As many as 89% of people living with Parkinson’s disease develop speech signs (Ramig et al., Movement Disorders 2018), which makes voice and speech one of the most widely shared consequences of a condition projected to reach 25.2 million people worldwide by 2050 (Su et al., BMJ 2025). The treatment evidence is stronger than it has ever been: in the 388-person PD COMM trial, LSVT LOUD cut Voice Handicap Index scores by 8.0 points against no therapy (Sackley et al., BMJ 2024). Access is the weak link, with 14.6% of US Medicare beneficiaries with Parkinson’s holding a speech therapy claim in the largest national analysis (Fullard et al., Neurology 2017) and 45.4% of UK speech and language therapy services offering LSVT in 2025 (Parkinson’s UK, UK Parkinson’s Audit 2025). We aggregated data from the Parkinson’s Foundation, the World Health Organization, the BMJ, Parkinson’s UK, the Michael J. Fox Foundation, the Cochrane Library, and more than a dozen peer-reviewed studies. For neighbouring clinical voice topics, see our roundups on acoustic voice analysis and dysphonia and voice prosthesis and laryngectomy rehabilitation.
This article reports published statistics only. It is not medical advice.
TL;DR
- About 1.1 million people in the US live with Parkinson’s, and nearly 90,000 are diagnosed each year (Parkinson’s Foundation, Statistics page 2026).
- Global cases are projected to rise from 11.9 million in 2021 to 25.2 million in 2050, a 112% increase (Su et al., BMJ 2025).
- As many as 89% of people with Parkinson’s develop speech signs (Ramig et al., Movement Disorders 2018).
- 69.6% of 125 speakers with Parkinson’s scored below the intelligibility mean of unaffected controls (Miller et al., JNNP 2007).
- 14.6% of 174,643 US Medicare beneficiaries with Parkinson’s had a speech therapy claim in 2007 (Fullard et al., Neurology 2017).
- UK speech and language therapy services offering LSVT fell from 61.0% in 2022 to 45.4% in 2025 (Parkinson’s UK, UK Parkinson’s Audit 2025).
- LSVT LOUD lowered Voice Handicap Index scores by 8.0 points versus no therapy at 3 months; standard NHS therapy showed no evidence of benefit (Sackley et al., BMJ 2024).
- A meta-analysis of eight trials found LSVT raised reading loudness by 4.34 dB over no intervention after 1 month of treatment (Yuan et al., European Journal of Neurology 2020).
- Less than 15% of 69 voice-based Parkinson’s detection studies used external or cross-dataset validation (Sedigh Malekroodi et al., Bioengineering 2025).
- Parkinson’s and atypical parkinsonisms cost the US $82.2 billion in 2024 (Michael J. Fox Foundation, Economic Burden report 2026).
- LSVT LOUD was not found to be cost-effective in the UK, at £197,772 per QALY gained versus no therapy (NIHR Health Technology Assessment, PD COMM report 2024).
1. Prevalence, Incidence and Projections
Every count of Parkinson’s is a moving target, and the direction of travel is the same in all of them. The US incidence estimate rose from a long-quoted 40,000 to 60,000 diagnoses a year to nearly 90,000 once five North American cohorts were pooled (Parkinson’s Foundation, Statistics page 2026). Globally, cases are projected to more than double to 25.2 million by 2050 (95% UI 21.7 to 30.1 million), and 89% of that growth comes from population ageing rather than any change in underlying risk (Su et al., BMJ 2025).
For speech services the implication is arithmetic. If roughly the same share of people develop voice and speech changes (section 2), caseloads grow in step with the population curve, and the over-80 group, where case counts are projected to rise 196% by 2050, is the fastest-growing part of it.
| Metric | Value | Source |
|---|---|---|
| People living with Parkinson’s in the US | About 1.1 million (1,112,643) | Parkinson’s Foundation, Statistics page 2026 |
| New US diagnoses per year | Nearly 90,000 | Parkinson’s Foundation, Statistics page 2026 |
| Incidence, North American adults aged 65+ | 108 to 212 per 100,000 | Willis et al., npj Parkinson’s Disease 2022 |
| Prevalence, North American adults aged 45+ | 572 per 100,000 (95% CI 537 to 614) | Marras et al., npj Parkinson’s Disease 2018 |
| Projected US cases aged 45+ in 2030 | 1,238,000 | Marras et al., npj Parkinson’s Disease 2018 |
| Global cases in 2021 | 11.9 million | Su et al., BMJ 2025 |
| Global cases projected for 2050 | 25.2 million, a 112% increase | Su et al., BMJ 2025 |
| Male-to-female ratio of age-standardised prevalence | 1.46 in 2021, projected 1.64 in 2050 | Su et al., BMJ 2025 |
Context: the World Health Organization fact sheet uses an older baseline of over 8.5 million people in 2019, with 5.8 million disability-adjusted life years (up 81% since 2000) and 329,000 deaths. The Parkinson’s Foundation adds that about 4% of people are diagnosed before age 50 and that men are 1.5 times more likely to have the disease than women. Most recent available data for North American prevalence and incidence: Marras et al., 2018 and Willis et al., 2022, the latter counting diagnoses made in 2012. The global projections are summarised in the BMJ Group release on Su et al.
2. How Common Speech and Voice Changes Are
The 89% figure describes how many people develop speech signs over the course of the disease. The more useful numbers describe how far speech has moved by the time someone is measured. In a sample of 125 speakers with Parkinson’s, 48% were perceived as worse than the lowest-scoring unaffected speaker (Miller et al., JNNP 2007).
The loudness gap itself is small on paper. In the Ramig trial, people with Parkinson’s read aloud at 71.0 dB against 72.6 dB for healthy controls, a difference of 1.6 dB (72.6 minus 71.0). The functional gap is far wider: their median communicative effectiveness score was 62 against 84 for controls. A modest acoustic deficit, in other words, travels with a large perceived loss of communication.
| Metric | Value | Source |
|---|---|---|
| People with Parkinson’s who develop speech signs | As many as 89% | Ramig et al., Movement Disorders 2018 |
| Scored below the control mean for intelligibility | 69.6% (87 of 125) | Miller et al., JNNP 2007 |
| Scored more than 1 SD below the control mean | 51.2% (64 of 125) | Miller et al., JNNP 2007 |
| Perceived as worse than the lowest unaffected speaker | 48% (60 of 125) | Miller et al., JNNP 2007 |
| Placed speech changes among their top four concerns | 38% (47 of 125) | Miller et al., JNNP 2007 |
| Early untreated patients with some vocal impairment | 78% (23 patients studied) | Rusz et al., JASA 2011 |
| Baseline reading loudness, Parkinson’s vs healthy controls | 71.0 dB vs 72.6 dB SPL at 30 cm (P = 0.04) | Ramig et al., Movement Disorders 2018 |
| Median communicative effectiveness (CETI-M), Parkinson’s vs controls | 62 vs 84 (P = 0.0001) | Ramig et al., Movement Disorders 2018 |
Outlier note: voice changes appear early. In a Czech study of 23 people with early, untreated Parkinson’s, 78% already showed some form of vocal impairment (Rusz et al., JASA 2011), which is why voice is being studied as a detection signal (section 6). An older patient survey found 70% of respondents with Parkinson’s reported speech or voice impairment after onset, and 29% named speech among their greatest problems (Hartelius and Svensson, Folia Phoniatrica et Logopaedica 1994).
Most recent available data: the prevalence studies in this section date from 1994 to 2018, and no large population survey of Parkinson’s speech symptoms has replaced them. The intelligibility figures come from Miller et al., JNNP 2007.
3. Who Actually Receives Speech Therapy
Prevalence near 89% and therapy use in the teens is the central mismatch of this topic. The largest US dataset, 174,643 Medicare beneficiaries with Parkinson’s, found 14.6% had a speech therapy claim in 2007, and about 75% received no rehabilitation therapy of any kind across three years (Fullard et al., Neurology 2017). Race changed the odds sharply: African American beneficiaries had 8.2% speech therapy use, with an adjusted odds ratio of 0.63 (95% CI 0.59 to 0.67) against Caucasian beneficiaries.
The UK data show the opposite problem: referral routes exist, but intensive treatment capacity is shrinking. Across 97 speech and language therapy services and 1,206 audited patients, the share of services offering LSVT fell by more than 15 percentage points in three years, and the report notes that around half of services can offer it only to a small number of people (Parkinson’s UK, UK Parkinson’s Audit 2025). The trial evidence in section 4 was published in between those two audit cycles.
| Metric | Value | Source |
|---|---|---|
| US Medicare beneficiaries with Parkinson’s with a speech therapy claim | 14.6% of 174,643 (2007) | Fullard et al., Neurology 2017 |
| Speech therapy use, African American beneficiaries | 8.2% (AOR 0.63, 95% CI 0.59 to 0.67) | Fullard et al., Neurology 2017 |
| Beneficiaries with no rehabilitation therapy over 3 years | About 75% | Fullard et al., Neurology 2017 |
| UK services offering LSVT | 45.4% in 2025, down from 61.0% in 2022 | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| UK referrals seen within 8 weeks | 59.2% | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| UK referrals made at the diagnosis stage | 12.7% (66.2% at the maintenance stage) | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| Patient survey respondents reporting access to speech and language therapy | 32.9% in 2025, down from 47.8% in 2019 | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| Survey respondents with Parkinson’s who had received any speech therapy | 3% | Hartelius and Svensson, Folia Phoniatrica et Logopaedica 1994 |
Context: specialist contact predicts therapy. In the Fullard analysis, 67% of beneficiaries saw a neurologist at least once a year, and those with more than three neurology visits a year had an adjusted odds ratio of 1.81 (95% CI 1.65 to 1.99) for a speech therapy evaluation. The Parkinson’s Foundation reports that 9% of people with Parkinson’s in the US see a movement disorders specialist. Most recent available data: no newer national US claims study of speech therapy use in Parkinson’s was found, so the 14.6% figure should be read as a 2007 baseline rather than a current rate.
Where therapy does reach people, satisfaction is high: 88.2% of UK patient survey respondents rated the quality of speech and language therapy services as good or excellent, with 55.0% choosing excellent (UK Parkinson’s Audit 2025). For people whose speech can no longer carry a conversation, funding rules for communication devices are covered in our speech-generating device insurance statistics, and the same referral-timing problem shows up at the other end of the age range in pediatric speech delay and early intervention.
4. Treatment Trial Outcomes: LSVT LOUD and Standard Therapy
The 2012 Cochrane position was that three small trials with 63 participants were not enough to support or refute speech and language therapy in Parkinson’s. Two larger trials have since changed the picture, and they point the same way: intensity and a single loudness target matter more than therapy contact in general.
In the US trial, the voice-focused and articulation-focused treatments each ran for one month, yet only the voice-focused arm moved loudness: reading loudness rose 6.3 dB with LSVT LOUD against 1.3 dB with LSVT ARTIC (Ramig et al., Movement Disorders 2018). In the UK, PD COMM randomised 388 people across 41 centres and found that standard NHS speech and language therapy, delivered at much lower intensity, was not distinguishable from no therapy on the primary outcome.
| Metric | Value | Source |
|---|---|---|
| Reading loudness gain at 1 month and 7 months, LSVT LOUD | +6.3 dB and +3.9 dB | Ramig et al., Movement Disorders 2018 |
| Reading loudness gain at 1 month, comparison arms | +1.3 dB LSVT ARTIC, +0.5 dB untreated | Ramig et al., Movement Disorders 2018 |
| Pooled LSVT advantage over no intervention after 1 month of treatment | 8.87 dB sustained ‘Ah’, 4.34 dB reading, 3.25 dB monologue | Yuan et al., European Journal of Neurology 2020 |
| Voice Handicap Index at 3 months, LSVT LOUD vs no therapy | -8.0 points (99% CI -13.3 to -2.6; P<0.001) | Sackley et al., BMJ 2024 |
| Voice Handicap Index at 3 months, NHS therapy vs no therapy | 1.7 points (99% CI -3.8 to 7.1; P=0.43) | Sackley et al., BMJ 2024 |
| Participants reporting adverse events, mostly vocal strain | 28% LSVT LOUD (36 of 130), 12% NHS therapy (16 of 129), none in control | Sackley et al., BMJ 2024 |
| Cochrane review of therapy versus placebo or no intervention | 3 trials, 63 participants: insufficient evidence | Herd et al., Cochrane Database of Systematic Reviews 2012 |
The head-to-head comparison in PD COMM favoured LSVT LOUD over NHS therapy by 9.6 points (99% CI -14.9 to -4.4; P<0.001), and the mean Voice Handicap Index score in the LSVT LOUD arm moved from 44.6 at baseline to 35.0 at 3 months. Only 59% of the LSVT LOUD arm (77 of 130) and 54% of the NHS therapy arm (70 of 129) met the trial’s adherence definition, and 84% (109 of 130) of those allocated to LSVT LOUD started it at all.
Three caveats sit behind those rows. PD COMM stopped at 71% of its 546-participant target, most participants were in the early stages of the disease with mild speech impairment, and no serious adverse events were recorded. In the Ramig trial, communicative effectiveness (CETI-M) improved for both treatments against the untreated group at 1 month (P = 0.02 each), but the between-group difference was no longer significant at 7 months (P = 0.08). And the primary outcome there, loudness, is the thing LSVT LOUD trains directly, a bias the authors themselves flag.
The Cochrane review predates both trials (its search covered studies published before 11 April 2011), so its conclusion describes the evidence base as it stood then. Full acoustic results are in Ramig et al., 2018.
5. Telehealth and Remote Delivery
Remote delivery is the obvious answer to an intensive protocol that asks for four visits a week, and the trial evidence, while small, does not show a penalty. The eight-trial meta-analysis reported no difference between face-to-face and online LSVT (Yuan et al., European Journal of Neurology 2020), and an earlier noninferiority trial reached the same conclusion with 31 randomised participants (Theodoros et al., American Journal of Speech-Language Pathology 2016).
Service data tell a less tidy story. In the UK audit, video consultations fell from 78.1% in 2022 to 68.0% in 2025 while telephone consultations rose from 68.8% to 76.3%, a drift from video back toward the phone. Only 6.7% of audited patients experienced technology in their sessions.
| Metric | Value | Source |
|---|---|---|
| Video consultations, UK speech and language therapy services | 68.0% in 2025, down from 78.1% in 2022 | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| Telephone consultations, UK speech and language therapy services | 76.3% in 2025, up from 68.8% in 2022 | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| PD COMM centres with access to remote LSVT companion software | 21; used for 7 LSVT LOUD participants at 5 sites | Sackley et al., BMJ 2024 |
| Pilot trial of online vs face-to-face LSVT LOUD: retention | 95% (95% CI 75.1 to 99.9) of 20 randomised | Kratz et al., BMC Neurology 2025 |
| Same pilot: adherence to supervised sessions | 99.1% (95% CI 68.7 to 99.6) | Kratz et al., BMC Neurology 2025 |
| Same pilot: text-reading loudness change | +1.80 dB online (n = 10) vs +2.92 dB face-to-face (n = 9) | Kratz et al., BMC Neurology 2025 |
| Noninferiority trial of home-based online LSVT LOUD | 31 randomised, plus 21 nonmetropolitan participants treated online | Theodoros et al., AJSLP 2016 |
| Australian speech-language pathologists providing online services to people with Parkinson’s | 23.1% of 63 surveyed (82.5% interested) | Swales et al., International Journal of Speech-Language Pathology 2020 |
Context: the 2025 Austrian pilot was a feasibility study with 19 completers. Its loudness results were mixed across tasks, and the sustained vowel task favoured face-to-face delivery (Hedges’ g = 1.416), so the authors state that further research is warranted (Kratz et al., 2025). The Australian survey dates from 2019 to 2020 and reflects practice at that time. Details of the earlier noninferiority trial are in Theodoros et al., 2016.
6. Voice as a Digital Biomarker
Because voice changes are measurable early (78% of early untreated patients in section 2), a phone recording looks like an ideal screening tool. The published accuracy numbers are high. The gap between those numbers and field performance is the real finding.
A 2025 systematic review covered 69 studies of voice-based Parkinson’s detection and reported accuracy from 68.56% up to 100%, with the top scores on small datasets. Less than 15% of those studies used external or cross-dataset validation, almost two-thirds relied on a single dataset, and most datasets held fewer than 100 participants (Sedigh Malekroodi et al., Bioengineering 2025). When one team tested telephone-quality recordings at scale, with 1,483 people with Parkinson’s and 8,300 controls across seven countries, sensitivity and specificity both came in below 70%.
| Metric | Value | Source |
|---|---|---|
| Studies in the 2025 systematic review (2020 to September 2025) | 69 | Sedigh Malekroodi et al., Bioengineering 2025 |
| Reported classification accuracy range | 68.56% to 100% | Sedigh Malekroodi et al., Bioengineering 2025 |
| Studies with external or cross-dataset validation | Less than 15% | Sedigh Malekroodi et al., Bioengineering 2025 |
| AUC range across the 38 studies reporting it | 0.73 to 1.00 | Sedigh Malekroodi et al., Bioengineering 2025 |
| Pooled sensitivity and specificity, speech biomarkers for movement disorders (20 studies, n = 852) | 0.81 (95% CI 0.76 to 0.85) and 0.83 (95% CI 0.78 to 0.88) | Yang et al., BMJ Evidence-Based Medicine 2026 |
| Telephone-quality screening, 1,483 Parkinson’s vs 8,300 controls | Sensitivity 64.90% (SD 2.90%), specificity 67.96% (SD 2.90%) | Arora et al., JASA 2019 |
| Voicemail recordings with a pre-trained image network, 50 Parkinson’s vs 50 controls | AUC 0.97 | Iyer et al., Scientific Reports 2023 |
Outlier note: the 0.97 result and the sub-70% result both used telephone audio. The difference is scale: 100 recruited participants (81 analysed after preprocessing, 40 with Parkinson’s plus 41 controls) against 9,783 (1,483 plus 8,300). The pooled estimate from the BMJ Evidence-Based Medicine meta-analysis sits between them, and that review notes 50 of its 96 included studies had fewer than 100 participants.
The Bioengineering review lists sustained vowel phonation as the most common recording task, used in 21 of the 29 datasets it catalogued. Related clinical measures are covered in our acoustic voice analysis statistics. Large-scale telephone results are in Arora et al., 2019 and the voicemail model in Iyer et al., 2023.
7. Cost, Economic Burden and Value for Money
The US cost of Parkinson’s has already passed a level that was forecast for 2037. The 2019 burden study estimated $52 billion for 2017 and projected $79 billion for 2037; the updated report puts 2024 costs at $82.2 billion, with 71% of that (58.4 divided by 82.2) classed as indirect and non-medical costs rather than medical care (Michael J. Fox Foundation, Economic Burden report 2026).
Against that backdrop, the one formal economic evaluation of intensive voice treatment is sobering. LSVT LOUD was more effective and more expensive than both comparators in PD COMM, and it was not found to be cost-effective. Dose explains much of the cost: participants received a mean of 1,216 minutes of LSVT LOUD against 404 minutes of standard NHS therapy.
| Metric | Value | Source |
|---|---|---|
| Total US cost of Parkinson’s and atypical parkinsonisms, 2024 | $82.2 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| Direct medical vs indirect and non-medical costs | $23.8 billion vs $58.4 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| Federal government cost; care partner lost earnings and productivity | More than $25 billion; $8.3 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| Projected annual cost by 2045 without intervention | More than $112 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| LSVT LOUD cost per QALY gained vs no therapy | £197,772 | NIHR Health Technology Assessment, PD COMM report 2024 |
| LSVT LOUD cost per QALY gained vs NHS therapy | £77,017 | NIHR Health Technology Assessment, PD COMM report 2024 |
| Mean therapy time delivered per participant | 1,216 minutes LSVT LOUD vs 404 minutes NHS therapy | Sackley et al., BMJ 2024 |
Context: Medicare and Medicaid pay approximately 90% of direct medical costs, and nearly 40% of people with Parkinson’s relied on unpaid care partners, according to the Parkinson’s Foundation summary of the burden report. The Foundation’s statistics page puts average medication costs at $2,500 a year. The cost-effectiveness result is UK-specific and is reported in the PD COMM Health Technology Assessment monograph; no equivalent US analysis of speech therapy for Parkinson’s was found.
Summary: Parkinson’s Speech and Voice Treatment by the Numbers
| Metric | Value | Source |
|---|---|---|
| People living with Parkinson’s in the US | About 1.1 million | Parkinson’s Foundation, Statistics page 2026 |
| New US diagnoses per year | Nearly 90,000 | Parkinson’s Foundation, Statistics page 2026 |
| Global cases, 2021 to 2050 | 11.9 million to 25.2 million | Su et al., BMJ 2025 |
| Share of projected growth due to population ageing | 89% | Su et al., BMJ 2025 |
| People with Parkinson’s who develop speech signs | As many as 89% | Ramig et al., Movement Disorders 2018 |
| Speakers below the control intelligibility mean | 69.6% | Miller et al., JNNP 2007 |
| Early untreated patients with vocal impairment | 78% | Rusz et al., JASA 2011 |
| US Medicare beneficiaries with a speech therapy claim | 14.6% (2007) | Fullard et al., Neurology 2017 |
| UK services offering LSVT | 45.4% in 2025 (61.0% in 2022) | Parkinson’s UK, UK Parkinson’s Audit 2025 |
| Respondents who had received any speech therapy, 1994 survey | 3% | Hartelius and Svensson, Folia Phoniatrica et Logopaedica 1994 |
| LSVT LOUD reading loudness gain at 1 month | +6.3 dB (untreated +0.5 dB) | Ramig et al., Movement Disorders 2018 |
| Voice Handicap Index at 3 months vs no therapy | -8.0 points LSVT LOUD (P<0.001); 1.7 points NHS therapy (P=0.43) | Sackley et al., BMJ 2024 |
| Pooled LSVT reading advantage over no intervention | 4.34 dB | Yuan et al., European Journal of Neurology 2020 |
| Cochrane review of therapy versus placebo or no intervention | 3 trials, 63 participants | Herd et al., Cochrane Database of Systematic Reviews 2012 |
| Online vs face-to-face LSVT LOUD pilot retention | 95% | Kratz et al., BMC Neurology 2025 |
| Voice detection studies with external validation | Less than 15% of 69 | Sedigh Malekroodi et al., Bioengineering 2025 |
| Telephone-quality voice screening at scale | Sensitivity 64.90%, specificity 67.96% | Arora et al., JASA 2019 |
| Total US cost of Parkinson’s and atypical parkinsonisms, 2024 | $82.2 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| Projected annual US cost by 2045 | More than $112 billion | Michael J. Fox Foundation, Economic Burden report 2026 |
| LSVT LOUD cost per QALY gained vs no therapy | £197,772 | NIHR Health Technology Assessment, PD COMM report 2024 |
Methodology and Sources
Figures were collected in October 2026 from primary publications, official reports and organisational statistics pages. Each number was read in the source document or its indexed abstract. Where a figure is older than three years it is flagged in the text as most recent available data. No VoxBooster data is used.
- Parkinson’s Foundation, Statistics page (2026): US prevalence, incidence, demographics, care access, medication costs.
- Marras et al., Prevalence of Parkinson’s disease across North America, npj Parkinson’s Disease 2018: prevalence per 100,000 and 2030 projection.
- Willis et al., Incidence of Parkinson disease in North America, npj Parkinson’s Disease 2022: incidence ranges by age group.
- BMJ Group release on Su et al., BMJ 2025;388:e080952: global projections to 2050; baseline figures read from the full paper.
- World Health Organization, Parkinson disease fact sheet: 2019 global prevalence, DALYs and deaths.
- Ramig et al., Speech treatment in Parkinson’s disease: randomized controlled trial, Movement Disorders 2018: loudness and CETI-M outcomes, baseline comparisons.
- Sackley et al., PD COMM trial, BMJ 2024;386:e078341: Voice Handicap Index outcomes, dose, adverse events, remote delivery.
- NIHR Health Technology Assessment 2024;28(58), The effect of two speech and language approaches on speech problems in people with Parkinson’s disease: the PD COMM RCT: cost per QALY results.
- Yuan et al., Lee Silverman Voice Treatment for dysarthria in Parkinson’s disease: systematic review and meta-analysis, European Journal of Neurology 2020: pooled loudness differences.
- Herd et al., Speech and language therapy versus placebo or no intervention, Cochrane Database of Systematic Reviews 2012: evidence base as of its April 2011 search cutoff.
- Parkinson’s UK, Transforming care: report of the 2025 UK Parkinson’s Audit (2026): LSVT availability, waits, referral stage, consultation modes, patient survey.
- Fullard et al., Utilization of rehabilitation therapy services in Parkinson disease in the United States, Neurology 2017: Medicare speech therapy claims.
- Miller et al., Prevalence and pattern of perceived intelligibility changes in Parkinson’s disease, JNNP 2007: intelligibility prevalence.
- Hartelius and Svensson, Speech and swallowing symptoms associated with Parkinson’s disease and multiple sclerosis: a survey, Folia Phoniatrica et Logopaedica 1994: patient survey of speech symptoms and therapy receipt.
- Rusz et al., Quantitative acoustic measurements for characterization of speech and voice disorders in early untreated Parkinson’s disease, Journal of the Acoustical Society of America 2011: early vocal impairment.
- Kratz et al., Comparing face-to-face and online LSVT LOUD speech training, BMC Neurology 2025: pilot trial feasibility and loudness changes.
- Theodoros et al., American Journal of Speech-Language Pathology 2016: noninferiority trial of home-based online treatment.
- Swales et al., International Journal of Speech-Language Pathology 2020: Australian telepractice survey.
- Sedigh Malekroodi et al., Voice-based detection of Parkinson’s disease: a systematic review, Bioengineering 2025: detection study characteristics.
- Yang et al., AI-driven speech biomarkers for disease diagnosis and monitoring, BMJ Evidence-Based Medicine 2026: pooled diagnostic accuracy.
- Arora et al., Journal of the Acoustical Society of America 2019: telephone-quality screening at scale.
- Iyer et al., Scientific Reports 2023: voicemail-based detection model.
- Michael J. Fox Foundation economic burden report, as summarised by the Parkinson’s Foundation (March 26, 2026): 2024 US costs and projections.
- Data watch: prevalence figures conflict by source and method. The Parkinson’s Foundation counts about 1.1 million people in the US today, while Su et al. (BMJ 2025) estimate 0.84 million for high-income North America in 2021 and project 0.9 million for the US in 2050; Marras et al. projected 930,000 US cases aged 45 and over for 2020. Globally, WHO cites over 8.5 million (2019), the Parkinson’s Foundation cites more than 10 million, and Su et al. count 11.9 million (2021). The US incidence headline of nearly 90,000 a year is the Parkinson’s Foundation figure; the Willis paper itself projects 86,000 diagnoses for 2020 and proposes a working rate of 62 per 100,000 person-years at ages 45 and over. The 89% speech-signs figure is cited by Ramig et al. and Yuan et al. from earlier literature rather than measured in those studies. The UK audit report does not state the denominator for its video and telephone consultation percentages or for its patient-survey access figure. The 2022 audit summary report split its LSVT figure into 34.4% of services offering it to all eligible candidates and 26.6% offering a limited service, which sum to the 61.0% quoted for 2022; the 2025 report text gives no equivalent split for 45.4%. The PD COMM cost per QALY figures are taken from the monograph abstract, which lists them in comparator order (no therapy, then NHS therapy). The US Medicare speech therapy figure dates from 2007 claims and the Australian telepractice survey dates from 2019 to 2020; newer national data on both are overdue.
Last updated: October 11, 2026. We update this roundup quarterly, and the next refresh is expected when the Global Burden of Disease collaborators publish their next round of Parkinson’s disease estimates.