Statistical controversies in clinical research: Value of adverse events relatedness to study treatment: analyses of data from randomized double-blind placebo-controlled clinical trials

Background

Collection and reporting of adverse events (AEs) and their relatedness to study treatment, known commonly as attribution, in clinical trials is mandated by regulatory agencies (the National Cancer Institute and the Food and Drug Administration). Attribution is assigned by the treating physician using judgment based on various factors including patient's baseline status, disease history, and comorbidity as well as knowledge about the safety profile of the study treatments. We evaluate the patterns of AE attribution (unrelated, unlikely, possibly, probably, and definitely related to the treatment) in treatment, symptom intervention (cancer patients) and cancer prevention (participants at high risk for cancer) setting.

Materials and methods

Nine multicenter placebo-controlled trials (two treatment, two symptom intervention, and five cancer prevention) were analysed separately (2155 patients). Frequency and severity of AEs were summarized by arm. Attribution and percentage of repeated AEs whose attribution changed overtime were summarized for the placebo arms. Percentage of physician over- or under-reporting of AE relatedness was calculated for the treatment arms using the placebo arm as the reference.

Results

Across all trials and settings, a very high proportion of AEs reported as related to treatment were classified as possibly related, a significant proportion of AEs in the placebo arm were incorrectly reported as related to treatment, and clinician-reported attribution over-estimated the rate of AEs related to treatment. Fatigue, nausea, vomiting, diarrhea, constipation, and neurosensory were the common AEs that were over reported by clinician as related to treatment.

Conclusions

These analyses demonstrate that assigning causality to AE is a complex and difficult process that produces unreliable and subjective data. In randomized double-blind placebo-controlled trials where data are available to objectively assess relatedness of AE to treatment, attribution assignment should be eliminated.

Favorable glycemic response of type 2 diabetics to low-calorie cranberry juice

Fruit and vegetable intake is typically low for type 2 diabetics, possibly due to a perceived adverse effect on glycemic control. Cranberry juice (CBJ) may represent an attractive means for increasing fruit intake and simultaneously affording positive health benefits. This single cross-over design compared metabolic responses of type 2 diabetics (n= 12) to unsweetened low-calorie CBJ (LCCBJ; 19 Cal/240 mL), carbohydrate sweetened normal calorie CBJ (NCCBJ; 120 Cal/240 mL), isocaloric low-calorie sugar water control (LCC), and isocaloric normal calorie sugar water control (NCC) interventions. CBJ flavonols and anthocyanins, and proanthocyanidins were quantified with HPLC, LC-MS, and MALDI-TOF that includes an original characterization of several large oligomeric proanthocyanidins. Blood glucose peaked 30 min postingestion after NCCBJ and NCC at 13.3 +/- 0.5 and 12.8 +/- 0.9 (mmol/L), and these responses were significantly greater than the LCCBJ and LCC peaks of 8.1 +/- 0.5 and 8.7 +/- 0.5, respectively. Differences in glycemic response remained significant 60 min, but not 120 min postingestion. Plasma insulin values 60 min postingestion for NCCBJ and NCC interventions were 140 +/- 19 and 151 +/- 18 (pmol/L), respectively, and significantly greater than the LCCBJ and LCC values of 56 +/- 10 and 54 +/- 10; differences were not significant 120 min postingestion. Metabolic responses within the 2 high and 2 low-calorie beverages were virtually identical; however, exposure to potentially beneficial nutrients was greater with CBJ. Relative to conventionally sweetened preparation, LCCBJ provides a favorable metabolic response and should be useful for promoting increased fruit consumption among type 2 diabetics or others wishing to limit carbohydrate intake.

Adenocarcinoma of the lung is strongly associated with cigarette smoking: further evidence from a prospective study of women

In a prospective cohort of 41,836 Iowa women aged 55-69 years with 13 years of follow-up from 1986 through 1998, the authors examined the association between cigarette smoking history and three common histologic subtypes of lung cancer (123 small cell, 115 squamous cell, and 234 adenocarcinoma). Using Cox proportional hazards and additive Poisson regression analysis, they estimated four epidemiologic measures of effect: age-adjusted incidence rate, relative risk, excess risk (or risk difference), and population attributable risk. Of the three major lung cancer subtypes, the excess risk for heavy smokers compared with never smokers was higher for adenocarcinoma (excess risk = 206) than for squamous cell (excess risk = 122) and small cell (excess risk = 104) carcinomas. Adenocarcinoma of the lung is more strongly associated with tobacco smoke exposure than previously recognized.

Physical activity in relation to cancer of the colon and rectum in a cohort of male smokers

We examined the association between occupational and leisure physical activity and colorectal cancer in a cohort of male smokers. Among the 29,133 men aged 50-69 years in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention study,152 colon and 104 rectal cancers were documented during up to 12 years of follow-up. For colon cancer, compared with sedentary workers, men in light occupational activity had a relative risk (RR) of 0.60 [95% confidence interval (CI), 0.34-1.04], whereas those in moderate/heavy activity had an RR of 0.45 (CI, 0.26-0.78; P for trend, 0.003). Subsite analysis revealed a significant association for moderate/heavy occupational activity in the distal colon (RR, 0.21; CI, 0.09-0.51) but not in the proximal colon (RR, 0.87; CI, 0.40-1.92). There was no significant association between leisure activity and colon cancer (active versus sedentary; RR, 0.82; CI, 0.59-1.13); however, the strongest inverse association was found among those most active in both work and leisure (RR, 0.33; CI, 0.16-0.71). For rectal cancer, there were risk reductions for those in light (RR, 0.71; CI, 0.36-1.37) and moderate/heavy occupational activity (RR, 0.50; CI, 0.26-0.97; P for trend, 0.04), and no association for leisure activity. These data provide evidence for a protective role of physical activity in colon and rectal cancer.

Helicobacter pylori seropositivity as a risk factor for pancreatic cancer

BACKGROUND

Pancreatic cancer is among the most fatal cancers worldwide and one for which few preventable risk factors have been established. Gastric carriage of Helicobacter pylori, particularly cytotoxin-associated gene-A-positive (CagA+) strains, is known to be a risk factor for peptic ulcer disease and gastric cancer and may have a similar etiologic relationship with pancreatic cancer.

METHODS

We investigated the association of H. pylori carriage and exocrine pancreatic cancer in a nested case-control study within the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study cohort of 29 133 male Finnish smokers aged 50-69 years at baseline. Case subjects (n = 121) were matched on date of baseline serum collection, study center, age, trial intervention, and completion of the dietary questionnaire to 226 control subjects who were alive at the time the matching case subject was diagnosed and who remained free of cancer, during up to 10 years of follow-up. Levels of immunoglobulin G antibodies to H. pylori whole-cell and CagA+ antigens from stored baseline serum were measured by enzyme-linked immunosorbent assay. Smoking-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by use of conditional logistic regression. Statistical tests were two-sided.

RESULTS

Seroprevalence of H. pylori was 82% and 73% among case and control subjects, respectively. Compared with seronegative subjects, those with H. pylori or CagA+ strains were at statistically significantly elevated risk of pancreatic cancer (OR = 1.87 [95% CI = 1.05 to 3.34]; OR = 2.01 [95% CI = 1.09 to 3.70], respectively).

CONCLUSIONS

Our findings support a possible role for H. pylori carriage in the development of exocrine pancreatic cancer.

Fecal calprotectin levels predict colorectal inflammation among patients with chronic diarrhea referred for colonoscopy

OBJECTIVE

Chronic diarrhea is a relatively common condition with multiple diverse etiologies. Stool testing may serve as a diagnostic aid to discriminate the presence or absence of organic pathology, such as colorectal inflammation. Calprotectin (a leukocyte-derived protein) and hemoglobin can be measured quantitatively from stool and represent candidate inflammation biomarkers. The aim of this study was to assess and compare the screening performance of fecal calprotectin and fecal hemoglobin among colonoscopy referral patients with chronic diarrhea of unknown origin or chronic colitis of unknown activity.

METHODS

All subjects were identified prospectively and each submitted a single stool sample before purgation. Fecal calprotectin (PhiCal; Nycomed Pharma, Oslo, Norway) and fecal hemoglobin (HemoQuant; Mayo Medical Laboratories, Rochester, MN) assays were performed in separate laboratories by masked technicians. Colonoscopic and histological findings served as criterion standards for establishing the presence or absence of colorectal inflammation.

RESULTS

Among 110 subjects who provided complete fecal assay data, 29 (26%) had and 81 (74%) did not have colorectal inflammation. Increased fecal calprotectin levels were significantly (p = 0.0001) associated with the presence of colorectal inflammation, whereas fecal hemoglobin levels were not (p = 0.61). Direct comparison of the fecal assays revealed that calprotectin was a more sensitive biomarker for colorectal inflammation at all specificity levels (p = 0.0001).

CONCLUSIONS

In this study of colonoscopy referral patients, colorectal inflammation was reflected by fecal calprotectin but not by fecal hemoglobin levels. Assay of fecal calprotectin holds promise as a triage tool to identify inflammatory causes of chronic diarrhea.

Fecal calprotectin levels predict colorectal inflammation among patients with chronic diarrhea referred for colonoscopy

OBJECTIVE

Chronic diarrhea is a relatively common condition with multiple diverse etiologies. Stool testing may serve as a diagnostic aid to discriminate the presence or absence of organic pathology, such as colorectal inflammation. Calprotectin (a leukocyte-derived protein) and hemoglobin can be measured quantitatively from stool and represent candidate inflammation biomarkers. The aim of this study was to assess and compare the screening performance of fecal calprotectin and fecal hemoglobin among colonoscopy referral patients with chronic diarrhea of unknown origin or chronic colitis of unknown activity.

METHODS

All subjects were identified prospectively and each submitted a single stool sample before purgation. Fecal calprotectin (PhiCal; Nycomed Pharma, Oslo, Norway) and fecal hemoglobin (HemoQuant; Mayo Medical Laboratories, Rochester, MN) assays were performed in separate laboratories by masked technicians. Colonoscopic and histological findings served as criterion standards for establishing the presence or absence of colorectal inflammation.

RESULTS

Among 110 subjects who provided complete fecal assay data, 29 (26%) had and 81 (74%) did not have colorectal inflammation. Increased fecal calprotectin levels were significantly (p = 0.0001) associated with the presence of colorectal inflammation, whereas fecal hemoglobin levels were not (p = 0.61). Direct comparison of the fecal assays revealed that calprotectin was a more sensitive biomarker for colorectal inflammation at all specificity levels (p = 0.0001).

CONCLUSIONS

In this study of colonoscopy referral patients, colorectal inflammation was reflected by fecal calprotectin but not by fecal hemoglobin levels. Assay of fecal calprotectin holds promise as a triage tool to identify inflammatory causes of chronic diarrhea.

Adenocarcinoma of the esophagus: risk factors and prevention

Esophageal cancer, with an estimated number of 12,300 new cases in the year 2000, is relatively uncommon in the United States but produces a high number of annual deaths, estimated at 12,100. Moreover, the incidence of the adenocarcinoma histologic type of esophageal cancer has been rising over the past two decades. Identification of risk factors could lead to primary prevention, as well as earlier diagnosis, treatment, and increased survival. Multiple risk factors are associated with the development of esophageal adenocarcinoma. These include Barrett's esophagus, acid peptic disorders, motor disorders of the esophagus, other malignancies, medications, environmental exposures, diet, and nutrition. However, no one particular risk factor is responsible for the rising incidence of esophageal cancer. Several preventive strategies are under investigation using such agents as nonsteroidal anti-inflammatory drugs (NSAIDs), selenium, alpha-difluoromethylornithine (DFMO), and retinoids. As we gain more insight into the biology of this disease, other risk factors will hopefully be identified that will enable us to develop effective prevention strategies and, thus, reverse the current rising incidence of esophageal carcinoma.

Immunodiscrimination of colorectal neoplasia using MUC1 antibodies: discrepant findings in tissue versus stool

Colorectal tumor-associated antigens are attractive targets for novel stool-screening assays. MUC1, a glycoprotein antigen, is aberrantly expressed in transformed colorectal mucosa and represents a candidate fecal biomarker. In this study, tissue staining and stool testing were performed to further clarify the discriminant potential of MUC1 in markedly different biologic media. One anti-MUC1 monoclonal antibody (MA5) was used for immunohistochemistry and two commercially available MUC1 assay kits (ELSA-CA 15-3 and Truquant BR) were used for stool detection. On tissue staining, MUC1 expression was strong in 40/40 (100%) adenocarcinomas, moderate in 42/55 (76%) adenomas, faint in 8/28 (29%) juxtatumoral mucosa specimens, and absent in 15/15 (0%) nonadjacent mucosa specimens. Conversely MUC1 levels in stool testing did not differ between colorectal cancer cases (N = 14) and controls (N = 14). Based on these results, MUC1 appears to be a functional tumor biomarker in colorectal tissue but not in stool. Bacterial metabolism within stool may unmask the core antigen of MUC1 and account for this discordance in immunoreactivity.

Pervasive occult gastrointestinal bleeding in an Alaska native population with prevalent iron deficiency. Role of Helicobacter pylori gastritis

OBJECTIVE

To confirm prevalent iron deficiency among Yupik Eskimos living in Alaska and to explore the frequency of and potential lesions accounting for occult gastrointestinal bleeding.

DESIGN

Descriptive survey.

SETTING

Rural Arctic community.

SUBJECTS

A total of 140 adult volunteers from 3 villages in the Yukon-Kuskokwim Delta region of western Alaska.

MAIN OUTCOME MEASURES

Daily iron intake, hematologic and biochemical indexes of iron status, fecal hemoglobin levels, stool parasites, and endoscopic findings.

RESULTS

While dietary iron intake by Yupiks was similar to that of a reference population, iron deficiency prevalence was increased 13-fold in Yupik men and 4-fold in Yupik women. Fecal hemoglobin levels were elevated in 90% of subjects contrasted with only 4% of a reference group; median levels were 5.9 and 0.5 mg of hemoglobin per gram of stool, respectively. Among 70 Yupik subjects with elevated fecal hemoglobin levels who had endoscopy performed, 68 (97%) had an abnormal gastric appearance consisting of erythema, mucosal thickening, diffuse mucosal hemorrhages, erosions, or ulcerations. Gastric biopsies revealed chronic active gastritis with associated Helicobacter pylori in 68 (99%) of 69. No other hemorrhagic gastrointestinal disease was detected.

CONCLUSIONS

Based on this study sample, occult gastrointestinal bleeding appears to be pervasive in the Yupik population and likely underlies the prevalent iron deficiency. An atypical hemorrhagic gastritis associated with H pylori infection is present almost universally and may represent the bleeding source.

Correlation of trunk muscle strength with age in children 5 to 18 years old

OBJECTIVE

To determine the correlation of trunk muscle strength with age in children and the effect of gender on changes in trunk muscle strength.

MATERIAL AND METHODS

Strength of the back extensors and back flexors was evaluated in 137 healthy boys and 109 healthy girls who were 5 to 18 years old. Anthropometric determinations of height and weight were done in all study subjects. Maximal back muscle strength was measured with an isometric dynamometer (BID-2000). Standardized positioning techniques were used to allow comparison with the follow-up evaluations.

RESULTS

Regression analysis revealed significant increases in trunk strength with increasing age; the strength of the boys began diverging from that of the girls at age 9 to 10 years. Back extensor strength also increased with increasing height and weight, and expected differences were noted between boys and girls. Strength levels ranged from 80 to 774 newtons (18 to 174 lb). Mean back extensor strength in 2-year age intervals ranged from 153 newtons (34.4 lb) in girls 5 to 7 years old to 504 newtons (113.3 lb) in boys 16 to 18 years of age. (Each newton is equal to 0.2248 lb.)

CONCLUSION

Results of this study demonstrated that age, height, and weight are all important predictors of trunk strength in children, and the effects of these factors are modulated by gender.

Muscle strength in osteoporotic versus normal women

Strong back muscles contribute to good posture and skeletal support. Osteoporosis, being a metabolic bone disease, should not affect muscle strength. In this study we were interested in comparing the back extensor strength (BES) of osteoporotic and normal women. Fifty-five women ages 40-85 years who had a documented diagnosis of osteoporosis and were referred for initiation of proper exercise programs were included in our study after meeting the inclusion criteria. They all had evaluation of their posture, back and upper extremity strength, and physical activity score through our Rehabilitation of Osteoporosis Program--Exercise (ROPE). In addition, to avoid the interference of pain on application of maximal effort, we did not include subjects with acute back pain or those who experienced back pain with maximal effort during the testing trial. BES for osteoporotic women ranged from 16 to 65 lb (mean +/- SD, 36.5 +/- 15.5) for ages 40-59 years, 9 to 55 lb (mean +/- SD, 29.9 +/- 10.6) for ages 60-69 years, 6 to 52 lb (mean +/- SD, 24.3 +/- 10.2) for ages 70-79 years, and 17 to 27 lb (mean +/- SD, 21.2 +/- 4.2) for ages 80 years or older. Comparison of these data with the BES of 25 normal women, with statistical adjustment for age, demonstrated that the osteoporotic women had significantly lower BES than the normal women. A longitudinal study of a larger group of women would be of great interest for clarifying whether the weakness of back extensors precedes and, indeed, contributes to compression fractures of the spine.

A useful technique for measurement of back strength in osteoporotic and elderly patients

Improvement of back extensor strength (BES) can be used as a therapeutic method for patients with chronic back pain and osteoporosis. The method of evaluation must be reliable and accurate without compromising the condition of the patient. We report the development of a back isometric dynamometer (BID-2000) designed specifically by two of us to address these concerns in elderly patients with osteopenia or osteoporosis. As the demographics of the general population change, increasing numbers of patients will need the type of monitoring that the BID-2000 provides. Aging has been shown to cause a reduction in the number of functional muscle motor units. To examine this effect on BES, we tested 50 normal, healthy women who were 30 to 79 years old. Proper testing of BES in patients with fragile vertebrae should include isometric measurement in the prone position, maneuverability of the device to allow comfortable positioning of the patient, and simplicity of technique to minimize repetitious performance of maximal contraction. The BID-2000 incorporates each of these features and also provides meaningful results inexpensively. The device offers a safe, reliable (coefficient of variation = 2.33%), and valid (P = 0.001) method of evaluation. The results of our study demonstrated moderate, steady reduction of BES with increasing age and with each successive decade.