We have investigated biases in physicians' decisions regarding the form of life support to withdraw from critically ill patients in whom the decision to withdraw has already been made. Using a specially designed instrument that solicited both self-reported preferences and also responses to experimentally varied clinical vignettes, we surveyed 862 American internists, of whom 481 (56%) responded. Physicians do have preferences about the form of life support withdrawn. From most likely to least likely the order is: blood products, haemodialysis, intravenous vasopressors, total parenteral nutrition, antibiotics, mechanical ventilation, tube feedings, and intravenous fluids. Four biases in decision making were also identified. Physicians prefer to withdraw forms of therapy supporting organs that failed for natural rather than iatrogenic reasons, to withdraw recently instituted rather than longstanding interventions, to withdraw forms of therapy resulting in immediate death rather than delayed death, and to withdraw forms of therapy resulting in delayed death when confronted with diagnostic uncertainty. Because these biases may have clinical, social, and ethical consequences counter to patient goals, and because they may affect the underlying decision whether to withdraw life support at all, they may represent impediments to rational and compassionate decision making in critical care.

To compare octreotide with injection sclerotherapy in the treatment of acute variceal haemorrhage, patients admitted with gastrointestinal bleeding and oesophageal varices confirmed by endoscopy were randomised to receive either emergency sclerotherapy with 3% sodium tetradecyl sulphate or octreotide (50 micrograms intravenous bolus plus 50 micrograms per h intravenous infusion for 48 h). At the end of the study period (48 h), the octreotide group also had sclerotherapy to obliterate the varices. 100 patients were recruited. Demographic features including the aetiology of portal hypertension and the Child-Pugh's grading of the two groups were similar. Bleeding was initially controlled in 90% of patients by emergency sclerotherapy and in 84% by octreotide infusion (95% confidence interval 0-19.5, p = 0.55). There were no significant differences between the two groups in early (within 48 h of randomisation) rebleeding (16% vs 14%), blood transfusion (3 units vs 3.5), hospital stay (5 days vs 6 days), or hospital mortality (27% vs 20%). No notable side-effects were associated with octreotide. We conclude that octreotide infusion and emergency sclerotherapy are equally effective in controlling variceal haemorrhage.

Randomised assessment of new laparoscopic surgical techniques is difficult. Surgeons need time to become experienced with the methods and tend, when they have experience, to favour one or other approach. We have carried out a prospective randomised comparison of laparoscopic and conventional appendicectomy done by surgeons of comparable experience in patients with suspected acute appendicitis. Postoperative management decisions were made by surgeons other than the operating surgeon. 140 patients were randomly assigned to open (OA) or laparoscopic (LA) appendicectomy (70 each). The age, sex ratio, duration of symptoms, and proportion of patients with histologically confirmed appendicitis was similar in the two groups. Operating time was longer for LA than for OA (mean 70.3 [SD 21.9] vs 46.5 [25.9] min; p < 0.001). There were no major intraoperative complications in either group. 14 (20%) patients in the LA group required conversion to an open operation. No significant differences between the groups were found postoperatively for pain score, analgesic requirement, time to reintroduction of diet, or hospital stay. 46 LA patients and 42 OA patients attended follow-up 3 weeks after surgery. Similar proportions had returned to work (36 [79%] vs 31 [74%]). The frequency of wound complications and wound pain after leaving hospital was lower after LA but not significantly so. We conclude that the postoperative course after LA and conventional OA does not differ significantly.

There is uncertainty over whether vitamin A supplementation reduces morbidity among children with subclinical deficiency of the vitamin. Hence a double-blind, placebo-controlled trial of the effect of vitamin A supplementation on childhood morbidity was conducted among 11,124 children aged 6-83 months in the northwest of Haiti. After a random start, children were sequentially assigned by household units to receive either megadose vitamin A or placebo in three distribution cycles 4 months apart. 2 to 8 weeks after each administration of the vitamin A and placebo capsules, indicators of childhood morbidity were reassessed through interviews conducted in the homes of participating families. The vitamin A group was found to have an increased 2-week prevalence of all symptoms and signs of childhood morbidity assessed, including diarrhoea (rate ratio [RR] = 1.09, 95% confidence interval 1.05-1.14), rhinitis (RR = 1.02, 95% confidence interval 1.00-1.04), cold/flu symptoms (RR = 1.04, 95% confidence interval 1.01-1.06), cough (RR = 1.07, 95% confidence interval 1.03-1.11), and rapid breathing (RR = 1.18, 95% confidence interval 1.09-1.27). The study shows an increased 2-week prevalence of diarrhoea and the symptoms of respiratory infections after vitamin A supplementation.

A double-blind, randomised, placebo-controlled trial was conducted to evaluate the safety and toxicity of vitamin A supplementation within the Expanded Programme on Immunisation (EPI) in rural Bangladesh. 191 infants received 3 doses of either 50,000 IU of vitamin A or placebo at about 1.5, 2.5, and 3.5 months and were examined on days 1, 2, 3, and 8 after supplementation. 11 infants (11.5%) supplemented with vitamin A had episodes of bulging of the fontanelle as opposed to 1 (1%) in the placebo group. 16 of the 17 events occurred in the vitamin A supplemented group. No other side effects were noted. There was a tendency towards a cumulative effect of toxicity with increasing doses.

10-40% of children with Down's syndrome have atlantoaxial instability. These children might run the risk of spinal cord compression if they play sport. The aim of our study was to assess this presumed risk. We obtained 282 radiographs of the cervical spine from a cohort of 400 children and young adults with Down's syndrome who attended special schools and who were between 4 and 20 years old (about 25% of all such children in the Netherlands). The atlantoaxial distance was more than 4 mm in 91 children. These children were randomly assigned to two groups, with the provision that all children at any particular school were assigned to the same group. Children of one group were allowed to continue their habitual sports and exercise activities, whereas those in the other group were advised not to play "risky" sports (as defined by a panel of four experts) and not to make "risky" movements during physical education lessons. The compliance of the experimental group was good. After a year, there were no differences between the groups in scores on a functional motor scale, the frequency of neurological signs, or changes in the atlantoaxial distance. The motor function of a third group of 44 children with Down's syndrome but normal atlantoaxial distances was similar to that of children in the other two groups, as was the frequency of neurological signs. These findings suggest there is no reason to stop children with Down's syndrome from playing certain sports and no need to screen them by radiography before they take up such sports activities.

Much of the lung damage that limits the life of young adults with cystic fibrosis is due to proteases and oxygen metabolites generated by neutrophils, which are recruited into the airway by the interaction between Pseudomonas aeruginosa and pulmonary macrophages. Leukotriene B4 (LTB4) has been proposed as a local mediator of this process; its production is susceptible to specific modulation with dietary eicosapentaenoic acid (EPA). We carried out a placebo-controlled trial of EPA (2.7 g daily for 6 weeks) to assess its effects on markers of clinical state, peripheral neutrophil function, and lung inflammation in sixteen patients with cystic fibrosis colonised with P aeruginosa. EPA was well tolerated and resulted in a significant reduction in sputum volume (median change with EPA -10 mL/day, placebo 0; p = 0.015), and improvements in Schwachman score (EPA 5%, placebo 0; p = 0.034), forced expiratory volume in 1 s (EPA 0.25 L, placebo -0.1 L; p = 0.006), and vital capacity (EPA 0.6 L, placebo 0; p = 0.011). Relative chemotaxis of circulating neutrophils to LTB4 increased from a subnormal baseline of 4 (median; range 0-10) microns/30 min before treatment, to a near normal value of 11 (5-18) microns/30 min after EPA. Relative chemotaxis to LTB4 of patients taking placebo did not change: the difference in response was highly significant (p = 0.001). Specific reduction of neutrophil chemotaxis to LTB4 is a sensitive assay of chronic in-vivo exposure to LTB4. Our results suggest that LTB4 has a pathogenetic role in the lung damage of cystic fibrosis. Longer-term clinical trials of EPA are warranted in a larger number of cystic fibrosis patients.

Routine use of steroids as adjunctive treatment of bacterial meningitis remains controversial. We have carried out a prospective, placebo-controlled, double-blind study of dexamethasone in 115 children with acute bacterial meningitis in Switzerland. The patients were randomly assigned to receive either placebo (n = 55) or dexamethasone (n = 60) in addition to optimum antibiotic treatment (100 mg/kg daily ceftriaxone). Dexamethasone therapy (0.4 mg/kg) was started 10 min before the first dose of ceftriaxone and given every 12 h for 2 days. Baseline demographic, clinical, and laboratory features of the two groups were similar. After 24 h treatment meningeal inflammation as shown by cerebrospinal fluid (CSF) glucose concentration was significantly less with dexamethasone than with placebo (mean increase in glucose 63 [76] vs 40 [75]%, p = 0.008). However, other indices of inflammation showed similar changes in both groups. Addition of dexamethasone did not affect the rate at which CSF became sterile. Both groups showed prompt clinical responses and similar frequencies of complications (15 vs 12%). Monitoring for possible adverse effects of dexamethasone revealed no abnormalities. At follow-up examinations 3, 9, and 15 months after hospital discharge, 9 (16%) of 55 placebo recipients and 3 (5%) of 60 dexamethasone recipients had one or more neurological or audiological sequelae (p = 0.066); the relative risk of sequelae was 3.27 (95% CI 0.93-11.47). Our results and those of similarly designed studies lead us to believe that adjunctive dexamethasone therapy improves outcome from bacterial meningitis in infants and children. We recommend its use, preferably in the dose regimen used in this study.

Blindness due to trachoma is a serious public health issue world wide. The currently recommended treatment of active trachoma with repeated doses of tetracycline eye ointment has many disadvantages. The new azalide antibiotic azithromycin is effective as a single oral dose in the chemotherapy of genital Chlamydia trachomatis infections, and we have assessed its efficacy for trachoma treatment. We carried out a randomised single-blind comparison of azithromycin (a single oral dose of 20 mg/kg) with conventional treatment (6 weeks of topical tetracycline plus erythromycin for severe cases) in two villages with endemic trachoma in The Gambia. The patients were followed up for 26 weeks from the start of treatment by an observer unaware of treatment allocation. By 6 months' follow-up, trachoma had resolved in 76 (78%) of 97 subjects who received azithromycin compared with 70 (72%) of 97 who were treated conventionally (95% CI for difference -6% to 18%). Compliance with both treatments was good, but that for conventional treatment could probably not be achieved outside the research setting. There were no significant differences in treatment effect, baseline characteristics, or re-emergent disease between the treatment groups. Azithromycin was well tolerated. As a systemic treatment effective in a single dose it has important potential for trachoma control.

Signs of infection with a central venous access device in situ raise the possibility of catheter sepsis. We evaluated three tests for diagnosis of infection in infants with suspected catheter sepsis. The acridine orange leucocyte cytospin (AOLC) test was 87% sensitive and 94% specific in the diagnosis of catheter-related sepsis defined by quantitative blood culture. The C-reactive protein and nitroblue tetrazolium tests were not as useful. Using the AOLC results, available in an hour, we now remove fewer catheters on suspicion of sepsis alone.

Injection sclerotherapy of bleeding oesophageal varices is undoubtedly beneficial but it is associated with a substantial complication rate, and variceal rebleeding is common during the treatment period before variceal obliteration is achieved. We aimed to find out whether endoscopic variceal banding ligation is safer and more effective. The two methods were compared in a randomised controlled trial of 103 patients (54 assigned to banding ligation, and 49 to injection sclerotherapy) of whom 21 (39%) and 23 (47%), respectively, had active bleeding at index endoscopy. Both treatments were highly effective in controlling active haemorrhage (91% and 92% respectively). Variceal obliteration was not achieved for 22 patients in each group, but among those whose varices were eradicated, banding ligation achieved obliteration more quickly than did sclerotherapy (mean 39 [SD 4] vs 72 [7] days, p = 0.004) and in fewer endoscopy sessions (3.4 [2.2] vs 4.9 [3.5], p = 0.006). Rebleeding was less common in the banding ligation group than in the sclerotherapy group (16 [30%] vs 26 [53%], p < 0.05). There was no difference in outcome between the groups, but 14 sclerotherapy patients were withdrawn from the trial (7 for orthotopic liver transplantation) compared with only 5 (1 for liver transplantation) in the banding ligation group (p < 0.05). Complication rates were similar in the two groups. Variceal banding ligation is a safe and effective technique, which obliterates varices more quickly and with a lower rebleeding rate than injection sclerotherapy.

We conducted a community intervention trial in 12 villages in Tamil Nadu, India to evaluate the benefits of growth monitoring. The villages were divided into 6 "growth-monitoring package" of intervention villages (GMP) and 6 "non-growth-monitoring package" of intervention villages (NGM). A functioning primary health care system was in place in all 12 villages implemented a set of interventions including health and nutritional education. About 550 children under the age of 60 months were studied over 4 years in GMP villages and a similar number of children in NGM villages. The interventions were identical in the two sets of villages except for the use of growth charts in education in the 6 GMP villages. The nutrition worker in the NGM villages had the same contact time as in the GMP villages but advised mothers without the benefit of growth charts. The research team, independently of the nutrition worker, did anthropometric studies on children in all villages every 4 to 5 months. Comparisons were done by calculating monthly gains in stature, and weight, and the significance of differences observed was adjusted for age and sex. After 30 months of interventions, similar improvements in growth were seen in GMP and NGM children. The interventions seemed to have improved the nutritional status of young children in both groups of villages. In view of the lack of additional benefit from growth monitoring over other educational interventions, we question its use as part of child survival programmes in India.

A sick premature baby who requires intensive care will undergo many uncomfortable procedures. It is now accepted that such babies perceive pain and need adequate analgesia, but little is known about the effects of sedation in these patients. We investigated the use of morphine to provide analgesia and sedation for ventilated preterm babies in a randomised, double-blind, placebo-controlled trial. 41 mechanically ventilated babies who had been treated with surfactant (Curosurf) for hyaline membrane disease were randomly assigned morphine in 5% dextrose (100 micrograms/kg per h for 2 h followed by 25 micrograms/kg per h continuous infusion) or 5% dextrose (placebo). Plasma catecholamine concentrations were measured 1 h after the first dose of surfactant and 24 h later. Blood pressure was measured at study entry and after 6 h. The morphine and placebo groups showed no differences in method of delivery, Apgar scores, birthweight, gestation, or catecholamine concentrations at baseline. Morphine-treated babies showed a significant reduction in adrenaline concentrations during the first 24 h (median change -0.4 [95% CI -1.1 to -0.3] nmol/L p < 0.001), which was not seen in the placebo group (median change 0.2 [-0.6 to 0.6] nmol/L, p = 0.79). There was a non-significant reduction in noradrenaline concentration in the morphine group. Blood pressure showed a slight but non-significant fall (median -4 mm Hg) in morphine-treated babies. The incidence of intraventricular haemorrhage, patent ductus arteriosus, and pneumothorax, the number of ventilator days, and the numbers of deaths did not differ significantly between the groups. Morphine, in the dose regimen we used, is safe and effective in reducing adrenaline concentrations in preterm ventilated babies.

Opioids produce analgesia by interacting with local opioid receptors in peripheral inflamed tissue. This study investigated whether endogenous ligands of these receptors are present in synovia and whether such opioid peptides can inhibit pain by activation of intra-articular opioid receptors. Samples of synovium from 8 patients undergoing arthroscopic knee surgery were examined by immunohistochemistry for the presence of beta-endorphin, met-enkephalin, and dynorphin. All tissue samples showed synovitis. Inflammatory cells stained strongly for beta-endorphin and met-enkephalin but not for dynorphin. To find out whether blockade of intra-articular opioid receptors affected pain, we randomly assigned 22 patients undergoing arthroscopic knee surgery to receive naloxone (0.04 mg) intra-articularly (n = 10) or intravenously (n = 12); each patient received a placebo injection into the other site. Postoperative pain was assessed by visual analogue scale, a numerical rating scale, the McGill pain questionnaire, and supplementary analgesic consumption during the next 24 h. All pain scores were higher in the intra-articular naloxone group than in the intravenous naloxone group. The differences were significant (p < 0.05) during the first 4 h. Supplementary analgesic consumption was significantly higher in the intra-articular group (52.5 [14.0] vs 15.6 [8.0] mg diclofenac, p < 0.05). Opioid peptides are present in inflamed synovial tissue and can inhibit pain after knee surgery through an action specific to intra-articular opioid receptors. These findings expand the gate control theory of pain and suggest new approaches such as the development of peripherally acting opioid analgesics without central side-effects.