Bayesian network data imputation with application to survival tree analysis

Retrospective clinical datasets are often characterized by a relatively small sample size and many missing data. In this case, a common way for handling the missingness consists in discarding from the analysis patients with missing covariates, further reducing the sample size. Alternatively, if the mechanism that generated the missing allows, incomplete data can be imputed on the basis of the observed data, avoiding the reduction of the sample size and allowing methods to deal with complete data later on. Moreover, methodologies for data imputation might depend on the particular purpose and might achieve better results by considering specific characteristics of the domain. The problem of missing data treatment is studied in the context of survival tree analysis for the estimation of a prognostic patient stratification. Survival tree methods usually address this problem by using surrogate splits, that is, splitting rules that use other variables yielding similar results to the original ones. Instead, our methodology consists in modeling the dependencies among the clinical variables with a Bayesian network, which is then used to perform data imputation, thus allowing the survival tree to be applied on the completed dataset. The Bayesian network is directly learned from the incomplete data using a structural expectation–maximization (EM) procedure in which the maximization step is performed with an exact anytime method, so that the only source of approximation is due to the EM formulation itself. On both simulated and real data, our proposed methodology usually outperformed several existing methods for data imputation and the imputation so obtained improved the stratification estimated by the survival tree (especially with respect to using surrogate splits).

Effects of baited crab pots on cultivated mussel (Mytilus edulis) survival rates

The shore crab, Carcinus maenas, is recognized as a voracious predator of blue mussels, Mytilus edulis, having the potential to greatly reduce stocks in the benthic cultivation industry. As a consequence, baited crab pots are often deployed on and around cultivated benthic mussel beds to trap and remove crabs, in an attempt to reduce predatory pressure. Little is known about how C. maenas behaves around crab pots, but for many other crustacean fisheries, the trapping efficiency of pots is often low. Crabs may be attracted towards but not enter pots, instead feeding on cultivated mussels outside pots on the surrounding substratum. We tested whether the rate of loss of mussels attached to plates differed in areas next to baited pots compared with unbaited pots and to areas without any pots, at two sea loughs (60 km apart) in Northern Ireland. In Strangford Lough, more mussels were lost from plates next to baited pots than the other treatments. In Carlingford Lough, however, we found no difference in the number of mussels lost from plates in any treatment. This difference could be attributed to the different assemblages of mobile benthic predators at the two loughs. The presence of the starfish Asterias rubens, which was absent from experimental sites in Carlingford Lough, was thought to be responsible for increased predation rates near baited pots in Strangford. It is, therefore, important to consider local predator communities when deploying crab pots as a predator mitigation technique to ensure predation rates are in fact reduced and not enhanced. This study is of relevance not only to attempts to limit predation on commercial stocks of benthic cultivated mussels but also in situations where baited traps are deployed close to species vulnerable to mobile benthic predators.

Digoxin use after diagnosis of breast cancer and survival: a population-based cohort study

Digoxin has been shown to have an estrogenic effect and is associated with increased risk of gynecomastia and estrogen-sensitive cancers such as breast and uterus cancer. These findings, particularly recent observations of increased breast cancer risk, raise questions about the safety of digoxin use in breast cancer patients. Therefore, we investigated whether digoxin use after breast cancer diagnosis increased the risk of breast cancer-specific mortality in breast cancer patients. A cohort of 17,842 breast cancer patients newly diagnosed from 1998 to 2009 was identified from English cancer registries (from the National Cancer Data Repository). This cohort was linked to the UK Clinical Practice Research Datalink (to provide digoxin and other prescription records) and to the Office of National Statistics mortality data (to identify breast cancer-specific deaths). Using time-dependent Cox regression models, unadjusted and adjusted hazard ratios (HR) and 95 % confidence intervals (CIs) were calculated for the association between post-diagnostic exposure to digoxin and breast cancer-specific and all-cause mortality. In 17,842 breast cancer patients, there were 2219 breast cancer-specific deaths. Digoxin users appeared to have increased breast cancer-specific mortality compared with non-users (HR 1.73; 95 % CI 1.39–2.15) but this association was entirely attenuated after adjustment for potential confounders (adjusted HR 0.91; 95 % CI 0.72–1.14). In this large population-based breast cancer cohort study, there was little evidence of an increase in breast cancer-specific mortality with digoxin use after diagnosis. These results provide some reassurance that digoxin use is safe in breast cancer patients.

Time and Cell Type Dependency of Survival Responses in Co-cultured Tumor and Fibroblast Cells after Exposure to Modulated Radiation Fields

Advanced radiotherapy techniques such as intensity-modulated radiation therapy (IMRT) achieve high levels of conformity to the target volume through the sequential delivery of highly spatially and temporally modulated radiation fields, which have been shown to impact radiobiological response. This study aimed to characterize the time and cell type dependency of survival responses to modulated fields using single cell type (SCT) and mixed cell type (MCT) co-culture models of transformed fibroblast (AG0-1522b) cells, and prostate (DU-145) and lung (H460) cancer cells. In SCT cultures, in-field responses showed no significant time dependency while out-of-field responses occurred early, and plateaued 6 h after irradiation in both DU-145 and H460 cells. Under modulated beam configurations MCT co-cultures showed cell-specific, differential out-of-field responses depending on the irradiated in-field and responding out-of-field cell type. The observed differential out-of-field responses may be due to the genetic background of the cells, in particular p53 status, which has been shown to mediate radiation-induced bystander effects (RIBEs). These data provide further insight into the radiobiological parameters that influence out-of-field responses, which have potential implications for advanced radiotherapy modalities and may provide opportunities for biophysical optimization in radiotherapy treatment planning.

Intracellular survival of Burkholderia cepacia complex in phagocytic cells

Burkholderia cepacia complex (Bcc) species are a group of Gram-negative opportunistic pathogens that infect the airways of cystic fibrosis patients, and occasionally they infect other immunocompromised patients. Bcc bacteria display high-level multidrug resistance, and chronically persist in the infected host while eliciting robust inflammatory responses. Studies using macrophages, neutrophils and dendritic cells, combined with advances to genetically manipulate these bacteria have increased our understanding of the molecular mechanisms of virulence in these pathogens and the molecular details of cell-host responses triggering inflammation. This article discusses our current view of the intracellular survival of B. cenocepacia within macrophages.

Statin use after colorectal cancer diagnosis and survival: a population-based cohort study

PURPOSE: To investigate whether statins used after colorectal cancer diagnosis reduce the risk of colorectal cancer-specific mortality in a cohort of patients with colorectal cancer.

PATIENTS AND METHODS: A cohort of 7,657 patients with newly diagnosed stage I to III colorectal cancer were identified from 1998 to 2009 from the National Cancer Data Repository (comprising English cancer registry data). This cohort was linked to the United Kingdom Clinical Practice Research Datalink, which provided prescription records, and to mortality data from the Office of National Statistics (up to 2012) to identify 1,647 colorectal cancer-specific deaths. Time-dependent Cox regression models were used to calculate hazard ratios (HR) for cancer-specific mortality and 95% CIs by postdiagnostic statin use and to adjust these HRs for potential confounders.

RESULTS: Overall, statin use after a diagnosis of colorectal cancer was associated with reduced colorectal cancer-specific mortality (fully adjusted HR, 0.71; 95% CI, 0.61 to 0.84). A dose-response association was apparent; for example, a more marked reduction was apparent in colorectal cancer patients using statins for more than 1 year (adjusted HR, 0.64; 95% CI, 0.53 to 0.79). A reduction in all-cause mortality was also apparent in statin users after colorectal cancer diagnosis (fully adjusted HR, 0.75; 95% CI, 0.66 to 0.84).

CONCLUSION: In this large population-based cohort, statin use after diagnosis of colorectal cancer was associated with longer rates of survival.

Explaining variation in cancer survival between 11 jurisdictions in the International Cancer Benchmarking Partnership: a primary care vignette survey

OBJECTIVES: The International Cancer Benchmarking Partnership (ICBP) is a collaboration between 6 countries and 12 jurisdictions with similar primary care-led health services. This study investigates primary care physician (PCP) behaviour and systems that may contribute to the timeliness of investigating for cancer and subsequently, international survival differences.

DESIGN: A validated survey administered to PCPs via the internet set out in two parts: direct questions on primary care structure and practice relating to cancer diagnosis, and clinical vignettes, assessing management of scenarios relating to the diagnosis of lung, colorectal or ovarian cancer.

PARTICIPANTS: 2795 PCPs in 11 jurisdictions: New South Wales and Victoria (Australia), British Columbia, Manitoba, Ontario (Canada), England, Northern Ireland, Wales (UK), Denmark, Norway and Sweden.

PRIMARY AND SECONDARY OUTCOME MEASURES: Analysis compared the cumulative proportion of PCPs in each jurisdiction opting to investigate or refer at each phase for each vignette with 1-year survival, and conditional 5-year survival rates for the relevant cancer and jurisdiction. Logistic regression was used to explore whether PCP characteristics or system differences in each jurisdiction affected the readiness to investigate.

RESULTS: 4 of 5 vignettes showed a statistically significant correlation (p<0.05 or better) between readiness to investigate or refer to secondary care at the first phase of each vignette and cancer survival rates for that jurisdiction. No consistent associations were found between readiness to investigate and selected PCP demographics, practice or health system variables.

CONCLUSIONS: We demonstrate a correlation between the readiness of PCPs to investigate symptoms indicative of cancer and cancer survival rates, one of the first possible explanations for the variation in cancer survival between ICBP countries. No specific health system features consistently explained these findings. Some jurisdictions may consider lowering thresholds for PCPs to investigate for cancer-either directly, or by specialist referral, to improve outcomes.

Survival and behaviour of the earthworms Lumbricus rubellus and Dendrodrilus rubidus from arsenate-contaminated and non-contaminated sites

Two arsenic- and heavy metal-contaminated mine-spoil sites, at Carrock Fell, Cumbria and Devon Great Consols Mine, Devon, were found to support populations of the earthworms Lumbricus rubellus Hoffmeister and Dendrodrilus rubidus (Savigny). L. rubellus and D. rubidus collected from the Devon site and an uncontaminated site were kept for 28 days in uncontaminated soil and in soil containing sodium arsenate (494 mg As kg^-1). The state of the specimens was recorded every 7 days using a semi-quantitative assessment of earthworm health (condition index, C. I.). The C. I. remained high for all specimens except those of L. rubellus and D. rubidus from uncontaminated sites, which displayed 60 and 10% mortality, respectively. L. rubellus collected from the Carrock Fell site, and L. rubellus and D. rubidus from an uncontaminated site, burrowed as rapidly into soil containing up to 1235 mg As kg^-1 in the form of sodium arsenate as into uncontaminated soil when placed on the soil surface. When earthworms were allowed a choice between uncontaminated soil and soil contaminated with sodium arsenate in concentrations of up to 1235 mg As kg^-1, the threshold concentration for avoidance of contaminated soil was lower for L. rubellus and D. rubidus from uncontaminated soil than for specimens from contaminated soil. There was no significant effect of pH on soil discrimination. The LC₅₀ concentration of As for L. rubellus from Devon Great Consols was significantly higher (P < 0.001) than for L. rubellus from the uncontaminated site: 1510 and 96 mg As kg^-1, respectively. © 2001 Elsevier Science Ltd. All rights reserved.

Molecular mechanisms of virulence of Burkholderia cepacia complex bacteria: Survival in Phagocytic Cells

Burkholderia cepacia complex (Bcc) species are a group of Gram-negative opportunistic pathogens that chronically infect the airways of cystic fibrosis patients, but they can also infect patients with various types of immunosuppressive disorders. Bcc members are multidrug resistant bacteria that have the ability to persist in the infected host and also elicit robust inflammatory responses. Studies using macrophages, neutrophils and dendritic cells, combined with dramatic advances in the ability to genetically manipulate these microorganisms have contributed to increase our understanding of the molecular mechanisms of virulence in these pathogens and the molecular details of the cell host responses triggering inflammation. This chapter reviews our understanding of the pathogenic mechanisms used by Bcc to establish an intracellular niche in phagocytic cells and modulate host cell responses that ultimately end up in cell death and a proinflammatory response.

Long-Term Imprisonment in Northern Ireland:Psychological or Political Survival?

The chapter reviews the literature on the affects of long-term imprisonment with a focus on concepts of psychological survival, learned helplessness and institutionalization. It then discusses trends in imprisonment in Northern Ireland and the exceptionally high proportion of prisoners serving life sentences. Through interviews with prisoners and their relatives, a critique of the literature is made in terms of the failure to take account of cases of collective prisoner organization and strong political motivation.

Quality of reporting of dental survival analyses

To explore the quality of reporting (writing and graphics) of articles that used time-to-event analyses to report dental treatment outcomes. A systematic search of the top 50 dental journals in 2008 produced the sample of articles for this analysis. Articles reporting treatment outcomes with (n = 95) and without (n = 91) time-to-event statistics were reviewed. Survival descriptive words used in the two groups were analysed (Pearson's chi-square). The quality of life tables, survival curves and time-to-event statistics were assessed (Kappa analysed agreement) and explored. Words describing dental outcomes 'over time' were more common in time-to-event compared with control articles (77%, 3%, P < 0.001). Non-specific use of 'rate' was common across both groups. Life tables and survival curves were used by 39% and 48% of the time-to-event articles, with at least one used by 82%. Construction quality was poor: 21% of life tables and 28% of survival curves achieved an acceptable standard. Time-to-event statistical reporting was poor: 3% achieved a high and 59% achieved an acceptable standard. The survival statistic, summary figure and standard error were reported in 76%, 95% and 20% of time-to-event articles. Individual statistical terms and graphic aids were common within and unique to time-to-event articles. Unfortunately, important details were regularly omitted from statistical descriptions and survival figures making the overall quality poor. It is likely this will mean such articles will be incorrectly indexed in databases, missed by searchers and unable to be understood completely if identified.

Accuracy of medical subject heading indexing of dental survival analyses

PURPOSE: To assess the Medical Subject Headings (MeSH) indexing of articles that employed time-to-event analyses to report outcomes of dental treatment in patients.

MATERIALS AND METHODS: Articles published in 2008 in 50 dental journals with the highest impact factors were hand searched to identify articles reporting dental treatment outcomes over time in human subjects with time-to-event statistics (included, n = 95), without time-to-event statistics (active controls, n = 91), and all other articles (passive controls, n = 6,769). The search was systematic (kappa 0.92 for screening, 0.86 for eligibility). Outcome-, statistic- and time-related MeSH were identified, and differences in allocation between groups were analyzed with chi-square and Fischer exact statistics.

RESULTS: The most frequently allocated MeSH for included and active control articles were "dental restoration failure" (77% and 52%, respectively) and "treatment outcome" (54% and 48%, respectively). Outcome MeSH was similar between these groups (86% and 77%, respectively) and significantly greater than passive controls (10%, P < .001). Significantly more statistical MeSH were allocated to the included articles than to the active or passive controls (67%, 15%, and 1%, respectively, P < .001). Sixty-nine included articles specifically used Kaplan-Meier or life table analyses, but only 42% (n = 29) were indexed as such. Significantly more time-related MeSH were allocated to the included than the active controls (92% and 79%, respectively, P = .02), or to the passive controls (22%, P < .001).

CONCLUSIONS: MeSH allocation within MEDLINE to time-to-event dental articles was inaccurate and inconsistent. Statistical MeSH were omitted from 30% of the included articles and incorrectly allocated to 15% of active controls. Such errors adversely impact search accuracy.