Performing lifestyle changes and reducing stroke risk — the implementation behind Make My Day

Clinical Question (PICO)

Population: Swedish-speaking adults aged 55–75 years in the Stockholm region, without previous stroke or TIA, with at least three high stroke-risk factors, including at least two modifiable factors. Intervention: Make My Day (MMD), a 10-week programme combining personally meaningful everyday activities, realistic activity-focused goals, six interdisciplinary group sessions and a mobile app. Comparator: For the parent trial, lifestyle assessment and individual goal setting with follow-up; for this process evaluation, there was no randomized comparison of delivery mechanisms. Outcomes: Reach, delivered and received intervention dose, fidelity, adaptations, and participants’ perceived mechanisms of lifestyle change through 12 months. No subgroup treatment effects were estimated in this report.

Bottom Line

The authors conclude that implementation depends on fidelity, context and adequately prepared healthcare professionals: realistic goals and supportive groups can help, but the same groups and technology can also obstruct engagement. All six sessions were delivered, yet allocated intervention participants attended a reported mean of 3.6 of six sessions (SD 2.2); this is implementation evidence, not evidence that the programme prevented clinical strokes.

Design

Study type: Preplanned, convergent mixed-methods process evaluation alongside a single-blinded randomized controlled trial; the process analysis itself was not blinded and did not randomize individual programme components.
N: Parent trial cohort, 122 adults.
- MMD intervention: 63.
- Active control: 59, as verified in the parent trial report.
- Additional process sources: 10 delivering professionals; 20 intervention participants interviewed; evaluation and app-usability surveys from 41 intervention participants.
Randomization: Parent trial used blocks of four with two allocations to each arm; allocation was performed by a researcher not involved in delivery or data collection. This does not make the process themes randomized causal findings.
Setting: Three primary-care rehabilitation units in greater Stockholm, Sweden; delivery occurred across five cohorts, also discussed as sites in parts of the article.
Enrollment and collection: Intervention cohorts ran from spring 2022 to spring 2023 in the parent report; process data were collected during 2022–2024.
Follow-up: Baseline, postintervention and 12 months; a mean follow-up duration was not separately reported.
Analysis: Descriptive statistics; qualitative content analysis of interviews, professional logbooks and researcher fieldnotes; integration of quantitative and qualitative findings in a convergent interpretation matrix. No process-endpoint intention-to-treat effect model or formal causal mediation analysis was presented.
Primary outcome: This process report did not specify one primary efficacy endpoint. Its prespecified purpose was to explain implementation and perceived mechanisms. The parent trial’s primary outcome was a stroke-risk-scorecard category, not incident stroke.

Population

Inclusion Criteria

At least three high stroke-risk factors, including at least two modifiable factors.
Age 55–75 years; residence in greater Stockholm; motivation to change lifestyle; no cognitive impairment preventing participation. These detailed eligibility criteria were verified in the linked parent trial.

Exclusion Criteria

Previous stroke or TIA.
Inability to use a smartphone application.
Insufficient understanding of Swedish.

Baseline Characteristics (process-evaluation Table 1)

Mean age 62 years (SD 5.2); 61 of 122 participants were women (50%).
Ninety-seven participants (80%) had more than 12 years of education; 121 (99%) were of European ethnicity; 84 (69%) were born in Sweden.
The intervention arm’s mean age was 61.7 years (SD 4.9), and 31 of 63 participants were women (49.2%).
Professionals comprised four occupational therapists, three physiotherapists and three dietitians; eight of 10 were women.
NIHSS and acute-stroke severity measures are not applicable: this was primary prevention, not acute-stroke treatment or post-stroke rehabilitation.

Interventions

MMD: Six themed 90-minute sessions over 10 weeks, combining information, reflection and activity. An occupational therapist led an interdisciplinary team with a physiotherapist and dietitian. Participants used an app to record goals, everyday activities, diet and physical activity. The focus was on activities that participants themselves found engaging, rather than generic instructions alone.
Preparation of professionals: Two face-to-face sessions, one digital group session and independent study over two weeks; intervention delivery was outside their usual routine duties.
Active control in the parent trial: The same baseline lifestyle analysis and three individualized activity-focused goals, with follow-up assessments, but without the six-session MMD package or its app. Controls were therefore not an untreated comparison.
Mechanism comparisons: No group was randomized to an app alone, goal setting alone or alternative facilitation. Their separate contributions remain unresolved.

Outcomes

Primary Process Findings (reach, dose, fidelity and perceived mechanisms):

Reach: Recruitment achieved gender balance after advertising adjustments, but socioeconomic disadvantage and non-Swedish-born populations remained difficult to reach. A mostly highly educated, European-ethnicity cohort should not be mistaken for proof of equitable access.
Delivered dose: All six planned sessions were delivered (100%).
Received dose: Reported mean attendance was 3.6 sessions (SD 2.2) among allocated participants and 4.4 (SD 2.1) among programme completers (n=53). The article separately reports mean session participation of 52.8% across sites, or 62.8% when calculated among completers. These are differently described attendance summaries; they should not be treated as interchangeable or used to reconstruct individual exposure.
Retention: Sixteen of 122 participants dropped out before postintervention follow-up, with three further dropouts by 12 months; the reported overall dropout rate was 15.6%. Follow-up retention is distinct from session attendance.
Fidelity: Programme content and professional education broadly followed the manual, but time pressure, extended discussions, facilities and weather altered or displaced activities.
Perceived mechanisms: Realistic goals, meaningful activities and belonging were valued. Group continuity and professional communication could support change, while dominant discussion, unclear concepts and inconsistent attendance could impede it. The focal role of professional communication emerged as an unanticipated mechanism, not a prespecified randomized effect.
Subgroups and uncertainty: No HR, OR, RR, confidence interval or P value was reported for these process mechanisms or for treatment effects within demographic subgroups. Qualitative themes identify hypotheses and implementation problems, not subgroup efficacy.

Additional Process Outcomes:

App logging peaked at 3.2 days per week in week two and declined to approximately one day per week in weeks nine and 10. Mean registration varied across sites, including 3.3 days per week at the final site versus 1.0 and 0.8 at two others.
Mean app usability was reported as 53.9 on a 0–100 scale, with a reported range of 20–97.5. Technical faults and cumbersome entry were recurrent concerns.
Individual-level attendance data were unavailable for an exposure–response analysis; the report cannot establish how many sessions are sufficient.

Parent-Trial Efficacy Context — a different publication:

The 2025 randomized report found adjusted odds of remaining in a high stroke-risk-scorecard category of 0.390 (95% CI 0.156–0.971; P=0.043) for MMD versus active control in its longitudinal mixed-effects analysis with 12 months as the reference time. Its unadjusted 12-month comparison was not statistically significant. The process article’s reference to a “61% reduction” should therefore be read as 61% lower adjusted odds of high-risk classification, not 61% fewer strokes or a demonstrated 61% reduction in absolute stroke risk. Goal-fulfilment mean scores did not differ significantly between groups at 12 months; both groups improved from baseline.

Adverse Events / Safety:

This process evaluation did not provide a formal comparative table of clinical adverse events, stroke, symptomatic intracranial hemorrhage or mortality. Absence of such reporting is not proof of safety. In the parent trial, self-reported stress or depression at 12 months was higher in MMD than control: 18/47 (38.3%) versus 7/39 (17.9%), P=0.039, against a pre-existing numerical baseline imbalance and multiple secondary comparisons. This is a signal to assess participant burden, not proof that MMD caused psychological harm.

Figures

Figure 1: Make My Day programme theory: resources and activities are linked to proposed mechanisms, near-term outcomes and sustained lifestyle change. The diagram is a logic model, not empirical proof of its causal pathways. COPM denotes the Canadian Occupational Performance Measure; EEA, engaging everyday activity; HP, healthcare professional; PHC, primary healthcare. — *Figure 1.* Make My Day programme theory: resources and activities are linked to proposed mechanisms, near-term outcomes and sustained lifestyle change. The diagram is a logic model, not empirical proof of its causal pathways. COPM denotes the Canadian Occupational Performance Measure; EEA, engaging everyday activity; HP, healthcare professional; PHC, primary healthcare.

Figure 2: Attendance percentages over the six sessions for the five delivery cohorts/sites. Site denominators differ, and the published plot excludes dropouts at each site and week. These curves do not supply individual exposure data. — *Figure 2.* Attendance percentages over the six sessions for the five delivery cohorts/sites. Site denominators differ, and the published plot excludes dropouts at each site and week. These curves do not supply individual exposure data.

Figure 3: The intended MMD mechanisms combine meaningful health-promoting activities, realistic activity-focused goals, mobile-health support and group processes. The process evaluation explores how these were experienced rather than separately randomizing them. — *Figure 3.* The intended MMD mechanisms combine meaningful health-promoting activities, realistic activity-focused goals, mobile-health support and group processes. The process evaluation explores how these were experienced rather than separately randomizing them.

Figure 4: Recruitment pathway for participants and primary-care rehabilitation units across five phases. Outreach was adapted during recruitment, but the final cohort remained predominantly highly educated and of European ethnicity. — *Figure 4.* Recruitment pathway for participants and primary-care rehabilitation units across five phases. Outreach was adapted during recruitment, but the final cohort remained predominantly highly educated and of European ethnicity.

Criticisms

Causal overreach: Participants’ accounts of what helped do not demonstrate which component caused the parent trial’s score-based effect. No factorial comparison or mediation analysis isolated goals, groups, professionals or the app.
Selection and missingness: Interviews included 20 intervention participants who accepted invitations after follow-up, not a random sample of nonattenders or controls. The 41 survey respondents and programme completers may not represent disengaged participants.
Investigators were not independent: The team conducted both the RCT and its process evaluation; analysts were not blinded. Reflexivity and triangulation improve credibility but cannot remove allegiance or interpretation bias.
Delivery was not directly observed: Professional logs and participant accounts may miss fidelity failures. Attendance was available only at group level, preventing an individual dose–response test.
External validity: Swedish language, smartphone ability, motivation and the largely highly educated sample limit transferability, particularly to populations with greater digital or social barriers. These results do not directly apply to stroke survivors, who were excluded.
Clinical endpoint gap: The parent trial’s risk-scorecard outcome is not a clinical stroke endpoint; the parent report itself notes limited psychometric testing of that scorecard. A positive adjusted model alongside a nonsignificant unadjusted comparison deserves more caution than the phrase “61% reduction” suggests.

Funding

The Research School in Health Sciences at Karolinska Institutet (FIH) and the Swedish Research Council for Health, Working Life and Welfare (FORTE; grant 2020-00175) funded the study; Karolinska Institute provided open-access funding. The funding statement does not specify the funders’ role in design, analysis or publication, so independence should not be inferred from silence. ScientificMed, now part of Cuviva AB, is acknowledged for app development. Parent-trial registration: NCT05279508.

The paper

Authors. Cecilia Johnsson et al.
Title. Performing lifestyle changes and reducing stroke risk: a process evaluation of a stroke prevention intervention.
Journal. BMJ Open.
Year. 2026; published September 30, 2026.
DOI. 10.1136/bmjopen-2026-123012
PMCID. PMC13629870
PMID. 42816085

Deep Dive — click to expand

What this is

Published in September 2026, this is the implementation companion to the 2025 MMD randomized trial, not another trial proving fewer strokes. It examines the distance between a programme on paper and the programme people actually experienced: six sessions delivered, a mean of 3.6 received, and app use falling to about one logged day per week. The practical question is whether a clinic can reproduce the conditions for lifestyle change, not whether a brochure or app can inherit a treatment effect by association. The analysis below is editorial interpretation; the numerical observations are from the process report unless explicitly attributed to its parent trial.

1. Shadow Audit

The headline can conceal two different denominators. First, 100% session delivery did not mean complete patient exposure: the reported mean was 3.6 of six sessions, and the article’s separate site-level participation summary was 52.8%. A service can successfully schedule everything while failing to make participation workable. Second, the parent trial’s 0.390 adjusted odds ratio concerned risk classification, not strokes. Conflating that finding with clinical event prevention would convert an implementation study into an efficacy claim it cannot support.

The active control is another consequential detail. Both arms received individualized goals and repeated assessment, and both improved in goal fulfilment. Realistic goals may be useful, but because they were shared across arms, their perceived value cannot explain the incremental MMD effect without additional evidence. The people least represented—those unable to use an app, unable to participate in Swedish, or poorly reached by recruitment—also matter most when planning equitable implementation.

2. Inversion Engine

To claim that the app is a necessary engine of sustained change, we would need evidence that keeping people using it changes outcomes independently of motivation, attendance and professional contact. Instead, registrations fell from 3.2 to 1.0 days per week, an approximately 69% decline, calculated from the reported means. That does not prove the app is useless; it rules out assuming sustained high engagement merely because the programme included technology.

For the opposite clinical claim—no added benefit over active control—the parent trial’s adjusted confidence interval already approaches the null: 0.156–0.971, with 1.0 representing no odds difference. The upper bound corresponds to only 2.9% lower odds, calculated as one minus 0.971; the unadjusted 12-month comparison was nonsignificant. A sufficiently precise replication whose interval includes only clinically trivial score changes, or whose clinical stroke outcomes do not improve, would materially weaken expansion. Conversely, a randomized dismantling study could show whether a simpler goal-setting programme preserves benefit. This process evaluation alone settles neither proposition.

3. Second-Order Catalyst

The first reasonable mover is a primary-care prevention or rehabilitation team already offering group lifestyle support, not an acute-stroke thrombectomy service. Its occupational therapist, physiotherapist and dietitian can use the findings to test protected preparation time, clearer examples of meaningful activity, balanced discussion and a non-digital way to participate. These are implementation hypotheses, not newly established treatment requirements.

A realistic first-cycle evaluation would record who was invited, who attended each session, why people missed it and whether participants could articulate one feasible activity goal. That is a proposed local quality-improvement approach, not an outcome measured by this paper. Testing accessibility with underrepresented patients should precede claims that the programme reaches the people at greatest need.

4. Asymmetric Leverage

The small lever is better delivery of what the clinic already promises. Six 90-minute sessions represent nine scheduled group hours, calculated from the programme specification. Extending the syllabus or buying more software may add less value than preventing core goal discussions from being squeezed out of those hours. At the reported mean attendance, exposure would be 5.4 hours if every attended session lasted 90 minutes; that is an illustrative calculation, not observed contact time or a validated minimum dose.

The larger denominator is the potential prevention population, but this report supplies no clinical event reduction, cost-effectiveness estimate or number needed to treat. The asymmetric opportunity is therefore a testable operational one: improve access and facilitator readiness before paying to scale an elaborate package. Whether a simplified model saves resources while preserving outcomes requires a direct comparison, not inference from the interviews.

5. Paradigm Destroyer

The habit this paper challenges is treating attendance, app installation and distribution of information as interchangeable with effective prevention. It does not justify removing medical risk-factor treatment, promising event reduction or mandating an app.

Proposed two-sentence prevention-service update: “For lifestyle groups, document one patient-chosen activity goal, the clinician responsible for reviewing it, and individual attendance barriers rather than counting sessions offered alone. Provide a non-digital participation route and check facilitator preparedness and participant burden before extending the programme.” This is a cautious service-design proposal, not a trial-validated protocol or a change to acute-stroke treatment.

MVP — Minimum Viable Proof

The minimum actionable observation is the mismatch between all six sessions delivered and a reported mean of 3.6 attended: a delivery audit that ignores received dose misses a central implementation problem. That supports measuring access, exposure and experience now. To claim that a revised service prevents strokes, the minimum proof would instead require a sufficiently powered study with clinical events and an appropriate comparator; neither 20 interviews nor a change in a risk-score category meets that standard.

Best Combination

Read this process evaluation beside the 2025 parent trial, which provides the randomized comparison, eligibility details and score-based endpoint. The trial gives a qualified signal of benefit on adjusted risk classification; the process report identifies plausible delivery failures and candidate mechanisms, not confirmation of why that signal occurred. The 2023 protocol, cited by both papers, supplies the prospective design context. This is a linked-study interpretation, not a systematic synthesis of all lifestyle-prevention research.

The productive next comparison would retain individualized goals in both arms and randomize a clearly defined additional component, with individual attendance, meaningful objective risk-factor measures, participant burden and longer-term clinical outcomes captured. Until then, use the qualitative findings to improve questions and implementation measurement—not to rank untested components as proven active ingredients.

Overvalue Warning

Do not turn “61%” into fewer strokes. It describes lower adjusted odds of a high-risk scorecard category in a separate trial, with a confidence interval close to the null and no significant unadjusted 12-month group difference. It is neither a relative risk of clinical stroke nor an absolute event reduction.
Do not declare a minimum effective dose or universal mechanism. Low average attendance does not prove that three or four sessions suffice, and perceived benefits among volunteers do not establish that the app, group format or professional mix caused better outcomes. Individual exposure data, control interviews and equitable recruitment remain important missing pieces.

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.