Yield of Diagnostic Testing in Infants Who Have Had an Apparent Life-Threatening Event

Study Site

The Maria Fareri Children's Hospital at Westchester Medical Center is a 114-bed tertiary care referral hospital located 20 miles north of New York City in Westchester County. The medical center is the major academic teaching affiliate of New York Medical College. It serves patients who represent a broad demographic and socioeconomic spectrum and a geographic region that includes urban, suburban, and rural areas in the Hudson Valley of New York State, northern New Jersey, western Connecticut, and New York City.

Patients

This report is based on data from a consecutive series of infants who were younger than 12 months and were admitted to the Children's Hospital between November 1, 1996, and June 30, 1999, after presenting to the hospital emergency department because of ALTE symptoms. These data were gathered as part of a larger study to investigate the possible causes of ALTEs and their relative frequencies.³ The 32-month data-collection period was expected to generate a sample size that would be sufficient to yield reasonably stable estimates of the proportions of ALTEs attributable to various causes.

To be enrolled in the study, an infant must have experienced the sudden occurrence of 1 or more of the following symptoms (from the initial history) that frightened the caregiver enough to seek medical attention:

• breathing irregularity (eg, apnea, labored or shallow breathing, choking, gagging)

• color change (eg, cyanosis or pallor)

• altered muscle tone or mental status (eg, hypotonia, hypertonia, clonic movements, unresponsiveness)

These enrollment criteria constituted our working definition of an ALTE. The criteria are consistent with the official definition established in 1987 by the National Institutes of Health Consensus Development Conference on Infantile Apnea and Home Monitoring.¹ When a patient was admitted for an ALTE more than once during the study period, only the first admission was included.

Data Collection

A physician data manager visited the pediatric floor of the hospital and the 14-bed PICU on a daily basis to examine the histories and admitting diagnoses in the charts of all newly admitted infants to identify study subjects. Patients who met the above criteria were enrolled in the study and followed until discharge.

A standardized data-collection instrument was used to record sociodemographic data, the initial history and physical examination, diagnostic testing done in the hospital, the hospital course, the final diagnosis, and the discharge disposition. The data manager obtained this information by reviewing the hospital chart and, when necessary, obtaining clarification by speaking with the patient's physicians. She collected the admission data within 24 hours of admission and updated the information on a daily basis until discharge. The study was approved by the institutional review board of New York Medical College.

Analysis

The analysis considered tests of blood and other body fluids, imaging studies, and monitoring tests such as electrocardiogram and pH probe that physicians ordered and might have considered useful in helping to identify the cause of an ALTE. The analysis did not consider tests that may have been ordered for other reasons but that would have had no likely connection to a known cause of an ALTE (eg, certain components of a comprehensive metabolic panel, such as serum cholesterol). For each patient, we noted whether a test was performed, whether the result was positive, and, if so, whether the test would have contributed to the diagnosis, that is, suggested or helped establish a cause. The definition of a positive laboratory test was based on age-specific ranges of normal from The Harriet Lane Handbook, 16th Edition.⁴ A culture was considered positive when it grew a clinically significant pathogen—contaminant species were excluded. An imaging or monitoring study was considered positive on the basis of the interpretation of the responsible subspecialist (eg, a radiologist for a neuroimaging study, a pediatric cardiologist for an echocardiogram). When a test was ordered more than once (eg, hemoglobin), we classified the test as positive when any result was positive.

The analysis also considered the role of the initial history and physical examination in guiding the subsequent evaluation. We classified each patient into 1 of 4 categories according to the sources of data that contributed to the diagnosis: history and physical examination, diagnostic testing, both, or neither (Table 1):

1. Diagnosis was based on tests ordered to investigate positive findings from the history and/or physical examination that suggested a cause (ie, the history/physical examination and diagnostic tests both were contributory).

2. Diagnosis was based on the history and/or physical examination alone (ie, the history/physical examination was contributory, but all diagnostic tests were noncontributory).

3. Diagnosis was based on positive tests obtained after the history and physical examination failed to suggest a cause (ie, the history and physical examination were noncontributory, but tests were contributory).

4. Cause unknown (ie, the history, physical examination, and tests all were noncontributory).

A contributory history included 1 or more positive findings from the history of present illness or medical history that could have had a connection with the final diagnosis (eg, 28 weeks' gestation in a patient with apnea of prematurity). A contributory physical examination included 1 or more physical findings suggestive of the final diagnosis (eg, congestion and cough in a patient with a final diagnosis of pneumonia). The reason for using this classification is that a physician's approach to an ALTE patient is likely to depend on results of the physician's initial clinical assessment.

Overview of Patients and Tests

A total of 243 patients met the enrollment criteria during the 32-month study (Table 2). Five patients had multiple ALTE admissions during the study period (4 patients had 2 admissions each, and 1 patient had 3 admissions). Only data from the first admission were used in the present analysis, as explained in Methods.

Of the 32 tests examined (Table 3), patients received 15.5 ± 5.2 (mean ± SD) tests (range: 0–26), 2.8 ± 2.2 of which were positive (range: 0–14). Because some blood tests are ordered in panels, the number of tests per patient reported here exceeds the number of separate test requisitions.

Yield of Diagnostic Testing

For each of the 32 tests, Table 3 shows the number of tests performed, number of positive tests, and number and percentage of positive tests (and of all tests) that contributed to the diagnosis. Table 3 also lists the diagnoses to which each positive test contributed.

Of the 669 positive tests, 224 (33.5%) contributed to the diagnosis that was believed to have caused the ALTE. As a percentage of all tests performed, these “contributory” tests accounted for 5.9% of the total (224 of 3776; Table 3).

A positive test was regarded as having contributed to the final diagnosis when there was any reasonable link, direct or indirect. For example, 1 patient whose final diagnosis of febrile seizure was believed to have caused the ALTE also had a urinary tract infection. In this case, the positive urine culture was considered to have contributed to the final diagnosis because the urinary tract infection would have caused the fever that caused the seizure. A positive test was considered noncontributory only when the abnormal result would clearly have had no connection to the ALTE (eg, low hemoglobin in a patient with bronchiolitis). We did not distinguish between a “weakly” positive test and a “strongly” positive test. For example, we did not consider how far above or below the range of normal a laboratory test fell in deciding whether the test was contributory. The above approach made us err on the side of calling a test contributory.

Several noncontributory positive tests in Table 3 warrant clarification. The 16 noncontributory positive metabolic screens (item 13) were transient abnormalities that were not linked to the cause of the ALTE. The noncontributory positive tests for infections (items 15, 17, and 20) include 5 urine cultures, 1 cerebrospinal fluid culture, and 2 respiratory syncytial virus (RSV) antigen tests. Each of the 5 patients with a noncontributory positive urine culture had another, unrelated condition (eg, gastroesophageal reflux, infantile spasms) that was believed to have been a more likely cause of the ALTE. Hence, the positive culture had no link to the final diagnosis associated with the ALTE. The noncontributory positive cerebrospinal fluid culture involved a patient whose specimen grew Streptococcus viridians and who also had pneumonia, considered to have been the more likely cause of the ALTE. Of the 2 infants with a noncontributory positive RSV antigen test, the first had Escherichia coli sepsis with hemodynamic instability and respiratory failure that required intubation. The patient was also RSV positive, but the presenting ALTE symptoms were attributed to the sepsis. The second infant was tested for RSV on 3 occasions over 4 days. The first and third tests were negative, and the second test was weakly positive. Because of the equivocal RSV test results and paucity of respiratory findings during hospitalization, the treating physicians did not consider the evidence sufficient to attribute the ALTE to RSV infection. Their discharge diagnosis was ALTE of unknown cause. In neither of these cases, therefore, was there a link between a positive RSV test and a diagnosis associated with the patient's ALTE.

The “c/b” and “c/a” columns in Table 3 measure the informational value of a test in 2 different ways. The former ratio provides an estimate of the probability that a positive test result would help to establish the cause, a concept similar to the positive predictive value used by clinical epidemiologists.⁵ The latter ratio provides an estimate of the pretest probability that a given test would prove useful in helping to establish the cause.

Contributions of the History, Physical Examination, and Diagnostic Testing

In 49% (119 of 243) of study patients, positive findings from the initial history and/or physical examination suggested a cause that was confirmed by subsequent testing (Table 1, category 1). For example, an 8-month-old boy who presented with respiratory distress, wheezing, cough, and a sudden episode of apnea had hyperinflation on chest radiograph and a positive RSV antigen test. These findings suggested RSV bronchiolitis as the probable cause of the ALTE. In 21% (52 of 243) of patients, the diagnosis was based on the history and physical examination alone, all tests having been noncontributory (category 2). For example, a 5-week-old girl with stridor and grunting presented with sudden apnea and received a diagnosis of laryngomalacia on the basis of classic breathing pattern and a negative workup for infectious, cardiac, and gastrointestinal causes. In 14% (33 of 243) of patients, the diagnosis was based on positive tests obtained after a noncontributory history and physical examination (category 3). For example, a 9-week-old boy with a 10-second episode of limpness and cyanosis was well appearing at presentation. A positive pH probe and otherwise normal comprehensive workup led to the diagnosis of gastroesophageal reflux. The remaining 16% (39 of 243) of patients were discharged without a diagnosis (category 4). For example, a 4-month-old girl with acute apnea, cyanosis, and limpness appeared well on presentation, and the entire workup was negative.

Table 4 shows the distribution of final diagnoses by category. In 119 patients, the initial history and physical examination prompted 1 or more tests that contributed to 22 different final diagnoses. In 52 patients, 16 different diagnoses were based on the history and physical examination alone. The 33 patients with a noncontributory history and physical examination but contributory testing included 27 patients with a final diagnosis of gastroesophageal reflux by pH probe or upper gastrointestinal series, 3 patients with urinary tract infections confirmed by culture, 1 patient with pneumonia by chest radiograph, 1 patient with periodic breathing demonstrated by pneumogram, and 1 patient with birth trauma identified by head computed tomography scan and brain MRI. This set of tests seems to have identified the occult causes of an ALTE without overlooking any serious disorders. Table 5 and Fig 1 show test yields for patients who had a contributory (categories 1–2) versus noncontributory (categories 3–4) history and physical examination.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Ranking of diagnostic tests on the basis of the percentage of tests that contributed to establishing the diagnosis. Numbers above the bars indicate contributory tests (top number) and total tests performed (bottom number).

Diagnostic Testing Strategy

In patients with suggestive findings from the history or physical examination (categories 1 and 2), those findings will necessarily guide the ensuing workup.² For example, positive respiratory findings would prompt a comprehensive evaluation for pulmonary pathology. Similarly, a history of maternal miscarriages would prompt a metabolic and genetic evaluation. That 70% of our study patients had a contributory initial clinical assessment underscores the importance of a thorough history and physical examination.

In patients with nonsuggestive initial findings (categories 3 and 4), the physician must consider the likelihood of a given cause in such a patient, the invasiveness and cost of various tests, and the potential harm that could result from missing or delaying a diagnosis. These patients present the greatest challenge to the physician.^6,7 They are also the patients for whom the results of the present investigation may offer the greatest benefit by highlighting tests that are likely to yield useful information. Our study found a subset of tests that led to the identification of all occult causes of an ALTE occurring in our series:

• screening for gastroesophageal reflux

• urine analysis and culture

• brain neuroimaging

• pneumogram

• white blood cell count

The absence of yield from the remaining tests is striking (Table 5, categories 3–4). In particular, broad evaluations for systemic infections, metabolic diseases, and blood chemistry abnormalities were not productive in this group of patients. Nevertheless, the need to supplement the above short list of tests must still be considered on an individual basis and dictated by clinical judgment. The zeros in Table 5 demonstrate that many tests have a low yield in category 3 and 4 patients, but larger samples will be required to determine the precise yields of these low-yield tests.

Although our series did not include any unsuspected cases of child abuse, we and others have previously reported ALTEs that presented without suggestive findings but were eventually linked to abusive head injury.^8–10 The potentially devastating consequences of missing such a diagnosis underscores the need for brain neuroimaging in the evaluation of ALTE patients with nonsuggestive initial presentations.

By presenting empirical estimates of diagnostic yield in a comparative format, Fig 1 can help clinicians choose tests on the basis of informational value. It is important to emphasize, however, that the rank order of tests in Fig 1 does not suggest a simple formula for deciding which tests to order and when to order them. Those decisions must take into account, in addition to the yield of a test, its invasiveness and cost in relation to the potential harm that could result from missing or delaying a diagnosis. For example, although “CSF culture” ranks near the bottom of the list of 23 tests in Fig 1A, this certainly does not suggest that a contributory rate of 3.8% (3 of 79) is too low to warrant obtaining the test. A physician who is concerned about the possibility of bacterial meningitis, even if he or she believes the likelihood to be <5%, must consider performing a lumbar puncture because of the potentially grave consequences of missing this diagnosis.

Previous Research on Diagnostic Testing

Three previous ALTE investigations have reported positivity rates (percentages of ordered tests that were positive). A prospective study of 69 infants who had an ALTE and underwent a standardized testing protocol¹¹ reported the positivity rate for each of the 17 tests included in the protocol: radioisotope milk scan, lactate, chest radiograph, white blood cell count, bicarbonate, urine culture, hemoglobin, postnasal swab, ammonia, metabolic screen, urine toxicology screen, sodium, potassium, urea, creatinine, and glucose. For patients with a nonsuggestive initial history and physical examination, the authors recommended 9 tests. Our data provide support for 2 of these (urine microscopy and chest radiograph), but 6 of their recommended tests (hemoglobin, glucose, metabolic screen, electrocardiogram, lactate, and pertussis testing) were not helpful in establishing a diagnosis in our category 3 and 4 patients. We did not have the data needed to evaluate their ninth recommended test (toxicology screen).

A retrospective review of 130 infants with an ALTE¹² reported positivity rates for the most commonly performed investigations: radioisotope milk scan, chest radiograph, blood count and chemistries, electroencephalogram, skull radiograph, and urine dipstick. The authors noted wide variation in the number of diagnostic tests performed in well-appearing infants (“from none to many”), an observation that underscores the need for a testing guideline to promote a systematic approach.

A prospective study of 128 infants with an ALTE¹³ reported positivity rates for toxicology screen, head computed tomography scan, and skeletal survey. This study focused exclusively on these tests because the aim of the study was to bring attention to inflicted injury as a possible cause.

A fourth investigation involving 150 ALTE patients¹⁴ noted the frequency with which specific tests were ordered in the emergency department but did not report positivity rates. The objective of that study was to ascertain the percentage of patients whose test results led to specific treatments in the emergency department or during hospitalization.

None of the above investigations reported the percentage of tests that were contributory, a more relevant measure of test performance when assessing diagnostic yield. Contributory rates, which are necessarily lower than positivity rates, are more relevant because they exclude false-positive findings and true-positive findings that have no relationship to the reason for the admission—ie, the ALTE.

Study Limitations