Incidence and risk factors for revision and contralateral anterior cruciate ligament reconstruction: A population-based retrospective cohort study

Abstract

There is a limited data on epidemiology of primary and recurrent anterior cruciate ligament reconstruction (ACLR) in Canada. The objectives of this study were to examine the incidence and factors associated with recurrent ACLR (revision and contralateral ACLR) in a western Canadian province of Alberta. We conducted a retrospective cohort study with an average follow up of 5.7 years. Albertans aged 10 to 60 years with a history of primary ACLR between 2010/11 to 2015/16 were included in the study. Participants were followed up until March 2019 to observe outcomes: Ipsilateral revision ACLR and contralateral ACLR. Kaplan Meir approach was used to estimate event free survival and Cox proportional hazard regression analysis was conducted to identify associated factors. Of the total participants with a history of primary ACLR on a single knee (n = 9292), n = 359, 3.9% (95% confidence interval: 3.5–4.3) underwent a revision ACLR. A similar proportion among those having primary ACLR on either knee (n = 9676), n = 344, 3.6% (95% confidence interval: 3.2–3.9) underwent a contralateral primary ACLR. Young age (<30 years) was associated with increased risk of contralateral ACLR. Similarly, young age (<30 years), having initial primary ACLR in winter and having allograft were associated with a risk of revision ACLR. Clinicians can use these findings in their clinical practice and designing rehabilitation plans as well as to educate patients about their risk for recurrent anterior cruciate ligament tear and graft failure.

1. Introduction

Anterior cruciate ligament reconstruction (ACLR) is a commonly performed procedure to restore functional stability of the knee after a torn anterior cruciate ligament (ACL).^[1] It is suggested that timely ACLR can prevent further damage to the meniscus and other knee ligaments.^[2] However, there is a variation in graft choice, surgical technique including tunnel placement as well as rehabilitation practices after ACLR.^[3]

Multiple studies have investigated revision ACLR rate and associated risk factors.^[3–8] Commonly reported risk factors are age, graft type, body mass index, smoking status, and concomitant ligament injuries. Previously reported revision rates ranged from 2.6% to 8.4%. Furthermore, the variables included in multivariable models greatly differ across studies with only a few studies including a combination of patient related, surgery related and provider related variables. Additionally, few studies have investigated the rate and factors associated with primary ACLR in the contralateral knee.^[5,9,10]

The purpose of this study was to investigate the incidence and factors associated with ipsilateral ACLR revision and contralateral primary ACLR. among individuals with a primary ACLR between the fiscal years 2010/11 and 2015/2016. This population-based study will provide a reference for monitoring trends of subsequent ACLR and may be helpful for future quality assessment projects in Canada. In addition, it may inform clinicians on factors that may help to improve patient outcomes. We hypothesized that patient related, provider related and surgery related factors would influence the risk of revision and contralateral ACLR. Specifically, we hypothesized that contralateral ACLR rate would be higher than ipsilateral revision rate and younger age have a high risk of recurrent ACLR compared to older age.

2. Methods

2.1. Setting

Alberta has a population of over 4.4 million. Alberta has a publicly funded health care system that guarantees universal access to hospital and medical services to all Albertans. Alberta has acute knee injury clinics in Edmonton and Calgary. In Calgary, the acute knee injury clinic (AKIC) program was launched full time in 2010 (personal communication) and in Edmonton since 2013 (personal communication). Alberta knee clinics offer specialized diagnostic and treatment services for knee injuries with a shorter wait time than in a general hospital setting. This study received ethical approval from the University of Alberta Health Research Ethics Board (Pro00090820).

2.2. Study design

We used a retrospective cohort study design with a minimum of 3 years of patient follow up.

2.3. Data

Administrative data from the fiscal years 2010/11 until 2018/19 were used to identify ACL reconstructions performed during the study period. Databases available from Alberta Ministry of Health were linked deterministically using a unique identifier. Population registry has demographic and geographic information on all individuals registered with the Alberta Health Care Insurance program. The National Ambulatory Care Reporting system database includes data on emergency department visits and same day procedures. The discharge abstract database has information on all inpatient cases from hospitals including free- standing rehabilitation hospitals. Practitioners claims database has information on processed claims on fee for service claims from all medical practitioners to Alberta Health.

2.4. Primary outcome

Practitioner billing codes for primary ACLR: (Primary ACL-93.45A), Primary ACLR with meniscal repair (93.45D) and Primary ACLR with meniscectomy (93.45 C) were used to identify cases of primary ACLR. There was no variable to determine laterality, hence, if there were 2 records for primary ACLR for the same person on 2 different dates then the second primary ACLR was considered as contralateral primary ACLR. Similarly, billing codes for revision ACL reconstruction: 93.45E (revision ACL), 93.45F(revision ACL with meniscus repair) and 93.45J (revision ACL with meniscectomy) were used to identify ACL revision. Canadian classification of health intervention (CCI) codes for ACLR were combined with practitioner billing codes to confirm the procedures performed. The CCI codes for ACLR were accessed from a previous study^[5] available in Table S1, Supplemental Digital Content, https://links.lww.com/MD/I931.

2.5. Demographic and clinical variables

We used demographic data available inpatient registers. Patient related variables were age, sex, income quintile, place of residence and Charlson Comorbidity Index.^[11] Age was divided into 6 categories: 10 to 19 years, 20 to 29 years, 30 to 39 years, 40 to 49 years and 50 to 59 years. Sex was categorized into males and females. Urban-rural residence status was obtained by looking at the initial 3 letter forward sortation areas in postal code. A forward sortation area with a zero “0” in the middle would indicate a rural residence.^[12] Socio economic status quintile was obtained by linking the postal codes with 2011 National Household Survey data to derive neighborhood income quintiles. The comorbidity index was created using Charlson Deyo procedure for using administrative hospital discharge data.^[11] Two categories: No morbidity, and index >=1 were created. Surgery related variables included: graft type, season of surgery, outpatient/inpatient surgery setting, and concomitant meniscus procedure during primary ACLR. Graft type was categorized into 3 categories: autograft, allograft, synthetic and combined techniques. Season of surgery was defined as Spring (March, April, May), Summer (June, July, August) and Fall (September, October, November) and winter (December, January, February). Surgery years were divided into 2 categories: 2010 to 2014 (representing early years of establishment of AKIC) and 2015 to 2016 (later years of establishment of AKIC). Two thousand and fifteen was chosen as a cutoff year since there was a big increase in the proportion of ACLR conducted in outpatient settings in 2015 compared to 2014. Setting of surgery had 2 categories: outpatient surgery versus inpatient surgery. Presence of concomitant meniscus procedure was divided into 3 categories: ACLR only, ACLR with meniscectomy and ACLR with meniscus repair. Provider related variables include surgeon volume in the last 365 days of initial primary ACLR and average annual hospital volume over 6 years. Surgeon volume 1 year prior to the index operation was calculated and 5 categories were formed as suggested by Wasserstein et al^[5] 0, 1 to 12, 13 to 50. Fifty-one to one hundred, >100. Annual hospital volume was categorized into 3 categories:1 to 24 (V1), 25 to 99 (V2) and > 100(V3) with some adjustment to classification suggested in a previous work.^[13]

2.6. Eligibility criteria

All Albertans who underwent primary ACL reconstructions over 6 years (between 2010/11 and 2015/16) were included in the analysis. Patients below the age of 10 years and above 60 years were excluded since ACL injury is less common in these age groups. If the index ACL reconstruction was recorded as ACL revision, then patients were excluded from the analysis. ACL repairs (93.45 B) were excluded. Primary ACL reconstructions that occurred after April 1, 2016, were excluded in the cox proportional hazard model to allow a minimum of 3 years of follow up period for all ACL reconstructions. Patients having posterior cruciate ligament tears were excluded. When estimating proportion of ACL revision and its associated factors, we excluded all records from individuals having contralateral knee reconstructions since there was no information on sidedness of the knee. To improve the specificity of outcome assessment, all primary reconstructions performed before April 2010 and subsequent contralateral reconstructions, and revision ACL on those patients during the study period were excluded. Sample selection charts for ipsilateral revision and contralateral ACLR are given in Figure S1 and Figure S2, Supplemental Digital Content, https://links.lww.com/MD/I932, respectively.

2.7. Statistical analysis

Outcomes were assessed until March 31, 2019. Proportions of revision and contralateral ACLR by patient related, surgery related and provider related factors were calculated. Survival analysis using the Kaplan Meir approach was used to estimate event free survival for both ipsilateral ACL revision and contralateral primary reconstruction. Censoring was done for death, emigration, occurrence of event of interest or end of follow up period (March 31, 2019), whichever is earlier. Event rates were calculated for age categories and sex expressed as an event rate per 1000 person years. Cox proportional hazard regression model was used to identify association of covariates with the outcomes. Hazard ratios with 95% confidence intervals were generated with level of statistical significance set at 0.05. Participants with missing demographics (age and gender) were excluded. Analyses were carried out on all available cases for respective variables assuming that data were missing completely at random. All analyses were carried out using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina).

3. Results

3.1. Cohort characteristics

We found a total of 17,793 ACLR conducted between 2010/11 and 2018/19 which had a CCI code available in National Ambulatory Care Reporting System or discharge abstract database database. After exclusions based on a predetermined criterion (Fig. S1 and Fig. S2, Supplemental Digital Content, https://links.lww.com/MD/I932), our analysis included a total of 9636 initial primary ACLR performed in either knee and 9292 ACLR performed in a single knee over 6 years between April 2010 and March 2016 in Alberta (Table 1). The annual volume of primary ACL reconstructions meeting our inclusion criteria increased from 1528 in 2010/11 to 1712 in 2015/16. Over half of the participants (56%) were males. Mean age of the participants was 30 years (SD 10.9) and over half of the participants (57%) were between 20 to 39 years of age (Table 1). The cohort was mostly healthy with 99.3% of the participants having a Charlson comorbidity index of 0. About 12% of the participants were from rural areas. Most participants (80%) underwent same day surgery. Of the total number of primary ACL reconstructions performed annually, slightly lower proportions were performed in the summer season (21.5%) compared to spring, fall and winter seasons (25–27%). Nearly 73% of primary ACL reconstruction were performed in hospitals that conducted over 100 ACL reconstructions annually. Average follow up period for this cohort was 5.7 years.

Table 1 - Characteristics of the cohort and proportion of revision and contralateral ACLR.

	Ipsilateral revision ACLR				Contralateral primary ACLR
Variables	Initial cohort (n = 9292)		Revision ACLR (n = 359, 3.9%)		Initial cohort (n = 9636)		Contralateral ACLR (n = 344, 3.6%)
Patient related factors
Age category	n	%	n	%	n	%	n	%
10–19 yr	1955	21.0	150	7.7	2084	21.6	129	6.2
20–29 yr	2955	31.8	127	4.3	3074	31.9	119	3.9
30–39 yr	2351	25.3	54	2.3	2410	25.0	59	2.4
40–49 yr	1502	16.2	23	1.5	1531	15.9	29	1.9
50–59 yr	529	5.7	5	0.9	537	5.6	8	1.5
Gender
Females	4055	43.6	170	4.2	4227	43.9	172	4.1
Males	5237	56.4	189	3.6	5409	56.1	172	3.2
charleson Index
0	9224	99.3	357	3.9	9566	99.3	342	3.6
>=1	68	0.7	2	2.9	70	0.7	2	2.9
Income Quintile
Q1 (poorest)	1319	14.2	50	3.8	1367	14.2	48	3.5
Q2	1373	14.8	53	3.9	1419	14.7	46	3.2
Q3	2081	22.4	82	3.9	2159	22.4	78	3.6
Q4	2148	23.1	77	3.6	2224	23.1	76	3.4
Q5 (richest)	2371	25.5	97	4.1	2467	25.6	96	3.9
Place of residence
Urban	8159	87.8	324	4.0	8458	87.8	299	3.5
Rural	1133	12.2	35	3.1	1178	12.2	45	3.8
Surgery related factors
Surgery years
2011–2014	5998	64.6	260	4.3	6253	64.9	255	4.1
2015–2016	3294	35.5	99	3.0	3383	35.1	89	2.6
Setting of surgery
Outpatient/same d surgery	7495	80.7	271	3.6	7741	80.3	246	3.2
Inpatient	1797	19.3	88	4.9	1895	19.7	98	5.2
Season of surgery
Spring	2460	26.5	82	3.3	2539	26.4	79	3.1
Summer	1991	21.4	69	3.5	2075	21.5	84	4.0
Fall	2494	26.8	99	4.0	2587	26.8	93	3.6
Winter	2347	25.3	109	4.6	2435	25.3	88	3.6
Graft type
Autograft	8379	90.2	325	3.9	8691	90.2	312	3.6
Allograft	594	6.4	28	4.7	618	6.4	25	4.0
Combined, synthetic, unspecified	319	3.4	6	1.9	326	3.4	7	2.1
Concomitant procedure
Primary ACL reconstruction only	1736	18.7	70	4.0	1818	18.9	82	4.5
Primary ACL reconstruction with meniscus repair	5419	58.3	186	3.4	5601	58.1	182	3.2
Primary ACL reconstruction with menisectomy	2137	23.0	103	4.8	2217	23.0	80	3.6
Provider related factors
Hospital volume (Annual)
Low (1–24)	205	2.2	8	3.9	211	2.2	6	2.8
Medium (24–99)	2324	25.0	76	3.3	2397	24.9	73	3.0
High (>=100)	6763	72.8	275	4.1	7028	72.9	265	3.8
Surgeon volume (previous yr)
0	58	0.6	0	0.0	62	0.6	4	6.5
1–12	938	10.1	44	4.7	971	10.1	33	3.4
13–50	3305	35.6	117	3.5	3414	35.4	109	3.2
51–100	1440	15.5	53	3.7	1491	15.5	51	3.4
>100	3551	38.2	145	4.1	3698	38.4	147	4.0

ACL = anterior cruciate ligament, ACLR = anterior cruciate ligament reconstruction, CI = confidence interval.

3.2. Ipsilateral revision ACL reconstruction

Of the 9292 participants with an initial primary ACL reconstruction in a single knee, n = 359 (3.9%, 95% confidence interval [CI]: 3.5–4.3) underwent ipsilateral ACL revision over a mean revision period of 3 years (SD 1.8) (Table 1). The Kaplan–Meier estimate of failure free survival for revision ACLR was 94.6% (Figure S3, Supplemental Digital Content, https://links.lww.com/MD/I933). The event rate for revision ACL was 13.5 per 1000 person years, and 7.5 per 1000 person years among 10 to 19 and 20 to 29 years of age, respectively (Table S2, Supplemental Digital Content, https://links.lww.com/MD/I934). Similarly, the event rate among females was 7.4 per 1000 person years in females and 6.3 per 1000 person years in males (Table S2, Supplemental Digital Content, https://links.lww.com/MD/I934).

Cox proportional hazard regression analysis revealed that age at the time of initial primary ACL reconstruction was strongly associated with the risk of revision (Table 2). Participants aged 10 to 19 years showed 3.5 times higher risk (hazard ratio (HR) = 3.5, 95% CI−2.6 to 4.8), and those aged 20 to 29 years were at nearly double (HR = 1.9, 95% CI−1.4 to 2.7) higher risk than the 30 to 39 years age group (Fig. 1). Participants aged 50 to 59 years showed lower risk compared to 30 to 39 years (HR = 0.4, 95% CI:0.2–1.0).

Table 2 - Multivariable Cox proportional hazard model for ACL revision and contralateral primary ACLR.

	Ipsilateral revision ACLR				Contralateral primary ACLR
Variables	Hazard ratio	95% CI		P value	Hazard ratio	95% CI		P value
Patient related factors
Age category
10–19 yr	3.5	2.6	4.8	<.0001	2.4	1.8	3.3	<.0001
20–29 yr	1.9	1.4	2.7	<.0001	1.6	1.2	2.2	.003
30–39 yr	ref				Ref
40–49 yr	0.7	0.4	1.1	.090	0.7	0.5	1.2	.180
50–59 yr	0.4	0.2	1.0	.049	0.6	0.3	1.2	.127
Gender
Females	Ref
Males	0.9	0.7	1.1	.461	0.8	0.7	1.0	.116
Charlson Index
0	0.9	0.2	3.7	.911	1.0	0.2	3.8	.938
>=1	Ref
Income Quintile
Q1 (poorest)	Ref
Q2	1.0	0.7	1.4	.868	0.9	0.6	1.4	.690
Q3	0.9	0.6	1.4	.659	1.1	0.8	1.7	.520
Q4	0.8	0.5	1.2	.302	1.0	0.7	1.6	.829
Q5 (richest)	1.0	0.7	1.4	.826	1.2	0.8	1.9	.327
Place of residence
Urban	Ref
Rural	0.7	0.5	1.1	.091	1.3	0.8	1.9	.278
Surgery related factors
Surgery yr
2011–2014	ref				Ref
2015–2016	1.1	0.9	1.4	.429	1.1	0.8	1.4	.526
Setting of surgery
Outpatient/same day surgery	Ref
Inpatient	1.3	1.0	1.7	.046	1.5	1.1	1.9	.003
Season of initial surgery
Spring	1.1	0.8	1.5	.576	0.8	0.6	1.2	.288
Summer					Ref
Fall	1.4	1.0	1.9	.054	1.0	0.8	1.4	.833
Winter	1.6	1.2	2.2	.003	1.0	0.8	1.4	.756
Graft type
Autograft	Ref
Allograft	1.5	1.0	2.3	.039	1.3	0.9	2.0	.178
Combined, synthetic, unspecified	0.5	0.2	1.1	.099	0.6	0.3	1.3	.234
Concomitant procedure
Primary ACL reconstruction only	Ref				Ref
Primary ACL reconstruction with meniscus repair	1.0	0.8	1.3	.986	0.8	0.6	1.1	.145
Primary ACL reconstruction with menisectomy	1.1	0.8	1.5	.455	0.7	0.5	1.0	.066
Provider related factors
Hospital volume (Annual)
Low (1–24)	Ref				Ref
Medium (24-99)	0.8	0.4	1.6	.479	1.1	0.4	2.5	.899
High (>=100)	0.9	0.4	1.8	.683	1.1	0.5	2.7	.780
Surgeon volume (previous yr)
0	0.0	0.0	0.0	.960	1.2	0.4	3.5	.706
1–12	1.2	0.8	1.8	.388	0.7	0.5	1.2	.188
13–50	1.0	0.7	1.3	.961	0.8	0.6	1.1	.227
51–100	0.9	0.6	1.2	.439	0.8	0.6	1.1	.146
>100	Ref				Ref

ACL = anterior cruciate ligament, ACLR = anterior cruciate ligament reconstruction.

Figure 1.: Kaplan–Meier survival curve for ipsilateral revision ACLR by age category. ACLR = anterior cruciate ligament reconstruction.

Having an initial surgery in an inpatient setting was associated with a 1.3 times higher chance of revision surgery (HR = 1.3, 95% CI: 1.0–1.7) compared to surgeries conducted in an outpatient setting/same day surgery (Table 2). Similarly, initial surgery in the winter season (HR = 1.6, 95% CI: 1.2–2.2) and having allograft were associated with higher risk of revision surgery compared to patients having an initial reconstruction using autograft (HR = 1.5 95% CI: 1.0–2.3). Provider related factors such as surgeon volume in the previous year, annual hospital volume did not show a statistically significant association.

3.3. Contralateral primary ACL reconstructions

Of the 9676 initial primary ACLR in either knee, n = 344 (3.6%, 95% CI: 3.2–3.9) underwent contralateral primary ACLR over a mean reconstruction period of 3.0 years (SD 2.1) (Table 1). The Kaplan–Meier estimate of failure free survival for contralateral ACLR was 94.7% (Figure S4, Supplemental Digital Content, https://links.lww.com/MD/I935). The event rate for contralateral primary ACLR was 10.9 per 1000 person years, and 6.8 per 1000 person years among 10 to 19 and 20 to 29 years age, respectively (Table S2, Supplemental Digital Content, https://links.lww.com/MD/I934). Similarly, the event rate was 7.1 per 1000 person years in females and 5.6 per 1000 person years in males. Females represent 50% of contralateral ACLR cases compared with 44% of the primary ACLR cases.

Cox proportional hazard regression analysis revealed that age at the time of initial primary ACLR was the only patient related factor associated with the risk of contralateral ACLR with participants 10 to 19 years having more than double (HR = 2.4 95% CI−1.8 to 3.3) higher risk compared to 30 to 39 years age group (Table 2). Similarly, those aged 20 to 29 years were at 1.6 times (HR = 1.6,95% CI−1.2 to 2.2) higher risk than the 30 to 39 years age group (Fig. 2). The participants undergoing contralateral ACLR were 5 years younger [mean age = 25.2 (SD-9.8) years] than the cohort [mean age = 30.0 years (SD-10.9)]. Place of residence or income quintile did not show association with contralateral ACLR.

Figure 2.: Kaplan–Meier survival curve for contralateral ACLR by age category. ACLR = anterior cruciate ligament reconstruction.

No provider related factors were associated with the risk of contralateral ACLR. Having an initial primary ACLR in an inpatient setting was associated with higher risk of contralateral ACLR (HR = 1.5 95% CI 1.1–1.9) compared to outpatient/ same day surgery.

4. Discussion

The most important finding in our study is that age is a significant risk factor for both revision ACLR and contralateral ACLR. Having allograft in primary ACLR showed an association with higher risk of revision ACLR. Finally, having ACLR in the winter was associated with an increased risk of revision ACLR, a finding that has not previously been shown in the literature.

Age was strongly associated with the risk of revision ACLR, and contralateral primary ACLR, a finding consistently reported in published literature.^[6,14,15] Participants aged 10 to 19 years and 20 to 29 were 3.5 times and 1.9 times higher risk than participants aged 30 to 39 years, respectively. Younger participants may have higher preoperative activity levels, tend to return to normal activity earlier and are less compliant with rehabilitation plan.^[3] We also found that older patients aged 50 to 59 years had a lower risk of revision ACLR than patients aged 30 to 39 years. Altogether, these findings suggest that young patients might benefit from patient education and rehabilitation plans to avoid early return to normal activity levels.

The revision ACLR rate (3.9%) and the contralateral primary ACLR rate (3.6%) in our cohort fall within the range reported in some of the previous studies^.[5,14] However, our revision rates and contralateral primary ACLR rate are higher than reported in some studies with 2 years or lower follow up.^[9,15,16] Longitudinal cohort studies with more than 5 or more follow up,^[3,17–19] report higher subsequent ACLR rates than in our study. In a US-based longitudinal follow up study with 6 years of follow up, Hettrich et al^[17] found a ipsilateral revision rate of 7.7% and contralateral ACLR rate of 6.4%. Higher median age of the participants (29 years vs 23 years) and a shorter follow up period (5.6 years versus 6 years) than in Hettrich et al^[17 study might be some of the reasons for lower rate in our study. Another study from Italy with 10 years of follow up among patients undergoing hamstring autograft reported ipsilateral revision ACLR of 3.4% and contralateral ACLR of 7.8%.^[18]

We found a similar revision rate than contralateral ACL rupture. Studies with 6 years or longer follow up have shown that contralateral ACL rupture occurs later than ipsilateral graft rupture.^[18] Of the total contralateral reconstructions within 6 years, Hettrich et al^[17] reported that 40% occurred within 2 years of follow up and 60% occurred after 2 years of follow up. Whereas 63% of graft rupture on the ipsilateral knee occurred within 2 years of follow up. In a 10 year follow up study, Grassi et al^[18] reported double the rate of contralateral ACLR compared to ipsilateral revision ACLR. Another study among professional football players also reported that mean time to ipsilateral graft rupture (23.5 months) was shorter than contralateral ACL rupture (31.5 months).^[20]

We found that use of allograft in the index ACLR was associated with higher risk of undergoing ACL reconstruction compared to use of autograft. This finding is consistent with the published literature.^[5,21] Previous studies have reported that early return to sports and use of allograft in combination further increase the risk of revision ACLR especially in young patients.^[22,23] Since the mean age of our cohort was 30 years, early return to activity might have increased the chances of graft failure among those who had an allograft.

Also of note was the association with surgery in the winter being a risk factor for revision ACLR. To our knowledge, this is the first-time association with season of surgery has been reported. We hypothesize that patients having primary ACLR in winter may be more likely to return to normal activity as soon as the spring/summer season arrives compared to patients having surgery in Summer or Fall. Research has shown that physical activity in Canada is higher in summer compared to spring and winter due to shorter days and extreme weather conditions especially in winter.^[24,25] However, this needs further investigation to fully understand why the season of primary ACLR is associated with a risk of revision ACLR.

Unadjusted rates for revision ACLR were higher among ACLRs performed by surgeons who performed > 100 surgeries in the past year compared with ACLRs performed by surgeons performing 13 to 100 surgeries (Table 1). Surgeons performing > 100 ACLRs in a year may be working in high volume referral hospitals where complicated cases with multiple ligament injuries are treated. However, this difference by surgeon volume was not significant in multivariable analysis.

4.1. Strengths and limitations

Using population-based data we present the epidemiology of contralateral primary ACLR and revision ACLR in Alberta for the first-time. We have an average follow up period of 5.7 years, one of the longest follow up periods used in similar studies. However, our study has some limitations. Many patients who have failed ACLR may choose not to perform revision ACLR or to have another ACLR in the next knee which might underestimate true graft failure rate or contralateral ACL tear. Findings may not be generalized to other jurisdictions with different health care systems including health financing mechanisms and population structure. Residual confounding due to unmeasured confounders is possible since we relied on administrative data. Additionally, our findings rely on accuracy of billing practices and electronic records for example, coding errors. However, previous studies have established the validity of administrative data in Alberta including physician claims to analyze service utilization patterns. Similarly, revision rates may differ by type of autograft or allograft used, however, we did not have information on subtypes of allografts or autografts used. Based on information from some of the Alberta knee injury clinics and interaction with surgeons hamstring tendon is the most common autografts used in Alberta followed by patellar tendon.^[26] Furthermore, although major acute knee injury clinics in Alberta have a structured rehabilitation program in place, the data on whether our participants took part in such programs was not available in the administrative database. Moreover, type of sports and time to return to sports after surgery is an important risk factor identified to be associated with risk of revision ACLR, however, we could not include these variables due to unavailability of such data.

5. Conclusion

We found a similar rates of ipsilateral revision ACLR (3.9%) and contralateral primary ACLR (3.6%) with a much higher rate in patients younger than 30 years compared to patients older than 30 years. The findings suggest that younger patients might benefit from patient education (about their higher risk of graft failure and contralateral ACL rupture) and improved rehabilitation plans. Future cohort studies are needed to accurately estimate graft rupture and contralateral ACL rupture as well as to disentangle the relationship between surgery setting, and season with risk of subsequent ACLR.

Acknowledgements

This paper is a part of the first author’s doctoral thesis. We would like to thank Alberta Ministry of Health for providing the data for this study.

Author contributions

Conceptualization: Yuba Raj Paudel, Mark Sommerfeldt, Don Voaklander.

Data curation: Yuba Raj Paudel.

Formal analysis: Yuba Raj Paudel.

Investigation: Yuba Raj Paudel, Mark Sommerfeldt, Don Voaklander.

Methodology: Yuba Raj Paudel, Mark Sommerfeldt, Don Voaklander.

Project administration: Don Voaklander.

Resources: Don Voaklander.

Software: Yuba Raj Paudel, Don Voaklander.

Supervision: Mark Sommerfeldt, Don Voaklander.

Validation: Yuba Raj Paudel, Mark Sommerfeldt, Don Voaklander.

Visualization: Yuba Raj Paudel.

Writing – original draft: Yuba Raj Paudel.

Writing – review & editing: Mark Sommerfeldt, Don Voaklander.