnitude of treatment effect noted in this study is similar to what has been noted in treatment studies for depression in veterans.48 Although the treatment response that we noted during the first 8 weeks was significant for both groups, others using behavioural activation with veterans have not noted significant improvement in depression scores.49 Study retention was excellent, well within the range consistently noted in psychotherapy trials for depression (65–85%),50 and session attendance was high. Assessment of longitudinal trajectories of BDI and GDS shows a slight worsening at 3 months in both groups, which coincides with the end of treatment sessions. The reason for improvement in the telemedicine group between 3 months and 12 months is unclear, so these findings will need to be assessed in future studies. This study has several important aspects. First, it is one of only a handful of methodologically rigorous, noninferiority-designed, randomised controlled trials of psychotherapy telemedicine interventions with any population,30 and the first involving major depression in older adults, a population that often faces barriers of access to care. Second, study implementation was rigorously controlled, including careful a-priori noninferiority analyses and sample size calculations, careful therapist fidelity monitoring, high study retention and session attendance, follow-up assessments up to 12 months, and examination of clinical outcomes in a difficult-to-treat and understudied clinical sample. Third, the study had broad inclusion criteria that allowed for high psychiatric comorbidity, and thus participants are reasonably representative of the patient population of interest. Finally, this study includes a large sample size and high proportions of rural residents (about 70%) and African Americans (about 40%). Along with its strengths, this study has limitations. First, we conservatively excluded participants with acute safety concerns (homicidal or suicidal), present substance dependence, and active psychosis or dementia. Second, clinical benefit was less than was anticipated www.thelancet.com/psychiatry Vol 2 August 2015
Panel: Research in context Systematic review We searched PubMed for articles published in English from Jan 1, 2000, to March 31, 2014, with the following search terms: “depression”, “veteran”, “elderly”, “psychotherapy”, “behavioral activation”, “telemedicine”, “telehealth”, “rural”, and “access”. We found little evidence of effectiveness of evidence-based treatments for depression or any other disorder delivered via office-based treatment versus home-based telemedicine in older patients. Findings from studies testing the use of home-based telemedicine to deliver psychotherapy show that relevant clinical outcomes are non-inferior to those obtained by traditional same-room care in treatment of post-traumatic stress disorder31–35 and that the method is cost effective.36 Other mental health disorders, including depression, have received substantially less attention than post-traumatic stress disorder has. Interpretation To our knowledge, this trial is one of the first demonstrations that evidence-based psychotherapy can be delivered, without modification, via home-based telemedicine. Our findings show that home-based telemedicine can be used to overcome barriers to care associated with distance from and difficulty with attendance at in-person sessions. The implications of this study are that resources should be devoted to offering of services directly into patients’ homes via telemedicine methods, with specific safety procedures in place, such as having the direct telephone number of the police department in which the home-based patient resides in case an emergency dispatch is needed.
when the study was designed. The effect on realised power is negligible in that sample size calculation with proportions is symmetrical, with differences in proportions of 0·7 needing similar power to differences in proportions of 0·3 (with differences in proportions of 0·2 yielding slightly higher power than 0·3). Recruitment exceeded that projected, resulting in slightly higher power for detection of between-group differences than was planned (with corresponding reduction in the falsenegative possibility for the comparison). We are uncertain why treatment response was lower than was expected. Our population was probably more sick and had more severe depression and psychiatric comorbidity than those in the studies that we based our estimates on. Third, the information technology used in the study is now somewhat obsolete; however, new technologies should only improve communication between therapists and patients. Fourth, we included very few women in the study sample, and generalisation of results from this subgroup is not guaranteed. Finally, although the focus of this study was on behavioural treatment of depression, some of the patients were also taking antidepressant medications during the study, and we did not specifically track their effects. Potential participants who had recently begun an antidepressant or other prescription medication needed to wait 4 weeks after assessment to ensure medication stabilisation before starting study treatment. However, some antidepressants take longer than 4 weeks to reach maximum effect, particularly in older people, which might have affected study findings. Nevertheless, we expect incremental treatment effects of medications would be the same between groups. 699
Articles
We have shown that telemedicine is a viable and effective means of delivery of evidence-based psychotherapy for depression to older adults. Future research should rigorously assess effectiveness of telemedicine in other clinical settings and with other information technologies. For example, evidence exists that landline telephones,51 mobile phones,52 and web-based53 interventions can be effectively used to treat mild-to-moderate depression and might be useful alternatives or adjuncts to videoteleconferencing, depending on patient preferences and access. Finally, implementation research is needed into how to most effectively disseminate telemedicine for populations with depression and integrate it with existing models of care of other disorders that elderly adults face. Contributors LEE, BCF, and RGK drafted the report. RA edited the initial draft. LEE, BCF, RGK, CL, and RA reviewed the report. LEE, BCF, RGK, CL, and RA edited the final version. MH-T drafted the literature review. EHP did data analyses and created the tables and figures. Declaration of interests CL is one of the developers of the manualised Behavioural Activation for Depression protocol used in this study. All other authors declare no competing interests. Acknowledgments This work was supported by grants IIR-04-421-3 from the Veterans Affairs Health Services Research and Development Program. We are deeply appreciative of the veterans and Veterans Affairs primary care and mental health providers who contributed to this research effort. All views and opinions expressed herein are those of the authors and do not necessarily reflect those of our respective institutions or the Department of Veterans Affairs. References 1 Centers for Disease Control and Prevention (CDC). Current depression among adults—United States, 2006 and 2008. MMWR Morb Mortal Wkly Rep 2010; 59: 1229–35. 2 WHO. Definition of an older or elderly person. http://www.who. int/healthinfo/survey/ageingdefnolder/en (accessed June 25, 2015). 3 Steffens DC, Skoog I, Norton MC, et al. Prevalence of depression and its treatment in an elderly population: the Cache County study. Arch Gen Psychiatry 2000; 57: 601–07. 4 National Alliance on Mental Illness. Depression in older persons. Oct, 2009. http://www2.nami.org/Content/NavigationMenu/ Mental_Illnesses/Depression/Depression_Older_Persons_ FactSheet_2009.pdf (accessed July 25, 2015). 5 Blazer D, Hughes DC, George LK. The epidemiology of depression in an elderly community population. Gerontologist 1987; 137: 439–44. 6 Bottino C, Barcelos-Ferreira R, Ribeiz S. Treatment of depression in older adults. Curr Psychiatry Rep 2012; 14: 289–97. 7 Alexopoulos GS, Borson S, Cuthbert BN, et al. Assessment of late life depression. Soc Biol Psychiatry 2002; 52: 164–74. 8 Hankin CS, Spiro A 3rd, Miller DR, Kazis L. Mental disorders and mental health treatment among US Department of Veterans Affairs outpatients: the Veterans Health Study. Am J Psychiatry 1999; 156: 1924–30. 9 Fiske A, Wetherell J, Gatz M. Depression in older adults. Ann Rev Clin Psychol 2009; 5: 363–89. 10 Tiemeier H, van Tuijl R, Hofman A, Kiliaan A, Breteler M. Plasma fatty acid composition and depression are associated in the elderly: the Rotterdam Study. Am J Clin Nutr 2003; 78: 40–46. 11 Valkanova V, Ebmeier K. Vascular risk factors and depression in later life: a systematic review and meta-analysis. Biol Psychiatry 2013; 73: 406–13. 12 Meneilly G, Tessier D. Diabetes in elderly adults. J Gerontol A Biol Sci Med Sci 2001; 56: 5–13. 13 Egede L. Diabetes, major depression, and functional disability among U.S. Adults. Diabetes Care 2004; 27: 421–28.
700
14
15
16
17
18 19
20
21
22 23 24
25
26
27
28
29
30
31
32
33
34
35
36
Egede L, Hernández-Tejada M. Effect of comorbid depression on quality of life in adults with type 2 diabetes. Expert Rev Pharmacoecon Outcomes Res 2013; 13: 83–91. Schulz R, Beach S, Ives D, Martire L, Ariyo A, Kop W. Association between depression and mortality in older adults: the Cardiovascular Health Study. Arch Intern Med 2000; 160: 1761–68. Van der Kooy K, van Hout H, Marwijk H, Marten H, Stehouwer C, Beekman A. Depression and the risk for cardiovascular diseases: systematic review and meta-analysis. Int J Geriatr Psychiatry 2007; 22: 613–26. Bolland R, Keller M. Course and outcome of depression. In: Gotlib I, Hammen C, eds. Handbook of Depression. New York: Guilford Press, 2009. Alexopoulos G. Depression in the elderly. Lancet 2005; 365: 1961–70. Centers for Disease Control and Prevention. Suicide. Facts at a glance 2012. http://www.cdc.gov/ViolencePrevention/pdf/Suicide_ DataSheet-a.pdf (accessed June 25, 2015). Zivin K, Kim M, McCarthy J, et al. Suicide mortality among individuals receiving treatment for depression in the Veterans Affairs health system: associations with patient and treatment setting characteristics. Am J Public Health 2007; 97: 2193–98. Gerritsen DL, Smalbrugge M, Teerenstra S, et al. Act in case of depression: the evaluation of a care program to improve the detection and treatment of depression in nursing homes. Study protocol. BMC Psychiatry 2011; 11: 91. Unutzer J, Park M. Older adults with severe, treatment-resistant depression. JAMA 308: 2012; 909–18. Birrer R, Vemuri S. Depression in later life: a diagnostic and therapeutic challenge. Am Fam Physician 2004; 69: 2375–82. Snowden M, Steinman L, Frederik J. Treating depression in older adults: challenges to implementing the recommendations of an expert panel. Prev Chronic Dis 2003; 5: A26. Liu C, Hedrick S, Chaney E, et al. Cost-effectiveness of collaborative care for depression in a primary care veteran population. Psychiatr Serv 2003; 54: 698–704. Ivanova J, Bienfait-Beuzon C, Birnbaum H, Connolly C, Emani S, Sheehy M. Physicians’ decisions to prescribe antidepressant therapy in older patients with depression in a US managed care plan. Drugs Aging 2011; 28: 51–62. Stojanović-Špehar S, Ozretić L, Blažeković-Milaković S, Bergman-Marković B. Late life depression: challenge or curse for the general practitioner (GP). A cohort study. Arch Gerontol Geriatr 2011; 52: 134–39. Fortney JC, Pyne JM, Mouden SB, et al. Practice-based versus telemedicine-based collaborative care for depression in rural federally qualified health centers: a pragmatic randomized comparative effectiveness trial. Am J Psychiatry 2013; 170: 414–25. Frueh BC, Deitsch SE, Santos AB, et al. Procedural and methodological issues in telepsychiatry research and program development. Psychiatr Serv 2000; 51: 1522–27. Richardson LK, Frueh BC, Grubaugh AL, Egede L, Elhai JD. Current directions in videoconferencing tele-mental health research. Clin Psychol 2009; 16: 323–38. Frueh BC, Monnier J, Yim E, Grubaugh AL, Knapp R. A randomized trial of telepsychiatry for post-traumatic stress disorder. J Telemed Telecare 2007; 13: 142–47. Morland LA, Greene CJ, Rosen CS, et al. Telemedicine for anger management therapy in a rural population of combat veterans with posttraumatic stress disorder: a randomized noninferiority trial. J Clin Psychiatry 2010; 71: 855–63. Greene CJ, Morland LA, Durkalski VL, Frueh BC. Noninferiority and equivalence designs: issues and implications for mental health research. J Trauma Stress 2008; 21: 433–39. Gros DF, Yoder M, Tuerk PW, Lozano BE, Acierno R. Exposure therapy for PTSD delivered to veterans via telehealth: predictors of treatment completion and outcome and comparison to treatment delivered in person. Behav Ther 2011; 42: 276–83. Strachan M, Gros DF, Ruggiero KJ, Lejuez CW, Acierno R. An integrated approach to delivering exposure-based treatment for symptoms of PTSD and depression in OIF/OEF veterans: preliminary findings. Behav Ther 2012; 43: 560–69. Morland LA, Raab M, Mackintosh M. Telemedicine: a cost-reducing means of delivering psychotherapy to rural combat veterans with PTSD. Telemed J E Health 19: 754–59.
www.thelancet.com/psychiatry Vol 2 August 2015
Articles
37
38
39
40
41
42 43
44
45
American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th edn, text revision (DSM-IV-TR). Washington, DC: American Psychiatric Association, 2000. Egede LE, Frueh BC, Richardson LK, et al. Rationale and design: telepsychology service delivery for depressed elderly veterans. Trials 2009; 10: 22. Lejuez CW, Hopko DR, Hopko SD. A brief behavioral activation treatment for depression. Treatment manual. Behav Modif 2001; 25: 255–86. Lejuez CW, Hopko DR, Acierno R, Daughters SB, Pagoto SL. Ten year revision of the brief behavioral activation treatment for depression: revised treatment manual. Behav Modif 2011; 35: 111–61. Yesavage JA, Brink TL, Rose TL, et al. Development and validation of a Geriatric Depression Screening Scale: a preliminary report. J Psychiatr Res 1982; 17: 37–49. Beck AT, Steer RA. Beck Depression Inventory manual. San Antonio: Psychological Corporation, 1987. Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc 1975; 10: 433–41. Spitzer M, Williams RL, Gibbon M. Structured clinical interview for DSM-IV (SCID). Washington, DC: American Psychiatric Association, 1995. Dimidjian S, Hollon SD, Dobson KS, et al. Randomized trial of behavioral activation, cognitive therapy, and antidepressant medication in the acute treatment of adults with major depression. J Consult Clin Psychol 74: 2006; 658–70.
www.thelancet.com/psychiatry Vol 2 August 2015
46 47 48
49
50
51
52
53
Machin D, Campbell MJ. Statistical tables for the design of clinical trials. Oxford: Blackwell Scientific Publications, 1987. Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials 2011; 12: 106. Schnurr P, Friedman M, Engel C, et al. Cognitive behavioral therapy for posttraumatic stress disorder in women: a randomized controlled trial. JAMA 2007; 297: 813–19. Jakupcak M, Wagner A, Paulson A, Varraand A, McFall M. Behavioral activation as a primary care-based treatment for PTSD and depression among returning veterans. J Trauma Stress 2010; 23: 491–95. Cuijpers P, van Straten A, Warmerdam L. Behavioral activation treatments of depression: a meta-analysis. Clin Psychol Rev 2007; 27: 318–26. Bombardier CH, Ehde DM, Gibbons LE, et al. Telephone-based physical activity counseling for major depression in people with multiple sclerosis. J Consult Clin Psychol 2013; 81: 89–99. Donker T, Petrie K, Proudfoot J, Clarke J, Birch MR, Christensen H. Smartphones for smarter delivery of mental health programs: a systematic review. J Med Internet Res 2013; 15: e247. Proudfoot J, Clarke J, Birch M, et al. Impact of mobile phone and web program on symptom and funcational outcomes for people with mild-to-moderate depression, anxiety, and stress: a randomized controlled trial. BMC Psychiatry 2013, 13: 312.
701
Psychotherapy for depression in older veterans via telemedicine: a randomised, open-label, non-inferiority trial Leonard E Egede, Ron Acierno, Rebecca G Knapp, Carl Lejuez, Melba Hernandez-Tejada, Elizabeth H Payne, B Christopher Frueh
Summary Background Many older adults with major depression, particularly veterans, do not have access to evidence-based psychotherapy. Telemedicine could increase access to best-practice care for older adults facing barriers of mobility, stigma, and geographical isolation. We aimed to establish non-inferiority of behavioural activation therapy for major depression delivered via telemedicine to same-room care in largely male, older adult veterans. Methods In this randomised, controlled, open-label, non-inferiority trial, we recruited veterans (aged ≥58 years) meeting DSM-IV criteria for major depressive disorder from the Ralph H Johnson Veterans Affairs Medical Center and four associated community outpatient-based clinics in the USA. We excluded actively psychotic or demented people, those with both suicidal ideation and clear intent, and those with substance dependence. The study coordinator randomly assigned participants (1:1; block size 2–6; stratified by race; computer-generated randomisation sequence by RGK) to eight sessions of behavioural activation for depression either via telemedicine or in the same room. The primary outcome was treatment response according to the Geriatric Depression Scale (GDS) and Beck Depression Inventory (BDI; defined as a 50% reduction in symptoms from baseline at 12 months), and Structured Clinical Interview for DSM-IV, clinician version (defined as no longer being diagnosed with major depressive disorder at 12 months follow-up), in the per-protocol population (those who completed at least four treatment sessions and for whom all outcome measurements were done). Those assessing outcomes were masked. The non-inferiority margin was 15%. This trial is registered with ClinicalTrials.gov, number NCT00324701. Findings Between April 1, 2007, and July 31, 2011, we screened 780 patients, and the study coordinator randomly assigned participants to either telemedicine (120 [50%]) or same-room treatment (121 [50%]). We included 100 (83%) patients in the per-protocol analysis in the telemedicine group and 104 (86%) in the same-room group. Treatment response according to GDS did not differ significantly between the telemedicine (22 [22·45%, 90% CI 15·52–29·38] patients) and same-room (21 [20·39%, 90% CI 13·86–26·92]) groups, with an absolute difference of 2·06% (90% CI –7·46 to 11·58). Response according to BDI also did not differ significantly (telemedicine 19 [24·05%, 90% CI 16·14–31·96] patients; same room 19 [23·17%, 90% CI 15·51–30·83]), with an absolute difference of 0·88% (90% CI –10·13 to 11·89). Response on the Structured Clinical Interview for DSM-IV, clinician version, also did not differ significantly (39 [43·33%, 90% CI 34·74–51·93] patients in the telemedicine group and 46 [48·42%, 90% CI 39·99–56·85] in the same-room group), with a difference of –5·09% (–17·13 to 6·95; p=0·487). Results from the intention-to-treat population were similar. MEM analyses showed that no significant differences existed between treatment trajectories over time for BDI and GDS. The criteria for non-inferiority were met. We did not note any adverse events. Interpretation Telemedicine-delivered psychotherapy for older adults with major depression is not inferior to sameroom treatment. This finding shows that evidence-based psychotherapy can be delivered, without modification, via home-based telemedicine, and that this method can be used to overcome barriers to care associated with distance from and difficulty with attendance at in-person sessions in older adults.
Lancet Psychiatry 2015; 2: 693–701 Published Online July 17, 2015 http://dx.doi.org/10.1016/ S2215-0366(15)00122-4 See Comment page 668 Health Equity and Rural Outreach Innovation Center, Ralph H Johnson Veterans Affairs Medical Center, Charleston, SC, USA (Prof L E Egede MD, Prof R Acierno PhD, Prof R G Knapp PhD, Prof C Lejuez PhD, E H Payne MS, Prof B C Frueh PhD); Center for Health Disparities Research, Division of General Internal Medicine (Prof L E Egede, M Hernandez-Tejada DHA), College of Nursing (Prof R Acierno, M Hernandez-Tejada), and Department of Public Health Sciences (Prof R G Knapp, E H Payne), Medical University of South Carolina, Charleston, SC, USA; Department of Psychology, University of Maryland, College Park, MD, USA (Prof C Lejuez); and Department of Psychology, University of Hawaii, Hilo, HI, USA, and The Menninger Clinic, Houston, TX, USA (Prof B C Frueh) Correspondence to: Prof Leonard E Egede, Center for Health Disparities Research, Medical University of South Carolina, Charleston, SC 29425-0593, USA [email protected]
Funding US Department of Veterans Affairs.
Introduction Depression is a serious and debilitating psychological disorder, with a lifetime prevalence of about 10% in the USA for all age groups.1 Prevalence is slightly lower for older adults (defined as ages 60 years and older)2 in the USA (ages 45–64 years 10%; ages 65 years and older 7%); nonetheless, about 6·5 million people older than 65 years have the disorder in the USA.3,4 Furthermore, 20% of older US adults have substantial symptoms of depression that do not meet DSM diagnostic criteria, but might still warrant treatment5,6 because untreated www.thelancet.com/psychiatry Vol 2 August 2015
minor depression is a risk factor for development of the full clinical disorder.7 Depression is particularly problematic for veterans, with substantial depressive symptoms 2–5 times more likely than in their civilian counterparts.8 Depression in old versus young adults seems to be represented by a somewhat different clinical presentation, characterised by increased physical complaints, fatigue, and apathy, and decreased attention and concentration, but fewer reported symptoms of sadness in older adults.9 Causes of these age-based differences might include neurological changes 693
Articles
associated with ageing, effects of acute disease states, effects of chronic disease states such as atherosclerosis10–14 or cardiovascular complications,15,16 and individual predisposition (ie, late onset of depression).17 Age-specific psychosocial factors include increasingly unstable psychosocial environments associated with old age, resulting from death or incapacitation of those giving social support (ie, friends and family die or become too ill to serve in social support roles), reduced financial freedom, and frequent bereavement.9,18 Depression is common in suicidal individuals, and suicide rates in older adults are disproportionally high compared with the rest of the population,19 particularly with respect to older veterans.20 Despite its wide prevalence and potentially increasing lethality in old age, depression is generally poorly detected and hence undertreated.21,22 Effective treatment of depression in older adults includes some specific forms of psychotherapy and pharmacotherapy (most notably the selective serotonin reuptake inhibitors).23 Cognitive behavioural therapies are the most recommended forms of psychotherapy for depression because of their simplicity and costeffectiveness.24 If collaborative care (interprofessional intervention) models are used, reduced costs are often evident secondary to reduced doctor visits and generally improved functioning.25 However, because most geriatric depression is diagnosed and treated in the primary care setting by non-mental-health specialists, pharmacological treatment is the most widely used, although it is slightly less effective than psychotherapy.26,27 Barriers to psychotherapy for elderly people include mobility issues, stigma concerns, and geographical isolation (eg, living in rural areas). Telemedicine service delivery strategies, such as videoteleconferencing, offer a medium to address these concerns and increase access to evidence-based care.28–30 Findings from studies testing use of telemedicine to deliver psychotherapy show that these methods are non-inferior to those obtained by traditional same-room care in treatment of post-traumatic stress disorder,31–35 and are cost efficient.36 Other disorders, including depression, have received substantially less attention. Thus, we designed this study to assess the efficacy of telemedicine in delivery of psychotherapy for depression to older adult veterans in their homes.
Methods Study design and participants In this randomised, controlled, open-label, noninferiority trial, we recruited participants from the Ralph H Johnson Veterans Affairs Medical Center (Charleston, SC, USA) and four associated community outpatientbased clinics (Goose Creek, Beaufort, and Myrtle Beach, SC, and Savannah, GA, USA). Male and female veterans (aged 60 years or older) meeting DSM-IV37 criteria for major depressive disorder were eligible. However, after we started recruitment, veterans from the Vietnam War 694
era already enrolled in the study requested that we lower the age limit to accommodate more veterans from that war era, so we lowered the age eligibility to 58 years to accommodate their request on Nov 1, 2007. We excluded actively psychotic or demented people, those with both suicidal ideation and clear intent, those meeting criteria for substance dependence, and those unable to provide informed consent from participation; however, to increase generalisation of results, presence of other forms of psychopathology (eg, anxiety disorders) were not cause for exclusion. The Veterans Affairs study site (Ralph H Johnson Veterans Affairs Medical Center, SC, USA) maintains a mailing list of all patients other than those who have opted out of this class of mailings, which is available for clinical research recruitment. We mailed postcards from the centre inviting Veterans Affairs patients aged 58 years and older (after the age limit was lowered from 60 years)2 inviting them to contact study personnel if they felt that they were sad or depressed and might be interested in a trial assessing telemedicine. We offered all participants who contacted study staff standard clinic-based assessment and services for depression as an alternative to the study. All study procedures were approved by the Medical University Institutional Review Board and Veterans Affairs Research and Development Committee. All patients gave written informed consent to participate in the study. The study design and methods have been published previously.38
Randomisation and masking The study coordinator verified all eligibility criteria before randomisation. The study coordinator randomly assigned participants (1:1) to one of two study groups delivering behavioural activation for depression—telemedicine and same-room treatment. The randomisation sequence was computer-generated by the senior biostatistician using a permuted-block randomisation scheme, stratified by race. We varied block size to minimise the chance that masking of assignment would be broken (block size 2–6). Patient allocation was provided in individual sealed envelopes to the study coordinator. Once study eligibility and informed consent for a patient were verified, the study coordinator opened the sealed randomisation envelope to ascertain the random treatment assignment for that patient. Once a randomisation assignment was made, the study coordinator entered that patient into the study. Those doing endpoint psychiatric interviews at baseline and 12 months, and those assessing outcomes, were masked to treatment assignment.
Procedures We did not collect data on antidepressant use because medication needed to be stabilised before randomisation. We maintained psychiatric medications at present doses for 3 months at the discretion of the treatment physician. We asked patients meeting major depressive disorder criteria to www.thelancet.com/psychiatry Vol 2 August 2015
Articles
maintain psychiatric medications at present doses when medically possible. Participants who had not initiated new prescription medications in the 4 weeks before enrolment into the study initiated treatment 1 week after randomisation and completion of the baseline assessment. However, potential participants who had recently begun trials of antidepressants or any other prescription medications needed to wait 4 weeks after assessment to ensure medication stabilisation before starting study treatment. All patients received the same individual mental health intervention—behavioural activation for depression—for 8 weeks according to a previously published manual.39,40 We did not modify treatment for the telemedicine group. Behavioural activation is based on the idea that what a patient does plays a part in how he or she feels.40 To increase the likelihood that reinforcing behaviours will be implemented and logistical obstacles to such behaviours overcome, we used daily planners and valued activity lists to schedule positively (eg, attendance at a grandson’s football game) or negatively (eg, completion of a necessary chore resulting in spousal praise) reinforcing behaviours that are of value to the patient. The planner allows obstacles to successful completion of these behaviours to be addressed in advance so that the likelihood that reinforcement is received is maximised. Patients had 60 min sessions about once a week. The intervention group received therapy via videophone, whereas the control group received face-to-face therapy. We did treatment sessions for the telemedicine group using in-home videoconferencing technology. So as not to need participants to have broadband internet access, we used an analogue videophone (KMEA TV500SP; KMEA, CA, USA) that operates via the standard telephone service. Apart from the video screen, this equipment looks and functions much like a basic touchtone telephone. It is a plug-and-play product, with built-in camera, full duplex speakerphone, 4” LCD colour screen (270k pixels), realtime motion display (18 frames per s), and oversized touchtone buttons for easy use by all patients. Therapists were masters-level counsellors with at least 5 years clinical experience. Each received training during a 2 day workshop from CL and participated in subsequent 1 h supervision meetings once a week with RA. We randomly audited (with a computer-generated random list) 20% of session audiotapes for treatment fidelity (adherence to the manualised behavioural activation treatment, ascertained with a checklist). We used previously validated self-report measures to assess depression at baseline, mid-treatment (4 weeks), post-treatment (8 weeks), and 3 months and 12 months of follow-up. We used the Geriatric Depression Scale (GDS)41 and Beck Depression Inventory (BDI)42 to assess depression, both self-administered in the presence of the interviewer. The GDS is a 30 item measure, and we used the generally accepted cutoff score of 11. The BDI is a 21 item self-report scale. We used the Short Portable Mental Status Questionnaire43 to screen for cognitive impairment www.thelancet.com/psychiatry Vol 2 August 2015
(cutoff >7). We administered endpoint psychiatric interviews (ie, Structured Clinical Interview for DSM-IV, clinician version [SCID])44 at baseline and 12 months. We assessed major depressive disorder and other psychopathologies with this structured clinical interview on the basis of the DSM-IV. We specified onset of major depressive disorder. Interviewers were masters-level counsellors trained to 90% agreement in rating scores on the fidelity checklist before doing any study assessments. We audiotaped interviews, and 20% of them were randomly selected (with a computer-generated random list) and rated by an independent masters-level counsellor to assess inter-rater reliability on an item-by-item basis.
Outcomes The primary clinical outcome measure was the proportion of patients who responded to treatment at the end of the 12 months of follow-up. We established this treatment responder status with each of the dependent measures of depression separately (GDS, BDI, and SCID). For the BDI and GDS, treatment responders were those who showed improvement of at least 50% from baseline on the total score ([baseline – post]/baseline ≥50%), as used in similar studies of treatments for depression.45 For the SCID, treatment responders were those who were no longer diagnosed with major depressive disorder at 12 months follow-up. Secondary efficacy outcomes were
780 patients assessed for eligibility
403 did not give consent 131 ineligible 5 did not return for baseline assessment
241 randomised
120 assigned telemedicine
121 assigned same room
10 discontinued treatment before week 4 assessment 8 moved 2 died 2 discontinued treatment before week 8 assessment 2 moved 3 discontinued treatment before 3 month assessment 3 died 5 discontinued treatment before 12 month assessment 5 moved
100 included in per-protocol analysis
14 discontinued treatment before week 4 assessment 12 moved 2 died 2 discontinued treatment before 3 month assessment 2 died 1 lost to follow-up 1 moved
104 included in per-protocol analysis
Figure 1: Trial profile
695
Articles
continuous BDI and GDS total scores over the longitudinal time trajectory, and response status at week 4, week 8, and 3 months.
Statistical analysis The power and sample size calculation showed that about 100 participants per group were needed to detect noninferiority of telemedicine to same-room methods of intervention delivery (power 85%; non-inferiority parameter Δ = 0·15; assumed response proportion for sameroom group 70%) with the per-protocol sample.46 For intention-to-treat analyses, we inflated the sample size by 10% to account for the dilution effect of intention to treat to yield a sample size of 112 patients in each group. We did descriptive analyses (mean and SD) for continuous variables and frequency distributions for categorical variables to describe demographic and baseline clinical characteristics for the total study sample and each treatment group. We carried out the primary analysis using the per-protocol sample, which we defined Total (n=241) Age (years)
Telemedicine (n=120)
Same room (n=121)
63·9 (5·1)
63·5 (4·4)
64·2 (5·6)
235 (98%)
116 (97%)
119 (98%)
White
143 (60%)
68 (58%)
75 (63%)
Black
94 (40%)
49 (42%)
45 (38%)
Married
165 (69%)
83 (70%)
82 (68%)
13·7 (2·6)
13·5 (2·3)
13·8 (2·9)
Male sex Race
Education (years) Insurance coverage Private
67 (28%)
32 (27%)
35 (29%)
Medicaid or Medicare
72 (30%)
41 (34%)
31 (26%)
Private and Medicaid or Medicare
22 (9%)
9 (8%)
13 (11%)
VA only
80 (33%)
38 (32%)
42 (35%)
Employed
50 (21%)
23 (20%)
27 (23%)
Working hours per week if employed
30·3 (14·8)
29·3 (16·1)
31·0 (14·1) 24 (20%)
Income
49 (21%)
25 (21%)
US$15 000 to US$24 999
56 (24%)
24 (20%)
32 (27%)
US$25 000 to
93 (39%)
49 (42%)
44 (37%)
≥$50 000
40 (17%)
20 (17%)
20 (17%)
Health status Better than last year About the same Worse than last year
39 (16%)
20 (17%)
19 (16%)
104 (44%)
61 (51%)
43 (36%)
96 (40%)
38 (32%)
58 (48%)
Smoker Present
49 (21%)
25 (21%)
24 (20%)
Former
132 (56%)
65 (55%)
67 (57%)
Never
56 (24%)
29 (24%)
27 (23%)
1·2 (1·2)
1·1 (1·1)
1·3 (1·2)
Days with at least 20 min moderate or vigorous activities per week Service-connected medical care
45·1 (40·4)
47·8 (39·7)
42·5 (41·1)
Years being a VA patient
16·2 (12·2)
16·1 (12·0)
16·3 (12·4)
(Table 1 continues on next page)
696
as participants who completed at least four treatment sessions and for whom we made all outcome measurements. We decided the cutoff on the basis of the expected minimum number of sessions that would yield a meaningful effect with behavioural activation treatment. We repeated analyses using the intention-to-treat sample, consisting of all randomly allocated participants. We established effect sizes as the difference in response proportions for telemedicine versus same-room methods of delivery, with a positive value for difference in response proportions showing a superior response proportion for telemedicine. We used 90% CIs,47 judging telemedicine to be non-inferior to same-room delivery as long as the response proportion for same-room does not exceed that of telemedicine by more than a clinically meaningful amount, specified by the non-inferiority margin Δ. We set Δ for the primary outcome at an absolute difference in proportions of 15% (Δ = 0·15).38 We used multiple imputation for imputing missing data for the continuous outcome variables to complete the analysis sets at each timepoint using SAS procedures MI (Markov Chain Monte Carlo method) and MIANALYZE (SAS version 9.3, NC, USA). We used age, sex, race and ethnic origin, education, marital status, employment, income, insurance type, and disability status in the imputation models for BDI and GDS. We assumed missing at random. We established percentage change from baseline using the completed (imputed) dataset at 12 months, and we created the primary dichotomous response outcome on the basis of whether the percentage change from baseline was greater than or equal to 50% (responders) or less than 50% (non-responders). We repeated the procedure for establishment of response proportion for the secondary timepoints: 4 weeks, 8 weeks, and 3 months. We used a mixed-effects longitudinal modelling (MEM) approach to compare the longitudinal profile of response for telemedicine with same room. MEM allows for missing data under the assumption that data are missing at random and it is robust to a range of different distributions of the outcome variable. The dependent dichotomous outcome (response status) was based separately on BDI and GDS. The model included treatment, visit, visit-by-treatment (interaction term) as primary (fixed) independent variables. A significant interaction term for these models would provide evidence against a conclusion of non-inferiority of telepsychiatry. We also used the MEM approach, treating BDI and GDS as continuous outcomes, to estimate baseline-adjusted treatment means across the time periods. In a final set of exploratory analyses, we examined differential response to treatment by sex, race, income, and age (putative moderator variables). For these analyses, we added putative moderator variables to the MEM model as treatment-by-moderator interaction terms, and significance of this term shows that treatment response differs by moderator variable. For MEM www.thelancet.com/psychiatry Vol 2 August 2015
Articles
analyses, we used SAS PROC MIXED and PROC GLIMMIX (SAS version 9.3). We report analyses for each measure of treatment response separately, but based on our analysis plan. When results based on each of these analyses do not converge, overall conclusions on treatment response are based on the two measures that concur. Thus, the indicator of treatment response is in terms of each measure and the converging data from at least two different measures. Both BDI and GDS have to support non-inferiority for that conclusion to be drawn, shown by both the per-protocol and intention-to-treat analyses.47 This trial is registered with ClinicalTrials.gov, number NCT00324701.
Total (n=241)
Telemedicine (n=120)
Same room (n=121)
In Vietnam War
209 (87%)
107 (89%)
102 (84%)
Disabled
167 (70%)
82 (69%)
85 (70%)
4·2 (3·4)
4·3 (3·6)
4·2 (3·3)
(Continued from previous page)
Charlson comorbidity score Mental status questionnaire score 0 error, normal mental function
162 (67%)
80 (67%)
82 (68%)
1 error
64 (27%)
34 (28%)
30 (25%)
≥2 errors
15 (6%)
6 (5%)
9 (7%)
Baseline depression severity GDS
20·8 (4·8)
20·9 (4·8)
20·6 (4·8)
BDI
26·8 (10·0)
26·7 (9·8)
26·8 (10·3)
Generalised anxiety disorder
99 (42%)
50 (42%)
49 (42%)
Panic disorder
19 (8%)
9 (8%)
10 (9%)
Alcohol misuse
59 (25%)
30 (25%)
29 (24%)
Alcohol dependence
29 (12%)
13 (11%)
16 (13%)
Cannabis misuse
18 (8%)
6 (5%)
12 (10%)
6 (3%)
4 (3%)
2 (2%)
147 (63%)
75 (65%)
72 (62%)
Symptomatic diagnosis of panic disorder in the past month
12 (17%)
6 (17%)
6 (18%)
Symptomatic diagnosis of PTSD in the past month
143 (79%)
76 (83%)
67 (76%)
Psychiatric comorbidity
Role of the funding source
Lifetime prevalence of:
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. LEE and RGK had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Cannabis dependence
Results Between April 1, 2007, and July 31, 2011, 377 (48%) of the 780 patients who were screened consented to participate in the study (figure 1). Of these, we randomly allocated 241 (64%) to either the telemedicine (120 [50%]) or sameroom (121 [50%]) treatment groups (intention-to-treat sample). In the telemedicine group, 100 (83%) returned for the final assessment and 104 (86%) of same-room participants returned (per-protocol sample). All study therapists achieved a more than 90% protocolspecified fidelity. All ratings exceeded 90% agreement for study assessments. At the week 4 assessment point for the telemedicine group, 8·3% of the data were missing and imputed, at week 8, 10·0% were imputed, at month 3, 12·5% were imputed, and at month 12, 16·7% were imputed. In the same-room group, 11·6% of data were missing at week 4, 11·6% were missing at week 8, 13·2% were missing at month 3, and 14·0% were missing at month 12. We did not find any systematic differences in the missing and not missing outcome value (BDI and GDS at 12 months) between treatment groups in measured demographic and baseline clinical data; the percentage missing GDS 12 month outcome in the per-protocol sample was very small (1·47%), and using GDS as a surrogate for BDI, we found no differences in mean GDS at 12 months for the missing and not missing BDI between treatment groups (appendix). Table 1 shows the characteristics of the intention-to-treat sample at baseline. We did not note any adverse events for any participant in the study. A high proportion of participants were rural residents (170 [71%]). Average session attendance was high. 192 (80%) participants completed all eight sessions (97 [81%] in telemedicine vs 95 [79%] in same room); 35 (15%) participants completed www.thelancet.com/psychiatry Vol 2 August 2015
PTSD
GAF scale* ≤50
4 (2%)
2 (2%)
2 (2%)
51–60
103 (50%)
51 (48%)
52 (52%)
61–75
98 (48%)
51 (49%)
47 (47%)
Data are mean (SD) or n (%). Information was not available for every patient in most categories. VA=Veterans Affairs. GDS=Geriatric Depression Scale. BDI=Beck Depression Inventory. PTSD=post-traumatic stress disorder. GAF=Global Assessment of Functioning.
Table 1: Demographic and baseline characteristics (intention to treat)
Telemedicine
Same room
Difference
1·05% (–8·30 to 10·41)
Intention to treat BDI
27 (22·54% [15·40 to 29·69])
26 (21·49% [14·72 to 28·25])
GDS
25 (20·96% [14·45 to 27·47])
23 (19·30% [13·29 to 25·31])
1·66% (–7·20 to 10·52)
SCID
53 (44·17% [35·78 to 52·55])
58 (47·85% [39·63 to 56·07])
–3·68% (–15·53 to 8·16) 0·88% (–10·13 to 11·89)
Per protocol BDI
19 (24·05% [16·14 to 31·96])
19 (23·17% [15·51 to 30·83])
GDS
22 (22·45% [15·52 to 29·38])
21 (20·39% [13·86 to 26·92])
2·06% (–7·46 to 11·58)
SCID
39 (43·33% [34·74 to 51·93])
46 (48·42% [39·99 to 56·85])
–5·09% (–17·13 to 6·95)
Data are n (% [90% CI]) or % (90% CI). BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Table 2: Treatment response at 12 months (primary outcomes)
fewer than eight sessions (19 [16%] in telemedicine vs 16 [13%] in same room); and 14 (6%) participants did not attend any session (five [4%] in telemedicine vs nine [7%] in same room). Detailed information about session attendance by treatment group is provided in the appendix. Table 2 shows the treatment response proportions for both the BDI and GDS outcomes at the 12 month
See Online for appendix
697
Articles
Telemedicine
Same room
A
Difference
27
4 weeks Intention to treat 10 (8·50% [3·89 to 13·11])
22 (18·47% [11·86 to 25·08])
–9·97% (–18·06 to –1·88)
GDS
6 (4·63% [1·36 to 7·89])
16 (13·14% [7·87 to 18·41])
–8·52% (–14·70 to –2·34)
BDI
7 (8·86% [3·60 to 14·12])
15 (22·06% [13·79 to 30·33])
–13·20% (–23·00 to –3·40)
GDS
5 (5·15% [1·46 to 8·85])
15 (15·46% [9·43 to 21·50])
–10·31% (–17·39 to –3·23)
BDI score
25
BDI Per protocol
23 21 19
8 weeks
0
Intention to treat BDI
25 (20·83% [14·39 to 27·28])
34 (28·22% [20·79 to 35·66])
–7·39% (–17·34 to 2·56)
GDS
21 (17·67% [11·72 to 23·62])
31 (25·45% [18·30 to 32·61])
–7·79% (–17·01 to 1·43)
B 21 20
Per protocol 19 (25·00% [16·83 to 33·17])
23 (33·82% [24·39 to 43·26])
–8·82% (–21·31 to 3·66)
GDS
18 (19·57% [12·76 to 26·37])
26 (27·66% [20·07 to 35·25])
–8·09%(–18·29 to 2·10)
Intention to treat BDI
23 (19·38% [12·46 to 26·29])
28 (23·10% [15·93 to 30·27])
–3·72% (–13·86 to 6·41)
GDS
19 (15·63% [9·95 to 21·30])
23 (18·64% [12·27 to 25·01])
–3·01% (–11·55 to 5·52)
BDI
15 (22·39% [14·01 to 30·76])
19 (26·76% [18·12 to 35·40])
–4·37% (–16·41 to 7·66)
GDS
15 (17·05% [10·45 to 23·64])
19 (20·65% [13·71 to 27·59])
–3·61% (–13·18 to 5·97)
Per protocol
19 GDS score
BDI
3 months
18 17 16 15 0
Data are n (% [90% CI]) or % (90% CI). BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Table 3: Treatment response over 3 months (secondary outcomes)
follow-up point for the per-protocol and intention-to-treat samples (primary outcome). 22 (22·45% [90% CI 15·52–29·38]) patients in the telemedicine group had at least a 50% reduction in symptom severity and were classified as treatment responders at the primary assessment point (12 months follow-up) compared with 21 (20·39% [90% CI 13·86–26·92]) of those in the sameroom group according to GDS in the per-protocol population. Results with BDI (primary outcome) showed a similar pattern, with 19 (24·05% [90% CI 16·14–31·96]) patients classified as responders in the telemedicine group compared with 19 (23·17% [90% CI 15·51–30·83]) in the same-room group. Estimated differences in response proportions were 0·88% (90% CI –10·13 to 11·89) for BDI and 2·06% (90% CI –7·46 to 11·58) for GDS. The CI lower limits for response proportions are within the allowable non-inferiority limit (ie, do not exceed –0·15), showing that telemedicine was noninferior to same-room therapy at the 12 month timepoint. Treatment response on the SCID (primary outcome) at 12 months was 39 (43·33% [90% CI 34·74–51·93]) patients in the telemedicine group and 46 (48·42% [90% CI 39·99–56·85]) in the same-room group, with a difference of –5·09% (90% CI –17·13 to 6·95), which was not significant (p=0·487). Treatment response at 4 weeks, 8 weeks, and 3 months (secondary outcomes) are shown in table 3. At week 8, non-inferiority of telemedicine could not be established because the lower limit of the 90% CI was not less than 698
Actual telemedicine Actual same room Model-derived telemedicine Model-derived same room
s 12
m
on
th
W ee k4 W ee 3m k8 on th s
0
Time
Figure 2: Actual and model-derived mean (A) BDI and (B) GDS scores over time BDI=Beck Depression Inventory. GDS=Geriatric Depression Scale.
Δ and superiority of same-room therapy could also not be established because 0 is contained in the CI. The trend favouring same-room therapy seemed to diminish with time, with telemedicine having a small advantage in response proportion at 12 months, with a conclusion of non-inferiority of telemedicine reached at both the 3 month and 12 month assessment points. The longitudinal trajectories of actual and modelderived BDI and GDS means are shown in figure 2. Both groups improve over time during the treatment phase, with same-room treatment (MEM model: BDI slope –3·90, p<0·0001; GDS slope –2·66, p<0·0001) improving at a slightly faster rate from baseline to week 8 than does telemedicine (BDI slope –3·07, p<0·0001; GDS slope –2·20, p<0·0001), then both get slightly worse at 3 months. This slight worsening is more pronounced in the telemedicine (BDI slope 2·48, p=0·1914; GDS slope 1·65, p=0·1469) than in the same-room (BDI slope 0·94, p=0·6309; GDS slope 0·78, p=0·5056) group, but neither are significant. The telemedicine group then improves between 3 months and 12 months (BDI slope –1·37, p=0·4818; GDS slope –1·19, p=0·3156), whereas the same-room group gets marginally worse (BDI slope 0·36, p=0·8523; GDS slope 0·36, p=0·7486), but neither are significant. MEM analyses showed that no significant differences existed between treatment trajectories over time for BDI and GDS for either the dichotomous response (interaction term for BDI: F value=1·52; p=0·2107; interaction term for www.thelancet.com/psychiatry Vol 2 August 2015
Articles
GDS: F value=2·36; p=0·0729) or continuous (interaction term for BDI: F value=0·76; p=0·5169; interaction term for GDS: F value=0·97; p=0·4058) outcome. In subgroup analyses exploring the effects of putative moderator variables on treatment response, we noted no significant moderator-by-treatment interaction terms for sex, race, income, and age.
Discussion To our knowledge, this is the first randomised controlled trial of manualised evidence-based psychotherapy for depression in older adults via telemedicine (panel). We have shown that this method is feasible and produces outcomes that are no worse than in-person delivery 12 months after treatment. Participants in both groups tolerated and clinically benefitted from behavioural activation for depression. The magnitude of treatment effect noted in this study is similar to what has been noted in treatment studies for depression in veterans.48 Although the treatment response that we noted during the first 8 weeks was significant for both groups, others using behavioural activation with veterans have not noted significant improvement in depression scores.49 Study retention was excellent, well within the range consistently noted in psychotherapy trials for depression (65–85%),50 and session attendance was high. Assessment of longitudinal trajectories of BDI and GDS shows a slight worsening at 3 months in both groups, which coincides with the end of treatment sessions. The reason for improvement in the telemedicine group between 3 months and 12 months is unclear, so these findings will need to be assessed in future studies. This study has several important aspects. First, it is one of only a handful of methodologically rigorous, noninferiority-designed, randomised controlled trials of psychotherapy telemedicine interventions with any population,30 and the first involving major depression in older adults, a population that often faces barriers of access to care. Second, study implementation was rigorously controlled, including careful a-priori noninferiority analyses and sample size calculations, careful therapist fidelity monitoring, high study retention and session attendance, follow-up assessments up to 12 months, and examination of clinical outcomes in a difficult-to-treat and understudied clinical sample. Third, the study had broad inclusion criteria that allowed for high psychiatric comorbidity, and thus participants are reasonably representative of the patient population of interest. Finally, this study includes a large sample size and high proportions of rural residents (about 70%) and African Americans (about 40%). Along with its strengths, this study has limitations. First, we conservatively excluded participants with acute safety concerns (homicidal or suicidal), present substance dependence, and active psychosis or dementia. Second, clinical benefit was less than was anticipated www.thelancet.com/psychiatry Vol 2 August 2015
Panel: Research in context Systematic review We searched PubMed for articles published in English from Jan 1, 2000, to March 31, 2014, with the following search terms: “depression”, “veteran”, “elderly”, “psychotherapy”, “behavioral activation”, “telemedicine”, “telehealth”, “rural”, and “access”. We found little evidence of effectiveness of evidence-based treatments for depression or any other disorder delivered via office-based treatment versus home-based telemedicine in older patients. Findings from studies testing the use of home-based telemedicine to deliver psychotherapy show that relevant clinical outcomes are non-inferior to those obtained by traditional same-room care in treatment of post-traumatic stress disorder31–35 and that the method is cost effective.36 Other mental health disorders, including depression, have received substantially less attention than post-traumatic stress disorder has. Interpretation To our knowledge, this trial is one of the first demonstrations that evidence-based psychotherapy can be delivered, without modification, via home-based telemedicine. Our findings show that home-based telemedicine can be used to overcome barriers to care associated with distance from and difficulty with attendance at in-person sessions. The implications of this study are that resources should be devoted to offering of services directly into patients’ homes via telemedicine methods, with specific safety procedures in place, such as having the direct telephone number of the police department in which the home-based patient resides in case an emergency dispatch is needed.
when the study was designed. The effect on realised power is negligible in that sample size calculation with proportions is symmetrical, with differences in proportions of 0·7 needing similar power to differences in proportions of 0·3 (with differences in proportions of 0·2 yielding slightly higher power than 0·3). Recruitment exceeded that projected, resulting in slightly higher power for detection of between-group differences than was planned (with corresponding reduction in the falsenegative possibility for the comparison). We are uncertain why treatment response was lower than was expected. Our population was probably more sick and had more severe depression and psychiatric comorbidity than those in the studies that we based our estimates on. Third, the information technology used in the study is now somewhat obsolete; however, new technologies should only improve communication between therapists and patients. Fourth, we included very few women in the study sample, and generalisation of results from this subgroup is not guaranteed. Finally, although the focus of this study was on behavioural treatment of depression, some of the patients were also taking antidepressant medications during the study, and we did not specifically track their effects. Potential participants who had recently begun an antidepressant or other prescription medication needed to wait 4 weeks after assessment to ensure medication stabilisation before starting study treatment. However, some antidepressants take longer than 4 weeks to reach maximum effect, particularly in older people, which might have affected study findings. Nevertheless, we expect incremental treatment effects of medications would be the same between groups. 699
Articles
We have shown that telemedicine is a viable and effective means of delivery of evidence-based psychotherapy for depression to older adults. Future research should rigorously assess effectiveness of telemedicine in other clinical settings and with other information technologies. For example, evidence exists that landline telephones,51 mobile phones,52 and web-based53 interventions can be effectively used to treat mild-to-moderate depression and might be useful alternatives or adjuncts to videoteleconferencing, depending on patient preferences and access. Finally, implementation research is needed into how to most effectively disseminate telemedicine for populations with depression and integrate it with existing models of care of other disorders that elderly adults face. Contributors LEE, BCF, and RGK drafted the report. RA edited the initial draft. LEE, BCF, RGK, CL, and RA reviewed the report. LEE, BCF, RGK, CL, and RA edited the final version. MH-T drafted the literature review. EHP did data analyses and created the tables and figures. Declaration of interests CL is one of the developers of the manualised Behavioural Activation for Depression protocol used in this study. All other authors declare no competing interests. Acknowledgments This work was supported by grants IIR-04-421-3 from the Veterans Affairs Health Services Research and Development Program. We are deeply appreciative of the veterans and Veterans Affairs primary care and mental health providers who contributed to this research effort. All views and opinions expressed herein are those of the authors and do not necessarily reflect those of our respective institutions or the Department of Veterans Affairs. References 1 Centers for Disease Control and Prevention (CDC). Current depression among adults—United States, 2006 and 2008. MMWR Morb Mortal Wkly Rep 2010; 59: 1229–35. 2 WHO. Definition of an older or elderly person. http://www.who. int/healthinfo/survey/ageingdefnolder/en (accessed June 25, 2015). 3 Steffens DC, Skoog I, Norton MC, et al. Prevalence of depression and its treatment in an elderly population: the Cache County study. Arch Gen Psychiatry 2000; 57: 601–07. 4 National Alliance on Mental Illness. Depression in older persons. Oct, 2009. http://www2.nami.org/Content/NavigationMenu/ Mental_Illnesses/Depression/Depression_Older_Persons_ FactSheet_2009.pdf (accessed July 25, 2015). 5 Blazer D, Hughes DC, George LK. The epidemiology of depression in an elderly community population. Gerontologist 1987; 137: 439–44. 6 Bottino C, Barcelos-Ferreira R, Ribeiz S. Treatment of depression in older adults. Curr Psychiatry Rep 2012; 14: 289–97. 7 Alexopoulos GS, Borson S, Cuthbert BN, et al. Assessment of late life depression. Soc Biol Psychiatry 2002; 52: 164–74. 8 Hankin CS, Spiro A 3rd, Miller DR, Kazis L. Mental disorders and mental health treatment among US Department of Veterans Affairs outpatients: the Veterans Health Study. Am J Psychiatry 1999; 156: 1924–30. 9 Fiske A, Wetherell J, Gatz M. Depression in older adults. Ann Rev Clin Psychol 2009; 5: 363–89. 10 Tiemeier H, van Tuijl R, Hofman A, Kiliaan A, Breteler M. Plasma fatty acid composition and depression are associated in the elderly: the Rotterdam Study. Am J Clin Nutr 2003; 78: 40–46. 11 Valkanova V, Ebmeier K. Vascular risk factors and depression in later life: a systematic review and meta-analysis. Biol Psychiatry 2013; 73: 406–13. 12 Meneilly G, Tessier D. Diabetes in elderly adults. J Gerontol A Biol Sci Med Sci 2001; 56: 5–13. 13 Egede L. Diabetes, major depression, and functional disability among U.S. Adults. Diabetes Care 2004; 27: 421–28.
700
14
15
16
17
18 19
20
21
22 23 24
25
26
27
28
29
30
31
32
33
34
35
36
Egede L, Hernández-Tejada M. Effect of comorbid depression on quality of life in adults with type 2 diabetes. Expert Rev Pharmacoecon Outcomes Res 2013; 13: 83–91. Schulz R, Beach S, Ives D, Martire L, Ariyo A, Kop W. Association between depression and mortality in older adults: the Cardiovascular Health Study. Arch Intern Med 2000; 160: 1761–68. Van der Kooy K, van Hout H, Marwijk H, Marten H, Stehouwer C, Beekman A. Depression and the risk for cardiovascular diseases: systematic review and meta-analysis. Int J Geriatr Psychiatry 2007; 22: 613–26. Bolland R, Keller M. Course and outcome of depression. In: Gotlib I, Hammen C, eds. Handbook of Depression. New York: Guilford Press, 2009. Alexopoulos G. Depression in the elderly. Lancet 2005; 365: 1961–70. Centers for Disease Control and Prevention. Suicide. Facts at a glance 2012. http://www.cdc.gov/ViolencePrevention/pdf/Suicide_ DataSheet-a.pdf (accessed June 25, 2015). Zivin K, Kim M, McCarthy J, et al. Suicide mortality among individuals receiving treatment for depression in the Veterans Affairs health system: associations with patient and treatment setting characteristics. Am J Public Health 2007; 97: 2193–98. Gerritsen DL, Smalbrugge M, Teerenstra S, et al. Act in case of depression: the evaluation of a care program to improve the detection and treatment of depression in nursing homes. Study protocol. BMC Psychiatry 2011; 11: 91. Unutzer J, Park M. Older adults with severe, treatment-resistant depression. JAMA 308: 2012; 909–18. Birrer R, Vemuri S. Depression in later life: a diagnostic and therapeutic challenge. Am Fam Physician 2004; 69: 2375–82. Snowden M, Steinman L, Frederik J. Treating depression in older adults: challenges to implementing the recommendations of an expert panel. Prev Chronic Dis 2003; 5: A26. Liu C, Hedrick S, Chaney E, et al. Cost-effectiveness of collaborative care for depression in a primary care veteran population. Psychiatr Serv 2003; 54: 698–704. Ivanova J, Bienfait-Beuzon C, Birnbaum H, Connolly C, Emani S, Sheehy M. Physicians’ decisions to prescribe antidepressant therapy in older patients with depression in a US managed care plan. Drugs Aging 2011; 28: 51–62. Stojanović-Špehar S, Ozretić L, Blažeković-Milaković S, Bergman-Marković B. Late life depression: challenge or curse for the general practitioner (GP). A cohort study. Arch Gerontol Geriatr 2011; 52: 134–39. Fortney JC, Pyne JM, Mouden SB, et al. Practice-based versus telemedicine-based collaborative care for depression in rural federally qualified health centers: a pragmatic randomized comparative effectiveness trial. Am J Psychiatry 2013; 170: 414–25. Frueh BC, Deitsch SE, Santos AB, et al. Procedural and methodological issues in telepsychiatry research and program development. Psychiatr Serv 2000; 51: 1522–27. Richardson LK, Frueh BC, Grubaugh AL, Egede L, Elhai JD. Current directions in videoconferencing tele-mental health research. Clin Psychol 2009; 16: 323–38. Frueh BC, Monnier J, Yim E, Grubaugh AL, Knapp R. A randomized trial of telepsychiatry for post-traumatic stress disorder. J Telemed Telecare 2007; 13: 142–47. Morland LA, Greene CJ, Rosen CS, et al. Telemedicine for anger management therapy in a rural population of combat veterans with posttraumatic stress disorder: a randomized noninferiority trial. J Clin Psychiatry 2010; 71: 855–63. Greene CJ, Morland LA, Durkalski VL, Frueh BC. Noninferiority and equivalence designs: issues and implications for mental health research. J Trauma Stress 2008; 21: 433–39. Gros DF, Yoder M, Tuerk PW, Lozano BE, Acierno R. Exposure therapy for PTSD delivered to veterans via telehealth: predictors of treatment completion and outcome and comparison to treatment delivered in person. Behav Ther 2011; 42: 276–83. Strachan M, Gros DF, Ruggiero KJ, Lejuez CW, Acierno R. An integrated approach to delivering exposure-based treatment for symptoms of PTSD and depression in OIF/OEF veterans: preliminary findings. Behav Ther 2012; 43: 560–69. Morland LA, Raab M, Mackintosh M. Telemedicine: a cost-reducing means of delivering psychotherapy to rural combat veterans with PTSD. Telemed J E Health 19: 754–59.
www.thelancet.com/psychiatry Vol 2 August 2015
Articles
37
38
39
40
41
42 43
44
45
American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th edn, text revision (DSM-IV-TR). Washington, DC: American Psychiatric Association, 2000. Egede LE, Frueh BC, Richardson LK, et al. Rationale and design: telepsychology service delivery for depressed elderly veterans. Trials 2009; 10: 22. Lejuez CW, Hopko DR, Hopko SD. A brief behavioral activation treatment for depression. Treatment manual. Behav Modif 2001; 25: 255–86. Lejuez CW, Hopko DR, Acierno R, Daughters SB, Pagoto SL. Ten year revision of the brief behavioral activation treatment for depression: revised treatment manual. Behav Modif 2011; 35: 111–61. Yesavage JA, Brink TL, Rose TL, et al. Development and validation of a Geriatric Depression Screening Scale: a preliminary report. J Psychiatr Res 1982; 17: 37–49. Beck AT, Steer RA. Beck Depression Inventory manual. San Antonio: Psychological Corporation, 1987. Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc 1975; 10: 433–41. Spitzer M, Williams RL, Gibbon M. Structured clinical interview for DSM-IV (SCID). Washington, DC: American Psychiatric Association, 1995. Dimidjian S, Hollon SD, Dobson KS, et al. Randomized trial of behavioral activation, cognitive therapy, and antidepressant medication in the acute treatment of adults with major depression. J Consult Clin Psychol 74: 2006; 658–70.
www.thelancet.com/psychiatry Vol 2 August 2015
46 47 48
49
50
51
52
53
Machin D, Campbell MJ. Statistical tables for the design of clinical trials. Oxford: Blackwell Scientific Publications, 1987. Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials 2011; 12: 106. Schnurr P, Friedman M, Engel C, et al. Cognitive behavioral therapy for posttraumatic stress disorder in women: a randomized controlled trial. JAMA 2007; 297: 813–19. Jakupcak M, Wagner A, Paulson A, Varraand A, McFall M. Behavioral activation as a primary care-based treatment for PTSD and depression among returning veterans. J Trauma Stress 2010; 23: 491–95. Cuijpers P, van Straten A, Warmerdam L. Behavioral activation treatments of depression: a meta-analysis. Clin Psychol Rev 2007; 27: 318–26. Bombardier CH, Ehde DM, Gibbons LE, et al. Telephone-based physical activity counseling for major depression in people with multiple sclerosis. J Consult Clin Psychol 2013; 81: 89–99. Donker T, Petrie K, Proudfoot J, Clarke J, Birch MR, Christensen H. Smartphones for smarter delivery of mental health programs: a systematic review. J Med Internet Res 2013; 15: e247. Proudfoot J, Clarke J, Birch M, et al. Impact of mobile phone and web program on symptom and funcational outcomes for people with mild-to-moderate depression, anxiety, and stress: a randomized controlled trial. BMC Psychiatry 2013, 13: 312.
701