Public health implications of new guidelines for lead in drinking water: a case study in an area with historically high water lead levels

Concern about the neurotoxicity of lead, particularly in infants and young children, has led to a revision of blood lead levels which are considered to involve an acceptable level of human exposure. Drinking water guidelines have also been reviewed in order to reduce this source of population exposure to lead. In the last 20 years, guidelines have been reduced from 100 to 50 to 10 microg/litre. Lead in tap water used to be a major public health problem in Glasgow because of the high prevalence of houses with lead service pipes, the low pH of the public water supply and the resulting high levels of lead in water used for public consumption. Following two separate programmes of water treatment, involving the addition of lime and, a decade later, lime supplemented with orthophosphate, it is considered that maximal measures have been taken to reduce lead exposure by chemical treatment of the water supply. Any residual problem of public exposure would require large scale replacement of lead service pipes. In anticipation of the more stringent limits for lead in drinking water, we set out to measure current lead exposure from tap water in the population of Glasgow served by the Loch Katrine water supply, to compare the current situation with 12 years previously and to assess the public health implications of different limits. The study was based on mothers of young children since maternal blood lead concentrations and the domestic water that mothers use to prepare bottle feeds are the principal sources of foetal and infant lead exposure. An estimated 17% of mothers lived in households with tap water lead concentrations of 10 microg/litre (the [WHO,] guideline) or above in 1993 compared with 49% in 1981. Mean maternal blood lead concentrations fell by 69% in 12 years. For a given water lead concentration, maternal blood lead concentrations were 67% lower. The mean maternal blood lead concentration was 3.7 microg/litre in the population at large, compared with 3.3 microg/litre in households with negligible or absent tap water lead. Nevertheless, between 63% and 76% of cases of mothers with blood lead concentrations of 10 microg/dl or above were attributable to tap water lead. The study found that maternal blood lead concentrations were well within limits currently considered safe for human health. About 15% of infants may be exposed via bottle feeds to tap water lead concentrations that exceed the WHO guideline of 10 microg/litre. In the context of the health and social problems which affect the well-being and development of infants and children in Glasgow, however, current levels of lead exposure are considered to present a relatively minor health problem.

A comparison of reports from referees chosen by authors or journal editors in the peer review process

The aim was to analyse the peer review process by comparing reports produced by referees selected by journal editors, with those of referees selected by the authors of a scientific manuscript. Some 104 consecutive papers from the UK submitted to the British Journal of Surgery (BJS) were included. Of these, 102 were reviewed blind both by referees chosen by the journal editors, and referees chosen by the paper's principal author. Manuscripts were marked using a standard sheet for four basic aspects: originality, clinical/scientific importance, clarity and analysis; a final overall recommendation about possible publication was given. The time taken and the number of completed referee reports were similar in each group. Referees chosen by the BJS editors were more critical (scored higher) of the submitted articles. Mean scores for all domains were higher than for authors' referees, significantly for scientific importance (p = 0.009) and decision to publish (p = 0.029). In conclusion, reports produced by referees selected by BJS editors were more critical than those chosen by authors of the papers. Authors might argue that this reduced their chance of publication but constructive criticism might improve the final article and assist editors to make decisions about acceptance or rejection.

Genetic polymorphisms of cytochrome p4502E1 and susceptibility to alcoholic liver disease and hepatocellular carcinoma in a white population: a study and literature review, including meta-analysis

AIMS

To investigate the associations between the Rsa I, Dra I, and Taq I genetic polymorphisms of cytochrome p4502E1 and susceptibility to alcoholic liver disease or to hepatocellular carcinoma.

METHODS

DNA samples isolated from 61 patients with alcoholic liver disease, 46 patients with hepatocellular carcinoma, and 375 healthy controls were subjected to polymerase chain reaction amplification followed by digestion with the endonucleases Rsa I, Dra I, or Taq I. Meta-analysis was performed using data from previous studies of Rsa I polymorphism and the risk of alcoholic liver disease.

RESULTS

No association was found between any of the three polymorphisms and susceptibility to hepatocellular carcinoma. The distributions of Rsa I and Dra I alleles among the patients with alcoholic liver disease were not significantly different from those among the control group. Meta-analysis of this data and previous data concerning Rsa I polymorphism and alcoholic liver disease risk failed to demonstrate any significant association between the two. However, the alcoholic liver disease group in this study showed a significantly lower frequency of the less common Taq I allele compared with the healthy control group (odds ratio, 0.33; 95% confidence interval, 0.12 to 0.78).

CONCLUSIONS

Possession of the less common Taq I cytochrome p4502E1 allele is associated with reduced susceptibility to alcoholic liver disease. There is no existing evidence that the Taq I polymorphism is directly associated with altered alcohol metabolism, but it might be in linkage disequilibrium with as yet unidentified protective factors.

Domiciliary occupational therapy for patients with stroke discharged from hospital: randomised controlled trial

OBJECTIVE

To establish if a brief programme of domiciliary occupational therapy could improve the recovery of patients with stroke discharged from hospital.

DESIGN

Single blind randomised controlled trial.

SETTING

Two hospital sites within a UK teaching hospital.

SUBJECTS

138 patients with stroke with a definite plan for discharge home from hospital.

INTERVENTION

Six week domiciliary occupational therapy or routine follow up.

MAIN OUTCOME MEASURES

Nottingham extended activities of daily living score and "global outcome" (deterioration according to the Barthel activities of daily living index, or death).

RESULTS

By eight weeks the mean Nottingham extended activities of daily living score in the intervention group was 4.8 points (95% confidence interval -0.5 to 10.0, P=0.08) greater than that of the control group. Overall, 16 (24%) intervention patients had a poor global outcome compared with 30 (42%) control patients (odds ratio 0.43, 0.21 to 0.89, P=0.02). These patterns persisted at six months but were not statistically significant. Patients in the intervention group were more likely to report satisfaction with a range of aspects of services.

CONCLUSION

The functional outcome and satisfaction of patients with stroke can be improved by a brief occupational therapy programme carried out in the patient's home immediately after discharge. Major benefits may not, however, be sustained.

Promoting good research practice

Embarking on a new millennium provides the stimulus both to take stock and also to look forward. In the field of medical statistics there is much to make us feel proud and excited. Rapid methodological developments together with parallel developments in computer technology have enormously expanded our statistical repertoire. At the same time, the high profile attained by the evidence-based medicine movement means that the importance of our discipline is recognized more widely then ever before. However, any medical statistician who is involved in medical publishing, or who is even a regular reader of the medical literature, must be aware of the yawning chasm between what is recognized as good statistical practice and what is actually published. Poor study design, inappropriate analysis and selective reporting are commonplace. In my opinion the most important challenge currently facing our profession is the task of bridging this chasm.

Effects of influenza vaccination of health-care workers on mortality of elderly people in long-term care: a randomised controlled trial

BACKGROUND

Vaccination of health-care workers has been claimed to prevent nosocomial influenza infection of elderly patients in long-term care. Data are, however, limited on this strategy. We aimed to find out whether vaccination of health-care workers lowers mortality and the frequency of virologically proven influenza in such patients.

METHODS

In a parallel-group study, health-care workers in 20 long-term elderly-care hospitals (range 44-105 patients) were randomly offered or not offered influenza vaccine (cluster randomisation, stratified for policy for vaccination of patients and hospital size). All deaths among patients were recorded over 6 months in the winter of 1996-97. We selected a random sample of 50% of patients for virological surveillance for influenza, with combined nasal and throat swabs taken every 2 weeks during the epidemic period. Swabs were tested by tissue culture and PCR for influenza viruses A and B.

FINDINGS

Influenza vaccine uptake in health-care workers was 50.9% in hospitals in which they were routinely offered vaccine, compared with 4.9% in those in which they were not. The uncorrected rate of mortality in patients was 102 (13.6%) of 749 in vaccine hospitals compared with 154 (22.4%) of 688 in no-vaccine hospitals (odds ratio 0.58 [95% CI 0.40-0.84], p=0.014). The two groups did not differ for proportions of patients positive for influenza infection (5.4% and 6.7%, respectively); at necropsy, PCR was positive in none of 17 patients from vaccine hospitals and six (20%) of 30 from no-vaccine hospitals (p=0.055).

INTERPRETATION

Vaccination of health-care workers was associated with a substantial decrease in mortality among patients. However, virological surveillance showed no associated decrease in non-fatal influenza infection in patients.

The value of the "worst" computed tomographic scan in clinical studies of moderate and severe head injury. European Brain Injury Consortium

OBJECTIVE

Computed tomographic (CT) scanning can reveal the pattern and severity of structural brain damage after head injury. With the proliferation of CT scanners in general hospitals, and with improvements in patient transport, the interval from injury to the first CT scan is decreasing. The potential result is an "admission" scan missing an evolving and potentially operable lesion. Furthermore, the literature is confusing regarding the timing and coding of CT findings. We sought to establish the frequency of deterioration in CT appearance from an admission scan to subsequent scans and the prognostic significance of such deterioration.

METHODS

In a survey organized by the European Brain Injury Consortium, data on initial severity, management, and subsequent outcome were gathered prospectively for 1005 patients with moderate or severe head injury admitted to one of 67 European neurosurgical units during a 3-month period in 1995. The findings of the initial and the final ("worst") CT scan were classified according to the Traumatic Coma Data Bank system and were related to outcome as assessed using the Glasgow Outcome Scale 6 months after injury.

RESULTS

Data on an initial and a final CT scan were available for 897 patients; of these, 724 patients were assessed using the Glasgow Outcome Scale at 6 months. The initial CT findings were classified as a diffuse injury for 53% of the cohort, with 16% of these diffuse injuries demonstrating deterioration on a subsequent scan. In 56 (74%) of 76 deteriorations, the change was from a diffuse injury to a mass lesion. When the initial CT scan demonstrated a diffuse injury without swelling or shift, evolution to a mass lesion was associated with a statistically significant increase in the risk of an unfavorable outcome (62% versus 38%). When the initial scan demonstrated evidence of swelling or shift, there was a nonsignificant trend in the opposite direction, although the numbers were limited.

CONCLUSION

When an admission CT scan demonstrates evidence of a diffuse injury, follow-up scans should be performed, because approximately one in six such patients will demonstrate significant CT evolution. In studies comparing series of head-injured patients, correspondence of timing of CT scans is necessary for valid comparison.

Evaluation of designs for clinical trials of neuroprotective agents in head injury. European Brain Injury Consortium

In a study involving the statistical modeling of potential head injury trials, we explore approaches to trial design that could enhance their power to detect treatment-related effects on clinical outcome. The study uses a survey organized by the European Brain Injury Consortium of over 1,000 head-injured patients to characterize the population from which trial participants can be selected. A variety of models are postulated for the effects of "neuroprotective" treatment on outcome, and their interaction with a range of strategies for targeting patients for inclusion in the trial is evaluated. A very simple strategy of targeting patients with an intermediate prognosis was found to allow a reduction in sample size by 30%, with no reduction in statistical power. This paper illustrates an important methodology for studying the characteristics of competing trial designs.

Head injuries in four British neurosurgical centres

An issue in the design of trials in traumatic brain injury is whether variation amongst centres in 'conventional' management could mask the impact of a powerful new pharmacological agent. We report the results of an observational study of 988 patients admitted to one of four British neurosurgical units between 1986 and 1988 within 3 days of a severe head injury. The centres fell into two pairs on the basis of the 'intensity' of management. In Edinburgh and Southampton, more frequent use of intracranial pressure monitoring, ventilation and osmotic diuretics was made than in Glasgow and Liverpool. The odds ratio for an independent outcome at 6 months in Edinburgh or Southampton, relative to Glasgow or Liverpool, controlling for case mix, was 1.43 (95% CI, 1.03-1.98, p = 0.033). Thus, there is weak evidence of an association between the approach to management and clinical outcome at 6 months.

Comparing outcomes of percutaneous transluminal coronary angioplasty with coronary artery bypass grafting; can routine health service data complement and enhance randomized controlled trials?

AIM

To compare outcomes of percutaneous transluminal coronary angioplasty (PTCA) with coronary artery bypass graft surgery (CABG) for a population stemming from routinely collected data, in order to assess the merits of such data sources as a complement, and possible enhancement, to randomized controlled trial results.

METHODS AND RESULTS

A population of Scottish patients were taken from a routine discharge summary and from this data source patients comparable to those from randomized controlled trial settings were identified. Between 1989 and 1995, 12 238 pseudo randomized controlled trial patients were identified from the routine data set, of which 3714 (30.3%) received PTCA and 8524 (69.7%) received CABG. The baseline characteristics of the pseudo randomized controlled trial and randomized controlled trial patients were similar. The evidence from both the randomized controlled trials and routine data indicate that for 1 year follow-up the risk of cardiac death and/or non-fatal myocardial infarction is not significantly different between the two treatment groups.

CONCLUSION

The outcomes expected of PTCA and CABG following trial evidence have been realized in the routine data which are representative of a complete, non-selective population. Due to the size of the routine data set it would be possible to set up hypotheses for potential subgroup effects at the outset.

The European Brain Injury Consortium survey of head injuries

To provide a picture of contemporary practice, a survey was carried out of severely and moderately head injured patients admitted to 67 'neuro' centres in 12 European countries. 1,005 adult head injuries were recruited over a three month period. Sixty items of information on demography, clinical features, investigations, management and early complications were captured on a simple, two-page questionnaire and, information on outcome at six months on a third page. The median age of the subjects was 38 years, 74% were male and 51% injured in road traffic accidents; 57% of patients were transferred to the 'neuro' centre from another hospital. Assessment of clinical responsiveness was limited by the use of sedation and intubation and information from four early time points (pre-hospital, arrival at the Accident and Emergency department, post-resuscitation, and arrival at the 'neuro' unit) was combined to stratify the subjects as severe (58%), moderate (17%) or intermediate (19%). In 48% of patients classified the CT scan showed features of a 'mass lesion' and in 40% showed a subarachnoid haemorrhage. Fifty-five centres provided the data on outcome for 94% of the cases recruited in these centres six months after injury. 31% died, 3% were vegetative, 16% severely disabled, 20% moderately disabled and 31% had made a good recovery. Comparison of the data from different parts of Europe showed differences in the frequency of secondary transfer, cause of injury, occurrence of major extracranial injury, CT scan findings, intracranial operation, clinical severity of injury and utilisation of the components of intensive care and the occurrence of a favourable outcome, although the latter difference was not statistically significant when variations in the initial severity of injury were taken into account. The findings in the present survey are compared with newly analysed information for three previous large series: the International Data Bank involving the UK, the Netherlands and the USA, the North American Traumatic Coma Data Bank, and data from four centres in the UK. The comparisons showed substantial similarities and also differences that may reflect variations in policy for admission of the head injury to 'neuro' units, and evolution in methods of assessment, investigation and management. The effects of these differences on outcome requires further, rigorous prospective study.

Factors influencing tackle injuries in rugby union football

OBJECTIVES

To assess the influence of selected aspects of lifestyle, personality, and other player related factors on injuries in the tackle. To describe the detailed circumstances in which these tackles occurred.

METHODS

A prospective case-control study was undertaken in which the tackling and tackled players ("the cases") involved in a tackle injury were each matched with "control" players who held the same respective playing positions in the opposing teams. A total of 964 rugby matches involving 71 senior clubs drawn from all districts of the Scottish Rugby Union (SRU) were observed by nominated linkmen who administered self report questionnaires to the players identified as cases and controls. Information on lifestyle habits, match preparation, training, and coaching experience was obtained. A validated battery of psychological tests assessed players' trait anger and responses to anger and hostility. The circumstances of the tackles in which injury occurred were recorded by experienced SRU coaching staff in interviews with involved players after the match.

RESULTS

A total of 71 tackle injury episodes with correct matching of cases and controls were studied. The following player related factors did not contribute significantly to tackle injuries: alcohol consumption before the match, feeling "below par" through minor illness, the extent of match preparation, previous coaching, or practising tackling. Injured and non-injured players in the tackle did not differ in their disposition toward, or expression of, anger or hostility. Some 85% of tackling players who were injured were three quarters, and 52% of injuries occurred when the tackle came in behind the tackled player or within his peripheral vision. Either the tackling or tackled player was sprinting or running in all of these injury episodes. One third of injuries occurred in differential speed tackles--that is, when one player was travelling much faster than the other at impact. The player with the lower momentum was injured in 80% of these cases. Forceful or crunching tackles resulting in injury mostly occurred head on or within the tackled player's side vision.

CONCLUSIONS

Attention should be focused on high speed tackles going in behind the tackled player's line of vision. Comparative information on the circumstances of the vast majority of tackles in which no injury occurs is required before any changes are considered to reduce injuries in the tackle.

Reproducibility of pulse wave velocity and augmentation index measured by pulse wave analysis

OBJECTIVE

The aim of this study was to determine the reproducibility of pulse wave velocity (PWV) and augmentation index (AIx) measured using pulse wave analysis (PWA), prior to its use in large-scale clinical trials.

METHODS

Arterial pressure waveforms were recorded and analysed using an established technique (Sphygmocor). Subjects with and without a range of recognized cardiovascular risk factors were studied to provide a wide range of values. Measurements were made after a brief introduction to the technique in a clinical setting. Two observers recorded aortic and brachial PWV in 24 subjects, each on two occasions, in a random order. In a separate study, two different observers used PWA to determine AIx in 33 subjects, each on two occasions, in a random order. Data were analysed using Bland-Altman plots and presented as mean +/- SD.

RESULTS

Brachial PWV was 8.65+/-1.58 m/s (range 6.16-10.95 m/s) and aortic PWV was 8.15+/-3.01 m/s (5.01-17.97 m/s). Within-observer variability was 0.14+/-0.82 m/s for brachial PWV and 0.07+/-1.17 m/s for aortic PWV. Corresponding between-observer values were -0.44+/-1.09 m/s and -0.30+/-1.25 m/s. AIx ranged from -15.0 to +45.0%, with a group mean of +19.6+/-12.0%. The within-observer difference was 0.49+/-5.37% and between-observer difference 0.23+/-3.80%.

CONCLUSION

PWA is a simple and reproducible technique with which to measure PWV and AIx. Reproducibility accords with that reported by other workers using different methodologies. PWA may, therefore, be suitable for large-scale population and intervention studies investigating the clinical relevance of vascular stiffness.

Oestrogen receptor status predicts local recurrence following breast conservation surgery for early breast cancer

AIMS

To evaluate factors predicting locoregional recurrence in patients treated for early breast carcinoma by breast conservation surgery with or without radiotherapy.

METHODS

A retrospective study of 256 patients was carried out, with special emphasis on the role of oestrogen receptor status. Other parameters studied included age, menopausal status, size of primary tumour, tumour type, axillary nodal status and adjuvant therapy.

RESULTS

Multivariate analysis showed the following parameters to be independent predictors of locoregional recurrence. Radiotherapy (57% reduction in hazard, P = 0.004): expression of oestrogen receptors (52% reduction in hazard, P = 0.008); tamoxifen therapy (46% reduction in hazard, P = 0.023); tumour size (40% increase in hazard per cm, P<0.001).

CONCLUSIONS

Within this study, lack of oestrogen receptor expression was a strong independent factor associated with a higher rate of locoregional recurrence. This may be of value in selecting a group of patients less suitable for breast conservation surgery.

Influence of graft material on blood rheology and plasma biochemistry following insertion of an infrainguinal bypass graft

BACKGROUND

Occlusive arterial disease causes alterations in blood rheology and levels of potential thrombotic and fibrinolytic mediators. The aim of this study was to investigate the effect of graft materials on these parameters in patients undergoing successful infrainguinal revascularization.

METHODS

Some 186 consecutive infrainguinal grafts were observed for 12 months. Venous blood was sampled before operation and at 3, 6, and 12 months after surgery. Samples were assayed for thrombotic and rheological parameters. An area under the curve analysis was used to compare the effects of vein and synthetic grafting on these parameters in 99 patients whose grafts remained patent and free from stenosis.

RESULTS

Plasma levels of fibrin degradation products were significantly higher in patients with synthetic grafts (n = 46) than in those with autogenous vein grafts (n = 53) (median 274 versus 150 ng/ml; P < 0.001). There were no significant differences in plasma fibrinogen or any other parameters between the two groups.

CONCLUSION

Patients with a synthetic infrainguinal graft have a higher fibrin turnover than those with a vein graft. Further studies are required to determine whether this increase in fibrin turnover is an essential requirement to maintain patency of a synthetic infrainguinal graft.

Service provision for people with schizophrenia. I. Clinical and economic perspective

BACKGROUND

The aim of this study was to provide information on patients current service use which could inform future decisions on service planning and resource allocation.

METHOD

Individuals with a diagnosis of schizophrenia, who had received in-patient care in the previous five years, were identified from the Lothian Case Register. Information was obtained from 193 subjects. Patients' service use over a six-month period was examined. The costs incurred in service provision were determined.

RESULTS

Patients differed markedly in their use of services. This was not found to be related to their mental state. Average care costs were high. In-patient care accounted for most of the overall expenditure.

CONCLUSIONS

There is considerable variation in the services used by patients with schizophrenia and in the costs incurred in service provision. When planning services it is therefore important that detailed information on the patient population is available if resources are to be allocated cost-effectively.

Service provision for people with schizophrenia. II. Role of the general practitioner

BACKGROUND

This second report of a study of service provision for patients with schizophrenia describes patients' contact with general practice and general practitioners' (GPs') views of the mental health services.

METHOD

A postal questionnaire was sent to the GPs, and patients' primary care records were examined.

RESULTS

Data were collected on 131 subjects. The majority of patients (96) (73%) were in regular contact with their GP and were consulting for many different reasons; 27 (21%) were posing particular difficulties for the primary care team. GPs reported that 27 (21%) patients required additional support and that the care arrangements for 50 (38%) patients could be improved if alterations were made to the roles of the professionals already involved.

CONCLUSIONS

GPs are central to service provision for patients with schizophrenia. Both additional resources and changes in working practices are required to improve patient care. The service implications of these findings are discussed.

Follow-up study of patients randomly allocated ramipril or placebo for heart failure after acute myocardial infarction: AIRE Extension (AIREX) Study. Acute Infarction Ramipril Efficacy

BACKGROUND

In the Acute Infarction Ramipril Efficacy (AIRE) Study, the effect of angiotensin-converting-enzyme (ACE) inhibition on the survival of patients with clinical heart failure after acute myocardial infarction (AMI), was assessed. At an average follow-up time of 15 months after randomisation, all-cause mortality was reduced from 22.6% (placebo group) to 16.9% (ramipril group) representing an absolute mortality reduction of 5.7% and a relative risk reduction of 27% (95% CI 11-40%; p = 0.002). Our aim in this study was to assess the long-term (3 years after the AIRE Study closed) magnitude, duration, and reliability of the survival benefits observed after treatment with ramipril (target dose 5 mg twice a day) when compared with placebo.

METHODS

We investigated the mortality status of all 603 patients recruited from the 30 UK centres involved in the AIRE Study. Through government records we were able to confirm the death or survival of all 603 patients exactly 3 years after the close of the AIRE Study. Follow-up was for a minimum of 42 months and a mean of 59 months. The average duration of treatment with masked trial medication was 13.4 months for placebo and 12.4 months for ramipril.

FINDINGS

By 0000 h March 1, 1996, death from all causes had occurred in 117 (38.9%) of 301 patients randomly assigned placebo and 83 (27.5%) of 302 patients randomly assigned ramipril, representing a relative risk reduction of 36% (95% CI 15-52%; p = 0.002) and an absolute reduction in mortality of 11.4% (114 additional 5-year survivors per 1000 patients treated for an average of 12.4 months).

INTERPRETATION

Our data provide robust evidence that administration of ramipril to patients with clinically defined heart failure after AMI results in a survival benefit that is not only large in magnitude, but also sustained over many years.

Is hyperglycaemia an independent predictor of poor outcome after acute stroke? Results of a long-term follow up study

OBJECTIVE

To determine whether raised plasma glucose concentration independently influences outcome after acute stroke or is a stress response reflecting increased stroke severity.

DESIGN

Long-term follow up study of patients admitted to an acute stroke unit.

SETTING

Western Infirmary, Glasgow.

SUBJECTS

811 patients with acute stroke confirmed by computed tomography. Analysis was restricted to the 750 non-diabetic patients.

MAIN OUTCOME MEASURES

Survival time and placement three months after stroke.

RESULTS

645 patients (86%) had ischaemic stroke and 105 patients (14%) haemorrhagic stroke. Cox's proportional hazards modelling with stratification according to Oxfordshire Community Stroke Project categories identified increased age (relative hazard 1.36 per decade; 95% confidence interval 1.21 to 1.53), haemorrhagic stroke (relative hazard 1.67; 1.22 to 2.28), time to resolution of symptoms > 72 hours (relative hazard 2.15; 1.15 to 4.05), and hyperglycaemia (relative hazard 1.87; 1.43 to 2.45) as predictors of mortality. The effect of glucose concentration on survival was greatest in the first month.

CONCLUSIONS

Plasma glucose concentration above 8 mmol/l after acute stroke predicts a poor prognosis after correcting for age, stroke severity, and stroke subtype. Raised plasma glucose concentration is therefore unlikely to be solely a stress response and should arguably be treated actively. A randomised trial is warranted.