Alan S. Rigby

Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins

OBJECTIVEnTwin concordance data for rheumatoid arthritis (RA) on their own provide only limited insight into the relative genetic and environmental contribution to the disease. We applied quantitative genetic methods to assess the heritability of RA and to examine for evidence of differences in the genetic contribution according to sex, age, and clinical disease characteristics.nnnMETHODSnData were analyzed from 2 previously published nationwide studies of twins with RA conducted in Finland and the United Kingdom. Heritability was assessed by variance components analysis. Differences in the genetic contribution by sex, age, age at disease onset, and clinical characteristics were examined by stratification. The power of the twin study design to detect these differences was examined through simulation.nnnRESULTSnThe heritability of RA was 65% (95% confidence interval [95% CI] 50-77) in the Finnish data and 53% (95% CI 40-65) in the UK data. There was no significant difference in the strength of the genetic contribution according to sex, age, age at onset, or disease severity subgroup. Both study designs had power to detect a contribution of at least 40% from the common family environment, and a difference in the genetic contribution of at least 50% between subgroups.nnnCONCLUSIONnGenetic factors have a substantial contribution to RA in the population, accounting for approximately 60% of the variation in liability to disease. Although tempered by power considerations, there is no evidence in these twin data that the overall genetic contribution to RA differs by sex, age, age at disease onset, and disease severity.

Statistical methods in epidemiology. v. Towards an understanding of the kappa coefficient

Purpose : This paper introduces readers to the problem of measuring interrater agreement in observer variation studies. The most usual statistic to quote is the kappa coefficient which measures agreement having corrected for chance. Method : The kappa coefficient for measuring agreement between two observers is introduced. Some pointers are given to determining sample size estimation. Results : Some properties of the kappa coefficient are illustrated by taking examples from the authors teaching experiences. Conclusion : The kappa coefficient is recommended for measuring agreement in observer variation studies.

Acute correction of lower limb deformity and simultaneous lengthening with a monolateral fixator

We have reviewed, retrospectively, all children with a lower limb deformity who underwent an acute correction and lengthening with a monolateral fixator between 1987 and 1996. The patients were all under the age of 19 years and had a minimum follow-up of eight months after removal of the fixator. A total of 41 children had 57 corrections and lengthening. Their mean age was 11.3 years (3.2 to 18.7) and there were 23 girls and 18 boys. The mean maximum correction in any one plane was 23 degrees (7 to 45). In 41 bony segments (either femur or tibia) a uniplanar correction was made while various combinations were carried out in 16. The site of the osteotomy was predominantly diaphyseal, at a mean of 47% (17% to 73%) of the total bone length and the mean length gained was 6.4 cm (1.0 to 17.0). Univariate analysis identified a moderately strong relationship between the bone healing index (BHI), length gained, maximum correction and grade-II to grade-III complications. For logistic regression analysis the patients were binary coded into two groups; those with a good outcome (BHI < or = 45 days/cm) and those with a poor outcome (BHI > 45 days/cm). Various factors which may influence the outcome were then analysed by calculating odds ratios with 95% confidence intervals. This analysis suggested a dose response between increasing angular correction and poor BHI which only reached statistical significance for corrections of larger magnitude. Longer lengthenings were associated with a better BHI while age and the actual bone lengthened had little effect. Those patients with a maximum angulatory correction of less than 30 degrees in any one plane had an acceptable consolidation time with few major complications. The technique is suitable for femoral deformity and shortening, but should be used with care in the tibia since the risk of a compartment syndrome or neurapraxia is much greater.

Disability in young adults following major trauma: 5 year follow up of survivors

BackgroundInjuries are a major cause of mortality and morbidity in young people. Despite this, the long-term consequences for young survivors of severe injury are relatively unexplored.MethodsPopulation based cohort study involving 5 year post injury structured interview of all cases of major trauma (Injury Severity Score > 15) identified retrospectively for 12 month period (1988 to 1989) within former Yorkshire Health Authority area of the United Kingdom.Results125 individuals aged 11–24 years at time of injury were identified. Of these, 109 (87%) were interviewed. Only 20% (95% CI 14–29%) of those interviewed reported no disability. Mean Office of Population Census and Surveys (OPCS) disability score of the remainder was 7.5 (median 5.8, range 0.5 to 19.4). The most commonly encountered areas of disability were behaviour (54%, 95% CI 45–63%), intellectual functioning (39%, 95% CI 31–49%) and locomotion (29%, 95% CI 22–39%). Many respondents reported that their daily lives were adversely affected by their health problems for example, causing problems with work, 54% (95% CI 45–63%), or looking after the home, 28% (95% CI 21–38%). Higher OPCS scores were usually but not always associated with greater impact on daily activities. The burden of caring responsibilities fell largely on informal carers. 51% (95% CI 42–61%) of those interviewed would have liked additional help to cope with their injury and disability.ConclusionThe study has revealed significant disability amongst a cohort of young people 5 years post severe injury. Whilst many of these young people were coping well with the consequences of their injuries, others reported continuing problems with the activities of daily life. The factors underpinning the young peoples differing experiences and social outcome should be explored.

Headache management—Are we doing enough? An observational study of patients presenting with headache to the emergency department

Objectives: To identify the causes of acute headache presenting to the emergency department (ED), assess the adequacy of history, examination, and investigation, and determine which clinical features are predictive of secondary headache. Method: A retrospective study of alert (GCS⩾14) patients presenting with headache, to an ED over a one year period. Patients were followed up for three months. The adequacy of history, examination, and investigation were compared with published standards. Analysis using Bayes’s theorem determined which clinical features were predictive of secondary headache. Results: Headache in alert patients accounted for 0.5% (nu200a=u200a353) of new patient episodes, 81.2% (nu200a=u200a280) of patients had a primary headache disorder. One patient (0.3%) had an adequate history recorded. No patient had a complete examination recorded. Seventy seven (21.8%) patients underwent computed tomography of the head; 80.5% (nu200a=u200a62) were normal. Lumbar puncture was performed in 23 (6.5%) cases; 18 (78.3%) were normal. A number of clinical features were found to be predictive of secondary headache. Conclusion: Headache is an uncommon symptom in alert patients presenting to the ED. The recorded history, examination, and subsequent investigation do not comply with published standards. A number of predictive features have been identified that may permit the development of a clinical prediction rule to improve the management of this patient group.

Statistical methods in epidemiology. III. The odds ratio as an approximation to the relative risk

PURPOSEnThis paper introduces readers to case-control studies and how to analyse them. Specifically, the odds ratio statistic is discussed. The importance of the odds ratio in epidemiology is that it is used as an approximation to the true relative risk.nnnMETHODnData are presented in the form of 2 x 2 contingency tables, and a method for calculating the odds ratio is presented. An extension to 2 x k tables (where k > 2) is given, as is a graphical method for plotting odds ratios. Confidence intervals based on the Normal approximation are introduced.nnnRESULTSnSome properties of the odds ratio statistic are illustrated by taking examples from the authors own teaching experiences.nnnCONCLUSIONnAs long as the odds ratio is not used uncritically as an estimate of the relative risk, it remains an attractive statistic for epidemiologists to calculate.

Radiographic assessment of bone formation in tibia during distraction osteogenesis.

The radiologic appearance of bone formation during limb lengthening is used to judge the competence of the underlying biologic process and predict the likely time to healing. Interpretation is, however, based upon subjective parameters that have never been clearly defined. Thirty anteroposterior radiographs from pediatric tibial lengthenings were classified by four pairs of surgeons using a three-part system. Across the group, interobserver consistency showed considerable variation for all parameters tested. Pairwise analysis indicated that the surgeons directly involved in limb lengthening procedures agreed better than those whose practice was of a different nature. A second series of radiographs contained 12 radiographs with a satisfactory and 12 with a poor bone healing index (BHI). These radiographs were digitized, and an interobserver study showed significantly improved observer concordance if the images were enhanced. No feature was associated with a 100% chance of a satisfactory outcome, but certain appearances in the regenerate were associated with a better BHI than others.

Statistical methods in epidemiology. II: A commonsense approach to sample size estimation.

PURPOSEnIt has been argued, by many, that mathematical formulae for estimating sample size are unnecessarily complex, so much so that researchers may be reluctant to seek statistical advice.nnnMETHODnThis paper reviews methods of sample size estimation arguing that two formulae (one based on comparison of proportion of successes, the other based on comparison of means of normally distributed data) suffice for many situations. This paper argues the case by taking examples drawn mainly from clinical trials research. However, the methods outlined can also be used in epidemiology specifically in both case-control and cohort studies with no loss of information.nnnRESULTSnFor the situations outlined, worked examples are provided.nnnCONCLUSIONSnSample size estimation need not necessarily be a complex process. Simple techniques exist which enable the clinician and the statistician to work together. Continued dialogue between both parties is required so that good ideas do not go to waste.

HLA haplotype sharing in rheumatoid arthritis sibships: Risk estimates subdivided by proband genotype

There is a well‐known association between rheumatoid arthritis (RA) and HLA‐DR4. Recent research has indicated that both DR4 haplotypes are important in disease predisposition (favoring a recessive mode of inheritance). Others have suggested that certain combinations of genotypes, in particular Dw4/Dw14 heterozygotes, may be more important than others. We examined the mode of inheritance of RA using data from the Arthritis and Rheumatism Councils national repository of family material [Worthington et al. (1994) Br J Rheumatol 33:970–976]. There were 85 affected sibships consisting of 77 sib pairs, 6 trios, 1 quintuplet, and 1 sextuplet. The affected sibs shared two, one, and zero parental HLA haplotypes in a ratio of 0.42:0.43:0.15, which was significantly different from random expectations (P = 0.00009). Risk estimates for RA to sibs were calculated based on an overall sibling recurrence risk of 3.9%. Risks for those sharing two, one, and zero parental HLA haplotypes were 6.5% [95% confidence interval (CI) = 5.1–7.9%], 3.3% (95% CI = 2.6–4.0%), and 2.5% (95% CI = 1.5–3.5%), respectively. We also examined the risk of RA based on the DRβ1 genotype status of sib and proband. After excluding genotypic combinations with small numbers, the highest genotype‐specific risks were seen for sibs sharing two haplotypes with either a DRβ1*0401/DRβ1*0404 (12.5%, 95% CI = 6.9–15.2%) or a DRβ1*0401/DRβ1*0408 (11.1%, 95% CI = 4.5–15.1%) proband. An independent assessment based on the AGFAP methodology confirmed the increase in risk for these genotypes, in particular for DRβ1*0401/DRβ1*0408. The excess being due to *0401/*0408 rather than to *0401/*0404 may explain why the Dw4/Dw14 effect is not always observed. Genet. Epidemiol. 15:403–418,1998.

Modelling 5-year functional outcome in a major traumatic injury survivor cohort.

Aim. To estimate associations between possible predictors of functional disability outcome at 5 years in a working adult population cohort of survivors of major traumatic injury and to develop a prognostic model of outcome. Design. Population based retrospective cohort study. Population. Persons who had experienced major traumatic injury (Injury Seventy Score >15) in the area of the former Yorkshire Regional Health Authority during the period 30 September 1988 to 1 October 1989 and who survived for 5 years (average 5.3 years). Methods. The same interviewer saw each survivor at their home and used the OPCS Adult Disability Schedule to ascertain levels of functional disability. Disability scores for each survivor were then combined in accordance with the OPCS guidance to calculate a level of disability between 0 (no disability) to 10 (maximum disability). The OPCS level was then dichotomised with a cut-point at 4/5. Possible predictor and confounding variables from pre-injury, injury and post-injury periods were modelled in a logistic regression to identify those that predicted outcome level. Two reduced models were developed to allow early prognosis of late outcome. Results. The full model correctly classified 91% of observed outcomes. Inpatient length of stay OR 1.031 (95% CI 1.014,1.048) per day predicted poorer 5-year outcome; Glasgow Coma Score OR 0.790(0.629,0.992) per 1 point increase; stay in Intensive Care Unit OR 0.931 (0.877,0.987) per day; attainment of degree-level education OR 0.014 (0.000,0.707); single civil status OR 0.110 (0.013,0.908); being taken initially to a hospital with a neurosurgical facility OR 0.064(0.010,0.420); being in paid work during the 2 weeks before index injury OR 0.093(0.009,0.969) predicted better 5-year disability outcome. Two reduced models were constructed that included a simple set of variables, one of these models excluded any rehabilitation variables but still correctly classified 85% of the observed outcomes. Conclusion. As well as level of traumatic brain injury (TBI) and total inpatient stay, Pre-injury educational attainment and employment, civil status, immediate care in a hospital with a neurosurgical facility and stay in an Intensive Care Unit determined 5-year outcome. It is possible to efficiently predict outcome at an early stage. Previous work on predictors of disablement have suffered from large selection and attrition biases.