Prevalence and temporal relationship of clinical co-morbidities in idiopathic dystonia: a UK linkage-based study

While motor and psychiatric phenotypes in idiopathic dystonia are increasingly well understood, a few studies have examined the rate, type, and temporal pattern of other clinical co-morbidities in dystonia. Here, we determine the rates of clinical diagnoses across 13 broad systems-based diagnostic groups, comparing an overall idiopathic dystonia cohort, and sub-cohorts of cervical dystonia, blepharospasm, and dystonic tremor, to a matched-control cohort. Using the SAIL databank, we undertook a longitudinal population-based cohort study (January 1st 1994–December 31st 2017) using anonymised electronic healthcare records for individuals living in Wales (UK), identifying those diagnosed with dystonia through use of a previously validated algorithm. Clinical co-morbid diagnoses were identified from primary health care records, with a 10% prevalence threshold required for onward analysis. Using this approach, 54,166 dystonia cases were identified together with 216,574 matched controls. Within this cohort, ten of the main ICD-10 diagnostic codes exceeded the 10% prevalence threshold over the 20-year period (infection, neurological, respiratory, gastrointestinal, genitourinary, dermatological, musculoskeletal, circulatory, neoplastic, and endocrinological). In the overall dystonia cohort, musculoskeletal (aOR: 1.89, aHR: 1.74), respiratory (aOR: 1.84; aHR: 1.65), and gastrointestinal (aOR: 1.72; aHR: 1.6) disorders had the strongest associations both pre- and post-dystonia diagnosis. However, variation in the rate of association of individual clinical co-morbidities was observed across the cervical, blepharospasm, and tremor dystonia groups. This study suggests an increased rate of specific co-morbid clinical disorders both pre- and post-dystonia diagnosis which should be considered during clinical assessment of those with dystonia to enable optimum symptomatic management. Supplementary Information The online version contains supplementary material available at 10.1007/s00415-024-12284-6.


Introduction
Dystonia is a hyperkinetic movement disorder characterised by abnormal movements or postures [1], with an estimated prevalence of 120/100,000 [2].Recent work has highlighted the co-morbid non-motor symptom profile of dystonia, focusing predominantly on psychiatric, cognitive, pain, and sleep symptoms [3].However, little work to date has examined whether those diagnosed with dystonia are at increased risk of other medical disorders, potentially complicating treatment options and care needs, as well as increasing healthcare utilisation and overall burden of disease.
Although dystonia does not reduce life expectancy [2], poor health-related quality of life is consistently reported [4,5].The few studies to date investigating dystonia comorbidities have had varying results.Some have suggested higher rates of thyroid dysfunction [6][7][8] and cardiovascular changes, including heart rate and blood pressure variability [9].Respiratory symptoms have also been recognised, including stridor, dyspnoea, and gasping [10,11].While a more recent population-based study of cervical dystonia found higher rates of cervical spondylosis, softtissue disorders, back pain, essential tremor, neck tension, dental caries, and abdominal and pelvic pain, compared to a control cohort [12].Autoimmune disorders, such as thyroid disease [13], myasthenia gravis [14], and Sjogren's syndrome [15], have also been reported at higher rates, with preliminary mixed immunoassays potentially indicating common autoimmune mechanisms in dystonia [16].Respiratory symptoms have been attributed to dystonic spasms affecting the respiratory muscles, with evidence of upper airway or diaphragmatic dysfunction [17,18].
To date, a few limited studies have examined the nature and rate of medical co-morbidities in dystonia, and none have examined their temporal relationship to dystonia motor symptom onset.Here, we examine a previously identified dystonia and matched-control population cohort with linked clinical data.This work determines those medical co-morbidities present at higher rates amongst those with dystonia, and their temporal pattern relative to dystonia motor symptom onset.Awareness of whether medical co-morbidities are at an excess in dystonia is key in developing more comprehensive clinical care pathways as well as potentially aiding in determining underlying causative mechanisms.

Study design and data sources
Using both retrospective and prospective analyses, the SAIL Databank (Swansea University, UK: www.saild ataba nk.com) was used to investigate the rate of medical co-morbidities in individuals diagnosed with dystonia.SAIL is an anonymised data repository containing health, education, and social care data covering the population of Wales.Clinical data are available from multiple sources, including primary (general practice) and secondary (hospital) care records, with data available for 86% and 100%, respectively, of the Welsh population (3.1 million).

Dystonia diagnosis
Derivation and diagnostic validation of the dystonia cohort has been described elsewhere [2].In brief, a reference population of 90 patients with a clinically confirmed diagnosis of adult-onset idiopathic focal cervical dystonia (AOIFCD) was linked to records in SAIL.Codes relevant to dystonia [Read code-primary care or International Classification of Diseases version 10 (ICD-10) codessecondary care] were reviewed and used to identify 54,966 individuals with idiopathic dystonia throughout the study period (January 1994-December 2017).

Study cohort
The dystonia cohort consisted of 54,166 individuals diagnosed with idiopathic dystonia (800 had no matched controls and were therefore not included) and 216,574 controls matched on: year of birth, sex, year of study entry, year of follow-up, and deprivation index quintile, their derivation has been described elsewhere [2].Controls were randomly matched on a 1:4 (case:control, n = 54,121, 99.9%) basis using individuals with Welsh Longitudinal General Practice (WLGP) records, the remaining 45 cases were matched on a ratio of 1:1-3.Individuals were required to be resident in Wales at the time of index date and have an age, sex, and GP registration date recorded.Controls were assigned the same index date as their matched cases.

Comorbidities
We obtained data on medical co-morbidities using primary care (WLGP) datasets (January 1994 and December 2017).With the exception of mental disorders that have been explored elsewhere [19], we examined all medical diagnostic categories, including: infectious and parasitic diseases, neoplasms, endocrine, nutritional, metabolic and immunity disorders, diseases of blood and blood-forming organs, nervous system and sense organ diseases, circulatory system diseases, respiratory system diseases, digestive system diseases, genitourinary system diseases, complications of pregnancy, childbirth and the puerperium, skin and subcutaneous tissue diseases, musculoskeletal and connective tissue diseases, congenital abnormalities, and perinatal conditions (Supplementary Table 1).To ensure consistent data coverage, only diagnoses recorded 10 years before and after the index date were analysed.

Statistical analysis
During the primary analyses, individuals having received ≥ 1 diagnosis of any disorder within the broader disease categories (main Read code diagnoses) were identified.Only those affecting a minimum of 10% of the dystonia cohort were further analysed to ensure clinical relevance and consistency with approaches used elsewhere (Table 1 and Supplementary Tables 3-5) [20].Subsequently, a nested case-control design was used to examine the association of each co-morbidity and idiopathic dystonia.Logistic regression analysis was used to estimate the risk of developing these co-morbidities during the retrospective time periods, with odds ratios (ORs) and 95% confidence intervals (CIs) calculated and adjusted for contact with medical services and/or clinical professionals.Medical contact was defined as any general practice interaction (diagnosis, symptom, and diagnostic testing) or secondary care contact (in-patient hospital admission and out-patient appointment), with each contact considered an independent event.A cohort study design was used to determine rates of medical diagnoses following dystonia diagnosis, with only those not diagnosed with each co-morbidity at the index date included for onward analysis.Hazard ratios (HR) and 95% CIs for each co-morbidity were calculated using Cox proportional hazard regression, again adjusted for the total number of medical contacts.The follow-up period was censored at the first co-morbid diagnosis, death, and loss to follow-up or 10 years after the index date.Bonferroni correction was used to adjust p values for both retrospective and prospective methods.

Ethics
This study design uses anonymized, routinely collected data, and therefore does not require ethical approval and written informed consent.The SAIL independent Information Governance Review Panel (IGRP), consisting of experts in information governance and members of the public, approved this study (Reference: 0768).

Results
Using the previously described dystonia case ascertainment algorithm, 54,166 dystonia cases were identified together with 216,574 matched controls.Demographic characteristics are provided in Supplementary Table 2. Overall, ten primary ICD-10 codes exceeded the 10% prevalence threshold in both the dystonia and control cohorts over the 20-year period examined (Table 1).Evaluation of the individual dystonia diagnostic categories (cervical dystonia, blepharospasm, and dystonic tremor) identified similar patterns to the overall cohort but with differences observed in the number and type of co-morbid medical disorders observed pre-dystonia diagnosis (Supplementary Tables 3-5).Here, infectious, nervous, respiratory, gastrointestinal (GI), genitourinary, dermatological, and musculoskeletal disorders were observed in those with cervical dystonia (Supplementary Table 3), with the addition of circulatory disorders in the blepharospasm cohort (Supplementary Table 4), and neoplastic and endocrinological disorders in the tremor cohort (Supplementary Table 5).All primary ICD-10 diagnostic groups exceeding the 10% threshold were taken forward for analysis of their relative risk of diagnosis both pre-and postdystonia diagnosis.

Cervical dystonia
Of those diagnosed with cervical dystonia (n = 36,846), the strongest associations were seen across the same three primary ICD-10 diagnostic categories-respiratory, infectious, and musculoskeletal disorders-both pre-and postdystonia diagnosis, with adjusted odds ratios being higher than their adjusted hazard ratio counterpart (Table 2).

Discussion
This study represents the first to systematically examine for the potential risk of other medical disorders, outside of psychiatric diagnoses, within a dystonia cohort using a longitudinal, linked clinical databank.As has been undertaken in the previous studies, application of the ≥ 10% prevalence rate to primary ICD-10 diagnoses enabled this work to focus on disorders with greatest clinical relevance [20], with disorders involving musculoskeletal, respiratory, and nervous system consistently observed across the dystonia diagnostic cohorts.Subsequent calculation of adjusted OR and HR, pre-and post-dystonia diagnosis, respectively, identified higher associations with infectious disorders in cervical dystonia, cutaneous and GI disorders in blepharospasm, and circulatory and GI disorders with dystonic tremor.Analysis of the more specific individual secondary diagnoses identified circulatory system disorders, cerebrovascular disease, and disorders of the digestive system in dystonic tremor, peripheral nervous system and other musculoskeletal or connective tissue disorders  with blepharospasm, and vertebral column syndromes in cervical dystonia.Analysis of adjusted OR and HR consistently observed musculoskeletal disorders to be one of the most common diagnoses to be associated with dystonia.Higher rates of musculoskeletal diagnoses have been observed for some time in dystonia [21], with recognition that initial clinical presentation may be via musculoskeletal and orthopaedic services, potentially contributing to the delay in dystonia diagnosis [22,23].Much of the reported literature date has highlighted this higher burden of musculoskeletal difficulties in those diagnosed with cervical dystonia, identifying disorders such as vertebral subluxation, fractures, and spinal degenerative changes [24,25].These are coupled with risk of secondary complications such as the vertebral column changes observed in the cervical dystonia cohort in this study post-dystonia diagnosis, with more severe complications such as spinal cord compression noted to be of higher risk amongst those with other forms of bone disease such as osteoporosis and rheumatoid arthritis [25,26].A previous Finnish national registry study focusing on comorbidity in those diagnosed with cervical dystonia found odds ratios for cervical intervertebral disc disorders [OR 5.9, (95% CI 3.1-1.1)]to be the highest for any diagnosis outside of psychiatric symptoms, with shoulder [OR 2.26, (95% CI 1.2-4.4)]and intervertebral disc disorders [OR 1.41, (95% CI 0.7-2.7)]being the two most common physical causes for retirement in those with cervical dystonia [12].Respiratory disorders have been previously linked with dystonia, primarily in conjunction with cranio-cervical dystonias and predominantly involving the upper airways, typically in the context of laryngeal, platysmal, and paranasal muscular involvement [10].In relation to infective illness, this has been predominantly described in paediatric dystonia cohorts, primarily in the form of Grisel's syndrome [27], although more recently dystonia has been described in the context of sequelae to COVID-19 infection [28].Disorders associated with the peripheral nervous system were associated at higher rates in the blepharospasm and dystonic tremor groups.This may in part relate to the sensory disturbances well described in dystonic cohorts, and in particular sensory discrimination thresholds as endophenotypes in genetic dystonia cohorts [29].Significant increases to sensory discrimination thresholds have also been more specifically described in dystonia tremor cohorts, distinct from essential tremor, as well as those cranial forms of dystonia, including blepharospasm [30,31].More recent work involving those diagnosed with blepharospasm has also found a reduction in activity of brainstem inhibitory interneurons following stimulation of both the supraorbital nerve and upper limb peripheral nerves [32].Additional neurological diagnoses identified in the blepharospasm cohort included those involving the ear and mastoid processes.Although not frequently reported, a few studies have described hearing loss in dystonia, with 50% of a single GCH1 mutation-positive cohort reported to experience symptoms of hearing loss, as well as there being the rare but recognised X-linked dystonia deafness syndrome [33,34].
Finally, GI disturbance was identified in the overall, blepharospasm and dystonic tremor cohorts, and more specifically involvement of the upper GI tract and digestive system.There are few case reports or case series highlighting these symptoms; however, a recent analysis of the gut microbiome in a dystonia cohort of (27 cervical dystonia, 20 dopa-responsive dystonia, and 24 myoclonus dystonia) identified changes to the gut flora, compared to controls, with this linked to non-motor symptom severity and plasma neurotransmitter levels.In addition, several metabolic pathways linked with the nervous system, notably tryptophan degradation, were present at lower levels in the dystonia cohort [35].By contrast, functional analysis in a distinct cohort of individuals diagnosed with isolated dystonia identified genes related to tryptophan and purine biosynthesis to be at higher levels in the gut microbiota, compared to controls, while serum metabolomic analysis found altered levels of l-glutamic acid, taurine, and d-tyrosine, again linking to changes in neurotransmitter metabolism [36].
Although this study provides a robust examination of potential co-morbid medical diagnoses across a spectrum of idiopathic dystonia diagnoses, this work is limited primarily by clinical detail being constrained to diagnostic codes, with no further specific symptomatic detail that may allow for greater understanding of the nature, variability, and severity of the associated phenotypes.In addition, we have used only primary care diagnostic codes, leading to the potential risk of coding errors or diagnoses being missed.Although primary care diagnostic codes reflect those entered in the primary care setting, in the instance of more specialised diagnoses, such as dystonia, this is information is typically entered after correspondence from secondary or tertiary specialist care services.In addition, delays in dystonia diagnosis are well recognised, which may have impacted our analyses, particularly in determining those diagnoses identified at being at increased rates prior to dystonia diagnosis.Finally, although, during our analyses, we have corrected for the number of individual clinical contacts across both primary and secondary care sectors, there remains potential for referral bias which too may be reflected in the diagnoses identified both pre-and post-dystonia diagnosis.
In summary, this study represents the first to systematically undertake a longitudinal evaluation of the rate and type of co-morbid clinical diagnoses in idiopathic dystonia, consistently demonstrating higher rates of respiratory, musculoskeletal, and nervous system disorders, with higher prevalence of rates prior to dystonia diagnosis rather than in subsequent years.This work highlights the need for greater awareness of these disorders in the context of dystonia, with specific evaluation to ensure identification and appropriate management.Recognition of these increased rates of comorbidity also represents an important component of any future care pathways designed for the comprehensive management of idiopathic dystonia.
by Welsh Government through Health and Care Research Wales.KJP is funded by an MRC Clinician-Scientist Fellowship (MR/P008593/1) and an MRC Transition Fellowship (MR/V036084/1).

Table 1
Frequency of co-morbid diagnoses within main diagnostic categories

Table 2
Adjusted Odds and Hazard Ratios for overall and cervical dystonia groups Table only includes disorders where statistically significant (p < 0.001) difference was observed between dystonia and control cohorts aHR adjusted hazard ratio, aOR adjusted odds ratio, 95% CI 95% confidence interval, GI gastrointestinal system, GU genitourinary system, ID infectious diseases, MSK musculoskeletal system

Table 3
Adjusted Odds and Hazard Ratios for blepharospasm and dystonic tremor groups Table only includes disorders where statistically significant (p < 0.001) difference was observed between dystonia and control cohorts aHR adjusted hazard ratio, aOR adjusted odds ratio, 95% CI 95% Confidence Interval, GI gastrointestinal system, GU genitourinary system,

Table 4
Adjusted Odds and Hazard Ratios for sub-diagnoses in the highest three main diagnostic groups across overall, cervical, blepharospasm and tremor dystonia cohorts