On repeated diagnostic testing in screening for a medical condition: How often should the diagnostic test be repeated?

In screening large populations a diagnostic test is frequently used repeatedly. An example is screening for bowel cancer using the fecal occult blood test (FOBT) on several occasions such as at 3 or 6 days. The question that is addressed here is how often should we repeat a diagnostic test when screening for a specific medical condition. Sensitivity is often used as a performance measure of a diagnostic test and is considered here for the individual application of the diagnostic test as well as for the overall screening procedure. The latter can involve an increasingly large number of repeated applications, but how many are sufficient? We demonstrate the issues involved in answering this question using real data on bowel cancer at St Vincents Hospital in Sydney. As data are only available for those testing positive at least once, an appropriate modeling technique is developed on the basis of the zero-truncated binomial distribution which allows for population heterogeneity. The latter is modeled using discrete nonparametric maximum likelihood. If we wish to achieve an overall sensitivity of 90%, the FOBT should be repeated for 2 weeks instead of the 1 week that was used at the time of the survey. A simulation study also shows consistency in the sense that bias and standard deviation for the estimated sensitivity decrease with an increasing number of repeated occasions as well as with increasing sample size.

Nonparametric estimation of the random effects distribution for the risk or rate ratio in rare events meta-analysis with the arm-based and contrast-based approaches

Rare events are events which occur with low frequencies. These often arise in clinical trials or cohort studies where the data are arranged in binary contingency tables. In this article, we investigate the estimation of effect heterogeneity for the risk-ratio parameter in meta-analysis of rare events studies through two likelihood-based nonparametric mixture approaches: an arm-based and a contrast-based model. Maximum likelihood estimation is achieved using the EM algorithm. Special attention is given to the choice of initial values. Inspired by the classification likelihood, a strategy is implemented which repeatably uses random allocation of the studies to the mixture components as choice of initial values. The likelihoods under the contrast-based and arm-based approaches are compared and differences are highlighted. We use simulations to assess the performance of these two methods. Under the design of sampling studies with nested treatment groups, the results show that the nonparametric mixture model based on the contrast-based approach is more appropriate in terms of model selection criteria such as AIC and BIC. Under the arm-based design the results from the arm-based model performs well although in some cases it is also outperformed by the contrast-based model. Comparisons of the estimators are provided in terms of bias and mean squared error. Also included in the comparison is the mixed Poisson regression model as well as the classical DerSimonian-Laird model (using the Mantel-Haenszel estimator for the common effect). Using simulation, estimating effect heterogeneity in the case of the contrast-based method appears to behave better than the compared methods although differences become negligible for large within-study sample sizes. We illustrate the methodologies using several meta-analytic data sets in medicine.

Modeling COVID-19 contact-tracing using the ratio regression capture-recapture approach

Contact-tracing is one of the most effective tools in infectious disease outbreak control. A capture-recapture approach based upon ratio regression is suggested to estimate the completeness of case detection. Ratio regression has been recently developed as flexible tool for count data modeling and has proved to be successful in the capture-recapture setting. The methodology is applied here to Covid-19 contact tracing data from Thailand. A simple weighted straight line approach is used which includes the Poisson and geometric distribution as special cases. For the case study data of contact tracing for Thailand, a completeness of 83% could be found with a 95% confidence interval of 74%-93%.

Assessing injury risk in male and female Royal Navy recruits: does the Functional Movement Screen provide understanding to inform effective injury mitigation?

INTRODUCTION

Musculoskeletal injuries (MSKIs) are common during military and other occupational physical training programmes, and employers have a duty of care to mitigate this injury risk. MSKIs account for a high number of working days lost during initial military training, contribute to training attrition and impact training costs. Poorer movement quality may be associated with increased MSKI risk.

METHODS

The present study evaluated the relationship between the Functional Movement Screen (FMS) Score, as a measure of movement quality, and injury risk in Royal Navy (RN) recruits. A cohort of 957 recruits was assessed using the FMS prior to the 10-week phase I training programme. Injury occurrence, time, type and severity were recorded prospectively during the training period.

RESULTS

Total FMS Score was associated with injury risk (p≤0.001), where recruits scoring ≥13 were 2.6 times more likely to sustain an injury during training. However, FMS Score accounted for only 10% of the variance in injury risk (R2=0.1). Sex was the only additional variable to significantly affect the regression model. Mean FMS Scores for men (14.6±2.3) and women (14.4±2.4) were similar, but injury occurrence in women was 1.7 times greater than in men. Examining the influence of individual FMS movement tests on injury prediction did not improve the model, where those movements that significantly contributed to injury prediction only accounted for a small amount of the variance (R2=0.01).

CONCLUSION

There was a weak relationship between FMS and injury risk in RN recruits. Evidence is provided that FMS score alone would not be appropriate to use as an injury prediction tool in military recruits.

Retrospective evaluation of factors affecting successful fit testing of respiratory protective equipment during the early phase of COVID-19

OBJECTIVES

Respiratory protective equipment is critical to protect healthcare workers from COVID-19 infection, which includes filtering facepiece respirators (FFP3). There are reports of fitting issues within healthcare workers, although the factors affecting fitting outcomes are largely unknown. This study aimed to evaluate factors affecting respirator fitting outcomes.

DESIGN

This is a retrospective evaluation study. We conducted a secondary analysis of a national database of fit testing outcomes in England between July and August 2020.

SETTINGS

The study involves National Health Service (NHS) hospitals in England.

PARTICIPANTS

A total of 9592 observations regarding fit test outcomes from 5604 healthcare workers were included in the analysis.

INTERVENTION

Fit testing of FFP3 on a cohort of healthcare workers in England, working in the NHS.

PRIMARY AND SECONDARY OUTCOME MEASURES

Primary outcome measure was the fit testing result, that is, pass or fail with a specific respirator. Key demographics, including age, gender, ethnicity and face measurements of 5604 healthcare workers, were used to compare fitting outcomes.

RESULTS

A total of 9592 observations from 5604 healthcare workers were included in the analysis. A mixed-effects logistic regression model was used to determine the factors which affected fit testing outcome. Results showed that males experienced a significantly (p<0.05) higher fit test success than females (OR 1.51; 95% CI 1.27 to 1.81). Those with non-white ethnicities demonstrated significantly lower odds of successful respirator fitting; black (OR 0.65; 95% CI 0.51 to 0.83), Asian (OR 0.62; 95% CI 0.52 to 0.74) and mixed (OR 0.60; 95% CI 0.45 to 0.79.

CONCLUSION

During the early phase of COVID-19, females and non-white ethnicities were less likely to have a successful respirator fitting. Further research is needed to design new respirators which provide equal opportunity for comfortable, effective fitting of these devices.

On the equivalence of one-inflated zero-truncated and zero-truncated one-inflated count data likelihoods

One-inflation in zero-truncated count data has recently found considerable attention. There are currently two views in the literature. In the first approach, the untruncated model is considered as one-inflated whereas in the second approach the truncated model is viewed as one-inflated. Here, we show that both models have identical model spaces as well as identical maximum likelihoods. Consequences of population size estimation are illuminated, and the findings are illustrated at hand of two case studies.

Estimating the undetected infections in the Monkeypox outbreak

While the number of detected Monkeypox infections are widely available, an understanding of the extent of undetected cases is urgently needed for an effective tackling of its spread. The aim of this study is to estimate the true number of Monkeypox (detected and undetected) infections in most affected countries. The question being asked is: How many cases have actually occurred? We propose a lower bound estimator for the true number of Monkeypox cases. The estimator is data-driven and can be easily computed from the cumulative distributions of weekly cases. We focused on the ratio of the total estimated cases to the observed cases on July 31, 2022: The proportion of undetected cases was relevant in all countries, with countries whose estimated true number of infections could be more than three times the observed one. We provided a practical contribution to the understanding of the current Monkeypox wave and reliable estimates on how many undetected cases are going around in several countries, where the epidemic spreads differently.

Fewer COVID-19 Neurological Complications with Dexamethasone and Remdesivir

OBJECTIVE

The objective of this study was to assess the impact of treatment with dexamethasone, remdesivir or both on neurological complications in acute coronavirus diease 2019 (COVID-19).

METHODS

We used observational data from the International Severe Acute and emerging Respiratory Infection Consortium World Health Organization (WHO) Clinical Characterization Protocol, United Kingdom. Hospital inpatients aged ≥18 years with laboratory-confirmed severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) infection admitted between January 31, 2020, and June 29, 2021, were included. Treatment allocation was non-blinded and performed by reporting clinicians. A propensity scoring methodology was used to minimize confounding. Treatment with remdesivir, dexamethasone, or both was assessed against the standard of care. The primary outcome was a neurological complication occurring at the point of death, discharge, or resolution of the COVID-19 clinical episode.

RESULTS

Out of 89,297 hospital inpatients, 64,088 had severe COVID-19 and 25,209 had non-hypoxic COVID-19. Neurological complications developed in 4.8% and 4.5%, respectively. In both groups, neurological complications were associated with increased mortality, intensive care unit (ICU) admission, worse self-care on discharge, and time to recovery. In patients with severe COVID-19, treatment with dexamethasone (n = 21,129), remdesivir (n = 1,428), and both combined (n = 10,846) were associated with a lower frequency of neurological complications: OR = 0.76 (95% confidence interval [CI] = 0.69-0.83), OR = 0.69 (95% CI = 0.51-0.90), and OR = 0.54 (95% CI = 0.47-0.61), respectively. In patients with non-hypoxic COVID-19, dexamethasone (n = 2,580) was associated with less neurological complications (OR = 0.78, 95% CI = 0.62-0.97), whereas the dexamethasone/remdesivir combination (n = 460) showed a similar trend (OR = 0.63, 95% CI = 0.31-1.15).

INTERPRETATION

Treatment with dexamethasone, remdesivir, or both in patients hospitalized with COVID-19 was associated with a lower frequency of neurological complications in an additive manner, such that the greatest benefit was observed in patients who received both drugs together. ANN NEUROL 2023;93:88-102.

Estimating risk and rate ratio in rare events meta-analysis with the Mantel–Haenszel estimator and assessing heterogeneity

Meta-analysis of binary outcome data faces often a situation where studies with a rare event are part of the set of studies to be considered. These studies have low occurrence of event counts to the extreme that no events occur in one or both groups to be compared. This raises issues how to estimate validly the summary risk or rate ratio across studies. A preferred choice is the Mantel–Haenszel estimator, which is still defined in the situation of zero studies unless all studies have zeros in one of the groups to be compared. For this situation, a modified Mantel–Haenszel estimator is suggested and shown to perform well by means of simulation work. Also, confidence interval estimation is discussed and evaluated in a simulation study. In a second part, heterogeneity of relative risk across studies is investigated with a new chi-square type statistic which is based on a conditional binomial distribution where the conditioning is on the event margin for each study. This is necessary as the conventional Q-statistic is undefined in the occurrence of zero studies. The null-distribution of the proposed Q-statistic is obtained by means of a parametric bootstrap as a chi-square approximation is not valid for rare events meta-analysis, as bootstrapping of the null-distribution shows. In addition, for the effect heterogeneity situation, confidence interval estimation is considered using a nonparametric bootstrap procedure. The proposed techniques are illustrated at hand of three meta-analytic data sets.

Estimation of Effect Heterogeneity in Rare Events Meta-Analysis

The paper outlines several approaches for dealing with meta-analyses of count outcome data. These counts are the accumulation of occurred events, and these events might be rare, so a special feature of the meta-analysis is dealing with low counts including zero-count studies. Emphasis is put on approaches which are state of the art for count data modelling including mixed log-linear (Poisson) and mixed logistic (binomial) regression as well as nonparametric mixture models for count data of Poisson and binomial type. A simulation study investigates the performance and capability of discrete mixture models in estimating effect heterogeneity. The approaches are exemplified on a meta-analytic case study investigating the acceptance of bibliotherapy.

Bayesian spatio-temporal joint disease mapping of Covid-19 cases and deaths in local authorities of England

The overwhelming spatio-temporal nature of the spread of the ongoing Covid-19 pandemic demands urgent attention of data analysts and model developers. Modelling results obtained from analytical tool development are essential to understand the ongoing pandemic dynamics with a view to helping the public and policy makers. The pandemic has generated data on a huge number of interesting statistics such as the number of new cases, hospitalisations and deaths in many spatio-temporal resolutions for the analysts to investigate. The multivariate nature of these data sets, along with the inherent spatio-temporal dependencies, poses new challenges for modellers. This article proposes a two-stage hierarchical Bayesian model as a joint bivariate model for the number of cases and deaths observed weekly for the different local authority administrative regions in England. An adaptive model is proposed for the weekly Covid-19 death rates as part of the joint bivariate model. The adaptive model is able to detect possible step changes in death rates in neighbouring areas. The joint model is also used to evaluate the effects of several socio-economic and environmental covariates on the rates of cases and deaths. Inclusion of these covariates points to the presence of a north-south divide in both the case and death rates. Nitrogen dioxide, the only air pollution measure used in the model, is seen to be significantly positively associated with the number cases, even in the presence of the spatio-temporal random effects taking care of spatio-temporal dependencies present in the data. The proposed models provide excellent fits to the observed data and are seen to perform well for predicting the location specific number of deaths a week in advance. The structure of the models is very general and the same framework can be used for modelling other areally aggregated temporal statistics of the pandemics, e.g. the rate of hospitalisation.

Combined statistical decision limits based on two GH-2000 scores for the detection of growth hormone misuse

The growth hormone-2000 biomarker method, based on the measurements of insulin-like growth factor-I and the amino-terminal pro-peptide of type III collagen, has been developed as a powerful technique for the detection of growth hormone misuse by athletes. Insulin-like growth factor-I and amino-terminal pro-peptide of type III collagen are combined in gender-specific formulas to create the growth hormone-2000 score, which is used to determine whether growth hormone has been administered. To comply with World Anti-Doping Agency regulations, each analyte must be measured by two methods. Insulin-like growth factor-I and amino-terminal pro-peptide of type III collagen can be measured by a number of approved methods, each leading to its own growth hormone-2000 score. Single decision limits for each growth hormone-2000 score have been introduced and developed by Bassett, Erotokritou-Mulligan, Holt, Böhning and their co-authors in a series of papers. These have been incorporated into the guidelines of the World Anti-Doping Agency. A joint decision limit was constructed based on the sample correlation between the two growth hormone-2000 scores generated from an available sample to increase the sensitivity of the biomarker method. This paper takes this idea further into a fully developed statistical approach. It constructs combined decision limits when two growth hormone-2000 scores from different assay combinations are used to decide whether an athlete has been misusing growth hormone. The combined decision limits are directly related to tolerance regions and constructed using a Bayesian approach. It is also shown to have highly satisfactory frequentist properties. The new approach meets the required false-positive rate with a pre-specified level of certainty.

Can digital rectal examination be used to detect cauda equina compression in people presenting with acute cauda equina syndrome? A systematic review and meta-analysis of diagnostic test accuracy studies

BACKGROUND

Digital rectal examination (DRE) is a commonly used test to help identify people with cauda equina compression (CEC).

OBJECTIVE

To determine the diagnostic accuracy of DRE in assessment of anal tone, squeeze, sensation and reflexes, as predictors of CEC.

DESIGN

A systematic review to investigate the diagnostic accuracy of DRE to detect CEC compared with lumbar Magnetic Resonance Imaging (MRI).

METHOD

Six electronic databases were searched from inception to 6 July 2020 for studies published in English. Two assessors independently performed screening, data extraction and risk of bias assessment (QUADAS-2). Meta-analysis was performed using STATA-16.

RESULTS

Six studies were included (n = 741). The sensitivity of anal tone was low across all studies (range: 0.23 to 0.53) with moderate quality evidence against the use of DRE of anal tone. One study on anal sensation found no correlation with CEC using Kendall's tau test: p = 0.102 and another found sensation had low test accuracy. One study identified sensitivity: 0.29 and specificity: 0.96 for anal squeeze, while another identified sensitivity: 0.38 and specificity: 0.6 for anal reflexes.

CONCLUSION

The diagnostic accuracy of DRE of anal tone to detect CEC is low and carries a high risk of false reassurance. It is therefore not recommended in any clinical setting. More research is needed to determine the diagnostic accuracy of DRE of anal squeeze, sensation and reflexes and if done the results should be interpreted with caution.

Sensitivity of contact-tracing for COVID-19 in Thailand: a capture-recapture application

BACKGROUND

We investigate the completeness of contact tracing for COVID-19 during the first wave of the COVID-19 pandemic in Thailand, from early January 2020 to 30 June 2020.

METHODS

Uni-list capture-recapture models were applied to the frequency distributions of index cases to inform two questions: (1) the unobserved number of index cases with contacts, and (2) the unobserved number of index cases with secondary cases among their contacts.

RESULTS

Generalized linear models (using Poisson and logistic families) did not return any significant predictor (age, sex, nationality, number of contacts per case) on the risk of transmission and hence capture-recapture models did not adjust for observed heterogeneity. Best fitting models, a zero truncated negative binomial for question 1 and zero-truncated Poisson for question 2, returned sensitivity estimates for contact tracing performance of 77.6% (95% CI = 73.75-81.54%) and 67.6% (95% CI = 53.84-81.38%), respectively. A zero-inflated negative binomial model on the distribution of index cases with secondary cases allowed the estimation of the effective reproduction number at 0.14 (95% CI = 0.09-0.22), and the overdispersion parameter at 0.1.

CONCLUSION

Completeness of COVID-19 contact tracing in Thailand during the first wave appeared moderate, with around 67% of infectious transmission chains detected. Overdispersion was present suggesting that most of the index cases did not result in infectious transmission chains and the majority of transmission events stemmed from a small proportion of index cases.

Novel Use of Capture-Recapture Methods to Estimate Completeness of Contact Tracing during an Ebola Outbreak, Democratic Republic of the Congo, 2018-2020

Despite its critical role in containing outbreaks, the efficacy of contact tracing, measured as the sensitivity of case detection, remains an elusive metric. We estimated the sensitivity of contact tracing by applying unilist capture-recapture methods on data from the 2018–2020 outbreak of Ebola virus disease in the Democratic Republic of the Congo. To compute sensitivity, we applied different distributional assumptions to the zero-truncated count data to estimate the number of unobserved case-patients with any contacts and infected contacts. Geometric distributions were the best-fitting models. Our results indicate that contact tracing efforts identified almost all (n = 792, 99%) of case-patients with any contacts but only half (n = 207, 48%) of case-patients with infected contacts, suggesting that contact tracing efforts performed well at identifying contacts during the listing stage but performed poorly during the contact follow-up stage. We discuss extensions to our work and potential applications for the ongoing coronavirus pandemic.

The combined effect of alcohol and body mass index on risk of chronic liver disease: A systematic review and meta-analysis of cohort studies

BACKGROUND & AIMS

Increasingly populations are both overweight/obese and consume alcohol. The risk of liver disease from the combination of these factors is unclear. We performed a systematic review and meta-analysis to address this important gap in evidence. Protocol registered with PROSPERO(CRD42016046508).

METHODS

We performed electronic searches of Ovid Medline, Embase Classic + Embase, until 17th June 2020 for cohort studies of adults without pre-existing liver disease. Primary outcome was morbidity/mortality from chronic liver disease. Exposures were alcohol consumption categorised as within or above UK recommended limits (14 units/112 g per week) and BMI categorised as normal, overweight or obese. Non-drinkers were excluded. A Poisson regression log-linear model was used to test for statistical interaction between alcohol and BMI and to conduct a one-stage meta-analysis.

RESULTS

Searches identified 3129 studies-16 were eligible. Of these, nine cohorts (1,121,514 participants) had data available and were included in the analysis. The Poisson model showed no significant statistical interaction between alcohol consumption and BMI on the risk of chronic liver disease. Compared to normal weight participants drinking alcohol within UK recommended limits, relative risk of chronic liver disease in overweight participants drinking above limits was 3.32 (95% CI 2.88 to 3.83) and relative risk in obese participants drinking above limits was 5.39 (95% CI 4.62 to 6.29).

CONCLUSIONS

This meta-analysis demonstrated a significantly increased risk of chronic liver disease in participants who were both overweight/obese and consumed alcohol above UK recommended limits. This evidence should inform advice given to patients and risk stratification by healthcare professionals.

Delayed antibiotic prescribing for respiratory tract infections: individual patient data meta-analysis

OBJECTIVE

To assess the overall effect of delayed antibiotic prescribing on average symptom severity for patients with respiratory tract infections in the community, and to identify any factors modifying this effect.

DESIGN

Systematic review and individual patient data meta-analysis.

DATA SOURCES

Cochrane Central Register of Controlled Trials, Ovid Medline, Ovid Embase, EBSCO CINAHL Plus, and Web of Science.

ELIGIBILITY CRITERIA FOR STUDY SELECTION

Randomised controlled trials and observational cohort studies in a community setting that allowed comparison between delayed versus no antibiotic prescribing, and delayed versus immediate antibiotic prescribing.

MAIN OUTCOME MEASURES

The primary outcome was the average symptom severity two to four days after the initial consultation measured on a seven item scale (ranging from normal to as bad as could be). Secondary outcomes were duration of illness after the initial consultation, complications resulting in admission to hospital or death, reconsultation with the same or worsening illness, and patient satisfaction rated on a Likert scale.

RESULTS

Data were obtained from nine randomised controlled trials and four observational studies, totalling 55 682 patients. No difference was found in follow-up symptom severity (seven point scale) for delayed versus immediate antibiotics (adjusted mean difference -0.003, 95% confidence interval -0.12 to 0.11) or delayed versus no antibiotics (0.02, -0.11 to 0.15). Symptom duration was slightly longer in those given delayed versus immediate antibiotics (11.4 v 10.9 days), but was similar for delayed versus no antibiotics. Complications resulting in hospital admission or death were lower with delayed versus no antibiotics (odds ratio 0.62, 95% confidence interval 0.30 to 1.27) and delayed versus immediate antibiotics (0.78, 0.53 to 1.13). A significant reduction in reconsultation rates (odds ratio 0.72, 95% confidence interval 0.60 to 0.87) and an increase in patient satisfaction (adjusted mean difference 0.09, 0.06 to 0.11) were observed in delayed versus no antibiotics. The effect of delayed versus immediate antibiotics and delayed versus no antibiotics was not modified by previous duration of illness, fever, comorbidity, or severity of symptoms. Children younger than 5 years had a slightly higher follow-up symptom severity with delayed antibiotics than with immediate antibiotics (adjusted mean difference 0.10, 95% confidence interval 0.03 to 0.18), but no increased severity was found in the older age group.

CONCLUSIONS

Delayed antibiotic prescribing is a safe and effective strategy for most patients, including those in higher risk subgroups. Delayed prescribing was associated with similar symptom duration as no antibiotic prescribing and is unlikely to lead to poorer symptom control than immediate antibiotic prescribing. Delayed prescribing could reduce reconsultation rates and is unlikely to be associated with an increase in symptoms or illness duration, except in young children.

STUDY REGISTRATION

PROSPERO CRD42018079400.

The effects of recombinant human insulin-like growth factor-1/insulin-like growth factor binding protein-3 administration on lipid and carbohydrate metabolism in recreational athletes

OBJECTIVE

Previous studies suggested that recombinant human IGF-1 (rhIGF-1) administration affects carbohydrate and lipid metabolism in healthy people and in people with diabetes. This study aimed to determine the effects of rhIGF-1/rhIGF binding protein-3 (rhIGFBP-3) administration on glucose homeostasis and lipid metabolism in healthy recreational athletes.

DESIGN AND SETTING

Randomized, double-blind, placebo-controlled rhIGF-1/rhIGFBP-3 administration study at Southampton General Hospital, UK.

PARTICIPANTS

56 recreational athletes (30 men, 26 women).

METHODS

Participants were randomly assigned to receive placebo, low-dose rhIGF-1/rhIGFBP-3 (30 mg/day) or high-dose rhIGF-1/rhIGFBP-3 (60 mg/day) for 28 days. The following variables were measured before and immediately after the treatment period: fasting lipids, glucose, insulin, C-peptide and glycated haemoglobin. The homeostatic model assessment (HOMA-IR) was used to estimate insulin sensitivity and indirect calorimetry to assess substrate oxidation rates. The general linear model approach was used to compare treatment group changes with the placebo group.

RESULTS

Compared with the placebo group, there was a significant reduction in fasting triglycerides in participants treated with high-dose rhIGF-1/rhIGFBP-3 (p = .030), but not in the low-dose group (p = .390). In women, but not in men, there were significant increases in total cholesterol (p = .003), HDL cholesterol (p = .001) and LDL cholesterol (p = .008). These lipid changes were associated with reduced fasting insulin (p = .010), C-peptide (p = .001) and HOMA-IR (p = .018) in women and reduced C-peptide (p = .046) in men.

CONCLUSIONS

rhIGF-1/rhIGFBP-3 administration for 28 days reduced insulin concentration, improved insulin sensitivity and had significant effects on lipid profile including decreased fasting triglycerides.

Population size estimation based upon zero-truncated, one-inflated and sparse count data

Estimating the size of a hard-to-count population is a challenging matter. In particular, when only few observations of the population to be estimated are available. The matter gets even more complex when one-inflation occurs. This situation is illustrated with the help of two examples: the size of a dice snake population in Graz (Austria) and the number of flare stars in the Pleiades. The paper discusses how one-inflation can be easily handled in likelihood approaches and also discusses how variances and confidence intervals can be obtained by means of a semi-parametric bootstrap. A Bayesian approach is mentioned as well and all approaches result in similar estimates of the hidden size of the population. Finally, a simulation study is provided which shows that the unconditional likelihood approach as well as the Bayesian approach using Jeffreys’ prior perform favorable.

Comparison of normal distribution-based and nonparametric decision limits on the GH-2000 score for detecting growth hormone misuse (doping) in sport

This paper is motivated by the GH-2000 biomarker test, though the discussion is applicable to other diagnostic tests. The GH-2000 biomarker test has been developed as a powerful technique to detect growth hormone misuse by athletes, based on the GH-2000 score. Decision limits on the GH-2000 score have been developed and incorporated into the guidelines of the World Anti-Doping Agency (WADA). These decision limits are constructed, however, under the assumption that the GH-2000 score follows a normal distribution. As it is difficult to affirm the normality of a distribution based on a finite sample, nonparametric decision limits, readily available in the statistical literature, are viable alternatives. In this paper, we compare the normal distribution-based and nonparametric decision limits. We show that the decision limit based on the normal distribution may deviate significantly from the nominal confidence level 1 - α or nominal FPR γ when the distribution of the GH-2000 score departs only slightly from the normal distribution. While a nonparametric decision limit does not assume any specific distribution of the GH-2000 score and always guarantees the nominal confidence level and FPR, it requires a much larger sample size than the normal distribution-based decision limit. Due to the stringent FPR of the GH-2000 biomarker test used by WADA, the sample sizes currently available are much too small, and it will take many years of testing to have the minimum sample size required, in order to use the nonparametric decision limits. Large sample theory about the normal distribution-based and nonparametric decision limits is also developed in this paper to help understanding their behaviours when the sample size is large.

General flation models for count data

The paper discusses very general extensions to existing inflation models for discrete random variables, allowing an arbitrary set of points in the sample space to be either inflated or deflated relative to a baseline distribution. The term flation is introduced to cover either inflation or deflation of counts. Examples include one-inflated count models where the baseline distribution is zero-truncated and count models for data with a few unusual large values. The main result is that inference about the baseline distribution can be based solely on the truncated distribution which arises when the entire set of flation points is truncated. A major application of this result relates to estimating the size of a hidden target population, and examples are provided to illustrate our findings.

Estimating the undetected infections in the Covid-19 outbreak by harnessing capture-recapture methods

OBJECTIVES

A major open question, affecting the decisions of policy makers, is the estimation of the true number of Covid-19 infections. Most of them are undetected, because of a large number of asymptomatic cases. We provide an efficient, easy to compute and robust lower bound estimator for the number of undetected cases.

METHODS

A modified version of the Chao estimator is proposed, based on the cumulative time-series distributions of cases and deaths. Heterogeneity has been addressed by assuming a geometrical distribution underlying the data generation process. An (approximated) analytical variance of the estimator has been derived to compute reliable confidence intervals at 95% level.

RESULTS

A motivating application to the Austrian situation is provided and compared with an independent and representative study on prevalence of Covid-19 infection. Our estimates match well with the results from the independent prevalence study, but the capture-recapture estimate has less uncertainty involved as it is based on a larger sample size. Results from other European countries are mentioned in the discussion. The estimated ratio of the total estimated cases to the observed cases is around the value of 2.3 for all the analyzed countries.

CONCLUSIONS

The proposed method answers to a fundamental open question: "How many undetected cases are going around?". CR methods provide a straightforward solution to shed light on undetected cases, incorporating heterogeneity that may arise in the probability of being detected.

Upper bound estimators of the population size based on ordinal models for capture-recapture experiments

Capture-recapture studies have attracted a lot of attention over the past few decades, especially in applied disciplines where a direct estimate for the size of a population of interest is not available. Epidemiology, ecology, public health, and biodiversity are just a few examples. The estimation of the number of unseen units has been a challenge for theoretical statisticians, and considerable progress has been made in providing lower bound estimators for the population size. In fact, it is well known that consistent estimators for this cannot be provided in the very general case. Considering a case where capture-recapture studies are summarized by a frequency of frequencies distribution, we derive a simple upper bound of the population size based on the cumulative distribution function. We introduce two estimators of this bound, without any specific parametric assumption on the distribution of the observed frequency counts. The behavior of the proposed estimators is investigated using several benchmark datasets and a large-scale simulation experiment based on the scheme discussed by Pledger.

Estimation of the adolescent pregnancy rate in Thailand 2008-2013: an application of capture-recapture method

BACKGROUND

Adolescent pregnancy is an important health and social issue that affects both individual and social well-being. However, deriving a national estimate is challenging in a country with multiple incomplete national databases especially the abortion statistics. The objective of this study was to estimate the adolescent pregnancy rates in Thailand using capture-recapture method.

METHODS

An application of capture-recapture method was conducted using two cross-sectional databases (i.e., the national birth registration and the Ministry of Public Health standard health databases) and one hospital-based data source from medical record reviews. A 3-sources capture-recapture with log-linear model was applied to estimate adolescent pregnancy rates.

RESULTS

A total number of 741,084, 290,922 and 25,478 records were respectively identified from the birth registrations, standard health databases and hospital-based survey data during 2008 to 2013. The estimated adolescent pregnancy rates /1000 adolescent women (95% confidence intervals (CI)) ranged from 56.3 (49.4, 66.9) to 70.3 (60.3, 76.6). The estimated rates were about 12-31% higher than adolescent birth rates reported by the Thailand Public Health Statistics.

CONCLUSIONS

With the capture-recapture method, more accurate adolescent pregnancy rates were estimated. This method should be able to apply to any setting with similar context.

The identity of two meta-analytic likelihoods and the ignorability of double-zero studies

In meta-analysis, the conventional two-stage approach computes an effect estimate for each study in the first stage and proceeds with the analysis of effect estimates in the second stage. For counts of events as outcome, the risk ratio is often the effect measure of choice. However, if the meta-analysis includes many studies with no events the conventional method breaks down. As an alternative one-stage approach, a Poisson regression model and a conditional binomial model can be considered where no event studies do not cause problems. The conditional binomial model excludes double-zero studies, whereas this is seemingly not the case for the Poisson regression approach. However, we show here that both models lead to the same likelihood inference and double-zero studies (in contrast to single-zero studies) do not contribute in either case to the likelihood.

Renewed: Protocol for a randomised controlled trial of a digital intervention to support quality of life in cancer survivors

INTRODUCTION

Low quality of life is common in cancer survivors. Increasing physical activity, improving diet, supporting psychological well-being and weight loss can improve quality of life in several cancers and may limit relapse. The aim of the randomised controlled trial outlined in this protocol is to examine whether a digital intervention (Renewed), with or without human support, can improve quality of life in cancer survivors. Renewed provides support for increasing physical activity, managing difficult emotions, eating a healthier diet and weight management.

METHODS AND ANALYSIS

A randomised controlled trial is being conducted comparing usual care, access to Renewed or access to Renewed with brief human support. Cancer survivors who have had colorectal, breast or prostate cancer will be identified and invited through general practice searches and mail-outs. Participants are asked to complete baseline measures immediately after screening and will then be randomised to a study group; this is all completed on the Renewed website. The primary outcome is quality of life measured by the European Organization for Research and Treatment of Cancer QLQ-c30. Secondary outcomes include anxiety and depression, fear of cancer recurrence, general well-being, enablement and items relating to costs for a health economics analysis. Process measures include perceptions of human support, intervention usage and satisfaction, and adherence to behavioural changes. Qualitative process evaluations will be conducted with patients and healthcare staff providing support.

ETHICS AND DISSEMINATION

The trial has been approved by the NHS Research Ethics Committee (Reference 18/NW/0013). The results of this trial will be published in peer-reviewed journals and through conference presentations.

TRIAL REGISTRATION NUMBER

ISRCTN96374224; Pre-results.

Virtual chromoendoscopy for the real-time assessment of colorectal polyps in vivo: a systematic review and economic evaluation

BACKGROUND

Current clinical practice is to remove a colorectal polyp detected during colonoscopy and determine whether it is an adenoma or hyperplastic by histopathology. Identifying adenomas is important because they may eventually become cancerous if untreated, whereas hyperplastic polyps do not usually develop into cancer, and a surveillance interval is set based on the number and size of adenomas found. Virtual chromoendoscopy (VCE) (an electronic endoscopic imaging technique) could be used by the endoscopist under strictly controlled conditions for real-time optical diagnosis of diminutive (≤ 5 mm) colorectal polyps to replace histopathological diagnosis.

OBJECTIVE

To assess the clinical effectiveness and cost-effectiveness of the VCE technologies narrow-band imaging (NBI), flexible spectral imaging colour enhancement (FICE) and i-scan for the characterisation and management of diminutive (≤ 5 mm) colorectal polyps using high-definition (HD) systems without magnification.

DESIGN

Systematic review and economic analysis.

PARTICIPANTS

People undergoing colonoscopy for screening or surveillance or to investigate symptoms suggestive of colorectal cancer.

INTERVENTIONS

NBI, FICE and i-scan.

MAIN OUTCOME MEASURES

Diagnostic accuracy, recommended surveillance intervals, health-related quality of life (HRQoL), adverse effects, incidence of colorectal cancer, mortality and cost-effectiveness of VCE compared with histopathology.

DATA SOURCES

Electronic bibliographic databases including MEDLINE, EMBASE, The Cochrane Library and Database of Abstracts of Reviews of Effects were searched for published English-language studies from inception to June 2016. Bibliographies of related papers, systematic reviews and company information were screened and experts were contacted to identify additional evidence.

REVIEW METHODS

Systematic reviews of test accuracy and economic evaluations were undertaken in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. Meta-analyses were conducted, where possible, to inform the independent economic model. A cost-utility decision-analytic model was developed to estimate the cost-effectiveness of VCE compared with histopathology. The model used a decision tree for patients undergoing endoscopy, combined with estimates of long-term outcomes (e.g. incidence of colorectal cancer and subsequent morbidity and mortality) derived from University of Sheffield School of Health and Related Research's bowel cancer screening model. The model took a NHS perspective, with costs and benefits discounted at 3.5% over a lifetime horizon. There were limitations in the data on the distribution of adenomas across risk categories and recurrence rates post polypectomy.

RESULTS

Thirty test accuracy studies were included: 24 for NBI, five for i-scan and three for FICE (two studies assessed two interventions). Polyp assessments made with high confidence were associated with higher sensitivity and endoscopists experienced in VCE achieved better results than those without experience. Two economic evaluations were included. NBI, i-scan and FICE are cost-saving strategies compared with histopathology and the number of quality-adjusted life-years gained was similar for histopathology and VCE. The correct surveillance interval would be given to 95% of patients with NBI, 94% of patients with FICE and 97% of patients with i-scan.

LIMITATIONS

Limited evidence was available for i-scan and FICE and there was heterogeneity among the NBI studies. There is a lack of data on longer-term health outcomes of patients undergoing VCE for assessment of diminutive colorectal polyps.

CONCLUSIONS

VCE technologies, using HD systems without magnification, could potentially be used for the real-time assessment of diminutive colorectal polyps, if endoscopists have adequate experience and training.

FUTURE WORK

Future research priorities include head-to-head randomised controlled trials of all three VCE technologies; more research on the diagnostic accuracy of FICE and i-scan (when used without magnification); further studies evaluating the impact of endoscopist experience and training on outcomes; studies measuring adverse effects, HRQoL and anxiety; and longitudinal data on colorectal cancer incidence, HRQoL and mortality.

STUDY REGISTRATION

This study is registered as PROSPERO CRD42016037767.

FUNDING

The National Institute for Health Research Health Technology Assessment programme.

Delayed antibiotic prescribing for respiratory tract infections: protocol of an individual patient data meta-analysis

INTRODUCTION

Delayed prescribing can be a useful strategy to reduce antibiotic prescribing, but it is not clear for whom delayed prescribing might be effective. This protocol outlines an individual patient data (IPD) meta-analysis of randomised controlled trials (RCTs) and observational cohort studies to explore the overall effect of delayed prescribing and identify key patient characteristics that are associated with efficacy of delayed prescribing.

METHODS AND ANALYSIS

A systematic search of the databases Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Ovid Embase, EBSCO CINAHL Plus and Web of Science was conducted to identify relevant studies from inception to October 2017. Outcomes of interest include duration of illness, severity of illness, complication, reconsultation and patient satisfaction. Study authors of eligible papers will be contacted and invited to contribute raw IPD data. IPD data will be checked against published data, harmonised and aggregated to create one large IPD database. Multilevel regression will be performed to explore interaction effects between treatment allocation and patient characteristics. The economic evaluation will be conducted based on IPD from the combined trial and observational studies to estimate the differences in costs and effectiveness for delayed prescribing compared with normal practice. A decision model will be developed to assess potential savings and cost-effectiveness in terms of reduced antibiotic usage of delayed prescribing and quality-adjusted life years.

ETHICS AND DISSEMINATION

Ethical approval was obtained from the University of Southampton Faculty of Medicine Research Ethics Committee (Reference number: 30068). Findings of this study will be published in peer-reviewed academic journals as well as General Practice trade journals and will be presented at national and international conferences. The results will have important public health implications, shaping the way in which antibiotics are prescribed in the future and to whom delayed prescriptions are issued.

PROSPERO REGISTRATION NUMBER

CRD42018079400.

Nurse staffing levels, missed vital signs and mortality in hospitals: retrospective longitudinal observational study

Background

Low nurse staffing levels are associated with adverse patient outcomes from hospital care, but the causal relationship is unclear. Limited capacity to observe patients has been hypothesised as a causal mechanism.

Objectives

This study determines whether or not adverse outcomes are more likely to occur after patients experience low nurse staffing levels, and whether or not missed vital signs observations mediate any relationship.

Design

Retrospective longitudinal observational study. Multilevel/hierarchical mixed-effects regression models were used to explore the association between registered nurse (RN) and health-care assistant (HCA) staffing levels and outcomes, controlling for ward and patient factors.

Setting and participants

A total of 138,133 admissions to 32 general adult wards of an acute hospital from 2012 to 2015.

Main outcomes

Death in hospital, adverse event (death, cardiac arrest or unplanned intensive care unit admission), length of stay and missed vital signs observations.

Data sources

Patient administration system, cardiac arrest database, eRoster, temporary staff bookings and the Vitalpac system (System C Healthcare Ltd, Maidstone, Kent; formerly The Learning Clinic Limited) for observations.

Results

Over the first 5 days of stay, each additional hour of RN care was associated with a 3% reduction in the hazard of death [hazard ratio (HR) 0.97, 95% confidence interval (CI) 0.94 to 1.0]. Days on which the HCA staffing level fell below the mean were associated with an increased hazard of death (HR 1.04, 95% CI 1.02 to 1.07), but the hazard of death increased as cumulative staffing exposures varied from the mean in either direction. Higher levels of temporary staffing were associated with increased mortality. Adverse events and length of stay were reduced with higher RN staffing. Overall, 16% of observations were missed. Higher RN staffing was associated with fewer missed observations in high-acuity patients (incidence rate ratio 0.98, 95% CI 0.97 to 0.99), whereas the overall rate of missed observations was related to overall care hours (RN + HCA) but not to skill mix. The relationship between low RN staffing and mortality was mediated by missed observations, but other relationships between staffing and mortality were not. Changing average skill mix and staffing levels to the levels planned by the Trust, involving an increase of 0.32 RN hours per patient day (HPPD) and a similar decrease in HCA HPPD, would be associated with reduced mortality, an increase in staffing costs of £28 per patient and a saving of £0.52 per patient per hospital stay, after accounting for the value of reduced stays.

Limitations

This was an observational study in a single site. Evidence of cause is not definitive. Variation in staffing could be influenced by variation in the assessed need for staff. Our economic analysis did not consider quality or length of life.

Conclusions

Higher RN staffing levels are associated with lower mortality, and this study provides evidence of a causal mechanism. There may be several causal pathways and the absolute rate of missed observations cannot be used to guide staffing decisions. Increases in nursing skill mix may be cost-effective for improving patient safety.

Future work

More evidence is required to validate approaches to setting staffing levels. Other aspects of missed nursing care should be explored using objective data. The implications of findings about both costs and temporary staffing need further exploration.

Trial registration

This study is registered as ISRCTN17930973.

Funding

This project was funded by the National Institute for Health Research (NIHR) Health Services and Delivery Research programme and will be published in full in Health Services and Delivery Research; Vol. 6, No. 38. See the NIHR Journals Library website for further project information.

A correction to the age-adjustment of the GH-2000 score used in the detection of growth hormone misuse

OBJECTIVE

The GH-2000 biomarker test has been introduced by the World Anti-Doping Agency as a method of detecting growth hormone misuse in professional sport. The test involves the measurement insulin-like growth factor-I and the amino-terminal pro-peptide of type III collagen (P-III-NP) which increase in a dose-dependent manner in response to GH. These measurements are combined in sex specific formulae that include an age adjustment. The original age adjustment overcorrects the effect of age in male athletes and could potentially place older men at a disadvantage. The purpose of this note is to investigate the performance of a previously suggested correction term in two new and larger data sets.

RESULTS

The GH-2000 score was calculated for 7307 samples obtained from 15 accredited WADA laboratories in 2017 and 3916 samples measured at Drug Control Centre, King's College London, UK between 2013 and 2017. The GH-2000 scores were investigated for positive age effects using standard regression modelling. As previously, all analyses confirmed a positive age effect. Applying the earlier suggested correction term of 0.032 × age showed a significant over-correction leading to a negative association of the GH-2000 score with age. We now suggest a smaller age correction of 0.020 × age, which corresponds to the smallest effect found in the earlier studies.

Non-invasive tests for the detection of oesophageal varices in compensated cirrhosis: systematic review and meta-analysis

INTRODUCTION

Conclusive data on the accuracy and clinical applicability of non-invasive screening tests for oesophageal varices (OV) in patients with compensated cirrhosis remain lacking. We conducted this study to identify currently available tests, estimate their diagnostic performance and then exemplify how these could be utilized in clinical practice.

MATERIALS AND METHODS

A systematic literature search was performed to identify all primary studies that reported accuracy using oesophagogastroduodenoscopy (OGD) as the gold standard. Sources searched included Ovid MEDLINE, Ovid EMBASE and The Cochrane Library databases.

RESULTS

Twenty-one studies with a total of 2471 patients were identified. Several tests were evaluated in more than three studies. Platelet count/spleen diameter ratio (PSR) had the highest summary area under the curve for detection of any size OV of 0.85 (95% confidence interval 0.78-0.92). At a cut-off of 909 (n = 4 studies) and prevalence rates of 10, 20, 30, 40 and 50% for OV, PSR screening correctly avoided the need for OGD in 70, 62, 55, 47 and 39% of patients, respectively.

CONCLUSIONS

PSR appears to be the most accurate and validated non-invasive screening test for OV in patients with compensated cirrhosis. At a cut-off of 909, PSR could be clinically useful to avoid OGDs in a significant proportion of patients.

Why do endocrine profiles in elite athletes differ between sports?

BACKGROUND

Endocrine profiles have been measured on blood samples obtained immediately post-competition from 693 elite athletes from 15 Olympic Sports competing at National or International level; four were subsequently excluded leaving 689 for the current analysis.

METHODS

Body composition was measured by bioimpedance in a sub-set of 234 (146 men and 88 women) and from these data a regression model was constructed that enabled 'estimated' lean body mass and fat mass to be calculated on all athletes. One way ANOVA was used to assess the differences in body composition and endocrine profiles between the sports and binary logistical regression to ascertain the characteristic of a given sport compared to the others.

RESULTS

The results confirmed many suppositions such as basketball players being tall, weightlifters short and cross-country skiers light. The hormone profiles were more surprising with remarkably low testosterone and free T3 (tri-iodothyronine) in male powerlifters and high oestradiol, SHBG (sex hormone binding globulin) and prolactin in male track and field athletes. Low testosterone concentrations were seen 25.4% of male elite competitors in 12 of the 15 sports and high testosterone concentrations in 4.8% of female elite athletes in 3 of the 8 sports tested. Interpretation of the results is more difficult; some of the differences between sports are at least partially due to differences in age of the athletes but the apparent differences between sports remain significant after adjusting for age. The prevalence of 'hyperandrogenism' (as defined by the IAAF (International Association of Athletics Federations) and IOC (International Olympic Committee)) amongst this cohort of 231 elite female athletes was the highest so far recorded and the very high prevalence of 'hypoandrogenism' in elite male athletes a new finding.

CONCLUSIONS

It is unclear whether the differences in hormone profiles between sports is a reason why they become elite athletes in that sport or is a consequence of the arduous processes involved. For components of body composition we know that most have a major genetic component and this may well be true for endocrine profiles.

Meta-Analysis

Objectives: This contribution provides a unifying concept for meta-analysis integrating the handling of unobserved heterogeneity, study covariates, publication bias and study quality. It is important to consider these issues simultaneously to avoid the occurrence of artifacts, and a method for doing so is suggested here.

Methods: The approach is based upon the meta-likelihood in combination with a general linear nonparametric mixed model, which lays the ground for all inferential conclusions suggested here.

Results: The concept is illustrated at hand of a meta-analysis investigating the relationship of hormone replacement therapy and breast cancer. The phenomenon of interest has been investigated in many studies for a considerable time and different results were reported. In 1992 a meta-analysis by Sillero-Arenas et al. [1] concluded a small, but significant overall effect of 1.06 on the relative risk scale. Using the meta-likelihood approach it is demonstrated here that this meta-analysis is due to considerable unobserved heterogeneity. Furthermore, it is shown that new methods are available to model this heterogeneity successfully. It is argued further to include available study covariates to explain this heterogeneity in the meta-analysis at hand.

Conclusions: The topic of HRT and breast cancer has again very recently become an issue of public debate, when results of a large trial investigating the health effects of hormone replacement therapy were published indicating an increased risk for breast cancer (risk ratio of 1.26). Using an adequate regression model in the previously published meta-analysis an adjusted estimate of effect of 1.14 can be given which is considerably higher than the one published in the meta-analysis of Sillero-Arenas et al. [1]. In summary, it is hoped that the method suggested here contributes further to a good meta-analytic practice in public health and clinical disciplines.

Erratum to: “Youden’s Index and the Likelihood Ratio Positive in Diagnostic Testing”

Unfortunately, the letter “Youden’s Index and the Likelihood Ratio Positive in Diagnostic Testing“ by D. Böhning contained an error. It contained the wrong Figure 4. The correct Figure is now supplied here.

D. Böhning sent us this statement: “I am grateful to Professor Hughes for pointing this out. There was no intention with my own letter to be critical on the letter by Professor Hughes or dispute its results but rather to add some thoughts of my own to the subject of interest.“

Log-likelihood positive as a function of a cut-off value c for two normal distributions with equal unit variance and mean difference of 2

Youden’s Index and the Likelihood Ratio Positive in Diagnostic Testing

We refer to a recent letter to the editor by Hughes [1] and show that, despite existing similarities between Youden’s index and the log-likelihood ratio positive, important differences between these two measures remain to exist which can play an important difference in clinical practice.

Exact statistical calculation of the uncertainty term in the decision limits of the GH-2000 score for growth hormone misuse (doping) detection

The GH-2000 score has been developed as a powerful and unique technique for the detection of growth hormone misuse by sportsmen and women. The score depends upon the measurement of two growth hormone sensitive markers, insulin-like growth factor-I and the amino-terminal pro-peptide of type III collagen. It also includes a term to adjust for the age of the athlete. Decision limits for the GH-2000 score have been developed and are incorporated into the guidelines of the World Anti-Doping Agency. These decision limits are derived by setting a 1 in 10,000 false-positive rate rule. As these decision limits are estimated from samples of GH-2000 scores, they carry uncertainty. In previous work, this uncertainty has been addressed by establishing an upper 95% confidence interval for the true decision limits based on a normal approximation which has been shown to be appropriate if sample sizes are large (such as 1000 and above). Here, we show that these approximations, whether reasonable or not, can be entirely avoided by developing an upper 95% confidence interval for the true decision limits using an approach based upon the t-distribution. While there are considerable differences for smaller sample sizes, these become negligible when the sample size is large such as 1000 and above.