Effects of cognitive training in Parkinson's disease: a randomized controlled trial.

The researchers compared two different cognitive training programs against a waiting-list control group in people with Parkinson's disease (PD):

**Structured training (NEUROvitalis):** 12 group sessions (90 minutes each, twice per week for 6 weeks) where each session focused on one specific cognitive domain — attention, memory, or executive functions. Each session began with a psychoeducational component explaining how that cognitive function works and strategies to improve it. Exercises included individual tasks, group tasks, and group games targeting the specific domain of the day.

**Unstructured training (Mentally fit):** Same format (12 sessions, 90 minutes, 6 weeks) but with cognitive tasks randomly assembled across sessions — no domain-specific focus, no psychoeducation. Instead of learning about cognitive strategies, participants had group conversations about topics like "dealing with the disease." This was designed to mimic typical "brain jogging" programs found in books and community memory groups.

**Waiting-list control group:** Received no training between pre- and post-testing, but were offered training after the study ended.

The primary outcomes were cognitive functions commonly impaired in PD: attention, verbal short-term memory, verbal long-term memory, visual long-term memory, working memory, and verbal fluency. Secondary outcomes were visuoconstruction, depression (Beck Depression Inventory-II), and quality of life.

**Sample size:** 65 non-demented Parkinson's disease patients (after 5 dropouts from an initial 70)

**Population:** Patients with idiopathic PD according to UK Brain Bank criteria, recruited from the Movement Disorders Unit at the University Hospital of Cologne, Germany (n=40) and regional PD support groups (n=30)

**Inclusion criteria:** Hoehn & Yahr stages I–III (mild to moderate PD), no suspected dementia (Mini Mental State Examination score ≥25), no other neurological or psychiatric diseases (except depression), no impaired hearing or sight, no deep brain stimulation treatment

**Baseline characteristics:** Mean age ~67 years, mean education ~14 years, mean disease duration ~7 years, mean UPDRS motor score ~20 (range 5–39), 50 patients classified as cognitively unimpaired and 15 as having PD-Mild Cognitive Impairment (MCI), 13 patients had at least mild depression (BDI-II >13)

**Setting:** Outpatient clinic and community support groups in Cologne, Germany; study conducted May 2010 to June 2012

The researchers used a comprehensive neuropsychological test battery administered within 10 days before and 10 days after the 6-week training period:

**Primary outcomes (cognitive functions):**

**Verbal short-term memory:** "Memo" word list learning test (immediate recall of a word list)

**Verbal long-term memory:** Delayed recall of the Memo word list

**Visual long-term memory:** Complex Figure Test (recall of a complex geometric figure after a delay)

**Working memory:** Digit span reverse subtest from the DemTect (repeating sequences of numbers backwards)

**Attention:** Various attention tests (specific tests not detailed in abstract)

**Verbal fluency:** Semantic verbal fluency subtest from the DemTect (naming as many items as possible from a category)

**Secondary outcomes:**

**Visuoconstruction:** Complex Figure Test copy task

**Depression:** Beck Depression Inventory-II (BDI-II, 0–63 scale, higher = more depressive symptoms)

**Quality of life:** Quality of life questionnaire (specific instrument not named in abstract)

**Overall cognitive screening:** DemTect (a screening tool for detecting MCI and dementia, includes five subtests)

**Clinical measures:**

**Motor severity:** Unified Parkinson's Disease Rating Scale Part III (UPDRS-III, motor score, 0–108, higher = worse motor function), videotaped and rated by a blinded MDS-certified physician during "on" phase

**Disease stage:** Hoehn & Yahr staging (I–V)

**Medication:** Levodopa equivalent daily dose (LEDD) calculated to standardize dopaminergic treatment across patients

**Study design:** Randomized controlled trial (RCT) with three parallel arms: two intervention groups and one waiting-list control group.

**Randomisation:** Patients were randomly allocated using a computer program (randomizer.org). However, randomisation was stratified by recruitment source — support group patients and university hospital patients were randomised separately, which created imbalance: the control group had more support group patients (16 vs 5 from university hospital), while intervention groups had more university hospital patients (NV: 18 vs 7; MF: 17 vs 7). This is a methodological concern because support group patients may differ systematically from clinic patients (e.g., more motivated, different disease management).

**Blinding:**

Patients in the treatment groups were blinded to which type of training they received (structured vs unstructured) — they knew they were getting "cognitive training" but not the specific hypothesis about structure

Examinators (the people administering the cognitive tests) were blinded to which condition patients were in

The UPDRS motor ratings were done by a blinded physician from video recordings

**Trainers were NOT blinded** — they knew which program they were delivering, which could introduce bias in how enthusiastically or competently they delivered each program

The control group knew they were on a waiting list (no placebo control), so expectation effects cannot be ruled out

**Duration:** 6 weeks of training (12 sessions of 90 minutes each, twice per week). Pre- and post-testing occurred within 10 days before and after the training period. No follow-up assessment was conducted, so long-term effects are unknown.

**Statistical approach:** The researchers treated the study as "three separate trials" for analysis — comparing NV vs CG, MF vs CG, and NV vs MF directly. They used repeated measures analysis of variance (ANOVA) with time (pre vs post) as the within-subject factor and group as the between-subject factor, controlling for age, education, and disease duration as covariates. Bonferroni correction was applied for multiple comparisons. Effect sizes were reported as partial eta-squared (η²p), where 0.01 = small, 0.06 = medium, 0.14 = large.

**What this design can prove:**

The RCT design with random allocation can establish that differences between groups are due to the intervention rather than pre-existing differences

The comparison of two active treatments (NV vs MF) can show whether structured training is superior to unstructured training

Blinding of examiners reduces assessment bias

**What this design cannot prove:**

**Cannot prove durability:** Without follow-up, we don't know if effects last beyond 6 weeks

**Cannot rule out expectation effects:** No placebo control group (the waiting list group knew they weren't getting training)

**Cannot separate social from cognitive effects:** Group training includes social interaction, which itself may improve cognition or mood — the design doesn't isolate the cognitive exercises from the social context

**Limited generalisability:** Non-demented PD patients only; results may not apply to those with dementia or more advanced disease

**No active control for attention:** The control group received no intervention at all, so any benefit could be from the extra attention and social engagement rather than the cognitive exercises specifically

**Major methodological weaknesses:**

Unbalanced randomisation by recruitment source

No blinding of trainers

No follow-up assessment

No placebo/sham control

Small sample size per group (NV: 25, MF: 22, CG: 18 after exclusions)

Multiple statistical comparisons without full correction (Bonferroni was applied but with three separate "trials" rather than a single omnibus test)

**Primary outcomes — Structured training (NV) vs Control:**

**Working memory (DemTect digit span reverse):** Significant improvement in NV compared to CG (F=5.36, p=0.03, η²p=0.12 — medium effect size)

**Short-term memory (Memo word list learning):** Significant improvement in NV compared to CG (p<0.01, exact F-value not reported in abstract)

**Other cognitive measures (attention, verbal long-term memory, visual long-term memory, verbal fluency):** No significant differences between NV and CG

**Primary outcomes — Unstructured training (MF) vs Control:**

No significant improvements in any cognitive measure compared to CG

**Primary outcomes — Structured (NV) vs Unstructured (MF):**

**Working memory (DemTect):** NV improved significantly more than MF (p<0.05)

**Other cognitive measures:** No significant differences between NV and MF

**Secondary outcomes:**

**Depression (BDI-II):** The MF (unstructured) group showed significant reduction in depression scores compared to CG (p<0.05). The NV group did NOT show significant depression reduction compared to CG

**Quality of life:** No significant differences between any groups

**Visuoconstruction:** No significant differences between any groups

**Subgroup considerations:**

At baseline, 15 patients met criteria for PD-MCI and 13 had at least mild depression — the study was not powered to analyze these subgroups separately

The authors note that the MF group's depression improvement may be related to the group conversations about "dealing with the disease" which were part of that program but not the structured program

**Working memory improvement:** The structured training group showed a medium effect size (η²p=0.12) for digit span reverse improvement. In practical terms, this means patients could remember approximately one additional digit when repeating numbers backwards — for example, going from recalling 4 digits backwards to 5 digits backwards. This is a modest but clinically meaningful improvement for daily tasks like mental arithmetic or following multi-step instructions.

**Short-term memory improvement:** The word list learning improvement (p<0.01) represents being able to recall approximately 1–2 more words from a list immediately after hearing it. For context, healthy adults typically recall 7±2 words from a 10-word list; PD patients often recall 4–6. An improvement of 1–2 words brings them closer to normal range.

**Depression reduction:** The unstructured training group showed a significant reduction on the BDI-II (p<0.05). The BDI-II ranges from 0–63, with scores 14–19 indicating mild depression, 20–28 moderate, and 29–63 severe. The average reduction was not reported numerically in the abstract, but given that the group included patients with scores from 15–20 at baseline, a clinically meaningful reduction would be 3–5 points (moving from moderate to mild range, or from mild to minimal).

**Comparison to other interventions:** The effect sizes here (medium for working memory) are comparable to those seen in cognitive training studies in healthy older adults, but smaller than the effects typically seen with exercise interventions in PD (which often show large effects on executive function). The depression reduction in the unstructured group is roughly equivalent to what might be expected from a low-dose antidepressant or 8 weeks of group psychotherapy.

**Acknowledged by authors:**

No long-term follow-up data — unknown whether effects persist

Small sample size, especially for subgroup analyses (PD-MCI, depressed patients)

Heterogeneous patient population (different disease durations, motor severities)

No control for medication changes (though patients with medication changes were excluded post-hoc)

Trainers were not blinded to treatment condition

**Additional critical observations:**

**No active placebo control:** The waiting-list group received no intervention, so the "attention effect" of coming to group sessions twice weekly cannot be separated from the cognitive training itself. A better design would include a social activity control group (e.g., book club or art class).

**Unbalanced randomisation:** The stratified randomisation by recruitment source created imbalance — more support group patients ended up in the control group. Support group attendees may be more socially engaged and have different disease trajectories than clinic patients, potentially biasing results.

**Multiple comparisons problem:** With 3 group comparisons × multiple cognitive tests, the chance of false positives is elevated. The authors used Bonferroni correction but only within each "trial" (comparison), not across all comparisons.

**No intention-to-treat analysis:** Only patients who attended ≥75% of sessions were included. This excludes those who dropped out, which could bias results toward positive findings (people who stick with training may be more motivated or healthier).

**Industry/funding bias:** Not explicitly stated in the abstract, but the structured program "NEUROvitalis" is a commercial product — potential conflict of interest if authors developed or benefit from it.

**Population limits:** Results apply only to non-demented PD patients with mild-to-moderate disease (Hoehn & Yahr I–III). Not applicable to patients with dementia, advanced disease, or deep brain stimulation.

**Self-report bias for depression:** BDI-II is a self-report questionnaire; patients in the unstructured group may have reported less depression due to social desirability or the group conversations about coping, rather than genuine mood improvement.

**No measure of real-world function:** The study used laboratory cognitive tests, not measures of everyday functioning (e.g., medication management, driving, financial decision-making). It's unclear whether the cognitive improvements translate to daily life.

For someone running their own n=1 experiment to test whether cognitive training improves thinking or mood:

### What to test

**Structured cognitive training** (like NEUROvitalis): Focus on one cognitive domain per session (e.g., Monday = attention exercises, Wednesday = memory strategies, Friday = executive function puzzles). Include a brief "lesson" about how that cognitive function works and strategies to improve it.

**Unstructured "brain jogging"** (like Mentally fit): Random mix of puzzles, word games, and number games without domain-specific focus. Include time for reflection or discussion about challenges.

**Dose:** 90-minute sessions, twice per week (or ~3 hours total per week)

### Minimum meaningful duration

**6 weeks** (12 sessions) — this is what the study used and showed effects

For cognitive effects, do not expect noticeable changes before 4 weeks

For mood effects, some benefit may appear by 3–4 weeks

**No data on optimal duration** — longer may be better, but the study didn't test it

### What to measure (specific metrics)

**Working memory:** Test yourself on digit span backwards (repeat increasingly long sequences of numbers in reverse order). Track the maximum length you can consistently recall. A positive result = improvement of 1+ digits over 6 weeks.

**Short-term memory:** Use a word list learning test (e.g., read 10 words, immediately recall as many as possible). Track number recalled. A positive result = improvement of 1–2 words.

**Depression/mood:** Beck Depression Inventory-II (free online) or a simple 1–10 mood rating daily. A positive result = 3+ point drop on BDI-II or consistent improvement in daily mood ratings.

**Quality of life:** PDQ-39 or a simple 1–10 satisfaction rating weekly

**Attention:** Use a free online reaction time test or Stroop test. Track reaction time and errors.

### Key confounds to control for

**Medication timing:** Test at the same time of day relative to your PD medication (ideally during "on" phase, ~1 hour after medication)

**Sleep quality:** Poor sleep impairs cognition — track sleep duration and quality alongside training

**Exercise:** Physical exercise improves cognition in PD — keep exercise constant or log it as a covariate

**Social engagement:** If you do group training, the social interaction itself may help — consider a solo control period

**Practice effects:** Cognitive tests improve with repetition — use alternate versions of tests or have a 2-week baseline period before starting training

**Expectation:** If you believe training will help,