The effect of interventions aiming to optimise the prescription of antibiotics in dental care-A systematic review

Abstract

Background: Abundant evidence in dentistry suggests that antibiotics are prescribed despite the existence of guidelines aiming to reduce the development of antibiotic resistance. This review investigated (1) which type of interventions aiming to optimise prescription of antibiotics exist in dentistry, (2) the effect of these interventions and (3) the specific strengths and limitations of the studies reporting on these interventions.

Method: Literature search was based on Medline, Embase, Global Health, Cochrane CENTRAL, ClinicalTrials.gov and Current Controlled Trials. Studies with one of the two primary outcomes were included: (1) The number of antibiotics prescribed and/or (2) the accuracy of the prescription, commonly measured as a percentage of adherence to local clinical guidelines.

Results: Nine studies met these inclusion criteria. Five studies reported on the prescription of antibiotics in primary dental care and four studies focused on outpatient dental care. Interventions used in primary dental care included a combination of audit, feedback, education, local consensus, dissemination of guidelines and/or academic detailing. Trials in the outpatient setting made use of expert panel discussions, educational feedback on previous acts of prescribing, the dissemination of guidelines and the establishment of internal guidelines. All studies successfully reduced the number of antibiotics prescribed and/or increased the accuracy of the prescription. However, most studies were confounded by a high risk of selection bias, selective outcome reporting and high variance across study groups. In particular, information relating to study design and methodology was insufficient. Only three studies related the prescriptions to the number of patients treated with antibiotics.

Conclusions: This systematic review was able to offer conclusions which took the limitations of the investigated studies into account. Unfortunately, few studies could be included and many of these studies were confounded by a low quality of scientific reporting and lack of information regarding study methodology. High-quality research with objective and standardised outcome reporting, longer periods of follow-up, rigorous methodology and adequate standard of study reporting is urgently needed.

Fig 1. Study selection process.

Note: PRISMA flow chart based on: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7): e1000097.

Fig 2. Decrease in the number of prescriptions for antibiotics in percentage, by studies assessing this outcome parameter.

Note: Chopra et al. (2014) and Zahabiyoun et al. (2015) do not report that outcome measure. Instead of reporting figures, odds ratios are reported in Seager et al. (2006). (a) Reduction of 1.0 antibiotic items per 100 NHS treatment claims in the intervention group compared to 0.4 antibiotic items in the control group. (b) Reduction from 2,951 prescriptions for antibiotics before the audit to 1,665 prescriptions after the audit. (c) Reduction from 2,316 prescriptions for antibiotics before the audit to 1,330 prescriptions after the audit. (d) The number of prescriptions was not stated in the paper, but the authors report an overall reduction of ~50%. (e) Reduction of the total number of prescriptions for antimicrobial agents from 253 prescriptions among 300 patients at baseline to 82 prescriptions among 300 patients at one month after the intervention. Three months after the intervention these prescriptions were at 63 among 300 patients and at 102 prescriptions among 300 patients six months after the intervention. (f) Reduction of preoperative prescriptions for antibiotics from 15 prescriptions among 80 patients before the audit to one prescription among 52 patients after the audit. A postoperative reduction of prescriptions for antibiotics was not intended.

Fig 3. Accuracy of the prescription before (pre) and after (post) intervention, by studies assessing this outcome parameter (measured as a percentage of adherence to guidelines).

Note: Palmer et al. (2001), Steed and Gibson (1997), Raunair et al. (2012) and Thomas and Hill (1997) did not report this outcome measure. Seager et al. (2006) provided information as odds ratios. (a) p<0.001. (b) p = 0.01. (c) p-values were not reported within this publication.

Fig 4. Risk of bias within included studies.

Note: Low risk of bias is indicated by green colour, moderate risk of bias by yellow colour and high risk of bias by red colour. The question mark indicates an unknown risk.

Fig 5. Risk of bias across included studies.

Note: * Allocation concealment and sequence generation apply only to RCTs (Seager et al. and Elouafkaoui et al.) and are not applicable to pre-post studies. RCT = randomised controlled trial.