A multi-centre, experimental randomised clinical trial is the study design chosen as it is the recommended approach to obtain a valid comparison of competing intervention strategies (Green, 1997). The independent variable is the instructional method which will be manipulated to see whether it has any affect on the dependent variables, which are adherence and exercise performance. The Digital Versatile Disc (DVD) group will represent the study group and the group receiving illustrations will act as the control group as illustrations are supported by literature and currently used in standard physiotherapy practice. It is necessary to include a control group to prevent overestimation of treatment effect. A true control group (receiving no instructional aid and making the trial a randomised controlled trial) has not been used as it would be unethical to withhold treatment clearly shown to be effective (Schneiders et al., 1998).
The trial aims to be as representative of the UK population as possible to increase population/external validity. The trial will include 84 females and 56 males to represent the 3:2 female/male ratio of osteoarthritis sufferers in the United Kingdom (ARC).
Under-representation of ethnic groups would affect the external validity of relative effect estimates. The National Statistics Office states that the 2001 Census found that 8.3% of the UK population belonged to other ethnic groups and this figure appears to have risen to 10.1% according to the Labour Force survey, 2006. Thirty percent of participants will be recruited from other ethnic groups so that 1) the strata is large enough and 2) ethnic minorities are fairly represented. This figure will however be checked with the statistician prior to trial commencement.
Research suggests that people with a lower educational attainment are less likely to adhere to exercise, it is therefore important to consider where participants will be recruited from. A study by the Organization for Economic Cooperation and Development (2005) showed a close correlation between educational underachievement and areas of social deprivation in London. Cooper (2003) found those living in inner London generally achieved lower grades than those in Outer London (appendix1).Therefore to increase external validity, 2 Inner London and 2 Outer London locations have been chosen.
Inclusion/exclusion criteria are used in this study to identify appropriate participants and ensure their safety. Strict criteria will ensure 1) investigators cannot bias the results by influencing selection to include or exclude specific prognostic factors 2) participants represent the populations to which the trial's findings will apply, thus maintaining external validity. It is also important to consider that if the criteria are too restrictive it could limit the size of trial and the findings will lack generality.
Participants will be eligible if they are aged 50+ as 1 in 5 adults have knee osteoarthritis by this age and it is uncommon in people younger. By the age of 80, it increases to 1 in 2 adults (ARC). The trial will have no upper age limit so that the findings are relevant to all age groups susceptible to knee osteoarthritis. Inclusion of older participants is particularly pertinent to this study as Henry et al (2000) found adults aged 65+ had greater difficulties remembering exercises which may affect adherence and exercise performance.
Participants must have adequate hearing, vision and understanding of written and spoken English to allow equal comprehension of the exercise instructions. In practice this may not be necessary; however, for the sake of this trial it must be controlled to prevent confounding results.
To prevent allocation bias, participants need access to a DVD player, irrespective of their allocation. Eighty-six percent of the UK population own a DVD player (Office of National Statistics, 2008) therefore this will only exclude a small number of the population so it is considered acceptable.
Radiographic evidence of osteoarthritis, grade 2-3 on the Kellgren and Lawrence scale is an inclusion criteria as Deyle (2000) found increased disease severity is associated with lower adherence. Ensuring all participants are mild-moderate makes the results comparable. Additionally, it would be unethical to provide unnecessary treatment. The Nice guidelines (2008) recommend strengthening exercises for osteoarthritis patients; therefore participants would be prescribed strengthening exercises irrespective of the trial. Radiographic evidence ensures patients are not given an inappropriate treatment by clinicians who may want to increase recruitment rate.
Participants with inflammatory arthritis will be excluded as strengthening exercises can exacerbate inflammation in the knee joint during a flare-up (Luqmani et al.,2006). Similarly, Sharma et al (2003) found that in certain joint environments, such as malalignment or laxity, higher quadriceps strength was associated with a greater risk of disease progression therefore such participants will also be excluded. Participants who have previously been prescribed lower-limb exercises or partaken in regular exercise classes will be excluded as they may show increased exercise performance and adherence. Inclusion would be unethical for participants who have cognitive impairments as informed consent may not be achievable. Such participants may also have impaired ability to retain information which could confound the results. Finally, participants will be excluded if they have cardiac problems or any serious pathology as they may have an increased risk of an adverse reaction or need onward specialist referral which may delay commencement of physiotherapy.
It is important someone other than the person randomising the participants determines eligibility, to help ensure allocation concealment. (Altman and Schluz, 2001).
Random allocation and allocation concealment
This trial will use sequentially numbered, otherwise identical, opaque, sealed envelopes (SNOSE) to randomise participants and ensure allocation concealment. Participants will be randomised as it is considered the most reliable method of creating homogeneous treatment groups in terms of known or unknown prognostic factors (Altman 1999). Internal validity is increased as randomisation controls for unsuspected effects and increases the probability that any observed differences are caused by the experimental condition.
An additional protection against bias during randomization is concealment of allocation. This will prevent the researcher from influencing which intervention group each participant is assigned to, therefore preventing a selection bias. A selection bias could result in an overestimation of the effects of the intervention by approximately 40% (Schulz, 2002a). SNOSE is an economical and straightforward means of assuring allocation concealment. The additional precautions (e.g. aluminium foil, carbon paper) increase the difficulty of unmasking the allocation scheme. By placing the group assignment card inside a folded sheet of aluminium foil ensures the envelope is impermeable to light and lining the inside of the envelope with carbon paper allows the envelope to transfer information onto the assignment card, creating an audit trail. Stratified randomisation is still possible using randomisation envelopes by having a set of envelopes for each combination of stratification factors.
Baseline characteristics such as age, gender, ethnicity and education will be recorded at the beginning of the trial. As this is a relatively small trial, there is an increased risk of an uneven distribution of patient characteristics between groups (Kernan 1999). Age, ethnicity and educational attainment are known to be strongly related to adherence and performance (prognostic) therefore stratification is needed (Schluz, 2002b). Sluijs et al. (1993) found adherence to home exercises increases with age and is inversely proportional to educational level. Rastall et al. (1999) and Henry et al (2000) found adults aged 65+ had greater difficulty remembering exercises which may affect performance. Evidence also suggests ethnic minorities are less likely to adhere to exercise (e.g. Muse, 2005). To allow for 3 strata the sample size had been increased. However, as the trial is to include only 30% of participants from other ethnic groups this may affect stratification. Therefore, this figure will be checked with the statistician prior to the commencement of trial and if found to be too low to adequately stratify ethnic minorities, either sample size or the number of ethnic minorities included will increase further.
Stratified randomization also helps prevent type I error (Feinstein and Landis, 1976) and improve power (Lachin and Bautista, 1995) and is most beneficial to trials (such as this) with less than 200 participants (Schluz, 2002b). As the author is uncertain as to what the actual number of participants in each stratum will be before the trial commences, permuted blocks within each strata will ensure balance between the main treatment arms. Schluz(2002b) suggests that it may be possible to anticipate the randomization sequence in some trials that are block-randomized using uniform block sizes, therefore this trial will use two different block sizes (2 and 4).
Ideally, both participant and investigator should be blinded to avoid the comparison of treatments from being distorted (Schluz, 2002c) and to prevent ascertainment bias. This trial is a single-blind study as active subject participation is required. The physiotherapist (who also acts as the outcome assessor) will be blinded as knowledge of group assignment could create a bias when assessing outcome measures due to preconceived notions of "expected" outcomes. Furthermore, as the Exercise Assessment Scale has a degree of subjectivity, blinding the outcome assessor will strengthen their objectivity.
Strengthening exercises were chosen as they have been shown to improve pain and function in osteoarthritis patients and are part of the NICE(2008) guidelines. Previous studies suggest a HEP can maintain improvements (Huang et al, 2003; Huang et al, 2005), however long term adherence was low. This suggests a need to establish ways to improve adherence to treatments clearly shown to be effective. The exercise intensity, frequency and repetitions will be tailored to the individual and based on findings of recent evidence (e.g. Richards,2004; Jan et al., 2008). Strengthening exercises are also more likely to require specific instruction than, for example, aerobic exercises and older individuals in particular may find exercise technique difficult to remember.
Various studies (e.g. Grove, 1999) find participants using exercise videos have increased motivation to exercise and are less likely to develop bad habits/techniques. It is therefore hypothesised DVD's will increase adherence and exercise performance. Additionally, if performance improves physiotherapist time correcting exercise should decrease.
The Exercise Assessment Scale (Smith et al.,2005, adapted from Rastall et al.,1999) was chosen as it enables detailed assessment of performance and had the highest reliability of all the measures found. Validity of scale has not been formally reported, however, inter-rater reliability has been assessed (Smith et al.,2005) using the intra-class correlation coefficient (ICC). The ICC was 0.99 indicating excellent inter-rater reliability (Shrout and Fleiss, 1979). Prior to the trial, training will be given to ensure accuracy and consistency of measurement is increased thus increasing statistical power. As the measurement has a degree of subjectivity, assessments will be recorded so that performance can be analyzed post-intervention to ensure participant grading is equivalent.
Participants will use a self-report log to measure adherence to their HEP. It is a commonly used measure which has been found to be reliable when compared to spouse reports (Follick et al., 1984) and objective pedometer values (Speck,2006) supporting the validity of the measure. The recognized benefits include the ability to collect data from a large number of people at low cost and that it does not alter the behaviour of the individual (increasing ecological validity). However, as the measure is subjective, Vitolins et al (2000) recommends more than one method should be used to assess adherence. A widely used objective measure of percentage attendance to physiotherapy sessions will also be used.
Instruction time will be assessed as correction of exercises uses valuable physiotherapy time that could be spent on adjunct treatments. It is therefore important to measure the effectiveness of any intervention in reducing time allocated to exercise correction.
Means and standard deviations will be calculated for all variables. For parametric data (e.g. number of repetitions, exercise frequency) the means of the two groups will be compared using the independent-sample t-test as the study is experimental in design, comparing 2 groups of subjects (Hicks, 2004). The Exercise Assessment Scale provides ordinal data (Smith et al, 2005), therefore a Mann Whitney U test (a non-parametric test) will be used. P-values are used to determine whether a null hypothesis formulated before the study commenced is to be accepted or rejected. The level of significance will be set at p <0.05 so that the risk of type II error is reduced. In exploratory studies such as this, p-values enable the recognition of any statistically noteworthy findings. Confidence intervals will also be used as they provide a range of values for the variable of interest (the measure of treatment effect). This range has a specified probability of including the true value of the variable, as well providing information about the direction and strength of the demonstrated effect. Conclusions can then be made about the clinical relevance and the statistical plausibility of the study's findings. Both statistical measures will be reported because they provide complementary types of information (Prel,2009).
Intention to treat analysis will be used as it retains balance in prognostic factors arising from the original random treatment allocation and gives an unbiased estimate of treatment effect.
The trial is 18-months to allow for a sufficient accrual period and for a long-term intervention. The author contacted a similar sized Outpatient department, which reported approximately 30-40 patients present with osteoarthritis over a 6-month period. In addition participants will be recruited from doctor's surgeries and orthopaedic waiting lists. It is estimated that 12-14 participants at each hospital will enter the trial during each recruitment phase. The minimum number required is 12 so that 140 participants are recruited by the end of recruitment stage-3. If only 2 participants at each hospital are recruited each week, it will require a 6-week accrual period for each phase of the trial. The time to recruit the target population is often under-estimated (Carter,2005) therefore the accrual period has been increased to 8-weeks to ensure adequate numbers are reached.
The intervention period will be for 1-year as a recent review (Jordan,2010) identified a need for trials with long-term follow ups which explicitly address adherence.
Four physiotherapists, one from each hospital will be required to work 153 hours over the 18-month trial. Participants will be seen 9 times by their physiotherapist. The initial session will be 45minutes and the remaining sessions will be 30minutes (as standard in physiotherapy practice). The 3 recruitment and intervention phases are staggered, therefore, if the physiotherapist sees 3 participants per week (the required 12 per month) and appointments are made for the same time each month, the physiotherapist will be required to work 0-4.5 hours per week (see physiotherapist timetable, appendix2). The physiotherapists will be employed on an hourly basis. The research assistant will be employed on a part-time basis to randomise participants and issue the instructional aids.
The study aims to be of "value" and to improve the health of patients with osteoarthritis of the knee.
It will be conducted in full conformity with the Declaration of Helsinki (WMA, 2004) and the ICH Guidelines for Good Clinical Practice (EMA,2002). This is vital as The Declaration of Helsinki sets out ethical principles regarding human experimentation and is "widely regarded as the cornerstone document of human research ethics" (Tyebkhan 2003). As this study is using NHS patients and facilities it will also gain ethical approval from the NHS Research Ethics Committee.