Pre-Run Stretch Study Protocol

Protocol for a Large, Randomized, Prospective Study of the Impact of a Pre-Run Stretch on the Risk of Injury in Teenage and Older Runners

April 9, 2007
Sponsor: USA Track & Field
Principal Investigator: Daniel Pereles, MD
Project Coordinator: Alan Roth, Ph.D.

Table of Contents

1. Introduction
2. Background and Rationale
3. Objectives
4. Study Design
5. Study Population
   • 5.1 Inclusion Criteria
   • 5.2 Exclusion Criteria
   • 5.3 Recruitment of Volunteers
6. Enrollment and Randomization
7. Study Procedures and Evaluations
   • 7.1 Data Collection
   • 7.2 Stretching Regimen
   • 7.3 Evaluation of Injuries
8. Statistical Considerations
   • 8.1 Outcomes
   • 8.2 Primary and Interim Analyses
   • Table 1: Presentation of results for determining increased/decreased risk
   • 8.3 Additional Supplemental Analyses
   • 8.4 Missing Data Considerations
9. Ethics and the Protection of Human Subjects
   • 9.1 Informed Consent
   • 9.2 Potential Risks to Volunteers
   • 9.3 Confidential Information
10. References

1. Introduction

Over-use injuries are a common problem among runners. An individual runner may have biomechanical deficiencies that make running very difficult to accomplish without injury. Some runners may run for many years and not experience injuries. But for many runners, injuries are a constant threat, and it will be of great benefit to identify means to help avoid injury.

There have been many studies of stretching prior to physical exercise. To date, a number of studies have found that stretching and warm-up activities improve flexibility. However, none of these studies has conclusively determined that a pre-run stretch will prevent injury. Some studies have shown that stretching before exercise (but not looking at pre-run stretching) prevents injury, others show no benefit.

A prospective, randomized study is needed to appropriately address this important question and provide a basis for recommendations to athletes concerned about minimizing the risk of injury.

2. Background and Rationale

Stretching either before or after running has been an accepted training analog for many years. Laboratory studies of muscle cell contraction have shown improved function and more efficient metabolite transfer after mild sarcomere lengthening. Clinically, it was always assumed cold tight muscles were more easily injured, and thus stretching those muscles would impart protective qualities. In fact, uncontrolled studies claimed significant benefits from pre-participation stretching in football, soccer, and field hockey players.

Recently, a study claiming to show no benefit in the pre-run stretch was published, casting doubt on this commonly accepted principle. Shrier, in the Clinical Journal of Sports Medicine, reviewed 23 articles relating stretching and performance. Stretching was defined as either acute, which meant that the stretch immediately preceded the activity, or regular, in which the stretch occurred at some other time unrelated to the activity. Articles specific to running and stretching were critically reviewed. The conclusions about acute stretching and running speed were equivocal, with one study showing benefit and one showing slight detriment. The majority of articles examining regular stretching suggested it was beneficial, improving both force production and velocity of contraction. Regular stretching was also shown not to decrease injury rate.

Another study was published in 2004, Thacker et al, The Impact of Stretching on Sports Injury Risk: A Systematic Review of the Literature, in Medicine & Science in Sports & Exercise, in which the study team found six of 361 identified articles compared stretching with other methods to prevent sports injuries. From these six articles, they reported "Stretching was not significantly associated with a reduction in total injuries (OR = 0.93, CI 0.78-1.11) and similar findings were seen in the subgroup analyses." However, their conclusion was, "There is not sufficient evidence to endorse or discontinue routine stretching before or after exercise to prevent injury among competitive or recreational athletes. Further research, especially well-conducted randomized controlled trials, is urgently needed to determine the proper role of stretching in sports." The Pereles/USATF study is intended to meet this need with specific attention to pre-run stretching.

Studies involving acute stretching and running were mixed in their conclusions. The studies themselves were done on small numbers and not specific to runners. In this Pereles/USATF study, the hypothesis is that a 3-5 minute pre-run stretch reduces running injuries. In an effort to better quantify the benefit or lack thereof of pre-run (acute) stretching, a study model was devised. Runners will be chosen in a randomized pattern to participate in or refrain from a pre-run stretching protocol of 4 minutes. The protocol will use static stretching which is the type of stretching most widely practiced by runners. The active study group will stretch the quadriceps, hamstrings, and calf muscles based on photographs showing the correct manner of stretching and the amount of time for each stretch. The control group will move directly into training without stretching. Injury patterns and related medical histories will be recorded for each participant. The data collection will take place for 3 months per participant. The forms and enrollment will be mostly internet-based with participants having the option to use mail or facsimile communications.

The risk is minimal as the protocol for stretching uses a mild form of stretching that does not acutely stress soft tissue. This small risk is worthwhile as the study is intended make a major contribution to our understanding of the value of stretching in regard to injury and the tens of millions of runners who run regularly may have more guidance on stretching so that they may incur fewer injuries.

Consideration was given to the role of warm-up exercises prior to stretching. Participants will be informed that they can warm up as they normally would and data will be collected on if they warm up, whether they warm up before stretching, after stretching or, for non-stretchers, if they warm up before running.

Consideration will also be given to the flexibility of participants at the start of their participation. They will be asked to try to touch their toes while standing with their legs straight without forcing the stretch. How far they get will be recorded.

The study will be based on participants reporting injuries over the 3-month period. We encourage participants to obtain injury diagnoses from healthcare practitioners, but we expect many participants to self-diagnose their injuries and to do so fairly accurately. The runners will know which muscle groups they are stretching and will know when those specific muscles are painful. Runners are generally well-educated and know about muscle and tendon damage to the extent that they can describe an injury. If participants are not sure, we ask that they look for healthcare practitioner diagnosis. The study identifies the source of the diagnosis to aid data analysis.

The value of the study and the ability to bring in the analysis of many of the data points will depend on having an adequate number of participants. A study team will review the data on a regular basis to see how well analyses can be conducted with the data collected to that point. Runners will be invited to participate in the study on an ongoing basis using media outlets. Road Runners Club of America is the study's Outreach Partner to facilitate access to runners across the country. The study will be concluded when conclusions can be drawn and is expected to occur in late -2007 or early-2008.

A study in which people are randomly assigned to either a pre-run stretch, or no pre-run stretching can begin to understand the relationship between this type of stretching routine and injury. The study must include enough subjects to adjust for all of the aspects of an individual's physical capabilities and history of injury that may affect the rate of injury. Also, the effect of a pre-run stretch on injury may be different with different sub-populations (like ultra-marathoners or people with prior injuries) so this large study will also examine injury rates in these groups of people as well.

3. Objectives

The primary objective of this study is to determine whether a pre-run stretch reduces the risk of injury in runners over the age of 13 who run at least 10 miles per week. The secondary objective is to investigate whether there are additional risk factors contributing to the risk of injury where data can be obtained. Some known risk factors such as overtraining are too subjective to provide reliable, quantitative data.

4. Study Design

This is a prospective, randomized, 2-arm, unmasked comparative study of a pre-run stretch versus no stretch in up to 10,000 volunteers.

Subjects will be asked to adhere to their standard running regimen but to perform the test assignment instead of their usual pre-run routine. A post-run stretch is permitted. Applicants must provide a valid email address to participate, but no other identifying information will be collected. For those applying through regular mail, identifying information will be used only to conduct mailings and will be treated as confidential. Throughout the study, online subjects will receive e-mails reminding them of their test assignment, the study web address, the projected date of completion of their participation, and a reminder to report any injuries.

For the purposes of this study, an injury will be defined as any impairment that prevents subjects from following their usual running regimen for at least 3 days. If subjects participate in other activities, such as tennis, basketball, hockey, skiing/snowboarding, etc., which have the potential to cause injury, injuries occurring due to these specific activities are not to be specifically reported. They will be asked to report if an injury occurred during activity other than running, or became apparent following running or that other activity. Subjects should follow whatever their typical stretching routine is prior to participating in those activities. The test assignment (stretch or no-stretch) should be performed only prior to the subjects' typical running routine.

The maximum sample size will be 10,000, but the study may stop sooner if either test arm exhibits a statistically significant risk to subjects.

Due to the nature of the stretching regimen proposed, masking subjects is not feasible. Bias in injury reporting will be unavoidable. Subject self-assessed compliance scores will be collected to adjust for the actual amount of stretching that may have been performed.

A Frequently-Asked-Questions (FAQ) webpage will be available on the study website to answer questions that are foreseen. Subjects and and prospective subjects who have questions will first be directed to that webpage. If that does not provide the needed information, the person will then be directed to a contact web page which will have an information request form to be sent to project management. There will be an administrative and a technical option on where to send the inquiry. If the person who wishes to inquire prefers to do so by telephone, there will a call center set up to handle inquiries. For the technical questions, Dr. Pereles will handle those calls. An early objective of Dr. Pereles is to identify questions that need to be added to the FAQ page. Administrative queries will go to Dr. Alan Roth, the USATF project coordinator, who also will be looking for common questions that need to be added to the FAQ. Since the study is expected to attract subjects slowly over time in the beginning months, call volume will be reasonable and steps can be taken to handle major concerns early. When the study is more heavily promoted, telephone messages can deal with the most common questions. USATF can recruit organization volunteers to help respond to callers if demand is high.

5. Study Population

5.1 Inclusion Criteria

Applicants will be provided a test assignment and invited to participate if they satisfy all of the following criteria:

1. 13 years of age or older
2. Willing to commit to either a pre-run stretch or no stretch prior to their current running routine
3. Provide an informed consent to participate in the study
4. Current running routine is at least 10 miles per week

5.2 Exclusion Criteria

Applicants will be excluded from the study if they satisfy any of the following criteria:

1. Have suffered any injury impeding their ability to run at any point within 6 weeks prior to randomization.
2. Plan to stop running at least ten miles per week at any time between randomization and 3 months post-randomization.

5.3 Recruitment of Runners

Interested parties will be informed that they may not receive any direct benefit from participating, but will enable investigators to learn more about the relationship between stretching and injury, which may benefit all future runners. Subjects will not be compensated for their participation.

Runners will be recruited through a variety of means. The Road Runners Club of America has about 180,000 members and it will be an official "Outreach Partner" to recruit volunteers through their newsletter and their website. A number of running publications have agreed to include information about the study in upcoming issues. USATF has a media office and transmits press releases almost daily to a large number of media companies, giving us a capability to frequently issue press releases for runner recruitment. USATF also has direct, frequent communication with its membership of approximately 100,000 people, most of whom are athletes. Press releases will include information about risk and that applicants will be randomly assigned to be stretchers or non-stretchers. The press releases will provide background information about the value of learning about stretching and injuries and will include information about past studies on this subject. Readers will be directed to the USATF website where they will find considerable background information in Study Protocol so that they can make an informed decision about participating in the study.

We expect some newspapers and magazines to include articles about the study. We will invite them to visit the USATF website to obtain background information, the Study Protocol and other information for their use in the articles.

6. Enrollment and Randomization

Study applicants may enroll in the study through a page on the USATF website, through the U.S. Postal Service, or by fax. Those who enroll online must complete an electronic form indicating that they satisfy all the study eligibility criteria. If completed successfully, a test assignment (pre-run stretch or no stretch) will be provided via email. Those enrolling through postal mail and fax will be sent a form to fill out and send back to USATF. They will be sent their assignment by mail or fax and they will be instructed to respond by mail or fax.

Test assignments will be made randomly using a random block size (from 10 to 50) to ensure partial balance between the test arms. Only one test assignment will be provided to each volunteer. If the volunteer chooses not to participate, that test assignment will not be reused.

7. Study Procedures and Evaluations

7.1 Data Collection

To apply for the study individuals will submit the Stretch Study Registration Form. This form will collect relevant background information on the subject, including current and past running and stretching routines as well as information about diseases and/or injuries that may impact their risk of injury.

Subjects will be asked to provide information on any injuries that occur during the study period, and complete an end of study form when their 3-month study period is complete. Subjects will be asked to submit their injury reports 3 weeks after the injury occurred.

Limited personal identifying information (name, and one of the following: e-mail address, street address, fax number) will be collected during the study. The identifying information that is collected will be kept confidential and in a secure location apart from the other study data. The only connection between the study data and the volunteer's identifying information will be a code number so that data can be processed anonymously (without identification other than a code number). (see 9.3 Confidential Information for more details).

The EMMES Corporation is providing pro-bono service to the study for statistical analysis. Quoting from the company's website: "The EMMES Corporation is led by professional statisticians and biomedical scientists whose highest priority is the scientific integrity of our research efforts...We are committed to conducting clinical research with a firm scientific basis that is fully compliant with national and international regulatory guidelines."

7.2 Stretching Regimen

The subjects who will stretch will be supplied with a very detailed stretch protocol that includes photographs of the correct stretching routine and have access to the images and descriptions online. The regimen consists of 3 stretches, described below.

Stretch 1 - Quadriceps: From a standing position, grab foot/ankle of same sided leg and pull foot gently up toward butt. Steady yourself with the opposite hand. You may lean back to increase the stretch if so desired. Without bouncing, hold steady for 10 seconds. Repeat 3 times for each leg.

Stretch 2 - Hamstring: From a standing position, place the leg to stretch on an object to elevate the leg. Lean forward over the thigh and reach for the foot. Without bouncing, hold for 10 seconds. Repeat 3 times for each leg.

Stretch 3 - Calf/Achilles: From a standing position, lean forward with the leg to be stretched straight out behind you, foot flat on the ground. Keep the knee straight for a full 10 seconds. Then flex the knee and repeat the stretch with the knee bent for 10 seconds. No bouncing. Repeat 3 times for each leg.

Applicants randomized to the no-stretch group will be instructed to perform no stretching activities prior to their regular running routine.

Subjects will be asked to adhere to their test assignments and follow their normal running routine for 3 months.

7.3 Evaluation of Injuries

At the conclusion of the 3 months, the subjects will be sent a reminder to fill out a final form indicating that they have either reported all of their injuries or were not injured during their participation in the study. An injury is defined as any impairment which prohibits subjects from participating in their normal running routine for at least 3 days. Injuries do not need to be diagnosed by a medical professional, but whether or not they are, subjects will be asked to report the diagnosis and what type of professional made the diagnosis, if a professional was consulted. Subjects will be asked to submit their injury reports 3 weeks after the injury occurred. That will allow information to be provided about the duration of the injury and the recovery.

8. Statistical Considerations

This study will utilize a fully Bayesian primary interim monitoring plan as described in Fayer et al 1997 and Vail et al 2001. The primary analysis will not consist of a frequentist hypothesis test, but will provide posterior probabilities that there is a sufficiently large increased or decreased risk of injury.

8.1 Outcomes

The primary outcome for analysis is whether each subject has experienced a study defined injury or not. An important, confirmatory, secondary outcome will be the occurrence of injuries by a physician or other medical professional.

8.2 Primary and Interim Analyses

The primary analysis will be a Bayesian estimation of the posterior probability of a certain increase or decrease in risk. The final study results will be presented in a table similar to Table 1.

Table 1: Presentation of results for determining increased/decreased risk

Where p_s and p_ns are the true probabilities of injury in the stretch and no-stretch groups respectively, and d is a real-valued and possibly negative measure of the incremental increase in the probability of injury:

p_s=p_ns+d

The primary (interim) analysis will be unadjusted by covariates and will detect only overall aggregate differences between the two test groups. Since the expected sample size will be large, the standard Z-statistic for difference between two proportions will be used. The likelihood of interest therefore is Gaussian, as will be the 4 priors and therefore the resulting 4 posteriors. Figure 1, below, displays the approximate locations of proposed prior distributions. With respect to clinical impressions, the skeptical prior represents the strict belief that there will be no difference between the stretching and no-stretching group while an enthusiastic prior strongly believes that one of the test groups will be superior and not equivalent. The semi-non-informative prior lends some weight to no difference between the groups, but does allow for substantial probability around the region of no difference. The enthusiastic priors will be 3 standard deviations above and below zero.

With respect to Table 1, there will be indication of risk or benefit if the posterior probability of d is at least as large (positively or negatively) as 95% for all 4 priors. With consideration of perceived risks in other subgroups (Section 8.4) there will be some indication that the open portion of the study should be stopped if there is an indication of risk or protection (ie: posterior probability >95% for all priors) of at least 5% (ie: all 4 posterior probabilities are at least 95% that d is less than -5% or greater than 5%. The study may continue with enrollment into specific subgroups as indicated by the supplemental analyses.

Posterior probabilities will be calculated after every 100 subjects completes 3 months of follow-up. Unlike a frequentist setting, no maximum sample size is required (other than administrative requirements) and no adjustments for Type I error will be made.

Figure 1: Prior Distributions

8.3 Additional Supplemental Analyses

The primary analysis will be repeated at all interim steps but based only on the number of injuries that are diagnosed by a physician or medical professional. Since the study is unblended, this analysis will contribute to an assessment of reporting bias between the two arms.

If stopping is indicated according to the guidelines above, a full standard Logistic model for the probability of injury (McCullagh and Nelder 1999) will be constructed through backward model-building steps incorporating all available covariates including stretch assignment, compliance, gender, age, extent of running, medications, medical history, etc. The model will also be fit for specific injuries and for injuries diagnosed by a physician or medical professional.

8.4 Missing Data Considerations

Given the nature of this trial, missing outcome and covariate data is to be expected. Covariate and outcome data will be imputed following methods as outlined by Huang et al 2005 and Tang et al 2005. Sensitivity of the analysis to the missing data and imputed data will also be assessed by running the primary and secondary analysis on both the imputed data and the complete data. There will be two complete data sets considered. With respect to the primary, unadjusted analysis, only subjects who are injured or complete the 3 month study without injury will be considered. For secondary analysis, model building will be performed based on the imputed data set, and the complete data will be considered those subjects for whom the reduced set of covariate information is available.

9. Ethics and the Protection of Human Subjects

This study will be conducted following the principles laid out in the Declaration of Helsinki and the Belmont Report. The investigators recognize that although a medical product is not being tested, this study involves human subjects and is considered research. Subjects will not be compensated for their participation.

9.1 Informed Consent

Every subject will provide informed consent to participate in the study and will have ample opportunity to ask questions of the study Principal Investigator. For minors, a parent or guardian will be requested to sign the Informed Consent. USATF does not have the means to ensure that a minor does not self-consent, but there are no direct rewards for participation or other incentives that are expected to attract minors. There are no scores and no individual feedback from submitted data.

For online enrollment there will be blocks of text and applicants must check a box beside each block of text to indicate that they've read it. It looks something like this:

"Please put a checkmark in the box beside each statement to indicate you have read and agree to it. Registration for this study is conditioned upon acceptance of this Informed Consent."

9.2 Potential Risks to Subjects

There is a slight risk of soft tissue strain resulting from performing the protocol-prescribed, pre-run stretch. It is extremely unlikely that the pre-run stretch will cause any permanent disability of any kind.

Inherently, there is a possibility that one of the test arms will have a higher risk of injury than the other. In that sense, subjects in one group may be at a higher risk of injury than those in the other group. The study's analysis plan considers this possibility and includes the opportunity to stop the study if it becomes clear that one group is subject to a higher risk. Injuries that are expected to occur during the study include plantar fascitis, ankle/foot tendonitis, Achilles tendonitis, iliotibial band tendonitis, knee ligament tear, meniscal tear, patellar-femoral syndrome/chondromalacia, muscle tears, disc herniation, sciatica, groin pull, and stress fracture.

9.3 Confidential Information

Those who register online to participate in the study will be providing only a name and e-mail address, street address or fax number as identifying information. This identifying information will be separated from the study data and will be used only for communication with the subject specifically and strictly for study purposes with one exception (below). The data will be coded for that subject and processed as an anonymous participant. The name and the communication information will be kept in a separate location from the study data and will be kept in a secured manner. For online users of our organization's website, we provide the following information:

For those study participants who wish to use the U.S. Postal Service or fax, the identifying information that they provide, which can be limited to a name and return address or fax number, will be kept in a separate location from the study operation and will be kept in a locked compartment. The subject will be coded for data identification so that the data can be processed anonymously. The identifying information will be accessed only for communicating with the subject specifically and strictly for study purposes with one exception (below).

Exception: All study participants will be given information on how to become a member of USATF. There will be no linkage between their participation in the study and any subsequent membership decision of study participants. The information will be provided only one time.

10. References

Fayers P, Ashby D, Parmar M. Tutorial in Biostatistics Bayesian Data Monitoring in Clinical Trials. Statistics in Medicine 1997, 16;1413-30.

Huang L, Chen M, Ibrahim J. An Analysis for Generalized Linear Models with Nonignorably Missing Covariates. Biometrics 2005 Sep: 61(3): 767-80.

McCullagh P, Nelder J. Generalized Linear Models 2nd Edition. Chapman & Hall/CRC, 1999.

Shrier I. Does Stretching Improve Performance?: A Systematic and Critical Review of the Literature. Clinical Journal of Sport Medicine. 14(5):267-273, September 2004.

Tang L, Song J, Belin T, Unutzer J. A Comparison of Imputation Methods in a Longitudinal Randomized Clinical Trial. Statistics in Medicine 2005 Jul 30; 24(14): 2111-28.

Thacker S, Gilchrist J Stroup D, and Kimsey C. The Impact of Stretching on Sports Injury Risk: A Systematic Review of the Literature. Medicine & Science in Sports & Exercise, Vol. 36, No. 3, pp. 371-378, 2004.

Vail A, Hornbuckle J, Spiegelhalter D, Thornton J. Prospective Application of Bayesian Monitoring and Analysis in and 'Open' Randomized Clinical Trial. Statistics in Medicine 2001; 20: 3777-3787.