Statistical analysis plan (SAP) for the 5- and 10-year follow-up assessments of the FIDELITY trial

The research objectives of the 5- and 10-year assessments of the Finnish Degenerative Meniscal Lesion Study (FIDELITY) are two-fold: 1) to assess the long-term efficacy of arthroscopic partial meniscectomy (APM) in adult (age 35 to 65 years) patients with a degenerative meniscus tear, and 2) to determine the respective effects of APM and degenerative meniscus tear on the development of radiographic and clinical knee osteoarthritis (OA). Methods and Design FDELITY is an ongoing multi-centre, randomized, participant and outcome assessor blinded, placebo-surgery controlled trial of 146 patients. This statistical analysis plan (SAP) article describes the overall principles for analysis of long-term outcomes (5- and 10-year follow-ups), including how participants will be included in each analysis, the primary and secondary outcomes and their respective analyses, adjustments for covariates, and the presentation of the results. In addition, we will present the planned sensitivity and subgroup analyses.


Introduction
Trial overview and purpose of the statistical analysis plan (SAP) Finnish Degenerative Meniscal Lesion Study (FIDELITY) is a trial to assess the efficacy of arthroscopic partial meniscectomy (APM) for patients with a degenerative meniscus tear.
The primary outcome assessment point of the trial was at 1-year post surgery. The original study protocol [1] and the results of 1-and 2-year analyses [2,3], as well as a secondary analysis focusing on mechanical symptoms [4], are published.
To safeguard against the imminent risk of outcome reporting bias, selective reporting, and data-driven interpretation of results, this statistical analysis plan (SAP, Version 1.0) of the 5-and 10-year follow-ups is published as an update to the previously published protocol 3 [1]. The original study protocol [1] provides more details on the trial rational, eligibility criteria, interventions, data management and methods for limiting bias. This SAP follows the guidelines for writing SAPs provided by Gamble et al [5] and describes the overall principles for analysis of long-term outcomes (5-and 10-year follow-ups), including how participants will be included in each analysis, the primary and secondary outcomes and their respective analyses, adjustments for covariates, and the presentation of the results.
In addition, we will present the planned sensitivity and subgroup analyses. The trial results will be reported according to the Consolidated standards of reporting trials (CONSORT) guidelines for RCTs [32].

Background
By the end of the 21th century, arthroscopic partial meniscectomy (APM) had become the most common orthopaedic procedure with well over half a million such surgeries performed annually in the US alone [4,5], most to middle-aged and older patients [5].
According to conventional wisdom, APM was thought to result in short-term improvement in knee function and quality of life. However, a series of rigorous trials, summarized in three recent systematic reviews and meta-analyses, provide compelling evidence that APM offers little short-to medium-term benefit for most patients with knee pain and degenerative meniscus tear above sham surgery or non-surgical management [6,7].
Recent evidence thus convincingly contradicts the widely-held contentions on the asserted benefits of APM on knee symptoms or function, but uncertainty still exists on the possible undesirable consequences of the procedure [8]. Overall, the risk of adverse events within 90 days of the procedure appears low, but serious adverse events (including pulmonary embolism and infection) have been associated with the surgery [9,10].
However, there is mounting evidence to suggest that APM is associated with increased risk of accelerated progression of knee OA and earlier need for "corrective" surgery (high tibial osteotomy, HTO or total knee replacement, TKR) in this patient category of middleaged to older patients [11,12]. It still remains unclear whether the increased risk is due to the meniscus tear per se, the surgical procedure (APM), or if there is an interaction between the two. This question cannot be addressed simply by evaluating the outcome of patients who have undergone APM, because the role of the underlying degenerative process and the surgical procedure cannot be disentangled in such design [13]. Given the existing uncertainty regarding the potential effect of APM on the development or 4 progression of knee osteoarthritis (OA), we are planning to address this particular issue by carrying out an adjunct, pre-registered analyses of the FIDELITY trial at 5-and 10-years from randomization. The biological rationale behind these studies is that the resection of torn meniscus tear (APM) has an effect on the progression of degenerative knee disease: some argue that APM cures symptoms and slows down the development of OA while others assert the contrary.

Objectives
The following two research questions capture the primary objectives of these 5-and 10-  The study process shown in Figure 1 provides a brief outline of the trial. The eligibility criteria for the study are presented in Table 1. To assess the efficacy of APM (vs. placebo surgery) on the functional outcome and knee symptoms in patients with an arthroscopically-verified degenerative tear of medial meniscus, we will be using the same three patient-relevant outcomes (PROMs) that were used as our primary outcomes in the previous, 1-year and 2-year follow-up publications from this data [2,3].
Western Ontario Meniscal Evaluation Tool (WOMET) score , a disease specific quality of life -instrument developed and validated for patients with meniscal pathology [14,15] 2.
Lysholm knee score, the most commonly used outcome instrument for various knee conditions [16,17] and a tool that has also been validated for patients with meniscal injury [18] 3. Knee pain after exercise, assessed on an 11-point scale ranging from 0 (no pain) to 10 (extreme pain).
Secondary outcome: The frequency of unblindings in the two study groups: Patients with inadequate relief of symptoms underwent unblinding of the treatment-group allocation These outcomes and their justifications have been elaborated in detail previously [1][2][3][4].
Objective #2: Development of knee OA (NCT01052233) To assess whether APM either accelerates or delays the development/progression of radiographic and clinical knee OA in these patients, we will use radiographs and The KL scale is a semi-quantitative instrument (ordered categorical grades 0-4) to assess the severity of radiographic tibiofemoral knee OA [19]. Patients who have undergone an osteotomy or a total knee replacement during follow-up will be considered to have progressed radiographically according to the definition above. Patients' return to normal activities 10.
The presence of mechanical symptoms [4] 11. Clinical knee examination 12. Serious adverse events

13.
Frequency of re-APMs, and the number of osteotomies and knee arthroplasties 14.
Possible derivates from the above noted outcomes For all radiographic outcomes, one experienced musculoskeletal radiologist (JK), unaware of the treatment allocation and clinical data, will grade the baseline, and the 5-and 10- year radiographs of the operated (index) knee of all participants.
All the analyses of secondary outcomes are supportive, exploratory and/or hypothesisgenerating.

Rationale for outcomes to be reported and for the statistical analyses
For the assessment of the efficacy of APM (NCT00549172), we will use the same PROMs used in the previous publications depicting the 1-year [2] and 2-year [3] follow-up findings. To safeguard against potential multiplicity effects [22] in this analysis, we will interpret the treatment effect estimates and their 95% CIs for all our three primary outcomes.
As for the evaluation of the development/progression of knee OA (NCT01052233), the 5- year follow-up is the first time point one can reasonably expect any OA-related changes to take place/to be quantifiable. Having said that, the outcome measures we originally registered in the ClinicalTrials.gov database (i.e., Kellgren-Lawrence grade and OA by ACR clinical criteria) are quite insensitive to change, we have decided to add the sum of OARSI atlas osteophyte grades and JSN grades as an additional primary outcome of radiographic progression of OA.

Statistical analysis
All the analyses will be performed according to the intention to treat (ITT) principle or, if impossible, using full analysis set [23]. In sensitivity analysis also per protocol (PP) analyses will be performed. For all outcomes, 95% confidence intervals for the relevant between group differences will be presented.
For the analysis of continuous outcomes related to the objective #1 (efficacy of APM), a mixed model linear regression will be used. In this model the patient will be included as random effect and time point (6,12,24, 60 and 120 months), treatment arm (APM or placebo) and their interaction, and randomization stratification factors, i.e., the age [35 to 50 years or 51 to 65 years], sex, absence or presence of minor degenerative changes on a radiograph [Kellgren-Lawrence grade 0 or 1], and study centre, will be included as fixed effects. The model will be adjusted for baseline values of the respective outcome variable.
For the analysis of the binary outcomes a logistic regression model will be used. The model will be adjusted for the baseline randomization stratification factors (age [35 to 50 years or 51 to 65 years], sex, and absence or presence of minor degenerative changes on a radiograph [Kellgren-Lawrence grade 0 or 1]). To obtain the adjusted risk ratio and the adjusted risk difference from the logistic model, the method of standardization will be u s e d [24]. Although the randomization was also stratified by the study site, the site will not be adjusted for in the logistic regression analysis due to the low number of participants in some centres and anticipated sparse data. A sensitivity analysis including the study site as a covariate will be performed.
For the analysis of continuous outcomes related to the objective #2 we will use a linear regression model adjusted for randomization stratification variables and the baseline value of the outcome.
Also, serious adverse events will be reported.

Study power considerations
We note that we originally powered the study to detect a minimal clinically important difference in the efficacy (patient-reported) outcomes -the Lysholm and WOMET scores (differences of at least 11.5 and 15.5 points, respectively) and in the score for knee pain after exercise (difference of at least 2.0 points) -between the APM and placebo-surgery groups. The trial was not primarily powered to detect differences in the progression of 11 knee OA and only a large difference between the groups has a chance of reaching the statistical significance level of 5% in these outcomes. Thus, the interpretation of the latter findings will be based on the range of values included in the 95% confidence intervals.

Blinded data interpretation
We will interpret the results of this trial according to a blinded data interpretation scheme [25]. However, as the trial statistician (AT) has already performed analyses on 2-year data of this trial [3], she will perform all statistical analyses using unblinded treatment groups.
She will then provide the Writing committee of the trial with blinded results from the analyses with the groups labelled group A and group B. The Writing Committee then contemplates on the interpretation of the results until a consensus is reached and agrees in writing on all alternative interpretations of the findings. Once a consensus is reached, we will record the minutes of this meeting in a document coined statement of interpretation, which will be signed by all members of the Writing Committee. Only after reaching this common agreement, the data manager and the trial statistician will break the randomization code and the correct interpretation is chosen. Draft of the manuscript will then be finalized. Detailed minutes of blinded data interpretation meetings will be provided as a supplement to the manuscript.

Ethical considerations
The study was approved by the Pirkanmaa Hospital District's committee of ethics (n:o R06157). Our application contained a specific, six-point ethical analysis focusing on the methodological rationale for use of placebo surgery, risk-benefit assessment and informed consent (for detail, see [1]).

Dissemination
The findings of this study, whether positive, negative or neutral, will be disseminated widely through peer-reviewed publications and conference presentations. 12 Trial Status The enrollment for the study was carried out between December 2007 and January 2012, and subsequently, the follow-up examinations took place between December 2013 and January 2017. We have now completed the 5-year follow-up examinations and data management and are ready to carry out blinded data interpretation of the 5-year data.
The 10-year follow-up examinations are ongoing.

Declarations
Ethics approval and consent to participate Central ethical approval was confirmed from the Pirkanmaa Hospital District's committee of ethics (ref approval no. R06157) and we did not begin recruiting at other centres in the trial until local ethical approval was obtained. Informed consent was obtained from all study participants at trial entry.

Consent for publication
Not applicable.

Availability of data and material
Given that the informed consent forms of the FIDELITY trial did not include a provision for data sharing (trial launched in 2007), the full dataset cannot be shared due to a potential breach of the Finnish Personal Data Act. Scientists with a specific question regarding the trial data are encouraged to contact the corresponding author (TLNJ).

Competing interests
All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.