The prevalence of acute pyelonephritis (PNA) is 60-75 000/year. They are traditionally classified as uncomplicated (60-65%) and complicated. If it is assumed that the uncomplicated PNA can be treated as outpatients with a cure rate of over 80%, the second group is very heterogeneous. Some patients are severely infected. But others, despite an older age, structural urologic abnormalities or a controlled history, have no risk factors and can be simply managed. The investigators propose to reclassify the PNA into 3 categories: uncomplicated PNA (PNA-1), the PNA of moderate severity (PNA-2), the major PNA (PNA-3) to test whether the PNA-2 can benefit from the same outpatient care that the PNA-1. The existence of biological markers of the severity of bacterial infections would further support a tailored approach. The pro-adrenomedullin (pro-ADM), successfully tested to identify severe community acquired pneumonia, is a an interesting candidate.

In Europe and North America, urinary tract infections (UTI) are the 2nd largest community-acquired infections. In the U.S., the number of annual doctor's consultations for UTI is estimated at 8 million and that of women's APN to 250,000. In Britain, an estimated number of 62 of 1000 women consult for UTI annually. Extrapolated to France, these figures would be 5-6 million annual consultations and 60 to 75 000 APN.

APN is an infection associated with urinary and pelvic and / or the renal parenchyma, marked by fever ≥ 38.5 ° C, spontaneous pain flank, pain caused to the cost vertebral angle and a positive urine dipstick (BU). Conversely, pain and fever can be reduced and 30% of PNA is as cystitis, a history of PNA is then so evocative.

Without a precise technology to detect possible renal micro-abscesses, scars of any APN, direct evidence of upper tract infection is rarely made. Positive blood cultures (30-50%of cases) are indirect evidence. In most cases, only infection of urine is affirmed by cyto-bacteriological examination (urinalysis), which results are available 48-72 hours after culturing. It is assumed that urine culture is positive when leukocyte's count is ≥ 104 cells per milliliter (GB / ml) and bacteriuria ≥ 105 colony forming unit (CFU) / ml of urine (maximum two) uropathogen germ. Therefore, diagnosis and treatment of PNA are probabilistic in the beginning, hence the interest of the BU. It is considered positive when the leukocyte count is ≥ 10 ± GB/mm3 with ± nitriturie. Done correctly, it has a positive predictive value (PPV) of 74% and NPV of 98%.

Classically there are two types of PNA, the complicated PNA and the non-complicated PNA, opposed by age, sex, severity, causative organisms and their sensitivity to antibiotics.

Uncomplicated PNA is the best defined. It occurs on a normal urinary tract in nonpregnant women, aged 15 to 65, with no systemic disease or urological surgery. It is not accompanied with by septic shock or renal abscesses. It is due to E. Coli in 90% of cases, sometimes to Klebsiella and Proteus sp, all susceptible to recommended antibiotics.

Its overall cure rate, defined as bacterial eradication and disappearance of signs and symptoms, observed in three visits over 6 weeks follow-up, is ≥ 80% treated at home or in hospital.

Other PNA, traditionally described as complicated, are not very well defined. It includes the PNA with septic shock, abscess or renal failure, the PNA in old women or in men. It also ranks the PNA occurring in urinary tract malformations, whether or not it has an impact in renal function (bifid pelvis or ureter, renal cyst). In all cases, the evolution is assumed less simple than that of PNA of young women without being unavoidably complicated. But there is no tool to measure the impact of chronic conditions on a PNA which degree of infection is not very severe.

A better understanding of the evolutionary potential of these various categories of PNA would adapt the therapeutic management. Different parameters would help this approach including the duration of antibiotic therapy, the place of care (home or hospital) and in this case, the optimal duration of hospitalization. It is recognized that the uncomplicated PNA in young women can be treated immediately at home by a single oral antibiotic for 7 days. For other categories of PNA, there generally recommend hospitalization to evaluate the effect of antibiotic therapy, which should not be less than two weeks. The benefit of this therapeutic approach is unknown.

Instead of the traditional terminology "complicated" and "uncomplicated" the investigators propose a new classification into three categories:

• non-severe or mild PNA, corresponding to the uncomplicated PNA (PNA-1)
• PNA of moderate severity, including less intense infectious syndrome which occurs on a chronic, stable and controlled disease,
• PNA with severe sepsis, septic shock or hemodynamic instability and / or immediately complicated by the fact of the presence of altered and / or progressive chronic disease with risk to decompensate (PNA-3).

In a prospective observational study in 34 French emergency services, 211 patients meeting PNA criteria were included in two weeks. They were divided into 62.5% of PNA-1, 24.6% of PNA-2 and 12.7% of PNA-3. This classification reflects the initial severity of the infection and the subsequent evolutionary potential, can distinguish a class 2. This differs from the other two by the lack of severity of the infection which it shares with the class 1 and by the existence of factors such as age or male, stable disease or abnormalities structural urological without systemic impact. The evolutionary potential risk associated with these factors, independent of the ITU, has never been measured.

The question is whether patients with a PNA-2 should be treated as a benign PNA or PNA with risk of further complications. The Infectious Diseases Society of America offers two treatment possibilities for these patients, and whose effectiveness has never been compared:

1. Treatment at home with oral monotherapy, preferably a fluoroquinolone (FQ) for 7 days, as for a PNA-1,
2. Or starting parenteral antibiotics followed by 12 to 24 hours of observation in short-term hospitalization unit (UHCD) before deciding if: return to home or prolong hospitalization.

Our hypothesis is that the prognosis of ANP is mainly related to the severity of initial infection and very secondarily to history or related conditions as they are controllable and little or no decompensated. In fact, the table being mild infection in the NAP-1 and NAP-2, they differ only by demographic factors (age, sex) or a history little influence on evolution. These two categories of PNA should logically receive the same support. To support this hypothesis, the investigators propose a controlled intervention trial in which the PNA-2 will be randomized between two treatment strategies, one of which corresponds to the management of PNA-1, to demonstrate their prognostic similarity.

In addition to this a priori classification, clinical data of the study will be used to construct a prognostic score. To this end, the investigators intend to measure the pro-adrenomedullin (pro-ADM) to all patients. Adrenomedullin, a powerful natural vasodilator,has also immunomodulatory and bactericidal properties which explain the increase OF ADM serum levels in severe systemic infections. However, the dosage is difficult because of its short half-life in blood. One of its degradation products, the pro-ADM, reflecting active ADM, is more stable and easily measurable. The dosage of pro-ADM has recently allowed the reliable assessment of individual prognosis of septic patients, particularly in the context of community-acquired pneumonia, the leading cause of death from infection. Although the pro-ADM is mainly used to predict the prognosis and that death is a rare complication of PNA community, the investigators will test its ability to distinguish between hospitalized PNA patients and outpatient. pro-ADM will be tested to improve the performance of the prognostic score.

Inclusion Criteria:

• man or woman
• age ≥ 18 years
• signed a written informed consent
• Social security coverage
• have all the necessary signs for a PNA diagnosis
• whose ranking to PNA-1, PNA or PNA-2-3 is possible

Fever ≥ 38.5 ° C

• Or the possibility of hypothermia in the single case of PNA-3
• Of a spontaneous pain of one or both sides
• Pain caused to the costovertebral angle
• And presence of leukocyturia GB/mm3 ≥ 10 with or without nitriturie
• Imaging examination at least abdominopelvic ultrasound (EAP) or CT, excluding any other possible diagnoses

Exclusion Criteria:

• Pregnant or lactating women because they are also mother-child care
• Patients refusing to provide reliable contact information, especially phone number
• In general, patients with significant cognitive impairment, no entourage, because adherence to guidelines, treatment and monitoring can be very random
• The PNA 3 emergency situation such as no time to start antibiotics ("antibiotic Emergency" = 30 to 60 minutes between arrival and administration of antibiotics) or reanimation can not be accepted, especially if a third person was required to sign the consent. The typical example is the patient with septic shock.
• Inclusion in another study,
• Any suspected nosocomial PNA, defined as any PNA appeared within 48 hours of discharge from a medical institution. The residential facilities for the frail elderly (retirement homes), and other institutions for the aged, not medicalized, are not affected by this limitation.
• All uptake within 24 hours of: ciprofloxacin, ofloxacin, levofloxacin, or ceftriaxone.

Taking antibiotics considered less active in E. coli (amoxicillin ± clavulanate, trimethoprim) or ineffective in the pna (norfloxacin), before arriving at the hospital regardless of its duration, is not an exclusion criteria. The approach should remain similar to those of other patients, and possible exclusion imposed only if the initial bacteriuria is ≤ 103 CFU / ml.