Latent class analysis: an illustrative application for education in the assessment of resident otoscopic skills

BACKGROUND

There is no gold standard readily available in several components of the routine physical exam: one example is the otoscopic exam, where the gold standard is confirmation by tympanocentesis. Resident education does not typically include routine assessment by the gold standard, making estimates of trainee performance extremely difficult. This is one reason why the otoscopic examination is difficult to teach. Available techniques can assess diagnostic exams when there is no gold standard-one of these is latent class analysis.

METHODS

We use latent class analysis, a form of regression analysis, to compare the ability of pediatric residents to diagnose effusion with pediatric otolaryngologists and tympanometry. We briefly outline the technique of how to complete latent class analysis and provide an operational plan to use the method to assess resident performance.

RESULTS

The sensitivity and specificity of pediatric resident otoscopic examination to diagnose the presence of effusion was 72% and 84%, respectively. Pediatric otolaryngologist sensitivity and specificity was 91% and 82%; tympanometry had a sensitivity of 70% and specificity 76%. Our estimates of the performance of otolaryngologists and tympanometry to diagnose effusion were the same as previously reported when these diagnosticians have been compared with the gold standard of tympanocentesis.

CONCLUSIONS

Latent class analysis can help estimate otoscopic examination performance of residents. This technique can be incorporated into assessment in medical education.

Comparative safety of amphotericin B lipid complex and amphotericin B deoxycholate as aerosolized antifungal prophylaxis in lung-transplant recipients

BACKGROUND

Aerosolized administrations of amphotericin B deoxycholate (AmBd) and amphotericin B lipid complex (ABLC) in lung transplant recipients were compared for safety and tolerability. The incidence of invasive fungal infections in patients receiving aerosolized amphotericin B formulations as sole prophylaxis was determined.

METHODS

A prospective, randomized (1:1), double-blinded trial was conducted with 100 subjects. AmBd and ABLC were administered postoperatively by nebulizer at doses of 25 mg and 50 mg, respectively, which were doubled in mechanically ventilated patients. The planned treatment was once every day for 4 days, then once per week for 7 weeks. Treatment-related adverse events and invasive fungal infections were quantitated for 2 months after study drug initiation.

RESULTS

Intent-to-treat analysis revealed study drug was discontinued for intolerance in 6 of 49 (12.2%) and 3 of 51 (5.9%) patients in the AmBd- and ABLC-treated groups, respectively (p=0.313). Subjects receiving AmBd were more likely to have experienced an adverse event (odds ratio 2.16, 95% confidence interval 1.10, 4.24, p=0.02). Primary prophylaxis failure within 2 months of study drug initiation was observed in 7 of 49 (14.3%) AmBd-treated patients and 6 of 51 (11.8%) ABLC-treated patients. No fungal pneumonias were observed. Only two (2%) patients experienced documented primary prophylaxis failure with Aspergillus infections within the follow-up period.

CONCLUSIONS

Both aerosol AmBd and ABLC appear to be associated with a low rate of invasive pulmonary fungal infection in the early posttransplant period. Patients receiving ABLC were less likely to experience a treatment-related adverse event.

Growth patterns reflect response to antiretroviral therapy in HIV-positive infants: potential utility in resource-poor settings

Laboratory monitoring of HIV-infected children is the current standard of care in the United States to guide the appropriate use of antiretroviral therapy (ART). Although ART is becoming a reality in some developing countries, laboratory monitoring of ART is costly, necessitating creative approaches to monitoring. As an initial step to guide monitoring of HIV progression in low resource settings, we assessed the utility of the physical examination to predict clinical progression of HIV. We conducted a retrospective cohort study of HIV-infected children using data from Pediatric AIDS Clinical Trials Group Protocol 300. We developed a clinical predictive model, and compared the utility of the clinical model to the change in HIV RNA viral load as diagnostic tests of ART failure. The clinical model incorporated treatment regimen, age, and height velocity: a three-level clinical predictive model provided likelihood ratios of 0.3, 3.9, and 14. For decline in RNA the likelihood ratios were 0.2 (> 1 log decline), 1.4, and 3.5 (> log increase). We developed a simple clinical predictive model that was able to predict clinical progression of HIV after initiation of new ART. The clinical model performed similarly to using changes in HIV RNA viral load. These data should be validated internationally and prospectively, because the test subjects were from a resource rich environment and growth patterns in undernourished children may be impacted differently by HIV and its treatment. The model was most pertinent to children 36 months of age or younger, and was conducted in children receiving monotherapy and dual therapy.

A comparison of height and weight velocity as a part of the composite endpoint in pediatric HIV

BACKGROUND

HIV adversely affects growth in children. Pediatric AIDS Clinical Trial Group (PACTG) protocols often use weight velocity [changes in weight z-score for age (WAZ)] as a part of the composite endpoint for phase II and III clinical trials. However, WAZ and height velocity (HAZ) have not been critically compared for their utility as part of the composite endpoint.

METHODS

HAZ and WAZ were compared to predict laboratory and clinical progression of HIV in a retrospective cohort study of HIV-infected children with data from PACTG Protocol 300.

RESULTS

In both bivariable and multivariable analyses, changes in HAZ were more closely linked to subsequent progression than WAZ. Children with improved HAZ were somewhat less likely to exhibit virological failure [odds ratio (OR), 0.76; 95% confidence interval (CI) 0.51-1.14], than children with improved WAZ (OR, 1.45; 95% CI, 0.99,2.11). Children who had improved HAZ were less likely to exhibit immunological failure (OR, 0.7; 95% CI, 0.49-1.00), than children with improved WAZ (OR, 1.13; 95% CI, 0.82-1.57). Children who had improved HAZ were less likely to have other forms of clinical progression of HIV (OR, 0.55; 95% CI, 0.31-0.99), than children who had improved WAZ (OR, 1.0; 95% CI, 1.58-1.94).

CONCLUSIONS

Increases in HAZ were associated with reduced risk of subsequent clinical progression and subsequent immune reconstitution and weakly associated with declines in HIV RNA. Changes in WAZ were not associated with laboratory outcomes relevant to pediatric HIV infection. Height velocity should be considered as a component of a composite clinical endpoint in future PACTG trials.

Candida bloodstream infection in neonates

Neonatal candidemia is poorly understood and is a leading cause of nosocomial infectious mortality in the nursery. Prevention of candidemia has been difficult, although a combined approach of antifungal prophylaxis and targeted empirical therapy may eventually reduce morbidity and mortality. Multicenter prospective testing of an integrated approach to early diagnosis of neonatal candidemia using newer molecular techniques is also needed. Candidemia in the infant is cause for prompt removal (or replacement) of central vascular catheters and institution of antifungal therapy. End-organ evaluation is also probably warranted to guide treatment and facilitate prognostication. Given the continuing progress in clinical research infrastructure and development of new diagnostic tests and antifungal agents, substantial improvement in the prevention, diagnosis, and management of neonatal candidemia is plausible over the next decade.

Neonatal candidemia and end-organ damage: a critical appraisal of the literature using meta-analytic techniques

OBJECTIVE

Neonatal candidemia is an increasing cause of infant morbidity and mortality. We evaluated the current medical literature in an effort to critique the literature and to document the reported prevalences of end-organ damage after neonatal candidemia.

METHODS

We analyzed all peer-reviewed articles of neonatal candidemia published in the English language; inclusion criteria included a cohort limited to all neonatal intensive care unit admissions or all episodes of candidemia in neonates. Articles that also incorporated older patients, did not define a numerator and a denominator for at least 1 form of end-organ damage, included patients from other reports, or did not include all episodes of candidemia in the source population were excluded from the analysis.

RESULTS

Thirty-four articles reported episodes of candidemia and mortality; 21 articles reported prevalence for at least 1 form of end-organ damage. Only 4 (19%) of 21 articles reported prevalence for >4 forms of end-organ damage from the following list: endophthalmitis, meningitis, brain parenchyma invasion, endocarditis, renal abscesses, positive cultures from other normally sterile body fluids, or hepatosplenic abscesses. The median reported prevalence of endophthalmitis was 3% (interquartile range [IQR]: 0%-17%), of meningitis was 15% (IQR: 3%-23%), of brain abscess or ventriculitis was 4% (IQR: 3%-21%), of endocarditis was 5% (IQR: 0%-13%), of positive renal ultrasound was 5% (IQR: 0%-14%), and of positive urine culture was 61% (IQR: 40%-76%). The medical literature concerning end-organ evaluation after episodes of neonatal candidemia is heterogeneous and consists largely of single-center retrospective studies. Year that the data were collected and prevalence of neonates infected with Candida albicans were associated with observed heterogeneity.

CONCLUSIONS

Given the heterogeneity of the medical literature, precise estimates of the frequencies of end-organ damage are not possible and a prospective multicenter trial is warranted, but the data from the published literature suggest that the prevalence of neonates with end-organ damage not only is greater than 0 but also is high enough that until such a prospective trial is completed, end-organ studies should be considered before the conclusion of antifungal therapy.

Empirical therapy for neonatal candidemia in very low birth weight infants

OBJECTIVE

Neonatal candidemia is often fatal. Empirical antifungal therapy is associated with improved survival in neonates and patients with fever and neutropenia. Although guidelines for empirical therapy exist for patients with fever and neutropenia, these do not exist for neonates.

METHODS

A multicenter, retrospective, cohort study was conducted of neonatal intensive care unit patients (N = 6172) who had a blood culture (N = 21,233) after day of life 3 and whose birth weight was <or=1250 g. We performed multivariable conditional logistic regression of risk factors for candidemia. From the regression modeling coefficients, we developed a candidemia score.

RESULTS

In multivariable modeling, thrombocytopenia (odds ratio [OR]: 3.56; 95% confidence interval [CI]: 2.68-4.74) and cephalosporin or carbapenem use in the 7 days before obtaining the blood culture (OR: 1.77; 95% CI: 1.33-2.29) were risk factors for subsequent candidemia. Children who were 25 to 27 weeks' estimated gestational age (OR: 2.02; 95% CI: 1.52-3.05) and children who were born at <25 weeks (OR: 4.15; 95% CI: 3.12-6.29) were at higher risk of developing candidemia than were children who were born at >or=28 weeks. We developed a candidemia score on the basis of the ORs from the multivariable model. Children with a candidemia score >or=2 points were classified as having a "positive" score, and a score of >or=2 points had a sensitivity of 85% and a specificity of 47%.

CONCLUSIONS

We developed a clinical predictive model for neonatal candidemia with high sensitivity and moderate specificity for candidemia. On the basis of our model, when a physician obtains a blood culture, the physician should consider providing antifungal therapy to neonates who are <25 weeks' estimated gestational age and to neonates who have thrombocytopenia at the time of blood culture. In addition, if a physician obtains a blood culture from a child who is 25 to 27 weeks' estimated gestational age and is not thrombocytopenic but has a history of third-generation cephalosporin or carbapenem exposure in the 7 days before the blood culture, then the physician should consider administration of empirical antifungal therapy.

Delivery of glucose-6-phosphatase in a canine model for glycogen storage disease, type Ia, with adeno-associated virus (AAV) vectors

Therapy in glycogen storage disease type Ia (GSD Ia), an inherited disorder of carbohydrate metabolism, relies on nutritional support that postpones but fails to prevent long-term complications of GSD Ia. In the canine model for GSD Ia, we evaluated the potential of intravenously delivered adeno-associated virus (AAV) vectors for gene therapy. In three affected canines, liver glycogen was reduced following hepatic expression of canine glucose-6-phosphatase (G6Pase). Two months after AAV vector administration, one affected dog had normalization of fasting glucose, cholesterol, triglycerides, and lactic acid. Concatamerized AAV vector DNA was confirmed by Southern blot analysis of liver DNA isolated from treated dogs, as head-to-tail, head-to-head, and tail-to-tail concatamers. Six weeks after vector administration, the level of vector DNA signal in each dog varied from one to five copies per cell, consistent with variation in the efficiency of transduction within the liver. AAV vector administration in the canine model for GSD Ia resulted in sustained G6Pase expression and improvement in liver histology and in biochemical parameters.

Facial cellulitis in childhood: a changing spectrum

BACKGROUND

Before conjugated Haemophilus influenzae type b (Hib) vaccination, a syndrome known as buccal cellulitis, usually caused by Hib and often accompanied by bacteremia, was seen. We investigated the incidence and cause of facial cellulitis at our hospital during the 10 years before and the 10 years after introduction of the vaccine.

METHODS

Records of patients discharged with a diagnosis of facial cellulitis or infections of the oral cavity were reviewed. Fisher's exact test was used to compare rates of cellulitis during the two decades.

RESULTS

Trauma was the most common antecedent to facial cellulitis in both eras. Buccal cellulitis accounted for 7/25 (28%) of cases before Hib vaccination and 0/19 cases since. Pneumococcal buccal cellulitis was not seen in either decade.

CONCLUSIONS

Buccal cellulitis due to Hib is a disappearing disease. Eighty-nine percent of recent inpatient cases of childhood facial cellulitis were related to trauma, tooth problems, or severe sinusitis. Facial cellulitis due to S pneumoniae is rare.

Pediatric residents' clinical diagnostic accuracy of otitis media

OBJECTIVE

Pediatric resident physicians' clinical diagnostic accuracy of otitis media is unknown. We attempted to correlate the clinical examination of pediatric house staff with pediatric otolaryngologists and tympanometry.

METHODS

Pediatric residents evaluated patients who were scheduled in the pediatric acute care clinic and completed a provider examination form detailing their otoscopic findings, interpretation, and treatment plan. Patients were then immediately reevaluated by a pediatric otolaryngologist using an identical form. Tympanometry was also performed by a pediatric audiologist. We used kappa statistics to calculate correlation of clinical findings and interpretation.

RESULTS

A total of 103 patients consented for the study; 70 patients were examined by 27 different pediatric residents with 43 patients (86 ears) examined by all 3 providers. Correlation of clinical findings between all pediatric residents and the pediatric otolaryngologists was a kappa statistic of 0.30 (fair agreement). The individual diagnostic finding with the greatest correlation was tympanic membrane erythema (kappa statistic: 0.40 [fair agreement]), and the worst correlate was tympanic membrane position (kappa statistic: 0.16 [slight agreement]). Resident interpretation and tympanometry yielded a kappa statistic of 0.20 (slight agreement), and the otolaryngologist interpretation and tympanometry yielded a kappa statistic of 0.32 (fair agreement).

CONCLUSIONS

Otitis media is the most common disease seen by practicing general pediatricians, but there is a paucity of formalized resident education. We demonstrated only a slight to moderate correlation between the clinical diagnostic examinations of pediatric residents and pediatric otolaryngologists and tympanometry.

Thirty-day postoperative death rate at an academic medical center

OBJECTIVE

To improve understanding of perioperative deaths at an academic medical center.

SUMMARY BACKGROUND DATA

Because published data have typically focused on specific patient populations, diagnoses, or procedures, there are few data regarding surgical deaths and complications in institutional or regional studies. Specifically, surgical adverse events and errors are generally not studied comprehensively. This limits the overall understanding of complications and deaths.

METHODS

Data from all operations performed in the main operating suite of the University of Virginia Health Sciences Center from January 1 to June 30, 1999, were compared with state death records to gain a dataset of patients dying within 30 days of surgery. All clinical records from patients who died were screened for adverse events and subsequently reviewed by three surgeons who identified adverse events and errors and performed comparisons with survivors.

RESULTS

One hundred nineteen deaths followed 7,379 operations performed on 6,296 patients, yielding a patient death rate of 1.9%. Patients dying within 30 days of surgery were older and had higher American Society of Anesthesiologists scores. Of 119 deaths, 86 (72.3%) were attributable to the patient's primary disease. Twenty-three patient deaths (19.3% of all deaths, 0.37% of all patients) could not be attributed to the patient's primary disease and thus were suspicious for an adverse event (AE) as the cause of the death. Of the 23 deaths suspicious for AE, 15 (12.6% of all deaths, and 65.2% of AE deaths) followed an error in care and thus were classified as potentially preventable, affecting 0.24% of the study population.

CONCLUSIONS

Overall, the 30-day postoperative death rate was low in the total surgical population at an academic medical center. Errors and AEs were associated with 12.6% and 19.3% of deaths, respectively. Retrospective review inadequately characterized the nature of AEs and failed to determine causality. Prospective audits of outcomes will enhance our understanding of surgical AEs.

Integration of statistical theory and practical clinical expertise. Polymerase chain reaction testing of the HIV-exposed infant

Testing of the human immunodeficiency virus (HIV)-exposed infant has improved markedly over the past decade. Polymerase chain reaction (PCR) technology has made accurate diagnosis possible by 4 months of age and improved sensitivity and specificity of PCR testing has obviated the need for serologic follow-up for most HIV-exposed infants. Clinicians may use PCR testing and simple statistical theory to develop a rational algorithm for diagnosis of HIV-exposed infants. Physicians should determine whether the infant is at low, moderate, or high risk for acquiring HIV. After risk-stratification the physician may proceed with 2 PCR tests, 3 PCR tests, or PCR testing and serologic follow-up. Infants of mothers who are on highly active antiretroviral therapy (HAART) at delivery, whose mothers have a low or undetectable viral load at delivery, and who do not breast-feed, should be considered low-risk. These low-risk infants should have 2 PCR tests, 1- and 4-months post partum. Infants whose mothers have an unknown or high viral load at delivery and who do not breast-feed should be considered moderate-risk. These infants should have 3 negative PCR tests, 1 during the first month of life, 1 after the first month of life, and 1 after 4 months of life. Infants who breast-feed should be considered high-risk and require at least 3 negative PCR tests, PCR testing every 3 months until breast-feeding stops, and serologic follow-up. Any positive PCR test requires virologic confirmation and serologic follow-up.

Rational testing of the HIV-exposed infant

OBJECTIVES

The objectives of this study were 1) to evaluate testing regimens of human immunodeficiency virus (HIV)-exposed infants and 2) to determine optimal methods of follow-up by enzyme-linked immunosorbent assay (ELISA) testing.

METHODS

We reviewed the results from 742 HIV-exposed infants in the state of North Carolina; 2474 samples were tested for HIV by DNA polymerase chain reaction (PCR) at the University of North Carolina Retrovirology Core Laboratory. We then reviewed the utility and costs of ELISA testing of all HIV-exposed infants who were seen at the Duke University Pediatric Infectious Disease Clinic between January 1, 1993, and May 5, 1998. We used likelihood ratios to model probability of HIV infection given 3 negative DNA (PCR) tests and to provide recommendations on the use of ELISA follow-up.

RESULTS

The overall sensitivity of the DNA PCR was 87.1%, and its specificity was 99.9%. We evaluated 224 HIV-exposed infants who were seen at Duke University and who had at least 3 negative diagnostic tests using either DNA PCR tests or HIV blood cultures. All 178 infants who subsequently underwent ELISA testing ultimately demonstrated seroreversion. The Duke University Pediatric Infectious Disease Clinic transferred the care of 65 patients to primary care physicians before ELISA testing and retained the care of the remaining 159 patients. Children who remained in Duke's care were more likely to have documentation of seroreversion (158 of 159 vs 20 of 65). We reviewed costs of travel, physician appointment, and HIV antibody testing in a tertiary care setting. Given 3 negative PCR tests, the expected cost per case of HIV detected by a positive ELISA assay is $23.8 million.

CONCLUSIONS

Documentation of seroreversion in this cohort was nearly complete in the multidisciplinary subspecialty clinic but not when such responsibility was left to the primary care physician. Given the low probability of disease in patients who have had 3 negative PCR tests, documentation of a negative ELISA may not be an appropriate use of medical resources.

Bacteremia, central catheters, and neonates: when to pull the line

OBJECTIVES

Physicians who treat neonates who become bacteremic while dependent on central venous catheters face a serious and common dilemma. We sought 1) to evaluate the relationship between central venous catheter removal and outcome in bacteremic neonates, 2) to determine species of bacteria that are associated with an increased risk of infectious complications if the central catheter is not removed promptly, and 3) to provide evidence-based recommendations for central catheter management.

METHOD

A retrospective cohort study of all neonates who had central venous access and developed bacteremia between July 1, 1995, and July 31, 1999, was conducted in the Duke University neonatal intensive care unit.

RESULTS

The outcome for patients in whom the central catheter was not removed within 24 hours of organism identification was significantly worse (odds ratio = 9.8) than it was for those whose catheters were removed promptly. For patients who were infected with Staphylococcus aureus or with nonenteric Gram-negative rods, delayed removal of the central catheter was associated with complicated bacteremia. Catheter sterilization was attempted in 27 neonates who were infected with enteric Gram-negative rods; only 10 of these infants retained their catheters without infection-related complications. Infants who had 4 consecutive blood cultures that were positive for coagulase-negative staphylococcus (CoNS) were at significantly increased risk for end-organ damage and death, compared with infants who had 3 or fewer positive blood culture for CoNS (odds ratio = 29.58).

CONCLUSIONS

Bacteremic infants experienced fewer infection-related complications when the central catheter was removed promptly. One positive blood culture for S aureus or a Gram-negative rod warrants central line removal in a neonate. Clinicians who are faced with a neonate who has 1 positive culture for CoNS may attempt medical management without central catheter removal, but documentation of subsequent negative blood cultures is crucial. Once a neonate has 3 positive blood cultures for CoNS, the central catheter should be removed.central line, neonate, bacteremia, bacteria, umbilical catheter, Broviac, percutaneous.

When to suspect fungal infection in neonates: A clinical comparison of Candida albicans and Candida parapsilosis fungemia with coagulase-negative staphylococcal bacteremia

OBJECTIVES

To determine the epidemiology of candidemia in our neonatal intensive care unit; to compare risk factors, clinical presentation, and outcomes for neonates infected with Candida albicans, Candida parapsilosis, and coagulase-negative staphylococcus (CoNS); and to suggest a rational approach to empiric antifungal therapy of neonates at risk for nosocomial infection.

DESIGN

Retrospective chart review of all neonatal intensive care unit patients with systemic candidiasis or CoNS infection between January 1, 1995 and July 31, 1998 at Duke University Medical Center.

RESULTS

Fifty-one patients were reviewed. Nine of 19 patients infected with C parapsilosis and 5 of 15 patients infected with C albicans died of fungemia. Seventeen neonates had >2 positive cultures for CoNS obtained within 96 hours and 1 died. There was no statistically significant difference in birth weight, gestational age, or age at diagnosis between patient groups; however, candidemic patients had a sevenfold higher mortality rate. Before diagnosis, candidemic patients had greater exposure to systemic steroids, antibiotics, and catecholamine infusions. Of the 51 patients, 32 received third-generation cephalosporins in the 2 weeks before diagnosis and 19 did not. Twenty-nine of the 32 who were treated with third-generation cephalosporins subsequently developed candidemia, while candidemia occurred in only 5 of 19 patients who were not treated with cephalosporins. At the time of diagnosis, candidemic patients were more likely to have required mechanical ventilation and were less likely to be tolerating enteral feeding. Multivariate clustered logistic regression analysis revealed that candidemic patients had more exposure to third-generation cephalosporins. Once the clinician was notified of a positive blood culture for Candida, patients infected with C parapsilosis retained their central catheters longer than patients infected with C albicans.

CONCLUSIONS

In this retrospective review, we were able to identify aspects of the clinical presentation and medication history that may be helpful in differentiating between candidemia and CoNS bacteremia. Those key features may be used by clinicians to initiate empiric amphotericin B therapy in premature neonates at risk for nosocomial infections. Prolonged use of third-generation cephalosporins was strongly associated with candidemia. There was no statistically significant difference in the morbidity and mortality between patients infected with C parapsilosis and those infected with C albicans. Observed delays in removal of the central venous catheter may have contributed to finding a mortality rate from C parapsilosis that was higher than was previously reported.

Candidal mycetoma in the neonatal kidney

OBJECTIVE

To determine the natural history of renal mycetoma (fungal balls) in the neonate.

DESIGN

Retrospective chart review of all neonatal intensive care unit patients with systemic candidiasis and sonographic evidence of renal mycetoma admitted to the Duke University Medical Center between January 1, 1993, and July 1, 1998.

RESULTS

Fourteen patients were reviewed. Three died from fungemia, and 3 died from other causes months after completing treatment. Ten patients had urine cultures obtained within 1 week of diagnosis; each had a positive routine or fungal urine culture for candida. The rate of improvement of renal mycetoma by ultrasound was variable, ranging from 10 days to 4 months and was not predictive of survival or long-term renal function. All patients who were discharged from the hospital had creatinine </=0.5 mg/dL on discharge. Only 1 patient had surgical intervention (nephrostomy tube placement). Of the 11 patients who survived fungemia, 7 were treated for 3 weeks from the time negative cultures were obtained, while 4 were treated for 5 weeks or more after negative cultures. A declining platelet count was suggestive of fungemia in the patients we reviewed.

CONCLUSIONS

For our patients with renal mycetoma without complete obstruction (patients continued to have urine output) surgical intervention was rarely necessary, the rate of sonographic improvement neither correlated with clinical course nor necessitated longer therapy, and long-term creatinine levels were normal. Sustained declines in platelet count of 10% per day or more in a neonate on broad-spectrum antibiotics for suspected sepsis may be indicative of fungemia.