Exome sequencing and array-based comparative genomic hybridisation analysis of preferential 6-methylmercaptopurine producers.

Preferential conversion of azathioprine or 6-mercaptopurine into methylated metabolites is a major cause of thiopurine resistance. To seek potentially Mendelian causes of thiopurine hypermethylation, we recruited 12 individuals who exhibited extreme therapeutic resistance while taking azathioprine or 6-mercaptopurine and performed whole-exome sequencing (WES) and copy-number variant analysis by array-based comparative genomic hybridisation (aCGH). Exome-wide variant filtering highlighted four genes potentially associated with thiopurine metabolism (ENOSF1 and NFS1), transport (SLC17A4) or therapeutic action (RCC2). However, variants of each gene were found only in two or three patients, and it is unclear whether these genes could influence thiopurine hypermethylation. Analysis by aCGH did not identify any unusual or pathogenic copy-number variants. This suggests that if causative mutations for the hypermethylation phenotype exist they may be heterogeneous, occurring in several different genes, or they may lie within regulatory regions not captured by WES. Alternatively, hypermethylation may arise from the involvement of multiple genes with small effects. To test this hypothesis would require recruitment of large patient samples and application of genome-wide association studies.

Dosing of dabigatran etexilate in relation to renal function and drug interactions at a tertiary hospital.

BACKGROUND: Plasma concentrations of the anticoagulant dabigatran are correlated with clinical outcomes, and are affected by renal function, intestinal P-glycoprotein (P-gp) activity and stomach acidity. AIMS: To determine the adherence to dabigatran etexilate renal dosing guidelines, the frequency of co-prescription of potentially interacting drugs in patients on dabigatran, and how these related to dabigatran dosing. METHODS: A retrospective chart review of 204 patients discharged from a tertiary hospital on dabigatran etexilate over a 12-month period. Creatinine clearance, using the Cockcroft-Gault equation, was used as the surrogate of renal function in the 86 patients where this was calculable. RESULTS: Prescribed dabigatran etexilate dose rates in relation to creatinine clearance and the manufacturer's guidelines were classified as 'standard', 'low' and 'high' in 47% (40/86), 49% (42/86) and 5% (4/86) of patients respectively. Co-prescribed drugs that potentially interact with dabigatran etexilate were present in 75% (154/204) of patients and included strong P-gp inhibitors (16%, 32/204), proton-pump inhibitors (46%, 94/204) and anti-platelet drugs (47%, 95/204). Co-prescription of strong P-gp inhibitors was associated with the prescription of 'low' dose rates relative to renal function (P = 0.025). CONCLUSIONS: Few patients were dosed excessively in relation to creatinine clearance. Around 50% was prescribed with 'low' dose rates in relation to creatinine clearance, which because of the association with co-prescription of strong P-gp inhibitors may be clinically appropriate. Most patients were co-prescribed with drugs that potentially interact with dabigatran etexilate.

Evidence of interaction of CARD8 rs2043211 with NALP3 rs35829419 in Crohn's disease.

The location of CARD8 within an inflammatory bowel disease (IBD) locus and its role in the NALP3 inflammasome and as a nuclear factor (NF)kappaB inhibitor make it an attractive candidate risk gene for IBD. However, studies testing for the association of the CARD8 loss-of-function single-nucleotide polymorphism (SNP) rs2043211 with IBD have yielded mixed results. A recent study provided evidence that this discordance may result from an interaction of rs2043211 with loss-of-function variants in nucleotide-binding oligomerization domain protein 2 (NOD2) and a gain-of-function SNP (rs35829419) in NALP3. To confirm this interaction, we conducted a replication in an independent IBD sample set (n=1009 patients, n=517 controls). We found that the presence of the minor allele of rs2043211 with the major allele of rs35829419 conferred a protective effect against Crohn's disease (and vice versa), which intensified in the absence of NOD2 mutations (P(1,2/1,1)=0.009, odds ratio (OR)=0.66, 95% confidence interval (CI) (0.48-0.90); P(1,1/1,2)=0.015, OR=0.35, 95% CI (0.15-0.82)). We propose that these genotype combinations protect against gut inflammation by preventing the NALP3 inflammasome from producing excessive interleukin-1beta.

Evidence that glioma-associated oncogene homolog 1 is not a universal risk gene for inflammatory bowel disease in Caucasians.

The transcription factor glioma-associated oncogene homolog 1 (GLI1) has a central function in gastrointestinal tract development and homeostasis. A non-synonymous single-nucleotide polymorphism (SNP) (rs2228226; Q1100E) in GLI1, which impairs GLI1 function in vitro, has been proposed as a risk factor for inflammatory bowel disease (IBD). In this study, we assessed the cumulative evidence for association of GLI1 with IBD. New genotype data for rs2228226 from New Zealand (907 controls, 990 IBD patients) and Belgian Caucasian case-control data sets (312 controls, 1214 IBD patients) were combined with data from the National Institute of Diabetes and Digestive and Kidney Diseases and three previously studied Caucasian case-control data sets. Meta-analysis of rs2228226 did not detect any association with ulcerative colitis (UC) (P=0.09, odds ratio (OR)=1.07, 95% confidence interval (CI)=0.92-1.24), Crohn's disease (CD) (P=0.29, OR=1.06, 95% CI=0.93-1.21) or overall IBD (P=0.15, OR=1.05, 95% CI=0.92-1.19). Our analyses of rs2228226 suggest that GLI1 is not a significant risk factor for IBD in Caucasians.

Gender-stratified analysis of DLG5 R30Q in 4707 patients with Crohn disease and 4973 controls from 12 Caucasian cohorts.

BACKGROUND: DLG5 p.R30Q has been reported to be associated with Crohn disease (CD), but this association has not been replicated in most studies. A recent analysis of gender-stratified data from two case-control studies and two population cohorts found an association of DLG5 30Q with increased risk of CD in men but not in women and found differences between 30Q population frequencies for males and females. Male-female differences in population allele frequencies and male-specific risk could explain the difficulty in replicating the association with CD. METHODS: DLG5 R30Q genotype data were collected for patients with CD and controls from 11 studies that did not include gender-stratified allele counts in their published reports and tested for male-female frequency differences in controls and for case-control frequency differences in men and in women. RESULTS: The data showed no male-female allele frequency differences in controls. An exact conditional test gave marginal evidence that 30Q is associated with decreased risk of CD in women (p = 0.049, OR = 0.87, 95% CI 0.77 to 1.00). There was also a trend towards reduced 30Q frequencies in male patients with CD compared with male controls, but this was not significant at the 0.05 level (p = 0.058, OR = 0.87, 95% CI 0.74 to 1.01). When data from this study were combined with previously published, gender-stratified data, the 30Q allele was found to be associated with decreased risk of CD in women (p = 0.010, OR = 0.86, 95% CI 0.76 to 0.97), but not in men. CONCLUSION: DLG5 30Q is associated with a small reduction in risk of CD in women.

Confirmation of association of IRGM and NCF4 with ileal Crohn's disease in a population-based cohort.

Genome-wide association studies have identified PHOX2B, FAM92B, IRGM and NCF4 as candidate susceptibility factors for ileal Crohn's disease (CD). Here we sought to determine whether these genes were also associated with ileal CD in New Zealand Caucasians, as well as with ileocolonic CD, colonic CD and ulcerative colitis (UC). A total of 507 CD patients, 475 UC patients and 576 controls were genotyped for the single nucleotide polymorphisms rs16853571 (PHOX2B), rs4821544 (NCF4), rs13361189 and rs4958847 (IRGM), and rs8050910 (FAM92B). NCF4 and IRGM were significantly associated with ileal CD (P-value(rs4821544)=0.0090, odds ratio (OR)=1.425, 95% confidence interval (CI): 1.092-1.859; P-value(rs13361189)=0.0017, OR=1.942, 95% CI: 1.274-2.959; P-value(rs4958847)=0.0022, OR=1.767, 95% CI: 1.224-2.558), but not with other forms of inflammatory bowel disease (IBD). No association of PHOX2B or FAM92B with IBD was detected. Our study has demonstrated that IRGM and NCF4 are ileal-specific CD susceptibility factors in New Zealand Caucasians.

Randomised clinical trial: efficacy, safety and dosage of adjunctive allopurinol in azathioprine/mercaptopurine nonresponders (AAA Study).

BACKGROUND: Thiopurine hypermethylation towards 6-methylmercaptopurine (6MMP) instead of 6-thioguanine nucleotides (6TGN) is associated with inefficacy in patients with IBD. Allopurinol reverses such hypermethylation. AIMS: To prospectively determine efficacy of allopurinol-thiopurine combination and to compare 2 doses of allopurinol. DESIGN: In a multicentre, double-blind trial, patients with clinically active or steroid-dependent IBD and thiopurine shunting were randomised to 50 or 100 mg/d allopurinol and 25% of their screening thiopurine dose, which was subsequently optimised, aiming for 6TGN of 260-500 pmol/8x108 RBCs. The primary endpoint was steroid-free clinical remission at 24 weeks. RESULTS: Of 73 patients, 39 (53% [95% CI 42-65]) achieved steroid-free remission, (54% with 50 mg/d and 53% with 100 mg/d). 81% were able to discontinue steroids. Therapeutic 6TGN levels were achieved in both groups. Final thiopurine doses were lower with 100 mg/d allopurinol (P < 0.005). 6MMP: 6TGN ratio decreased from mean 64 to 4 (P < 0.001), being higher with 50 mg/d (6 ± 1.83) than for 100 mg/d ([1 ± 0.16], P = 0.003). Three patients on 50 mg/d failed to sustain low ratios at 24 weeks. Toxicity was minimal; three patients on 50 mg/d allopurinol developed transient leukopenia. Alanine aminotransferase concentrations decreased (P < 0.001) similarly in both arms. Faecal calprotectin levels at study end were lower in patients who achieved the primary endpoint (median 171 [85-541] vs 821[110-5892] ug/g, P = 0.03). CONCLUSIONS: Low-dose allopurinol-thiopurine combination safely reverses shunting and optimises 6TGN with associated improvement in disease activity. 100 mg/d allopurinol is preferable due to greater metabolite profile stability and lower thiopurine dose without additional toxicity.

Determination of in vivo absorption, metabolism, and transport of drugs by the human intestinal wall and liver with a novel perfusion technique.

BACKGROUND AND AIMS: The contribution of the gastrointestinal tract in comparison with the liver for the low and variable bioavailability of orally administered drugs is still poorly understood. Here we report on a new intestinal perfusion technique for the direct assessment of absorption, metabolism, and transport of drugs by the intestinal wall. METHODS: In 6 healthy volunteers a multilumen perfusion catheter was used to generate a 20-cm isolated jejunal segment that was perfused with 80 mg verapamil. Simultaneously, 5 mg [(2)H(7)]verapamil was given intravenously. Blood, perfusate, and bile samples were analyzed for parent verapamil and its major metabolites. RESULTS: The mean fraction of the verapamil dose absorbed from the 20-cm segment was 0.76 but substantial interindividual variability (0.51-0.96) was shown. Bioavailability was low (19.3%). The intestinal wall contributed to the same extent as the liver to extensive first-pass metabolism (mean extraction ratio, 0.49 versus 0.48). Substantial transport of verapamil metabolites from the systemic circulation via the enterocytes into the intestinal lumen was observed. Compared with biliary excretion, intestinal secretion into a 20-cm jejunal segment contributed to drug elimination to a similar extent. CONCLUSION: First-pass metabolism by the intestinal wall is extensive and contributes to the same extent as the liver to low bioavailability of some drugs such as verapamil. Moreover, intestinal secretion is as important as biliary excretion for the elimination of metabolites.

What is the evidence for once-daily aminoglycoside therapy?

Aminoglycosides are important antibacterial agents for the treatment of serious infection. Evidence suggests that high peak plasma concentrations must be achieved early in the course of treatment if these agents are to be effective, but prolonged high concentrations may cause ototoxicity and nephrotoxicity. Peak plasma concentrations of 6 to 10 mg/L and trough concentrations of less than 2 mg/L for gentamicin and tobramycin have been traditional goals of therapy. Extensive recent evidence from in vitro, animal and human studies suggests that these target concentrations need revision. Aminoglycosides display concentration-dependent bacterial killing, have a long postantibiotic effect, and induce adaptive resistance in Gram-negative bacteria. All of these factors support the use of larger doses of aminoglycosides that are given less frequently than conventional therapy. Studies in vitro support this approach, showing greater activity when aminoglycosides are given less frequently. Animal studies comparing different dosage intervals have shown varying results, with only a slight bias favouring the longer dosage interval. However, the short elimination half-lives for the drugs in animals limit the applicability of these models to humans. Importantly, there is convincing evidence in animal studies that nephrotoxicity and ototoxicity are both reduced when the same total daily dose is administered in less frequent doses. There have been at least 29 clinical trials comparing once-daily administration of aminoglycosides with conventional administration 2 to 4 times daily. In general, efficacy has not been shown to be different between regimens, although one trial showed an advantage for once-daily administration compared with administration 3 times daily. A small number of trials have shown less nephrotoxicity and ototoxicity with once-daily administration, leading several authors to suggest that there is sufficient evidence to warrant a change to once-daily administration of aminoglycosides. However, once-daily administration has not been well studied in the paediatric population, or in patients with renal failure or endocarditis, and cannot be recommended in these patients as yet. The choice of a 24-hour dosage interval is somewhat arbitrary, and the optimal interval may not necessarily be 24 hours. No studies have included dosage adjustment based on pharmacokinetic modelling methods, and the effect of this on treatment outcome needs to be assessed. The best method of administering aminoglycosides once daily is yet to be determined.

Once daily aminoglycoside therapy. Is it less toxic than multiple daily doses and how should it be monitored?

After 50 years of clinical experience with the aminoglycoside agents, there is continuing debate over the most appropriate administration regimen for these drugs. In recent years, once daily administration has been used increasingly, in the hope of both improving efficacy and reducing toxicity. At least 30 controlled clinical trials have compared once versus conventional multiple daily administration. Efficacy was assessed in some, but not all, studies using clinical and/or bacteriological cure. Toxicity was generally determined using rather nonsensitive end-points such as measurement of serum creatinine for nephrotoxicity and clinically detectable hearing loss for ototoxicity. The results of individual clinical trials and subsequent meta-analyses have been variable. However, 5 of 9 meta-analyses found clinical efficacy to be significantly better with once daily administration, and in 3 of the 9 there were significantly less nephrotoxicity with once daily administration. The results were not significant for ototoxicity in any of the meta-analyses. There is debate about how therapeutic drug monitoring should be performed, and whether it is still required with once daily administration. Previous experience with the aminoglycosides, especially in patients with impaired drug clearance caused by renal impairment, suggests that monitoring is still prudent. Results from the once daily administration trials appear to support this. Various methods of monitoring and dose adjustment have been proposed. The most common is to measure a 24-hour trough concentration and to adjust the dose to maintain the trough concentration below a value of 2, 1 or 0.5 mg/L. However, this method allows for greater total aminoglycoside exposure than has been permitted with conventional dosages, increasing the likelihood of toxicity in patients with impaired aminoglycoside clearance. Other methods measure drug concentrations at a time-point or points within the dose interval (when the concentration is still measurable), and adjust the dose according to concentration-time curve nomograms or to a target area under the concentration-time curve. This allows the use of higher doses in those with high drug clearance. Furthermore, in patients with impaired clearance, drug exposure is limited to the same extent as, or less than, that with conventional multiple daily administration. To date no controlled trials have compared methods of dose-individualisation. In summary, in addition to a slight overall improvement in efficacy, once daily administration has resulted in a small reduction in nephrotoxicity. In the studies using more sensitive measures of toxicity, the differences in toxicity were greater, strengthening the case for once daily administration. Therapeutic drug monitoring is probably required with once daily administration. Methods which use mid-dosage interval concentrations to gauge drug exposure would seem to be preferable over trough concentration measurement.

Aminoglycoside adaptive resistance: importance for effective dosage regimens.

There are various pharmacodynamic features of the aminoglycosides that are thought to contribute to the benefits of once-daily administration, of which the ability to induce adaptive resistance is the least understood and discussed. However, this may be the most important characteristic conferring increased efficacy with extended interval dose administration. Adaptive resistance describes a reversible refractoriness to the bactericidal effect of an antibacterial agent. It is well documented for the aminoglycosides but has also been seen with the quinolones. It does not appear to be caused by a genetic mutational change but rather by a protective phenotypic alteration in bacterial characteristics. This includes reversible down-regulation of the active transport of aminoglycosides into gram-negative bacteria. In vitro, animal and clinical studies have shown that marked adaptive resistance of gram-negative bacteria to aminoglycosides occurs within 1-2 hours of the first dose. The duration of adaptive resistance relates directly to the half-life of elimination of the aminoglycoside. With normal human aminoglycoside pharmacokinetics, the resistance may be maximal for up to 16 hours after a single dose of aminoglycoside, followed by partial return of bacterial susceptibility at 24 hours and complete recovery at around 40 hours. With conventional dosage regimens, second and subsequent doses of aminoglycoside are given at the time of maximal resistance and this practice is also likely to reinforce the resistance. Dose administration at 24 hour intervals, or longer, may increase efficacy by allowing time for adaptive resistance to reverse.

Review article: comparison of the pharmacokinetics, acid suppression and efficacy of proton pump inhibitors.

Proton pump inhibitors have dramatically influenced the management of acid-peptic disorders in recent years. They all have a broadly similar mechanism of action and are extensively metabolized in the liver via cytochromes P450 2C19 and 3A4. There is some variation in their potential for drug interactions due to differences in enzyme inhibition. Relatively few serious adverse effects have been reported for the proton pump inhibitors. Comparative studies of acid suppression suggest that lansoprazole and pantoprazole have a potency similar to that of omeprazole on a mg for mg basis; however, rabeprazole may have a greater potency than omeprazole. Lansoprazole and rabeprazole display a more rapid onset of maximal acid suppression than the other proton pump inhibitors. Comparative studies using proton pump inhibitors for the treatment of reflux oesophagitis, duodenal ulcer healing and Helicobacter pylori eradication show little overall difference in outcome between the proton pump inhibitors when used in their standard doses. Lansoprazole and rabeprazole provide earlier and better symptom relief than the other proton pump inhibitors in some studies of peptic ulcer treatment. The few studies of gastric ulcer treatment suggest that there is an advantage in using the proton pump inhibitors that have a higher standard daily dose.

Lansoprazole pharmacokinetics differ in patients with oesophagitis compared to healthy volunteers.

AIM: To compare the pharmacokinetics of lansoprazole in patients with reflux oesophagitis and in healthy volunteers, after a single dose and at steady-state. PATIENTS AND METHODS: A 30 mg dose of lansoprazole was administered orally daily for 7 days in eight healthy male volunteers aged 21-24 years, and in 16 patients aged 29-65 years with grade 2 or 3 reflux oesophagitis. The pharmacokinetics were assessed over the 24 h dose interval following the first dose and again after the 7th dose. RESULTS: Within both the patient and volunteers groups, there were no significant differences between day 1 and day 7 in any of the pharmacokinetic parameters including maximum concentration (Cmax), area under the concentration-time curve (AUC), and terminal half-life of elimination (t(1/2)). However, on both days 1 and 7, values were significantly higher in the patients than in the healthy volunteers. On day 7, Cmax was 1343 ng/mL in patients compared with 765 ng/mL in healthy volunteers, AUC was 3458 ng.h/mL vs. 1350 ng.h/mL and t(1/2) was 1.62 h vs. 0.90 h. CONCLUSION: The differences in results for the pharmacokinetics reflect reduced lansoprazole clearance in the patient group. Other research has not found a difference in pharmacokinetics when comparing healthy volunteers with patients with acid-related disorders. The difference in lansoprazole clearance in this study may be related to a variety of factors that are different in patients compared with young normal volunteers, such as age, gender, other drugs, and reduced general well-being.

Thiopurine S-methyltransferase (TPMT) genotype does not predict adverse drug reactions to thiopurine drugs in patients with inflammatory bowel disease.

BACKGROUND: Azathioprine and mercaptopurine (MP) are well established treatments for inflammatory bowel disease but they have severe adverse effects that prevent their use in some patients. The likelihood and type of adverse effect may relate to thiopurine methyltransferase (TPMT) enzyme activity and genotype. AIM: To compare the TPMT genotype frequencies in patients with inflammatory bowel disease who have had severe adverse effects to those who tolerate azathioprine or MP (controls). METHODS: Patients with inflammatory bowel disease who had been treated with azathioprine or MP in Christchurch between 1996 and 2002 were identified. Patients with adverse effects, and controls, were invited to provide a peripheral blood sample for analysis of TPMT genotype. The genotype frequencies were then compared between the two groups. RESULTS: Fifty-six patients were identified with adverse effects requiring cessation of therapy, of which 50 were genotyped. Reactions included allergic-type (25%), hepatitis (33%), nausea/vomiting (14%), bone marrow suppression (10%), pancreatitis (6%) and other (12%). Five of 50 patients with reactions had TPMT genotype *1/*3, one had *3/*3, and the rest had the wildtype genotype *1/*1. The patient with genotype *3/*3 had severe pancytopenia requiring hospitalization. Three of 50 controls had the *1/*3 genotype and the rest were *1/*1. CONCLUSIONS: The TPMT allele frequency in our population with inflammatory bowel disease is similar to that reported elsewhere. There was a slight trend for more frequent TPMT mutations in the patients with adverse reactions, but this was not statistically significant. Most patients with reactions did not have gene mutations.

A computer model of uterine contractions based on discrete contractile elements.

OBJECTIVE: To predict uterine contraction waveforms using a microcomputer-based model of uterine activity based on discrete contractile elements, varying the shape of the model, total number of cells, and pacemaker locations. METHODS: The model is a hollow ovoid composed of discrete contractile elements (cells) that propagate electrical impulses, generate tension, and have defined contracting and refractory periods. Each cell contacts eight surrounding cells and propagates impulses iteratively from cell to cell. Contraction pressure is the sum of the tension contributions by contracting cells. Sample contraction waveforms were generated based on various numbers of cells organized in ovoids with long:short axis ratios of 1:1, 3:2, and 2:1, with one or two pacemakers at varying positions. RESULTS: Contraction waveforms are altered by altering the shape of the matrix, but not by increasing the number of contractile elements. The vertical placement of the pacemaker has a dramatic effect on the shape and symmetry of contractions, including the development of patterns characteristic of "dysfunctional" uterine contractions. CONCLUSION: Abnormal uterine contraction patterns may result from pacemaker activity in unusual locations, such as mid-uterus. Further refinement of this computer model of uterine activity may contribute to a better understanding of the genesis of normal and abnormal intrauterine pressure waveforms and their relationship to the progress of labor.

A new technique to overcome failed second-trimester amniocentesis due to membrane tenting.

Membrane tenting during amniocentesis is a significant cause of dry taps, leading to immediate procedure failure or multiple needle insertions. Because of the increased risk of fetal loss and other minor complications reported with multiple taps, a means of accomplishing fluid retrieval on a single pass is desirable. We describe a new single-pass technique that involves further needle penetration into the posterior myometrium, under ultrasound guidance, physically displacing the obstructing membranes down the shaft away from the tip.

Diurnal variations in unconjugated and total plasma estriol levels in late normal pregnancy.

Diurnal variations in plasma unconjugated and total estriol were assessed in 11 third-trimester subjects with uncomplicated pregnancies. Commercially available 125I-labeled radioimmunoassay kits were used. Total plasma estriols reach a nadir during the hours of sleep (400 and 700 hours) which exceeds the episodic fluctuations seen from day to day or during a given 90-minute period. Plasma unconjugated estriol fluctuations over 24 hours did not significantly exceed our previously reported episodic fluctuation of 15.6 +/- 8.2%. The data are interpreted as showing no circadian rhythm, but reflecting, in the case of total plasma estriols, an effect of improved renal clearance during hours of rest. Plasma unconjugated estriol emerges as the test of choice in the monitoring of high-risk pregnancies.

Plasma estriol in the evaluation of third-trimester gestational age.

This study attempts to confirm previous reports of a clinically useful serum unconjugated estriol surge at 36 weeks' gestation. Although an apparently physiologic estriol surge occurred at 36 +/- 2.1 weeks in 25 of 32 patients, clinical reality makes weekly plasma sampling difficult. In individual cases, use of the "surge point" predicted gestational age within a 4-week range with only 66% accuracy, and potentially serious errors in dating occurred. Other biochemical (lecithin:sphingomyelin, phosphatidylglycerol) and sonographic methods are superior in resolving problems with dating gestational age in the third trimester.

Effect of search experience on sustained MEDLINE usage by students.

PURPOSE: While educators agree that medical students should learn to use MEDLINE for clinical application, there is a lack of consensus on an optimal level of exposure to this resource during training that will result in sustained usage. This study sought to identify the level of search experience (1) to increase the odds that the student searcher will continue to search MEDLINE in the absence of search assignments, and (2) to make an appreciable difference in the odds of retrieving items of relevance from the MEDLINE database. METHOD: Search frequencies of MEDLINE via the PaperChase interface by 184 fourth-year students (class of 1992) at the University of Michigan Medical School were analyzed using the log cross-product technique. The students were required to take the Comprehensive Clinical Assessment, an examination that included a search assignment, as they entered their fourth year of medical school. Their levels of MEDLINE use and their retrieval performances before the examination were compared with those achieved during the subsequent five months as fourth-year medical students. RESULTS: For those who searched an average of at least once a month during their first three years of medical school, there was a 7.38:1 chance that they would conduct three searches per month in the fourth year, compared with those who searched less frequently. The odds of retrieving at least one item of definite relevance were 8.27:1 for those who had searched at least one and one-half times per month before the search assignment. CONCLUSION: Searching once a month through the first few years of medical school provided an experience level that improved the odds that a student would continue to search MEDLINE: Data indicated that a history of a minimum of 1.5 online sessions per month increased the odds of retrieving relevant items to 8.27:1. Implications for educational strategy are clear.