Enamel Caries Detection and Diagnosis: An Analysis of Systematic Reviews.

Detection and diagnosis of caries-typically undertaken through a visual-tactile examination, often with supporting radiographic investigations-is commonly regarded as being broadly effective at detecting caries that has progressed into dentine and reached a threshold where restoration is necessary. With earlier detection comes an opportunity to stabilize disease or even remineralize the tooth surface, maximizing retention of tooth tissue and preventing a lifelong cycle of restoration. We undertook a formal comparative analysis of the diagnostic accuracy of different technologies to detect and inform the diagnosis of early caries using published Cochrane systematic reviews. Forming the basis of our comparative analysis were 5 Cochrane diagnostic test accuracy systematic reviews evaluating fluorescence, visual or visual-tactile classification systems, imaging, transillumination and optical coherence tomography, and electrical conductance or impedance technologies. Acceptable reference standards included histology, operative exploration, or enhanced visual assessment (with or without tooth separation) as appropriate. We conducted 2 analyses based on study design: a fully within-study, within-person analysis and a network meta-analysis based on direct and indirect comparisons. Nineteen studies provided data for the fully within-person analysis and 64 studies for the network meta-analysis. Of the 5 technologies evaluated, the greatest pairwise differences were observed in summary sensitivity points for imaging and all other technologies, but summary specificity points were broadly similar. For both analyses, the wide 95% prediction intervals indicated the uncertainty of future diagnostic accuracy across all technologies. The certainty of evidence was low, downgraded for study limitations, inconsistency, and indirectness. Summary estimates of diagnostic accuracy for most technologies indicate that the degree of certitude with which a decision is made regarding the presence or absence of disease may be enhanced with the use of such devices. However, given the broad prediction intervals, it is challenging to predict their accuracy in any future "real world" context.

Influence of El Niño on atmospheric CO2 over the tropical Pacific Ocean: Findings from NASA's OCO-2 mission.

Spaceborne observations of carbon dioxide (CO2) from the Orbiting Carbon Observatory-2 are used to characterize the response of tropical atmospheric CO2 concentrations to the strong El Niño event of 2015-2016. Although correlations between the growth rate of atmospheric CO2 concentrations and the El Niño-Southern Oscillation are well known, the magnitude of the correlation and the timing of the responses of oceanic and terrestrial carbon cycle remain poorly constrained in space and time. We used space-based CO2 observations to confirm that the tropical Pacific Ocean does play an early and important role in modulating the changes in atmospheric CO2 concentrations during El Niño events-a phenomenon inferred but not previously observed because of insufficient high-density, broad-scale CO2 observations over the tropics.

Cost-effectiveness of emergency versus delayed laparoscopic cholecystectomy for acute gallbladder pathology.

BACKGROUND: The optimal timing of cholecystectomy for patients admitted with acute gallbladder pathology is unclear. Some studies have shown that emergency cholecystectomy during the index admission can reduce length of hospital stay with similar rates of conversion to open surgery, complications and mortality compared with a 'delayed' operation following discharge. Others have reported that cholecystectomy during the index acute admission results in higher morbidity, extended length of stay and increased costs. This study examined the cost-effectiveness of emergency versus delayed cholecystectomy for acute benign gallbladder disease. METHODS: Using data from a prospective population-based cohort study examining the outcomes of cholecystectomy in the UK and Ireland, a model-based cost-utility analysis was conducted from the perspective of the UK National Health Service, with a 1-year time horizon for costs and outcomes. Probabilistic sensitivity analysis was used to investigate the impact of parameter uncertainty on the results obtained from the model. RESULTS: Emergency cholecystectomy was found to be less costly (£4570 versus £4720; €5484 versus €5664) and more effective (0·8868 versus 0·8662 QALYs) than delayed cholecystectomy. Probabilistic sensitivity analysis showed that the emergency strategy is more than 60 per cent likely to be cost-effective across willingness-to-pay values for the QALY from £0 to £100 000 (€0-120 000). CONCLUSION: Emergency cholecystectomy is less costly and more effective than delayed cholecystectomy. This approach is likely to be beneficial to patients in terms of improved health outcomes and to the healthcare provider owing to the reduced costs.

Cost-utility analysis of operative versus non-operative treatment for colorectal liver metastases.

BACKGROUND: Surgical resection of colorectal liver metastases (CRLMs) is the standard of care when possible, although this strategy has not been compared with non-operative interventions in controlled trials. Although survival outcomes are clear, the cost-effectiveness of surgery is not. This study aimed to estimate the cost-effectiveness of resection for CRLMs compared with non-operative treatment (palliative care including chemotherapy). METHODS: Operative and non-operative cohorts were identified from a prospectively maintained database. Patients in the operative cohort had a minimum of 10 years of follow-up. A model-based cost-utility analysis was conducted to quantify the mean cost and quality-adjusted life-years (QALYs) over a lifetime time horizon. The analysis was conducted from a healthcare provider perspective (UK National Health Service) in a secondary care (hospital) setting. RESULTS: Median survival was 41 and 21 months in the operative and non-operative cohorts respectively (P < 0·001). The operative strategy dominated non-operative treatments, being less costly (€22,200 versus €32,800) and more effective (4·017 versus 1·111 QALYs gained). The results of extensive sensitivity analysis showed that the operative strategy dominated non-operative treatment in every scenario. CONCLUSION: Operative treatment of CRLMs yields greater survival than non-operative treatment, and is both more effective and less costly.

Sentinel lymph node biopsy in vulval cancer: systematic review and meta-analysis.

BACKGROUND: The purpose of this study was to determine the accuracy of sentinel lymph node (SLN) biopsy with technetium 99 (99mTc) and/or blue dye-enhanced lymphoscintigraphy in vulval cancer. METHODS: Sensitive searches of databases were performed upto October 2013. Studies with at least 75% of women with FIGO stage IB or II vulval cancer evaluating SLN biopsy with 99mTc, blue dye or both with reference standard of inguinofemoral lymphadenectomy (IFL) or clinical follow-up were included. Meta-analyses were performed using Meta-Disc version 1.4. RESULTS: Of the 2950 references, 29 studies (1779 women) were included; most of them evaluated 99mTc combined with blue dye. Of these, 24 studies reported results for SLN followed by IFL, and 5 reported clinical follow-up only for SLN negatives. Pooling of all studies was inappropriate because of heterogeneity. Mean SLN detection rates were 94.0% for 99mTc, 68.7% for blue dye and 97.7% for both. SLN biopsy had pooled sensitivity of 95% (95% CI 92-98%) with negative predictive value (NPV) of 97.9% in studies using 99mTc/blue dye, ultrastaging and immunohistochemistry with IFL as reference. Pooled sensitivity for SLN with clinical follow-up for SLN-negatives was 91% (85-95%) with NPV 95.6%. Patients undergoing SLN biopsy experienced less morbidity than those undergoing IFL. CONCLUSIONS: Sentinel lymph node biopsy using 99mTC, blue dye and ultrastaging with immunohistochemistry is highly accurate when restricted to carefully selected patients, within a rigorous protocol, with close follow-up and where sufficient numbers for learning curve optimisation exist. Patients must make an informed choice between the slightly higher groin recurrence rates of SLN biopsy vs the greater morbidity of IFL.

Cost-effectiveness of sentinel lymph node biopsy vs inguinofemoral lymphadenectomy in women with vulval cancer.

BACKGROUND: This study examines the cost-effectiveness of sentinel lymph node biopsy, a potentially less morbid procedure, compared with inguinofemoral lymphadenectomy (IFL) among women with stage I and stage II vulval squamous cell carcinoma. METHODS: A model-based economic evaluation was undertaken based on clinical evidence from a systematic review of published sources. A decision tree model was developed with the structure being informed by clinical input, taking the perspective of the health-care provider. RESULTS: For overall survival for 2 years, IFL was found to be the most cost-effective option and dominated all other strategies, being the least costly and most effective. For morbidity-free related outcomes for 2 years, sentinel lymph node (SLN) biopsy with 99mTc and blue dye and haematoxylin & eosin (H&E) histopathology, with ultrastaging and immunohistochemistry reserved for those that test negative following H&E is likely to be the most effective approach. CONCLUSION: SLN biopsy using 99mTc and blue dye with ultrastaging may be considered the most cost-effective strategy based on the outcome of survival free of morbidity for 2 years. The findings here also indicate that using blue dye and H&E for the identification of the SLN and the identification of metastasis, respectively, are not sensitive enough to be used on their own.

Acupuncture and other physical treatments for the relief of pain due to osteoarthritis of the knee: network meta-analysis.

OBJECTIVE: To compare the effectiveness of acupuncture with other relevant physical treatments for alleviating pain due to knee osteoarthritis. DESIGN: Systematic review with network meta-analysis, to allow comparison of treatments within a coherent framework. Comprehensive searches were undertaken up to January 2013 to identify randomised controlled trials in patients with osteoarthritis of the knee, which reported pain. RESULTS: Of 156 eligible studies, 114 trials (covering 22 treatments and 9,709 patients) provided data suitable for analysis. Most trials studied short-term effects and many were classed as being of poor quality with high risk of bias, commonly associated with lack of blinding (which was sometimes impossible to achieve). End of treatment results showed that eight interventions: interferential therapy, acupuncture, TENS, pulsed electrical stimulation, balneotherapy, aerobic exercise, sham acupuncture, and muscle-strengthening exercise produced a statistically significant reduction in pain when compared with standard care. In a sensitivity analysis of satisfactory and good quality studies, most studies were of acupuncture (11 trials) or muscle-strengthening exercise (9 trials); both interventions were statistically significantly better than standard care, with acupuncture being statistically significantly better than muscle-strengthening exercise (standardised mean difference: 0.49, 95% credible interval 0.00-0.98). CONCLUSIONS: As a summary of the current available research, the network meta-analysis results indicate that acupuncture can be considered as one of the more effective physical treatments for alleviating osteoarthritis knee pain in the short-term. However, much of the evidence in this area of research is of poor quality, meaning there is uncertainty about the efficacy of many physical treatments.

Estimating the variability in the risk of infection for hepatitis C in the Glasgow injecting drug user population.

Glasgow (Scotland's largest city) has a high prevalence of injecting drug use and has one of the highest prevalences of hepatitis C virus (HCV) infection in injecting drug users (IDUs) in Western Europe. HCV prevalence data from surveys of Glasgow's IDUs from 1990 to 2007 were utilized and a model was applied that described the prevalence of HCV as a function of the rate (force) of infection. Force-of-infection estimates for HCV that may vary over time and injecting career length over a range of variables were investigated. New initiates to injecting were found to be at increased risk of HCV infection, with being recruited from a street location and reporting injecting in prison leading to a significant increase in the risk of infection in new initiates. These results indicate areas of importance for the planning of public health measures that target the IDU population.

Modelling HIV in the injecting drug user population and the male homosexual population in a developed country context.

In many high income countries men who have sex with men (MSM) and injecting drug users (IDUs) are the two groups with the highest HIV prevalence. Yet these two groups are not mutually exclusive, and those MSM who are also IDUs (MSM-IDUs) may be particularly vulnerable to HIV infection. This may be particularly relevant to the IDU population in countries, like the UK, with a much lower HIV prevalence amongst IDUs than MSM, as the MSM-IDUs could provide a route of HIV infection into the IDU population. In this research two alternative modelling approaches that describe the transmission dynamics of HIV within the IDU, MSM, and heterosexual populations are proposed. These models are constructed with two aims. The first is to investigate the possible impact of interventions that target HIV transmission in the MSM and IDU populations, and the second aim is to investigate the impact of the model structure on the model results. An examination of the assortativity of mixing between risk groups is also undertaken. The models are parameterised for England and Wales. While the MSM-IDU population is small, targeting MSM-IDUs was the most efficient intervention strategy in terms of cases averted per 100 individuals targeted with the intervention. Sensitivity analysis showed that variations in the assumed assortativity of mixing between the population groups in both models have a large impact on model results. This means that to generate quantitatively robust estimates for the impact of different intervention strategies it will be necessary to obtain estimates for assortativity values through empirical work.

Modelling the transmission dynamics of Theileria annulata: model structure and validation for the Turkish context.

A mathematical model that describes the transmission dynamics of Theileria annulata is proposed that consists of 2 host components: the Hyalomma tick population and a compartmental model of T. annulata infection in the cattle population. The model was parameterized using data describing tick infestation and the infection status of cattle in Turkey from 2006 to 2008. The tick attachment rates are highly seasonal and because of the temporal separation of infectious and susceptible ticks virtually all ticks are infected by carrier cattle, so that annual peaks of disease in cattle do not impact on infection in the Hyalomma tick population. The impact of intervention measures that target the tick population both on the host and in the environment and their impact on the transmission of T. annulata were investigated. Interventions that have a limited 'one-off' impact and interventions that have a more permanent impact were both considered. The results from the model show the importance of targeting ticks during the period when they have left their first host as nymphs but have yet to feed on their second host.

A meta-analysis and metaregression analysis of factors influencing mortality after endovascular repair of ruptured abdominal aortic aneurysms.

OBJECTIVE: To determine factors that may influence the perioperative mortality after endovascular repair of ruptured abdominal aortic aneurysms (RAAAs) using metaregression analysis. METHODS: A meta-analysis of all English-language literature with information on mortality rates after endovascular repair of RAAAs was conducted. A metaregression was subsequently performed to determine the impact on mortality of the following 8 factors: patient age; mid-time study point; anaesthesia; endograft configuration; haemodynamic instability; use of aortic balloon; conversion to open repair; and abdominal compartment syndrome. RESULTS: The pooled perioperative mortality across the 46 studies (1397 patients) was 24.3% (95% CI: 20.7-28.3%). Of the 8 variables, only bifurcated approach was significantly associated with reduced mortality (p = 0.005). A moderate negative correlation was observed between bifurcated approach and haemodynamic instability (-0.35). There was still a strong association between bifurcated approach and mortality after simultaneously adjusting for haemodynamic instability, indicating that the latter was not a major factor in explaining the observed association. CONCLUSIONS: Endovascular repair of RAAAs is associated with acceptable mortality rates. Patients having a bifurcated endograft were less likely to die. This may be due to some surgeons opting for a bifurcated approach in patients with better haemodynamic condition. Further studies will be needed to clarify this.

Quality control in systematic reviews and meta-analyses.

Systematic reviews and meta-analyses are being submitted to, and being published by biomedical journals with increasing frequency. In order to maintain the utility of such publications and avoid misguidance it is important that these studies are conducted to a high standard. This article aims to provide guidance both for those researchers undertaking and reporting such studies and for the readers of such articles. Details of a suggested method for conducting a systematic review are given, including methods for literature searches, data abstraction and data extraction followed by a brief overview of common methods used for meta-analyses and the interpretation of the results of meta-analysis.

Dissemination and publication of research findings: an updated review of related biases.

OBJECTIVES: To identify and appraise empirical studies on publication and related biases published since 1998; to assess methods to deal with publication and related biases; and to examine, in a random sample of published systematic reviews, measures taken to prevent, reduce and detect dissemination bias. DATA SOURCES: The main literature search, in August 2008, covered the Cochrane Methodology Register Database, MEDLINE, EMBASE, AMED and CINAHL. In May 2009, PubMed, PsycINFO and OpenSIGLE were also searched. Reference lists of retrieved studies were also examined. REVIEW METHODS: In Part I, studies were classified as evidence or method studies and data were extracted according to types of dissemination bias or methods for dealing with it. Evidence from empirical studies was summarised narratively. In Part II, 300 systematic reviews were randomly selected from MEDLINE and the methods used to deal with publication and related biases were assessed. RESULTS: Studies with significant or positive results were more likely to be published than those with non-significant or negative results, thereby confirming findings from a previous HTA report. There was convincing evidence that outcome reporting bias exists and has an impact on the pooled summary in systematic reviews. Studies with significant results tended to be published earlier than studies with non-significant results, and empirical evidence suggests that published studies tended to report a greater treatment effect than those from the grey literature. Exclusion of non-English-language studies appeared to result in a high risk of bias in some areas of research such as complementary and alternative medicine. In a few cases, publication and related biases had a potentially detrimental impact on patients or resource use. Publication bias can be prevented before a literature review (e.g. by prospective registration of trials), or detected during a literature review (e.g. by locating unpublished studies, funnel plot and related tests, sensitivity analysis modelling), or its impact can be minimised after a literature review (e.g. by confirmatory large-scale trials, updating the systematic review). The interpretation of funnel plot and related statistical tests, often used to assess publication bias, was often too simplistic and likely misleading. More sophisticated modelling methods have not been widely used. Compared with systematic reviews published in 1996, recent reviews of health-care interventions were more likely to locate and include non-English-language studies and grey literature or unpublished studies, and to test for publication bias. CONCLUSIONS: Dissemination of research findings is likely to be a biased process, although the actual impact of such bias depends on specific circumstances. The prospective registration of clinical trials and the endorsement of reporting guidelines may reduce research dissemination bias in clinical research. In systematic reviews, measures can be taken to minimise the impact of dissemination bias by systematically searching for and including relevant studies that are difficult to access. Statistical methods can be useful for sensitivity analyses. Further research is needed to develop methods for qualitatively assessing the risk of publication bias in systematic reviews, and to evaluate the effect of prospective registration of studies, open access policy and improved publication guidelines.

Is volcanic air pollution associated with decreased heart-rate variability?

OBJECTIVES: To determine the autonomic cardiovascular control among residents of Hawaii who are exposed to varying levels of volcanic air pollution (vog), which consists largely of sulfur dioxide (SO(2)) and acid aerosols. METHODS: In a cross-sectional study between April 2006 and June 2008, the authors measured cardiovagal autonomic function by heart-rate variability (HRV) in 72 healthy individuals who lived in four exposure zones on Hawaii Island: vog-free (n=18); episodic exposure to SO(2) >200 ppb and acid aerosol (n=19); chronic exposure to SO(2) ≥30 ppb and acid aerosol (n=15); and chronic exposure to acid aerosols (n=20). Individuals with diabetes or heart disease, or who had smoked in the preceding month were excluded. HRV was measured in all subjects during rest, paced breathing and active standing (Ewing manoeuvre). HRV was analysed in time and frequency domains and compared between the four exposure zones. RESULTS: There were no significant differences between exposure zones in HRV, in either time or frequency domains, even after adjustment for age, gender, ethnicity and body mass index. There was no significant HRV change in three individuals in whom HRV was measured before and during an exposure to combined SO(2) 100-250 ppb and concentration of respirable particles of diameter ≥2.5 µ (PM(2.5)) >500 µg/m(3). Age was significantly correlated with time-domain parameters during paced breathing and the Ewing manoeuvre. CONCLUSIONS: This study of healthy individuals found no appreciable effects of vog on the autonomic nervous system.

Lymph node retrieval in colorectal cancer resection specimens: national standards are achievable, and low numbers are associated with reduced survival.

OBJECTIVES: The current guidelines identify the retrieval of at least 12 lymph nodes as crucial for accurate staging of colorectal cancer. We set out to review our figures from a single centre to see whether this standard has been met, and to examine for factors which may influence the number of lymph nodes retrieved. The influence of a low lymph node harvest on survival in patients with Dukes' A and B cancers was specifically investigated. METHOD: Data were collected from all patients with colorectal cancer undergoing resectional surgery from our prospectively compiled database between June 1998 and May 2007. A multivariate analysis was performed to identify factors resulting in low lymph node yields in those patients undergoing formal resection. Survival analyses were performed in patients with Dukes' A and B cancers to assess whether a low lymph node yield negatively impacted on survival. RESULTS: A total of 2449 patients underwent formal resection and were included in the analysis. The median lymph node retrieval was 13 nodes (range 0-136). On multivariate analysis, preoperative chemo-radiotherapy, operation type, specimen length and patient age all independently influenced lymph node retrieval. Patient gender, ethnicity, operative mode, operative team and consultant presence had no influence. Survival in patients with Dukes' A and B cancers was significantly reduced if <12 nodes were sampled. CONCLUSIONS: As a unit, we are achieving the national standard for lymph node harvest. This standard was maintained whether the surgeon performing the surgery was a consultant or a trainee, and also when the surgery was performed in the emergency setting. These data support the concept of 12 nodes being required for accurate staging.

Subintimal angioplasty: meta-analytical evidence of clinical utility.

OBJECTIVES: The aim of this study was to determine accurate estimates of the success rate of subintimal angioplasty in terms of ability to recanalise occluded vessels, patency over time and limb salvage rates. DESIGN: A meta-analysis of published literature. MATERIALS: All studies reporting unique patient data published in English language between 1989 and 2008. METHODS: Separate meta-analyses were performed for immediate technical success, 12-month patency rates and 12-month limb salvage rates. Longer-term outcomes were analyzed in separate meta-analyses. Meta-regression was applied to determine whether any of these outcomes had improved over time. RESULTS: Pooled estimates for technical success, primary patency at 12 months and limb salvage at 12 months were 85.7% (95% confidence interval: 83.3%-87.7%, 2810 limbs), 55.8% (95% confidence interval: 47.9%-63.4%, 1342 limbs), and 89.3% (95% confidence interval: 85.5%-92.2%, 2810 limbs), respectively. Regression analysis demonstrated no significant change in outcomes over time. There was some evidence of publication bias, however, after adjusting for this there was little change in the pooled outcome estimates. CONCLUSIONS: This study demonstrates that the outcomes for subintimal angioplasty are good and that this method should be considered as an alternative to surgical bypass.

The effect of education and home safety equipment on childhood thermal injury prevention: meta-analysis and meta-regression.

OBJECTIVE: To evaluate whether home safety education and safety equipment provision increases thermal injury prevention practices or reduces thermal injury rates and whether the effect of interventions differs by social group. METHODS: Systematic review and meta-analysis using individual participant data (IPD) evaluating home safety education with or without provision of free or discounted safety equipment provided to children or young people aged 0-19 years. MAIN OUTCOME MEASURES: possession of functional smoke alarm, fitted fireguard and fire extinguisher; keeping hot drinks or food and keeping matches or lighters out of reach; having a safe hot water temperature and rate of medically attended thermal injuries. RESULTS: Home safety interventions were effective in increasing the proportion of families with a functional smoke alarm (odds ratio (OR) 1.83, 95% CI 1.22 to 2.74) and with a safe hot tap water temperature (OR 1.35, 95% CI 1.01 to 1.80). There was some evidence they increased possession of fitted fireguards (OR 1.39, 95% CI 1.00 to 1.94), but there was a lack of evidence that interventions reduced medically attended thermal injury rates (incident rate ratio (IRR) 1.12, 95% CI 0.81 to 1.56). There was no consistent evidence that the effectiveness of interventions varied by social group. CONCLUSIONS: Home safety education, especially with the provision of safety equipment, is effective in increasing some thermal injury prevention practices, but there is insufficient evidence to show whether this also reduces injury rates.

Measuring risk of HIV and HCV among injecting drug users in the Russian Federation.

BACKGROUND: The aim of the study was to measure risk of HIV and HCV infection among injecting drug users (IDUs) through force of infection (FOI) models in three cities of the Russian Federation and assess the value of behavioural data and FOI in predicting risk of infection as a method of second-generation surveillance. METHODS: FOI models were fitted to prevalence data collected through an anonymous, cross-sectional community-recruited survey of IDUs with oral fluid sample collection for antibodies to HIV and HCV. Risk of infection was estimated from FOI estimates obtained by fitting a model to prevalence data by length of injecting career for each city and then overall. Risk behaviours were examined by injecting career length. RESULTS: A total of 1473 IDUs were recruited. Prevalence of HIV was 8.1% (95% CI 6.7-9.6%) and HCV 63.4% (95% CI 60.9-65.9%). A higher FOI in new initiates to injecting (injecting career length <1 year) was found for both HIV and HCV compared with experienced IDUs (injecting career length <5 years). Increased risk of infection was not corroborated by injecting risk behaviours among new initiates into injecting (n = 38). Only 5.7% (n = 2) reported receptive sharing in the last 4 weeks, 57.9% (n = 22) sharing any injecting paraphernalia, 2.6% (n = 1) frontloading and 8.5% (n = 3) ever injecting with used needles/syringes. However, 29% of new initiates reported exchanging sex in the last 4 weeks (29%) compared with 11% long term IDUs. CONCLUSIONS: FOI models can play an important role in surveillance of HIV but caution is needed in the interpretation of behavioural data for predicting current or future risk of HIV.

Systematic review and meta-analysis of mortality in crop protection product manufacturing workers.

OBJECTIVES: The potential health effects of the manufacture and use of crop protection chemicals were investigated through systematic review and meta-analysis of studies of cohorts of workers in the crop protection product manufacturing industry. METHODS: Several computerised literature databases were searched from inception until December 2003, with references listed in identified articles checked for further relevant articles. Random effects meta-analyses of log standardised mortality ratios (SMRs) were carried out. Heterogeneity was explored through subgroup analyses and meta-regression; sensitivity analyses of different approaches for zero events were performed. RESULTS: 21 references reporting information on 37 separate cohorts for mortality were identified. The meta-SMR for all cause mortality was 0.94 (95% CI 0.88 to 1.00) (37 cohorts). Significantly raised mortality was found for cancers of the buccal cavity and pharynx, oesophagus, rectum, larynx, lung, and lymphatic and haematopoietic system with little heterogeneity being observed. Excluding studies with zero events identified additional excesses. CONCLUSIONS: Evidence of multiple excesses, particularly in subgroups exposed to phenoxy herbicides contaminated with dioxins, substantiates previous findings. The importance of careful treatment of zero cases was highlighted. Future systematic reviews and meta-analyses would benefit from availability of results for a standard list of causes of disease.