Cancer risk in information technology workers: a UK Biobank study.

BACKGROUND: The information technology (IT) workforce has been growing more rapidly than others, with occupational health (OH) risks of sedentary behaviour, physical inactivity and poor diet, yet studies of their non-communicable disease risk, notably cancer, are lacking. AIMS: To investigate cancer risk in IT workers compared to others in employment and the nine major Standard Occupational Classification (SOC) groups. METHODS: We evaluated incident diagnosed cancers in the UK Biobank cohort through national cancer registry linkage. Cox proportional hazard regression models, with 15-year follow-up, were used to compare incident cancer risk among IT workers with all other employed participants and with the nine major SOC groups. RESULTS: Overall, 10 517 (4%) employed participants were IT workers. Adjusting for confounders, IT workers had a slightly lower cancer incidence compared to all other employed participants (Model 2: hazard ratio = 0.91, 95% confidence interval [CI] 0.83-1.01). Compared to the nine major SOC groups, they had a similar (Major Groups 2, 5 and 8) or lower (Major Groups 1, 3, 4, 6, 7 and 9) cancer incidence. CONCLUSIONS: Despite their occupational risks of sedentary behaviour, poor diet and physical inactivity, IT workers do not have an increased cancer incidence compared to all other employed participants and the nine major SOC groups. This study paves the way for large, longitudinal health outcome studies of this under-researched and rapidly growing occupational group.

Health, lifestyle and occupational risks in Information Technology workers.

BACKGROUND: Information technology (IT) and the IT workforce are rapidly expanding with potential occupational health implications. But to date, IT worker health is under-studied and large-scale studies are lacking. AIMS: To investigate health, lifestyle and occupational risk factors of IT workers. METHODS: We evaluated self-reported health, lifestyle and occupational risk factors for IT workers in the UK Biobank database. Using logistic regression, we investigated differences between IT workers and all other employed participants. Regression models were repeated for IT worker subgroups (managers, professionals, technicians) and their respective counterparts within the same Standard Occupational Classification (SOC) major group (functional managers, science and technology professionals, science and technology associate professionals). RESULTS: Overall, 10 931 (4%) employed participants were IT workers. Compared to all other employed participants, IT workers reported similar overall health, but lower lifestyle risk factors for smoking and obesity. Sedentary work was a substantially higher occupational exposure risk for IT workers compared to all other employed participants (odds ratio [OR] = 5.14, 95% confidence interval [CI]: 4.91-5.39) and their specific SOC group counterparts (managers: OR = 1.83, 95% CI: 1.68-1.99, professionals: OR = 7.18, 95% CI: 6.58-7.82, technicians: OR = 4.48, 95% CI: 3.87-5.17). IT workers were also more likely to engage in computer screen-time outside work than all other employed participants (OR = 1.42, 95% CI: 1.35-1.51). CONCLUSIONS: Improved understanding of health, lifestyle and occupational risk factors from this, the largest to date study of IT worker health, can help inform workplace interventions to mitigate risk, improve health and increase the work participation of this increasingly important and rapidly growing occupational group.

Towards transforming community eye care: an observational study and time-series analysis of optometrists' prescribing for eye disorders.

OBJECTIVES: This study aimed to provide evidence on the therapeutic prescribing activity by community optometrists in Scotland and to determine its impact on workload in general practice and ophthalmology clinics. STUDY DESIGN: Scottish administrative healthcare data for a 53-month period (November 2013-April 2018) were used to analyse non-medical prescribing practice by optometrists. METHODS: Using interrupted time-series regression (Autoregressive Integrated Moving Average), we assessed the impact of optometrist prescribing on ophthalmology outpatient attendances and general practice prescribing for eye disorders. RESULTS: A total of 54,246 items were prescribed by 205 optometrists over the study period. Since the commencement of data recording, optometrist prescribing activity increased steadily from a baseline of zero to 1.2% of all ophthalmic items prescribed. Neither the monthly number of items prescribed nor the size of optometric workforce were associated with a reduction in ophthalmology outpatient appointments over time. CONCLUSIONS: Optometrists increasingly contribute to community ophthalmic prescribing in Scotland, releasing capacity and lessening general practice, but not secondary care workload. There appears to be an underutilisation of optometrists related to the management of dry eye, which represents an opportunity to release further capacity.

Complication rate among people with diabetes at low risk of foot ulceration in Fife, UK: an analysis of routinely collected data.

AIMS: To estimate the rate at which people with diabetes and a low risk of foot ulceration change diabetic foot ulceration risk status over time, and to estimate the rate of ulceration, amputation and death among this population. METHODS: We conducted an observational study of 10 421 people with diabetes attending foot screening in an outpatient setting in NHS Fife, UK, using routinely collected data from a national diabetes register, NHS SCI Diabetes. We estimated the proportion of people who changed risk status and the cumulative incidence of ulceration, amputation and death, respectively, among people with diabetes at low risk of diabetic foot ulceration at 2-year follow-up. RESULTS: At 2-year follow-up, 5.1% (95% CI 4.7, 5.6) of people with diabetes classified as low risk at their first visit had progressed to moderate risk. The cumulative incidence of ulceration, amputation and death was 0.4% (95% CI 0.3, 0.6), 0.1% (95% CI 0.1, 0.2) and 3.4% (95% CI 3.1, 3.8), respectively. CONCLUSIONS: At 2-year follow-up, 5% of people at low risk of diabetic foot ulceration changed clinical risk status and <1% of people experienced foot ulceration or amputation. These findings provide information which will help to inform the current debate regarding optimal foot screening intervals.

Explaining trends in alcohol-related harms in Scotland, 1991-2011 (I): the role of incomes, effects of socio-economic and political adversity and demographic change.

OBJECTIVE: This paper tests the extent to which differing trends in income, demographic change and the consequences of an earlier period of social, economic and political change might explain differences in the magnitude and trends in alcohol-related mortality between 1991 and 2011 in Scotland compared to England & Wales (E&W). STUDY DESIGN: Comparative time trend analyses and arithmetic modelling. METHODS: Three approaches were utilised to compare Scotland with E&W: 1. We modelled the impact of changes in income on alcohol-related deaths between 1991-2001 and 2001-2011 by applying plausible assumptions of the effect size through an arithmetic model. 2. We used contour plots, graphical exploration of age-period-cohort interactions and calculation of Intrinsic Estimator coefficients to investigate the effect of earlier exposure to social, economic and political adversity on alcohol-related mortality. 3. We recalculated the trends in alcohol-related deaths using the white population only to make a crude approximation of the maximal impact of changes in ethnic diversity. RESULTS: Real incomes increased during the 1990s but declined from around 2004 in the poorest 30% of the population of Great Britain. The decline in incomes for the poorest decile, the proportion of the population in the most deprived decile, and the inequality in alcohol-related deaths, were all greater in Scotland than in E&W. The model predicted less of the observed rise in Scotland (18% of the rise in men and 29% of the rise in women) than that in E&W (where 60% and 68% of the rise in men and women respectively was explained). One-third of the decline observed in alcohol-related mortality in Scottish men between 2001 and 2011 was predicted by the model, and the model was broadly consistent with the observed trends in E&W and amongst women in Scotland. An age-period interaction in alcohol-related mortality was evident for men and women during the 1990s and 2000s who were aged 40-70 years and who experienced rapidly increasing alcohol-related mortality rates. Ethnicity is unlikely to be important in explaining the trends or differences between Scotland and E&W. CONCLUSIONS: The decline in alcohol-related mortality in Scotland since the early 2000s and the differing trend to E&W were partly described by a model predicting the impact of declining incomes. Lagged effects from historical social, economic and political change remain plausible from the available data.

Explaining trends in alcohol-related harms in Scotland 1991-2011 (II): policy, social norms, the alcohol market, clinical changes and a synthesis.

OBJECTIVE: To provide a basis for evaluating post-2007 alcohol policy in Scotland, this paper tests the extent to which pre-2007 policy, the alcohol market, culture or clinical changes might explain differences in the magnitude and trends in alcohol-related mortality outcomes in Scotland compared to England & Wales (E&W). STUDY DESIGN: Rapid literature reviews, descriptive analysis of routine data and narrative synthesis. METHODS: We assessed the impact of pre-2007 Scottish policy and policy in the comparison areas in relation to the literature on effective alcohol policy. Rapid literature reviews were conducted to assess cultural changes and the potential role of substitution effects between alcohol and illicit drugs. The availability of alcohol was assessed by examining the trends in the number of alcohol outlets over time. The impact of clinical changes was assessed in consultation with key informants. The impact of all the identified factors were then summarised and synthesised narratively. RESULTS: The companion paper showed that part of the rise and fall in alcohol-related mortality in Scotland, and part of the differing trend to E&W, were predicted by a model linking income trends and alcohol-related mortality. Lagged effects from historical deindustrialisation and socio-economic changes exposures also remain plausible from the available data. This paper shows that policy differences or changes prior to 2007 are unlikely to have been important in explaining the trends. There is some evidence that aspects of alcohol culture in Scotland may be different (more concentrated and home drinking) but it seems unlikely that this has been an important driver of the trends or the differences with E&W other than through interaction with changing incomes and lagged socio-economic effects. Substitution effects with illicit drugs and clinical changes are unlikely to have substantially changed alcohol-related harms: however, the increase in alcohol availability across the UK is likely to partly explain the rise in alcohol-related mortality during the 1990s. CONCLUSIONS: Future policy should ensure that alcohol affordability and availability, as well as socio-economic inequality, are reduced, in order to maintain downward trends in alcohol-related mortality in Scotland.

Mortality differences and inequalities within and between 'protected characteristics' groups, in a Scottish Cohort 1991-2009.

BACKGROUND: Little is known about the interaction between socio-economic status and 'protected characteristics' in Scotland. This study aimed to examine whether differences in mortality were moderated by interactions with social class or deprivation. The practical value was to pinpoint population groups for priority action on health inequality reduction and health improvement rather than a sole focus on the most deprived socioeconomic groups. METHODS: We used data from the Scottish Longitudinal Study which captures a 5.3 % sample of Scotland and links the censuses of 1991, 2001 and 2011. Hazard ratios for mortality were estimated for those protected characteristics with sufficient deaths using Cox proportional hazards models and through the calculation of European age-standardised mortality rates. Inequality was measured by calculating the Relative Index of Inequality (RII). RESULTS: The Asian population had a polarised distribution across deprivation deciles and was more likely to be in social class I and II. Those reporting disablement were more likely to live in deprived areas, as were those raised Roman Catholic, whilst those raised as Church of Scotland or as 'other Christian' were less likely to. Those aged 35-54 years were the least likely to live in deprived areas and were most likely to be in social class I and II. Males had higher mortality than females, and disabled people had higher mortality than non-disabled people, across all deprivation deciles and social classes. Asian males and females had generally lower mortality hazards than majority ethnic ('White') males and females although the estimates for Asian males and females were imprecise in some social classes and deprivation deciles. Males and females who reported their raised religion as Roman Catholic or reported 'No religion' had generally higher mortality than other groups, although the estimates for 'Other religion' and 'Other Christian' were less precise.Using both the area deprivation and social class distributions for the whole population, relative mortality inequalities were usually greater amongst those who did not report being disabled, Asians and females aged 35-44 years, males by age, and people aged <75 years. The RIIs for the raised religious groups were generally similar or too imprecise to comment on differences. CONCLUSIONS: Mortality in Scotland is higher in the majority population, disabled people, males, those reporting being raised as Roman Catholics or with 'no religion' and lower in Asians, females and other religious groups. Relative inequalities in mortality were lower in disabled than nondisabled people, the majority population, females, and greatest in young adults. From the perspective of intersectionality theory, our results clearly demonstrate the importance of representing multiple identities in research on health inequalities.

Early parental physical punishment and emotional and behavioural outcomes in preschool children.

OBJECTIVE: To determine whether there is an association between being smacked by your main caregiver in the first two years and emotional and behavioural problems at age four. DESIGN: Secondary analysis of data from the Growing Up in Scotland Prospective Study (GUS). SETTING: Scotland, UK. PARTICIPANTS: GUS birth cohort children, whose main caregiver had no concerns about their behaviour at 22 months. EXPOSURE: Ever smacked by main caregiver in first 22 months, as measured by caregiver self-report at 22 months. MAIN OUTCOME: Emotional and behavioural problems as measured by parental assessment and the Strengths and Difficulties Questionnaire (SDQ) at 46 months. RESULTS: Preschool children exposed to main caregiver smacking in the first two years were twice as likely to have emotional and behavioural problems as measured by parental assessment [odds ratio (OR) 2.5, 95% confidence interval (CI) 1.9-3.2; absolute risk reduction (ARR) 17.8%, 95% CI 12.1-23.5] and SDQ (OR 2.5, 95% CI 1.7-3.7; ARR 7.5%, 95% CI 3.7-11.5), as children never smacked by their main caregiver. The association remained significant after adjusting for child age and sex, caregiver age, sex, ethnicity, educational attainment and mental health status, sibling number, structural family transitions and socioeconomic status (adj. OR 2.4, 95% CI 1.8-3.2 for parental assessment and adj. OR 2.2, 95% CI 1.4-3.5 for SDQ). CONCLUSIONS: Parental use of physical punishment in the first two years may be a modifiable risk factor for emotional and behavioural difficulties in preschool children.

Sex-specific time trends in first admission to hospital for peripheral artery disease in Scotland 1991-2007.

BACKGROUND: This study examined trends for all first hospital admissions for peripheral artery disease (PAD) in Scotland from 1991 to 2007 using the Scottish Morbidity Record. METHODS: First admissions to hospital for PAD were defined as an admission to hospital (inpatient and day-case) with a principal diagnosis of PAD, with no previous admission to hospital (principal or secondary diagnosis) for PAD in the previous 10 years. RESULTS: From 1991 to 2007, 41,593 individuals were admitted to hospital in Scotland for the first time for PAD. Some 23,016 (55.3 per cent) were men (mean(s.d.) age 65.7(11.7) years) and 18,577 were women (aged 70.4(12.8) years). For both sexes the population rate of first admissions to hospital for PAD declined over the study interval: from 66.7 per 100,000 in 1991-1993 to 39.7 per 100,000 in 2006-2007 among men, and from 43.5 to 29.1 per 100,000 respectively among women. After adjustment, the decline was estimated to be 42 per cent in men and 27 per cent in women (rate ratio for 2007 versus 1991: 0.58 (95 per cent confidence interval 0.55 to 0.62) in men and 0.73 (0.68 to 0.78) in women). The intervention rate fell from 80.8 to 74.4 per cent in men and from 77.9 to 64.9 per cent in women. The proportion of hospital admissions as an emergency or transfer increased, from 23.9 to 40.7 per cent among men and from 30.0 to 49.5 per cent among women. CONCLUSION: First hospital admission for PAD in Scotland declined steadily and substantially between 1991 and 2007, with an increase in the proportion that was unplanned.

Urinary Sodium Excretion, Blood Pressure, and Risk of Future Cardiovascular Disease and Mortality in Subjects Without Prior Cardiovascular Disease.

Hypertension is a risk factor for cardiovascular disease. Increased urinary sodium excretion, representing dietary sodium intake, is associated with hypertension. Low sodium intake has been associated with increased mortality in observational studies. Further studies should assess whether confounding relationships explain associations between sodium intake and outcomes. We studied UK Biobank participants (n=457 484; mean age, 56.3 years; 44.7% men) with urinary electrolytes and blood pressure data. Estimated daily urinary sodium excretion was calculated using Kawasaki formulae. We analyzed associations between sodium excretion and blood pressure in subjects without cardiovascular disease, treated hypertension, or diabetes mellitus at baseline (n=322 624). We tested relationships between sodium excretion, incidence of fatal and nonfatal cardiovascular disease, heart failure, and mortality. Subjects in higher quintiles of sodium excretion were younger, with more men and higher body mass index. There was a linear relationship between increasing urinary sodium excretion and blood pressure. During median follow-up of 6.99 years, there were 11 932 deaths (1125 cardiovascular deaths) with 10 717 nonfatal cardiovascular events. There was no relationship between quintile of sodium excretion and outcomes. These relationships were unchanged after adjustment for comorbidity or excluding subjects with events during the first 2 years follow-up. No differing risk of incident heart failure (1174 events) existed across sodium excretion quintiles. Urinary sodium excretion correlates with elevated blood pressure in subjects at low cardiovascular risk. No pattern of increased cardiovascular disease, heart failure, or mortality risk was demonstrated with either high or low sodium intake.

Survival after liver transplantation in the United Kingdom and Ireland compared with the United States.

BACKGROUND AND OBJECTIVE: Surgical mortality in the US is widely perceived to be superior to that in the UK. However, previous comparisons of surgical outcome in the two countries have often failed to take sufficient account of case-mix or examine long-term outcome. The standardised nature of liver transplantation practice makes it uniquely placed for undertaking reliable international comparisons of surgical outcome. The objective of this study is to undertake a risk-adjusted disease-specific comparison of both short- and long-term survival of liver transplant recipients in the UK and Ireland with that in the US. METHODS: A multicentre cohort study using two high quality national databases including all adults who underwent a first single organ liver transplant in the UK and Ireland (n = 5925) and the US (n = 41,866) between March 1994 and March 2005. The main outcome measures were post-transplant mortality during the first 90 days, 90 days to 1 year and beyond the first year, adjusted for recipient and donor characteristics. RESULTS: Risk-adjusted mortality in the UK and Ireland was generally higher than in the US during the first 90 days (HR 1.17; 95% CI 1.07 to 1.29), both for patients transplanted for acute liver failure (HR 1.27; 95% CI 1.01 to 1.60) and those transplanted for chronic liver disease (HR 1.18; 95% CI 1.07 to 1.31). Between 90 days and 1 year post-transplantation, no statistically significant differences in overall risk-adjusted mortality were noted between the two cohorts. Survivors of the first post-transplant year in the UK and Ireland had lower overall risk-adjusted mortality than those transplanted in the US (HR 0.88; 95% CI 0.81 to 0.96). This difference was observed among patients transplanted for chronic liver disease (HR 0.88; 95% CI 0.81 to 0.96), but not those transplanted for acute liver failure (HR 1.02; 95% CI 0.70 to 1.50). CONCLUSIONS: Whilst risk-adjusted mortality is higher in the UK and Ireland during the first 90 days following liver transplantation, it is higher in the US among those liver transplant recipients who survived the first post-transplant year. Our results are consistent with the notion that the US has superior acute perioperative care whereas the UK appears to provide better quality chronic care following liver transplantation surgery.

The role of pain and function in determining patient satisfaction after total knee replacement. Data from the National Joint Registry for England and Wales.

A postal questionnaire was sent to 10,000 patients more than one year after their total knee replacement (TKR). They were assessed using the Oxford knee score and were asked whether they were satisfied, unsure or unsatisfied with their TKR. The response rate was 87.4% (8231 of 9417 eligible questionnaires) and a total of 81.8% (6625 of 8095) of patients were satisfied. Multivariable regression modelling showed that patients with higher scores relating to the pain and function elements of the Oxford knee score had a lower level of satisfaction (p < 0.001), and that ongoing pain was a stronger predictor of this. Female gender and a primary diagnosis of osteoarthritis were found to be predictors of lower levels of patient satisfaction. Differences in the rate of satisfaction were also observed in relation to age, the American Society of Anesthesiologists grade and the type of prosthesis. This study has provided data on the Oxford knee score and the expected levels of satisfaction at one year after TKR. The results should act as a benchmark of practice in the United Kingdom and provide a baseline for peer comparison between institutions.

Continuous monitoring of the performance of hip prostheses.

New brands of joint prosthesis are released for general implantation with limited evidence of their long-term performance in patients. The CUSUM continuous monitoring method is a statistical testing procedure which could be used to provide prospective evaluation of brands as soon as implantation in patients begins and give early warning of poor performance. We describe the CUSUM and illustrate the potential value of this monitoring tool by applying it retrospectively to the 3M Capital Hip experience. The results show that if the clinical data and methodology had been available, the CUSUM would have given an alert to the underperformance of this prosthesis almost four years before the issue of a Hazard Notice by the Medical Devices Agency. This indicates that the CUSUM can be a valuable tool in monitoring joint prostheses, subject to timely and complete collection of data. Regional or national joint registries provide an opportunity for future centralised, continuous monitoring of all hip and knee prostheses using these techniques.

A cardiovascular disease policy model: part 2-preparing for economic evaluation and to assess health inequalities.

OBJECTIVES: This is the second of the two papers introducing a cardiovascular disease (CVD) policy model. The first paper described the structure and statistical underpinning of the state-transition model, demonstrating how life expectancy estimates are generated for individuals defined by ASSIGN risk factors. This second paper describes how the model is prepared to undertake economic evaluation. DESIGN: To generate quality-adjusted life expectancy (QALE), the Scottish Health Survey was used to estimate background morbidity (health utilities) and the impact of CVD events (utility decrements). The SF-6D algorithm generated utilities and decrements were modelled using ordinary least squares (OLS). To generate lifetime hospital costs, the Scottish Heart Health Extended Cohort (SHHEC) was linked to the Scottish morbidity and death records (SMR) to cost each continuous inpatient stay (CIS). OLS and restricted cubic splines estimated annual costs before and after each of the first four events. A Kaplan-Meier sample average (KMSA) estimator was then used to weight expected health-related quality of life and costs by the probability of survival. RESULTS: The policy model predicts the change in QALE and lifetime hospital costs as a result of an intervention(s) modifying risk factors. Cost-effectiveness analysis and a full uncertainty analysis can be undertaken, including probabilistic sensitivity analysis. Notably, the impacts according to socioeconomic deprivation status can be made. CONCLUSIONS: The policy model can conduct cost-effectiveness analysis and decision analysis to inform approaches to primary prevention, including individually targeted and population interventions, and to assess impacts on health inequalities.

A cardiovascular disease policy model that predicts life expectancy taking into account socioeconomic deprivation.

OBJECTIVES: A policy model is a model that can evaluate the effectiveness and cost-effectiveness of interventions and inform policy decisions. In this study, we introduce a cardiovascular disease (CVD) policy model which can be used to model remaining life expectancy including a measure of socioeconomic deprivation as an independent risk factor for CVD. DESIGN: A state transition model was developed using the Scottish Heart Health Extended Cohort (SHHEC) linked to Scottish morbidity and death records. Individuals start in a CVD-free state and can transit to three CVD event states plus a non-CVD death state. Individuals who have a non-fatal first event are then followed up until death. Taking a competing risk approach, the cause-specific hazards of a first event are modelled using parametric survival analysis. Survival following a first non-fatal event is also modelled parametrically. We assessed discrimination, validation and calibration of our model. RESULTS: Our model achieved a good level of discrimination in each component (c-statistics for men (women)-non-fatal coronary heart disease (CHD): 0.70 (0.74), non-fatal cerebrovascular disease (CBVD): 0.73 (0.76), fatal CVD: 0.77 (0.80), fatal non-CVD: 0.74 (0.72), survival after non-fatal CHD: 0.68 (0.67) and survival after non-fatal CBVD: 0.65 (0.66)). In general, our model predictions were comparable with observed event rates for a Scottish randomised statin trial population which has an overlapping follow-up period with SHHEC. After applying a calibration factor, our predictions of life expectancy closely match those published in recent national life tables. CONCLUSIONS: Our model can be used to estimate the impact of primary prevention interventions on life expectancy and can assess the impact of interventions on inequalities.

Using routine data to complement and enhance the results of randomised controlled trials.

BACKGROUND: Randomised controlled trials (RCTs) are widely accepted as the best way to assess the outcomes and safety of medical interventions, but are sometimes not ethical, not feasible, or limited in the generalisability of their results. In such circumstances, routinely available data could help in several ways. Routine data could be used, for example, to conduct 'pseudo-trials', to estimate likely outcomes and required sample size to help design and conduct trials, or to examine whether the expected outcomes observed in an RCT will be realised in the general population. OBJECTIVES: The project was undertaken to explore how routinely assembled hospital data might complement or supplement RCTs to evaluate medical interventions: in contexts where RCTs are not feasible for defining the context and design of an RCT for assessing whether the benefits indicated by RCTs are achieved in wider clinical practice. METHODS: The project was based on the system of linked Scottish morbidity records, which cover 100% of acute hospital care episodes and statutory death records from 1981 to 1995. Three case studies were undertaken as a way of investigating the utility of these records in different applications. First, an attempt was made to analyse the link between the timing of surgery for subarachnoid haemorrhage (SAH) and subsequent outcomes (a question not easily susceptible to RCT design). A subsample was derived by excluding patients for which a diagnosis of SAH may not have been established or that may not have been admitted to a neurosurgical unit, and the data were assessed to attempt to inform the design of a trial of early versus late surgery. Transurethral prostatectomy (TURP), the second case study, has become the surgery of choice for benign prostatic hyperplasia without systematic assessment of its effectiveness and safety, and an RCT would now be considered unethical. However, there is a need to investigate long-term effects and the influence of co-morbidities on outcomes. A retrospective comparison of mortality and re-operation following either open prostatectomy (OPEN) or TURP was, therefore, undertaken. Patients for whom it was not possible to establish the initial procedure were excluded. The third case study compared coronary artery bypass grafting (CABG) with percutaneous transluminal angioplasty (PTCA) for coronary revascularisation. RCTs have been conducted in limited patient subgroups with short follow-up periods. A meta-analysis of RCTs could be augmented by routine data, which are available for large populations. This would allow assessment of subgroup effects, and outcomes over a long period. A subgroup of patients was therefore constructed for whom relevant routine data were available and who reflected the entry criteria for major RCTs, thus enabling a comparison between the results expected from this subgroup and those of the general population. RESULTS AND CONCLUSIONS: The uses of routine data in these contexts had strengths and weaknesses. The SAH study suggested a means of assessing outcomes and survival rates following haemorrhage, which could have value in informing the design of more precise trials and in evaluating changes in outcome following the introduction of new treatments such as embolisation. However, the potential of the data was not realised because their scope and content were insufficient. For example, lack of data on the time of onset of symptoms and patients' conditions at hospital admission made it difficult to establish the link between timing of surgery and the outcome, and there was insufficient information on patients' conditions at discharge to enable a comparison of outcomes. The prostatectomy study was able to address questions not answered by RCT literature because the large number of cases it included allowed exploration of subgroup effects. (ABSTRACT TRUNCATED)

The utility of the zero-inflated Poisson and zero-inflated negative binomial models: a case study of cross-sectional and longitudinal DMF data examining the effect of socio-economic status.

OBJECTIVES: To examine the utility of the zero-inflated Poisson (ZIP) and zero-inflated negative binomial (ZINB) modelling approaches for modelling four sets of dental caries data from the same cohort study [with particular attention to the influence of childhood socioeconomic status (SES)]: cross-sectional data on the deciduous dentition at age 5 years; cross-sectional data on the permanent dentition at age 18 and 26 years; and longitudinal data on caries increment between ages 18 and 26 years. METHODS: Data on dental caries occurrence at ages 5, 18 and 26 years were obtained from the Dunedin Multidisciplinary Health and Development Study (DMHDS). ZIP and ZINB models were fitted to the cross-sectional (n = 745) and longitudinal (n = 809) data sets using Stata (Intercooled Stata 7.0). The dependent variables for the three cross-sectional analyses were the DMFS indices at age 5, 18, and 26 years, and net DFS increment (NETDFS) was the dependent variable for the longitudinal analysis. RESULTS: The empty ZIP model was a poor fit for all four data sets, whereas the empty ZINB model showed good fit; consequently both the cross-sectional and longitudinal analyses were conducted using ZINB modelling. Being in the high-SES group during childhood was associated with a greater probability of being caries-free by age 18 years, over and above that which would be expected from the negative binomial process. Low childhood SES also had the largest coefficient in the modelling of the negative binomial process, but at age 5 years, where the adjusted mean dmfs score in the low-SES group was 6.8 (compared with 4.7 and 2.9 in the medium- and high-SES groups, respectively). The substantial SES differences which existed at age 5 years (in the deciduous dentition) had reduced somewhat by age 18 years, and had widened again by age 26 years. In the longitudinal analysis, "baseline" caries experience (age 18-year DMFS) was a predictor both of being an extra zero and of caries severity. CONCLUSION: This investigation of the utility of the zero-inflated approach for modelling both cross-sectional and longitudinal caries data has shown that ZIP/ZINB models can provide new insight into disease patterns. It is anticipated that they will become increasingly useful in epidemiological studies that use the DMF index as the outcome measure.

An introduction to meta-analysis within the framework of multilevel modelling using the probability of success of root canal treatment as an illustration.

OBJECTIVE: To introduce the statistical methodology of meta-analysis within the framework of multilevel modelling (MLM) using an illustrative example. BASIC RESEARCH DESIGN: In meta-analysis it is important that the quantitative pooling of study results should be carried out in conjunction with careful consideration of the variation apparent between studies. If statistical heterogeneity is found to be significant, it is due, at least in part, to clinical heterogeneity. It is possible to account for clinical heterogeneity by including covariates that are thought to be responsible, using meta-regression. CLINICAL SETTING: A total of 38 studies of root canal treatment outcome were identified as being suitable for introducing the meta-analysis methodology. Two covariates were considered for modelling: a 'loose' or 'strict' (loose--incomplete radiographic healing; strict--complete radiographic healing) criterion for judging outcome of treatment and the year in which the study was performed. RESULTS: There was considerable statistical heterogeneity between the study results. The effect of employing loose criteria for judging success significantly increased the probability of success when compared to employing strict criteria. Furthermore, the variance between studies was significantly reduced when this covariate was included in the modelling process when compared to the variation estimated in the model which did not consider covariates. CONCLUSION: MLM is a good facilitator for meta-analysis and meta-regression.

Is modelling dental caries a 'normal' thing to do?

OBJECTIVE: To introduce and encourage the use of generalised linear models (GLMs) in analysing caries data that do not require the response to be treated necessarily as a sample from a normal distribution. BASIC RESEARCH DESIGN: At the present time, it is most likely that the sampling distribution of dmf/DMF in industrialised countries will not approximate normality. Generalised linear modelling can be conducted assuming many underlying distributions which, in fact, includes the normal distribution. In this paper three GLMs are employed (normal, Poisson, negative binomial) for modelling an example caries data set. In addition, a binomial model is used to model the dichotomous outcome of caries-free/caries-present. CLINICAL SETTING: The data comprised 871 Old Trafford, Manchester primary school children aged between 4 years 0 months and 5 years 11 months. RESULTS: The effect of one study covariate was prominent in a normal model applied to all available dmf data but not in two non-normal models which used dmf > 0 data only. Furthermore, the same covariate was significant at the 5% level in a binomial model indicating that it influenced whether or not caries was present and not the level of dmf. CONCLUSION: A suitable modelling approach for caries data is to employ a Poisson or a negative binomial model for the dmf/DMF response and a binomial model for the caries-free/caries-present outcome. This allows separate estimation of those factors which influence the magnitude of caries and those factors which influence whether caries is actually present or not.

The initial dental health status of British Army Gurkha recruits.

AIM: To describe the initial dental health status of British Army Gurkhas who were recruited in 1999 and to compare the present caries data with previous unpublished data from before 1970 and 1983. DESIGN AND SETTING: A clinical examination was conducted on the 228 Gurkhas, the entire UK intake for 1999, during their second week of military training. The focus of the examinations was on caries experience. A questionnaire was employed to collect demographic data as well as information on the recruits' reported dental behaviour and beliefs. RESULTS: 1999 recruits who reported a dental problem within the past year were significantly more likely to have visited a dentist before compared to those recruits who reported no dental problems. The frequency distributions of D3MFT for the 1983 and 1999 recruits were very similar. CONCLUSIONS: Gurkha men are at relatively low risk of dental caries and predicted treatment time suggests a relatively small use of resources would be needed to make this group dentally fit. These recruits are an extremely homogeneous group who remain discernible from the general Nepalese population.