Myfanwy Griffith

Computerised linking of medical records: methodological guidelines.

OBJECTIVES--To report on the development of computer assisted methods for linking medical records and record abstracts. DESIGN--The methods include file blocking, to put records in an order which makes searching efficient; matching, which is the process of comparing records to determine whether they do or do not relate to the same person; linkage, which is the process of assembling correctly matched records into a time sequenced composite record for the individual; and validation checks and corrections, in which any inconsistencies between different records for the same person are identified and corrected. SETTING--The dataset comprising the Oxford record linkage study which includes hospital inpatient records and vital records. RESULTS AND CONCLUSIONS--Probability matching, using an array of identifiers, achieves much higher levels of correct matching than is generally achievable by exact character by character comparisons. The increasing use of information technology to store data about health and health care means that there is increasing scope to link records for research and for patient care. Sophisticated methods to achieve this on a large scale are now available.

Mortality Rates for Stroke in England From 1979 to 2004. Trends, Diagnostic Precision, and Artifacts

Background and Purpose— Stroke mortality appears to be declining more rapidly in the UK than in many other Western countries. To understand this apparent decline better, we studied trends in mortality in the UK using more detailed data than are routinely available. Methods— Analysis of datasets that include both the underlying cause and all other mentioned causes of death (together, termed “all mentions”): the Oxford Record Linkage Study from 1979 to 2004 and English national data from 1996 to 2004. Results— Mortality rates based on underlying cause and based on all mentions showed similar downward trends. Mortality based on underlying cause alone misses about one quarter of all stroke-related deaths. Changes during the period in the national rules for selecting the underlying cause of death had a significant but fairly small effect on the trend. Overall, mortality fell by an average annual rate of 2.3% (95% confidence interval 2.1% to 2.5%) for stroke excluding subarachnoid hemorrhage; and by 2.1% (1.7% to 2.6%) per annum for subarachnoid hemorrhage. Coding of stroke as hemorrhagic, occlusive, or unspecified varied substantially across the study period. As a result, rates for hemorrhagic and occlusive stroke, affected by artifact, seemed to fall substantially in the first part of the study period and then leveled off. Conclusion— Studies of stroke mortality should include all mentions as well as the certified underlying cause, otherwise the burden of stroke will be underestimated. Studies of stroke mortality that include strokes specified as hemorrhagic or occlusive, without also considering stroke overall, are likely to be misleading. Stroke mortality in the Oxford region halved between 1979 and 2004.

Perinatal and early life risk factors for inflammatory bowel disease.

AIM To investigate associations between perinatal risk factors and subsequent inflammatory bowel disease (IBD) in children and young adults. METHODS Record linked abstracts of birth registrations, maternity, day case and inpatient admissions in a defined population of southern England. Investigation of 20 perinatal factors relating to the maternity or the birth: maternal age, Crohns disease (CD) or ulcerative colitis (UC) in the mother, maternal social class, marital status, smoking in pregnancy, ABO blood group and rhesus status, pre-eclampsia, parity, the infants presentation at birth, caesarean delivery, forceps delivery, sex, number of babies delivered, gestational age, birthweight, head circumference, breastfeeding and Apgar scores at one and five minutes. RESULTS Maternity records were present for 180 children who subsequently developed IBD. Univariate analysis showed increased risks of CD among children of mothers with CD (P = 0.011, based on two cases of CD in both mother and child) and children of mothers who smoked during pregnancy. Multivariate analysis confirmed increased risks of CD among children of mothers who smoked (odds ratio = 2.04, 95% CI = 1.06-3.92) and for older mothers aged 35+ years (4.81, 2.32-9.98). Multivariate analysis showed that there were no significant associations between CD and 17 other perinatal risk factors investigated. It also showed that, for UC, there were no significant associations with the perinatal factors studied. CONCLUSION This study shows an association between CD in mother and child; and elevated risks of CD in children of older mothers and of mothers who smoked.

Day case surgery: geographical variation, trends and readmission rates.

Data from the Oxford Record Linkage Study were analysed to determine the amount of work undertaken in day case surgery for 12 surgical conditions in five districts in the Oxford Region during the years 1976 to 1985. Record linkage was used to study readmission rates, comparing day surgery with inpatient care. The use of day surgery gradually increased in some conditions (eg, termination of pregnancy, female sterilisation) but did not increase from a fairly low base for others (eg, inguinal hernia repair, operations on varicose veins and haemorrhoids). There were striking differences between the districts in the use of day case care. For example, the use of day case care as a percentage of all hospital admissions for termination of pregnancy varied from 1% in one district to 24% in another; that for dilatation and curettage varied from 1% to 43%; and that for female sterilisation varied from less than 1% to 35%. Emergency readmission rates after day surgery were similar to those following inpatient treatment. We conclude that the use of day surgery for some conditions judged suitable for day care is still low and, even within one region, variation in the use of day surgery is considerable. The reasons for continued reluctance in some places to undertake more day surgery merit investigation.

Trends in death certification for multiple sclerosis, motor neuron disease, Parkinson’s disease and epilepsy in English populations 1979–2006

The objective of this study is to report trends in mortality, as certified on death certificates, from multiple sclerosis (MS), motor neuron disease (MND), primary Parkinson’s disease (PD), and epilepsy, analysing not only the underlying cause of death but also all certified causes for each disease. Death records in the Oxford region, 1979–2006, and England, 1996–2006, were analysed for ascertaining the trends in mortality. The percentage of deaths coded as the underlying cause changed over time, coinciding with changes to the rules for selecting the underlying cause of death. Changes over time to coding rules had a large impact on apparent trends in death rates for PD when studied by underlying cause alone. They also had significant, though smaller, effects on trends in death rates for MS, MND and epilepsy. Nationally, in the last period of the study, underlying cause mortality identified 64% of deaths with a mention of MS, 88% of MND, 56% of PD, and 48% of epilepsy. In the longstanding Oxford data from 1979 to 2006, death rates based on all certified causes of death showed no significant change for MS; an upward trend for MND (notably in women over 75), though only in the last few years of the study; a significant but small decline for PD; and no significant change for epilepsy. When mortality statistics are analysed by underlying cause only, their value is reduced. A substantial percentage of neurological deaths are missed. Time trends may be misleading. All certified causes for each disease, as well as the underlying cause, should be analysed.

Psychiatric disorders certified on death certificates in an English population

BackgroundPsychiatric disorders are sometimes certified on death certificates, but seldom selected as the underlying cause of death. The majority of deaths with a certified psychiatric cause are usually omitted from official mortality statistics, which are typically based on the underlying cause alone.AimTo report on death rates for psychiatric disorders, as certified on death certificates, including all mentioned causes as well as the underlying cause of death.MethodAnalysis of database including all certified causes of death in 1979–1999, in three time periods defined by coding rule changes.ResultsStatistics on the underlying cause of death grossly under-estimated certified psychiatric disorders. For example, in the first period of our study they missed 88% of deaths in which schizophrenia was a certified cause, 98% of affective psychosis, and 96% of depression. Over time, considering all certified causes, age-standardised death rates for schizophrenia declined, those for affective psychosis showed no change, and those for depression and dementia increased.ConclusionThe decline in mortality rates for schizophrenia, and the increase for depression and dementia, may reflect real changes over time in disease prevalence at death, although other explanations are possible and are discussed.

Hospital admission and mortality rates for venous thromboembolism in Oxford region, UK, 1975–98

Multiple-cause-coded mortality rates for thromboembolism in the Oxford region of the UK showed a steady decline fom 1975 to 1998. The trend for admission rates was less clear-cut, but recent changes were not specific to women of ages associated with use of oral contraceptives or hormone-replacement therapy.

Childhood cancer incidence in a cohort of twin babies

We studied childhood cancer incidence in a population-based twin cohort using record linkage to the National Registry of Childhood Tumours. After correcting for mortality, an incidence deficit was observed (Standardized Incidence Ratio (SIR) 79; 95% Confidence Interval (CI) 39–120). Pooled analysis with data from published cohort studies indicates a similar significant incidence reduction (SIR 81, 95% CI 67–96). Further studies are warranted.

Trends in mortality for cancers, comparing multiple- and underlying-cause rates, in an English population 1979-1999.

In compiling official mortality statistics, rules for selecting the underlying cause of death have changed twice in the last 20 years in England. Mortality statistics for most types of cancer were not greatly affected, but there were significant effects on coding for cancers of colon, liver, breast, prostate, testis and bladder, and for lymphoma and leukaemia.

Trends in mortality from appendicitis and from gallstone disease in English populations, 1979–2006: study of multiple-cause coding of deaths

Objective To report on trends in mortality from appendicitis and from gallstone disease. Design and setting Analysis of data from death certificates including all certified causes of death, termed ‘mentions’, not just the underlying cause, using data from Oxford (‘mentions’ available from 1979 to 2006) and all England (‘mentions’ available from 1995 to 2006). Outcome measures Mortality rates across all years studied, and in periods defined by changes to national rules in selecting the underlying cause of death. Results In the Oxford region, mortality rates for appendicitis based on underlying cause fluctuated between periods defined by coding rule changes. Those based on mentions were more stable: they were 3.3 per million population in 1979–83 and 3.5 in 2001–6. Nationally, mentions-based rates fell from 3.4 per million in 1995–2000 to 3.1 in 2001–2006, a small but statistically significant decline. For gallstone disease in Oxford, rates fell in the early years but then levelled off. Mortality rates for gallstones in England actually rose slightly between 1995–2000 (15.0 deaths per million) and 2001–6 (15.6 per million). The majority of deaths from appendicitis and gallstone disease were in the elderly. Conclusions Appendicitis and gallstones are considered to be avoidable causes of death. Mortality rates for each cause have not shown any substantial falls in the recent periods covered by this study. It is possible that deaths were not, in fact, avoidable, but audits into the circumstances of deaths from these diseases may be warranted. Mortality data based on underlying cause alone substantially underestimate deaths from these two conditions.