Double-bundle anterior cruciate ligament reconstruction is superior to single-bundle reconstruction in terms of revision frequency: a study of 22,460 patients from the Swedish National Knee Ligament Register

The SNKLR is a nationwide database established in January 2005, and 92.9% of all eligible units for ACL reconstruction in Sweden are linked to the register. Over 90% of the ACL reconstructions annually performed in Sweden are therefore registered [6, 15]. The register is sectioned in one data-registration for surgeons and one for patients. Information about activity at the time of injury, date of injury and surgery, fixation method and graft choice is reported. Concomitant injuries, and previous surgery to the knee or the contralateral knee are also registered along with information of all interventions made to the injured knee. Revision surgery or reoperation is separately registered, and the event is correlated with the primary ACL reconstruction. The patient reported outcome is evaluated through two questionnaires, the EQ-5D (European Quality of Life-5 Dimensions) and the KOOS (Knee injury and Osteoarthritis Outcome Score). The patients’ response rate preoperatively and at 1, 2 and 5 years range from 38 to 72% [15].

Through an online-based questionnaire, detailed information about surgical technique(s) performed by ACL surgeons in Sweden could be collected and matched to a specific ACL reconstruction, a method first described by Desai et al. [4]. The questionnaire was based on the items in the AARSC [23]. All the 175 surgeons registered in the SNKLR as of 31 December 2014 had the possibility to complete the questionnaire online between Jan 2015 and 30 April 2015. The response rate to the questionnaire was 61.7% [4]. The design of the questionnaire made it possible to identify which surgical technique(s) each surgeon performed during a specific period of time, thereby matching each patient’s ACL reconstruction to a specific technique.

The single-bundle reconstructions were sorted into groups defined by which surgical techniques had been implemented. In the present study, the same method and pre-defined groups were used as it was first described by Desai [4]. By combining eight different items from the AARSC, five groups were generated that were characterized by a specific combination of fulfilled items from the AARSC. The combinations of items were selected in order to produce homogenous groups with regard to surgical technique and size. Thus, for a specific ACL reconstruction to be placed in one of the groups the surgeon who performed the reconstruction must have answered to meet the requirements of the group. Each group has at least a few mandatory “yes” or “no” answers regarding the technique used, while some items are considered non-mandatory in some of the groups (Table 2). All single-bundle procedures meeting the inclusion criteria of the study could not be matched with a subgroup since not all surgeons responded to the survey and their surgical technique remains unknown. Furthermore, if a surgeon for any reason could not specify which techniques he or she implemented during a period of time, the reconstructions performed by that surgeon during that interval was not included in a subgroup. The limited amount of double-bundle reconstructions registered during the period of interest made it difficult to further subgroup these reconstructions. The double-bundle group was analysed as one unit, regardless of which surgical techniques had been implemented in the reconstructions included.

This cohort study was conducted according to the WMA Declaration of Helsinki. Participation in the Swedish National Knee Ligament Register is voluntary for patients and surgeons. No written consent is necessary for national databases in Sweden. Investigators had access only to unidentifiable patient data. The study was approved by the Regional Ethics Committee in Gothenburg, Sweden (reference number: 760-14).

Tables and diagrams were generated using Microsoft Excel for Windows (Version 14.0.7, Microsoft Corp, Redmond, Washington, USA). A statistician assigned to the SNKLR performed all statistical analyses. Statistical analysis was performed in IBM SPSS statistics (version 23.0, IBM Corp, Armonk, New York, USA). The data were summarized using counts and percentages for descriptive data and means ± SDs and median and range for patient-reported outcome data. The end point of revision surgery was analysed as time-to-event outcomes using Cox proportional hazards regression. Kaplan–Meier curves and log minus log plots were used to visually test the assumption of proportionality. All survival estimates and hazard ratios (HRs) were reported with 95% confidence intervals (CI). Statistical significance was defined as a 95% CI for hazard ratios not including 1.0, and alpha was set to 0.05. Additionally, a multivariate analysis adjusted for possible confounding factors (age and patient sex) was performed using a Cox proportional hazards regression expressed as HR and 95% CI.