A Phase II Multicentered, Single-Blind, Randomized, Controlled Trial of the Stroke Self-Management Program
Background and Purpose—The benefits of chronic disease self-management programs for stroke survivors are uncertain because individuals with severe impairments have been excluded from previous research. We undertook a phase II randomized controlled trial to determine whether a self-management program designed for survivors (SSMP; 8 weeks) was safe and feasible compared to standard care (control) or a generic self-management program (generic; 6 weeks).
Methods—Stroke survivors were recruited from 7 South Australian hospitals via a letter or indirectly (eg, newspapers). Eligible participants were randomized at a 1:1:1 ratio of 50 per group. Primary outcomes were recruitment, participation, and participant safety. Secondary outcomes were positive and active engagement in life using the Health Education Impact Questionnaire and characteristics of quality of life and mood at 6 months from program completion.
Results—Of 315 people screened, 149 were eligible and 143 were randomized (48 SSMP, 47 generic, 48 control); mean age was 69 years (SD, 11) and 59% were female. Demographic features were similar between groups and 41% had severe cognitive impairment; 57% accessed the interventions, with 52% SSMP and 38% generic completing >50% of sessions (P=0.18). Thirty-two participants reported adverse events (7 control, 12 generic, 13 SSMP; P=0.3; 34% severe); however, none was attributable to the interventions. Potential benefits for improved mood were found.
Conclusions—SSMP was safe and feasible. Benefits of the stroke-specific program over the generic program included greater participation and completion rates. An efficacy trial is warranted given the forecast growth in the stroke population and improved survival trends.
More than 60,000 strokes occur in Australia each year and stroke is a leading cause of disease burden.1,2 At 10 years after a first-ever stroke, the cumulative risk of a recurrent event is ≈43%, and the risk of being disabled or deceased is ≈86%.3,4 Reduced quality of life, depression and cognitive impairment are common.5 Stroke survivors and their families often find it difficult to manage this long-term condition given the abrupt transition from being “healthy” to having disability. The situation is exacerbated by the lack of community-based programs that may help survivors reduce the risk of recurrent events and improve quality of life.
Self-management programs have emerged as a mechanism for enhancing the care offered by health services.6,–,9 Components of successful self-management programs include a focus on patients' needs, goal setting, practice of skills obtained, attention to emotional and role management, and empowering patients to work effectively with health care professionals.10,11
Previous studies have excluded survivors with severe physical disability and cognitive impairments, and this has limited the generalizability to the broader stroke population. Recent evidence supports the inclusion of disadvantaged patients with poor disease outcomes in self-management interventions.12
In Australia, the most commonly available self-management course is the generic Stanford University program (http://patienteducation.stanford.edu/programs/cdsmp.html).11 However, disease-specific self-management programs may have advantages over generic programs.13 Given the lack of evidence, data are urgently needed to determine the value of both generic and tailored self-management programs for this population. In 2004, the National Stroke Foundation, with experts and consumers, piloted a Stroke-specific Self-Management Program (SSMP). The pilot program provided evidence that the SSMP was a viable alternative to the Stanford program.14 The aims of this phase II randomized controlled trial were to evaluate whether the SSMP was safe and feasible.
Materials and Methods
A protocol detailing the trial has been published14 (www.strokefoundation.com.au). Briefly, it is a multicentered, single-blind, phase II, randomized controlled trial. Treatment groups included the SSMP compared to the Stanford chronic condition self-management program (referred to here as the generic program) and standard care (control group) in stroke survivors.
The SSMP intervention is an 8-week co-facilitated program. It is different from the generic (Stanford) intervention in the following ways: (1) only includes stroke survivors; (2) has greater contact time; (3) is only delivered by health professionals and peer leaders skilled in stroke and trained by the National Stroke Foundation; (4) provides targeted stroke-specific information each week; and (5) revisits information provided in other weeks to ensure retention of learning and skills. The SSMP intervention was performed 5 times by the same stroke educator, but the peer leaders may have differed.
The generic intervention is a 6-week co-facilitated program and, in this study, participants were enrolled in 11 generic programs performed by up to 8 different Stanford-trained leaders. We ensured that each program did not include more than one-third of stroke survivors to reflect how survivors might experience generic programs. The generic program covers a range of topics such as appropriate use of medicines, communicating effectively with family and friends, and nutrition. The standard care group received usual care, that is, access only to the information and education provided by the hospital team or their local general practitioner.
We hypothesized that a greater proportion of stroke survivors would participate and complete the SSMP compared with a generic program, and the number of adverse events would be similar across comparator groups. Feasibility was assessed using levels of participation and included numbers of randomized participants who had: (1) an intention to participate (or had booked to start program); (2) accessed a program (ie, had attended at least 1 session); and (3) completed the program (attended ≥50% of sessions). We also assessed which recruitment method attracted the most potential participants. The study was conducted in the metropolitan region of South Australia.
Secondary outcomes included comparisons of positive and active engagement in life (range, 1–6 [strongly disagree]), a domain of the Health Education Impact Questionnaire,15 quality of life measured by the Assessment of Quality of Life16 tool (range, −0.04 to 1 [good health]), and mood measured by the irritability, depression, and anxiety scales (range, 0–15 [worst]).17 A self-reported health service utilization diary, initially included as a secondary measure, was later omitted because of inconsistent and missing data.
Eligibility criteria included confirmed stroke diagnosis of ≥3 months, age 18 years or older, ability to participate in a program for 2.5 hours per week for 8 weeks, English as a first language, living in the community (but not in an aged care facility at time of enrollment), and no previous participation in self-management programs. Presence of cognitive impairment, language difficulties, marked physical disability, or recurrent stroke events were not exclusions.
Recruitment included advertising in local newspapers, promotion through Divisions of General Practice, local peer support group newsletters, and letters of invitation to discharged survivors at 1 of 7 participating hospitals. Potential participants were invited to make direct telephone contact with research staff who undertook eligibility screening. If eligible, then participants formally gave consent to a research assistant undertaking a home visit to complete a baseline assessment. The sample size was arbitrarily set at 50 participants per group (total n=150) to generate preliminary evidence of safety and feasibility while generating data for sample size estimations for a larger phase III efficacy study. When possible, survivors who were randomized to an intervention arm, but who declined participation or did not complete the intervention, were asked to provide reasons for nonparticipation or noncompletion.
Randomization and Masking
After written informed consent and baseline assessments, participants were randomized (1:1:1 ratio) using a remote Web-based telephone randomization program.14 Randomization was undertaken by the stroke educator who also organized the programs. This ensured that research staff remained unaware of the allocation group when conducting the follow-up assessments and data processing.
All participants were followed-up using a modified Dillman protocol.14 The first assessment was completed 2 to 4 weeks after program completion for individuals randomized to an intervention; for standard care, assessment was completed ≈1 month after the baseline assessment. The final assessment was undertaken at 6 months after program completion.
Statistical analyses were undertaken according to the protocol.14 Participants' baseline characteristics were reported using descriptive statistics. Outcomes were reported according to intention-to-treat and as an “on-program” (participants who completed at least 50% of the program sessions) analysis. To explore potential differences in health outcomes, hierarchical, multilevel, generalized least-square random effects modeling was used. This is a superior approach when sample sizes are small and variance in program factors, as well as participant factors, may influence outcomes. Effect size (Cohen D) was calculated using the difference between the mean change in the intervention and standard care groups and dividing by the pooled SD. The Cohen D was estimated to assist future sample size calculations and to permit comparisons with other self-management trials. The magnitude of the effect was classified as small (0.20 to 0.49), medium (0.50 to 0.79), and large (>0.8).18
Missing Data Replacement
A small number of observations were missing for individual questions within a dimension (Health Education Impact Questionnaire, n=7; Assessment of Quality of Life, n=2). Imputation of values was only used for the intention-to-treat analysis, whereby for missing questions within a domain the average of the 2 known questions was calculated and this value was replaced for the missing data. If 6-month follow-up data were missing for an entire question or domain, then the “last value carried forward” approach was used.
The numbers of participants randomly assigned, receiving intended treatment, completing the study protocol, and analyzed for the primary outcome analysis are outlined in Figure 1. The median time between randomization and commencing a program varied according to the intervention group (SSMP: median, 39; interquartile range, 21–50 days; generic group: median, 52; interquartile range, 46–77 days; P=0.001).
Participant recruitment occurred between March 2008 and April 2009. The majority of screened participants were obtained by response to a hospital letter (89%; n=278). We were unable to report the total number of letters sent to achieve this response, but at least 2112 letters from the hospitals were posted (≈13% return rate). Based on the available data for 115 cases, most participants were sourced from acute hospitals (75%) compared to rehabilitation hospitals (13%) or the community (12%).
Of the 226 individuals who were screened by telephone, 39 (17%) were ineligible. For 12 participants (5%) who refused during the telephone interview, eligibility was not determined. Among those who were eligible, 26 (12%) declined to participate and 149 (66%) consented (Figure 1). Reasons for refusing to participate included time constraints, lack of interest, and inability to attend (eg, because of transportation issues). Sixteen participants did not provide a reason for refusing. Of the 149 consented participants, 6 declined before baseline assessments, resulting in 143 (96%) participants randomized.
Response Rates for Data Collection Periods
All randomized participants completed baseline questionnaires, 123 (86%) completed the immediate postintervention assessments, and 122 (85%) completed health outcome assessments at 6 months. The standard care group completed more follow-up questionnaires than the intervention groups (Figure 1). Supplemental Table 1 (http://stroke.ahajournals.org) provides the demographic characteristics of participants who did and did not complete the 6-month outcome assessment. Participants with a history of a first-ever stroke were more likely to complete the 6-month assessments for standard care and SSMP groups. The participants without comorbidities in the standard care group were less likely to complete the follow-up assessments.
Participants ages ranged from 62 to 78 years (mean, 69.4; SD, 11.4) and 59% were female (Table 1). Almost two-thirds were born in Australia and lived with partners. The overall mean score for the domain “positive and active engagement in life” was 3.83 (SD, 0.88). Age, gender, and education levels did not differ between participants in the standard care compared to the intervention groups. The generic group had slightly more patients with hypertension, hypercholesterolemia, and ischemic stroke. There were potentially clinically relevant differences in stroke severity. The standard care group appeared to have lower quality of life, worse depression scores, and fewer were independent (modified Rankin scale score, 0–1) at baseline. Randomization appeared balanced for cognitive level and language impairments across the groups (Supplemental Table 2, http://stroke.ahajournals.org). Severe cognitive impairment was found for 59 participants (41%).
Primary Outcome Analyses
Overall, 25 (52%) survivors completed the SSMP and 18 (38%) completed the generic intervention (P=0.18; Figure 1). Figure 2 shows the number of participant numbers and reasons for withdrawal. Participants were able to provide >1 reason. Participants indicated that they did not start or complete the program because dates were unsuitable, adverse medical events not related to the program (eg, cancer), lack of transportation, and no longer being interested. More individuals in the generic group who withdrew reported that the program was not appropriate for their recovery and/or they were no longer interested.
Overall, 36 adverse events were reported among 32 participants (Table 2). All were reported at first follow-up assessment. Most (78%) were classified as moderate or severe (eg, an event that causes interference with current level of activity or, if severe, is incapacitating, life-threatening, and involving a hospital admission). There were 11 severe adverse events reported and none was directly attributed to the study interventions. There were no differences between groups in the proportion of severe adverse events. (χ2=5.6; P=0.47).
Secondary Outcome Analyses
Improvements from baseline to 6 months after intervention were observed for each health domain, irrespective of the participant's intervention group (Table 3). No statistically significant differences between the groups for the intention-to-treat or on-program analyses were observed using generalized least square multilevel modeling that adjusted for recruitment (Table 3).
Estimation of Cohen D provided evidence of potentially large effect sizes mostly in the expected (benefits) direction for both interventions against standard care for reducing depression and anxiety and improving positive and active engagement in life (Supplemental Table 3, http://stroke.ahajournals.org). The effect on positive and active engagement in life for the on-program analysis suggested that there may be a large potential effect over the generic program in this Health Education Impact Questionnaire domain (generic, 0.66; SSMP, 1.47).
There is limited evidence to guide best practice in long-term stroke recovery despite a large stroke population that is expected to treble over the next 10 years.19 Stroke recovery is a dynamic process, with survivors and their caregivers needing to develop their own effective behavioral strategies to overcome barriers, often in the absence of health professional support.20 Self-management programs have emerged as key strategies to address this gap.21 This trial provides the first international evidence to our knowledge in self-management for the broad stroke population including those with cognitive, language, or severe physical impairments. Both the stroke-specific and generic self-management programs were safe and feasible interventions.
There were no significant differences in severe adverse events between the interventions and standard care. Furthermore, the differences in effect size using the on-program analyses for positive and active engagement in life, depression, and anxiety for the SSMP compared to standard care and generic program were large. This on-program analysis is significant because of the large proportion of participants who did not access either intervention. There are a number of potential explanations for this encouraging result. However, it requires further investigation in a phase III trial with a larger sample size because this study was not powered to achieve statistically significant changes in health outcomes. In addition, standard care participants were followed-up at regular intervals with questionnaires about their health, which may have diluted potential differences between groups.
The results suggest that the SSMP is, in fact, more feasible than the generic program because >90% of survivors in the SSMP intervention who started were able to complete the intervention compared to only 69% of participants in the generic program. We could speculate that the role modeling provided by facilitators skilled in stroke management and/or living with stroke may influence social persuasion in a stroke-specific group more than in the generic program. Alternately, the design aspects of the SSMP, which runs for 2 weeks longer than the generic intervention, has stroke-specific content, is deliberately repetitive, and uses multiple educational modalities, may have contributed to increased attendance.
To determine who benefits most from self-management programs, stratification by stroke severity may need to be considered in a future trial of efficacy. Access to the programs by randomized participants was only moderate (57%) because of lack of transportation, unsuitable dates, and delay between recruitment and commencement. Alternative access approaches include delaying assessment and randomization until courses are available. Qualitative evaluation of barriers and enablers to accessing the programs and positive and negative aspects of course content and delivery would enhance a phase III study.
This is the first study to our knowledge to trial a stroke-specific self-management program that is generalizable to all people with stroke. We found that the SSMP was more feasible than a generic program because more participants completed the intervention and found it relevant. SSMP appears warranted and may provide an important opportunity to address health and social issues associated with long-term recovery, which also helps address the gaps in community-based health services. A phase III efficacy study would be required to determine whether the stroke-specific program was superior to the standard generic programs in reducing the consequences of stroke, such as mood disorders.
Sources of Funding
Dominique Cadilhac is supported by a NHMRC/National Heart Foundation postdoctoral fellowship. Richard Osborne was supported, in part, by a NHMRC Population Health Fellowship (Career Development Award). The research was supported by a grant from the J.O. and J.R. Wicking Trust and in-kind support from the National Stroke Foundation. Richard Lindley is supported by an infrastructure grant from New South Wales Health.
The authors thank the research officers, program facilitators (Nicole Prideaux and Esther Westra), and the study participants. The authors acknowledge the input from the Advisory Committee, Rebecca Naylor, Associate Professor Leonid Churilov for advice on the multivariate modeling, and Assistant Professor Velandai Srikanth for his advice on categorizing participants with cognitive impairments.
The online-only Data Supplement is available at http://stroke.ahajournals.org/cgi/content/full/STROKEAHA.110.601997/DC1.
- Received September 7, 2010.
- Accepted December 29, 2010.
- © 2011 American Heart Association, Inc.
- Cadilhac DA,
- Carter RC,
- Thrift AG,
- Dewey HM
- Hardie K,
- Hankey GJ,
- Jamrozik K,
- Broadhurst RJ,
- Anderson C
- Hardie K,
- Hankey GJ,
- Jamrozik K,
- Broadhurst RJ,
- Anderson C
- Paul SL,
- Sturm JW,
- Dewey HM,
- Donnan GA,
- Macdonell RA,
- Thrift AG
- Lorig K
- Redman BK
- Battersby M,
- Hoffmann S,
- Cadilhac D,
- Osborne R,
- Lalor E,
- Lindley R
- Snaith RP,
- Constantopoulos AA,
- Jardine MY,
- McGuffin P
- Cohen J
- Senes S
National Health Priority Action Council. National Service Improvement Framework for Heart, Stroke and Vascular Disease: Australian Health Ministers' Conference. Nat Chronic Dis Strat. 2006; 58: 91.