In the past few years several articles have suggested routinely culturing urine of febrile infants to determine if their fever is due to a urinary tract infection (UTI). The reasons for diagnosing UTI in febrile infants are twofold: an immediate goal of identifying a treatable cause of the child's fever, and a longer-term goal of preventing kidney damage in those particularly susceptible. The pursuit of the second goal has led to an additional set of recommendations: that children with UTI have their urinary tracts imaged to determine whether they have scarring, vesicoureteral reflux, or obstruction. The thoughts are that UTI is more likely to occur if there are underlying urinary tract abnormalities, that in the presence of these abnormalities UTI carry a greater risk of sequelae and that early identification of these sequelae can lead to effective preventive strategies.

A. Focus: febrile infants 2-3 months to 2-3 years. Infants are the toughest group because they can't describe symptoms, can't void on command, and may be at higher risk of sequelae. As children get older, the decisions get easier: clean voided specimens become possible, imaging probably becomes (even) less necessary, and typical symptoms are more likely to be present. In febrile infants < 2-3 months you should have a low threshold for checking urine, as they are at greater risk of accompanying bacteremia. (Although, even in this high-risk age group, acute symptoms of most UTIs probably resolve even without treatment.) The other reason for focussing on older infants is that I think they are more likely to be psychologically traumatized by urethral catheterization than are the infants < 3 months old.

B. Decision analytic approach: I've tried to identify some of the key decisions, and the data one needs to make them optimally. This approach involves estimating PROBABILITIES and UTILITIES.

1. Probabilities are easier because (at least theoretically), they can be measured empirically. Examples are the probability that a circumcised boy with a negative UA will have a positive culture, or the probability that a girl vesicoureteral reflux will have progression of scarring while on prophylactic antibiotics.

2. Utilities are much harder. They may involve specific value judgements that vary across patients and physicians. (In fact, Kramer et al have found systematic differences between parents' and physicians' utilities related to diagnostic evaluation for UTI. Parents tend to be more willing to risk late sequelae in order to avoid the discomfort of diagnostic tests.) If we disagree about recommendations, it may be because we have different utilities for possible outcomes. Some questions we'll need to address that illustrate the difficulty estimating utilities are:

a. How many urethral catheterizations are worth doing to prevent one ambiguous or falsely positive culture result?

b. How many VCUGs (voiding cystourethrograms) is it worth doing to prevent one recurrent infection? To prevent one case of scarring? One case of hypertension or ESRD (end-stage renal disease)?

c. How many VCUGs is it worth doing to prevent giving prophylactic antibiotics to one child with UTI who does not have VUR (vesicoureteral reflux)?

Since there is no single answer to these questions, there is no single approach to the diagnosis and management of UTI that is right for everybody, even given a particular set of clinical and laboratory findings.

This decision depends on two things that are likely to vary from patient to patient: the likelihood that a UTI is present, and the potential benefit (and perceived discomfort) of diagnosing it.

A. Estimating the likelihood of UTI:

There is strong evidence for associations between risk of UTI and temperature, race, sex, and circumcision, and presence of other sources for the fever.. It's hard to quantify the effects of all of these predictors because so far most publishing studies have not presented the data in a way that allows that. As a rough ballpark, for infants with T ³ 39 degrees and no source, figure about a 1% risk of UTI in a circumcised boy < 1 year (even less thereafter), 10% in an uncircumcised boy, about 3-5% in a black girl, and 15% in a white girl. Decrease it somewhat if there's a source or lower temperature; increase if it the temperature is higher or the child is younger. Unfortunately, we know there's a big racial difference between white and black girls, much less data on infants of other races or ethnicities. Two studies suggest Latina girls are at (~50%) lower risk than whites. (The relationship in Latino boys is confounded by low circumcisions rates; but the lower risk in Hispanics probably also holds for boys. I've provided some likelihood ratios at the end of the handout.

At what probability of UTI is it worth obtaining urine? I don't know. Maybe 5 - 10%? Is the cost of 10 - 20 UAs (and a bunch of cultures) to make one diagnosis of UTI reasonable? It depends partly on the benefit of diagnosing it, as discussed below. Most UTIs, like most episodes of bacteremia, probably resolve on their own without treatment.

B. The benefit of diagnosing UTI is likely to be greater the younger and sicker the child. First, the sicker the child, the more you have to gain from quicker relief of symptoms. In addition, with high fever and young age the risk of immediate sequelae (bacteremia, shock) and (possibly) late sequelae (renal scarring; studies are mixed on this) are greater. This is the reason for recommending urine be checked in febrile infants less than 3 months old -- the importance of making the diagnosis is greatest in that age group. Beyond the first few months after birth, bacteremia as a complication of UTI is extremely rare. (Unlike the case in younger infants, in large series of 2-24 month-olds with bacteremia, enteric gram negatives are uncommon organisms.)

C. Timing: It's not unreasonable to check the urine on the first day of a fever, but a lot of fevers go away by themselves in a day or two. Thus (although I haven't seen good data), I suspect that the yield of urine cultures goes up with the number of days of unexplained fever. (in the 0-3 month age group, duration of fever > 24 hours increased the risk of UTI by about 80%. I'm also reluctant to recommend urine cultures on all infants the first couple of days of illness because that would be inconsistent with the advice I give over the telephone--I don't think all such infants need to be seen by a physician or have tests done. Of course if the child looks acutely ill, it's fine to look for a source without waiting. But as a general recommendation, I think you can wait at least a couple of days before routinely culturing urine. In at least one study that examined delayed diagnosis of UTI as a risk factor for renal scarring,, the definition of "delay in diagnosis" was no urine culture when there were UTI symptoms or fever for at least 5 days.

D. "Source" for fever: As might be expected, infants with no source for their fever are at higher risk of UTI than infants who have a definite source.

1. Otitis media: If you've diagnosed otitis media and will be treating the child with antibiotics anyway, the main reason to diagnose a UTI is because it might trigger further evaluation and treatment that might prevent future UTIs or their sequelae. However, I'm comfortable recommending against urine testing in this setting because (1) the infants will receive antibiotics anyway; (2) the yield of a urine culture will be very low and (3) the future benefits of making the diagnosis are currently unproven. The possible exception would be infants who are too young, febrile, or sick to treat with oral antibiotics.

2. URI: As you know, the symptom burden and specificity of a diagnosis of "URI" can vary a lot. Some infants will have a definite URI, with lots of symptoms and usually a positive exposure history. In such infants parents and the clinician are usually quite clear on what is causing the baby's fever, and it doesn't seem necessary to look for another explanation. However, if the infant mostly has high fever, with just a little cough or red throat, it would have to qualify as a "possible" as opposed to "definite" source of fever on exam, and it becomes reasonable to look for another source, particularly in those at highest risk (e.g. uncircumcised boys, white girls).

3. GI symptoms: Since a UTI can sometimes cause GI symptoms (especially vomiting), it would be unwise to assume vomiting is due to viral gastroenteritis. Unless there is a classic history (e.g., vomiting for 12 hrs followed by watery diarrhea in a child not very sick or febrile), GI symptoms should probably not count as a "definite" source of fever.

E. Tentative recommendation: Obtain a urine sample when the risk of UTI is about 10% or more, when the child is ill enough that you want to give parenteral antibiotics and have no other source, or when fever has gone on for 3-4 days with no definite source. Urine testing on children seen earlier in an illness is not unreasonable if it can be done at minimal cost and inconvenience, because there is certainly cost and inconvenience involved with a return visit for urine testing if the fever continues. However, the yield of testing in this situation will be low and it clearly is not mandatory.

1. Suprapubic aspiration (SPA) may be the best method, especially in the youngest infants. Experience with this technique in older infants is limited, and may be diminishing. In a recent study of infants < 6 months, the success rate of SPA was only about 50%, compared with 100% for urethral catheterization. Except for microscopic (common) and gross hematuria (~1%), reported complication rates are very low -- on the order of 1 in several thousand. Ultrasound guidance increases the success rate of this procedure to close to 100%, but also presumably its expense and inconvenience.
2. Urethral catheterization has become very popular. It yields a more easily interpretable result than a urine bag, available faster, but is invasive and runs the risk of introducing infection. The only data I could find suggest that the risk of iatrogenic infection is at least a few per cent (see below). Another problem is that some parents (and many infants) have strong negative feelings about this procedure. If you do it over their objections (as opposed to involving them in the decision), they may not want to come back the next time their child has a fever. (In fact, they may not want to come back at all.) Sometimes it's hard to find the urethral meatus in girls; it may help to gently retract the introital mucosa downward with a Q-tip. There's a particular concern about introducing infection in young uncircumcised boys, although good data are lacking. In both boys and girls the contamination rate will be lower of the first part of the sample is discarded.
3. Clean catch is probably the best choice! Have a specimen cup handy, so if the child starts to void as you are preparing for urethral catheterization, suprapubic aspiration or bag, you can catch a midstream urine! (This happens surprisingly often. I've also found that standing in front of little boys wearing nice clothes seems to trigger voiding.) Ramage et al recently reported sensitivity of 16/18 (89%) and specificity of 38/40 (95%) comparing clean-catch samples to (ultra-sound guided) suprapubic aspiration. They had the parents hold the infants on their laps (undressed) over a foil bowl to catch the urine; this reportedly led to a sample in less than an hour in a majority of patients. However, in another study, parents preferred
4. Urine Bag: The obvious advantage of a urine bag is that it is noninvasive. The disadvantages are that you have to wait for the infant to void, and if you send the urine for culture, the result can be falsely positive, falsely negative, or impossible to interpret. Bags are harder to use if the child has diarrhea, diaper rash or both. Contamination can probably be reduced by careful washing of the periurethral area first, and taking care that the urine sample is removed from the bag as soon as the child voids, rather than incubating against the baby's skin. However, somewhat surprisingly, Al-Orifi et al. recently found that the contamination rate for bag urines was actually higher when they were carefully obtained at a specialized Pediatric Test Center by nurses who reapplied the bag every 30 minutes if the infant had not voided, than when bag urines were obtained by E.D. nurses (69% vs 56%; in that study contamination was defined as any mixed growth or growth of a single organism at 1000-9999 CFU/ml) The same study found contamination was associated with adverse outcomes (e.g., unnecessary or delayed treatment) in about 2% of cases, although most of these could have been prevented by just considering contaminated urines to be negative. The contamination rate for bag urine may be low (<5%) in circumcised boys , but estimates from the literature for other infants vary widely, depending on definitions and other factors, and are often > 20%. If the sample is to be sent for culture (not enthusiastically recommended) it should be put on ice. As discussed below, the danger of contaminated or wrong culture results depends on how you react to them.
5. Diapers, Collection pads, etc. There are a few studies looking at urine collection from disposable diapers or pads. Cohen et al. found urine from the diaper gave similar results on culture to urine fom SPA or UC in 38 febrile infants (sensitivity 5/5; specificity 31/33). Others studies have not been as favorable, at least as far as culturing. For example, Liaw et al had parents of 44 well infants 1-18 months collect urine by bag, clean catch and collection pad. They found that parents preferred the collection pad, but that 25-34% of infants in all groups had bacterial growth at ³ 10^4 CFU/ml, including 9 false positive cultures (³ 10^5 CFU of a single organism, with UTI ruled out by subsequent culture). Ahmad et al. reported "contamination" rates (any mixed growth) of 10/39 and 17/39 for "nappy" (diaper) and bag collections, respectively. My assessment is that these methods are probably about like a bag -- OK for a UA, probably OK when the culture is negative, but unless you know your local contamination rates are lower than what is often reported, not good for culture. (Note that you can't get urine out of newer "super-absorbable" diapers because the urine gets sucked into a gel and won't come out.)
6. Recommendations: I think the best way to collect a specimen depends on the individual patient. Some families may be in a hurry and want their infant catheterized, others may very much want to avoid that procedure. I agree with the AAP that if a child looks sick enough to need immediate antibiotics, it's best to get the sample invasively and be done with it. For infants that don't look that sick, a compromise strategy (OK'd by the AAP) is to use the UA from a bag specimen to select samples for culture by UC or SPA (see below). This is reasonable, especially in children whose prior probability is close to the threshold at which culturing is worthwhile (whatever that is!).
   
   If a urine bag is placed on the infant soon after triage, waiting time can be minimized. If you cut a hole though the disposable diaper and bring the bag out through it, the parents can immediately notify the nurse when the child voids. If the dipstick is negative, you can probably toss it; see decision #3 below. If it is dipstick positive, you can obtain a catheterized or SPA specimen for culture. If you do go the bag route, you should explain the process to the parents at the outset, so they are not taken by surprise if you later want to obtain urine invasively.
   
   Can you send a bag urine for culture? I'm not enthusiastic about it, but it really depends what decisions you are going to make as a result of the culture. If you are going to recommend imaging if the culture is positive, you don't want to rely on a bag. If, on the other hand, you are only going to decide whether to continue (already purchased) antibiotics for a week or stop them after 2 days, and parents don't want to obtain urine invasively, it doesn’t seem like it's worth a battle to get urine invasively.
7. It's not really the topic of this handout, but I definitely wouldn't catheterize older kids. It's scary and traumatic and midstream specimens work fine. (Girls should face backwards on the toilet.)

A. Rationale: As with decision #1 above, this depends on the chance that the child has a UTI and the benefit of diagnosing it. It comes down to: how many urine cultures is it worth doing to find one UTI? The initial treatment decision will generally be based on the UA. Therefore, the potential benefits of the culture in the short term are either discontinuing antibiotics after a day or two in someone who doesn't need them, or starting them a day or two late in someone who does. I feel pretty comfortable that if the chance of a positive culture is < 2 -3%, it's probably not worth doing, and very comfortable about not culturing if the chance of a positive culture is < 1%. This is because 1% is getting down to the prevalence range of asymptomatic bacteriuria, which we know there is no benefit to detecting. (But this is arbitrary. Bachur and Harper recently reported they are comfortable not culturing when the probability of a positive culture is below 0.4%.)

B. If the UA is positive for nitrite, culture is probably unnecessary for diagnosis (because nitrite is so specific). So then the only reason to do it would be to identify the organism and its sensitivities. For a healthy, non-toxic child with community-acquired UTI, this is probably not necessary, but it is also not unreasonable. So I have no strong recommendation on this one. To be 100% honest, I'd probably send it.

C. If just the esterase is positive, it's probably worth sending the urine culture -- chance that it will be positive is about 30% (depending on other factors).

D. MICROSCOPIC UA probably adds little sensitivity to the dipstick, because the esterase usually is positive with more than 2 WBC/HPF. But the microscopic UA can still be helpful because just doing the esterase you don't know if there are 2 or 50 WBC/HPF and the latter is much more suggestive of UTI.

E. If the UA is entirely negative, I'd toss out urine obtained by bag (i.e., not bother with a culture) unless the prior probability was very high (e.g. in someone with a history of past UTI or known anatomic abnormality) or the patient very young (< 3 months) or ill. If you look at it from the point of view of sensitivity this may seem like a mistake -- about 1 in 5 UTIs will be missed. But the proper perspective is negative predictive value. If you start out with a 10% chance of UTI, and the UA is negative, the chance of a UTI goes down to about 2%. (If we can do an "enhanced UA," we can get it even lower.) The significance of these UTIs that didn't have much pyuria is very questionable, and you'll do 50 urine cultures for each one you diagnose! Plus, of course, your immediate management will not be affected anyway!

Here's another way of looking at it. It has become pretty well accepted now to use the WBC count to help decide which children with fever should have a blood culture. But the WBC as a predictor of bacteremia performs much more poorly than the UA as a predictor of bacteriuria. And somehow I find it hard to believe that identifying bacteriuria unaccompanied by pyuria (which we know can be asymptomatic and harmless) is that much more important than identifying bacteremia unaccompanied by leukocytosis!

What should you do with dipstick-negative urine obtained by catheterization? If you're a bit selective about whom you catheterize (i.e., you catheterize younger or sicker infants), then it makes sense to go ahead and culture the urine even if the dipstick is negative. However, the yield will be quite low, and even if you find bacteriuria, it won't be clear what it means.

Note: all you could want to know about test characteristics of UA is included in the technical report to the AAP practice guideline written by Stephen Downs, the URL is at the end of this handout.

A. Decision to hospitalize should be based on the age and toxicity of the child and whether they have demonstrated ability to take fluids and medications orally. Most children with UTIs probably can be treated as outpatients. A shot of ceftriaxone to begin treatment may be a good way to start if you are worried. Especially in young (<6 months) infants, UTIs make me nervous. If you are going to treat orally, I recommend observing the child taking the medication in the emergency room and watching for at least an hour to make sure he or she does not vomit.

B. Choice of antibiotics: Hoberman et al did a randomized trial of oral cefixime vs IV cefotaxime in children 1-24 months with febrile UTIs. They found that the urine was sterile in 24 hours in all children in both groups. There was also no difference in rates of scarring at follow-up. So I'm tempted to recommend cefixime, particularly for younger patients. An alternative, based on sensitivities, cost, and taste, is a first generation cephalosporin. Amoxicillin and TMP/SMX are cheap and usually effective, but at least in vitro there's quite a bit more resistance. Ceftriaxone is a good choice IM, and ampicillin and gentamicin or a third generation cephalosporin should be good choices IV.

A. This, of course, is a major controversy. One thing to keep in mind is that as we look harder for UTIs and scarring, we will begin to find them in children in whom they previously were not noted. Such children are likely to be more mildly affected and have a better prognosis than those identified and studied previously.

B. I'd start with a history. For boys in particular, make sure the urine comes out in a nice stream, rather than dribbling. Other simple things, like a growth curve, blood pressure and serum creatinine are reasonable. Although the literature on this is not great, constipation is probably worth identifying and treating because it can probably contribute to UTI by interfering with bladder emptying.

C. For imaging, I'd start with ultrasound. There's been a lot of criticism of ultrasound as being too insensitive an imaging modality, but it mostly misses lower grades of VUR and some renal scarring, neither of which I care much about. The main reason for imaging is to identify obstruction, and ultrasound is fine for that. In fact, if the mother had an ultrasound in the third trimester, the yield of obstruction on ultrasound after UTI (in the absence of a history of abnormal voiding) is close to zero. If there has been no late prenatal ultrasound, I'd recommend an ultrasound in circumcised boys for after a first clear UTI. It's probably also reasonable for girls and uncircumcised boys less than 1 or 2 after a one or more febrile UTIs, though their risk of obstruction is a lot lower.

D. VCUG? Most published recommendations include a VCUG, and there's a recent tendency towards recommending them sooner (after the first UTI), as opposed to later. However, my trouble with the VCUG is that I don't know what to do with information about vesicoureteral reflux. A perfectly acceptable treatment strategy for VUR is close follow-up with early treatment of any UTI recurrences, and I'd recommend that whether the child had VUR or not. So if the ultrasound is normal, I'd stop. I don't see the purpose of a DMSA scan, since I don't know what to do with information on scarring. I don't think anyone needs an IVP. I think prophylactic antibiotics may be reasonable (although I worry about selecting out resistant organisms), but I'd base that decision on frequency and severity of recurrences. (More discussion on this below.) . Note that in the AAP practice guideline the wording is: "a VCUG or RNC is strongly encouraged (strength of evidence: fair)." (There is a mistake in the print version, which says it should be done.) In fact, in response to some excellent letters-to-the-editor about the guideline, the Subcommittee wrote:

I agree with this assessment, except that I would say the evidence of benefit from prophylactic antibiotics is somewhere between poor and nonexistent.

E. AT LAST, more people are beginning to question many assumptions underlying imaging. A recent article in Pediatric Nephrology questions the notion that urinary tract infections cause renal failure and a recent review of VUR concludes "A review of the available data does not support many of the roles attributed to vesicoureteral reflux or found conflicting data. Vesicoureteral reflux does not seem to be …a predisposing factor for the development of UTI nor a contributing factor inducing renal parenchymal scars…" Additional selected quotes:

• "[VCUG] is often emotionally traumatic…carries a small risk of introducing infection…and involves ionizing radiation. In the absence of controlled studies to demonstrate that either prophylaxis or antireflux procedures are superior to early diagnosis of intercurrent infection and short courses of appropriate treatment, it is difficult to persuade doctors and parents that [VCUGs] are really necessary…"
  --Verrier Jones, K. Arch Dis Child 1995;72:255
• "Operation and long-term antibiotic prophylaxis are equally ineffective...We have at present no scientific basis on which to tell our patients and their parents that their reflux needs to be [treated]...It is psychologically difficult to accept results that suggest that time-honored methods that are generally recommended and applied are of no or doubtful value."
  --Winberg, J. Infection 1994Suppl 1;22:S4-7
• This study suggests that the currently advocated treatment of vesicoureteral reflux in children, which is invasive and costly, may be of no benefit in the prevention of ESRD.
  Craig JC et al. Does treatment of VUR in childhood prevent end-stage renal disease attributable to reflux nephropathy? Pediatrics 2000;105:1236-41.
• "An appreciation of the epidemiology of ESRD…should lead to reconsideration of the imaging evaluation of the urinary tract recommended for a pediatric patient following a UTI."
  -- Sreenarasimhaiah S, Hellerstein S. UTI per se do not cause end-stage kidney disease. Pediatr Nephrol 1998;12:210-3.
• 'It’s time to question the routine use of an invasive test for what some experts call an 'innocent bystander.'
  --Ortigas AP, Cunningham AS. Three facts to know before you order a VCUG. Contemp Peds 1997; Sept:69-79.
• "In God We Trust. All others must bring their data."
  -- W. E. Deming

F. Follow-up. I think what is probably most important is emphasizing that recurrences of UTI should be treated promptly. The study by Smellie suggests that if recurrences are treated promptly (< 5 days of fever or less than a MONTH of nonspecific symptoms!) the risk of scarring is low.

If it were MY kid (and years ago it was, since my daughter had several UTIs), I'd want access to dipsticks, so I could check the urine myself when I had any suspicion of UTI, but I wouldn't want to give prophylactic antibiotics all the time or have surgery. (I did acquiesce to an ultrasound (which was normal), but not to the VCUG.)

I said at the beginning that the twin purposes of diagnosing UTI were immediate relief of symptoms and prevention of kidney damage. The more I read about VUR, the more skeptical I get about purpose #2. The goal, of course, is laudable--we all would love to prevent kidney damage, especially if it translates into something the patient can notice, like pregnancy complications or end-stage renal disease. However, good evidence that diagnosing and treating VUR will help accomplish that goal simply does not exist.

Many authors indicate that the thing to avoid is reflux of infected urine into the kidneys -- that either reflux alone or infection alone is not so bad, but that the combination can lead to scarring and eventually destroy the kidneys. But the randomized trials of medical vs surgical treatment of reflux do not provide much support for this view. In both the Birmingham trial and the International Reflux Study in Children (IRSC) (both European and US arms), the surgery was highly successful at curing reflux. Since the rate of infection was the same in the two groups, there should have been less scarring in those treated surgically. But the rates of scarring were similar (if anything a little higher with surgery) suggesting that cure of reflux does not prevent scarring. Furthermore, in the IRSC, children with recurrent infections actually had LESS scarring than those who did not! Here's a table that can be extracted from that paper:

Similarly, I would expect surgery to reduce the frequency of upper tract infection more than lower tract infection, whereas medical treatment should have prevented both upper and lower tract infections. Thus I would have expected fewer total infections in the antibiotic group, but perhaps a higher proportion of pyelonephritis. But this was NOT observed. Surgical treatment apparently did keep some infections confined to the bladder, but the total number of infections in the two groups was the same, suggesting that medical treatment did not prevent any infections. Thus, one conclusion that is consistent with the data from clinical trials is that neither medical nor surgical management makes much difference in this condition. (Surgery did have a statistically significant benefit, in terms of reduction in pyelonephritis episodes, but the effect size was small: about 1 case of pyelonephritis prevented per 5 children receiving ureteral reimplantations.)

One line of "evidence" you may encounter is that children with breakthrough infections or otherwise bad outcomes are described as having been noncompliant, or children with ESRD have a history of fevers of unknown etiology in infancy, that are later presumed to have been UTIs. But these sorts of uncontrolled observations do not provide good evidence for efficacy of treatment. Noncompliance in those with a good outcome is less likely to come to the attention of the investigators (and will be less salient when it does). And possible recall bias and lack of a control group make histories of possible undiagnosed UTIs during infancy unconvincing.

We in medicine have a long history of doing things to patients based on beliefs rather than good evidence. Compared with many doctors, I think I like to see stronger evidence of benefit before recommending invasive and expensive tests and procedures. There is no question that others have looked at the same evidence I have and found it more convincing. I don't have a problem with people disagreeing with me and recommending VCUGs. But I do have a problem with dogmatic recommendations that do not acknowledge the poor quality of the evidence, and that do not allow room for decisions to be influenced by preferences of individual patients.

1. Obtain urine for UA and/or culture based on level of concern about UTI (see table). Level of concern depends both on the probability of UTI and its potential consequences.

 

1. The method of urine collection can be individualized. Sick patients needing immediate antibiotics should probably have their urine obtained by suprapubic aspiration or urethral catheterization. Bag urine is OK for most others. Early application of a urine bag in febrile infants is recommended.
2. If the dipstick is negative, culture of urine obtained by bag is generally not necessary. (Note that "trace" positive leukocyte esterase is considered positive.)
3. If the dipstick is positive, urine should be cultured. If the urine was obtained by bag, consider obtaining a second bag sample, or a sample by SPA or urethral catheterization, especially if you believe in imaging of infants with UTIs. In most cases it is reasonable to treat empirically pending culture results. (See below.)
4. (Following applies to UCSF only): When culturing urine, request a "SCREENING CULTURE." (The box to check has been moved to the SOURCE part of the requisition.) This provides information of the form "10^5 Gram-negative rods." Identification of the organism and antibiotic susceptibilities can always be added on later (e.g., if the patient is not responding) by calling the microbiology laboratory.
5. Interpretation of the urine culture depends on the method of collection as well as the colony count. In ambiguous cases, the level of concern about UTI before the culture guides interpretation. (See Table 2.)


Table 2: Interpretation of culture results




6. Decisions to treat orally vs parenterally and as an in- or outpatient should be individualized based upon age, toxicity, vomiting, and reliability of follow-up. Initial oral treatment should generally be with a first generation cephalosporin for nonallergic patients, initial parenteral treatment can be with ceftriaxone (IM or IV), or ceftizoxime, or ampicillin and gentamycin (IV).
7. Consider a renal ultrasound if no clinical improvement in 48 hr.

9. Prophylaxis and imaging may be considered after the acute episode. Many authorities recommend both ultrasound and voiding cystourethrography (VCUG). However, the yield of ultrasound if very low if there was a normal 3^rd trimester ultrasound, and in patients with normal ultrasonography, VCUG primarily serves to diagnose vesicoureteral reflux. As there is currently no good evidence that patients with vesicoureteral reflux should be treated differently from other patients, acceptable (and, in TN's opinion, preferable) alternatives to invasive imaging are either (1) close follow-up, aimed at early diagnosis and treatment of recurrences or (2) antimicrobial prophylaxis (without imaging) in highest risk infants. Almost no one needs an IVP or DMSA scan.

APPENDIX: Additional data and references

I. Data for Decision #1: Whether to obtain a urine sample.

A. Prevalence of UTI: 4-5% overall 1-2% circumcised boys; 10 times higher in uncircumcised boys; 5-15% in girls; see below.

B. Risk factors: LR

1. Source for fever?

a. No 1.5

b. Maybe .7

c. Yes .3

2. Fever

a. <39 .7

b. 39-40 1.4

c. >40 2.5

3. Younger age

a. Boys < 6 mos 1.5

Boys > 6 mos 0.7

b. Girls < 12 mos 1.4

Girls ³ 12 mos 0.5

II Data for decision #2: How to obtain the urine specimen

A. False positive rate (>10^5 orgs) for bag urine

1. Circumcised boys: 0-2%

2. Girls: 2%?

If culture result is < 10^5 organisms, it's probably negative. The problem comes interpreting >= 10^5 CFU of a single organism (could perhaps be a false positive) and especially interpreting >10^5 CFU of multiple organisms (could be overgrowth of what had been a pure culture of a pathogen, or could be just contaminated.)

B. Contamination rate:

B. Complication rate from suprapubic aspiration: In competent hands, probably on the order of .2%; may be higher in a teaching setting.

C. Complication rate for catheterization for urine: Jodal et al reported infection rates of 1.4% following "instrumentation" of the urinary tract,(many of these may have been VCUGs or cystoscopies, which might carry a higher risk). Of these, 9/16 were pyelonephritic. Lohr et al. prospectively studied inpatients whose urethras were catheterized and found about a 10% infection rate. Most of these patients were subjected to multiple intermittent catheterizations or had indwelling catheters, but 2 had had a single in-and-out catheterization. The denominator for these 2 is not provided, but cannot be > 64 (which would give a 3% rate), the total number of catheterized children who did not have indwelling catheters. It seems more likely that the denominator is about 20-30, giving a 7-10% infection rate. The Oski text (p. 1969) cites a study by Kass from the 1950s (I haven't seen it yet) that 2% of asymptomatic women developed UTI symptoms after being catheterized.

D. Caveats: You need to take the bag off and get the specimen on ice or refrigerated as soon as the infant voids, or contamination rates will be much higher. Pure growth of > 10^5 CFU of a pathogen on a bag urine is very suggestive of a UTI, and colony counts < 10,000 pretty much rule one out. Other results probably need to be repeated.

A. Performance of elements of UA at predicting UTI (Definitions vary, but generally similar to whats in Table 2, i.e. > 10^5 orgs for a bag urine, sometimes fewer for cath or SPA .(Data below are estimated from reviews by Lohr,, Downs and Gorelick and Shaw. Of recent systematic reviews, the one by Huicho, is much less user-friendly. The one by Gorelick and Shaw is excellent except for an internal inconsistency: both sensitivity and specificity for (nitrite+ or LE+) are higher than LE+ alone.

1. Esterase LR

a. >= trace + (sens~.85; spec~.85) 5.7

b. - .17

2. Nitrite (sens~.4; spec~..98)

a. + 40

b. - .6

3. Dipstick positive for esterase OR nitrite (sens .88; spec .85)

a. Any 5.9

b. None .17

4. Urine WBC/mm3(Sens .91, spec .965)

a. < 10 .09

b. >= 10 26

5. Urine WBC/HPF (Sens ~.7; spec ~.8)

a. >5 3.5

b. <5 .375

IV. Data for Decision #5: Diagnosis and treatment of VUR. There was a nice summary of this on the U.W. EBM website, but as of May 20, 2002 it seems to be gone.

1. Birmingham reflux study group--104 children entered if they had grade 2 reflux plus scarring, or grade 3 reflux (note: there were only 3 reflux grades here--not the international system!). Surgical group had reimplantation; BOTH groups got prophylactic antibiotics, usually TMP or TMP/SMX or nitrofurantoin. No difference between the groups in scarring, renal function, or infections.

2. International Reflux Study in Children (European Branch)306 children (75% girls; 37% of girls and 67% of boys < 1 year old) with grade 3 or 4/5 VUR documented after a UTI randomized to surgical or medical (low dose prophylactic antibiotic) treatment. Identical rates of new scarring (based on IVP) 19/155 = 12% in the medical group, vs 20/151 (13%) in the surgical group in 5 years; new parenchymal thinning in 11/155 (7%) medical vs 15/151 (10%) surgical. There was no overall difference in UTIs, but the medical group had more pyelonephritis and less lower UTI. But this difference amounted to only .2 episodes of pyelo over 5 years per patient. Reflux surgery was followed by post-operative obstruction (probably leading to scarring) in 10/151 patients (7%). In this study, those treated with surgery had a 74% chance of an uncomplicated course: (none of the following: obstruction, persisting reflux, pyelonephritis or severe renal damage.)

3. Smellie looked at children from the International Reflux Study in Children (IRSC) who developed scarring and found that delay in diagnosis of UTI was the overwhelming risk factor, occurring in 41 of 52 cases. But note their definition of delay in diagnosis included: no urine sample cultured when there was a period of acute urinary tract symptoms or fever for at least 5 days or ill-defined symptoms (such as abdominal pain or screaming attacks) for at least one month.

4. International Reflux Study in Children (US Branch) : 132 infants and children (90% girls). New scarring (IVP) in 22% of medical and 31% of surgical patients over 5 years. VUR resolved at about 8% per year. In this study, new scars were NOT associated with recurrent infections. In BOTH groups there was continuing decrease in kidney size. "This is a particularly troublesome observation that should make us pause and reconsider the approach we use in the diagnosis and treatment of moderately severe vesicoureteral reflux." Like the European study, less pyelonephritis in the surgical group, again, a difference of about 0.24 episodes of pyelo per person over 5 years. Better results from surgery than in Europe?--no obstructions.

5. Summary: In a commentary on the IRSC, Jan Winberg notes the lack of difference between the groups, and states: "Since no nonprophylactic group has been included, it is impossible to judge whether the treatments studied actually had any effect at all on the course of the disease."

I've finally obtained some data on end-stage renal disease (ESRD) in children (age < 20).(See the U.S. Renal Data System website at: http://www.med.umich.edu/usrds/. ) This permits a "back-of-the-envelope" calculation to estimate what might be required to prevent a single case of ESRD from undiagnosed UTI:

Data:

First we need to estimate the cumulative risk of ESRD due to reflux nephropathy up to age 30. For data, see http://www.med.umich.edu/usrds/chapters/ch02/ch02.html

Cumulative risk up to age 19

13/million (= annual risk of ESRD age 0-19 yr) x 2.3% (% of ESRD in this age group attributed to reflux nephropathy) x 18 years (assuming average age 1 yr at time of UTI) = 5.4/million

Cumulative risk ages 20-30 = 117/million (= annual risk of ESRD 20-44 yr) x 1% (Estimated % of ESRD in this age group attributed to reflux nephropathy, based on published 0.6% for the entire 20-44 yr age group, and 0.4% for the ³ age 65 age group, which suggests that the risk in the 20-30 range is a bit higher than that for the entire 20-44 range) x 10 years = 11.7/million

So total 30-year cumulative risk of ESRD from reflux from age 1is about 5.4 + 11.7 = 17.1/million. On the other hand, about 1.5% of girls will have a diagnosed febrile UTI before age 2.. At least an equal number probably never get their UTI diagnosed or treated. Even if we assume ESRD is 100% preventable, i.e., that every case occurs among patients whose UTIs were missed, the 30-year cumulative risk of ESRD in someone whose UTI is not diagnosed or treated would be (17.1/ million)/1.5% or about 1/1100. Thus (under these very optimistic assumptions) we might prevent 1 case of ESRD for every 1100 UTIs we diagnose that otherwise would have been missed. To identify these UTIs, however, we will probably need to do ten times that many urine cultures (say 11,000 urine cultures), do VCUGs and sonograms on all 1100 who have UTI, repeat the VCUG and give prophylactic antibiotics to those with reflux, etc.

Thus to prevent one case of ESRD occurring some time in the 30 years following a UTI, we'd need a minimum of

• 11,000 urine cultures
• 1,100 sonograms
• 1,100 first VCUGs
• 300 repeat VCUGs (at least 30% of VCUGs abnl)
• 300 person-years of antibiotic treatment (if treatment continued for 1 year).
• And we haven't counted doctor visits, hospital admissions, etc.!

Something tells me this might not be a great way to spend health care resources, especially since these estimates optimistically assume treatment is 100% effective and there is no evidence that the treatment actually works AT ALL!