Meta-analysis of clinical trials: summary of an international conference

An international conference of clinicians, clinical investigators and biostatisticians discussed meta-analysis in relation to clinical trials, i.e. the combination of data from separate studies for the purpose of obtaining information that cannot be derived from the individual studies. Meta-analysis can be a helpful tool for generating hypotheses for future trials, for studying the consistency of trials of the same or similar goals, and for generating more precise estimates of effect. There are still a number of unresolved questions about the methodology and interpretation of meta-analysis. Better and more uniform reporting of primary studies would increase the usefulness of meta-analysis.

Publication bias and clinical trials

A study was performed to evaluate the extent to which the medical literature may be misleading as a result of selective publication of randomized clinical trials (RCTs) with results showing a statistically significant treatment effect. Three hundred eighteen authors of published trials were asked whether they had participated in any unpublished RCTs. The 156 respondents reported 271 unpublished and 1041 published trials. Of the 178 completed unpublished RCTs with a trend specified, 26 (14%) favored the new therapy compared to 423 of 767 (55%) published reports (p less than 0.001). For trials that were completed but not published, the major reasons for nonpublication were "negative" results and lack of interest. From the data provided, it appears that nonpublication was primarily a result of failure to write up and submit the trial results rather than rejection of submitted manuscripts. The results of this study imply the existence of a publication bias of importance both to meta-analysis and the interpretation of statistically significant positive trials.

Meta-analysis of clinical trials as a scientific discipline. II: Replicate variability and comparison of studies that agree and disagree

The replicate variability of meta-analyses of controlled clinical trials has been assessed as a measure of scientific precision. 46 of 91 known meta-analysis papers were divided into 20 cohorts of studies of the same therapies. Ten cohorts contained meta-analyses with different statistical conclusions; 14 contained differing clinical conclusions with a wider spread than the statistically differing studies. Possible causes of variability, such as different trials included, different policies regarding the inclusion of non-randomized and unpublished trials, and different statistical methodologies, were not obvious causes of differing conclusions. Further work in this area should include multivariate analyses in order to explore possible interactions in the factors accounting for the variability found in replicate meta-analyses.

Meta-analysis of randomised controlled trials of nicotine chewing-gum

A meta-analysis was made of fourteen randomised controlled trials that evaluated the efficacy of nicotine chewing-gum in stopping patients smoking. The combined success rates in specialised cessation clinics are significantly higher with nicotine gum (27%) than with placebo gum (18%) at 6 months (n = 734), and 23% and 13% at 12 months, respectively. In contrast, success rates in general medical practices are similar with nicotine gum (11.4%) and with placebo gum (11.7%) at 6 months (n = 1022). However, in general practices, the success rates are 17% for nicotine gum and 13% for the no gum control at 4-6 months, and 9% and 5% at 12 months, a significant difference between the treatments at each time (n = 2238). The data suggest that proper use of nicotine gum in specialised clinics will increase the rate of stopping patients smoking. The use of the gum in general medical practices is questionable.

Medullary carcinoma of the thyroid, pancreatic nesidioblastosis and microadenosis, and pancreatic polypeptide hypersecretion: a new association and clinical and hormonal responses to long-acting somatostatin analog SMS 201-995

We describe a 63-yr-old man with disseminated medullary carcinoma of the thyroid and pancreatic nesidioblastosis and microadenosis with pancreatic polypeptide (PP) hypersecretion. His major symptoms were watery diarrhea, flushing, and abdominal bloating; these and the elevated plasma PP levels did not change after resection of the distal two thirds of the pancreas, which contained a 2-cm mass of nesidioblastotic tissue. Postoperatively, a long-acting somatostatin analog, SMS 201-995 (100 micrograms/day), normalized PP secretion acutely and chronically (7 months) and ameliorated his symptoms. The analog had no side-effects and did not alter glucose tolerance, calcitonin hypersecretion, or growth of the medullary carcinoma, but it did inhibit GH secretion. After withdrawal from therapy for 1 month, PP hypersecretion and all symptoms except diarrhea recurred. The coexistence of medullary carcinoma of the thyroid and PP cell nesidioblastosis represents a new variant of the overlap syndromes between multiple endocrine neoplasia types I and II. Patients with medullary carcinoma and unexplained watery diarrhea should have fasting gastroenteropancreatic hormone assays done to screen for a potential gastrointestinal or pancreatic origin for the diarrhea.

Meta-analysis of clinical trials as a scientific discipline. I: Control of bias and comparison with large co-operative trials

Meta-analysis is an important method of bridging the gap between undersized randomized control trials and the treatment of patients. However, as in any retrospective study, the opportunities for bias to distort the results are widespread. Attempts must be made to introduce the controls found in prospective studies by blinding the selection of papers and extraction of data and making blinded duplicate determinations. Informal and personalized methods of obtaining data are probably more liable to error and bias than employing only published data. Publication bias is a serious problem requiring further research. There also need to be more comparisons of meta-analysed small studies with large co-operative trials.

Meta-analyses of randomized controlled trials

A new type of research, termed meta-analysis, attempts to analyze and combine the results of previous reports. We found 86 meta-analyses of reports of randomized controlled trials in the English-language literature. We evaluated the quality of these meta-analyses, using a scoring method that considered 23 items in six major areas--study design, combinability, control of bias, statistical analysis, sensitivity analysis, and application of results. Only 24 meta-analyses (28 percent) addressed all six areas, 31 (36 percent) addressed five, 25 (29 percent) addressed four, 5 (6 percent) addressed three, and 1 (1 percent) addressed two. Of the 23 individual items, between 1 and 14 were addressed satisfactorily (mean +/- SD, 7.7 +/- 2.7). We conclude that an urgent need exists for improved methods in literature searching, quality evaluation of trials, and synthesizing of the results.

Pharmacokinetics of mezlocillin in patients with hepatobiliary dysfunction

The pharmacokinetics of mezlocillin were investigated in 26 patients with alcoholic liver disease. Serum concentrations of mezlocillin were measured following intravenous administration of 3 g doses over 30 min. The mean peak serum concentration (+/- standard deviation) of mezlocillin at the end of infusion was 138.8 +/- 55.7 mg/l and the mean terminal half-life (T 1/2 beta) was 2.10 +/- 0.9 h. The 24 h urinary recovery of mezlocillin was 35.4 +/- 12.4% of the administered dose. Serum clearance was found to be inversely correlated with alkaline phosphatase and with total bilirubin. The T 1/2 beta was related to the following clinical measurements: age, SGOT and prothrombin time. This relationship suggests it may be prudent to adjust the dosage and the dosage interval of mezlocillin in patients with hepatobiliary dysfunction.

Loss of patients in clinical trials that measure long-term survival following myocardial infarction

Loss of patients from clinical trials can nullify adequate randomization if the loss is unequally distributed among treatment groups. This study was designed to assess the magnitude of the problem in randomized control trials evaluating long-term therapy for survivors of myocardial infarction (MI). Only 19 of 52 trials reported having an explicit policy on withdrawals in the design stage; only 2 reported blinding the decision for withdrawal and only 7 reported accounting for withdrawals in sizing. In addition, only 16 gave the reader enough information to calculate the effect of withdrawals on trial results. In 2 of these 16 trials a p less than 0.05 result obtained by including withdrawals (intention to treat method) was reduced to p less than 0.05 when withdrawals were excluded. It is evident that many long-term trials do not contain adequate data on withdrawals. Readers of published trials are seldom able to judge whether or not withdrawals might affect the final results.

Transcutaneous oxygen monitoring during fiberoptic bronchoscopy

A transcutaneous oxygen sensor was used to detect the development of hypoxemia in patients undergoing fiberoptic bronchoscopy. Measurements of transcutaneous PO2 (tcPO2) correlated well with arterial PO2 (PaO2), and correctly detected all instances when PaO2 was less than 65 mm Hg. The sensor was noninvasive and caused no adverse effects.

Meningococcal pneumonia and empyema

A case of pneumonia and empyema caused by Neisseria meningitidis is described in which the diagnosis was initially missed. Meningococcal pneumonia and empyema are rare. It is important to recognize unusual manifestations of meningococcal infection promptly because of the risk of spread to contacts including health care personnel and other patients.

Lumbar puncture in asymptomatic late syphilis. An analysis of the benefits and risks

We evaluated the treatment of asymptomatic patients with untreated syphilis of more than one year's duration (asymptomatic late syphilis) using a decision-analysis model. Two strategies were compared: treatment with 7.2 million units of penicillin G benzathine, or performing a lumbar puncture to test for asymptomatic neurosyphilis followed by penicillin and management based on cerebrospinal fluid analysis. Estimates of probabilities of disease prevalence, test sensitivity, and cure and complication rates were derived from published studies. Both strategies resulted in a cure rate of at least 99.7% using the best estimates. Although the strategy using lumbar puncture results in a 0.2% higher cure rate, its rate of complications (0.3%) exceeds its marginal benefit. We conclude that a lumbar puncture offers little additional benefit and may increase morbidity in patients with asymptomatic late syphilis.

Should the risk of acquired immunodeficiency syndrome deter hepatitis B vaccination? A decision analysis

The current epidemic of acquired immunodeficiency syndrome (AIDS) and fear that its causative agent contaminates the currently available hepatitis B vaccine may have deterred vaccine use. We formulated a decision-analytic model that compares the risk of death from hepatitis B and AIDS in those vaccinated with the risk of death from hepatitis B alone in those who wait two years for a synthetic vaccine. For individuals with 5% annual risk of hepatitis B, the best current estimate is that vaccination now would save 25 lives per 100,000. The best current estimate of the rate of vaccine-induced AIDS is zero, and one can be 95% confident that the rate is less than eight per 100,000. The rate would have to be considerably higher before postponement of vaccination would be rational for those for whom vaccination has been recommended.

Malignancy in primary biliary cirrhosis. High incidence of breast cancer in affected women

Primary biliary cirrhosis is characterized by abnormalities in both cellular and humoral immunity. It is associated with presumably autoimmune diseases such as Sjögren's syndrome, rheumatoid arthritis, and scleroderma. Sjögren's syndrome and scleroderma have been noted to have an increased frequency of malignancy. Of 208 patients with primary biliary cirrhosis, followed for one month to 15.9 years, extrahepatic malignancies developed in 11, six of whom were women with breast cancer, and one with hepatocellular carcinoma. The incidence of breast cancer was 4.4 times (p less than 0.01) the incidence expected from the rate prevailing in the same age range in a comparable normal population. The incidence of cancer in sites other than the breast and of primary hepatocellular tumor was not significantly increased.

Longitudinal study of persistent generalised lymphadenopathy in homosexual men: relation to acquired immunodeficiency syndrome

42 homosexual or bisexual men with persistent generalised lymphadenopathy not attributable to an identifiable cause have been followed longitudinally since February, 1981. Lymphadenopathy was accompanied by fatigue, low-grade fever and/or night sweats (57%), splenomegaly (29%), leucopenia (40%), hypergamma - globulinaemia (76%), and diminished proportion and absolute numbers of helper T cells (95%). Of the 26 patients who had lymph node biopsy, all showed benign reactive hyperplasia. After 15-30 (median 22) months, 8 patients have met criteria for the diagnosis of acquired immunodeficiency syndrome (AIDS). This outcome was associated with previous heavy nitrite inhalant use, with the presence of night sweats, with leucopenia, and with the triad of constitutional symptoms, splenomegaly, and leucopenia. In addition, a lower mean absolute helper T cell count and an increased frequency of anergy to mumps intradermal antigen and of herpes simplex virus isolation distinguished these patients from those remaining in the cohort, who seem to be stable and in some cases to have improved.

Peer review using a paired-comparison technique

A paired-comparison technique is used to enable surgical residents and attending surgeons to make peer judgments of each other. Comparative peer judgments were made in three areas: the ability of the surgeon to make a diagnosis and to decide on a plan of active medical care, the operating ability of the surgeon, and the postoperative care of the patient. This method of peer judgment ensures the confidentiality of those making the judgments, and the analysis results in a final ranking of the surgeons and surgical residents with indicated significant differences (P less than 0.05) among them. The technique also includes an assessment of how consistent each judge is in making comparisons of fellow surgeons.

Somatostatin metabolism: differences in clearance of N-terminal and central portions of molecule during perfusion of rat liver

Clearance of cyclic somatostatin (SRIF) from a plasma-free recirculating medium containing human erythrocytes and a bovine albumin fraction was measured with site-specific N-terminal (sheep B) and central core-directed (R101) radioimmunoassays during perfusion of the isolated rat liver (3-4 g). With the N-terminal radioimmunoassay (RIA), the t 1/2, hepatic clearance, and extraction of somatostatinlike immunoreactivity (SLI) were 20.9 +/- 2.0 (SE) min, 2.82 +/- 0.27 ml/min, and 35.2 +/- 3.4%. Corresponding values for the centrally directed assay were 51.0 +/- 6.3 min, 1.16 +/- 0.14 ml/min, and 14.4 +/- 1.8%. Clearances of immunoprecipitable 125I-Tyr-SRIF and [125I-Tyr11]SRIF were 6.56 and 1.06 ml/min, respectively, and were not saturable by 1 microM Tyr-SRIF and SRIF, respectively. SRIF (1.26 +/- 0.09 nM) and SRIF-28 (1.34 +/- 0.14 nM) clearances determined by R101 RIA were similar. After SRIF-28 perfusion, high-performance liquid chromatographic analysis of SLI showed 86% to be retained with the SRIF-28 peak and 14% with the SRIF peak, suggesting no major conversion of SRIF-28 to SRIF. Des-(Ala1,Gly2)-N3-Ac-SRIF and dihydrosomatostatin were cleared more rapidly than SRIF. Clearance of SLI by the perfusate without the liver was 12-43% of liver clearance, depending on the peptide examined. These results support the hypothesis that aminopeptidase and endopeptidase activities are involved in SRIF clearance by the intact liver. The activities appear to function independently. The intrachain disulfide bond of SRIF may confer relative stability during its hepatic metabolism.

Bias in treatment assignment in controlled clinical trials

Controlled clinical trials of the treatment of acute myocardial infarction offer a unique opportunity for the study of the potential influence on outcome of bias in treatment assignment. A group of 145 papers was divided into those in which the randomization process was blinded (57 papers), those in which it may have been unblinded (45 papers), and those in which the controls were selected by a nonrandom process (43 papers). At least one prognostic variable was maldistributed (P less than 0.05) in 14.0 per cent of the blinded-randomization studies, in 26.7 per cent of the unblinded-randomization studies, and in 58.1 per cent of the nonrandomized studies. Differences in case-fatality rates between treatment and control groups (P less than 0.05) were found in 8.8 per cent of the blinded-randomization studies, 24.4 per cent of the unblinded-randomization studies, and 58.1 per cent of the nonrandomized studies. These data emphasize the importance of keeping those who recruit patients for clinical trials from suspecting which treatment will be assigned to the patient under consideration.

Deficiencies of clinical trials of alcohol withdrawal

Eighty-one therapeutic trials of alcohol withdrawal were found that have been published in English since 1954; controls were randomized in 29 (RCTs). Two thousand three hundred thirteen patients were randomized. Variable pretreatment description prevented estimates of delirium tremens and convulsion prevalence, but only four deaths were reported. Endpoints were thus entirely subjective in these moderately ill patients. Protocol quality of the RCTs was graded by a previously developed system for evaluating adequacy of descriptions, blinding, and essential measurements. Mean score obtained was .49 +/- .03 (1 SE). (A perfect paper would score 1.00.) Data presentations and statistical analyses scored .18 +/- .03. There was little evidence of improvement of scores over time. Papers lacked confidence intervals, proper handling of dropouts, and adequate details of side effects. In five RCTs, six comparisons showed that benzodiazepines are clearly superior to placebo (p less than .001), but conclusions about comparisons with other drugs were not possible. In none of eight "negative" comparisons was the probability of a type II error (beta) considered. Discovery of more effective symptomatic agents or methods of reducing the death rate will require more rigid protocols and analyses as well as larger studies to allow the use of more critical endpoints such as occurrence of delirium tremens, convulsions, or death.

Association of adrenocorticosteroid therapy and peptic-ulcer disease

We reexamined the association between corticosteroid therapy and subsequent peptic ulceration or gastrointestinal hemorrhage by pooling data from 71 controlled clinical trials in which patients were randomized to systemic corticosteroids (or ACTH) or to nonsteroid therapy. Of 3064 steroid-treated patients evaluated for peptic ulcer, 55 (1.8 per cent) had ulcers, as compared with 23 of 2897 controls (0.8 per cent) (relative risk, 2.3; 95 per cent confidence interval, 1.4 to 3.7). Of 3135 steroid-treated patients evaluated for gastrointestinal hemorrhage, 78 (2.5 per cent) had bleeding, as compared with 48 of 2976 controls (1.6 per cent) (relative risk, 1.5; 95 per cent confidence interval, 1.1 to 2.2). The incidence of ulcers varied directly with the dosage of steroids. When separate analyses were performed for studies that were double-blind, used only oral steroids, used only parenteral steroids, or excluded patients with a history of ulcer, the trend remained consistent but did not always reach statistical significance. This study strongly suggests that corticosteroids do increase the risk of peptic ulcers and gastrointestinal hemorrhage.

Sensitivity and specificity of clinical trials. Randomized v historical controls

The relative accuracy of randomized control trials (RCTs) and historical control trials (HCTs) in determining effective therapies has not been compared since there is no external verification of efficacy. We reviewed six therapies studied by both methods. Most HCTs concluded therapy was better than control, but few RCTs agreed. We calculated sensitivity and specificity for each type of trial by combining published results with all possible combinations of effectiveness. The sensitivity of HCTs was 0.80 to 1.00 (mean, 0.90) and specificity was 0.0 to 0.27 (mean, 0.11). The sensitivity of RCTs was 0.0 to 0.27 (mean, 0.12) and specificity was 0.67 to 1.00 (mean, 0.88). Defects of RCTs are more easily corrected than those of HCTs. Readers should consider trial design and the probability of errors when deciding how much credence to give to a clinical trial.

Physician attitudes toward applications of computer data base systems

To determine the attitudes of internists and surgeons to problems in clinical information management and to the use of a computer data base, a survey of 529 internists and 487 otolaryngologists was conducted. The survey group consisted of a national distribution of academic physicians, private practitioners, and residents. Results of 296 completed questionnaires were analyzed for this report. More than one third of all responders were not satisfied with their ability to keep up with new developments; less than 10% were very satisfied. Ninety percent thought that a computer data base would improve their access to information in the literature, and 85% thought it would improve their practice of medicine. Among internists, randomized control trials (RCTs) were viewed as clearly the most useful source of data. Otolaryngologists indicated that RCTs and patient registries were both very helpful. Most physicians indicated significantly greater preference for literature summary and patient registry features over probability estimation capabilities of a computer data base. Sources of data to be incorporated into a successful data base would be different for systems designed for internists and otolaryngologists.

A survey of clinical trials of antibiotic prophylaxis in colon surgery: evidence against further use of no-treatment controls

To evaluate the use of antibiotics given prophylactically of colon surgery, we examined 26 trials published from 1965 to 1980 in which patients given various antibiotic regiments were compared with controls given no antibiotic treatment. In 22 (85 per cent of these trials) antibiotics reduced postoperative wound infection (p less than 0.05 in 14). Combining the results of the trials published from 1965 to 1975 reveals a 95 per cent confidence interval from the true difference in infection rates of 14 +/- 6 per cent (36 per cent for control group vs. 22 per cent for treatment group) and the true difference in death rates of 6.7 +/- 4.4 per cent (11.2 per cent for control group vs 4.5 per cent for treatment group). Yet trials employing control groups given no treatment continue to be reported. Since the use of such controls is justified only when no effective alternative therapy exists, we believe that any further trials of antibiotic prophylaxis in colon surgery should employ a previously proved standard. However, steadily increasing efficacy of treatment means that comparisons of new therapies with standard therapies will become prohibitively expensive because of the large number of patients required.

Adult respiratory distress syndrome in Rocky Mountain spotted fever

A culture-proved case of Rocky Mountain spotted fever is described in which the typical findings of the adult respiratory distress syndrome developed. Rocky Mountain spotted fever should be added to the list of conditions associated with the adult respiratory distress syndrome. This case also illustrates the difficulties in distinguishing Rocky Mountain spotted fever from atypical measles.

Similar responsiveness of diabetic ketoacidosis to low-dose insulin by intramuscular injection and albumin-free infusion

We compared low-dose insulin regimens in a prospective randomized trial in 30 patients with diabetic ketoacidosis. One group received a loading dose of 0.44 U/kg body weight of regular insulin half intramuscularly and half intravenously followed by 7 U/h intramuscularly, whereas the other group received a loading dose of 0.44 U/kg intravenously followed immediately by a constant infusion of 7 U/h in albumin-free saline. The time for metabolic control of the ketoacidosis was not significantly different in the two groups. Five patients in each group developed mild hypokalemia (serum potassium, 3.0 to 3.4 meq/litre). No patient became hypoglycemic, and there were no deaths within the follow-up period (24 h). In the treatment of diabetic ketoacidosis, low doses of insulin administered by the priming dose-intermittent intramuscular route are as effective as the constant infusion method.

Clostridium botulinum type E toxin: effect of pH and method of purification on molecular weight

The toxin of Clostridium botulinum type E was isolated from intact cells and from toxic culture filtrates by column chromatography at three pH values, 4.5, 5.3, and 6.0. At pH 6.0 and 5.3, the isolated toxin was in a form with a molecular weight (MW) of 86,000. This toxin was homogeneous on polyacrylamide gel electrophoresis and gel filtration and had an optical density ratio, 280 nm/260 nm, greater than 2.0. It did not dissociate at higher pH levels, but was dissociated into nontoxic components of approximately 12,000 MW when reduced and alkylated in the presence of 6 M guanidine hydrochloride. At pH 4.5, smaller amounts of an impure toxic moiety with a MW of 12,000 were found. After storage for 6 months, the 86,000-MW moiety had lost 60% of its lethality. Gel filtration revealed that the bulk of the toxicity was associated with a component having a MW of 150,000. Toxic components with MW of 12,000 and over 200,000 were also found. The toxin appears to polymerize or aggregate when in a pure form, so that most, if not all, of the MW previously reported for the toxin may belong to different polymers of a monomer with a MW of 12,000 or less. Treatment of the 86,000-MW toxin with trypsin resulted in an 18- to 128-fold increase in lethality, but no detectable change in MW.