Background:
From November 2014 to November 2015, an experiment in French community pharmacies replaced traditional pre-packed boxes by per-unit dispensing of pills in the exact numbers prescribed, for 14 antibiotics.

Methods:
A cluster randomised control trial was carried out in 100 pharmacies. 75 pharmacies counted out the medication by units (experimental group), the other 25 providing the treatment in the existing pharmaceutical company boxes (control group). Data on patients under the two arms were compared to assess the environmental, economic and health effects of this change in drug dispensing. In particular, adherence was measured indirectly by comparing the number of pills left at the end of the prescribed treatment.

Results:
Out of the 1185 patients included during 3 sessions of 4 consecutive weeks each, 907 patients experimented the personalized delivery and 278 were assigned to the control group, consistent with a 1/3 randomization-rate at the pharmacy level. 80% of eligible patients approved of the per-unit dispensing of their treatment. The initial packaging of the drugs did not match with the prescription in 60% of cases and per-unit dispensing reduced by 10% the number of pills supplied. 13.1% of patients declared that they threw away pills residuals instead of recycling—no differences between groups. Finally, per-unit dispensing appeared to improve adherence to antibiotic treatment (marginal effect 0.21, IC 95, 0.14–0.28).

Conclusions:
Supplying antibiotics per unit is not only beneficial in terms of a reduced number of pills to reimburse or for the environment (less pills wasted and non-recycled), but also has a positive and unexpected impact on adherence to treatment, and thus on both individual and public health.

This study evaluated the impact of infectious disease (ID) specialist referrals on outcomes in a tertiary hospital in France. This study tackled methodological constraints (selection bias, endogeneity) using instrumental variables (IV) methods in order to obtain a quasi-experimental design. In addition, we investigated whether certain characteristics of patients have a bearing on the impact of the intervention. We used the payments database and ID department files to obtain data for adults admitted with an ID diagnosis in the North Hospital, Marseille from 2012 to 2014. Comparable cohorts were obtained using coarsened exact matching and analysed using IV models. Mortality, readmissions, cost (payer perspective) and length of stay (LoS) were analysed. We recorded 15,393 (85.97%) stays, of which 2,159 (14.03%) benefited from IDP consultations. The intervention was seen to significantly lower the risk of inpatient mortality (marginal effect (M.E) = –19.06%) and cost of stay (average treatment effect (ATE) = – €5,573.39). The intervention group was seen to have a longer LoS (ATE = +4.95 days). The intervention conferred a higher reduction in mortality and cost for stays that experienced ICU care (mortality: odds ratio (OR) =0.09, M.E cost = –8,328.84 €) or had a higher severity of illness (mortality: OR=0.35, M.E cost = –1,331.92 €) and for patients aged between 50 and 65 years (mortality: OR=0.28, M.E cost = -874.78 €). This study shows that ID referrals are associated with lower risk of inpatient mortality and cost of stay, especially when targeted to certain subgroups.

-

Background:
Recently, French policymakers have tried to improve care at the end-of-life, by improving access to community-based palliative care, particularly for patients with cancer and neurological diseases. If effective, these efforts should reduce the proportion of such patients who die in the hospital. In light of these policies, we sought to determine the effectiveness of these efforts on reducing inpatient deaths by conducting a retrospective, observational analysis of patients aged 65 and older who were admitted to hospitals in France between 2010 and 2013 for 1 of 3 non-surgical conditions.

Methods:
We calculated department-specific age- and sex-adjusted inpatient death rates for 3 types of non-surgical admissions and modeled expected number of inpatient deaths had their rates for patients with cancer or neurological disease tracked those of patients with non-cancer non-neurological diseases.

Results:
We found that patients admitted with a cancer diagnosis experienced 20,394 (13.0%) fewer inpatient deaths that expected had non-surgical cancer diagnosis admission rates tracked those of non-surgical non-cancer and non-neurological admission rates; patients admitted with a primary neurological disease diagnosis experienced 513 (4.5%) fewer inpatient deaths than expected. During the study period, observed-to-expected inpatient deaths fell more dramatically and consistently for patients admitted with cancer diagnoses than for those admitted with neurological diseases. Observed-to-expected ratios fell least in departments that were on the periphery of the French mainland.

Conclusions:
Our findings suggest that, in France, efforts to reduce inpatient death rates among patients with cancer or neurological disease diagnoses appear to be effective. However, their effectiveness varies geographically, suggesting that targeted efforts to improve lower performing departments may generate substantial performance improvements.