Abstract

The definition of acute pyelonephritis is controversial. There are two contrasting approaches: (1) acute pyelonephritis is a severe infectious disease involving the kidney parenchyma, and specific imaging techniques are required for diagnosis; (2) acute pyelonephritis is a urinary tract infection, and diagnosis and therapy follow simplified clinical and laboratory pathways. In this study, recent randomized controlled trials (RCTs) were systematically reviewed and the diagnostic and therapeutic approaches to acute ‘uncomplicated’ pyelonephritis were analysed. Medline, Embase, Cochrane Central Register of Controlled Trials (CCTR) and Chinal were searched employing Mesh, Emtree and free terms on ‘pyelonephritis’. Limits included human, period (1995–2004), and trials-reviews (where available). In total, 904 references and 175 full-text were retrieved; 29 were pertinent RCTs. Seven RCTs were added from reference lists (indexed on urinary tract infections). Imaging examinations were performed in 11 of 14 studies on children (diagnostic requisite in two) and in two studies on adults; scarring was not analysed in adults. Clinical definitions varied widely (fever >37.8 to >39 °C, culture titres 10⁴ >10⁵). Studies on adults were limited to short-term end-points (microbiological sterilization, clinical improvement). Duration of therapy was 4–20 days. The trend was towards shorter periods of therapy, mainly on an outpatient basis; intravenous therapy, if performed, was usually limited to the first 1–3 days. For acute uncomplicated pyelonephritis, the tendency is towards 2 weeks of mainly oral antibiotic therapy. However, the recent literature on adults does not discriminate among different upper urinary tract infections nor does it provide data on renal scarring. While cost constraints point towards short-term therapies, further studies are needed to assess the prevalence and long-term effect of kidney scars.

Keywords: Acute pyelonephritis; Evidence-based medicine; Randomized controlled trials; Antibiotic therapy; Renal scars

Article Outline

1. Introduction

2. Materials and methods

2.1. Search strategy

2.2. Selection of papers

2.3. Data analysis

3. Results

3.1. Retrieving the evidence

3.2. Diagnostic definitions and inclusion criteria

3.3. Imaging data

3.4. Exclusion criteria

3.5. Duration of follow-up

3.6. Therapeutic protocols

3.7. Definitions of the outcomes

3.8. Success rate and failure outcomes

4. Discussion

References

1. Introduction

Almost every ‘classic’ nephrology textbook defines acute pyelonephritis as a severe infectious disease involving the pelvis, calices and kidney parenchyma [1], [2] and [3]. According to this view, the life-threatening risks of the disease are due to Gram-negative sepsis, occurring either during the acute phase or as a result of kidney abscesses, and that long-term sequelae are a result of the formation of parenchymal kidney scars at the sites of infectious foci [1], [2] and [3].

This interpretation has recently been challenged, owing to the pressure of cost constraints and the tendency to reduce hospitalization times. As highlighted by Talner et al. about 10 years ago, there was no agreement on the terminology nor, as a consequence, on diagnostic definitions [4]. This situation may be one of the reasons for the lack of conformity on therapeutic approaches [4].

Interest in the demonstration of parenchymal involvement is due to the postulate that renal scars could have a detrimental effect on renal function and play a role in the pathogenesis of secondary hypertension [5], [6], [7], [8] and [9]. However, because pyelonephritis is typically a disease of the young, long-term data on the pathogenetic effect of pyelonephritic scars are lacking.

The literature on acute pyelonephritis shows a progressive change from older therapeutic schemes; that is, up to 6 weeks of i.v. antibiotic therapy [10] and [11], to shorter courses, mainly on the basis of outpatient treatment and favouring oral antibiotic therapy.

Furthermore, the literature published to date shows that different diagnostic and treatment pathways are followed for adults compared with children. In the latter, diagnosis is mainly based on the demonstration of infectious foci in the kidney parenchyma. Consequently, the avoidance of renal scars is usually a main therapeutic goal, and imaging data are analysed during follow-up [12], [13], [14], [15], [16], [17] and [18]. In adults, diagnosis is usually based solely on clinical grounds, thus failing to differentiate between pyelitis (without parenchymal kidney involvement) and pyelonephritis (with parenchymal involvement), and the therapeutic response is mainly defined according to short/medium-term microbiological data [19], [20], [21] and [22].

These different definitions can affect the different therapeutic protocols used in clinical practice. Therefore, the aim of the present study was to systematically review the therapeutic approaches to acute pyelonephritis and to discuss them in the light of the diagnostic criteria, outcome definitions and enrolment–exclusion criteria employed in randomized controlled trials (RCT) of antibiotic treatment, as indexed in the main medical databanks (Medline, Embase, Chinal and the Cochrane Central Register of Controlled trials (CCTR)) in the past 10 years (1994–2004). This time interval was chosen in an attempt to reach a balance between retrieving an adequate number of papers and selecting only the most recent studies, given the rapid evolution of antibiotic therapy in recent years.

2. Materials and methods

2.1. Search strategy

The search was performed separately on Medline, Embase and Chinal, the world's largest databanks, and on CCTR (Ovid). The search was deliberately broad to increase sensitivity, according to the guidelines of the Cochrane Collaboration [23]. Acute pyelonephritis, lobar nephronia and acute upper urinary tract infections were initially employed as Mesh, Emtree or free terms. However, as the nomenclature of acute pyelonephritis is not universally accepted, the simple term ‘pyelonephritis’ was finally chosen. The search was limited to ‘human’, studies published in the past decade (1994–2004), and to RCTs or clinical trials. Review studies were also retrieved to allow screening for references not identified with previous search strategies.

2.2. Selection of papers

The search was refined by manual sieving of titles and abstracts, selecting those dealing with the treatment of ‘uncomplicated’ or ‘primary’ acute pyelonephritis (i.e. pyelonephritis occurring in the absence of predisposing factors), but excluding pyelonephritis during pregnancy. This step was performed in duplicate by Piccoli and Mesiano. The reference lists of the selected papers, of the main clinical texts and of relevant secondary studies (guidelines, reviews and systematic reviews), were also screened for papers not found with our search strategy. Discrepancies were resolved by discussion. The full papers considered potentially relevant were retrieved and efforts were made to overcome language barriers. The data were extracted in duplicate by Piccoli and Consiglio.

2.3. Data analysis

The analysis focused on the definition of diagnostic criteria, enrolment and exclusion criteria, duration of therapy, and type of outcome. Efficacy data were also gathered. The decision to perform a narrative or a systematic meta-analytical review was subordinated to the analysis of the type and quality of evidence retrieved. A descriptive, narrative review was planned in the case of insufficient data or high heterogeneity of data, and particular attention was paid to the ‘semantic’ analysis of the different definitions, from diagnostic criteria to outcomes.

3. Results

3.1. Retrieving the evidence

The broad search strategy retrieved a large number of titles and abstracts: 403 titles and abstracts on Medline, 255 on Embase, 124 on Chinal and 122 on the Cochrane Database of Systematic Reviews (CDSR), the American College of Physicians (ACP) Journal Club, the Centers for Reviews and Dissemination Database of Abstracts of Reviews of Effects (DARE), and the CCTR (Fig. 1).

Of the 175 full papers obtained, 29 were selected for inclusion. An analysis of the reference lists allowed us to add seven more papers, presumably not identified with the previous search because of the choice of ‘urinary tract infection(s)’ (UTI) as index and free terms [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59] and [60].

All selected papers reported results of a RCT of acute ‘uncomplicated’ or primary pyelonephritis. However, several papers also included ‘complicated’ UTI, and one paper misreported acute primary pyelonephritis under the title of ‘complicated’ UTI [56]. We selected only those papers in which some information on acute pyelonephritis (at least a clinical definition and a description of the outcome) was available separately. Two papers reported on different trials with different diseases; one reported data from industry archives [48, Eli Lilly and Company archives]. Since these papers also gave relevant information on the definitions of the diseases and on the outcomes, they were included in the present analysis.

Of the selected papers, 22 related to adults, and 14 to children. Twenty papers came from European countries, seven from the USA, two from Asia, two from Australia/New Zealand, two from Turkey, and three collected information from various continents.

3.2. Diagnostic definitions and inclusion criteria

Table 1 shows the diagnostic definitions and the inclusion criteria of the selected studies. The main clinical data employed for clinical diagnosis, both in adults and children, were fever, bacteriuria–leucocyturia and urinary symptoms–flank pain. However, the definitions were heterogeneous for all items.

Fever was a cardinal symptom; it was defined in 26 of 36 studies of adults, ranging from >37.4 °C (axillary) to >39 °C (rectal), and also included complex definitions such as ‘>38 °C (oral) or >39 °C (rectal) or history of fever’ (Table 1).

Among the other symptoms, the presence of flank pain or costovertebral angle tenderness was explicitly required in 25 of 36 studies, although it was not included in 11 studies. Conversely, flank pain and/or loin or costovertebral angle tenderness was the only clinical symptom required in 11 of 36 studies.

‘Urinary syndrome’ was defined as different combinations of dysuria, urgency, frequency or lower abdominal pain, and was required for diagnosis in only eight of 22 studies of adults and in five of 14 studies of children. Studies of children either defined a broader category of symptoms according to age, such as failure to thrive, or did not include specific diagnostic clinical criteria.

Laboratory parameters required for diagnosis and enrolment in a clinical trial also varied widely. For example, the definition of bacteriuria varied from ≥10⁴ to >10⁵ CFU/mL in midstream urine. Some of the studies also included the option of suprapubic puncture (lower values or any CFU were required for inclusion). Since microbiological evaluation was a major outcome, most studies included only those patients with positive urine cultures. An intent-to-treat analysis was performed in a minority of cases (six studies).

Leucocyturia was explicitly required in most studies, with different standards and ranges. Other laboratory tests, mainly regarding acute phase reactants, were occasionally included, again with different definitions (Table 1).

3.3. Imaging data

The role of imaging techniques was different for adults compared with children. In four of 22 adult studies, renal ultrasound or other imaging tests were required for the exclusion of any secondary forms. Yet, none of the studies required the demonstration of parenchymal involvement for the diagnosis of pyelonephritis. In children, renal scintiscan was performed at least once during follow-up in 11 of 14 studies. However, only two studies required the presence of a positive scintiscan for diagnosis and enrolment (Table 1).

3.4. Exclusion criteria

Exclusion criteria varied widely. As a general rule, patients with severe pyelonephritis or underlying diseases (various definitions) were excluded. The conditions most often reported were chronic kidney disease, dialysis and transplantation, HIV infection, diabetes, and diseases that impair the immune response.

Recent antibiotic therapy for at least 1 day was an exclusion criterion in most studies (19/36). However, the criteria differed widely. For example, seven studies excluded all patients who had any kind of antibiotic therapy within 2 days, seven within 3 days, three within 7 days, one within 8 days, and one within 1 month. Ten studies specified that the antibiotic therapy had to have been taken in full doses for at least 1 day within the specified limits.

The prevalence of patients included (or excluded) in trials varied widely (5–70%) (Fig. 2). The criteria and definitions were also different. For example, the intent-to-treat analysis was indicated in some cases as the ‘clinical and safety’ analysis of all patients who started the randomization pathway, and in others as the cases who were enrolled and started the therapy, including drop-outs.

3.5. Duration of follow-up

The duration of the clinical follow-up varied according to the type of outcome chosen (<