The severe asthma is a major source of expenses in term of public health, while it concerns no more than 5 % of the asthmatics. The expenses is direct (medicines, hospitalizations, care) but especially indirect (absenteeisms, etc.). The forward-looking follow-up of cohort of more than 500 severe asthmatic patients multicentrique in an already widely established cohort (COBRA, at present in Visit 9 (one every 6 months) is an once-in-a-lifetime opportunity, coupled with the data of the CPAM, to identify well the evolution in time of a real medical economic variable. The possibility of dynamic follow-up of the expenses compared to the medical data offers perspectives of evaluation cost-efficiency of the informed therapeutic procedures. It is possible to couple in a forward-looking and dynamic way the data of health stemming from a cohort with the economic data stemming from the CPAM. This variable included in a Cluster's algorithm has to allow to identify the interventions the more and the less cost effective. The main objective of this study is to realize a cost estimate of care of the severe asthma. The variation of the costs will be also studied. The recruited patients are patients already included in the cohort COBRA in the centers of Marseille, Montpelier or Nice, classified GINA 4 and agreeing to participate.

The following direct medical costs will be collected according to the point of view of the Social Security on a duration of 2 years:

• Hospitalizations (valuation from the GHS + medicines except GHS)· Consultations (valuation AM)
• Complementary(additional) examinations except hospitalization (valuation AM)
• Pharmacy (valuation AM)
• Transport (valuation AM)The estimate of average costs will be made with a reliable interval for 95 %.

The average cost by patient of the care will be calculated for every year of the follow-up. Relationship individual cost-efficiency: by analogy with the studies cost-efficiency comparative clauses, the investigators shall calculate a relationship cost-efficiency at the individual level as follows: (cost-year2 - cost-year1) / (efficiency-year2 - efficiency-year 1). Differences in these relationships will be looked for between various groups by tests of comparisons of averages.The construction of patients' clusters will be made according to the algorithm of the study SARP but also by using techniques statistics of classification non supervising to define our own algorithm. Variables used on our algorithm will include the relationship cost efficiency individual in addition to the clinical variables. The temporal stability of the relationship cost efficiency will be analyzed. The risk factors of a modification of this relationship will be looked for.

Inclusion Criteria:

• Patients included in the observationnal cohort COBRA for at least 6 months
• Patient members in one of the partner health insurance funds

Exclusion Criteria:

• The patient refusing his participation to the research

The recruited patients are patients already included in the observationnelle cohort COBRA in the centers of Marseille, Montpelier or Nice, classified GINA 4 and agreeing to participate.