Linkage, evaluation and analysis of national electronic healthcare data: application to providing enhanced blood-stream infection surveillance in paediatric intensive care

Background

Linkage of risk-factor data for blood-stream infection (BSI) in paediatric intensive care (PICU) with bacteraemia surveillance data to monitor risk-adjusted infection rates in PICU is complicated by a lack of unique identifiers and under-ascertainment in the national surveillance system. We linked, evaluated and performed preliminary analyses on these data to provide a practical guide on the steps required to handle linkage of such complex data sources.

Methods

Data on PICU admissions in England and Wales for 2003-2010 were extracted from the Paediatric Intensive Care Audit Network. Records of all positive isolates from blood cultures taken for children <16 years and captured by the national voluntary laboratory surveillance system for 2003-2010 were extracted from the Public Health England database, LabBase2. “Gold-standard” datasets with unique identifiers were obtained directly from three laboratories, containing microbiology reports that were eligible for submission to LabBase2 (defined as “clinically significant” by laboratory microbiologists). Reports in the gold-standard datasets were compared to those in LabBase2 to estimate ascertainment in LabBase2. Linkage evaluated by comparing results from two classification methods (highest-weight classification of match weights and prior-informed imputation using match probabilities) with linked records in the gold-standard data. BSI rate was estimated as the proportion of admissions associated with at least one BSI.

Results

Reporting gaps were identified in 548/2596 lab-months of LabBase2. Ascertainment of clinically significant BSI in the remaining months was approximately 80-95%. Prior-informed imputation provided the least biased estimate of BSI rate (5.8% of admissions). Adjusting for ascertainment, the estimated BSI rate was 6.1-7.3%.

Conclusion

Linkage of PICU admission data with national BSI surveillance provides the opportunity for enhanced surveillance but analyses based on these data need to take account of biases due to ascertainment and linkage error. This study provides a generalisable guide for linkage, evaluation and analysis of complex electronic healthcare data.

Public health costs for tuberculosis suspects in Wake County, North Carolina, United States

SETTING

As the incidence of tuberculosis (TB) declines in high-income countries, resources to control TB are also declining. A portion of these resources are utilized for the evaluation and treatment of persons initially suspected of, but who do not actually have, TB (TB suspects).

OBJECTIVE

To describe the cost of TB suspects to public health departments, and determine whether part of this cost can be averted using improved diagnostic tools.

DESIGN

We evaluated resource utilization for all TB suspects as well as a random sample of TB cases managed at the Wake County public health clinic during 2008-2010. The proportion of total health department costs attributable to TB suspects was estimated. A sensitivity analysis assessed the potential impact of a rapid, accurate diagnostic test to avert suspect-associated costs.

RESULTS

Of 135 patients evaluated for TB, 36 (27%) were suspects, accounting for 14% (US5,885) of the total estimated costs for managing all patients. A perfect diagnostic test with a 3-day turnaround would have averted US27,975 (53%) of the costs attributable to suspects.

CONCLUSION

A substantial proportion of public health resources is utilized to manage persons whose final diagnosis is not TB. Efficient implementation of novel rapid tests could avert substantial public health costs.

Is chest CT useful in newborn screened infants with cystic fibrosis at 1 year of age?

Rationale

Sensitive outcome measures applicable in different centres to quantify and track early pulmonary abnormalities in infants with cystic fibrosis (CF) are needed both for clinical care and interventional trials. Chest CT has been advocated as such a measure yet there is no validated scoring system in infants.

Objectives

The objectives of this study were to standardise CT data collection across multiple sites; ascertain the incidence of bronchial dilatation and air trapping in newborn screened (NBS) infants with CF at 1 year; and assess the reproducibility of Brody-II, the most widely used scoring system in children with CF, during infancy.

Methods

A multicentre observational study of early pulmonary lung disease in NBS infants with CF at age 1 year using volume-controlled chest CT performed under general anaesthetic.

Main results

65 infants with NBS-diagnosed CF had chest CT in three centres. Small insignificant variations in lung recruitment manoeuvres but significant centre differences in radiation exposures were found. Despite experienced scorers and prior training, with the exception of air trapping, inter- and intraobserver agreement on Brody-II score was poor to fair (eg, interobserver total score mean (95% CI) κ coefficient: 0.34 (0.20 to 0.49)). Only 7 (11%) infants had a total CT score ≥12 (ie, ≥5% maximum possible) by either scorer.

Conclusions

In NBS infants with CF, CT changes were very mild at 1 year, and assessment of air trapping was the only reproducible outcome. CT is thus of questionable value in infants of this age, unless an improved scoring system for use in mild CF disease can be developed.

Evolution of lung function during the first year of life in newborn screened cystic fibrosis infants

Rationale

Newborn screening (NBS) for cystic fibrosis (CF) allows early intervention. Design of randomised controlled trials (RCT) is currently impeded by uncertainty regarding evolution of lung function, an important trial end point in such infants.

Objective

To assess changes in pulmonary function during the first year of life in CF NBS infants.

Methods

Observational longitudinal study. CF NBS infants and healthy controls were recruited between 2009 and 2011. Lung Clearance Index (LCI), plethysmographic lung volume (plethysmographic functional residual capacity (FRC_pleth)) and forced expired volume (FEV_0.5) were measured at 3 months and 1 year of age.

Main results

Paired measurements were obtained from 72 CF infants and 44 controls. At 3 months, CF infants had significantly worse lung function for all tests. FEV_0.5 improved significantly (0.59 (95% CI 0.18 to 0.99) z-scores; p<0.01) in CF infants between 3 months and 1 year, and by 1 year, FEV_0.5 was only 0.52 (0.89 to 0.15) z-scores less than in controls. LCI and FRC_pleth remained stable throughout the first year of life, being on average 0.8 z-scores higher in infants with CF. Pulmonary function at 1 year was predicted by that at 3 months. Among the 45 CF infants with entirely normal LCI and FEV_0.5 at 3 months, 80% remained so at 1 year, while 74% of those with early abnormalities remained abnormal at 1 year.

Conclusions

This is the first study reporting improvements in FEV_0.5 over time in stable NBS CF infants treated with standard therapy. Milder changes in lung function occurred by 1 year than previously reported. Lung function at 3 months predicts a high-risk group, who should be considered for intensification of treatment and enrolment into RCTs.

Community-acquired, healthcare-associated and hospital-acquired bloodstream infection definitions in children: a systematic review demonstrating inconsistent criteria

BACKGROUND

Historically, bacterial infections were categorized as either community-acquired (CA) or hospital-acquired (HA). However, the CA/HA dichotomy no longer adequately reflects patterns of emerging healthcare-associated (HCA) infections in complex patients managed between hospital and the community. Studies trying to define this evolving epidemiology often excluded children.

AIM

To identify what criteria have been used to distinguish between CA, HCA and HA bloodstream infections (BSIs) in children, and the proportional distribution of CA, HCA and HA among total BSIs and by organism.

METHODS

We systematically reviewed published literature from PubMed, UK Department of Health and US Centers for Disease Control and Prevention websites.

FINDINGS

Results from 23 studies and the websites highlighted the use of inconsistent criteria. There were 13 and 15 criteria variations for CA and HA BSI respectively, although a 48h cut-off for cultures sampled post admission was most commonly reported. Five studies used variable clinical criteria to define HCA. The mean proportion of paediatric CA BSI in nine studies was 50%. Only four BSI organisms from five studies were predominantly CA (Streptococcus pneumoniae, Salmonella spp.) or HA (coagulase-negative staphylococci, Enterococcus spp.), whereas Pseudomonas spp., Klebsiella spp. and Enterobacter spp. did not clearly fit into either category.

CONCLUSIONS

Our study reveals inconsistent use of criteria, and a lack of evidence upon which to base them, to distinguish between CA, HCA and HA BSI in children. Criteria for CA, HCA and HA BSI need to be developed using population-based studies that consider patients' clinical characteristics, recent healthcare exposure as well as isolated organism species.

Post-mortem MRI versus conventional autopsy in fetuses and children: a prospective validation study

BACKGROUND

Post-mortem MRI is a potential diagnostic alternative to conventional autopsy, but few large prospective studies have compared its accuracy with that of conventional autopsy. We assessed the accuracy of whole-body, post-mortem MRI for detection of major pathological lesions associated with death in a prospective cohort of fetuses and children.

METHODS

In this prospective validation study, we did pre-autopsy, post-mortem, whole-body MRI at 1·5 T in an unselected population of fetuses (≤24 weeks' or >24 weeks' gestation) and children (aged <16 years) at two UK centres in London between March 1, 2007 and Sept 30, 2011. With conventional autopsy as the diagnostic gold standard, we assessed MRI findings alone, or in conjunction with other minimally invasive post-mortem investigations (minimally invasive autopsy), for accuracy in detection of cause of death or major pathological abnormalities. A radiologist and pathologist who were masked to the autopsy findings indicated whether the minimally invasive autopsy would have been adequate. The primary outcome was concordance rate between minimally invasive and conventional autopsy.

FINDINGS

We analysed 400 cases, of which 277 (69%) were fetuses and 123 (31%) were children. Cause of death or major pathological lesion detected by minimally invasive autopsy was concordant with conventional autopsy in 357 (89·3%, 95% CI 85·8-91·9) cases: 175 (94·6%, 90·3-97·0) of 185 fetuses at 24 weeks' gestation or less, 88 (95·7%, 89·3-98·3) of 92 fetuses at more than 24 weeks' gestation, 34 (81·0%, 66·7-90·0) [corrected] of 42 newborns aged 1 month or younger, 45 (84·9%, 72·9-92·1) of 53 infants aged older than 1 month to 1 year or younger, and 15 (53·6%, 35·8-70·5) of 28 children aged older than 1 year to 16 years or younger. The dedicated radiologist or pathologist review of the minimally invasive autopsy showed that in 165 (41%) cases a full autopsy might not have been needed; in these cases, concordance between autopsy and minimally invasive autopsy was 99·4% (96·6-99·9).

INTERPRETATION

Minimally invasive autopsy has accuracy similar to that of conventional autopsy for detection of cause of death or major pathological abnormality after death in fetuses, newborns, and infants, but was less accurate in older children. If undertaken jointly by pathologists and radiologists, minimally invasive autopsy could be an acceptable alternative to conventional autopsy in selected cases.

FUNDING

Policy research Programme, Department of Health, UK.

New reference equations to improve interpretation of infant lung function

RATIONALE

With increasing use of infant pulmonary function tests (IPFTs) in both clinical and research studies, appropriate interpretation of results is essential.

OBJECTIVES

To investigate the potential bias associated with "normalising" IPF by expressing results as a ratio of body size and to develop reference ranges for tidal breathing parameters, passive respiratory mechanics (compliance [Crs] and resistance [Rrs]) and plethysmographic functional residual capacity (FRCp ) for white infants during the first 2 years of life.

METHODS

IPFTs were measured using the Jaeger BabyBody system and standardized protocols. Reference equations, adjusted for body size, age, and sex where appropriate, were created using multilevel modeling.

RESULTS

The ratio of lung function to body length changes markedly with growth, thereby precluding its use for any outcome. While the ratio of tidal volume and Crs to body weight remained relatively constant with growth, this was not the case for FRCp . Even in healthy infants, a strong inverse relationship was observed between lung function/body weight and weight z-score which could distort interpretation of results in growth-restricted infants with lung disease, such as cystic fibrosis. Reference equations were derived from 153 healthy white infants on 232 test occasions (median age 35.5 weeks [range: 2.6-104.7]). Crown-heel length was the strongest predictor of IPF.

CONCLUSIONS

When reporting IPF, use of size-corrected ratios should be discouraged, with interpretation instead based on appropriate reference equations. The current equations are applicable to white infants and young children up to 2 years of age, studied using the same commercially available equipment. The extent to which these equations are applicable to infants and young children of other ethnic backgrounds or who are tested with different equipment needs to be established.

Therapeutic hypothermia for neonatal encephalopathy in low- and middle-income countries: a systematic review and meta-analysis

Although selective or whole body cooling combined with optimal intensive care improves outcomes following neonatal encephalopathy in high-income countries, the safety and efficacy of cooling in low-and middle-income countries is not known.

Normal ascent of the conus medullaris: a post-mortem foetal MRI study

OBJECTIVES

The position of the conus medullaris is considered abnormal if it ends below lumbar vertebrae three (L3) at birth. We used magnetic resonance imaging (MRI) to measure the position of the conus in post-mortem foetuses, to investigate the timing of normal ascent.

METHODS

The position of the conus in 84 post-mortem foetuses (mean gestation 26.3 weeks; range 14-41 weeks) was identified using 3D MRI datasets. A numerical scale was used for vertebral levels, from 1 (S2) to 15 (T12).

RESULTS

There was significant ascent of the conus medullaris with increasing gestation. At 20 weeks gestation, an estimated 84.2% (95% confidence interval (CI): 72.9, 93.2%) of foetuses have a conus position of L4/5 or higher, but only 22.8% (95% CI 11.7, 34.9%) at L3 or higher. By 26 weeks, an estimated 50.7% (95% CI 34.1, 67.5) will have reached L3, and 94.8% (95% CI 87.0, 98.5%) reach L3 by 40 weeks.

CONCLUSION

There is regular ascent of the conus throughout foetal life. Although growth for each individual foetus may be non-linear, most foetuses have a conus level within the normal adult range by 33 weeks gestation.

Repeatability and bronchodilator reversibility of lung function in young children

Knowledge of short- and longer-term repeatability of lung function in health and disease is essential to determine bronchodilator reversibility thresholds and to recognise if changes in lung function represent disease progression, therapeutic intervention or normal variability. Multiple-breath washout indices (lung clearance index, conductive ventilation inhomogeneity (Scond)) and specific airway resistance (sRaw) were measured in healthy children and stable wheezers. Measurements were performed at baseline and after 20 min without intervention to assess repeatability and determine bronchodilator reversibility thresholds. Bronchodilator reversibility was assessed by repeating baseline measurements 20 min after inhaled salbutamol. 28 healthy controls, mean±sd age 6.1±0.7 years and 62 wheezers 5.4±0.6 years were tested. Baseline variability in multiple-breath washout indices and sRaw was not significantly different between wheezers and healthy controls. Significant bronchodilator reversibility was only observed in wheezers for Scond (16%), but in both wheezers (37%) and healthy controls (20%) for sRaw. Some wheezers and healthy controls demonstrated increases in multiple-breath washout indices post-bronchodilator. Lung clearance index and sRaw demonstrate low baseline variability in healthy and diseased subjects. Neither multiple-breath washout indices nor sRaw are ideal for assessing bronchodilator reversibility in young children with stable wheeze. These findings will help to interpret the effect of therapeutic interventions in children with respiratory diseases.

Age and height dependence of lung clearance index and functional residual capacity

The lung clearance index (LCI) is more sensitive than spirometry in detecting abnormal lung function in children with cystic fibrosis. LCI is thought to be independent of age, but recent evidence suggests that the upper limit of normal is higher in infants and preschool children than in older subjects. This study examines whether LCI remains independent of body size throughout childhood. Multiple-breath washout data from healthy children and adolescents were collated from three centres using the mass spectrometer system and the inert gas sulfur hexafluoride. Reference equations for LCI and functional residual capacity (FRC) were constructed using the LMS (lambda-mu-sigma) method. Data were available from 497 subjects (2 weeks to 19 years of age) tested on 659 occasions. LCI was dependent on body size, decreasing in a nonlinear pattern as height increased. Changes were particularly marked in the first 5 years of life. Height, age and sex were all independent predictors of FRC. Minimal between-centre differences allowed unified reference equations to be developed. LCI is not independent of body size. Although a constant upper normal limit would suffice for cross-sectional clinical assessments from 6 years of age, appropriate reference equations are essential for accurate interpretation of results during early childhood.

The analysis of record-linked data using multiple imputation with data value priors

Probabilistic record linkage techniques assign match weights to one or more potential matches for those individual records that cannot be assigned 'unequivocal matches' across data files. Existing methods select the single record having the maximum weight provided that this weight is higher than an assigned threshold. We argue that this procedure, which ignores all information from matches with lower weights and for some individuals assigns no match, is inefficient and may also lead to biases in subsequent analysis of the linked data. We propose that a multiple imputation framework be utilised for data that belong to records that cannot be matched unequivocally. In this way, the information from all potential matches is transferred through to the analysis stage. This procedure allows for the propagation of matching uncertainty through a full modelling process that preserves the data structure. For purposes of statistical modelling, results from a simulation example suggest that a full probabilistic record linkage is unnecessary and that standard multiple imputation will provide unbiased and efficient parameter estimates.