A medical error is defined as an act of omission or commission in planning or execution that contributes or could contribute to an unintended result [16]. A complication was defined according to the Association of Surgeons of the Netherlands as a condition or event, unfavourable to the patient’s health, causing irreversible damage or requiring a change in therapeutic policy [17]. An error has the potential to cause a complication. Both are considered to be preventable when there is a failure to follow accepted practice at the individual or system level. The definitions used were accepted by the entire surgical staff, and were used when documenting errors and complications.

This study was conducted in a secondary referral hospital and level 1 trauma centre, with a capacity of 673 beds. The surgical department consisted of 12–15 surgical residents, 8–10 consultant surgeons and 3–4 junior staff surgeons.

At the beginning of 1995 an electronic medical record was introduced in the hospital. The software used for the electronic medical record was an Oracle^® Forms (Redwood City, CA, USA) application with an Oracle database as back-end, which was developed by the hospital itself.

The electronic medical record has an integrated system for classifying complications developed by the Trauma Registry of the American College of Surgeons (TRACS which is further described elsewhere [18]). The TRACS system was originally developed as a complication list to record morbidity in trauma populations [19]. The list explicitly defines complications and uses four-digit codes. An advantage of the system is that it also allows registration of medical errors by specific codes [20, 21]. In the hospital, this integrated TRACS system allows physicians to register errors and complications in the operating room, in the wards, or in the outpatient clinic, regardless of patient outcome.

All events recorded are discussed during the daily surgical conference before final storage in the database. Only events judged by consensus to meet the definitions mentioned above were recorded in the complication registry. Furthermore, all procedures performed were analysed during the daily surgical conference along with the radiographs from the procedure and postoperative period.

In the present study, all osteosynthesis procedures performed between January 1, 2000 and December 31, 2010 were collected by performing a search in the Electronic Medical Record (EMR) database. For all patients operated during this study period, any error or complication due to the osteosynthesis procedure until discharge from outpatient follow-up was registered in the TRACS system.

The data of all osteosynthesis procedures performed during the study period were collected from the operating room database. This included fracture location, patient’s age, surgeon’s specialisation, presence of an orthopaedic trauma surgeon, elective or emergency setting, starting time, duration of surgery and need for a reoperation.

During the study period no osteosynthesis procedure was performed by residents without the supervision of a surgeon, independent of the field of interest of this surgeon. In this study, only surgeons who completed the full training (general training plus trauma differentiation) and received their certification by the Dutch Society of Trauma Surgery were regarded as orthopaedic trauma surgeon. During the study period the average time of experience per orthopaedic trauma surgeon was about 5–10 years after completion of their training. The involvement of an orthopaedic trauma surgeon in this study means the surgeon has scrubbed in and played an active role in the osteosynthesis procedure.

The EMR also provides a fracture template that can be completed by the surgeon after the procedure. It contains a fracture classification (Müller AO Classification of Fractures [22]) and a description of the soft tissue injury (Gustilo-Anderson Classification [23]), the degree of contamination (Surgical wound classification by the National Academy of Sciences [24, 25]) and the type of osteosynthesis performed.

Osteosynthesis procedures that are scheduled, at least 24 h ahead, are labelled as elective (non urgent) procedures. Emergency procedures are defined as procedures performed within 24 h after injury presentation, with a subcategory of procedures that are performed within 2 h.

The starting times of the procedures are divided into office hours (between 7:30 a.m. and 5:00 p.m.) and outside office hours (between 5:00 p.m. and 7:30 a.m., including Saturday and Sunday).

All errors in osteosynthesis were reviewed and subdivided in the following subcategories: inadequate fracture reduction, use of wrong implant, wrong length of implant, incorrect implant positioning, incorrect use of implant, or error in surgical approach. A certified orthopaedic trauma surgeon determined the fracture type using the Müller AO Classification of Fractures on the available conventional radiographic recordings [22].

Descriptive analysis was performed in order to compare the circumstances (operation variables and patient variables) between procedures in which an error occurred and procedures in which no error occurred. The same was done for procedures followed by a complication and procedures not followed by a complication.

The characteristics of procedures in which an error occurred and procedures in which no error occurred were compared. Similarly, procedures that were followed by a complication were compared with those without complications. Finally, the rate of errors and complications linked to osteosynthesis procedures performed before and after July 1, 2009 were compared. From that date onwards, the involvement of a dedicated orthopaedic trauma surgeon became a requirement for conducting an osteosynthesis procedure. Pearson’s Chi squared tests were used in order to compare categorical variables. Odds ratios were calculated for categorical independent variables. Independent Student’s T tests were used to compare parametric continuous data. Mann–Whitney U tests were used to compare nonparametric continuous data. Differences were considered to be significant at a p level < 0.05.

Data were analysed using the Statistical Package for the Social Sciences (SPSS) version 16.0 (SPSS, Chicago, IL, USA).