**4. Discussion**

Sixteen studies described the outcome of different implants and postoperative therapy regimes for MCP arthroplasty. Two of these studies compared the efficacy of one regimen over another, one of these being prospective. Sambandam, Gul and Priyanka (2007)(25) state that 'most studies undermined the importance of this aspect (post-operative treatment) of the procedure' with regards to first carpometacarpal joint arthroplasty, but their claim could be expanded to arthroplasty of other joints of the hand. Post-operative protocols for splinting, activity and exercises are not always well-described, so although there were numerous studies about MCP arthroplasty, they are not included in this review.

Hand therapy for other conditions such as flexor tendon repair also offers multiple postoperative regimens. For example, healing tendons of the hand usually receive motion, but it may be passive, active, or a combination of all of these. The rationale for the various exercise regimens is based on biological healing of the tendon and the strength of the surgical repair, thus its ability to withstand stress without rupturing or gap formation (26,27). These patients usually have normal anatomy preoperatively, leaving few patient variables. Postoperative therapy regimens for MCP arthroplasty are also based on principles of healing and scar formation, but are not prescribed according to the patient's preoperative hand impairment, the type of implant used, or soft tissue balancing procedures performed. For example, patients having undergone extensor tendon rebalancing and recentralization may benefit from avoidance of passive flexion or avoidance of the extremes of flexion, much like a postoperative extensor tendon repair regime. The literature suggests that postoperative therapy for MCP arthroplasty has not been prescribed in this manner; rather, standard protocols have been designed and applied to consecutive patients.

To compare the efficacy of a new protocol, many patients would be required for allocation to various postoperative therapy groups. Their outcomes would have to be analyzed according to what protocol they received with the implant, surgery, and preoperative status as variables. The first difficulty in forming control or comparison groups lies in the infrequency of this procedure; for example Ring et al.(8) took three years to include 25 hands in their study.

The most common source of bias in the studies was selection bias, which occurs when patients are chosen for treatment or control groups as a result of characteristics that are expected to affect their outcome. Randomization is designed to control the confounding effects of differences between subjects at baseline, and the randomized trial is recommended as the best method of determining treatment efficacy. Here lies the second difficulty in forming control or comparison groups. Patients undergo MCP arthroplasty at all stages of their disease, evidenced by the wide range of motion deficits between the studies of Burr and Pratt, in which the case study patient had nearly normal preoperative MCP motion, and Burr et al., in which some patients had only 25° of MCP flexion, Measures of pain also varied widely in the latter study, ranging from "zero" to "eight out of ten." These baseline measurements demonstrate the difficulty in obtaining a homogeneous, comparable group of patients with rheumatoid arthritis.

The other three sources of bias described by the Cochrane Collaboration(7) were present in the reviewed studies. Performance bias occurs when patients receive a variation in duration,

Sixteen studies described the outcome of different implants and postoperative therapy regimes for MCP arthroplasty. Two of these studies compared the efficacy of one regimen over another, one of these being prospective. Sambandam, Gul and Priyanka (2007)(25) state that 'most studies undermined the importance of this aspect (post-operative treatment) of the procedure' with regards to first carpometacarpal joint arthroplasty, but their claim could be expanded to arthroplasty of other joints of the hand. Post-operative protocols for splinting, activity and exercises are not always well-described, so although there were

Hand therapy for other conditions such as flexor tendon repair also offers multiple postoperative regimens. For example, healing tendons of the hand usually receive motion, but it may be passive, active, or a combination of all of these. The rationale for the various exercise regimens is based on biological healing of the tendon and the strength of the surgical repair, thus its ability to withstand stress without rupturing or gap formation (26,27). These patients usually have normal anatomy preoperatively, leaving few patient variables. Postoperative therapy regimens for MCP arthroplasty are also based on principles of healing and scar formation, but are not prescribed according to the patient's preoperative hand impairment, the type of implant used, or soft tissue balancing procedures performed. For example, patients having undergone extensor tendon rebalancing and recentralization may benefit from avoidance of passive flexion or avoidance of the extremes of flexion, much like a postoperative extensor tendon repair regime. The literature suggests that postoperative therapy for MCP arthroplasty has not been prescribed in this manner; rather, standard protocols have been designed and

To compare the efficacy of a new protocol, many patients would be required for allocation to various postoperative therapy groups. Their outcomes would have to be analyzed according to what protocol they received with the implant, surgery, and preoperative status as variables. The first difficulty in forming control or comparison groups lies in the infrequency of this procedure; for example Ring et al.(8) took three years to include 25

The most common source of bias in the studies was selection bias, which occurs when patients are chosen for treatment or control groups as a result of characteristics that are expected to affect their outcome. Randomization is designed to control the confounding effects of differences between subjects at baseline, and the randomized trial is recommended as the best method of determining treatment efficacy. Here lies the second difficulty in forming control or comparison groups. Patients undergo MCP arthroplasty at all stages of their disease, evidenced by the wide range of motion deficits between the studies of Burr and Pratt, in which the case study patient had nearly normal preoperative MCP motion, and Burr et al., in which some patients had only 25° of MCP flexion, Measures of pain also varied widely in the latter study, ranging from "zero" to "eight out of ten." These baseline measurements demonstrate the difficulty in obtaining a homogeneous, comparable group of

The other three sources of bias described by the Cochrane Collaboration(7) were present in the reviewed studies. Performance bias occurs when patients receive a variation in duration,

numerous studies about MCP arthroplasty, they are not included in this review.

**4. Discussion** 

applied to consecutive patients.

patients with rheumatoid arthritis.

hands in their study.

quality, or quantity of the treatment being studied, which was suspected in the continuous passive motion (CPM) study by Ring et al. Ring et al. describe the application of CPM in detail, except passive forces are described as "low" and treatment quantity is described as "as tolerated." As a result, the reader remains unsure of what amount of passive force is ineffective, as well as what quantity of treatment per day is ineffective.

Detection bias is determined if the timing of assessment, the outcome assessment used, or knowledge of the assessor of the patient's previous state could miss any relevant aspect of the outcome. This may have occurred in the study by Groth et al.,(10) in which some preoperative data were unavailable and patients were assessed at different postoperative time frames. Detection and comparison of outcomes between studies are only possible when the same outcome measures are used in a standardized manner. The researchers in this review all measured range of motion, but at different time frames (Table 1). Those who measured pain, cosmesis, and function applied different assessments at different time frames. The challenge of outcome measurement in rheumatology has led to the formation of focus groups such as OMERACT (Outcome Measures in Rheumatoid Arthritis Clinical Trials), who have made recommendations for outcome measures to be used in drug trials. OMERACT recommendations are not fully relevant to hand therapy research; however, the process of forming a focus group, and the development of assessment guidelines that allow comparison between homogeneous patients, is possible (28).

Attrition bias is determined if the loss of patients in the study is significant or varies between the treatment and control groups. This is common in long-term studies involving patients with rheumatoid arthritis, and was experienced by Groth et al.,(10) who were unable to obtain long-term follow-up of the patient group who received their extension protocol. Long-term follow-up is an issue with rheumatoid populations. These patients undergo numerous surgical and drug interventions, while their disease progresses and fluctuates, making the long-term effects of the MCP surgery and therapy difficult to define. Once more, large numbers of patients in each treatment group would be required to decrease the effects of attrition bias and to dilute the effects of subsequent interventions.

The difficulties of past studies guide the planning of future studies. Although the issues of low patient numbers, variable preoperative status, additional surgical and drug interventions, and chronic disease cannot be altered, study designs can. Large randomized trials may not be possible; however, samples of patients, paired according to preoperative status, may be allocated to different treatment protocols. Standardized measurement of pain, cosmesis, impairment, disability, and impact on the patient, made at similar postoperative time frames, would further assist in determining treatment efficacy.
