Methods |
Study type: Cluster‐randomised control trial (with 2 x 2 factorial design) Duration of study: 18 months: six‐month fee intervention period from September 2003 to February 2004 followed by a 12‐month follow‐up period to allow for the recording of any continuing effects of the interventions on dentists’ decisions to place fissure sealants Setting: Scotland, in areas representing the four most deprived categories (out of seven) in the Scottish post code based system defining deprivation |
|
Participants |
Unit of randomisation: Dentists Unit of assessment/analysis: Patient‐level outcomes were measured, but the results were presented at dentist‐level i.e. the mean percentage of 12‐ to 14‐year‐olds receiving fissure sealants for second permanent molars per dentist (weighted by number of children seen) Method of recruitment:
Inclusion criteria:
Exclusion criteria:
External validity: The participating dentists may have had different characteristics to the non‐participating dentists. Out of the 284 dentists who were eligible to take part, 131 (46%) refused or did not respond (and a further four were not randomised due to being late recruits etc). In choosing to take part, the dentists would have the possibility (depending on which arm they were randomised to) of claiming fee‐for‐service remuneration on top of their normal capitation remuneration, therefore those that chose not to take part may be less motivated to claim the small sums associated with the fee‐for‐service remuneration, at least in the context of the study (the fees were to be claimed from the research team rather than via the routine National Health Service channel). In addition, all the dentists were working in deprived areas, which would have different levels of need compared to Scotland as a whole |
|
Interventions |
Interventions:
Control: no specific intervention Applicability:
|
|
Outcomes |
Primary outcomes: 1) Measures of clinical behaviour of primary care dentists
2) Healthcare costs (including costs of (i) introducing the incentives, (ii) the transaction, (iii) the information systems and (iv) monitoring)
Adverse outcomes: None stated |
|
Source of funding | Chief Scientist Office, Scottish Executive and Scottish Higher Education Funding Council | |
Notes | ||
Risk of bias | ||
Bias | Authors' judgement | Support for judgement |
Random sequence generation (selection bias) | Low risk |
“Sampling, randomisation, and analysis were conducted at arm’s length from the study base by the Health Services Research Unit, University of Aberdeen" The authors state that they carried out randomisation using minimisation, a form of covariate‐adaptive randomisation. The process involved four practice‐based variables which were obtained pre‐randomisation from National Health Service records: the deprivation category for the area of practice, the number of partners in the practice, the throughput of 11‐ to 13‐year‐olds and the number of restorative sealant claims in 2002 |
Allocation concealment (selection bias) | Low risk |
“Sampling, randomisation, and analysis were conducted at arm’s length from the study base by the Health Services Research Unit, University of Aberdeen" As minimisation was used to randomise the dentists, the dentists would have to have been recruited before the random sequence was generated, which therefore reduces selection bias |
Blinding of outcome assessment (detection bias) Dental health | High risk |
"A random sample of 25 [patients] per dentist was taken, and data collection forms were sent to each dentist for completion from practice records 12 [months] post‐intervention" Blinding was not possible for the dentists, who acted as the outcome assessors after they had received the intervention i.e. data collection forms were sent to each dentist for completion using patient records, which means that the outcome measure depended on the accuracy of the dentists' reports. The authors stated that given that the sealant placement did not attract a National Health Service fee‐for‐service payment – and therefore did not generate a National Health Service‐held record – it was not possible to corroborate the dentists' sealant placement records. Equally, National Health Service‐held records of the dental visit during which the sealant was placed are not reliable as no such record would occur if no other fee‐attracting treatment was given during the visit and the child did not require re‐registration. The authors stated that other reasons for mismatches are scanning error, transcription error, and delayed submission of claim forms from the practice to the National Health Service body responsible for fee payment. The authors carried out a crosscheck of National Health Service‐held records of dental visit dates and practice‐reported dates of sealant placement and found matches in 28% (109) of cases. The authors stated that while it is possible that bias was introduced, this would have applied equally across the intervention arms but this statement is not verifiable. However, the outcome was objective, which lowers the risk of bias even though the outcome was not assessed blindly |
Incomplete outcome data (attrition bias) All outcomes | Unclear risk | There does not appear to be an imbalance of missing data across the fee‐for‐service and control arms (for example, four dentists in both the fee‐for‐service and control arms were lost to follow‐up), nor an imbalance in the reasons behind missing data in these arms (for example, 7.1% of children were excluded from the fee‐for‐service arm and 10.4% of children were excluded in the control arm because they did not have erupted second permanent molars). However, the authors do not appear to have conducted statistical analyses to check for possible imbalances, which may have occurred, particularly between the arm where dentists received fee‐for‐service remuneration and education (six dentists were lost to follow‐up in this arm) and the education arm (two dentists were lost to follow‐up in this arm). The data were analysed using the intention‐to‐treat principle. For example, dentists who did not attend the education intervention were mailed the course material and retained in the study on an intention‐to‐treat basis, thus reducing the use of incomplete outcome data |
Selective reporting (reporting bias) | Unclear risk | No protocol is cited and it is not stated whether all the prespecified primary outcomes have been reported |
Other bias | Unclear risk |
"There was a lower baseline of sealant treatment of second permanent molars in the fee and both arms. No other significant baseline differences in practice or practitioner characteristics were found" The baseline characteristics and baseline outcomes of the arms were generally balanced (partly due to the fact that minimisation using four dental practice‐level variables was carried out at the beginning of the study), thereby reducing the possibility of confounding. No statistically significant differences in baseline characteristics of dental practices/dentists were found between the arms. However, there was a slight imbalance between arms in the patient‐level characteristics measured at baseline i.e. there were statistically significant lower percentages of children having at least one sealant treatment of second permanent molars at baseline in the fee‐for‐service and both fee‐for‐service and education arms compared to the education and control arms. Even so, the primary analysis (which adjusted for a number of variables including the number of sealants placed in first permanent molars pre‐intervention) found a statistically significant difference in favour of the dentists receiving fee‐for‐service remuneration. However, the secondary analysis (which did not adjust for baseline differences) did not reach statistical significance. The risk of contamination is not a concern as dentists (rather than patients) were randomised and it is unlikely that communication between dentists in the different arms could have occurred as a maximum of one dentist per dental practice was selected |