Beneficial and limiting factors for return to work following anterior cruciate ligament reconstruction: a retrospective cohort study

A retrospective cohort study was combined with a cross-sectional questionnaire survey. All consecutive patients, aged 18–65 years, who underwent primary ACL reconstruction (single bundle semitendinosus-gracilis autograft) at Noordwest Ziekenhuisgroep, Alkmaar, the Netherlands, between April 1st 2012 and April 1st 2014, were screened for inclusion. Patients included had to be part of the working population and have sufficient knowledge of the Dutch language to complete a questionnaire to be eligible for inclusion. Patients were excluded if they had concomitant collateral ligament injuries and/or posterior cruciate injury and/or ACL revision. This study was performed in accordance with the guidelines on Good Clinical Practice of the International Conference on Harmonisation [23] and the Declaration of Helsinki [24]. However, medical ethical approval of the study was not required.

The electronic database of the hospital was searched for patients that fulfilled the inclusion criteria listed above. All patients eligible for inclusion received an invitation letter, between April 1st 2013 and June 1st 2014, containing information regarding the study, an informed consent form, a questionnaire and a stamped return envelope. Patients not responding to this invitation were contacted by telephone and asked if they had received the invitation letter. If not, a new invitation was sent. If so, they were asked to respond to the invitation. Reasons for not wishing to participate were registered if given. The files of patients that agreed to participate were screened and relevant information was extracted.

The primary outcome variable was the number of days between the date of surgery and the date the patient fully returned to work. To identify potential beneficial and limiting factors for a rapid RTW, the continuous RTW variable was dichotomized into rapid and prolonged RTW with the median value of RTW as cut-off point [25, 26].

The independent (predictor) variables of interest on the basis of previous literature and also on clinical expertise were patient-, injury-, surgery-, sports-, work-, and rehabilitation-related factors (Table 1). These factors were retrospectively assessed from the patient files or the questionnaire.

Patient characteristics assessed from the patient files were gender (male–female), date of birth, and body mass index (BMI). Pain, knee disability, and function were assessed using the validated Dutch version of the subjective International Knee Documentation Committee (IKDC) [27, 28]. The IKDC contains 18 items that measure symptoms, function, and sports activities in patients with a variety of knee problems. The IKDC was proven valid when evaluating patients with recent ACL ruptures or those within 1 year of an ACL reconstruction [27, 28]. The score varies between 0 and 100. A higher score means a higher level of function and a lower level of symptoms. In addition, the quality of life (QoL) as perceived by the patient was assessed using the EuroQol-5D (EQ-5D). This questionnaire is a generic standardized health-related QoL measurement tool and is widely used in clinical trials, observational studies, and other studies. The questionnaire measures five dimensions: mobility, self-care, daily activities, pain/discomfort, and anxiety/depression. Each dimension can be assessed at three levels: no problems, some problems, and extreme problems. These five dimensions provide a health profile that can be converted into a weighted health index for each patient. In this study, Dutch tariffs were used to estimate the weighted health index [29]. The weighted health index results in a score between 0 (death) and 1 (completely healthy). The EQ-5D uses also a visual analog scale for overall health.

Injury-related characteristics assessed from the patient files were the side of the ACL injury (left–right), extension lag of the affected knee before the ACL surgery (in degrees), and time interval between injury and surgery (dichotomized into short or prolonged with a cut-off point at three months [14, 15, 30]).

Characteristics of surgery assessed from the patient files were meniscal injury as observed by the surgeon (yes–no) and cartilage damage assessed using the guidelines of the International Cartilage Repair Society, grade zero (normal) up to grade four (severely abnormal) [31].

Sports characteristics assessed using the questionnaire were practicing knee-demanding sports before injury (yes–no), practicing contact sports before injury (yes–no), and practicing pivoting sports before injury (yes–no). In addition, the activity level was assessed using the Tegner Activity Score (TAS). The TAS is retrieved from patient files before ACL injury and between ACL injury and ACL surgery, both assessed one week before surgery by the physician assistant. At follow-up, the TAS was assessed by means of a questionnaire. For this, patients had to rate their level of activity by an integer between 0 (‘sick leave or disability pension because of knee problems’) and 10 (‘competitive sports, soccer—national and international elite’) [32, 33]. In addition, the patients were asked whether they had physical contact with another person at the time of injury (yes–no).

Work-related characteristics were assessed using the Work Osteoarthritis or joint Replacement Questionnaire (WORQ) to determine knee-demanding workload [34]. This is a questionnaire for RTW, first validated for patients with a Total Knee Prosthesis [22]. The questionnaire focuses on knee-demanding activities during work. Heavy knee-demanding work in this study is defined as ‘often’ or ‘nearly always’ performing one of the following five work-related activities: crouching, kneeling, clambering, lifting or carrying, or taking the stairs [22]. It also assesses the difficulty patients experience when performing the above-mentioned set of knee-demanding activities during work. A score between 0 and 100 is generated, a lower score indicating that the patient experiences more difficulty during knee-demanding activities.

Characteristics of rehabilitation assessed using the questionnaire were the number of physiotherapy visits per week and the duration of physiotherapy after surgery in months, additional fitness training during rehabilitation (yes–no), additional home exercise performed during rehabilitation (yes–no), and number of weeks of walking with the aid of crutches.

Descriptive statistics were used to characterize the study population. Mean and standard deviation (SD) were used for normally distributed variables, and median and interquartile range (IQR) for not normally distributed variables. The normality of distributions of the variables was explored visually using histograms, q–q plots, and box-plots, and using the Kolmogorov–Smirnov test.

To determine whether the respondents of the questionnaire were representative for the study population, a non-response analysis was performed. Patient and surgical characteristics from the patient files were compared between the patients that responded to the questionnaire and those that did not. In the case of normally distributed continuous variables, independent t tests were performed. In the case of not normally distributed continuous variables, the Mann–Whitney U test was used and for categorical or dichotomous variables differences were explored using Chi-square tests.

The number of days between the date of surgery and the date the patient fully returned to work was used to determine when patients achieved full RTW. The continuous RTW variable was dichotomized into rapid and prolonged RTW with the median value of RTW as cut-off point [25, 26]. To explore whether and which of the independent variables were most strongly associated with the dependent variable fully RTW, a logistic backward stepwise regression was used. Hot deck imputation was used for missing data [35]. To further reduce the number of independent variables that were initially entered into the regression model, considering the number of cases in the analyses, only those independent variables were used that were univariately associated with fully RTW at a p value of ≤0.25 [36]. Before running the logistic backward stepwise regression analyses, the independent variables were checked for multicollinearity (r > 0.80) [37]. For the remaining independent variables, the logistic backward stepwise regression was performed with time between surgery and filling in the questionnaire forced into the regression. For each of the remaining independent variables that remained in the regression model after backward stepwise regression, their odds ratios (OR) and corresponding 90% confidence intervals (CI) were determined. Because this is the first study on this topic, a 90% CI was chosen to prevent possibly relevant clinical variables being opted out. The overall fit of the logistic regression models was quantified by Nagelkerke’s R ² (R _N ² ), which can be interpreted as R ² in linear regression [38].

All statistical analyses were performed using IBM SPSS Statistics version 20.0 (IBM Corporation, Armon, NY, USA).

Patient inclusion is displayed in a flow diagram (Fig. 1). A total of 218 patients had received an ACL reconstruction in the hospital in the period between April 1st 2012 and April 1st 2014. After the initial screening, 33 patients appeared not to be eligible for inclusion, because they were younger than 18, had other ligament injuries, had an ACL revision, or had insufficient knowledge of the Dutch language. Therefore, 185 patients were contacted and received a questionnaire. One hundred and twenty-five (125) patients returned the questionnaire, a response rate of 68%. Only five patients gave a reason for not returning the questionnaire: they did not want to participate, lived abroad, or had no surgery. Additional 55 patients did not respond at all. Non-response analysis showed no significant differences in pre-operative age, body mass index, TAS before ACL injury, TAS between ACL injury and ACL reconstruction, meniscal injury, or cartilage damage between the response and non-response group.

Descriptive data of the study participants are displayed in Table 2. Study participants had a median (IQR) age at surgery of 31 (24–44) years, a mean (SD) body mass index of 24 (3) kg/m², a median (IQR) TAS before ACL injury of 7 (6–9), a mean (SD) TAS between ACL injury and surgery of 3 (1), and a median (IQR) time from injury to surgery of 8 (4–13) months. Characteristics of study participants (median (IQR)) assessed by follow-up questionnaire were an IKDC score of 81 (66–89), an EQ-5D index score of 0.8 (0.8–1), a WORQ difficulty score of 89 (77–95), and a number of weeks of crutch walking of 5 (3–6) weeks.

Of the 125 patients that received the questionnaire, 36 patients had no work and were not part of the working population, and so they were excluded. Thus, 89 patients were eligible for inclusion. Eighty-two patients (92%) had fully returned to work at the moment of filling in the questionnaire, and seven patients had started to work but had not yet returned to their full work activities (Table 2). The median fully RTW time was 78 days with an IQR of 49–112 days.

Three predictor factors were associated with the binary outcome variable fully RTW (≤78, >78 days) at a p value equal to or less than 0.25 in the univariate analyses (Table 3). Time between ACL injury and ACL reconstruction (p = 0.22), knee-demanding work (p ≤ 0.01), and number of weeks walking with the aid of crutches (p = 0.01) were entered in the backward stepwise regression analysis with time between surgery and filling in the questionnaire forced in the model. Table 4 presents the final model. The odds for fully RTW ≤78 days were 5.4 times greater for patients who perform light knee-demanding work compared with patients who perform heavy knee-demanding work (OR = 5.40, 90% CI 2.24–13.06). Furthermore, also walking aided by crutches was significantly associated with fully RTW with the OR (OR = 1.47, 90% CI 1.16–1.85), indicating an increase in risk for RTW > 78 days for an increase in the number of weeks that patients needed to walk with crutches. The model explained 29% of the variance of fully RTW (Nagelkerke R ²).

The strongest predictive factor for RTW after ACL reconstruction was the level of knee-demanding work, assessed with the WORQ [34]. One study [22] previously used the WORQ for defining the level of knee-demanding work and a similar definition is also used in the current study. The WORQ was applied in patients that received a primary total knee arthroplasty. When comparing the time to fully RTW after a total knee arthroplasty with the time to fully RTW after ACL reconstruction, the results are better for ACL reconstruction, with 71% of ACL patients returning within 3 months, while 50% of total knee arthroplasty patients returned within three months. The rate of RTW is also better for ACL reconstruction compared with total knee arthroplasty, respectively, 92 and 80%. There are no other knee-related surgery studies that describe RTW. Following ACL reconstruction, the first 3 months is most vulnerable for the fixation of the graft. Therefore, it was expected that patients who have knee-demanding work will have a prolonged RTW, because the graft fixation is too weak for knee-demanding work. Thirty-eight patients (46%) performed heavy knee-demanding work in this study. Two-thirds of the patients (68%) that perform heavy knee-demanding work had a prolonged RTW, while one-third of the patients that perform light knee-demanding work had a prolonged RTW.

The other predictive factor that was significantly associated with full RTW > 78 days in the final regression model was the number of weeks that patients needed to walk with the aid of crutches following surgery. The significant OR of 1.54 indicates that for each week, the patient walks longer with the aid of crutches, the risk for prolonged RTW is 1.54 times greater. This means that for each four weeks, the OR increases to 4.57. The number of weeks of walking with the aid of crutches that are needed following ACL reconstruction has never been investigated in the previous studies, as far as we know. The main question is whether walking with the aid of crutches for more weeks—for instance, because the medical specialists advise doing so—actually increases the risk of prolonged RTW. Another reason might be that patients avoid the workplace anyway, because they are walking with the aid of crutches, possibly because they cannot travel to work or they can perform only sedentary work for the first few weeks, instead of their actual knee-demanding tasks, or it was too painful to work. It may be that patients with insufficient knee function need to walk with crutches for a longer period of time and that post-operative knee function is the actual cause of the prolonged RTW. In clinical settings, the patients are usually advised to walk with crutches as long as they have no active knee extension during walking. This suggests that walking with crutches is not an independent predictor for RTW but an intermediate factor for functional recovery. Considering the arguments stated above, both the knee-demanding workload and the number of weeks walking with the aid of crutches should be carefully considered as predictive factors for RTW. For further research, it is recommended to measure post-operative knee function and active knee extension during rehabilitation. In addition, it is recommended to know the primary factors why patients do not return to work in the first few weeks, or understand the factors that ensured RTW. Furthermore, it should be known whether the patient is able to start working, while they are still walking with the aid of crutches. Therefore, more scientific evidence is required for RTW after knee surgery, especially after ACL reconstruction.

In contrast with the beneficial and limiting factors for return to sports following ALC reconstruction, no significant association was found for factors, such as body mass index, TAS before ACL injury, TAS between ACL injury and ACL reconstruction, physical contact during injury, physiotherapy, additional home exercises, or the other included potential predictive factors (Table 1) [2, 11, 17‐21]. Return to sports is possible after 9–12 months of rehabilitation [14], while fully RTW is possible around 3 months after surgery. This difference in time between return to sports and fully RTW might explain the differences in the limiting and beneficial factors between return to sports and RTW. After 9–12 months, the patients will have recovered more than after 3 months: this might explain the difference in factors at stake. Extension lag of the affected knee before the ACL surgery is a limiting factor for return to sports [2, 12, 14]. In this study, there were three patients who had an extension lag before surgery. Since these three patients had a longer fully RTW time than 78 days, it is possible that this factor also has an effect on fully RTW. However, the number of patients is too small to find an effect on RTW. Furthermore, the median age in this study population was 31 years, which is higher than in other ACL studies focusing on return to sports. This can be explained by the inclusion criteria—patients had to be part of the working population. However, with respect to the other characteristics, the study population in this study was comparable with study populations of other ACL studies, so that these cannot account for the differences in beneficial and limiting factors.

This study shows that patients who perform heavy knee-demanding work and need to undergo ACL reconstruction should be advised to discuss with their employers and/or occupational physician the possibility to perform adapted work activities with low demands on the knees. This might result in a more rapid RTW. In addition, with respect to the number of weeks in which walking is aided with crutches, it may be important to practice walking without the aid of crutches and performing active knee extensions guided by a physiotherapist to reduce the number of weeks walking with crutches.

This was the first, exploratory, study conducted to investigate beneficial and limiting factors for RTW following ACL reconstruction. Both the medical literature concerning the predictive factors for return to sports and clinical expertise were used to determine potential beneficial and limiting factors evaluated in this retrospective cohort study, which resulted in a large number of potential predictive factors. Another strength of the study was the use of several questionnaires with sufficient clinimetric qualities, such as the IKDC, WORQ, and EQ-5D. In addition, since non-response analysis showed no significant differences between the patients that responded (68%) and the group that did not respond, the responders were deemed a representative sample of the actual population. The seven patients that had not yet returned to work did vary largely in the time between ACL reconstruction and filling out the follow-up questionnaire, which means that there was no association between the rate of RTW and follow-up time, this could be expected, since a longer follow-up time increases the chance of fully RTW. Moreover, almost 85% of the patients with a follow-up time less than 3 months returned to work. These results were similar for patients with a follow-up period of up to 24 months, showing that, although time between surgery and follow-up was purposely included in the final regression model, follow-up time could not have biased the results.

An important limitation of the study is that the patients answered the questionnaire retrospectively, thereby introducing the risk of recall bias. We have partly tried to overcome this bias by making categories for number of variables, most often in two categories. A prospective study including the factors found for RTW might give more reliable and precise estimates. Another possible limitation is that several studies have reported that a higher activity level before ACL injury increases the likelihood of return to sports [12, 13]. In this study, the activity level was measured using the TAS. This score has not yet been validated in Dutch, which is potentially a restriction in this study [14]. Moreover, the TAS before ACL injury had a large number of missing values. These missing values were imputed, but might result in imputation bias. This is another argument for conducting a prospective cohort study on RTW and ACL reconstruction.