Effectiveness of extended- versus normal-release nitrofurantoin for cystitis: an instrumental variable analysis

Background: It is unknown whether nitrofurantoin 50 mg normal-release every 6 h (NF50) and nitrofurantoin 100 mg extended-release every 12 h (NF100) are equally effective for treating cystitis in primary care. In the Netherlands, GP prescription of either option largely depends on pharmacy procurement, rather than on patient-related factors. Methods: GP data between January 2013 and July 2018 were retrospectively collected. Inclusion criteria were the use of nitrofurantoin for uncomplicated cystitis, complicated cystitis or cystitis in pregnancy. Criteria for early and late failure were a second antibiotic prescription for cystitis or pyelonephritis within 14 and 28 days post-prescription, respectively. Crude and confounder-adjusted (CA) risk differences (RDs) were estimated using linear regression. Instrumental variable analysis and CA instrumental variable analysis used GP practice proportion of NF50 versus NF100 use as the instrumental variable. Results: For uncomplicated cystitis (n=46855), treatment with NF50 and NF100 resulted in late failure in 9.7% and 9.6%, respectively. The CA RD, instrumental variable RD and CA instrumental variable RD were 0.2% (95% CI=-0.5 to 0.8), -0.7% (95% CI=-1.7 to 0.3) and 0.0% (95% CI=-0.9 to 1.0), respectively. In complicated cystitis (n=10767), late failure occurred in 10.9% and 11.1% after using NF50 and NF100, respectively [CA RD=0.5% (95% CI=-1.2 to 1.8), instrumental variable RD=-0.8% (95% CI=-3.4 to 1.8) and CA instrumental variable RD=-0.3% (95% CI=-3.0 to 2.4)]. For cystitis in pregnancy (n=1087), NF50 and NF100 resulted in late failure in 13.4% and 7.8%, respectively [CA RD=-5.4% (95% CI=-10.0 to -1.4), instrumental variable RD=-8.9% (95% CI=-16.0 to -1.8) and CA instrumental variable RD=-8.9% (95% CI=-16.0 to -1.7)]. No differences were observed in early failure. Conclusions: In patients with cystitis in pregnancy, NF100 was associated with a lower incidence of late clinical failure compared with NF50. We found no differences in clinical failure between NF50 and NF100 for uncomplicated and complicated cystitis.