Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The strong association between childhood urinary tract infection (UTI), vesicoureteric reflux (VUR) and kidney scarring has been recognised for many years,1 2 but their relationship is inconsistent. This is in part because many infants with UTIs only have non-specific symptoms and are not diagnosed, in part because children may scar while they have VUR and subsequently outgrow it,3 and in part because some dysplastic lesions are present from birth. The relationship was also in part unexplained because some kidneys were recognised not to scar despite having VUR and recurrent UTIs. Particularly striking are children with severe bilateral VUR where one kidney is extensively damaged and the other is pristine. Ransley and Risdon’s piglet experiments in the 1970s provided a possible explanation for these variations.4
By inducing unilateral VUR surgically in piglets and then introducing a bladder infection, Ransley and Risdon showed that the combination of VUR and a UTI caused coarse renal damage which closely resembled human kidney scarring, with dead segments adjacent to unaffected ones.4 They then showed that this pattern was related to the morphology of the individual renal pyramids: segments with nipple-shaped tips did not scar, whereas flattened and compound ones did. In the latter, the refluxing infected urine entered the parenchyme via the collecting ducts (intra-renal reflux) causing inflammation and the subsequent death of that segment.5 They showed that just one brief UTI could cause permanent damage6 (the “big-bang” concept). They also showed that about 70% of children’s kidneys had similar, apparently vulnerable papillae, predominantly at the poles where human kidney scars are commonest.7 Thus, Ransley and Risdon provided a model that explained much about renal scarring in children, apparently justifying the clinical term “reflux nephropathy”.8
In the 1980s it was suggested that UTIs could occasionally initiate scarring in children up to about 10 years of age,9 but that study used intravenous urography which may not identify scars until they have been present for several years.10 The upper age was subsequently shown to be about 4 years by using dimercaptosuccinic acid (DMSA) scans which detect scars immediately they occur.11 It was widely assumed (including by the authors at the time) that as well as reflux tending to resolve with time,12 human kidneys must also mature in such a way that they lose their vulnerability to scarring by 4 years of age. However, this view has now been challenged in a way which, if sustained, will have major implications. It leads to important new conclusions about how and when human kidneys scar, and consequently about how we should manage children with VUR and with UTI. These points will be addressed after considering whether most focal defects are due to reflux nephropathy or congenital dysplasia.
ARE MOST FOCAL DEFECTS DUE TO CONGENITAL DYSPLASIA?
Reflux nephropathy is not the only cause of children having irregularly shaped, poorly functioning kidneys. Some are abnormal from before birth because of dysplasia, which may affect the development of the whole urinary tract, producing ureteric abnormalities including dilatation and VUR. There may be a genetic or obstructive cause, such as posterior urethral valves. Overall, dysplasia is commoner in boys. There is often loss of cortico-medullary differentiation on ultrasound, and there may be thinned parenchyme, which differs from the typical appearance of reflux nephropathy (which is commoner in girls), where some segments look normal while adjacent ones are not, although they are not always easily distinguished in individual cases, and indeed may coexist. Increasingly, it is being suggested that most coarse renal defects in children are congenital in nature,13 although this is necessarily speculative since few have had prior DMSA scans.
Three arguments are used to support this view. The first is that most kidney defects are detected at the child’s first imaging. However, this only rules out reflux nephropathy if it is assumed that infants are likely to have their first UTI diagnosed and treated so promptly that scarring is prevented. I will argue below that the UK primary care management of infants with UTIs does not come close to guaranteeing this.
The second argument for most defects being congenital is that attempts to prevent their development have failed. However, this has never been tested in the right subjects. To determine if babies born with VUR are at high risk of scarring with a UTI, and whether this could be prevented by specific interventions or by radical changes to the delivery of primary care, these arrangements would need to be in place before the babies presented with their first UTI. No such studies exist. Instead, conclusions have been projected from studies of older children who predominantly already have widespread scarring and are then treated with interventions such as prophylactic antibiotics or anti-reflux surgery.14–17 Despite this lack of evidence, the concept that little useful can be done by managing childhood UTI appears to be the basis for the National Institute for Health and Clinical Excellence (NICE) guideline recommendations,18 which minimise the previous diagnostic and imaging schedules.19
The third argument is that some children are reported to have scars without a history of a UTI. However, having a negative history of recognised UTIs has a very low sensitivity, as discussed below.
LESSONS FROM REFLUX NEPHROPATHY IN HUMAN TRANSPLANTS, AND IN ADULT PIGS
VUR into kidney transplants has generally been considered unimportant, despite the fact that UTIs are common after transplantation.20–22 Young children’s kidneys are seldom used for transplantation, so reflux nephropathy was not expected to occur. However, scars with typical pyelonephritic histology have been reported in transplant nephrectomies,23 and shown to be very common, and often severe, on DMSA (despite never being recognised on ultrasound).24 This appears to conform closely to Ransley and Risdon’s model for native kidneys,4 with about 70% of transplants with VUR acquiring focal scars after a UTI. Scarring can occur alarmingly quickly; two patients’ kidney transplants were extensively and permanently damaged despite immediate treatment after a single UTI of <48 h, with symptoms lasting less than 1 day.24
Two important lessons can be drawn. First, human kidneys remain vulnerable to reflux nephropathy scarring into adult life, and second scarring can sometimes occur so quickly after a UTI that prevention may be very difficult.
The possibility that these adult kidneys were different or especially vulnerable because they had been transplanted and immunosuppressed was effectively ruled out by showing that adult sows’ kidneys scar as readily as those of piglets if exposed to VUR and a UTI.25 The question, then, is not why do transplanted kidneys scar so readily, but why do older children and adults not initiate scarring with UTIs? The (depressing) conclusion we can suggest is that most children born at risk of scarring (that is with VUR and vulnerable pyramids) have already developed reflux nephropathy scars before the age of 4.26 The numbers seem approximately right, as can be seen from the following crude “guesstimate”. Around 1% of children are born with VUR,26 so if about 70% of these have compound papillae,7 0.7% should acquire kidney scars. We have demonstrated that at least 0.54% of girls in Newcastle have scars,27 which is reasonably close, given this is inevitably an underestimate. If correct, this conclusion carries huge implications. It suggests that our present medical system for managing children with UTIs is preventing few, if any, scars. Below, I consider whether it may be possible to improve on this.
WHAT WOULD BE NEEDED TO PREVENT UTIS FROM CAUSING SCARS?
If native kidneys that have VUR and are exposed to a UTI can scar as quickly as transplanted kidneys, prevention must involve very much quicker diagnosis and treatment, probably within a day or two of the first symptom. In the UK, this would inevitably require radical changes in primary care management. Let us consider what actually happens now.
First, there is indirect, but powerful evidence that general practitioners (GPs) currently fail to diagnose at least three-quarters of UTIs in infants, but there are ways in which primary care can be supported to improve on this.28 A randomised study of GPs showed that the UTI diagnosis rate was double the control level for children overall, and quadruple for infants, when the GPs received education and ongoing support from a specialist childhood UTI nurse, and were given protocol-driven direct access to investigations. Using parent-friendly urine collection techniques for babies29 and toddlers30 and initiating immediate treatment were emphasised. The higher diagnosis rates were immediate and sustained.28 If most GPs continue only to recognise a quarter of infant UTIs, there is little chance of reducing scarring rates. The advice that NICE gives about which infants to investigate could have the unintended result of reducing the diagnosis rate yet further. While listing various (mainly non-specific) symptoms, they emphasise that an infant without a fever above 38°C18 is unlikely to require urine testing. However, their recommendation referring all infants younger than 3 months with a fever of ⩾38°C to a paediatric specialist and for urgent urine microscopy and culture is to be welcomed.18
These diagnostic failures will be multiplied if nitrite testing of urine is used as a diagnostic screen, and negative samples are assumed not to have a UTI (unless they also have white cells present, when clinicians are invited to “use their clinical judgement”), as advocated by NICE.18 Having a specificity of 99% but a sensitivity just over 50%31 32 makes nitrite testing almost valueless in excluding a UTI, especially when one is clinically likely.33 Tests require a high specificity to rule diagnoses in (“SPIN”) but a high sensitivity to rule them out (“SNOUT”).34 Following this recommendation would guarantee missing half the children with UTI.
econd, treatment delays are inherent if antibiotics are not started until laboratory culture and sensitivity are known, a minimum of 2 days later. In general practice, the normal interval between consultation and microbiology confirmation is probably similar to the infant kidney scarring time. Children presenting from Thursdays to Sundays (over half the week) will have a much more delayed result, if specimens are collected at all.35 There is therefore a case for either treating on strong clinical suspicion and reviewing the decision according to the culture result (thus, aiming to prevent scarring, but balancing this against the potential to overuse antibiotics), or using a more reliable near-patient urine screening test.36 GPs supported by a specialist nurse service treated 60% more children without delay than did control GPs.28
COULD THIS BE PRACTICALLY DELIVERED?
At least two models suggest it may be possible to deliver sufficiently rapid and effective diagnosis and treatment of childhood UTI to reduce renal scarring: the Swedish experience and the Newcastle direct access model. Febrile children in Sweden are primarily managed by paediatricians, and there is a high degree of awareness of the need to exclude UTIs, especially in infants. No cases of end-stage renal failure were recorded among children in Sweden between 1986 and 1994.37
In the UK, our study in which a specialist nurse worked directly with Newcastle GPs, not only multiplied the UTI diagnosis rate, improved diagnostic standards, reduced treatment delay, and improved GP and parent satisfaction28 but may also have prevented scarring. Part of the intervention facility was the provision of an acute urine collection and rapid microscopy36 diagnostic service available to GPs on our children’s day unit. We saw many infants post-UTI who had VUR but no scarring, which was a new phenomenon for us, although others have described it.38 We speculated that such children might have had scarring prevented28 (one has since acquired a scar following a UTI after his family stopped antibiotic prophylaxis). Post-study, we have now provided a nurse-led service for all Newcastle children for over 3 years. It is encouraging, but not conclusive, that only four children born since then have scars, compared to 11 predicted from historical rates27 (we are collecting prospective data). This suggests, but does not prove, that it should be feasible to deliver locally-organised services within the UK that may be able to reduce kidney scarring.
SHOULD WE SCREEN FOR AND MANAGE CHILDREN WITH VUR WITHOUT UTI?
The evidence that VUR and UTI may result in rapid, silent, permanent scarring in infancy, plus the fact that kidneys do not outgrow their scarring risk, has major implications for whether VUR should be screened for in high-risk groups, and how it should be managed.
Primary non-syndromic VUR is genetic.39 Its prevalence is up to one-third in babies whose close relatives have VUR,40 41 and in those with a multicystic-dysplastic kidney.42 43 Screening these higher-risk infants might provide an opportunity to prevent scarring if their chance of having an unrecognised prolonged UTI could be reduced, perhaps by increased diagnostic vigilance with prompt treatment, or by antibiotic prophylaxis, or a combination. This remains unproven but may be tested by a national prospective study being developed by the British Association for Paediatric Nephrology (BAPN).
Previously, many paediatricians have aimed to maintain children with VUR free of UTIs up to 4 years of age, believing them to be safe after that. It is now clear this may not be so,24 25 and until children outgrow their VUR, their risk of scarring with future UTIs remains. Figure 1A shows an example of this. We are not aware of any other circumstances in which scarring has been demonstrated to begin after 4 years of age, apart from in kidney transplants. It is therefore our policy to actively manage children with VUR until they outgrow it.
In practice, we prescribe low-dose night-time trimethoprim and provide rapid access to an immediate UTI diagnosis service using phase-contrast microscopy, until the VUR has been shown to resolve on indirect cystography at 4 years of age, or on scans at intervals thereafter. Using this approach, the very few children who have developed new scars have all been in families admitting non-compliance (fig 1B) or where management has been extremely complicated (fig 1C).
WHO NEEDS RENAL IMAGING?
The NICE childhood UTI guidelines have advised18 using less imaging tests on fewer children than previous national recommendations.19 This may reflect the authors’ views that little can be done to improve outcomes, and that smaller kidney scars are unimportant, so there is little benefit in documenting them. I disagree.
I have argued that scarring rates are likely to be reduced if more attention is focussed on minimising treatment delay and attempting to prevent UTIs in children with VUR. Although this case is not proven, it is certainly not disproved by any evidence presented by NICE. However, if there is any chance of an active approach reducing the burden of renal scarring, it is essential to use reliable imaging to decide on management (eg, does the infant have VUR?) and to evaluate outcomes (eg, has the child acquired a scar?).
Imaging has no value unless it is done effectively. For example, if it is considered important to determine whether a child has acquired renal scarring, there is no point in advocating using ultrasound to detect it, as NICE has done.18 While ultrasound is very sensitive for detecting abnormalities such as hydronephrosis, as NICE’s own background analysis shows, its sensitivity for detecting scars (including large ones) is hugely variable and only averages about 50%. The statistical implications of this are precisely the same as the argument already presented for nitrite sticks. A negative ultrasound does not exclude scarring, and relying on this technique guarantees missing half the cases.
NICE’s description of a DMSA scan being invasive18 is interesting. It does require a venepuncture (we use a 23-butterfly and local anaesthetic cream) and provides an effective dose of 0.7 mSv of radiation,44 but this is equivalent to the increase in radiation that a Geordie from northern England would receive by spending 7 weeks in Cornwall in southern England.45 We do not consider these concerns justify avoiding its use after a first UTI.
IS IT ALL WORTH IT?
Although UTIs may produce renal scarring, and this in turn may occasionally lead to hypertension or very rarely cause end-stage renal failure, most children with UTI do not sustain any sequelae. From the perspective of GPs or general paediatricians, energy invested in diagnosing, treating and imaging children with UTI can understandably feel fruitless because most cases are normal, so it may seem that little is achieved. From that perspective, the NICE guidelines18 might appear attractive because they will inevitably reduce the current clinical workload; UTI diagnosis rates could fall by a further half through false-negative nitrite stick results, and among those children, only half the scars will be identified without DMSA. However, while the individual risk of sequelae is low, childhood UTI is so common that many people are affected. The question is whether potentially preventing chronic kidney disease is worth the cost of actively managing childhood UTI.
When the sequelae of reflux nephropathy are seen in adult life, the label of pyelonephritis is usually used. This was the primary diagnosis in 12% of patients having a kidney transplant in the UK northern health region (3.1 million total population) for the first 5 years of this millennium, which translates to one patient (mostly adults) every month. If 12% of the dialysis population in the same area also have pyelonephritis (a reasonable assumption), then approximately 100 patients are being dialysed at any one time in the north-east of England for the same reason. Apart from the obvious quality of life issues, this is very costly. Using national tariffs, the annual cost of keeping these patients alive is approximately £4 000 000, without even considering the expenditure on commoner sequelae such as hypertension. Even if a significant proportion of cases labelled as pyelonephritis were dysplastic in origin, the savings are still immense.
I believe that it may be possible to substantially reduce renal scarring in children, and hence in the next generation of adults, and that this is worth doing. Whenever possible, prevention is better than cure. However, if the NICE childhood UTI guidelines18 are widely adopted we will lose this important opportunity.
Competing interests: The author declares that the ideas and writing of this manuscript were entirely his own, and that there are no competing interests.