Frozen shoulder is a severely painful condition that causes stiffness and a limited range of motion of the shoulder joint and may lead to a reduced quality of life. The condition is around five-times more likely to occur in people with Diabetes Mellitus (DM) compared to those without. This study will estimate the proportion of people in the UK with DM that develop frozen shoulder, and we will investigate some reasons why people with DM are more likely to develop frozen shoulder; such as, through coexisting comorbidities. We will also estimate the proportion of people that are newly diagnosed with DM after already being diagnosed with frozen shoulder, then we will test whether this rate is higher than a population without frozen shoulder. We will explore whether people with DM are more likely to require/receive surgery for frozen shoulder than those without DM. This will help indicate whether people with DM have worse outcomes from frozen shoulder than those without DM.
Despite the prevalence of frozen shoulder being around five times more in people with DM than people without DM (1), there is still a lack of high-quality studies investigating how DM and frozen shoulder are associated. My recent systematic review has highlighted that poor adjustment for confounding and insufficient follow-up length have compromised the quality of evidence for DM as a risk factor for frozen shoulder. We will use time-to-event models to investigate the association between DM and frozen shoulder, whilst ensuring suitable adjustment for confounding (see section P) and attempting to account for missing data (see section Q). Causal mediation analysis using weighted Cox regression will be used to investigate the extent to which the DM and frozen shoulder relationship is mediated through other variables (see section N for potential mediators).
A meta-analysis of five studies has shown that the prevalence of DM in the frozen shoulder population is around 30% (1). However, my recent systematic review found only two studies that have investigated whether the presence of frozen shoulder could be indicative of a subsequent diabetes diagnosis. We will estimate the incidence of DM after a diagnosis of frozen shoulder and compare this to a population without frozen shoulder.
We will also investigate whether people with DM have worse outcomes compared to those without DM. To do this, we will test whether frozen shoulder patients with DM are more likely to require/receive surgery than those without DM using Cox regression. Since initially frozen shoulder is treated using conservative methods, and surgery is reserved only for stubborn cases of frozen shoulder, this may help to indicate whether the outcomes of frozen shoulder are worse in those with DM.
Incident frozen shoulder and time to incident frozen shoulder (objective 1); Incident Diabetes Mellitus and time to incident Diabetes Mellitus (objective 2); Frozen shoulder surgery and time to frozen shoulder surgery (objective 3).
Note: Objectives are listed below in section E.
Danielle van der Windt - Chief Investigator - Keele University
Brett Dyer - Corresponding Applicant - Keele University
Claire Burton - Collaborator - Keele University
Milica Bucknall - Collaborator - Keele University
Trishna Rathod-Mistry - Collaborator - Keele University
HES Admitted Patient Care;Patient Level Index of Multiple Deprivation