Diabetes (DM) has many well-known and well-understood complications, such as diabetic retinopathy, nephropathy, and polyneuropathy, which are currently screened for during the periodic check-ups [1
]. However, studies have shown that DM is also associated with an increased prevalence of numerous musculoskeletal (MSK) disorders [2
], including shoulder, hand, and lower extremity disorders [6
], which are currently not screened for.
The pathophysiological mechanism of MSK disorders is not fully understood, but evidence suggests that increased accumulation of advanced glycation end-products (AGEs) plays an important role [1
]. AGEs are formed by the non-enzymatic condensation of metabolic intermediates and glucose, and this process is increased or stimulated in chronic hyperglycemia. AGE accumulation occurs in connective tissue causing damage to the tendons, joint capsule, ligaments and nerves, which leads to structural and functional deterioration. AGE formation leads to collagen disposition in the periarticular connective tissue causing the damage [9
Several international studies have shown that MSK disorders have an increased prevalence in T2DM. For example, the prevalence of frozen shoulder ranges from 5% to 30% in patients with DM and from 2% to 5% in patients without DM [6
]. The wide range in observed prevalence in patients with DM might be caused by differences in used study methods, i.e., database and questionnaire studies. Moreover, patients with T2DM have higher odds if developing MSK disorders compared to patients without DM (OR 1.1, 95% CI:1.0, 1.3), and age seems to be associated: patients with DM aged <60 years have a higher odd ratio than those without DM aged <60 years (OR 1.6, 95% CI:1.2, 2.2) [12
Inadequate management of MSK disorders leads to a decrease in functional ability, an increasingly inactive lifestyle and a poorer quality of life; factors that undermine DM treatment [13
]. Therefore, it is important to diagnose and treat MSK disorders at an early stage. Even though international studies have already concluded that physical examination of the hand and shoulder should be included in the evaluation of DM patients [15
], this is currently not implemented in the Dutch nor international guidelines. To better inform developers of guidelines and policymakers, the next step is to investigate the prevalence of MSK disorders in the Netherlands. In the well-organized Dutch healthcare system, almost all patients with T2DM are treated by care groups composed of general practitioners (GPs) and trained diabetes nurses. Routine check-up visits including screening for complications take place at least once a year [16
]. Therefore, it is questionable if the prevalence is as high as reported in the international literature.
Our cross-sectional study aims to investigate the prevalence of overall and specific upper extremity MSK disorders in patients with T2DM in the Netherlands using two approaches, namely (1) a primary care medical database study and (2) a questionnaire that was handed out only to patients with T2DM attending their routine DM check-ups in general practice. The reason for combining two methods instead of selecting just one is that a much better insight into the MSK prevalence and its potential modifiers can be achieved.
We carried out a cross-sectional study composed of the two different approaches. The study was approved by the Medical Ethics Committee of Zuyderland Medical Centre (METC-Z 17-T-138, date of approval: 23 November 2017).
2.1. Approach 1 (RNFM Database)
We used data from the Research Network Family Medicine (RNFM), which is a large anonymized medical database of a GP network in the Maastricht University region, the Netherlands. RNFM was developed in 1988 and reflects the national healthcare system where patients are registered with a GP and access all healthcare through their GP. The network consists of 65 GPs from 22 GP practices. RNFM is composed of computerized medical data of approximately 105,000 patients (reference year 2017). Current and relevant past health problems (i.e., diseases, diagnoses and prescribed drugs of all patients) are recorded systematically and updated continuously along with the basic sociodemographic characteristics of the patients. Registration of these medical data is part of daily routine in the participating GP practices, and every three months registered health problems are added and uploaded to the RNFM database [17
]. The International Classification of Primary Care (ICPC) is used to code and register health problems. Next to the ICPC code for T2DM (T90.02), the following ICPC codes were used to define upper extremity MSK disorders: Shoulder symptoms/complaints (L08), Shoulder syndromes (L92), Wrist symptoms (L11), Hand/finger symptoms/complaints (L12), Carpal tunnel syndrome (CTS) (N93), Dupuytren’s contracture (L99.03), Trigger finger (L99.04), Rheumatoid arthritis and related conditions (L88), and Osteoarthritis (L91). However, there is no ICPC code for MSK. All patients with T2DM aged between 18 and 70 years old who were registered in January 2017 were included and matched with patients without any type of DM to compare the difference in prevalence. Matching was in a ratio of 1:1, based on age, gender, and general practice [19
]. Additionally, a maximum age of 70 was chosen because we believe that MSK disorders in patients older than 70 years are more likely based on age related degenerative processes.
2.2. Approach 2 (Questionnaire in General Practice)
This approach was carried out in the Meditta region, the Netherlands. Meditta is a company organized by GPs in the area of Sittard-Geleen, Roermond and Weert. Part of their activities includes providing T2DM care in the so-called diabetes care group. In this region, T2DM care is delivered by specially trained diabetes nurses employed by Meditta, who work in GP practices under supervision of the GP. T2DM care in the Netherlands is usually delivered by GPs and specially trained diabetes nurses. During a six month-period, T2DM patients between 30 and 70 years of age were approached upon visiting their diabetes nurse during their annual routine check-ups and were asked to fill in a questionnaire inquiring about MSK pain and medical history.
Additionally, to increase the recruitment, an announcement was placed on the website of the Dutch Diabetic Association (Diabetesvereniging Nederland, DVN) and in their periodic Diabetes Magazine.
All included patients withT2DM received the questionnaire during their check-up, and the diabetes nurse filled in a short case report form per patient, containing a last reading of HbA1C, the current body mass index (BMI) and year of T2DM diagnosis. Patients were asked to fill in the questionnaire and send it to our research center in the provided return envelope.
The outcome measures were categorized into a complaint and disease level. On a complaint level, the following subcategories were distinguished: point prevalence of painful body sites (shoulder, elbow, wrist, hand or fingers), lifetime prevalence of painful body sites, prevalence of the most painful body site, and on a diagnosis level: prevalence of specific MSK disorders (frozen shoulder, CTS, trigger finger and Dupuytren’s contracture). We defined point prevalence as the proportion of patients with T2DM with pain at time of filling in the questionnaire while lifetime prevalence was defined as the proportion of patients with T2DM who has suffered from pain any time in the past for at least 4 weeks.
The questionnaire was compiled on the basis of existing ones:
The Douleur Neuropathique 4 questionnaire (DN4) [20,21,22,23]
: A commonly used questionnaire for screening and diagnosing neuropathic pain in patients with neurological complaints, valid for the Dutch population and validated to be used in patients with DM. For our study purpose, we left out the physical examination questions, and only used the two interview questions, which composed of 7 items (yes/no answers). The cut-off point for neuropathic pain is considered to be 4 out of 7 points.
Pain questionnaire 
: A six-item questionnaire that classifies chronic MSK pain is adapted from the version used in epidemiological research by the Arthritis Research Campaign in the United Kingdom, translated in Dutch, and previously used in the Maastricht Study.
Epidemiology of diabetes intervention and complications association questionnaire for cheiroarthropathy 
: This questionnaire was used to assess the medical history of upper extremity MSK disorders (yes/no answers). We incorporated eight questions concerning history of symptoms and previous diagnosis while the examination part was excluded. These questions were translated from English to the Dutch language.
Self-reported comorbidity. A list of diseases derived from two sources was used; lists of the Study of Medical Information and Lifestyles in Eindhoven (SMILE), and the National Institute for Public Health and the Environment (RIVM) [26
2.3. Sample Size Calculation
In this calculation, we have assumed a prevalence of 12.8% for frozen shoulder [28
], and prevalence for specific MSK disorders starting from 5–15% for trigger finger [6
] up to 20% for frozen shoulder [29
]. Therefore, for approach 2, we used a conservative expected proportion of 0.12. Assuming a 10% response rate, we have handed over 1900 questionnaires to the diabetes nurses. This results in an expected number of 190 patients. By using a conservative expected proportion of 0.12, the width of the corresponding 95% CI is then equal to about 0.09, i.e., a 95% CI of 0.07 to 0.16. For approach 1, all patients with T2DM will be included, which is expected to exceed this number of 190 patients, implying an accurate estimate of the prevalence of MSK disorders.
2.4. Data Analysis
2.4.1. Approach 1: RNFM Database
Proportions were calculated for the prevalence of overall and specific upper extremity MSK disorders in DM patients and non-diabetes. Patients with T2DM were compared with non-diabetes using logistic regression correcting for the matching variables age, gender and general practice. Additionally, for patients with T2DM, logistic regression was used to assess which of the variables age, gender, duration of DM, rheumatoid arthritis and osteoarthritis were independently related to the outcomes of MSK in general and specific MSK disorders.
2.4.2. Approach 2: Questionnaire Study
For patients with T2DM in this approach, proportions were calculated for both overall and specific upper extremity point and lifetime prevalence of painful body sites on a complaint level, and for specific MSK disorders on diagnosis level. Logistic regression analyses were used to assess which variable (age, gender, duration of T2DM, body mass index (BMI), HbA1C, rheumatoid arthritis, osteoarthritis and other joint inflammation) was independently related to the outcomes of MSK in general and specific MSK disorders. Interaction between several variables, suggested in the literature or who have possible biological influences, were tested.
For both approaches, linearity assumption for numerical variables was assessed by testing whether a quadratic centered term improved the model fit significantly. In case the linearity assumption was violated, the analysis was repeated using a categorized variable (including dummy variables) instead of linear and quadratic terms, where these results were compared with those with linear and quadratic terms to see whether the same quadratic trend was represented by both analyses. The numerical variables were categorized using cut-off values based on the number of subjects per category and on sensible values, for example 50, 55, 60, and 65 years for age in approach 1 or 25, 30, and 40 for BMI [30
]. Multicollinearity was checked using variance inflaction factors (VIF), where VIF >10 indicate a collinearity problem, and influential outliers were defined as Cook’s distance > 1.
Odds ratios (OR) with corresponding 95% confidence intervals (CI) and two-sided p-values were reported, where p-values ≤ 0.05 were considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics for Windows (version 25.0, Armonk, NY, USA).
This is the first study evaluating the prevalence of upper extremity MSK disorders in patients with T2DM in general practice. This study was conducted with two different approaches and showed a prevalence for MSK disorders in patients with T2DM ranging from 16% based on by GP registered disorders and complaints (database study) to 67% based on self-reported diagnosis and pain (questionnaire study). This difference can be explained by the nature of the two approaches. The medical database study contains data of patients with T2DM who sought medical attention for their MSK disorders, otherwise these disorders would not have been registered by the GP, while in the questionnaire study, patients were attending their regular DM check-up and not primary seeking medical attention for MSK complaints. Therefore, it is plausible that the results of the database study might be an underestimation of the real prevalence and the questionnaire study might have caused an overestimation, as patients with T2DM suffering from pain might be more eager to participate. In this approach, half of the patients reported to have pain at time of filling in the questionnaire.
Additionally, we observed that the shoulder is the most affected body site in both studies, and that age, duration of T2DM, and gender show conflicting, statistically significant associations between the studies, except for the duration of T2DM and shoulder disorders/complaints, females, and CTS, which show a statistically significant positive association in both studies.
4.1. Comparison with the Literature
When comparing the observed prevalence with international studies, we noticed that there are no studies conducted in general practice. A cross-sectional study conducted in an outpatient diabetes centre in the USA reported a prevalence of shoulder pain with or without disability in 63% of the patients with DM. This result was obtained by using the Shoulder Pain and Disability Index (SPADI). This is a higher prevalence compared to the observed 39% in our questionnaire study. This difference may be explained by the fact that we only asked for shoulder pain and not for disability, and that our population consistent only of patients with T2DM [31
]. A tertiary hospital-based study conducted in Pakistan investigating MSK disorders of the upper limb extremity using a survey and physical examination, showed prevalences somewhere between our database and questionnaire studies [32
]. The differences can be explained by the difference in study settings and healthcare system. Another cross-sectional population-based study conducted in Norway, that also used a questionnaire approach, reported a prevalence of chronic MSK complaints of 58% in patients with T2DM, which is approximately 9% lower than observed in our study [12
]. They defined MSK complaints as pain and/or stiffness ≥3 months during the last year, where we used a duration of ≥4 weeks, which might explain the difference between the two studies. A hospital-based study conducted in Turkey where patients were physically examined showed a prevalence of frozen shoulder in patients with T2DM of 13% and 1.3% for CTS [28
], which is much lower than we observed. This large difference might be explained by the study design; we calculated lifetime prevalence in our questionnaire study, while in the hospital-based study, the point prevalence is estimated in patients having shoulder pain at the time of consultation. Therefore, we can conclude that study design, setting, healthcare system, and definition of MSK disorders might influence prevalence.
4.2. Strengths and Limitations
The RNFM database contains medical data representative for the Dutch population [18
], and made it possible to compare the prevalence of MSK disorders in patients with T2DM and patients without DM, which are major strengths of this work. GPs affiliated with the RNFM register the data of their patients, including ICPC codes, in a uniform manner and meet twice per year for training.
To overcome the problem of using only a single approach to determine the prevalence, we also performed a questionnaire study, which has two main advantages. First, unregistered MSK disorders and complaints that are missed in database studies can be included, because patients did not report them to the GP or because the GP judged them to be clinically not significant. Second, it enabled us to include BMI, HbA1C, rheumatoid arthritis, osteoarthritis, and other joint inflammation disorders in the analysis. However, a disadvantage is that recruiting patients for a questionnaire study might result in selection bias or reporter bias.
Despite the large sample size in the database study, the number of patients with T2DM diagnosed with rheumatoid arthritis and osteoarthritis was too small to correct for in the regression analysis, and additionally, we were unable to test the influence of BMI and HbA1C as this information was not registered in the database yet, which can be seen as limitations. Additionally, there might be other confounding factors associated with diabetes and MSK disorders, including medication, depression and health services use. Although statins may cause MSK pain in patients with T2DM, yet all patients with T2DM in The Netherlands actually are advised to always use statins, which is why we did not report about number of patients using this medication. Regarding co-morbidities, we have chosen to select only the most relevant ones. Unfortunately, we were unable to correct for the volume of health services used as this is not registered in the RNFM database.
Last, we have chosen a 1:1 matching between patients with T2DM and patients without DM, while other proportions were possible. However, we expected to have enough power of the study using 1:1 matching, which is proven by the statistical significance of the results found.
4.3. Clininal Implications and Future Research
In the context of clinical practice, our findings indicate that MSK disorders have a high prevalence in patients with T2DM and that screening for these disorders seems to be advisable. We define screening as a protocolled history taking and focused physical examination addressing MSK disorders during periodic regular DM check-up visits. Early detection and treatment of these disorders may play a role in preventing the development of chronic disorders, which might negatively influence DM treatment. However, to better inform guideline and policymakers, it would be useful to conduct a trial to investigate the effectiveness in two study groups, one with MSK screening and early management incorporated and a second group without.