OBJECTIVE: To estimate and examine the incidence and determinants of initiation of migraine-prophylactic therapy as well as the corresponding drug of choice over a period of 5 years following the use of specific abortive migraine drugs.METHODS: Byccessing data from a large prescription database, an identification of patients treated with ergotamine or a triptan from 1 January 1994 to 31 December 1998 was made. The cumulative incidence of initiation of migraine-prophylactic drugs (beta-blockers, serotonin antagonists, specific calcium antagonists, amitriptyline, clonidine and valproic acid) was estimated in patients following the use of ergotamine or a triptan. An assessment of the migraine-prophylactic drug of first choice was also performed. Afew baseline determinants were analysed to highlight a possible association with the initiation of prophylactic therapy: age, gender, type of abortive migraine drug use and year of prophylaxis. Additional determinants included the analysis of drug-utilisation patterns, such as the consumption and switch patterns of abortive migraine drug use as well as co-medication use prior to prophylaxis. For this particular analysis a reference group (patients not having commenced prophylaxis) was selected from theinitial study population.RESULTS: After having satisfied eligibility criteria, a total of 3999 first-time users of ergotamine and triptans were included of whom 479 (12%) had initiated migraine-prophylactic therapy. This corresponded to an incidence density of 6.0 per 100 person-years and was highest for patients younger than 45 years and for multiple abortive migraine drug users. The incidence fell considerably from 12.0 person-years in 1994 to 5.1 person-years in 1998. More than half of the patients had been prescribed a beta-blocker as the migraine-prophylactic drug of first choice by both general practitioners and neurologists. The use of antidepressants and/or benzodiazepines and oral contraceptives was significantly higher in patients starting prophylaxis compared with those who did not. The consumption of abortive migraine drug use (4.0 defined daily doses per month vs 3.7 defined daily doses per month), and switch patterns (27.1% vs 30.9%) were similar for patients starting and not starting prophylaxis.CONCLUSION: The overall incidence of initiation of migraine-prophylactic therapy following the use of abortive migraine analgesics was 6.0% per year and fell considerably during 5 years of the study. Beta-blockers were the migraine-prophylacticdrugs of first choice for general practitioners and neurologists. In our study we could not determine any factors clearly associated with the initiation of migraine prophylaxis besides prior use of antidepressants and benzodiazepines. A future assessment of the usage patterns of migraine-prophylactic drugs may provide detailed information concerning the effectiveness and tolerability.