Complications in ankle arthroscopy

Arthroscopy of the ankle joint has become an important therapeutic tool for the management of post-traumatic and chronic ankle problems. Although the anatomy of the ankle is rather complex, ankle arthroscopy is generally perceived as ‘no problem surgery’. When compared to arthrotomy, ankle arthroscopy is related to decreased morbidity and rapid rehabilitation. For treatment of posterior ankle problems, a 2-portal hindfoot approach was developed and published [40]. Several publications have shown the safety of the portals [38, 40].

Between July 1987 and 2006, 1305 consecutive ankle arthroscopies were performed at the University Hospital of the University of Amsterdam. The procedures were performed by 33 different orthopaedic surgeons, at a single surgery centre. Minimal follow-up was 24 months. Charts of all patients, including surgery reports, were reviewed to identify and verify all pre- or post-operative complications. In addition, it was used to identify the most recent follow-up. Patient characteristics (side, gender, age at time of surgery, whether it concerned a day care procedure, surgeon, indication, treatment and duration of surgery) were retrieved. Patients with a pre- or post-operative complication were identified. A complication was defined as ‘every event that arises as an additional problem during or following the procedure and is secondary to it’. Our hospital has an active complication surveillance system. All complications are systematically scored in a database. Apart from this active prospective scoring system, we reviewed all patient charts and looked for any uneventful post-operative course. All patients with a reported complication were asked to visit our outpatient department for a clinical follow-up examination.

At follow-up, the patients’ history was taken and a physical examination of the ankle was performed. The patient was asked for pain, limitations in activities, support requirement, limitation in walking distance and difficulties of walking on uneven surfaces. During physical examination, gait abnormality, sagittal range of motion, alignment of the foot and hindfoot and ankle-hindfoot stability were assessed. This information was used to determine the AOFAS Ankle/Hindfoot score [24].

Patients were asked how the complication had developed over time. In case of a neurological complication, usually consisting of hyposensitivity of a cutaneous part of the foot, a pin prick test was used to determine the area of altered sensation. The area was marked. For documentation, feet were photographed and nerve damage was assessed.

Median follow-up was 7.5 years (range 2–20). Altogether, a total of 1,305 ankle arthroscopies in 1176 patients were performed. Of the 122 patients that were operated more than once, 52 were operated bilateral. Seven patients were operated three times, in three cases bilateral. Three patients were lost to follow up, as their medical charts were lost. We did, however, have their surgical files as well as their complication documentation. They did not have a registered complication. Sixty-three per cent was male and 37 female. The mean age at surgery was 33 years (SD 11.0). In 54 %, it involved the right and in 46 % the left ankle.

Nine hundred and five anterior ankle arthroscopies (69.3 %), 311 posterior (23.8 %) and 17 combined anterior and posterior arthroscopies (1.3 %) were performed. Twenty-one patients had a combined right and left ankle arthroscopy (in one session). Median duration of surgery was 37 min (range 4–195).

Forty-six complications in 43 patients were identified (Table 2). The overall complication rate was 3.5 %. The complication rate for hindfoot endoscopy alone was 2.3 %. Pulmonary embolism in a 32-year-old woman 1 week post-operatively was the most serious complication. The patient had undergone arthroscopy to determine the articular damage after sustaining an acute ruptured anterior talofibular ligament (ATFL). Because of (open) suturing of the ATFL, she received a cast for 6 weeks. She did not receive anticoagulation therapy while she was wearing the cast. The most common complication that occurred was neurological injury: 25 complications. These can be subdivided into the nerves that were injured (Table 2). Six of 7 patients with superficial infection were treated with 5 days of prophylactic antibiotics. None of them had a positive culture. In all patients, symptoms resolved within a week. Two patients with a deep infection had re-surgery and received intravenous antibiotics. One of them was a patient with sinus tract formation that did not receive prophylactic antibiotics and developed a deep infection 12 days after surgery.

Thirty-eight patients with a complication were seen at our hospital for re-evaluation. Five patients were lost to follow up. Among these were the patient with vascular damage and the one where instrument breakage had occurred. Both patients had a recorded uneventful recovery. Two patients with sinus tract formation and one patient with hypoesthesia in the area of the superficial peroneal nerve could not be traced. Chart review showed that the first two patients had no complaints at their last follow-up.

The 38 patients seen at follow-up accounted for 41 complications. Of the 41 complications, 22 healed without complaints, 17 had persistent complaints, and these were all of neurologic origin. Two could not be assessed because of situations not related to the arthroscopy. These situations included degenerative spine disease with loss of sensibility in both feet in one patient and the development of hereditary motor and sensory neuropathy in another patient.

Seventeen of 25 patients with a neurological complication reported persisting symptoms at follow-up (68 %). This consisted of hypoesthesia of a cutaneous part of the foot. None of these patients had functional limitations. Five patients were symptom free. One patient was lost to follow up, and two could not be assessed because of the reasons previously mentioned.

The most important finding of the present study was an overall complication rate of 3.5 % for ankle arthroscopy related to the dorsiflexion method. All patients in whom a complication occurred were satisfied with the outcome of the arthroscopic procedure at long-term follow-up. None of the initial complications led to a functional limitation. Our complication rate of 3.5 % compares favourably to the average of 10.3 % found in literature. The difference is most likely even bigger as we scored our complications prospectively, instead of retrospectively like other studies [1‐3, 11, 13, 14, 17‐19, 27, 34, 37, 45]. Retrospective studies tend to score lower complication rates when compared to prospective documentation.

Neurological complications (1.9 %) were most common and accounted for half of the total amount of complications in our series. In a recent study, in which continuous distraction was used, a neurological complication rate of 5.4 % was found [45]. Other authors that use continuous distraction report a similar high percentage of neurological complications. The most serious complication in our series was a pulmonary embolism in a 32-year-old woman. Apart from the ankle arthroscopy, she underwent open surgery during the same anaesthesia, requiring 6 weeks of cast immobilization. The combination of cast immobilization, her use of oral contraception and smoking (a package a day) may have made her more prone to the development of embolism. She did not receive prophylactic anticoagulants.

In order to find an explanation for the difference in complication rate in our series (3.5 %) and those reported in other series (average 10.3 %), we analysed to following aspects: definition of complications, type of complication registration, type of surgical procedures, surgical experience or technique of the procedure. Concerning definition of complications, we used the following definition: ‘every event that arises as an additional problem during or following the procedure and is secondary to it’. When we compare the type of our complications to those reported in literature, there is no difference [2, 14, 18, 26, 37]. We feel that an underestimation of complications did not occur since we used a prospective national registration system. Apart from that, we checked all patient charts. Concerning the type of operative procedures, we have listed our indications. They do not differ from other authors [2, 11, 14, 17, 18, 26, 27, 37]. Concerning surgical experience, in our series 33 surgeons were involved. Other studies often include arthroscopic procedures of only one experienced orthopaedic surgeon. Our low complication rate in spite of the involvement of a large number of low volume surgeons is therefore remarkable and in favour of the procedure.

We conclude that the dorsiflexion technique that we use to perform the arthroscopic procedure is the main reason for the low percentage of complications when compared to other series using a continuous distraction system. Some details and technical conditions of our approach are important to mention. Like others we identify the superficial peroneal nerve by plantarflexing and inverting the foot and marking it on the skin [13, 34]. It was shown by De Leeuw et al. [9] that the superficial peroneal nerve moves in a lateral direction in relation to the skin when the ankle is moved back from this inverted position to the neutral position. This movement is mean 3.6 mm to the lateral side. It is therefore important to stay on the medial side of the marking. We enter the joint by blunt dissection with the joint in dorsiflexion. In this dorsiflexed position, the nerves and vessels are relaxed and can thus move away when a blunt instrument is introduced (Fig. 2). When compared to the distracted and slightly plantarflexed position, nerves and vessels are stretched and cannot move, making them more vulnerable to iatrogenic damage (Fig. 3). In a recent series concerning ankle arthroscopy, foot distraction straps and continuous application of distraction were used during portal creation and throughout the procedure. The continuous distraction was held responsible for the high percentage (5.4 %) of neurological complications [45]. Our percentage of neurological complications of 1.9 % compares favourably to the average of 3.7 % reported in the literature [2, 3, 11, 14, 17, 18, 26, 27, 37]. We believe that this technique of portal creation and introduction and changing of instruments in a dorsiflexed position without joint distraction is the main reason for the low percentage of nerve complications in our series. Another important aspect of working in the anterior joint area is the distance between the anterior joint space and the overlying neurovascular structures. It has been recently shown that the distance between the anterior tibial rim and the neurovascular structures is significantly reduced with the ankle in the distracted position [8]. The neutral or dorsiflexed position without distraction increases the anterior safe working area [8].

Several authors mention the occurrence of a false aneurysm or other vascular lesions [5, 7, 20‐23, 25, 28, 30, 31, 43]. Although the anatomical course of the anterior tibial artery is known, medial deviation occurs in approximately 3.5 % and lateral deviation in 5.5 % of patients in anatomical studies [42, 44]. In an anatomical analysis, it was found that in 6.2 % the anterior tibial artery was located near the anterolateral portal [33]. Vessels are easily damaged when they do not follow the best-known anatomical route [12]. It is important to always pre-operatively palpate the pulsations to detect where arteries are located. Working in dorsiflexion helps to prevent vascular complications not only because it allows for a larger safe working area, but the vessels are also less likely to be damaged since they can move aside when accidently touched by a blunt instrument.

Another aspect in the prevention of complications is the position of the shaver. In the anterior working area, the opening of the shaver should be directed towards the ankle joint, and not towards the skin (Fig. 4). In a dorsiflexed, undistracted position, this is easily accomplished, because relaxation of the anterior structures of the ankle allows for a considerable working space. In a distracted ankle, however, this space is limited [8].

Looking specifically at complications in hindfoot endoscopy, we found a rate of 2.3 %. This is lower than what was found in a recent study, specifically on this subject, by Nickisch et al. [27]. One explanation may be that in 31.7 % a distraction technique was used in which a tensioned wire was placed transversely through the calcaneus. However, separate rates were not described. Also for hindfoot endoscopy, we recommend intermittent non-invasive distraction to prevent tension on neurovascular structures.

In this series there were 2 patients with a deep infection (0.15 %). In one patient this occurred 12 days after surgery following sinus tract formation, which was not treated with prophylactic antibiotics. The other occurred a few days after surgery. Our percentage of infection compares favourably to the average 0.6 % infection rate in other series (Table 1).

Ferkel et al. [13] have summarized some aspects in prevention of infection after ankle arthroscopy. These include immobilization of the ankle joint the first week post-operatively and the administration of prophylactic antibiotics. We do not believe that these two factors play an important role. On the contrary, we do not immobilize our patients post-operatively but stimulate the patient to move the ankle in active dorsiflexion a few times an hour starting on the day of surgery. One of the great benefits from the minimally invasive surgery over open surgery is this early mobilization. We tolerate patients to mobilize with partial weight bearing on the day of surgery. Patients are stimulated to use their crutches for 4–5 days maximum. Neither do we give prophylactic antibiotics. Our low infection rate seems to justify our protocol.

We agree with Ferkel et al. that it is important to suture the portals, and we do this in all patients. The most important precaution to prevent infection, as well as to prevent sinus tract formation, however, is probably to take care not to remove the subcutaneous tissue at the portal site. In a dorsiflexed position the anterior working area opens up when fluid is brought into the joint. The skin and subcutaneous tissues thus move away from the joint. In this position it is possible to perform the operative procedure without removing subcutaneous tissue. In a distracted, slightly plantarflexed position the anterior working area disappears. The skin and stretched subcutaneous layers move towards the joint [8], making them more vulnerable to iatrogenic removal. Removal of subcutaneous fatty tissue at the portal site creates an easy entry point for bacteria. Limited tourniquet time is another important factor. Before surgery the patient should have a proper diagnosis. Current additional diagnostics like MRI or CT scan make it possible to achieve this. A treatment plan must have been made pre-operatively. Diagnostic arthroscopy preceding an arthroscopic intervention unnecessarily lengthens the procedure. We therefore feel that diagnostic arthroscopy is obsolete.

The application of traction has been mentioned earlier as a possible cause of complications [45]. Concerning distraction, we can identify three options: invasive skeletal distraction, continuous soft tissue distraction or intermittent soft tissue distraction [29, 39]. Invasive skeletal distraction has been associated with a number of specific complications like fracture of the fibula, stress fracture of the tibia, pin breakage, vascular damage, ligament damage, pintrack drainage and pintrack infection [14, 34]. Neurological complications following skeletal distraction and continuous soft tissue distraction have been described by several authors [1, 14, 19, 37].

Soft tissue distraction increases the risk of complications because it creates tension on neurovascular structures and gives direct pressure on the skin. Dorsiflexion without distraction provides relaxed structures that are subsequently less prone to be injured. Tension on nerves and vessels should be avoided when the incision is made. Portals should therefore be created while the ankle is in a neutral or dorsiflexed position. If during the procedure distraction is desired to facilitate working in the posterior joint compartment, it can be accomplished by using a resterilizable non-invasive ankle distraction device [39]. The main advantage of the device is that the choice to perform distraction can be made at any time during the arthroscopic procedure. Applying the distraction only when indicated limits the overall distraction time. In our series of patients, none of the complications could be related to the use of this intermittent soft tissue distraction.

For most operative procedures, there is a learning curve. Consequently, a higher percentage of complications can be expected during the initial procedures. We could not demonstrate a learning curve in our series, however. For the surgeons in this study that performed more than 20 procedures, the complication rate was indeed lower when compared to the low volume surgeons. It was 5.7 % for their first 10 procedures, versus 3.5 % overall. However, the difference was not statistically significant. Most likely, there is a learning curve, but we could not demonstrate this because of the low number of complications.

We analysed whether we could identify intrinsic factors like demographics or type of procedure that influence the percentage of complications. Although there were differences in complication rates between the indication groups, no significant difference could be demonstrated between them, which could be due to the low overall rate of complications. The only variable of significant influence on the complication rate was age of the patient.

The present study has some limitations, concerning study design and possible under-registration of complications. Our study design did not allow us to compare the complication rate of different surgical techniques within our study. Since we only use one well-established technique in our hospital, we cannot compare complication rates for different surgical techniques within our study. This may lead to bias. Furthermore, we failed to identify statistically significant risk factors, accept for age and the effect of a learning curve. It is very well possible that some risk factors did not reach significance because a larger group of patients is needed. Also, our patient population was relatively heterogenous, having undergone a variety of procedures for several different diagnoses. Finally, certain complications like iatrogenous cartilage damage are difficult to examine and therefore not scored. This will have led to an under-registration of complications. However, this complication is not scored in other studies either, and complication rates in published literature can still be compared.

The current study is relevant because it shows a low complication rate related to a specific arthroscopy technique. Today, many different techniques are used to perform anterior and posterior ankle arthroscopy. One should always aim for the lowest possible number of complications. Use of a meticulous technique can aid in reaching this goal.

We conclude that anterior ankle arthroscopy by means of a dorsiflexion approach and a 2-portal hindfoot approach for posterior ankle arthroscopy lead to low complication rate. This fact points towards the benefits of working with the dorsiflexion method and with the use of intermittent soft tissue distraction. The overall percentage of complications for hindfoot endoscopy compares favourably to anterior ankle arthroscopy (2.3 vs 3.5 %).