Objective. To determine reasons for and adequacy of follow-up testing of suspect results of metabolic screening in infants born in Alberta in 1992.
Study Design. Of 42 392 live births, 41 553 infants were deterministically matched using birth registry data. Infants requiring repeat analyses were determined from notes made on the screening report. Characteristics of infants needing repeat screening, and obtaining a repeat screen results, were determined by logistic regression using variables from the birth registry and the screening record.
Results. A total of 1375 infants required repeat screening. Infants with unsatisfactory samples were more likely to be born in a smaller community, of low birth weight, and to have the sample obtained after 7 days of age. Infants with biologically suspect results were more likely to be of low birth weight, to die in week 1 of life, and to be born in a large hospital. Repeat analyses were found for 663 infants. Boys, infants from smaller communities, and low birth weight infants were more likely to have the required repeat screening. Infants of single mothers were less likely to undergo repeat screening.
Conclusions. The results of this study demonstrate the need for a clear, time-oriented protocol of follow-up of newborn metabolic screening results.
Screening of newborns between 1 and 7 days of age detects certain metabolic diseases such as congenital hypothyroidism and phenylketonuria (PKU), which can impair the health and mental development of the child permanently if not diagnosed and treated promptly.1 Newborn screening in Alberta, Canada, started in 1967 and is administered through the Department of Laboratory Medicine of the Walter Mackenzie Health Sciences Centre (WMC) in Edmonton, Alberta. Infants are screened for congenital hypothyroidism, PKU, biotinidase deficiency, and tyrosinemia.
A companion publication2 addressed the issue of coverage of the Alberta program in 1992 and described the characteristics of children who were never screened. It showed that of the 42 392 infants born alive in Alberta in 1992, 40 593 were good matches (GM), 960 were possible matches (PM), and 839 infants were not matched (NM) and thus not screened. Factors found to be related independently to absence of screening were death in week 1 of life, low birth weight, birth out of hospital, and birth to a mother who was single or formerly married. This article addresses the timing of the initial screen, the reasons for follow-up testing, and the adequacy of follow-up of suspect results.
SUBJECTS AND METHODS
Normal Screening Protocol
Although guidelines exist for the provincial institution and management of a newborn screening program (NSP),3-5the current program lacks a well-defined protocol that ensures that all children will be screened and suspect results followed to resolution. In 1992 the Alberta program screened infants between 24 hours and 7 days of age. Infants screened before 24 hours underwent repeat analysis, preferably before 7 days of age. All samples were analyzed at the WMC. Normal results were mailed to physicians. Abnormal results usually were conveyed to the physician by telephone, with a request for a repeat sample. Atypical or less significant abnormal results were mailed to the physician with a request for a repeat sample. The names of infants needing repeat samples were logged and the name checked off when the repeat sample was obtained. The repeat log was, and is, a static document in that there was no time frame after which a reminder was sent to the physician to obtain a repeat sample.
All repeat analyses for PKU, biotinidase deficiency, or elevated tyrosine were performed at the WMC laboratory or at the Biochemical Genetics Laboratory at the Alberta Children's Hospital (ACH) in Calgary, Alberta. Repeat thyroid function tests were performed either as a repeat newborn screen or by serum analysis at a private laboratory. In the latter event, the screening program was not necessarily notified of the result. When samples analyzed at the ACH were identified as repeats or when an abnormal phenylalanine, biotinidase, or tyrosine result was obtained, a copy of the result was sent to the WMC. If the sample was not identified as a repeat and not otherwise identified as a newborn screen, no report would be sent to the WMC.
Data Preparation and Matching
Details of data preparation and the matching process have been reported.2 An infant required repeat analysis if 1) the first sample was obtained before the infant was 24 hours of age, 2) it was an unsatisfactory specimen, or c) there was an abnormal result. The physician was requested to obtain a repeat specimen. Records were identified as a repeat if they were labeled as a repeat or if multiple records were found with matching personal identifiers. Repeat records were linked by a unique identifying number to the initial record and by the temporal sequence of sampling determined. The search for repeat records stopped after finding the first repeat, although sometimes several repeats were found for an infant. For infants needing follow-up, the primary data were those of the NSP. Vital statistics data from the Government of Alberta were used to describe demographic characteristics of the infants.
Analysis was limited to the GM group. Data for the NM group are not applicable, and characteristics of the PM infants are unreliable. Summary statistics were obtained relating to time when blood samples were taken and number and type of results needing repeat analysis. Characteristics of infants requiring a repeat specimen and the way these characteristics varied with the reason for a repeat were compared with GM infants with normal results. Characteristics of infants who did not have a requested repeat screen obtained were compared with infants who did have a requested repeat screen. The statistical significance of categorical data was assessed using contingency table analysis and χ2 statistics, and of continuous data using analysis of variance with post hoc comparisons using Scheffé's procedure and a probability criterion of P < .05. Birth weight and maternal age were stratified for use in contingency table analysis. After initial exploration of the data, logistic regression was used to control for several factors simultaneously, using need for repeat analysis versus no need as one dependent variable and repeat analysis done versus no repeat done as another dependent variable. The statistical packages SPSS6 and STATA7 were used for the analysis.
There were 43 155 records of screen data; 43 125 from the WMC and 30 from ACH. Of these, 1602 were repeat records from 1485 infants; however, 850 records were for infants for whom repeat analysis was not requested. A repeat sample was required for 1375 GM infants. The reasons for repeat screens are outlined in Table1. Characteristics of infants needing repeat sampling varied with the reason for the repeat sample and are summarized in Table 2.
Logistic regression analysis of these data show that infants with nonsufficient or unsatisfactory specimens (NSQ/US) were more likely to be native, born in a smaller community, and of low birth weight; to have a mother who was formerly married; or to have the first sample taken after 7 days of age (Table 3). Infants with samples taken at <24 hours of age were more likely to be boys and of low birth weight. Infants with biologically suspect results were more likely to be of low birth weight, native, die in the first week of life, or to have been born in a hospital with >2000 births per year.
Blood samples were taken before 24 hours of age from 235 infants. Among these infants, 17 abnormal biochemical results, representing 7.2% of the records, were obtained. This figure is 2.5 times the proportion of abnormal results seen in infants first sampled after 24 hours of age. A high thyroid stimulating hormone (TSH) level was the most common abnormality; it was 30 times more common than that found among infants whose blood was taken after 24 hours.
Infants Obtaining the Requested Repeat
Repeat samples were found for only 663 of the 1375 GM infants requiring a repeat screen. Infants with birth weight <1500 g were more likely to have a repeat sample than were infants >2500 g (odds ratio [OR]: 3.11) (Table 4). Infants of single mothers were less likely than those of married mothers to get the necessary repeat specimen (OR: 0.64). Boys were more likely than girls to get the required repeat specimen (OR: 1.39), and infants born in communities where there were fewer than one birth per day were more likely to undergo the required repeat screen than were infants born in communities having more than 10 births per day (OR: 1.59).
The average age at repeat was 27.2 ± 22.4 (mean, SD) days, with a maximum of 270 days; 65% of the samples were obtained by 30 days of age (data not shown). All infants with high phenylalanine levels on initial screen underwent repeat screen at a mean age of 11.6 days and a maximum age of 23 days. Two had phenylketonuria and six had hyperphenylalaninemia. One child with hyperphenylalaninemia had two repeat samples obtained. Both results were abnormal, but the child was then lost to follow-up until tracked down to resolve inquiries related to this project. Review of the case shows that the child probably had a benign variant of hyperphenylalaninemia.
Biotinidase activity was reported initially as decreased in 72 infants. For 39 infants, repeat samples were obtained at an average age of 34.7 ± 18.6 days (range, 3 to 92 days). One infant was diagnosed with biotinidase deficiency.
For 713 GM infants, initial screens found high tyrosine levels. A repeat sample obtained after 1 month of age was requested, but only 315 (44.2%) repeat samples were found. The average age at repeat analysis was 34.5 ± 21.3 days (range, 2 to 236 days). Among the repeat screens, 14 samples showed a high tyrosine level; of these, 8 were repeated and found to be normal. A high tyrosine level often is seen in premature infants and, in this study, 163 (22.8%) of the GM infants with a high tyrosine level had birth weights <2500 g, whereas only 5.5% of all GM infants weighed <2500 g.
Initial TSH values were >25 mIU/L in 59 infants tested after 24 hours of age. Ten of 11 infants with a TSH >50 mIU/L were retested, some by private laboratories, and hypothyroidism was found in all. No record of repeat analysis was found for the 11th infant. Repeat analysis was found for only 12 of 47 infants whose TSH level was between 25 and 50 mIU/L.
There were 142 NSQ/US samples, of which 92 were rescreened at a mean infant age of 25.0 ± 28.6 days (range, 2 to 270 days). Relatively more of the NSQ/US samples came from centers where there was less than one birth a day or where the child was born out of hospital (probably a home birth).
In 235 instances, blood was obtained before 24 hours of age; however, only 109 had repeat analysis, after a mean interval of 9.1 ± 13.8 days (range, 1 to 91 days; median, 3 days). Of these 109 repeat screens, 7 showed abnormal results, and none of these have a record of additional repeat screening.
Infants Having a Repeat Sample That Was Not Indicated
For 884 infants, a repeat sample was obtained although the initial sample was normal. Among these 884, which were obtained at an average age of 20.5 days (median, 7 days), 33 had abnormal or NSQ/US results. Analysis of the data (not shown) suggests that these samples were taken from children who had neonatal problems. Nearly 28% had a birth weight <2500 g, but only 5% of GM children with no repeat had a birth weight <2500 g. These children may have been graduates of a newborn ICU who receive a second screen routinely before discharge.
One objective of this study was to determine the degree to which infants who underwent repeat screening were followed appropriately until a final disposition could be made. Consistent with newborn screening guidelines, the assumption made in this report is that a newborn with an initial abnormal screen result, an unsatisfactory sample, or a sample obtained before 24 hours requires a repeat sample. The speed and completeness of this process reflect the effectiveness of the entire NSP because these children are at greatest risk for metabolic disease.
In our analysis, repeat samples were detected as such because they were labeled repeat or because records with equivalent identifiers were found during the editing, visual-matching, or computer-matching procedures. Although some records considered repeats could have been initial records, this is unlikely because the method of matching for repeat screens was always deterministic and almost always verified visually.
Some repeat samples may have been missed because they were not labeled as neonatal screens, but were requested specifically for phenylalanine, tyrosine, or biotinidase measures. This seems unlikely because these determinations are performed by only two laboratories in the province, and the files of both laboratories were searched specifically for all test terms that would report these values. A repeat thyroid screen may have been missed when a serum sample was analyzed at a private laboratory.
Follow-up of repeat screens is an area in which the NSP did not function well. Only 48% of the infants who required it actually underwent repeat analysis. Most of these were infants who had a high tyrosine level. All infants with high phenylalanine had repeat samples, but only 10 of 11 infants with a TSH value of >50 mIU/L and only 12 of 47 infants with a TSH value between 25 and 50 mIU/L were known to have had a repeat analysis. Finally, among those infants who did have the required repeat analysis, the analysis was often obtained after 30 days of age, diminishing the potential benefits of early intervention.
Ensuring that required follow-up samples are obtained is a common problem among all newborn screening systems. In the United States, the Council of Regional Networks for Genetic Services, National Newborn Screening Report, 1992, reported that 3.4% of abnormal phenylalanine results and 1.7% of abnormal thyroid function test results were lost to follow-up.1 Data are not directly comparable with Alberta because of different diagnostic criteria, but given the very small sample, the Alberta data for these two disorders appear to be comparable. Ms J. Tuerck of the Oregon Screening Program and a member of the Maternal and Child Health Select Panel on Newborn Screening Systems, which has evaluated 14 states in the United States noted that “follow up problems seem to be common to all the states … follow up which works best seem to be those [sic] who have a designated person(s) at the program manager level, use medical consultants for infants with significant abnormal results, have written protocols which are followed and evaluated on a regular basis, and who have close effective communication with other parts of the program (lab and practitioners) and with other state or provincial programs” (J. Tuerck, April 11, 1996, personal communication).
Several explanations may account for Alberta's poor performance. There is no clear, time-oriented protocol to handle follow-up requests, and there are no personnel dedicated to this specific task. Although serious metabolic disease is being detected and clearly abnormal results are communicated by telephone to the physician, follow up is not well structured and the possibility of a “missed repeat and affected child” is real.
Repeat screens are not always labeled as such and thus may not be identified if patient demographics do not match the initial screen. Physicians may be unaware of what is good practice. For example, Sinai8 documented that many infants who were discharged by 24 hours of age did not undergo repeat analysis for PKU, apparently because pediatricians and other physicians were unaware of the guidelines for newborn screening and thus for the need to obtain repeat analysis. Some physicians, on reading the message in the screening report “suggest clinical follow-up and serum thyroid function tests” or “suggest repeat screen in about 1 month or sooner if clinically indicated”, may assume that if they think it is not clinically indicated, the screening need not be repeated. Unfortunately, these diseases cannot be excluded reliably on the basis of clinical impression only; nondisease requires laboratory verification.
Finally, analysis of some thyroid function samples could have been repeated in a private laboratory and no record of repeat analysis would be reported to the screening program. This is illustrated by the fact that the screening program had repeat records for only 24.1% of infants with high TSH values, whereas 37.3% of infants with high TSH values were known to have undergone repeat analysis.
In 142 cases, the sample was inadequate or unsatisfactory. Only 46.4% of these were repeated. Among the repeat analyses, there were several abnormal results that required additional verification. Where data are available, all eventually showed normal values. The higher rates of poor-quality samples from smaller centers may reflect inexperience on the part of personnel obtaining the sample. Very low birth weight infants were more likely than term infants to have NSQ/US samples.
A raised tyrosine level was the most common abnormal screen result, seen in 1.8% of those screened after 24 hours of age. Although the most common cause of a raised tyrosine level is transient tyrosinemia attributable to immaturity of the enzyme involved in its metabolism, tyrosinemia may have long-term effects for this small subset of infants.9
Among other requirements, an effective screening program must test all eligible subjects, ensure rapid follow-up and verification of suspect results, and promptly provide and institute definitive therapy.10 Our earlier report showed that 98% of infants in Alberta were screened in 1992 and described the characteristics of nonscreened infants. This report evaluates further other aspects of the performance of the Alberta Newborn Screening Program with explicit reference to GM infants who were screened. The most important finding is that nearly half of the required repeat screens were not performed. Review of other jurisdictions suggests that the best follow-up occurs when a protocol, detailing the procedures for the initial process of screening and the subsequent process of follow-up of abnormal or unsatisfactory results, is conceived and implemented. Such protocols appear to be most effective when there is an integrated system with designated responsibility and authority to ensure that the program runs effectively and efficiently.
- Received November 20, 1997.
- Accepted April 8, 1998.
Reprint requests to (D.W.S.) Department of Pediatrics, 2C3.62 WMC, University of Alberta, Edmonton, Alberta, Canada T6G 2R7.
- PKU =
- phenylketonuria •
- WMC =
- Walter Mackenzie Centre •
- GM =
- good match •
- PM =
- possible match •
- NM =
- not matched •
- NSP =
- newborn screening program •
- ACH =
- Alberta Children's Hospital •
- NSQ/US =
- nonsufficient quantity or unsatisfactory sample •
- TSH =
- thyroid stimulating hormone •
- OR =
- odds ratio
- ↵Newborn Screening Committee. The Council of Regional Networks for Genetic Services (CORN), National Newborn Screening Report—1992. Atlanta, GA: CORN; 1995
- ↵Spady DW, Saunders LD, Bamforth F. Who gets missed: coverage in a provincial newborn screening program. Pediatrics. 1998;102(2). URL: http://www.pediatrics.org/cgi/content/full/102/2/e21
- American Academy of Pediatrics, Committee on Genetics
- Gruters A,
- Delange F,
- Giovannelli G,
- et al.
- LaFranchi S,
- Dussault JH,
- Fisher DA,
- et al.
- ↵SPSS. SPSS for Windows. Release 6.1.1. Chicago, IL: SPSS Inc; 1994
- ↵Statacorp. Stata Statistical Software. Release 5.0. College Station, TX: Stata Corporation; 1997
- ↵Sinai LN, Kim SC, Casey R, Pinto-Martin JA. Phenylketonuria screening: effect of early newborn discharge. Pediatrics, 1995;96:605–608
- ↵Stanbury JB, Wyngaarden JB, Fredrickson DS. Tyrosinemia and related disorders. In: Scriver CR, Beaudet AL, Sly WS, Valle D, eds. The Metabolic Basis of Inherited Disease. 6th ed. New York, NY: McGraw-Hill; 1989:556
- ↵Wilson JMG, Jungner G. Principles and Practice of Screening for Disease. Geneva, Switzerland: World Health Organization; 1968
- Copyright © 1998 American Academy of Pediatrics