Using Natural Language Processing to Improve Discrete Data Capture From Interpretive Cervical Biopsy Diagnoses at a Large Health Care Organization.

CONTEXT.­: The terminology used by pathologists to describe and grade dysplasia and premalignant changes of the cervical epithelium has evolved over time. Unfortunately, coexistence of different classification systems combined with nonstandardized interpretive text has created multiple layers of interpretive ambiguity. OBJECTIVE.­: To use natural language processing (NLP) to automate and expedite translation of interpretive text to a single most severe, and thus actionable, cervical intraepithelial neoplasia (CIN) diagnosis. DESIGN.­: We developed and applied NLP algorithms to 35 847 unstructured cervical pathology reports and assessed NLP performance in identifying the most severe diagnosis, compared to expert manual review. NLP performance was determined by calculating precision, recall, and F score. RESULTS.­: The NLP algorithms yielded a precision of 0.957, a recall of 0.925, and an F score of 0.94. Additionally, we estimated that the time to evaluate each monthly biopsy file was significantly reduced, from 30 hours to 0.5 hours. CONCLUSIONS.­: A set of validated NLP algorithms applied to pathology reports can rapidly and efficiently assign a discrete, actionable diagnosis using CIN classification to assist with clinical management of cervical pathology and disease. Moreover, discrete diagnostic data encoded as CIN terminology can enhance the efficiency of clinical research.

Serum Fibrosis Marker Panels FIB-4 Index and Aspartate Aminotransferase (AST)-to-Platelet Ratio Index (APRI) Are Equivalent to AST Alone at Predicting Liver Fibrosis in a Cohort of 1731 Patients Infected with Hepatitis C Virus.

BACKGROUND: Efficient tools are needed to stage liver disease before treatment of patients infected with hepatitis C virus (HCV). Compared to biopsy, several studies demonstrated favorable performance of noninvasive multianalyte serum fibrosis marker panels [fibrosis-4 (FIB-4) index] and aspartate aminotransferase (AST)-to-platelet ratio index (APRI), but suggested cutoffs vary widely. Our objective was to evaluate FIB-4 index and APRI and their component tests for staging fibrosis in our HCV-infected population and to determine practical cutoffs to help triage an influx of patients requiring treatment. METHODS: Transient elastography (TE) results from 1731 HCV-infected patients were mapped to an F0-F4 equivalent scale. Each patient's APRI and FIB-4 index were calculated. Areas under the receiver operator curve (AUROCs) and false-positive and false-negative rates were calculated to retrospectively compare the performance of the indices and their component tests. RESULTS: The highest AUROCs for distinguishing severe (F3-F4) from mild-to-moderate (F0-F2) fibrosis had overlapping 95% CIs: APRI (0.77; 0.74-0.79), FIB-4 index (0.76; 0.73-0.78), and AST (0.74; 0.72-0.77). Cutoffs had false-negative rates of 2.7%-2.8% and false-positive rates of 6.4%-7.4% for all 3 markers. CONCLUSIONS: AST was as effective as FIB-4 index and APRI at predicting fibrosis. Published cutoffs for APRI and FIB-4 index would have been inappropriate in our population, with false-negative rates as high as 11%. For our purposes, no serum fibrosis marker was sufficiently sensitive to rule-out significant fibrosis, but cutoffs developed for AST, FIB-4 index, and APRI all had specificities of 79.2%-80.3% for ruling-in severe fibrosis and could be used to triage 1/3 of our population for treatment without waiting for TE or liver biopsy.

Risk of Sensorineural Hearing Loss and Bilirubin Exchange Transfusion Thresholds.

BACKGROUND AND OBJECTIVES: High bilirubin levels are associated with sensorineural hearing loss (SNHL). However, few large studies of relative and excess risk exist. We sought to quantify the risk of SNHL in newborns who had bilirubin levels at or above American Academy of Pediatrics exchange transfusion thresholds (ETT). METHODS: Infants born at ≥35 weeks gestation in 15 Kaiser Permanente Northern California hospitals from 1995-2011 were eligible (N = 525 409). We used a nested double cohort design. The exposed cohort included subjects with ≥1 bilirubin level at or above ETT. The unexposed cohort was a 3.6% random sample of subjects with all bilirubin levels below ETT (10 unexposed per exposed). An audiologist, blinded to bilirubin levels, reviewed the charts of children in whom SNHL had been diagnosed before age 8 years to confirm the diagnosis. We calculated Cox proportional hazard ratios for time to diagnosis of SNHL. RESULTS: SNHL was confirmed in 11 (0.60%) of the 1834 exposed subjects and in 43 (0.23%) of the 19 004 unexposed. Only bilirubin levels ≥10 mg/dL above ETT were associated with a statistically significant increased risk of SNHL (hazard ratio: 36 [95% confidence interval (CI): 13 to 101]). Likewise, only bilirubin levels ≥35 mg/dL were associated with a statistically significant increased risk of SNHL (hazard ratio: 91 [95% CI: 32 to 255]). For subjects with total serum bilirubin levels 0 to 4.9 mg/dL above ETT, the upper limit of the 95% CI for excess risk was 0.5%. CONCLUSIONS: Only bilirubin levels well above ETT were associated with SNHL. At lower bilirubin levels, the excess risk of SNHL was low.

Risk for cerebral palsy in infants with total serum bilirubin levels at or above the exchange transfusion threshold: a population-based study.

IMPORTANCE: Exchange transfusion is recommended for newborns with total serum bilirubin (TSB) levels thought to place them at risk for cerebral palsy (CP). However, the excess risk for CP among these infants is unknown. OBJECTIVE: To quantify the risks for CP and CP consistent with kernicterus that are associated with high TSB levels based on the 2004 American Academy of Pediatrics exchange transfusion threshold (ETT) guidelines. DESIGN, SETTING, AND PARTICIPANTS: We enrolled 2 cohorts from a population of 525,409 infants in the Late Impact of Getting Hyperbilirubinemia or Phototherapy (LIGHT) birth cohort. Eligible infants were born at a gestational age of at least 35 weeks at 15 hospitals within the Kaiser Permanente Northern California integrated medical care delivery system from January 1, 1995, through December 31, 2011. EXPOSURES: The exposed cohort included all 1833 infants with at least 1 TSB measurement at or above the ETT based on age at testing, gestational age, and results of direct antiglobulin testing. The unexposed cohort was a 20% random sample of 104 716 infants with TSB levels below the ETT. MAIN OUTCOMES AND MEASURES: A pediatric neurologist blinded to the TSB levels reviewed medical records to determine the presence of CP, defined as a nonprogressive congenital motor dysfunction with hypertonia or dyskinesia. Cerebral palsy was judged to be consistent with kernicterus if magnetic resonance imaging of the brain revealed bilateral globus pallidus injury in the setting of dyskinetic CP. RESULTS: We identified CP in 7 of 1833 exposed (0.4%) vs 86 of 104 716 unexposed (0.1%) infants (relative risk, 4.7 [95% CI, 2.2-10.0]). Absolute risk differences were 0.2% (95% CI, 0%-0.5%) for a TSB level 0 to 4.9 mg/dL above the ETT (n = 1705), 0.9% (95% CI, 0.1%-5.3%) for a TSB level 5.0 to 9.9 mg/dL above the ETT (n = 102), and 7.6% (95% CI, 2.1%-24.1%) for a TSB level 10 mg/dL or more above the ETT (n = 26). Cerebral palsy consistent with kernicterus occurred in 3 infants (incidence, 0.57 per 100,000 births); all 3 had TSB levels of more than 5.0 mg/dL above the ETT and at least 2 risk factors for neurotoxicity, such as prematurity, glucose-6-phosphate dehydrogenase deficiency, or hypoxia-ischemia. CONCLUSIONS AND RELEVANCE: Cerebral palsy consistent with kernicterus occurred only in infants with 2 or more risk factors for neurotoxicity and TSB levels of more than 5 mg/dL above the ETT. Among infants with lower degrees of TSB level elevation, the excess risk for CP is minimal.

Incidence, etiology, and outcomes of hazardous hyperbilirubinemia in newborns.

BACKGROUND AND OBJECTIVES: Total serum bilirubin (TSB) levels ≥ 30 mg/dL are rare but potentially hazardous. A better understanding of their incidence, causes, and outcomes could help inform preventive efforts. METHODS: We identified infants born ≥ 35 weeks' gestational age from 1995-2011 in Kaiser Permanente Northern California (n = 525409) and examined the medical records of infants with a TSB ≥ 30 mg/dL to determine etiology and the occurrence of acute bilirubin encephalopathy. We reviewed inpatient and outpatient encounters through 2013 for evidence of sensorineural hearing loss (SNHL) or cerebral palsy (CP). RESULTS: We identified 47 infants with TSB ≥ 30 mg/dL (8.6 per 100000 births). In 44 infants (94%), the hyperbilirubinemia occurred after the initial birth hospitalization. The etiology was not identified in 33 (70%). Glucose-6-phosphate dehydrogenase (G6PD) activity was measured in only 25 (53%) of whom 10 (40%) were deficient. Four children had acute bilirubin encephalopathy of whom 2 developed both CP and SNHL, and 1 developed isolated SNHL. These 3 infants all had G6PD deficiency and TSB >40 mg/dL. One additional 35-week infant with TSB 38.2 mg/dL had SNHL. CONCLUSIONS: Hazardous (≥ 30 mg/dL) hyperbilirubinemia is a rare event. No etiology could be identified from the clinical record in most cases. G6PD deficiency was the leading cause of hazardous hyperbilirubinemia when an etiology was identified, but many were not tested. Chronic, bilirubin-induced neurotoxicity was uncommon and occurred only in the setting of additional risk factors and TSB values well over (>15 mg/dL) the American Academy of Pediatrics exchange transfusion thresholds.

Combining immature and total neutrophil counts to predict early onset sepsis in term and late preterm newborns: use of the I/T2.

BACKGROUND: The absolute neutrophil count and the immature/total neutrophil ratio (I/T) provide information about the risk of early onset sepsis in newborns. However, it is not clear how to combine their potentially overlapping information into a single likelihood ratio. METHODS: We obtained electronic records of blood cultures and of complete blood counts with manual differentials drawn <1 hour apart on 66,846 infants ≥ 34 weeks gestation and <72 hours of age born at Kaiser Permanente Northern California and Brigham and Women's Hospitals. We hypothesized that dividing the immature neutrophil count (I) by the total neutrophil count (T) squared (I/T) would provide a useful summary of the risk of infection. We evaluated the ability of the I/T to discriminate newborns with pathogenic bacteremia from other newborns tested using the area under the receiver operating characteristic curve (c). RESULTS: Discrimination of the I/T (c = 0.79; 95% confidence interval: 0.76-0.82) was similar to that of logistic models with indicator variables for each of 24 combinations of the absolute neutrophil count and the proportion of immature neutrophils (c = 0.80, 95% confidence interval: 0.77-0.83). Discrimination of the I/T improved with age, from 0.70 at <1 hour to 0.87 at ≥ 4 hours. However, 60% of I/T had likelihood ratios of 0.44-1.3, thus only minimally altering the pretest odds of disease. CONCLUSIONS: Calculating the I/T could enhance prediction of early onset sepsis, but the complete blood counts will remain helpful mainly when done at >4 hours of age and when the pretest probability of infection is close to the treatment threshold.

Stratification of risk of early-onset sepsis in newborns ≥ 34 weeks' gestation.

OBJECTIVE: To define a quantitative stratification algorithm for the risk of early-onset sepsis (EOS) in newborns ≥ 34 weeks' gestation. METHODS: We conducted a retrospective nested case-control study that used split validation. Data collected on each infant included sepsis risk at birth based on objective maternal factors, demographics, specific clinical milestones, and vital signs during the first 24 hours after birth. Using a combination of recursive partitioning and logistic regression, we developed a risk classification scheme for EOS on the derivation dataset. This scheme was then applied to the validation dataset. RESULTS: Using a base population of 608,014 live births ≥ 34 weeks' gestation at 14 hospitals between 1993 and 2007, we identified all 350 EOS cases <72 hours of age and frequency matched them by hospital and year of birth to 1063 controls. Using maternal and neonatal data, we defined a risk stratification scheme that divided the neonatal population into 3 groups: treat empirically (4.1% of all live births, 60.8% of all EOS cases, sepsis incidence of 8.4/1000 live births), observe and evaluate (11.1% of births, 23.4% of cases, 1.2/1000), and continued observation (84.8% of births, 15.7% of cases, incidence 0.11/1000). CONCLUSIONS: It is possible to combine objective maternal data with evolving objective neonatal clinical findings to define more efficient strategies for the evaluation and treatment of EOS in term and late preterm infants. Judicious application of our scheme could result in decreased antibiotic treatment in 80,000 to 240,000 US newborns each year.

Prevalence and neonatal factors associated with autism spectrum disorders in preterm infants.

OBJECTIVES: To determine the prevalence of autism spectrum disorders (ASD) across gestational age, examine the risk of ASD by gestational age controlling for other risk factors, and identify potential risk factors in the neonatal intensive care unit. STUDY DESIGN: A retrospective cohort of infants born at ≥ 24 weeks between January 1, 2000, and December 31, 2007 at 11 Kaiser Permanente Northern California hospitals (n = 195,021). ASD cases were defined by a diagnosis made at a Kaiser Permanente ASD evaluation center, by a clinical specialist, or by a pediatrician. Cox proportional hazards regression models were used to evaluate the association between gestational age and ASD as well as potential risk factors in the neonatal intensive care unit and ASD. RESULTS: The prevalence of ASD in infants <37 weeks was 1.78% compared with 1.22% in infants born ≥ 37 weeks (P < .001). Compared with term infants, infants born at 24-26 weeks had an adjusted hazard ratio (HR) for a diagnosis of ASD of 2.7 (95% CI 1.5-5.0). Infants born at 27-33 weeks (adjusted HR 1.4, 95% CI 1.1-1.8) and 34-36 weeks (adjusted HR 1.3, 95% CI 1.1-1.4) were also at increased risk. High frequency ventilation and intracranial hemorrhage were associated with ASD in infants < 34 weeks. CONCLUSIONS: ASD was ~ 3 times more prevalent in infants <27 weeks compared with term infants. Each week of shorter gestation was associated with an increased risk of ASD. High frequency ventilation and intracranial hemorrhage were associated with ASD among infants <34 weeks.

Estimating the probability of neonatal early-onset infection on the basis of maternal risk factors.

OBJECTIVE: To develop a quantitative model to estimate the probability of neonatal early-onset bacterial infection on the basis of maternal intrapartum risk factors. METHODS: This was a nested case-control study of infants born at ≥34 weeks' gestation at 14 California and Massachusetts hospitals from 1993 to 2007. Case-subjects had culture-confirmed bacterial infection at <72 hours; controls were randomly selected, frequency-matched 3:1 according to year and birth hospital. We performed multivariate analyses and split validation to define a predictive model based only on information available in the immediate perinatal period. RESULTS: We identified 350 case-subjects from a cohort of 608,014 live births. Highest intrapartum maternal temperature revealed a linear relationship with risk of infection below 100.5°F, above which the risk rose rapidly. Duration of rupture of membranes revealed a steadily increasing relationship with infection risk. Increased risk was associated with both late-preterm and postterm delivery. Risk associated with maternal group B Streptococcus colonization is diminished in the era of group B Streptococcus prophylaxis. Any form of intrapartum antibiotic given >4 hours before delivery was associated with decreased risk. Our model showed good discrimination and calibration (c statistic = 0.800 and Hosmer-Lemeshow P = .142 in the entire data set). CONCLUSIONS: A predictive model based on information available in the immediate perinatal period performs better than algorithms based on risk-factor threshold values. This model establishes a prior probability for newborn sepsis, which could be combined with neonatal physical examination and laboratory values to establish a posterior probability to guide treatment decisions.

Interpreting complete blood counts soon after birth in newborns at risk for sepsis.

BACKGROUND: A complete blood count (CBC) with white blood cell differential is commonly ordered to evaluate newborns at risk for sepsis. OBJECTIVES: To quantify how well components of the CBC predict sepsis in the first 72 hours after birth. METHODS: For this retrospective cross-sectional study we identified 67 623 term and late-preterm (≥ 34 weeks gestation) newborns from 12 northern California Kaiser hospitals and 1 Boston, Massachusetts hospital who had a CBC and blood culture within 1 hour of each other at <72 hours of age. We compared CBC results among newborns whose blood cultures were and were not positive and quantified discrimination by using receiver operating characteristic curves and likelihood ratios. RESULTS: Blood cultures of 245 infants (3.6 of 1000 tested newborns) were positive. Mean white blood cell (WBC) counts and mean absolute neutrophil counts (ANCs) were lower, and mean proportions of immature neutrophils were higher in newborns with infection; platelet counts did not differ. Discrimination improved with age in the first few hours, especially for WBC counts and ANCs (eg, the area under the receiver operating characteristic curve for WBC counts was 0.52 at <1 hour and 0.87 at ≥ 4 hours). Both WBC counts and ANCs were most informative when very low (eg, the likelihood ratio for ANC < 1000 was 115 at ≥ 4 hours). No test was very sensitive; the lowest likelihood ratio (for WBC count ≥ 20 000 at ≥ 4 hours) was 0.16. CONCLUSION: Optimal interpretation of the CBC requires using interval likelihood ratios for the newborn's age in hours.

Numbers needed to treat with phototherapy according to American Academy of Pediatrics guidelines.

OBJECTIVES: Our aims were to estimate the efficacy of hospital phototherapy for neonatal jaundice and the number needed to treat to prevent one infant from reaching the exchange transfusion level. METHODS: From a cohort of 281 898 infants weighing > or =2000 g born at > or =35 weeks' gestation at 12 Northern California Kaiser hospitals from 1995 to 2004, we identified 22 547 who had a "qualifying total serum bilirubin level" within 3 mg/dL of the American Academy of Pediatrics 2004 guideline phototherapy threshold. We used multiple logistic regression to estimate the efficacy of hospital phototherapy within 8 hours at preventing the bilirubin level from exceeding the 2004 guideline's exchange transfusion threshold within 48 hours. We combined this efficacy estimate with other predictors of risk to estimate the numbers needed to treat at different values of covariates. RESULTS: Of the 22 547 eligible newborns, 5251 (23%) received hospital phototherapy within 8 hours of their qualifying bilirubin level. Only 354 (1.6%) ever exceeded the guideline exchange transfusion threshold; 187 (0.8%) did so within 48 hours. Among infants who did not have a positive direct antiglobulin test, hospital phototherapy within 8 hours was highly effective (adjusted odds ratio, 0.16; 95% confidence interval, 0.07-0.34). For infants with bilirubin levels 0-0.9 mg/dL above the phototherapy threshold, the estimated number needed to treat at mean values of covariates was 222 (95% CI: 107-502) for boys and 339 (95% CI: 154-729) for girls, ranging from 10 (95% CI: 6-19) for <24-hour-old, 36-week gestation boys to 3,041 (95% CI: 888-11 096) for > or =3-day-old 41-week girls. Hospital phototherapy was less effective for infants direct antiglobulin test-positive infants (adjusted odds ratio 0.55; 95% CI: 0.21-1.45; P = 0.01 for the direct antiglobulin test x phototherapy interaction). CONCLUSIONS: While hospital phototherapy is effective, the number needed to treat according to current guidelines varies considerably across different infant subgroups.

Factors influencing women's acceptance of prenatal screening tests.

OBJECTIVE: To determine the factors influencing a woman's acceptance of the expanded alpha-fetoprotein (AFP) test. METHODS: A population-based case-control study. All women (age < 35) who declined the expanded AFP test were identified as eligible cases. Controls were randomly selected from all women (age < 35) who accepted the test. RESULTS: We interviewed 199 cases and 229 controls before 30 weeks of gestation. While 47% of cases reported opposition to abortion as one of their reasons for declining the test (Group A), the remaining 53% of cases had a variety of other reasons for declining (Group B). After controlling for potential confounders, factors significantly associated with declining the test included: skepticism of the usefulness of the test results (odds ratio (OR) = 33.0), influence from family members (OR = 11.4), low educational level (OR = 7.1), willingness to keep a malformed fetus (OR = 6.2), failure to use providers as useful sources of information (OR = 5.0), and misunderstanding of the purpose of the test (OR = 2.0). Polytomous logistic regression revealed that Groups A and B had different determining factors as well as common factors. CONCLUSION: While many influential factors for participating in prenatal screening remain unmodifiable, some of them may be addressed to improve women's acceptance of prenatal screening tests.

Risk factors for severe hyperbilirubinemia among infants with borderline bilirubin levels: a nested case-control study.

OBJECTIVE: To estimate the effect of phototherapy and other predictors on the risk of total serum bilirubin (TSB) >or= 25 mg/dL in infants with a TSB of 17 to 22.9 mg/dL at age >or= 48 hours. STUDY DESIGN: From a cohort of 285295 infants >or= 34 weeks gestation and >or= 2000 g born between 1995 and 2004 in northern California Kaiser hospitals, we identified 17986 with a TSB of 17 to 22.9 mg/dL at age >or= 48 hours. All infants exhibiting a TSB >or= 25 mg/dL were selected as cases for the study. Four randomly selected controls were matched to each case based on the difference between their qualifying TSB and the American Academy of Pediatrics' phototherapy threshold. RESULTS: A total of 62 cases were identified (0.4%). Six of these (10%) received inpatient phototherapy within 8 hours, along with 101 controls (41%) (adjusted odds ratio [AOR] 0.15; 95% confidence interval [CI] 0.06 to 0.40). Cases more often had lower gestational age (AOR 3.24; 95% CI 1.24 to 8.47 for 38 to 39 weeks and AOR = 3.70; 95% CI 0.61 to 22.4 for 34 to 37 weeks compared with >or= 40-week infants), bruising, (AOR 2.52; 95% CI 1.16 to 5.50), exclusive breast-feeding (AOR 2.09; 95% CI 1.05 to 4.03), and TSB increase of >or= 6 mg/dL/day (AOR 2.39; 95% CI 1.18 to 4.85). CONCLUSIONS: Phototherapy was 85% effective in preventing TSB >or= 25 mg/dL. The strongest predictors of TSB >or= 25 mg/dL were gestational age, bruising, family history, and rapid rise in TSB.

A population-based prospective cohort study of personal exposure to magnetic fields during pregnancy and the risk of miscarriage.

To study the effect of magnetic fields on the risk of miscarriage, we conducted a population-based prospective cohort study among pregnant women within a large health maintenance organization. All women with a positive pregnancy test at less than 10 weeks of gestation and residing in the San Francisco area were contacted for participation in the study. We conducted in-person interviews to obtain information on risk factors for miscarriage and other potential confounders. All participants were also asked to wear a magnetic field-measuring meter for 24 hours and to keep a diary of their activities. Pregnancy outcomes were obtained for all participants by searching the health maintenance organization's databases, reviewing medical charts, and telephone follow-up. We used the Cox proportional hazard model for examining the magnetic field-miscarriage association. A total of 969 subjects were included in the final analyses. Although we did not observe an association between miscarriage risk and the average magnetic field level, miscarriage risk increased with an increasing level of maximum magnetic field exposure with a threshold around 16 milligauss (mG). The rate ratio (RR) associated with magnetic field exposure > or = 16 mG (vs <16 mG) was 1.8 [95% confidence interval (CI) = 1.2-2.7]. The risk remained elevated for levels (in tertiles) of maximum magnetic field exposure > or = 16 mG. The association was stronger for early miscarriages (<10 weeks of gestation) (RR = 2.2, 95% CI = 1.2-4.0) and among "susceptible" women with multiple prior fetal losses or subfertility (RR = 3.1, 95% CI = 1.3-7.7). After excluding women who indicated that their daily activity pattern during the measurements did not represent their typical daily activity during pregnancy, the association was strengthened; RR = 2.9 (95% CI = 1.6-5.3) for maximum magnetic field exposure > or = 16 mG, RR = 5.7 (95% CI = 2.1-15.7) for early miscarriage, and RR = 4.0 (95% CI = 1.4-11.5) among the susceptible women. Our findings provide strong prospective evidence that prenatal maximum magnetic field exposure above a certain level (possibly around 16 mG) may be associated with miscarriage risk. This observed association is unlikely to be due to uncontrolled biases or unmeasured confounders.