The meeting, which took place on 20 July at the University of Manchester, was organised by Tim Pickles (Cardiff University) on behalf on the RSS Medical Section. Tim, along with Salma Ayis (KCL) and David Reeves (University of Manchester), is a Primary Health Care Special Interest Group (PHCSIG) legacy member of the Medical Section committee, and thus tasked with ensuring a continued primary health care discussion.
There were presentations by Rebecca Playle (Cardiff University), Jing Kang (University of Leeds) and Helen Worthington (University of Manchester) and a joint presentation by Nicola Innes and Thomas Lamont (University of Dundee).
Rebecca Playle's talk, titled 'Seal or Varnish? Addressing adherence in a dental clinical trial' presented analyses related to the Seal or Varnish (SoV) trial, which compared effectiveness of fissure sealants to fluoride varnish to prevent dental caries. Complier Average Causal Effect (CACE) was undertaken to investigate the effect of adherence but is designed where there is an intervention arm compared to a control arm, unlike the two intervention arms here. An extended CACE analysis found the primary outcome unaffected by outcome. There was an interesting discussion around adherence in clinical trials and whether this would reflect adherence in practice.
Jing Kang talked about the association between sugar consumption and dental caries in children in England, Wales and Northern Ireland—Analysis of Children Dental Health Survey 2013 using zero-inflated negative binomial model. This presentation involved the Child Dental Health Survey 2013 data. The aim was to examine sugar consumption in different demographic groups and its association with dental decay. Decay was measured by DMFT (the number of decayed, missing and filled teeth). However, analyses were complicated by the data being skewed towards the high number of zeros. The zero-inflated negative binomial model was most suitable for analyses. Results showed that higher tooth decay levels were associated with age, region and non-regular dental attendance while the chance of experiencing decay was related to gender, region, lower socio-economic status and sugar consumption frequency.
Nicola Innes and Thomas Lamont's presentation titled: 'Composite outcomes + core outcomes = Chaotic outcomes? Various outcomes and outcome measures are used when different interventions are compared'. This paper presented a systematic review of randomised control trials investigating interventions for prevention and management interventions. The majority of outcomes measured were composite outcomes and, using Social Network Analysis, gaps between interventions compared in these trials were identified. This impacts on the level of data synthesis that can be carried out and the clinical recommendations that can be made to guide clinical decision making.
Finally, Helen Worthington talked about meta-analysis of split-mouth studies. In Dentistry, around 10% of studies are split mouth in design. This design has some similarities, and similar issues, to cross-over studies. Over- and under-estimation of precision and weighting are associated with failure to account for clustering and correlation. Incorporating correctly reported study data within meta-analysis is relatively straightforward (using generic inverse variance). However, for studies with incorrectly reported data, the analysis can be more complicated although there are a number of methods that can be used to allow the data to be included within meta-analysis.