Validation of a UPDRS-/MDS-UPDRS-based definition of functional dependency for Parkinson's disease
Ramsay N., Macleod AD., Alves G., Camacho M., Forsgren L., Lawson RA., Maple-Grødem J., Tysnes OB., Williams-Gray CH., Yarnall AJ., Counsell CE., Rachael A Lawson None., Guido Alves S., Forgren L., Caslake R., Taylor KSM., McGhee DJM., Gordon J., Harris C., Forbes H., Barker RA., Foltynie T., Mason SL., Cummins G., Evans JR., Breen DP., Wijeyekoon RS., Linder J., Edström M., Andersson J., Eriksson L., Bäckström D., Hariz GM., Domellöf M., Gjerstad MD., Pedersen KF., Forsaa EB., Frantzen VH., Laugaland A., Lange J., Simonsen K., Fiske E., Dalen I., Müller B., Skeie GO., Renså M., Telstad W., Labusau A., Kastet J., HogenEsch I., Kjerandsen M., Håland LK., Herlofson K., Ongre S., Bruun S., Burn D., Rochester L., Duncan GW., Khoo TK.
Introduction: Functional dependency in basic activities of daily living (ADLs) is a key outcome in Parkinson's disease (PD). We aimed to define dependency in PD, using the original and MDS versions of the Unified Parkinson's Disease Rating Scale (UPDRS). Methods: We developed two algorithms to define dependency from items of UPDRS Part 2 and MDS-UPDRS Part 2 relating to basic ADLs (feeding, dressing, hygiene and walking, and getting out of a chair). We validated both algorithms using data from 1110 patients from six community-based PD incidence cohorts, testing concurrent validity, convergent validity, and predictive validity. Results: Our optimal algorithm showed high specificity and moderate to high sensitivity versus Schwab & England <80% (specificity 95% [95% confidence interval (CI) 93–97] and sensitivity 65% [95% CI 55–73] at baseline; 88% [95% CI 85–91] and 85% [95% CI 79–97] respectively at five-years follow-up). Convergent validity was demonstrated by strong associations between dependency defined by the algorithm and cognition (MMSE), quality of life (PDQ39), and impairment (UPDRS part 3) (all p < 0.001). Algorithm-defined dependency status also predicted mortality: HR for mortality in those dependent vs independent at baseline was 1.6 (95%CI 1.2–2.1) and in those dependent vs independent at five-years’ follow-up was 2.2 (1.6–3.0). Discussion: We have demonstrated the concurrent validity, convergent validity, and predictive validity of a UPDRS-/MDS-UPDRS-based algorithm to define functional dependency in PD. This can be used for studying dependency in any study where UPDRS or MDS-UPDRS part 2 data have been collected.