BACKGROUND Acute pyelonephritis often leaves children with permanent renal scarring.
AIMS To compare the prevalence of scarring following initial treatment with antibiotics administered intravenously for 10 or three days.
METHODS In a prospective two centre trial, 220 patients aged 3 months to 16 years with positive urine culture and acute renal lesions on initial DMSA scintigraphy, were randomly assigned to receive intravenous ceftriaxone (50 mg/kg once daily) for 10 or three days, followed by oral cefixime (4 mg/kg twice daily) to complete a 15 day course. After three months, scintigraphy was repeated in order to diagnose renal scars.
RESULTS Renal scarring developed in 33% of the 110 children in the 10 day intravenous group and 36% of the 110 children in the three day group. Children older than 1 year had more renal scarring than infants (42% (54/129) and 24% (22/91), respectively). After adjustment for age, sex, duration of fever before treatment, degree of inflammation, presence of vesicoureteric reflux, and the patients' recruitment centres, there was no significant difference between the two treatments on renal scarring. During follow up, 15 children had recurrence of urinary infection with no significant difference between the two treatment groups.
CONCLUSION In children with acute pyelonephritis, initial intravenous treatment for 10 days, compared with three days, does not significantly reduce the development of renal scarring.
- urinary tract infection
- radionuclide imaging
Statistics from Altmetric.com
Renal scarring is a frequent sequel of pyelonephritis in children.1-3 Appropriate antibiotics are used to eradicate urinary tract infections. Epidemiological studies4 and experiments on animals5 6 have shown that antibiotics were equally effective in preventing or limiting the progression of permanent renal lesions. In children with acute pyelonephritis the administration of intravenous antibiotics is often recommended. However, there is controversy regarding the duration of intravenous treatment, ranging from a few days7-9 to more than 10–15 days.10 11 A potential advantage of prolonged parenteral treatment might be that it would reduce the frequency of development of renal scarring.
We therefore carried out a randomised study to compare the effects of initial parenteral treatment for 10 days or for three days on the incidence of renal scarring. To diagnose the acute lesions and renal scars we used scintigraphy with technetium-99m labelled dimercaptosuccinic acid (DMSA), currently one of the most reliable methods of detecting renal parenchymal damage.12-14
This randomised two centre study was carried out between June 1995 and April 1999. The study protocol was approved by the local ethics committees of the two Swiss centres, Geneva and Zurich, which recruited the patients. Written informed consent was obtained from the patients' parents before enrolment. Children with acute pyelonephritis were treated with antibiotics administered either intravenously for 10 days then orally for five days, or intravenously for three days then orally for 12 days. The primary end point was the development of renal scarring.
Eligible children were those aged between 3 months and 16 years with probable acute pyelonephritis, hospitalised in the department of paediatrics in the Cantonal University Hospital of Geneva or in the University Children's Hospital of Zurich. Exclusion criteria were: age less than 3 months (in this age group pyelonephritis is often associated with bacteraemia or sepsis15; in our institutions these infections are therefore treated with parenteral antibiotics for more than three days); a history of abnormalities of the urinary tract; and hypersensitivity to cephalosporins.
Laboratory tests on admission included blood count, C reactive protein, blood cultures, urinary dipstick or urinalysis, and urine culture. Urine samples were collected by suprapubic puncture or in sterile bags from the younger children, and clean voided midstream from the older children. Diagnosis of acute pyelonephritis was considered probable in children with an abnormal urinary dipstick test (leucocyte esterase ⩾1+, or nitrite positive) or urinalysis (pyuria with at least 10 white blood cells per high power field in centrifuged urine, and bacteriuria with any bacteria per high power field on an unstained specimen of urinary sediment) and who had at least one of the following clinical or biological signs: fever with rectal temperature of 38°C or higher; abdominal or flank pain in children old enough to report pain accurately; general, non-specific signs such as irritability, vomiting, diarrhoea, or feeding problems in infants; or C reactive protein concentrations above 10 mg/l.
For children who had a positive urine culture (⩾104colony forming units/ml for voided urine, any colony forming units/ml for suprapubic collection), urine culture and C reactive protein were repeated on day 3–4 of the treatment, and a renal scintigraphy with DMSA and ultrasonography were performed. At this stage, children whose scintigraphy showed signs of acute lesions were finally enrolled in the study. For practical reasons, the patients were not randomised at the time when they met both criteria for final enrolment (a positive initial urine culture and a first scintigraphy showing signs of acute pyelonephritis). As the results of urine cultures and scintigraphy (carried out only on the working days of the week) were not consistently available on the third day of treatment (that is, before the antibiotic treatment was carried out either intravenously or orally), patients were randomised at the time of admission by using blocks of 20 sealed opaque envelopes containing an equal number of assignments for the two antibiotic treatments, with stratification for the centre. Investigations were completed six weeks after infection by voiding cystourethrography to detect vesicoureteric reflux and, after an interval of three months, by a second DMSA scintigraphy to follow the evolution of renal lesions. The three month interval between the first and second scintigraphy was chosen according to a study by Goldraich et al.12
During the three month follow up, recurrent urinary tract infection constituted a secondary end point. Specimens for urine culture were obtained at the time of the voiding cystourethrography, and when the children had fever or symptoms of urinary tract infection. All patients who completed the study were reviewed at the time of their voiding cystourethrography and at the end of the follow up period. Parents were asked whether their child had had fever, recurrences of urinary tract infection, side effects of treatment or prophylaxis, and whether the prophylaxis had been given as prescribed.
TREATMENT AND ANTIBIOTIC PROPHYLAXIS
The antibiotics were administered intravenously for 10 days or for three days, then orally to 15 days. Taking account of local epidemiological data, the antibiotics chosen were ceftriaxone (50 mg/kg once daily; Rocéphine, Roche) for the intravenous treatment, and cefixime (4 mg/kg twice daily; Cephoral, Merck) for the oral treatment. The therapy was given immediately after a urine sample had been taken.
At the end of treatment, antibiotic prophylaxis with co-trimoxazole (1–2 mg trimethoprim, 5–10 mg sulphamethoxazole per kg once daily) was introduced up to the time of the voiding cystourethrography. This was then stopped for those children who had no vesicoureteric reflux and no other malformations of the urinary tract shown by ultrasonography.
RENAL DMSA SCINTIGRAPHY
Scintigraphy was done with intravenous injection of DMSA (CIS Bio, Medipro, Switzerland) labelled with 99mTc (3.7 MBq per kg; minimum 18.5 MBq, maximum 185 MBq). Three hours after injection, six views (one posterior, two posterior oblique, one anterior, and two anterior oblique projections) were obtained with a three heads gamma camera (Toshiba GCA-9300 A/HG, Toshiba Medical Systems Inc., Japan) or, at the minimum two views (one posterior and one anterior) with a two heads gamma camera (Body Scan, Siemens, Erlanger, Germany).
All scintigrams were independently interpreted by two experienced paediatric radiologists who were unaware of the treatment assigned to the patients. Interpretation was based on the standard criteria previously defined by Patel et al.16 The progression of renal lesions was assessed by topographic analysis of each lesion. We defined renal scars as persistent changes in the same location, complete or partial reversible lesions as complete or partial resolution of changes that had been observed on first scintigraphic examination, and new lesions as lesions not present during the acute phase of urinary tract infection.
The relative size of each lesion was estimated by relating the surface of the lesion to the surface of the kidney, in the view of the scintigram where the size of the lesion was most pronounced (for an atrophic kidney, the contralateral kidney was used as a reference). The lesions were considered small if the relation to the surfaces was less than 10%; moderate if this relation was 10–30%; and large if it was 30% or greater.
As scintigraphy with DMSA does not always distinguish between pre-existing renal scar and an acute lesion,17 we analysed the results of two subgroups of children to reduce the risk of overestimating the incidence of renal lesions at the time of the second scintigraphy. These were composed of children who had presented with their first documented episode of urinary tract infection, and those whose renal lesions had partially or totally regressed between the first and second scintigraphy (regression was interpreted as the sign of a recent lesion, and in the case of partially regressive lesions, the persistent part of the lesion as a new renal scar).
On the basis of previous studies, we estimated that the minimum incidence of renal scarring was approximately 35%.1 2 In order to detect a difference of 20% between the rate of renal scarring of the two treatment groups (from 35% to 55%) with a power of 80% and a value α of 0.05 (two tailed), the sample size should be 106 children completing the study in each group. The proportions were compared by χ2 test or Fisher's exact test. The differences between the proportions and 95% exact confidence intervals were calculated.
A logistic regression model was used to assess the effect of treatment on renal scarring after adjustment for the following potential prognostic factors: age, sex, duration of fever before treatment, degree of inflammation at start of treatment (C reactive protein), presence or absence of vesicoureteric reflux, and the patients' recruitment centre. Age was modelled as a continuous or binary variable (⩽1 year and >1 year). All reported p values are two tailed.
After initial evaluation, 206 of the 435 children randomised (100 in the 10 day intravenous group and 106 in the three day intravenous group) did not fulfil the criteria for final enrolment in the study: 84 children had a negative initial urine culture and in 122 the first renal scintigraphy showed no signs of acute pyelonephritis.
For the 229 patients finally included in the study, the baseline characteristics of the two treatment groups were similar, except for age, where the children in the 10 day intravenous group were significantly younger than those in the three day intravenous group (median age 1 year and 2.4 years, respectively; p = 0.002 by the Mann–Whitney test; table 1). There was a similar difference in age when the comparison was made with all the randomised children (median age 1.2 years versus 1.9 years; p = 0.01 by the Mann–Whitney test). Nine patients finally included (eight in the 10 day intravenous group and one in the three day intravenous group) did not complete the study at their parents' request, and were therefore excluded from the analysis of results. Seven children who completed the study (four in the 10 day intravenous group and three in the three day intravenous group), unintentionally deviated from the protocol: the duration of intravenous treatment was prolonged in two children who had positive blood cultures (one in the 10 day intravenous group and one in the three day intravenous group); antibiotics other than those prescribed for the study were given to one child who showed poor clinical evolution and to one child who had repeated skin rashes; the initial urine culture of two children was lost, and was negative in one child after an antibiotic dose was administered by mistake before samples were taken.
Except for the child who had skin rashes, both treatments were well tolerated. All urinary cultures taken on the third or fourth day of treatment were negative. Defervescence occurred on an average 34 hours after therapy was begun in both groups of children with acute pyelonephritis.
Seventy six of 220 children (35%) who completed the study developed renal scarring. In the three day intravenous group, 40 of the 110 children (36%) had renal scars, compared with 36 of the 110 children (33%) in the 10 day intravenous group (p = 0.57). Similarly, in the two subgroups defined to reduce the risk of confusion with pre-existing renal scars, there were no significant differences between the rates of renal scarring according to treatment group (table2).
In univariate analyses, age and sex were the two factors that had a significant effect on the development of renal scarring. Children aged greater than 1 year developed renal sequelae more frequently than infants of 1 year of age or less (42% (54/129) and 24% (22/91), respectively; p = 0.007); as did girls compared to boys (39% (67/171) and 18% (9/49) respectively; p = 0.007). In contrast, the duration of fever before treatment, the degree of inflammation at the beginning of treatment (C reactive protein), the presence or absence of vesicoureteric reflux, or the patient recruitment centre had no significant effect on the incidence of renal scarring (p > 0.3). After adjustment for age (as a continuous or binary variable), sex, duration of fever before treatment, degree of inflammation, presence or absence of vesicoureteric reflux, and the patients' recruitment centres, the difference between the two treatments on renal scarring remains non-significant (p ⩾ 0.84).
We calculated size of the renal lesions in 156 children (75 in the 10 day intravenous group, 81 in the three day intravenous group), where the six views for their first and second renal scintigraphy were available (table 3). Evolution to renal scarring was more frequent as the size of the acute renal lesions increased: thus, 9% of the small acute lesions, 26% of the moderate, and 46% of the large acute lesions evolved into scars. In taking the largest size category among children who had acute lesions of different sizes, evolution to renal scarring involved 14% of children who had small acute lesions, 33% of children who had moderately large acute lesions, and 56% of children who had large acute lesions. There was no significant interaction between the size of the initial renal lesions and the treatment in relation to the rate of renal scarring (p = 0.10 in the homogeneity test).
RECURRENCE OF URINARY TRACT INFECTION
In the course of the three month follow up, 15 of the 220 children (7%) had at least one recurrence of urinary tract infection (table 4). Bacteriological eradication was achieved on the third day of treatment in 14 of these children; for one child the urine sample for culture was not obtained. In two children, an asymptomatic urinary tract infection was diagnosed at the time of voiding cystourethrography. Eight children on antibiotic prophylaxis developed a urinary tract infection (in one child the organism was resistant to the antibiotic used for prophylaxis), six of whom had an abnormality of the urinary tract (table 4). DMSA scintigraphy was performed in seven children who had a recurrence of febrile urinary tract infection: two children had new acute renal lesions (both were in the three day intravenous group), which later completely regressed in one child and partially regressed in the other.
The therapeutic modalities used to treat renal infection effectively and prevent renal sequelae in children are subject to controversy. We have found that antibiotics initially administered intravenously for 10 days do not confer a significant advantage in reducing the incidence of renal scarring when compared with an initial intravenous administration of three days.
To diagnose the renal lesions, we used DMSA scintigraphy, one of the most sensitive methods in detecting damage to renal tissue.12-14 However, the rate of renal scarring resulting from recent episodes of pyelonephritis may be overestimated by this method, as it can be difficult to clearly distinguish between pre-existing renal scars and new lesions.17 In children, pre-existing renal scarring may be a result of past renal infections, undiagnosed urinary tract infection,18-20 or other pathological conditions such as renal dysplasia.17 21 22In our study, patients with pre-existing renal scarring were not known. For this reason, we determined the incidence of renal scarring in two subgroups, formed respectively by children with a first documented urinary tract infection and those in whom the lesions had partially or totally regressed between the first and second scintigraphy (regression was interpreted as the sign of recent lesion). In these two subgroups, the incidence of renal scarring was 33%, no different from that of the entire group of patients (35%).
Data from previous scintigraphy investigations following acute pyelonephritis in children have documented the incidence of renal scarring to be 35–60%.1-3 However, in these studies, either the information on antibiotic treatment administered was incomplete, or treatment was not standardised. Hoberman and colleagues23 reported a particularly low incidence of renal scarring (9.6%), also diagnosed by DMSA scintigraphy, among children who had been treated for 14 days, either orally or by combined intravenous and oral treatment. The rate of renal sequelae in the two treatment groups was similar, and in their analyses none of the factors likely to influence the incidence of these lesions explained this result. However, they recruited only patients 1 to 24 months old who were not severely ill.
Among the different factors likely to have an effect on the development of renal scarring, one of the most important is the time interval between the onset of renal infection and the beginning of appropriate treatment.2 4 24 On this basis, experimental studies have shown that damage to renal tissue was a direct consequence of the acute inflammatory reaction, the duration of this reaction determining the severity and extent of the lesions.5 6 But in our study, neither the duration of fever before treatment, nor the degree of inflammation were significantly associated with the development of renal scarring; this was also reported by Jakobssonet al.1 Our results have shown that age has an effect on the development of renal scarring. Compared with infants, children aged greater than 1 year developed renal scarring more often, suggesting that the vulnerability of renal tissue varies with age. Although this difference of susceptibility has been reported previously,1 23 the reason is not known. In the univariate analyses, we found that sex had a significant effect on the rate of renal scarring, but this result should be interpreted with caution, bearing in mind the small number of boys aged over 1 year who had acute pyelonephritis. Vesicoureteric reflux is the most common abnormality of the urinary tract in children.25 Hellstromet al found that a minority of children with renal scarring after a urinary tract infection did not have reflux.26 As with other data,3 27 28 our results do not confirm the association between reflux and renal scarring, reflux being identified in only 39% of children with renal scarring. In the logistic regression models that we used, after adjustment for these different potential prognostic factors, the difference between the two treatments on renal scarring remains non-significant (p ⩾ 0.84).
Oral antibiotics alone as a treatment for pyelonephritis is an attractive alternative to combined intravenous and oral treatments. This avoids hospitalisation of children, thus reducing health care costs, but increases the risks of renal sequelae associated with “inadequate” treatment,4 whether this is caused by poor compliance or vomiting (36% of the children in our study who had acute pyelonephritis had vomited). Pyelonephritis, especially in infants, is frequently associated with bacteraemia.15 Oral treatment of bacteraemia, even when compared with parenteral treatment, limited to one single intramuscular injection of antibiotic, significantly exposes the child to serious infectious complications, such as meningitis, sepsis, or pneumonia.29 Under these conditions the value of giving oral treatment to young patients with acute pyelonephritis is debatable.
Following the episode of acute pyelonephritis, a high percentage of the children (35%) developed renal sequelae, particularly those aged over 1 year. This response to treatment suggests that effectiveness in preventing scarring is limited, despite antibiotics administered intravenously, at least initially. One possible strategy to improve the treatment of renal infections would be to act directly on the acute inflammation, as shown in studies carried out on animals.30 However, the modality of treatment using anti-inflammatory drugs together with antibiotics has yet to be defined and tested in children.