Selecting the language of the publications included in a meta-analysis: Is there a tower of babel bias?

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 Three hundred eighteen authors of published trials were asked whether they had participated in any unpublished RCTs. The 156 respondents reported 271 unpublished and 1041 published trials. Of the 178 completed unpublished RCTs with a trend specified, 26 (14%) favored the new therapy compared to 423 of 767 (55%) published reports (p less than 0.001).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.

Nonsteroidal anti-inflammatory drugs (NSAIDs) have well-known gastrointestinal and renal toxic reactions. Effects of NSAIDs on blood pressure are less appreciated. A meta-analysis was performed to determine the hypertensive effects of NSAIDs and rank them by magnitude of change in mean arterial pressure (MAP). A literature search of published English-language studies of NSAIDs and their effects on blood pressure was done. Studies were included if they met the following criteria: (1) the studies were intervention studies; (2) NSAIDs at any dose or aspirin at doses of 1.5 g/d or greater were included; (3) documentation of blood pressure was provided; and (4) the studies were 24 hours in duration. Studies were excluded if 20% or more of their participants dropped out or if the dose of antihypertensive drugs was adjusted while the subjects were taking NSAIDs. The major outcome was change in MAP while patients were receiving NSAIDs. Each NSAID arm was extracted from its trial. Information on possible confounders, including subject age, trial quality, amount of dietary salt intake, and whether study subjects were hypertensive or normotensive, was recorded. We calculated the average change in MAP on each NSAID, adjusting for confounders. Fifty-four studies with 123 NSAID treatment arms met inclusion criteria. The mean age of subjects was 46 years. Of the 1324 participants, 1213 subjects (92%) were hypertensive. The effects of NSAIDs on blood pressure were found solely in hypertensive subjects. Among these, the increase in MAP after adjusting for amount of dietary salt intake was 3.59 mm Hg for indomethacin (57 treatment arms), 374 mm Hg for naproxen (four arms), and 0.49 mm Hg for piroxicam (four arms). The MAP decreased by 2.59 mm Hg for placebo (10 arms), 0.83 mm Hg for ibuprofen (six arms), 1.76 mm Hg for aspirin (four arms), and 0.16 mm Hg for sulindac (23 arms). The effects on MAP by using placebo, sulindac, and aspirin were statistically significantly different from indomethacin. In short-term use, NSAIDs vary considerably in their effect on blood pressure. Of the drugs studied, indomethacin and naproxen were associated with the largest increases in blood pressure. The average effects of piroxicam, aspirin, ibuprofen, and sulindac were negligible.

To determine, by a quantitative meta-analysis of randomized trials, the effectiveness of strategies to improve patient compliance with screening, referral, and clinic appointments for health services that are provided at the time of the visit. Computerized searches of MEDLINE (1966 through 1990) were done using two search strategies: (1) (Patient Compliance OR Adhere* OR Dropout*) AND (Appointment*) AND (Screen* OR Follow* OR Refer*); and (2) (Patient Compliance OR Adhere* OR Dropout*) AND (Attend* OR Screen*) OR (Appointment*). A computerized search of PSYCHLIT was done with the terms Compliance AND Appointment*. In addition, the reference list of each retrieved article was reviewed and relevant citations retrieved. Only randomized trials with quantitative data concerning the effect of interventions to improve attendance at appointments for supervised administration of care were considered for detailed review. Studies of appointment keeping for self-administered treatments or tests were excluded. Two independent reviewers assessed each article for inclusion (kappa, for agreement, 0.66 for MEDLINE; 0.95 for PSYCHLIT) and validity (kappa, 0.62) using a priori criteria. Twenty-three (26%) of 88 relevant articles met all criteria. Data on study populations, interventions, and outcomes were extracted and analyzed using pooled odds ratios (ORs). The average rate of compliance with appointments was 58%. Mailed reminders and telephone prompts were consistently useful in reducing broken appointments (OR, 2.2; 95% confidence interval [CI], 1.7 to 2.9; and OR, 2.9, CI, 1.9 to 4.3, respectively). An "orientation statement" (OR, 2.9; CI, 1.5 to 5.6), "contracting" with patients (OR, 1.9; CI, 1.04 to 3.5), and prompts from physicians (OR, 1.6; CI, 1.4 to 2.0) showed positive effects as well. In clinic settings where kept appointments can be an accurate measure of patient compliance with health care interventions, broken appointments can be reduced by mail, telephone, or physician reminders; orienting patients to the clinic; or contracting with patients.