| Sign In to gain access to subscriptions and/or personal tools. |
The Surgical Treatment of Cubital Tunnel Syndrome: A Decision AnalysisFrom the Harvard Center for Risk Analysis, Harvard School of Public Health, Boston, MA, USA and the University of Toronto, University Health Network, Toronto, Ont., Canada Correspondence: Dr Brent Graham, Toronto Western Hospital, 399 Bathurst Street, Toronto, Ont., Canada M5T 2S8. Tel.: +1 416 603 5647; fax: +1 416 603 5716. E-mail:Brent.Graham{at}uhn.on.ca
The objective of our study was to use decision analysis to compare four common surgical treatments for cubital tunnel syndrome: simple decompression of the cubital tunnel, medial epicondylectomy, anterior subcutaneous transposition and anterior submuscular transposition. The variables used for this decision analysis model were based on data from the literature. Extensive sensitivity analyses were carried out to test the impact of the values given to these variables on the outcome of the model. The highest expected utility, 0.973, was associated with simple decompression. The expected utility was 0.969 for subcutaneous transposition and 0.965 for submuscular transposition. Medial epicondylectomy had the lowest expected utility at 0.961. Simple decompression remained the preferred strategy in extensive one-way sensitivity analyses.
Key Words: cubital tunnel syndrome • decision analysis • utility • nerve compression • ulnar nerve Compression of the ulnar nerve at the cubital tunnel is the most common cause of numbness on the ulnar side of the hand and, after carpal tunnel syndrome, the second most common compressive neuropathy affecting the upper extremity (McPherson and Meals, 1992; Rayan, 1992; Idler, 1996). Failure of non-operative treatment for this condition may be an indication for surgery. However, the best operative intervention remains controversial (Dellon, 1989; Mowlavi et al., 2000; Lowe et al., 2001). The most commonly performed operative procedures each have advantages and disadvantages (Table 1) and all have been shown to be associated with both satisfactory and poor outcomes (Dellon, 1989; Osterman and Davis, 1996; Posner, 2000; Lowe et al., 2001). There have been few well-executed randomised trials comparing these procedures reported in the literature and none that make a direct comparison of more than two of the possible surgical interventions. Issues of varying case definitions for cubital tunnel syndrome, inconsistencies in identifying the stage of the condition and the absence of an established consensus on measuring the outcome of treatment limit the feasibility of a convincing randomised trial (Graham, 2005) comparing treatment with the four most common surgical treatments used in this condition, viz. simple decompression of the cubital tunnel, medial epicondylectomy, anterior subcutaneous ulnar nerve transposition and anterior submuscular ulnar nerve transposition.
The objective of our study was to use decision analysis to compare these four interventions.
Decision analysis models Implicit in the process of clinical decision-making is the evaluation of risk and benefit associated with the various choices available for management. When actual evidence to guide clinicians in their decisions is either conflicting or completely absent, the element of uncertainty is more than usual. Decision analysis allows a quantitative comparison of the various options for addressing a clinical problem (Kassirer, 1976; Detsky et al., 1997a, b). A decision analysis attempts to consider all possible outcomes of a given strategy. The probability that any particular outcome will occur is estimated from the literature. The value of a given outcome is expressed in terms of utility, which is a measure of the desirability of the health state encompassed by that outcome (Torrance, 1987). Utility for a health state is expressed on a scale of 0 to 1.0, where 0 represents death and 1 represents perfect health. Techniques such as the standard gamble and the time tradeoff can be used to establish the utility of different health states (Naglie et al., 1997; Torrance, 1987; Sox et al., 1988). Some examples of the utility associated with various health states that have been reported previously in the literature are listed in Table 2. The decision analysis model establishes which strategy is associated with the highest expected utility and, thus, helps guide clinicians towards the best clinical policy for a particular clinical problem.
The concept of disutility is the converse of utility and represents a transient health state that temporarily downgrades the quality of life. For the example of cubital tunnel syndrome, disutilities would include perioperative discomforts and inconveniences, such as hospitalisation and immobilisation. A disutility can also be established for the complications specific to each treatment. For example, a haematoma, or having to drain a wound, would be a temporary state associated with quantifiable disutility. The disutilities associated with any of the short-term states during treatment are subtracted from the utility associated with each treatment. For example, the overall expected value of a procedure such as medial epicondylectomy would be: the utility of the procedure (the desirability of the postoperative state of a surgical scar and bone removal combined with a complete relief of symptoms), minus the disutility of the procedure (surgical wounds, perioperative pain, hospitalisation, immobilisation, etc.) and minus the disutility of any complications of the procedure (the negative impact of complications such as medial instability or ectopic bone formation occur) if they should occur. Important undesirable consequences of treatment that are permanent are not accounted for as disutilities, but rather as a decreased utility associated with a health state. To take the example of medial epicondylectomy again, the outcome from the standpoint of the relief of sensory symptoms might be considered good, but the utility of that health state might be decreased by the co-occurrence of permanent elbow stiffness, in comparison with a health state characterised by a full relief of sensory symptoms and normal elbow motion. Decision analysis models have two additional features that make them particularly useful where actual evidence to support one clinical strategy over another is not available. First, the explicit nature of the model makes it clear which assumptions were made, so that the reader can determine whether these seem valid. Second, the model allows a sensitivity analysis to establish the stability of the conclusions reached by the analysis. In a sensitivity analysis, each variable in the model is varied throughout the entire range of its possible values to determine whether the conclusion suggested by the model is sensitive to the value of one or more important variables. If changing the value of a certain variable changes the outcome of the model, e.g., a different treatment strategy becomes preferred, then the model is sensitive to that variable. For example, the model may indicate that surgical treatment A is recommended over surgical treatment B when the probability of a postoperative infection is 1%, but treatment B becomes the recommended strategy when the probability of infection is 5%. Users of this information may determine the best strategy for their patient population, based on how their probability of infection compares with these values.
Surgical treatment of cubital tunnel syndrome modelled using decision analysis Similarly, the choice of submuscular transposition as the salvage procedure may not correspond to the customary practice of some, perhaps even a majority, of surgeons and, as a result, the model may not be helpful to these individuals. However, all decision models require some assumptions of this nature and the designation of a different intervention as the most appropriate salvage procedure may also be open to the same criticism. One of the benefits of modelling the issue using this approach is the transparency of the process. Consumers of the information can easily determine how much, or little, the model fits in with their personal practices and beliefs. The possible treatment strategies can be established only at the starting point for the model. In actual practice, a patient who has a failure of one surgical procedure might elect to live with the symptoms of cubital tunnel syndrome rather than undergo any further treatment. This option cannot be evaluated within the model and is one of the shortcomings of this type of analysis. We assumed that submuscular transposition of the ulnar nerve was the most definitive management for this condition and would be the surgical treatment of choice if simpler procedures failed (Dellon, 1989; Holmberg, 1991; Rogers et al., 1991; Osterman and Davis, 1996; Siegel, 1996). A failure of submuscular transposition itself, either as a primary treatment or as a salvage procedure, was considered an endpoint of the model. It is possible that the outcome of a submuscular transposition may be worse when it is done as a salvage procedure, after either a medial epicondylectomy or subcutaneous transposition, than when it is the primary procedure. However, for the purposes of the model, it was assumed that the outcome of submuscular transposition was not affected by previous surgery. In decision analysis, outcomes are characterised as health states. The operative definition of these health states depends on the issue under examination. For the model of surgical treatment of cubital tunnel syndrome, a good outcome was defined a complete relief of all sensory symptoms for at least 2 years. All other results were classified as bad outcomes. It could be argued that a full relief of sensory symptoms with no recovery of lost motor function would not be considered a good outcome. However, the focus on sensory symptoms resulted from the observation that most, if not all, patients with ulnar nerve compression are affected by numbness and only a subset of these have motor loss. In addition, most reports in the literature have emphasised sensory recovery as an important outcome (Artico et al., 2000; Bartels et al., 2005; Biggs and Curtis, 2006; Black et al., 2000; Clark, 1979; Gervasio et al., 2005; Goldberg et al., 1989; Greenwald et al., 1999; Nathan et al., 1995; Osterman and Davis, 1996). Once again, this definition of a good outcome is an assumption that must be understood in interpreting the results of the decision analysis.
Basic tree structure
All the possible outcomes associated with each treatment were assigned a utility that was estimated from the literature.
Probabilities, utilities and disutilities The rates of complications for each of the procedures obtained from the literature and used in the base case analysis of the model are listed in Table 3. The probability of complications for each treatment was estimated by considering the risk of individual complications and summing them to establish an overall rate. Using these figures, the probability of complications associated with simple decompression was estimated to be 0.02, for anterior subcutaneous transposition 0.07, for anterior submuscular transposition 0.09, and for medial epicondylectomy 0.27.
A similar process for estimating the probabilities of unsatisfactory outcomes was used. The success of any of these procedures was defined as a complete relief of all sensory symptoms at follow-up, which ranged in the literature from months to years. A range for the probability of a bad outcome after each procedure was taken from the literature and, then, a value within the range was chosen to perform the base case analysis. We considered only the data for those patients treated for moderate to severe compression, based on the staging system established by Dellon (1989). For example, the literature reports a range of 3% to 50% failure to relieve symptoms after anterior submuscular transposition in patients with moderate to severe compression. For anterior subcutaneous transposition, the range is 10% to 73%. We set the probability of failure at 0.10 for anterior submuscular transposition and at 0.38 for anterior subcutaneous transposition after looking at the distribution of the results. The literature appears to indicate that the rate of success, defined as a full resolution of sensory symptoms, of simple decompression and medial epicondylectomy, is less than that of either subcutaneous or submuscular transposition. Based on this literature, the probability of an unsatisfactory outcome was estimated to be 0.60 for simple decompression and 0.45 for medial epicondylectomy. The complications for simple decompression, subcutaneous transposition and submuscular transposition were considered to be transient and to have little effect on the probability of failure of the treatment. In contrast, medial epicondylectomy can be associated with complications of medial instability or ectopic bone formation, both of which may decrease the probability of a good outcome. Therefore, a slightly higher probability of failure was assigned to medial epicondylectomy with a complication. The estimation of utility of each of the health states was based, in part, on published guidelines for determining impairment and permanent disability (AMA, 2001) and, partly, on assessments of utility in individual patients with cubital tunnel syndrome as well as other conditions affecting the hand (Graham and Detsky, 2001). Chronic sensory and motor disturbances, while potentially disabling for many vocational and recreational activities, are not life-threatening, do not affect ambulation and still allow independent completion of all activities of daily living for most patients. The utility of this health state was estimated to be 0.95. The utility of a small scar without pain, such as that associated with a simple decompression, was estimated to be relatively high at 0.99. The utility of a good outcome with a larger scar, such as those needed for medial epicondylectomy or anterior transposition, to either a subcutaneous or a submuscular location, was set at 0.98. Since the majority of complications associated with any of the four procedures are temporary, a conservative assumption was made that the utility of the outcome associated with the procedure was the same, whether or not a complication occurred. The effect of most complications was measured as a disutility. Estimates of disutilities of the procedures were based on the need for hospitalisation, discomfort associated with the intervention, the length of postoperative immobilisation and the potential need for rehabilitation after surgery. The magnitude of disutility values was based on similar analyses in the literature (Detsky et al., 1997a, b), including a study that estimated disutility associated with various forms of wrist reconstruction (Graham and Detsky, 2001). We assigned a disutility of 0.01 to medial epicondylectomy, since it required the removal of bone. There is moderate pain and immobility of the limb, associated with postoperative swelling, which may persist for a period of 3 to 6 weeks, and there is a short period of immobilisation of the limb, varying from a few days to a several weeks. The anterior submuscular transposition may also be associated with moderate postoperative pain and swelling and requires a longer period of immobilisation, usually 2 to 3 weeks in the literature. These combined factors were assigned the same disutility of 0.01. Anterior subcutaneous transposition also requires immobilisation, but it does not require muscle dissection. A slightly smaller disutility of 0.008 was assigned to this procedure. Simple decompression, which does not require extensive dissection, an excision of bone or prolonged immobilisation, was given the smallest disutility, 0.005. The values given to all the variables in the model appear in Table 3.
The base case analysis indicated that the preferred surgical treatment of moderate-to-severe cubital tunnel syndrome was simple decompression (Table 4). The expected utility for this procedure was 0.973. The next best strategy was anterior subcutaneous transposition, which had an expected utility of 0.969. Anterior submuscular transposition had a slightly smaller expected utility of 0.965. Medial epicondylectomy was the least supported surgical strategy, with an expected utility of 0.961.
Extensive sensitivity analyses were carried out to compare the two strategies with the highest expected value, viz. simple decompression and subcutaneous transposition (Table 5). This analysis showed that the result recommending simple decompression as the primary treatment for cubital tunnel syndrome was robust, indicating that this conclusion was, in general, not sensitive to the values of other variables in the model. The only variables to which this result was sensitive were those related to a bad outcome from either simple decompression or submuscular transposition and the disutilities of the simple decompression, subcutaneous transposition and submuscular transposition.
The preferred strategy shifted from simple decompression to subcutaneous transposition if the probability of a bad outcome from simple decompression increases to 0.82 from the base case value of 0.60. Similarly, subcutaneous decompression is preferred to simple decompression if the probability of a bad outcome after submuscular transposition is increased to 0.82 from its assumed value of 0.10. In neither case does it seem likely that the threshold value is the true value for the variable. In other words, it is unlikely that the probability of a bad outcome after either simple decompression or sub muscular transposition is actually as high as 0.82. As a result, the conclusion favouring simple decompression is probably accurate. Similarly, subcutaneous transposition is also the preferred strategy under three conditions related to the disutility of the other procedures. Subcutaneous transposition is preferred if the disutility of simple decompression increases to 0.0158. Subcutaneous transposition is also preferred if its disutility decreases to 0.001. This value is much lower than the disutility assigned to simple decompression. Finally, subcutaneous transposition would be preferred if the disutility of submuscular decompression increases to 0.0296. None of the other variables in the model affected the result.
Our decision analysis model indicates that the best surgical strategy for the treatment of moderate to severe ulnar nerve compression at the cubital tunnel is simple decompression, with submuscular transposition as a salvage procedure if sensory symptoms are not addressed. Decision analysis has not been used frequently in addressing controversial areas in clinical medicine (Birkmeyer and Birkmeyer, 1996), although there have been some reports that use this approach more recently (Brothers et al., 2004; Graham and Detsky, 2001; Kwok et al., 2001; Leblanc and Worsley, 1995; Sarasin et al., 1995; Richard et al., 2002). Where evidence from well-designed randomised trials is unavailable, decision analysis offers an opportunity for providing insight into a clinical problem (Graham, 2005). As with the result of a randomised trial, a decision analysis is best used to establish a clinical policy that has a basis in evidence. It is understood that individual cases may not fit the assumptions of the decision analysis model and may require treatment that does not necessarily correspond to the strategy suggested by these models. In this study, the decision analysis favours simple decompression as the first surgical treatment in the cases of moderate or severe cubital tunnel syndrome, over the other strategies of medial epicondylectomy, subcutaneous transposition and submuscular transposition, where submuscular transposition is also used as the salvage procedure for failures of the other three interventions. In the base case analysis, this was the result, even where the probability of failure of simple decompression was 0.60. In other words, it is preferable to carry out a simple decompression in all cases and to require a second procedure to perform a submuscular transposition in as many as 60% of cases, rather than to use any of the other more extensive treatments as a first intervention. In fact, the sensitivity analysis indicates that the failure rate for simple decompression would have to be 0.82 before subcutaneous decompression would be the preferred strategy. Although this may appear to be counter-intuitive, the reason for this result relates to the disutility of the other procedures and their potential complications, in comparison with simple decompression. Simply stated, simple decompression is such a small intervention that a substantial rate of failure is acceptable if it means that a more invasive, complicated procedure is avoided in a substantial number of individuals. Obviously, this does not take into account issues like the effect on individuals of the requirement for two operative procedures, with the associated absences from work and other activities. In fact, the probability of failure with simple decompression is probably much lower than 0.82. Recent prospective trials comparing simple decompression with anterior transposition have reported good or excellent results with this procedure in 60–80% of patients (Bartels et al., 2005; Biggs and Curtis, 2006; Gervasio et al., 2005). The conclusion that simple decompression is the best initial strategy for the surgical treatment of cubital tunnel syndrome has been indirectly supported by three recent randomised trials (Bartels et al., 2005; Biggs and Curtis, 2006; Gervasio et al., 2005). Two of these studies compared simple decompression with submuscular transposition (Biggs and Curtis, 2006; Gervasio et al., 2005) and the third compared simple decompression with anterior subcutaneous transposition (Bartels et al., 2005). No statistically significant differences were found with respect to the neurological outcome, although both anterior transposition procedures were found to be associated with a higher rate of complications than simple decompression. The sensitivity analysis provides several other interesting insights into this problem. The preferred strategy also shifts to subcutaneous transposition from simple decompression if the probability of a bad outcome from a submuscular transposition increases to 0.82. The reason for this is that the probability of needing submuscular transposition as a second procedure is much higher if the first treatment is simple decompression (base case probability of failure = 0.60) than if it is subcutaneous transposition (base case probability of failure = 0.38). It is unlikely that the risk of a bad outcome from a submuscular decompression would be as high as 0.82. The base case analysis set the value for this variable at a much lower level of 0.10. The sensitivity analysis also indicated that the result recommending simple decompression as the treatment of choice was sensitive to the value for disutility assigned to any of the procedures, other than medial epicondylectomy. In the case of simple decompression, the disutility of this procedure, 0.005 in the base case analysis, would have to be more than three times higher, 0.0158. This would place it at a higher level than either submuscular transposition or medial epicondylectomy, which is unlikely to be true. Similarly, the disutility of subcutaneous transposition would have to drop from 0.008 to 0.001 to displace simple decompression as the preferred strategy. It seems unlikely that subcutaneous transposition would be associated with less disutility than simple decompression. The only other variable that shifted the preferred strategy from simple decompression to subcutaneous transposition was the disutility associated with submuscular transposition. In the base case analysis the value of this variable was 0.01. The threshold in the sensitivity analysis was 0.0296. This, again, relates to the greater probability of requiring a submuscular transposition as a salvage procedure if simple decompression is the first surgical management than if subcutaneous transposition is the initial surgical procedure. It is possible that the disutility of submuscular transposition could be this high. This is the same level of disutility associated with scaphoid excision and midcarpal arthrodesis in a previous study (Graham and Detsky, 2001). Although both procedures are extensive, it is unlikely that a submuscular transposition would be associated with as much disutility as this type of wrist reconstruction. None of the variables associated with medial epicondylectomy, which scored the lowest expected utility among the four procedures, had an impact on the preferred strategy indicated by the model. In other words, changes to the values for disutility and for the probability of complications with this procedure did not change the outcome of the model. More complicated two, or three, way sensitivity analyses may have shown an effect of varying some of these variables, but the overall influence of this procedure on the analysis appears to be minimal. The values for expected utility of the four procedures were relatively close and covered a small range from 0.973 to 0.961. The differences between these values cannot be analysed using statistical methods associated with hypothesis testing concepts, such as calculating a p value. The meaning of differences in expected utility of competing strategies depends on how those utilities are interpreted. For example, if they are used to calculate quality adjusted life years (QALY), the magnitude of the difference would depend on the age of the patient at the time of intervention. In general, these results should be taken qualitatively to indicate that, among these procedures, under the stated assumptions of the model, the preferred strategy is simple decompression. In addition to imposing some assumptions on the model that may not be accepted in all circumstances, an additional drawback to our analysis is the assumption that all the bad outcomes occur simultaneously. The horizon for defining outcome in this study was two years. Of course, a more realistic approach would be to create a model in which bad outcomes accumulate over time. This would require advanced modelling procedures such as Markov processes (Sox et al., 1988). In general, the qualitative results of simple models, like the one used in this study, do not change with more complicated Markov models (Naimark et al., 1997). This is especially true if the conclusion of the model is robust in extensive sensitivity analyses, as was the case in this instance.
Articles used in determining variables used in the decision analysis model:
Manuscript received November 18, 2005. Accepted for publication July 3, 2007.
Amako M, Nemoto K, Kawaguchi M, Kato N, Arino H, Fujikawa K (2000). Comparison between partial and minimal medial epicondylectomy combined with decompression for the treatment of cubital tunnel syndrome. Journal of Hand Surgery, 25A: 1043–1050.[Medline] [Order article via Infotrieve]American Medical Association (AMA). Guides to the evaluation of permanent impairment, Chicago, American Medical Association, 2001.Arle JE, Zager EL (2000). Surgical treatment of common entrapment neuropathies in the upper limbs. Muscle Nerve, 23: 1160–1174.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]Artico M, Pastore FS, Nucci F, Guiffre R (2000). 290 surgical procedures for ulnar nerve entrapment at the elbow: physiopathology, clinical experience and results. Acta Neurochirurgia, 142: 303–308.[CrossRef]Bartels RH, Verhagen WI, van der Wilt GJ, Meulstee J, van Rossum LG, Grotenhuis JA (2005). Prospective randomized controlled study comparing simple decompression versus anterior subcutaneous transposition for idiopathic neuropathy of the ulnar nerve at the elbow: Part 1. Neurosurgery, 565: 22–30.Biggs M, Curtis J (2006). Randomized, prospective study comparing ulnar neurolysis in situ with submuscular transposition. Neurosurgery, 58: 296–304.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]Birkmeyer JD, Birkmeyer NO (1996). Decision analysis in surgery. Surgery, 120: 7–15.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]Black BT, Barron OA, Townsend PF, Glickel SZ, Eaton RG (2000). Stabilized subcutaneous ulnar nerve transposition with immediate range of motion. Long-term follow-up. Journal of Bone and Joint Surgery, 82A: 1544–1551.Brothers TE, Cox MH, Robison JG, Elliott BM, Nietert P (2004). Prospective decision analysis modeling indicates that clinical decisions in vascular surgery often fail to maximize patient expected utility. Journal of Surgical Research, 120: 278–287.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]Caputo AE, Watson HK (2000). Subcutaneous anterior transposition of the ulnar nerve for failed decompression of cubital tunnel syndrome. Journal of Hand Surgery, 25A: 544–551.[CrossRef][Medline] [Order article via Infotrieve]Clark CB (1979). Cubital tunnel syndrome. Journal of the American Medical Association, 241: 801–802.
Journal of Hand Surgery (European Volume), Vol. 32, No. 6,
654-662 (2007)
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
Top
Abstract
MATERIALS AND METHODS