The study was conducted in a prepaid, closed-panel health maintenance organization (HMO) providing comprehensive medical care, including pharmacy, to more than 430,000 members in the Portland, Oregon, metropolitan area. Although the race of participants in this study was not recorded, the general population of this HMO's patients is 87% white and English-speaking. This study was reviewed and approved by our institutional review board.
Recruitment occurred between June 1997 and December 1998 and included the referral of dyspeptic patients by primary care providers, supplemented by the use of an automated records system that identified patients aged 18 or older who had been diagnosed with dyspepsia and who had a history of medication use for dyspepsia (histamine-2-receptor blocker antagonists [H2RA]). For the purpose of this study, dyspepsia was defined as a chronic upper abdominal pain, with or without relation to meals, for which no clear cause has been established, as evaluated by primary care professionals.18,19,20
Referrals included patients with dyspepsia for longer than 4 weeks but excluded those with symptoms limited to gastroesophageal reflux disease, irritable bowel syndrome, cholelithiasis, and angina. Patients were also excluded if they were pregnant or trying to conceive, had cancer diagnosed in the past 2 years, had been hospitalized within a month of recruitment, or had a known sensitivity to the study drugs.
Potential participants were sent a study description and were then called by a recruiter who confirmed interest and eligibility. Those who agreed to screening were scheduled for an H pylori test and sent a copy of the consent form. At the time of the test, potential participants were met by a project staff member, given an opportunity to ask any questions, and then asked to sign the consent form. Participants then completed a baseline symptom questionnaire and an H pylori test.
An additional 11 patients with dyspepsia were referred by gastroenterologists. These patients had tested positive for H pylori on the rapid urease test for Campylobacter-like organism performed during endoscopy. These patients signed the informed consent form and completed the study questionnaires but did not have the baseline urea breath test (UBT) because the presence of H pylori had already been confirmed.
All participants completed a symptom questionnaire rating the intensity of dyspepsia, heartburn, nausea, constipation, and diarrhea. The dyspepsia question was, “During the past 30 days, have you been bothered by stomach aches or pains? Stomach ache refers to all kinds of aches or pains in your stomach or belly,” with a graded scale of none, mild, moderate, or severe. Additional questions were asked on general health, pain interfering with usual daily activity (not at all, little bit, moderate, quite a bit, and extreme), and satisfaction with pharmacy services, “How satisfied were you with the pharmacy services you received for your stomach problem (very satisfied, satisfied, dissatisfied, very dissatisfied)?”
Testing for H pylori
was done using a carbon 14 UBT (Pytest; Tri-Med Specialties, Inc, Charlottesville, VA). To ensure an accurate test result, potential participants were screened, by phone and at the time of testing, for any antibiotic or bismuth use in the previous month, use of proton-pump inhibitors or sucralfate in the past 14 days, or use of H2RAs or antacids in the past 24 hours. This commonly used diagnostic test has a sensitivity of 97% and a specificity of 100%.21,22,23
Breath samples were collected in a Mylar balloon according to the manufacturer's protocol. Values were reported as positive if greater than 200 disintegrations per minute occurred.
The primary outcome measure was eradication of H pylori, as measured by a UBT at the 3-month follow-up. Secondary outcome measures included symptoms as reported on the symptom questionnaire and satisfaction with treatment.
Treatment consisted of a 7-day regimen of omeprazole, 20 mg daily; bismuth subsalicylate, 2 tablets 4 times a day with food; metronidazole, 250 mg 4 times a day with food; and tetracycline hydrochloride, 500 mg 4 times a day with food. This regimen is referred to as OBMT. These drugs were provided in a blister pack with each day's dose clearly indicated.
Participants who tested positive for H pylori were randomly assigned to either usual care or special counseling using a computer-generated random sequence. The participating pharmacies were provided with a supply of opaque randomization envelopes, and the pharmacists were trained to open the top envelope to determine the treatment assignment for each new research participant.
Patients in both groups received the same OBMT medication regimen. In the usual pharmacy care condition, patients met for 5 minutes with the dispensing pharmacist. The pharmacist described the proper protocol for taking the prescribed drugs, and the patients had the opportunity to ask questions. This procedure is consistent with the current standard of pharmacy practice for dispensing these drugs.
Participants assigned to the special intervention received a 15-minute counseling session with the pharmacist, including a detailed review of possible side effects, emphasis on the importance of completing the entire drug regimen, discussion about possible barriers to adherence and coping strategies, and encouragement to call the pharmacist in the event of any problems. In addition to this extended counseling session, the pharmacist also scheduled a follow-up telephone call with the patient 2 to 3 days after the start of therapy to check on adherence to the drug regimen. All of the participating pharmacists received a special 4-hour training session in counseling techniques.
Follow-up data collection
Participants in both groups were contacted by telephone 8 days after they started the medication regimen and were asked to report their adherence to the regimen and their current symptoms. All participants were contacted again 90 days after receiving the treatment drugs and asked to complete a second UBT and the symptom questionnaire. All data collectors were masked to treatment assignments.
Analyses and power calculation
Statistical analyses were conducted using statistical software (Statistical Analysis Software, version 6.12; SAS Institute, Inc, Cary, NC). Participants with missing follow-up data were not included in the analyses of follow-up data. The intended sample size for the study was 150 participants per treatment. This sample size would have been sufficient to detect a difference in eradication rates between treatments of 0.80 versus 0.65, with a P of 0.05 and a power of 90%. Given the cost of providing added counseling by pharmacists, we determined that this level of difference in eradication rates would be required to justify changing current practice patterns.