The epidemiology, prognosis and management of carpal tunnel syndrome in primary care

Abstract

Carpal tunnel syndrome (CTS) is a compression neuropathy of the median nerve. Patients may experience discomfort in the hand and wrist, altered sensation and functional deficit. The management of mild to moderate CTS involves local corticosteroid injection (CSI) and night splinting (NS). The aims of the studies presented in this thesis were to describe the epidemiology of CTS and to develop prognostic models to predict future likely outcomes in patients presenting with CTS in a primary care setting. Evidence for predictors of treatment effect (CSI or NS) were also explored.
The estimated crude prevalence of patients presenting with CTS in UK primary care (using data from the Clinical Practice Research Datalink - CPRD) increased from 26.03 per 10,000 person years in 1993 (95% CI 25.10 to 27.00) to 36.08 per 10,000 person years (95% CI 35.45 to 36.72) in 2013. The proportion of patients having carpal tunnel release surgery (CTR) changed over the study period increasing from 19.35% in 1993 to a peak in 2009 at 29.32% and then decreasing to 27.3% in 2013.
A systematic review of 16 cohort studies reporting on the course and prognosis of CTS showed that prolonged symptom duration, a positive Phalen’s test, and thenar wasting were associated with poor outcome over a follow-up period of between six months and 12 years. However, not all associations were statistically significant and many studies were deemed to be at increased risk of bias (particularly relating to study attrition, confounding, and/or statistical analysis or reporting), hence the overall judgement regarding their predictive value remained inconclusive. A CPRD cohort of 91,412 patients with CTS was developed. 20.24% of the cohort had a recorded episode of CTR within 3 years of diagnosis. A prognostic model predicting time to surgical intervention was developed using Cox regression and included age, obesity, alcohol use, smoking and having multi-site pain, an inflammatory condition or a neck symptoms (adjusting for region and deprivation). The predictive capability of the model was limited, C statistic 0.59 (95% CI 0.58 to 0.59). Multivariable linear regression modelling using data from a randomised controlled trial (INSTINCTS) was used to develop a prognostic model predicting patient reported outcome at six months. The final model included the baseline Boston Carpal Tunnel Questionnaire score, the symptom severity score, and the absence of any other neck or upper limb symptom. The optimism-adjusted model calibrated poorly, overestimating the severity of outcome in patients with less severe observed CTS, and underestimating the severity of outcome in patients with more severe observed CTS.
A second systematic review including four RCTs summarised evidence for predictors of treatment effect. The results from one trial indicated that the effect of CSI was larger in patients with more severe nerve conduction and baseline symptom scores. Exploratory analysis of a small set of a priori defined candidate predictors of treatment effect using INSTINCTS data suggested that at six months, CSI was less effective than NS in patients with unilateral symptoms compared to those with bilateral CTS. However, results need to be carefully interpreted given the lack of significant interaction, the possibility of a unit of analysis error and a very small sample size.
In summary, patients with CTS presented in primary care with increasing frequency between 1993 and 2013. However, since around 2008, the proportion of patients receiving surgical treatment was observed to decrease, despite being considered clinically effective for most patients. Lower rates of surgery may be associated with changes in access to the procedure. This highlights the need for optimal management to be provided in primary care. Assuming the CPRD population to be representative, at least 20% of patients presenting with CTS did not respond well to their initial management in primary care. There is likely to be clinical benefit in identifying this group early in the course of their symptoms, and explore any differential treatment response, in order to better target treatment to the individual and identify those who are likely to require surgery. However, the prognostic models developed in this thesis performed poorly. It seems that the prediction of CTS is complex and potentially includes prognostic factors not measurable in CPRD or trial data. Similarly, no confirmatory evidence was found that could be used to match treatment options to individuals. Therefore, patients with CTS can be initially managed in primary care using current guidance and should be routinely followed up and referred for surgery if they fail to experience initial satisfactory improvement.