The best magazine
Analysis Plan
Statistical Analyses
The factorial design of the trial allows for the main effects and interactions between interventions to be examined. Reflecting the clustering in the data, the outcomes will be compared using multilevel models, with adjustment for minimisation variables. Statistical significance will be at the 2.5% level and corresponding confidence intervals will be derived. All participants will remain in their allocated group for analysis (intention to treat). Subgroup analyses using interaction terms will explore the possible effect modification of a number of factors (See Sub group analysis), all using stricter levels of statistical significance (p < 0.01). Missing patient reported outcomes will not be imputed at the follow-up time points for the primary analyses. However, we will investigate the mechanism of missingness using regression models and apply an appropriate missing data model as a sensitivity analysis. All trial analyses will be according to a statistical analysis plan that will be agreed in advance by the Trial Steering Committee (TSC). The Data Monitoring and Ethics Committee (DMEC) will meet at 9, 24 and 36 months to review progress and recommend any divergences from planned trial design.
A single main analysis will be performed at the end of the trial when all follow-up has been completed. Unblinded interim analyses will be conducted for the DMEC meetings as required.
Sub Group Analysis
Patient participant age (years):- < 45, 45 to 64, ≥ 65;
Smoking:- non-smoker or smoker;
Periodontal disease severity:- no clinical signs, presence of gingival bleeding on probing, pocket depth ≤ 4 mm or > 4 mm;
Intervention provider:- dentist or practice hygienist.
Economic Analyses
Estimation of Costs. Health care resource utilisation data will be combined with unit cost information for the use of specific resources provided by the participating practices; use of routine data sources and patient participants' time, travel and out-of- pocket costs (for the latter this will only include costs not otherwise collected from participating practices). Data on costs for each area of service use will summed to provide an average cost per patient participant. Sensitivity analysis will be used to explore the impact of price paid by patient participants on the uptake of dental services. These data will be used to consider whether use of services systematically varies by the extent of NHS coverage.
Estimation of Benefits. The benefit side of the economic evaluations will firstly be based upon the effectiveness data detailed in outcome measures. Patient's may place different weight on these different outcomes and also have preferences for the way in which services are organised. A DCE will be used to provide a framework to weight different process and outcomes measures. DCEs are increasingly used in the evaluation of health care interventions to produce overall benefit scores for treatments as well as examine the absolute and relative importance of different outcomes considered as important. This approach has been adopted as measures such as quality adjusted life years typically used in economic evaluations may not be sufficient to capture the strength of preferences for differences in the process and outcomes of care associated with each intervention. Briefly, DCEs describe an intervention in terms of a number of characteristics or outcomes (attributes). The extent to which an individual values an intervention depends upon the levels of these characteristics. The technique involves presenting choices to individuals that imply a trade-off in terms of the levels of the attributes. Experimental design techniques are used to define the set of choices presented to respondents and logit regression techniques are used to analyse the response data.
The DCE will be administered to a separate sample of the public obtained from an online marketing company. Respondents will be part of a large online panel who will be invited to complete on online survey via email. Panellists are rewarded for the time they take to complete the survey through a structured incentive scheme. They receive a cash reward for participating in individual surveys – the amount is clearly stated in the invitation email and related to the survey length, interest and complexity (range between 50p-£5). Each panellist will be assigned an individual ID, allowing the company to monitor panellist activity and distinguish between contact rate (e.g. those who were initially contacted and did/did not complete the survey) and completion rate (e.g. those who completed the survey and did not drop out). This is an approach that we have successfully used in previous studies and overcomes the problems caused by very poor response rates from samples drawn from the general population.
The sample size required reflects the need for the sample to be larger than the number of independent variables; provide an adequate sample for each predetermined subgroup e.g. dental attendance (regular, non-regular), non-smoker or current smoker, social economic group (high, medium and low), country (England, Scotland) (19 subgroups in total and 30–100 per subgroup. Allowing for individuals to be present in a number of groups, the questionnaire will be administered online and a maximum of 950 individuals (19 × 50) completed questionnaires will be sought.
In addition to the outcomes included in the DCE one further attribute included will be patient cost. By including this attribute, the willingness to pay (WTP) for a change in the level of any other attribute will be estimated. This information will be combined with the clinical outcomes provided obtained from the trial for each participant and about type of care provided to provide an estimate of the mean WTP for each intervention considered.
Costs and Benefits to the Practitioner Participant. Different frequencies of PI visits will impact upon clinicians' costs and benefits. The effect on incomes, job satisfaction and changes to the level of fees on the provision of PI will be assessed using self-reported questionnaires administered to clinicians over the duration of the trial.
Presentation of Results. Results will be presented both as a cost-consequence analysis (presentation of costs and outcomes, including those to practitioners) and as incremental net benefits. Net benefits will be calculated by combining estimates of mean WTP with estimates of mean cost for each intervention. The intervention with the greatest net benefit would be considered the most efficient. The evaluation will include both deterministic and probabilistic sensitivity analysis, using methods developed for previous analyses.
Sample Size
An OHA exploratory trial in the same population as the proposed trial demonstrated that at baseline 35% of gingival sites were bleeding on probing with sd = 25%. The PI Cochrane review suggested that a reduction of 15% of sites with bleeding was a plausible reduction for 6 monthly PI. If the effect is assumed linear, halving the number of PIs should half the expected difference of 15% of sites. If the effect is non linear and larger than 7.5%, the trial will be adequately powered. If the effect is smaller it would be of questionable clinical significance. There is some evidence that personalised OHA can reduce the number of gingival sites bleeding on probing by approximately 7.5%. The following calculations are based on estimating main effects from the trial. All calculations assume a significance level of 2.5% to give some protection against multiple testing.
OHA. To calculate the sample size required to estimate the main effect of OHA, it is recognised that the data are contained within a cluster RCT. Assuming a conservative estimate of the intracluster correlation (ICC) of 0.0531, a cluster RCT of 50 dentists collecting information from 25 patient participants each (25*25 = 625 patients per arm) will have 90% power to detect a difference of 7.5%. Should the correlation be 0.1, the trial will still have approximately 80% power to detect a difference of 7.5%.
PI. Given that the comparison of routine versus personalised OHA requires 625 patient participants in each arm, equal randomisation 1:1:1 (no PI; 6 monthly; 12 monthly) of patient participants implies 208 in each of the six groups. Assuming no interaction effect, the corresponding PI groups can be combined across both routine and personalised advice groups giving 416 patients allocated to each PI group. Assuming a sample size of 416 in each group, the trial will have in excess of 95% power for each pairwise comparison to detect a difference of 7.5% in the percentage of gingival sites that bleed on probing.
Interaction. We do not anticipate a substantive interaction effect between the PI interventions and the personalised OHA. Assuming an ICC of 0.05, the trial has 80% power to detect an interaction effect of 7.5%. Should the ICC be 0.1, the trial has approximately 80% power to detect an interaction of 10%.
At trial endpoint the total number of dentists required is 50 and the total number of participants is 1248 (6*208). Our previous trials in general dental practice suggests that we may lose a small number of dental practices in the trial for reasons such as practices amalgamating with other practices or restricting NHS patients. We have therefore very conservatively assumed 17% attrition for dentists and 20% for participants. These assumptions imply that 60 dentists and 1860 participants will be required. Each dentist will be required to recruit on average 31 participants to ensure 25 at follow-up.
Recruitment Plans
The trial will recruit 60 dental practitioners from 60 general dental practices in Scotland and North East England (Newcastle). Participating dentists will represent a cross-section of practitioners operating in a range of different circumstances (e.g. urban or rural, high, middle or low income communities, employing or not employing a dental hygienist). The target recruitment is for 40 dental practitioners to be in Scotland with the remainder in Newcastle.
Recruitment of 60 general dental practices is projected to take 12 months and the 1860 participants recruited within 16 months. The recruitment projection is shown in Figure 3.
(Enlarge Image)
Figure 3.
Recruitment projections.
Ethical Considerations
The project will be coordinated by a Trial Co-ordinating Office (TCOD) in the Dental Health Services Research Unit in the University of Dundee and CHaRT in the University of Aberdeen. Both institutions are committed to the highest standards of research governance and seek to conform to all relevant governance guidelines and codes of practice as detailed in the Research Governance Framework and ICH guidelines for Good Clinical Practice (GCP). As well as ensuring that research is conducted according to the requirements set out in these documents, all research will be conducted with the written agreement of the relevant Multi-Centre and/or Local Research Ethics Committee(s), and/or other relevant ethics committee(s) before starting recruitment. Favourable ethical opinion for the IQuaD study was confirmed by the East of Scotland Research Ethics Service on 24 March 2011 (REC reference number 10/S0501/65).
A study information leaflet will be given to each potential participant to inform them of the anticipated risks and benefits of taking part in the study. In particular, the trade-offs between possible short-term benefits and long-term risks will be explained.
Informed signed consent forms will be obtained from the participants in all centres, by an individual who is trained in GCP. Patients will be given sufficient time to accept or decline involvement and are free to withdraw from the study at any time.
Data Protection and Archiving
Patients will be reassured that all data which are collected during the course of the research will be kept strictly confidential. All patients' details will be anonymised and stored on a database under the guidelines of the 1998 Data Protection Act. The relevant research documentation will be archived at the University of Dundee for at least five years after completion of the trial as required by the applicable regulatory requirement(s).
Governance Arrangements
Research Governance applies to everyone working in the Dental Health Services & Research Unit and CHaRT. As such, all research will be conducted within the appropriate legislative and regulatory environment and in accordance with GCP. All staff involved in the trial at the two centres will have undertaken appropriate GCP training (to a level of knowledge that reflects their exposure to the principles). The three main groupings that contribute to the governance arrangements for this study are: the Trial Management Committee; an independent Trial Steering Committee (TSC); and an independent Data Monitoring Committee (DMC). The Trial Steering Committee (TSC) includes an independent Chairperson (Elizabeth Treasure, Professor in Dental Public Health, University of Cardiff), other independent members include Eleanor Grey, Consumer Representative, Tina Halford-McGuff, and James McCaul and will oversee the trial. The TSC also comprises a selection of the co-applicants including the Principal Investigators (Clarkson and Ramsay), the trial statistician and the Director of CHaRT. There will only be two voting members drawn from any of the co-applicants. The TSC will meet annually throughout the course of the study.
The Data Monitoring and Ethics Committee (DMEC) will be chaired by Damian Walmsley, (Professor of Restorative Dentistry, University of Birmingham) and include Peter Robinson and Pollyanna Hardy. It will meet early in the trial to agree it's terms of reference and other procedures and will likely have further meetings at 9, 24 and 36 months. The DMEC will report any recommendations to the Chair of the Steering Committee.
The University of Dundee has agreed to act as sponsor. As such, the TCOD will undertake to communicate promptly and effectively with the sponsor to satisfy and reassure the sponsor that the sponsor's obligations on the authorisations, the financing and the progress reporting (including emerging safety data) of the trial are being met. This may include providing comprehensive information before the start of a trial for the purposes of risk assessment for the sponsor.
Arrangements for Day-to-Day Management of the Trial
The trial will be co-ordinated from the TCOD in the Dental Health Services Research Unit, Dundee, and will provide day to day support for the clinical centres and outcome assessors/research nurses. The TCOD will be responsible for transacting the randomisation, collecting all trial data (including postal questionnaires), co-ordination of patient participant appointments, follow-up and data processing. CHaRT, Health Services Research Unit, Aberdeen University will provide the database applications and IT programming for the TCOD, and host the randomisation system, co-ordinate the patient follow-up questionnaires, provide experienced trial management guidance, and take responsibility for all statistical aspects of the trial (including interim reports to the TSC and DMEC). The outcome assessors will be responsible for recruiting participants (including initiating the randomisation call) and performing all clinical outcome assessments. An Operations Management Committee, led by the Trials Manager, will meet weekly in the early stages at the TCOD to ensure smooth running of the trial, trouble-shooting issues as they arise, and ensuring consistency of action across the participating centres. CHaRT staff in Aberdeen will join this group as required, weekly by teleconference, and in person every 4–6 weeks. These face to face meetings will become less frequent as the trial progresses successfully, and increase again in frequency as the trial enters its closedown phase. A Trial Management Committee will meet biannually and be chaired by the Principal Investigators, and include co-investigators and key members of the TCOD and CHaRT. Their remit will be to oversee the progress of the trial, and they will report to the independent TSC.
Trial Oversight
As described above, the trial will be overseen by a Trial Steering Committee and a Data Monitoring and Ethics Committee. In addition an expert Periodontal Advisory Committee has been convened to provide expert clinical advice to the Trial Management Committee throughout the duration of the study.
Data Monitoring
A Data Monitoring and Ethics Committee (DMEC) met early in the trial and agreed it's terms of reference and other procedures. The DMEC will make any recommendations to the chair of the Steering Committee.
Safety Concerns
The design of the study ensures that adults for whom allocation to a no-PI intervention may be detrimental are not eligible to be included the study. Periodontal disease and caries progress very slowly. During the trial participants will be monitored as per routine practice, possibly more frequently than might otherwise have been the case, and they may receive more frequent preventive oral hygiene advice. It is made clear to both the patients and their dentists that, within the design of the study, it is acknowledged that patients may attend anytime a dental appointment is needed and that these visits may be in addition to any study-specified recall visits. Thus no dental treatment, whether delivered in the dental surgery or following referral to specialist services will be withheld from patients as a result of taking part in this study. The PI intervention being evaluated has been routine in the NHS for many years and has no known safety concerns.
Sponsorship
The University of Dundee is the sponsor of the research.
Finance
The study is supported by a grant from the National Institute Health Research (NIHR) Health Technology Assessment Programme (ref 09/01/45).
Publication
The results of the study will be reported first to study collaborators. A main report will be drafted by the project management group and circulated to all clinical co-ordinators for comment before a final version is considered for publication by the steering committee.
Dissemination
The results of this trial will be disseminated widely and actively through professional, primary care, public and scientific routes. Results will be communicated directly to all participating dental practices and an open workshop will be held with them discussing the next steps in getting the findings of the study to influence clinical practice. The trial results will be used to update Cochrane reviews, inform policy (through targeted feedback to all of the UK Health Departments and the British Association for the Study of Community Dentistry and its Consultants in Dental Public Health Group); practice (through specific communications to the National Institute for Health and Clinical Excellence (NICE), the British Dental Association and the Faculty of General Dental Practice (UK)); the public (through INVOLVE and patient organisations) as well as with dental education and training (through a range of communications to postgraduate dental Deans, the undergraduate dental schools and if appropriate to aid the development of educational support material developed from the training CD-ROMs.
Given the current dearth of directly applicable evidence around this important research question, it is anticipated that the impact of this trial will also be felt at the International level as well as closer to home (specific presentations will be made to the International Association for Dental Research and its Evidence Based Dentistry Network as well as to organisations such as the European Association for Dental Public Health and related European specialty societies for research and practice.
Milestones for the IQuaD Trial
Dental practice recruitment began in month 7. Patient recruitment began in month 11 and is planned to continue until month 27. Follow up assessments will be made at three years, so the last patient will be seen in month 63.
Source: ...