FormalPara Key Points

Consensus was achieved for eight statements addressing bicycle set-up dimensions and nine statements on cycling kinematic assessment.

The recommendations will improve transparency, reproducibility, standardisation and interpretation of bicycle measurements and cycling kinematic data.

These guidelines are dedicated to assisting researchers, bicycle fitters and cycling-related practitioners.

1 Introduction

Cycling is a popular physical activity with numerous health benefits [1], but many cyclists are exposed to non-traumatic injuries (52–65%) [2, 3], with a suboptimal position during cycling suggested as one of the possible causes [4,5,6]. Bicycle fitting has been utilised to improve comfort, reduce pain and potentially mitigate the risk of non-traumatic injuries [6, 7]. Bicycle fitting has gained popularity in the cycling community. Still, there is a great variety of sources of information on which the practice of bicycle fitting is based (e.g. websites, books, research papers, etc.), as well as the methods and processes used [8]. There is a lack of evidence-based guidelines on how bicycle set-up dimensions should be measured, which may have resulted in inconsistent and varied bicycle fitting protocols. One example is the measurement of bicycle saddle height, which has a range of different methods reported (e.g. taken from the top edge of the saddle, the nose or from the broadest section of the saddle measured in line with the seat tube to the pedal spindle or the bottom bracket and/or adding the crank length), despite being one of the variables most studied in literature [9, 10].

Although a bicycle could be fitted based on anthropometrical measures when equipment or expertise is unavailable [11], research evidence suggests that bicycle fitting should be informed and guided by kinematic data obtained while cycling [10, 12, 13]. While standard recommendations are available for other forms of kinematic assessment [14], such as walking, running [15] and foot mechanics [16], no recommendations are available for cycling. One example of a limitation in cycling kinematic assessment is the inaccessibility of the anterior–superior iliac spine while cycling, which then affects the options for landmarks from which to model the pelvis [17]. The lack of consensus concerning the collection of cycling kinematic data and bicycle measurements limits the validity of the data and the ability to compare between studies. Guidelines for bicycle measurement and collecting kinematic data are seen to be essential.

There are an increasing number of professional bicycle fitting services based on kinematics, utilising two-dimensional (2D) or three-dimensional (3D) data [13, 18]. In addition to the technical challenges in implementing these measurements [18], exercise characteristics can influence kinematic outcomes [19,20,21]. Previous research has shown that exercise intensity [20], cadence [21] and fatigue [22, 23] can influence pedalling kinematics, but these often vary substantially across studies. Some of the authors in the current paper pointed out these challenges in an editorial that suggested the necessity of creating consensus regarding critical elements for assessing body position during cycling [18]. We, therefore, considered that a consensus statement using the Delphi method can assist researchers and cycling related practitioners in obtaining bicycle set-up dimensions and cycling kinematic data. The Delphi method consists of a structured group process to survey expert opinion and reach a group response [24]. This paper aims to obtain consensus about bicycle set-up dimensions and recommendations for performing cycling kinematic assessments.

2 Methods

2.1 Participants

Four core members (J.I.P.Q., R.R.B., F.P.C. and A.E.-M.) initiated the study, being responsible for recruiting the panellists, preparing the document to review in each round for panel members, analysing the scores and comments of panellists and reporting the decisions, as well as communicating with panellists. Core members did not participate as panel members, and were affiliated with institutions in Spain (J.I.P.Q. and A.E.-M.), Australia (R.R.B.) and Brazil (F.P.C.). None of the core members and panellists had conflicts of interest relevant to participation in the study.

Panellists were selected based on their expertise in cycling kinematics and/or bicycle fitting studies. The inclusion criteria were to have at least five publications about cycling biomechanics in journals indexed in the Web of Science or three publications about cycling kinematics in journals with impact factors in the first or second quartile. A total of 22 experts were invited, and 14 agreed to participate as panel members in the study. Two experts declined the invitation because they were too busy to participate, one declined due to a lack of interest in the topic and five did not respond. They received an information document outlining the study and the methodology with consent indicated by completion of the Delphi survey. Of the 14 panellists (mean ± standard deviation age of 46 ± 9 years old, minimum age 31 years old and maximum age 63 years old), 13 were currently working in academic and/or research institutions, and 6 of these experts (43%) work in a bicycle fitting service. The panellists were affiliated with institutions from the USA (n = 3; 21%), Spain (n = 2; 14%), South Africa (n = 1; 7%), France (n = 2; 14%), Canada (n = 1; 7%), Brazil (n = 1; 7%), Slovenia (n = 1; 7%), United Kingdom (n = 1; 7%), the Netherlands (n = 1; 7%) and Australia (n = 1; 7%). A search on the Scopus database was performed on 29 March 2023 to ensure that panellists had published a median of 6 (range 3–24) full peer-reviewed articles related to cycling kinematics and bicycle fitting, and they had a median H index of 11 (range 3–25).

2.2 Procedures

A Delphi procedure was applied as previously described [24, 25]. For this procedure, panel expert evaluation, judgment, phrasing and scoring were completed independently for each round. Core members wrote an initial document (Supplementary file) with a list of statements to be scored and commented on by the panel members. Each of these statements also included research background information to understand the reason for each statement. Moreover, a five-point Likert scale (strongly agree, moderately agree, neutral, moderately disagree, strongly disagree; from 5 to 1, respectively) was used to rate each statement, and an open box included comments from the panellists. They could also propose a new statement for the next round. The panellists were encouraged to support their opinions with evidence whenever possible. The consensus was reached when more than 80% of the panel members scored the statement with values of ‘4’ (moderately agree) or ‘5’ (strongly agree), with an interquartile range less than or equal to 1 [25]. However, considerable modifications were made to improve transparency whenever the consensus was reached based on panellist comments, and a re-evaluation was performed.

Three rounds were performed. In every round, all panellists had the opportunity to comment on the statements and suggest possible rephrasing. The panellists had 15 days to respond to each round, and all communications were conducted by electronic mail. For the second and third rounds, the document included each statement: the statement proposed in the previous round, the quantitative analysis of the first round (percentage of answers for each point of the Likert scale and interquartile range), the qualitative analysis (a summary of all comments), the core members' conclusion, whether the statement was accepted or not and the new statement proposed. The decision made in each round for each statement comprised seven distinct actions [26]: (1) modify (the statement was substantially modified), (2) rephrase (to improve understanding without changing the meaning), (3) divide (one statement being split in two or more statements) (4) join (the statement resulting from merging two or more previous statements), (5) exclude (the statement was excluded from the consensus), (6) include (a new statement was included as suggested by a panel member) and (7) approval (when consensus was reached on a statement).

2.3 Equity, Diversity and Inclusion Statement

The core members attempted to recruit panellists of both sexes and from around the world, including researchers in developing countries. However, of the 14 panellists, there were only 2 women, and 12 panellists were Caucasian. The established inclusion criteria limited the ability to select more women and people from other communities, which may have limited the generalisability of our results.

3 Results

Figure 1 shows the flowchart of the Delphi process with the results for the three rounds of evaluation.

Fig. 1
figure 1

Flowchart of the process

Full size image

Table 1 shows the results and decisions of each round. Decisions in the first round included modification (11 statements; 65% of the statements), joining (4 statements; 24%), rephrasing (1 statement; 6%) and approval (1 statement; 6%). Although most of the statements met the approval criteria (12 statements; 71%), only 1 was approved in the first round (item 9), and the others were revised based on panellists’ comments and were re-evaluated in the second round. Moreover, considering comments from panellists, three new statements were created for the second round: kinematic knee flexion at the bottom of the revolution (item 19), angle definitions (item 20) and foot position on the pedal (item 21). Statements showing a higher level of disagreement between panel members were those related to the saddle setback (item 4), the diagonal distance between the handlebar and saddle (item 8) and the offset by the individual’s standing posture for bicycle fitting (item 13).

Table 1 Results of the panellist’s evaluation ( of scores of five-point Likert scale for each item and interquartile) through the three rounds with core panel decisions for each statement
Full size table

In the second round, most of the statements improved their percentages of responses with ‘strongly agree’ and ‘moderately agree’ levels and met the approval criteria (13 items; 81%). However, this round did not approve five of these (31%) statements. Thus, the decisions in the second round included modification (eight statements; 50%) and approval (eight statements; 50%). They were edited considering panellists’ suggestions and re-evaluated in a third round. A new statement was created for the third round consisting of sample characterisation (item 22). Statements that had the highest level of disagreement among panel members were two of the statements with high levels of disagreement in the first round (item 8—the diagonal distance between the handlebar and saddle—and item 12—the offset by the individual’s standing posture for bicycle fitting) and two of the newly included statements related to angle definitions (item 20) and foot position on the pedal (item 21).

Decisions in the third round included approval (eight statements; 89%) and exclusion one statement; 11%). Except for the statement about the angle definitions (item 20), which was excluded due to a high level of disagreement in the second round, the other eight statements improved their scores after the third round.

The initial proposal for item 20 suggested adopting the International Society of Biomechanics (ISB) conventions for defining angles in 3D analysis. While recognising the need for further investigation into angles related to the spine and trunk due to their complex degrees of freedom, a proposal was also made for measuring angular kinematics in 2D. However, due to inconsistencies in using different joint angles (e.g. ankle, shoulder and elbow) versus supplementary angles (e.g. knee and hip), practical issues with using the posterior superior iliac spine (PSIS) and anterior superior iliac spine (ASIS) to analyse hip angles and discrepancies observed between dynamic measurements and static measurements obtained with different instruments, the item was ultimately rejected.

Therefore, the final consensus was achieved for eight statements for bicycle measurements (Table 2) and nine statements for kinematic methodology (Table 3).

Table 2 Statements that obtained a consensus for bicycle set-up dimensions after the Delphi procedure
Full size table
Fig. 2
figure 2

Example of saddle mechanisms adjustment. A Continuous adjustments system (adapted from https://wattbike.com/). B Discrete adjustment system (original image, Bodytone Monster model; Bodytone International Sport SL, Murcia, Spain)

Full size image
Fig. 3
figure 3

Proposed saddle height measurement. The red dashed line represents the sum of the yellow (from the bottom bracket to the centre of the saddle) and blue lines (crank length)

Full size image
Fig. 4
figure 4

Proposed saddle setback measurement. Red dashed line represents the sum of the yellow (horizontal distance from bottom bracket to the centre of the saddle) and blue lines (crank length)

Full size image
Fig. 5
figure 5

Effective seat tube angle

Full size image
Fig. 6
figure 6

Proposed vertical difference between handlebar and saddle height measurement

Full size image
Fig. 7
figure 7

Proposed vertical measurement for integrated handlebars

Full size image
Fig. 8
figure 8

Diagonal distance between handlebar and saddle

Full size image
Table 3 Statements that obtained a consensus for kinematic analysis after the Delphi procedure
Full size table
Fig. 9
figure 9

Foot pedal position (red arrow)

Full size image

4 Discussion

This research aimed to obtain a consensus about best practices for measuring and reporting bicycle set-up dimensions and collecting cycling kinematic data. Four core members and fourteen experts agreed on eight statements regarding bicycle measurements and nine statements regarding cycling kinematic assessment. The statements are presented to assist a range of users with differing access to equipment and technology. Best practice guidelines in bicycle measurement and kinematic data analysis can help improve research in the topic and the practical assessment of bicycle fitting. We are confident that further investigation will improve the current proposal and enhance the overall quality of cycling science. We intentionally omitted variables such as apparel, performance, environmental conditions and evaluator experience to maintain methodological clarity and consistency in the present study. We acknowledge the importance of these factors and suggest that future research explores their interactions and impacts in more detail. This approach aims to provide a solid foundation for standardising methods, which can be built upon in subsequent studies to address specific performance-related questions.

The consensus regarding bicycle set-up dimensions and cycling kinematic methodology is closely aligned with existing research, but there are ongoing gaps in the literature. Therefore, it is recommended that the present consensus be considered a guide, not hermetic, with scientific evidence for all its statements. Some statements may have a considerable amount of scientific research behind them, e.g., statement 2.3 regarding the differences between 2 and 3D kinematic measurements [27, 28] or statement 2.6 about the interrelationship between body segments during cycling [29, 30]. A recent study supporting statement 2.6 showed that increased pelvic tilt results in higher hip flexion and a more dorsiflexed ankle angle [30]. However, there are statements for which more investigation is needed (e.g. statement 2.6 ‘Normalisation by the individual’s standing posture for bicycle fitting’). Therefore, this consensus will need ongoing review and modification based on future evidence.

The concluding statements (Tables 2 and 3) are not intended to end the discussion and analysis around bicycle set-up dimensions and cycling kinematics but to act as a starting point for measurement, reporting and communicating. For example, in statement 1.3, aimed at determining saddle height, it is important to consider that while the point of contact on the saddle for the ischial tuberosities could be an option as a reference point, the pelvis can have variability in its antero-posterior position on the saddle. Another option could be assuming the centre of the saddle as a reference. However, this option can also be affected by the geometry of the saddle [31]. Therefore, it is possible that even after standardising the protocol for measuring saddle position, there are still questions about which method can provide the most reliable result. To mitigate these limitations, it was decided to provide multiple options and include clear and justified explanations for replication.

Two statements regarding kinematic data measurement caused great controversy among panellists. These involved the statement around ‘angle definitions’ (for which no consensus was achieved) and statement 2.4 ‘Normalisation by the individual’s standing posture for bicycle fitting’. Concerning ‘angle definitions,’ while the International Society of Biomechanics (ISB) makes recommendations for 3D analysis [32], there is great diversity in the definition of 2D angles for the spine and trunk [7, 33, 34]. These findings and the failure of our expert panel to reach a consensus further demonstrate the complexity of these measures, which seem to result from diverse approaches regarding the number of segments and reference points and suggest the need for future methodological studies. Therefore, carefully describing the methodology for determining 2D angles is recommended. Considering statement 2.4, a previous study using 2D video-based analyses observed a mean bias of ~ 11° in knee flexion comparing absolute angles and angles when normalisation by the individual’s standing posture was performed [35]. However, literature on this topic is scarce, and most studies only provide absolute angles. For this reason, the statement suggests that future research is needed to provide references for both absolute and normalised angles.

This Delphi process yielded expert consensus on performing bicycle measurements and kinematic analysis for cycling research and testing. This consensus is an important step towards the standardisation of measures to optimise the bicycle fitting process for recreation, rehabilitation and competition.

The review highlighted key areas where additional research is needed and suggested important next steps:

  1. (1)

    Generally accepted recommendations for reporting the whole body’s cycling kinematics (e.g. joint coordinate systems) in 2D and 3D are necessary before the field can move forward.

  2. (2)

    Research is required to determine if and how normalising kinematic variables should be performed; factors to consider are 2D and 3D data, joints and rider type (competitive, recreational or rehabilitation), as the developed statements are not specific to any rider type. Furthermore, it is important to understand that the statements may require adjustment to cater to populations with disabilities or injuries [36, 37], and future consensus could focus on such populations.

  3. (3)

    Future research and consensus statement development must consider different measurement conditions, e.g. 2D versus 3D, static versus dynamic and optical (camera) versus inertial measurement unit (IMU), to provide recommendations on body position during cycling.

  4. (4)

    Future consensus statements should consider measuring and reporting additional aspects of cycling biomechanics, e.g. how forces and moments are collected, how these data are processed, how joint kinetics are reported (joint reaction, contact forces, or joint moments and power) and how muscle activation (electromyography, simulation-based approaches) are computed and reported.

Despite the limitations, we argue that this consensus will benefit the field of study and improve the quality of bicycle fitting and cycling research. Moreover, the statements can help reviewers during the peer review process of cycling studies and guide authors to adhere to best practices in their reports. In this sense, two appendices (Appendix 1 and Appendix 2) are provided to ensure that all essential data were considered and included in the report. Core and panel members suggest using the present consensus to guide research studies.

5 Conclusions

We present a consensus of eight statements for measuring and reporting bicycle set-up dimensions and nine statements for cycling kinematics data collection for a wide array of researchers and cycling related practitioners. We encourage scientists and professionals to apply these statements in the field, aiming to improve the reproducibility, standardisation and interpretation of bicycle fitting assessments and cyclist testing protocols.