Introduction: The National Institute of Dental and Craniofacial Research has made significant investments in dental practice-based research networks dating back to 2005. By connecting practitioners with experienced clinical investigators and addressing questions faced by dental health practitioners on a daily basis, PBRNs have the potential to generate research findings that are immediately relevant to practitioners. These networks hold great potential to improve the translation of research evidence into daily practice. Prior efforts to examine this issue have relied on questionnaires to assess the dentist's stated clinical practice in a given clinical scenario. Clinical data examining actual practice patterns is needed to better answer the question of whether participation in topic related studies and related dissemination meetings changed their practice patterns to align with the evidence. This project utilized clinical data from the electronic heath record to examine actual practice patterns at baseline when HealthPartners dentists first enrolled in the network and examined practice patterns at a later time period after completion of the study and related meetings.
Methods: Multivariate logistic regression models adjusting for clustering at the clinic, provider, and patient level were used to compare 2005 and 2009 restoration rates. Preparation of the analytic datasets was done in four steps using two data sources. First, using the electronic dental record (EDR) all F80, F81, and F82 findings corresponding to preventive dental visits occurring in the calendar years 2005 and 2009 were identified. Visits with a finding were classified as index visits. Second, using the EDR, each finding's treatment was determined by identifying any treatment codes occurring for a period of up to six-months following the date of the finding code's index visit. Treatments were placed into one of four categories: Fluoride, Remineralization, Restoration, and Unknown. If multiple treatment codes were identified, the following order was applied: Restoration, Remineralization, and Fluoride. Third, using the EDR, the number of co-occurring dental finding for that patient were identified and classified as enamel (F80, F81, F82), dentin (F83, F84), or sealant (F892, F893) damage. Fourth, each HPDG provider's level of DBPRN engagement was determined using a separate database maintained by the study coordinator.
Results: Three levels of increasing DBPRN involvement were identified: None (n=6), Low (n=15), and High (n=14). The multivariate, cross-level models yielded two important findings. The first was that there was a significant overall reduction on restoration rates for the entire dental group. In 2005, prior to DBPRN involvement, restoration rates across the three groups of providers varied significantly. HPDG providers with no future DBPRN involvement restored 85% of all F80, F81 and F82 findings. Providers with a future low level of DBPRN involvement restored 78% of F80, F81, and F82 findings in 2005; and, HPDG providers with a high level of future DBPRN involvement only restored 73% of similar findings. In 2009, providers with no DBPRN involvement restored 54% (37% relative decrease) of F80, F81, and F82 findings. Those with a low level of DBPRN involvement restored 48% (39% relative decrease) of F80, F81, and F82 findings, and those with a high level of DBPRN involvement restored 46% (38% relative decrease). The second key finding was that the level treatment variation across providers was far greater than the level of variation across levels of DBPRN involvement. In 2005, restoration rates ranged from high of 95% to a low of 43% across providers. In 2009, restoration rates ranged from a high of 70% to a low of 24%.
Conclusions: There are multiple factors impacting the decision to restore or remineralize a tooth found to have a carious lesion. While involvement with the DBPRN appears to correlate with lower overall rates of restoration, it did not correlate with practice change defined as significant decreases in restoration rates. A more detailed analysis of how treatment patterns evolved over the study timeframe is needed to better understand the impact of DPBRN involvement upon dental treatment.